Abstract
Free full text
Pathogenesis of Proteus mirabilis Infection
Abstract
Proteus mirabilis, a Gram-negative rod-shaped bacterium most noted for its swarming motility and urease activity, frequently causes catheter-associated urinary tract infections (CAUTI) that are often polymicrobial. These infections may be accompanied by urolithiasis, development of bladder or kidney stones due to alkalinization of urine from urease-catalyzed urea hydrolysis. Adherence of the bacterium to epithelial and catheter surfaces is mediated by 17 different fimbriae, most notably MR/P fimbriae. Repressors of motility are often encoded by these fimbrial operons. Motility is mediated by flagella encoded on a single contiguous 54 kb chromosomal sequence. On agar plates, P. mirabilis undergoes a morphological conversion to a filamentous swarmer cell expressing hundreds of flagella. When swarms from different strains meet, a line of demarcation, a “Dienes line”, develops due to the killing action of each strain’s type VI secretion system. During infection, histological damage is caused by cytotoxins including hemolysin and a variety of proteases, some autotransported. The pathogenesis of infection, including assessment of individual genes or global screens for virulence or fitness factors has been assessed in murine models of ascending UTI or CAUTI using both single-species and polymicrobial models. Global gene expression studies carried out in culture and in the murine model have revealed the unique metabolism of this bacterium. Vaccines, using MR/P fimbria and its adhesin, MrpH, have been shown to be efficacious in the murine model. A comprehensive review of factors associated with urinary tract infection is presented, encompassing both historical perspectives and current advances.
I. Introduction
Proteus mirabilis, a Gram-negative rod-shaped bacterium, is well-known for its urease production and distinctive ability to differentiate into elongated swarm cells and characteristic bull’s-eye pattern of motility on agar plates. P. mirabilis belongs to the class Gammaproteobacteria, and has long been recognized as a member of the order Enterobacteriales, family Enterobacteriaceae. However, one group recently created a reconstructed phylogenetic tree based on shared core proteins, ribosomal proteins, and four multilocus sequence analysis proteins, and has proposed that the order Enterobacteriales be reclassified, placing Proteus within a new Morganellaceae family (1).
P. mirabilis can be found in a wide variety of environments, including soil, water sources, and sewage, but it is predominantly a commensal of the gastrointestinal tracts of humans and animals (2, 3). While the bacterium is capable of causing a variety of human infections, including those of wounds, the eye, the gastrointestinal tract, and the urinary tract, it is most noted for infections of the catheterized urinary tract, known as catheter-associated urinary tract infections (CAUTI) (4–9). These infections are common in long-term catheterized patients, such as those who reside in nursing homes and chronic care facilities, and may be of particular danger to spinal cord injury patients (10). Urinary tract infections (UTIs) and CAUTIs involving P. mirabilis are typically complicated by the formation of bladder and kidney stones (urolithiasis) and permanent renal damage (11–13), and may progress to bacteremia and sepsis (14, 15). Indeed, CAUTI is the most common source of bacteremia in nursing homes, bacteremia involving P. mirabilis most frequently occurs following UTI or CAUTI compared to other sources of infection, and bacteremia and sepsis due to P. mirabilis carry a high mortality rate (14–17). CAUTIs are also often polymicrobial (8, 17), and P. mirabilis is one of the most common organisms present during polymicrobial urine colonization and infection (4, 9) (Table 1).
Table 1
Single-Species | Dual-Species | ||||
---|---|---|---|---|---|
Microorganism | a Total Urine Cultures | b Urine Cultures | c % | d Urine Cultures | e % |
Proteus mirabilis | 48 | 28 | 22 | 20 | 35 |
Enterococcus spp. | 38 | 15 | 12 | 23 | 40 |
Escherichia coli | 37 | 23 | 18 | 14 | 25 |
Pseudomonas aeruginosa | 34 | 18 | 14 | 16 | 28 |
Staphylococcus aureus | 20 | 11 | 9 | 9 | 16 |
Klebsiella pneumoniae | 14 | 6 | 5 | 8 | 14 |
Citrobacter spp. | 9 | 3 | 2 | 6 | 11 |
Morganella morganii | 7 | 2 | 2 | 5 | 9 |
Providencia stuartii | 7 | 3 | 2 | 4 | 7 |
Yeast | 6 | 3 | 2 | 3 | 5 |
Acinetobacter baumanii | 4 | 2 | 2 | 2 | 3 |
Enterobacter spp. | 5 | 5 | 4 | 0 | 0 |
Serratia marscesens | 1 | 1 | 1 | 0 | 0 |
Corynobacterium spp. | 1 | 1 | 1 | 0 | 0 |
Other | 8 | 4 | 3 | 4 | 7 |
Total | 182 | 125 | 100 | 57 | 100 |
P. mirabilis is an agent of catheter biofilm formation, quickly fouling the surface of a newly inserted urinary catheter. Surface organelles such as fimbriae and other adhesins appear to play a significant role in this process. The enzyme urease also contributes dramatically to this process. Urea, our means of eliminating excess nitrogen, is present in high concentrations in urine (~400 mM), is the substrate of urease, and is hydrolyzed to CO2 and NH3. The liberated ammonia raises the pH of the urine and initiates the precipitation of otherwise soluble polyvalent anions and cations present in urine. The result is urolithiasis, the formation of struvite (MgNH3PO4) or apatite (CaPO4) stones. These crystals can form on and within the lumen of catheters, blocking urine flow and necessitating catheter removal and replacement. Stones may also form in the renal tubules or renal pelvis, causing inflammation and often requiring surgical removal. This bacterium is capable of invading bladder epithelial cells, and produces a variety of cytotoxins that damage the epithelium, leading to significant histopathology.
Over the last four decades, our laboratory and others have developed a variety of approaches to study the virulence of P. mirabilis. Importantly, two primary models of ascending urinary tract infections have been developed in mice, in which organisms are transurethrally delivered to the bladder and bacteria can be enumerated in the urine, bladder, and kidneys following a suitable interval, depending on the experimental question asked. One model requires no manipulation of the murine host, yet it still represents a model of complicated urinary tract infection for P. mirabilis as urolithiasis quickly occurs, offering a substrate for bacterial colonization and sometimes blockage of proper urine flow along with significant pathology. In a more recent modification of this model, a small silicone catheter segment is introduced into the bladder, providing an immediate substrate for bacterial colonization. This latter model may more closely reflect the catheterized urinary tract and CAUTI in humans. It has also been demonstrated as an excellent model for polymicrobial bacteriuria, a common feature of CAUTI.
The advent of straightforward genome sequencing and comparative genomics studies has provided a view of the heterogeneity of the species, and has allowed for a more global understanding of metabolic pathways, regulatory schemes, and virulence determinants present in the bacterium. The importance of individual genes and operons must be addressed experimentally. Nevertheless, sequencing information allows for generation of testable hypotheses. This information has also opened the door for transcriptomic studies, both under in vitro conditions and during experimental infection. While not as facile as in other genera, allelic exchange methods for specific gene mutation may be routinely conducted in P. mirabilis strains. In addition, genome-wide transposon mutagenesis is also straightforward and has facilitated a variety of screens, including signature-tagged mutagenesis (STM) and transposon insertion-site sequencing (Tn-Seq).
Clearly, P. mirabilis possesses an impressive arsenal of virulence factors (Fig. 1). Urease is a critical feature of this species, but the bacterium also expresses a startling number of fimbriae and other adhesins. The most well studied fimbria is the mannose-resistant Proteus-like (MR/P) fimbria, whose expression is phase-variable. The mrp operon also encodes a non-structural protein, MrpJ, which directly represses flagella synthesis, thus shutting down motility while the bacterium adheres, and flagella themselves contribute to pathogenesis. As well, a variety of potent toxins and proteases compound virulence. Similar to other members of the Enterobacteriaceae, P. mirabilis carries numerous secretion systems including types I, III, IV, V, and VI. To provide co-factors and regulate intracellular metabolism, P. mirabilis also carries a myriad of ion importers and exporters. Lastly, the bacterium carries an integrative and conjugative element named ICEPm1 that can self-replicate and self-transfer to other strains and species, transferring virulence genes and antibiotic resistance.
There are no currently licensed vaccines available for this organism, and multidrug-resistant isolates are becoming increasingly common. Thus, efforts to generate effective vaccines or therapeutic treatments are warranted. The majority of experimental vaccine studies have targeted P. mirabilis fimbriae, and successes have been described in the murine model for the MR/P fimbriae themselves as well as the tip adhesin of the fimbria, MrpH. However, none of the experimental vaccines have provided complete protection against infection, and additional targets remain to be explored.
In this chapter, we will focus on the current state of knowledge regarding how this fascinating bacterium is capable of infecting the urinary tract and causing disease. We will summarize studies that have made a clear connection between specific genes, gene clusters, and operons and their role in pathogenesis. While there is an enormous literature on the fascinating phenotype of swarming motility, this has been extensively summarized elsewhere (3, 18–20). We will therefore address only those studies that relate to flagellum-mediated motility and virulence in the murine model.
II. Host Interactions
A. The catheterized urinary tract
Numerous bacteria colonize the periurethral area, but are generally prevented from establishing an infection of the urinary tract by the regular flushing of the urethra during micturition (the passing of urine). Urothelial cells lining the bladder also provide a barrier against bacterial adhesion and invasion, partly due to a coating of glycosaminoglycan mucin as well as their role in the innate immune response.
When an indwelling urinary catheter is inserted, it bypasses many of the natural host defenses against urinary tract infection. In contrast to normal micturition and flushing of the urethra at regular intervals, the catheter allows for continuous bladder drainage as urine accumulates, which is not of sufficient volume or force to effectively flush the urinary tract. The design of the traditional Foley catheter and draining tubing also causes retention of 10–100 ml of urine within the bladder, providing a reservoir for bacteria to replicate (21). The urothelium is only 3 to 4 cell layers thick, and is easily damaged during rushed or multiple attempts to place a catheter (22). Any rough surfaces on the catheter can also facilitate bacterial biofilm formation or cause damage to the urethral lining and bleeding during insertion or withdrawal. Furthermore, improper insertion of the catheter or inflation of the balloon can create mucosal and submucosal tears (22). All of these potential complications of catheter insertion provide P. mirabilis with additional sites of attachment, nutrients, and routes for establishing more severe infection. The presence of an indwelling urinary catheter also elicits a robust inflammatory response, both in humans and in experimental animal models, and the resulting accumulation of fibrinogen on the catheter surface can provide bacteria with an ideal substrate for attachment (23–25).
Approximately 50% of individuals with long-term catheterization (>28 days) experience catheter blockage from crystalline deposits, and the urease activity of P. mirabilis is the most common cause of this blockage (5, 26, 27) (link to urease section) (Fig. 2). Urease catalyzes the hydrolysis of urea to ammonia and carbon dioxide, thereby raising urinary pH. As urine pH increases, calcium and magnesium phosphates begin to precipitate out of solution, leading to the formation of struvite (magnesium ammonium phosphate) and apatite (calcium phosphate) crystals (11, 28). The urinary pH at which precipitation occurs is referred to as the nucleation pH (pHn), and if the pH of voided urine (pHv) falls within the range of the pHn (common during colonization with urease-positive organisms), crystallization and catheter blockage are likely to occur (29).
The struvite and apatite crystals that form during P. mirabilis colonization deposit on the catheter surface, facilitating the formation of crystalline biofilms (link to biofilm section). Indeed, it has been experimentally determined that P. mirabilis typically begins establishing crystalline biofilms on a urine-bathed catheter when urinary pH reaches pHn (30). Adherence to the catheter and the formation of crystalline biofilms provides protection from the action of host cells recruited to the site of infection, particularly neutrophils. Crystalline biofilms formed by P. mirabilis can also result in blockage of the catheter lumen, which obstructs urine flow and may cause reflux of infected urine to the kidneys (31, 32). The lifestyle of P. mirabilis within the catheterized urinary tract is depicted in Fig. 3.
B. Urolithiasis
The urease of P. mirabilis is unambiguously associated with the development of infection-induced stone formation, known as urolithiasis (5, 11, 33–36) (Fig. 4). Indeed, Proteus species have been isolated in 70% of cases of bacteria-induced stone formation (37). The cytoplasmic enzyme urease catalyzes the hydrolysis of urea, the nitrogenous waste product of mammals which is maintained normally at 400–500 mM in human urine. Ammonia, generated from the breakdown of urea, results in a dramatic elevation of the pH of urine, and normally soluble polyvalent anions and cations precipitate at high pH to form struvite (MgNH4PO46H2O) and carbonate apatite [Ca10(PO4)6CO3]. In vitro studies inoculating urine with P. mirabilis demonstrated that increasing the concentration of Mg2+, Ca2+, or PO43− ions intensified the magnitude of crystallization (38).
Experimental evidence for the involvement of urease in P. mirabilis pathogenicity was provided by infection of mice via transurethral inoculation into the bladder (39, 40). Separate groups of mice were inoculated with the wild-type strain or a urease-negative mutant of P. mirabilis, and struvite stones were found in the renal pelvis of mice infected with the wild-type strain one week after inoculation. Quantitatively, 12 of 39 mice (31%) developed stones after one week and 8 of 20 mice (40%) developed stones after two weeks, while no stones were found in the 38 mice infected with the urease-negative mutant. A caveat to the study was that the urease-negative mutant did not colonize the mouse urinary tract to the same extent as the wild-type strain. However, follow-up studies with higher inocula of the urease-negative mutant construct still failed to result in development of urolithiasis.
Because P. mirabilis is often part of a polymicrobial infection, the impact of polymicrobial colonization on urease activity and urolithiasis in a murine CAUTI model has been examined for coinfection of P. mirabilis and its common CAUTI partner Providencia stuartii (25, 41). Coinfection resulted in bacterial loads that were similar to monoinfections with either pathogen. However, there was a significantly increased incidence of urolithiasis and bacteremia during coinfection. Coinfection was also accompanied by a significant increase in urease activity that was manifested in a synergistic manner (link to urease section).
Adding to the mechanism of stone formation, it has been noted that GFP-expressing P. mirabilis could be observed within the matrix of urinary stones in mice experimentally infected with P. mirabilis (42). Genome sequences of P. mirabilis have also been detected in urinary calculi by PCR (43). Furthermore, Schaffer and colleagues (44) found, using confocal microscopy, that P. mirabilis formed extracellular clusters in the bladder lumen that served as the basis for focused mineral deposition, consistent with nascent stone formation (Fig. 5).
C. Internalization, cytotoxicity, and histopathology
The ability of P. mirabilis to invade and lyse host cells has been explored for decades, and has been demonstrated to contribute to infection progression and severity of disease in animal models. The level of host cell invasion and cytotoxicity achieved by P. mirabilis in vitro varies dramatically by the bacterial strain being tested, growth phase, and morphology of the bacteria (for instance, log phase growth versus stationary phase and vegetative cells versus differentiated swarm cells), host cell line, multiplicity of infection, pH of the medium, and the duration of the experiment. Similarly, pathological changes in the bladder and kidneys of infected animals vary to some extent based on the bacterial strain, inoculating dose, and infection model. However, P. mirabilis has been demonstrated to be more invasive than Salmonella enterica serovar Typhimurium (Table 2) (45). Although not an exhaustive list, the invasion and cytotoxic properties of various P. mirabilis isolates are summarized in Table 3.
Table 2
Strain | Growth condition | Internalized bacteria (CFU/ml)a |
---|---|---|
P. mirabilis WPM111 | Broth; aerated; 18 h | 22,420 |
E. coli HB101 | Broth; aerated; 18 h | 67 |
S. Typhimurium | Broth; aerated; 18 h | 18,130 |
Modified from (45).
Table 3
Strain Source | Strain name | Invasion | Cytotoxicity | ||
---|---|---|---|---|---|
Cell Line | Ref | Cell Line | Ref | ||
Catheterized urinary tract | HI4320 | Human embryonic kidney (HEK293), monkey kidney (Vero), human urothelial (UMUC-3) | (46) | Human primary renal proximal tubular epithelial cells (HRPTEC), human embryonic kidney (HEK293) | (25, 47) |
BA6163 | ND | Human primary renal proximal tubular epithelial cells (HRPTEC), human bladder epithelial (T24), human B-cell lymphoma (Daudi and Raji), human monocyte (U-937), monkey kidney (Vero) | (47, 48) | ||
CFT295 HU2450 CFT106 SA1387 MI159 EL1131 CFT37 MA2489 DR3282 CFT403 | ND | Human primary renal proximal tubular epithelial cells (HRPTEC) | (47) | ||
C6 C7 C11 C31 | Human ureter (Hu 609), and human bladder epithelium (HCV 29) | (49) (49) | ND | ||
UTI | T1 31 | Monkey kidney (Vero) | (50) | Monkey kidney (Vero) | (50) |
Urinary stone | PM7002 | Human urothelial (UMUC-3) | (51) | ND | |
K8 K608 | Human ureter (Hu 609) | (49) | ND | ||
U6450 | ND | Human urothelial (EJ/28) | (52) | ||
Wound | KR37 | Human ureter (Hu 609) and human | (49) | ND | |
KW28 | bladder epithelium (HCV 29) | ||||
Unspecified | N2 P19 | Human urothelial (NTUB1) | (53, 54) | ND |
While it is clear that P. mirabilis invades host tissues during infection, it does not appear to establish a significant intracellular niche as observed for uropathogenic E. coli (44). Several virulence factors have been implicated in contributing to cell invasion and cytotoxicity in vitro, as well as histopathological changes in vivo. For instance, flagella contribute to invasion in part by allowing the bacterial cells to come into close proximity to the host cells, and mutants lacking flagella are unable to invade cells unless centrifuged directly onto the host cell monolayer (45, 52). The specifics of P. mirabilis invasion of bladder and kidney cells and the contributions of select virulence factors are detailed further below.
Bladder invasion and histopathology
Internalization of P. mirabilis by bladder epithelial cells has been directly demonstrated in vitro using a hemolysin (hpmA) mutant to avoid confounding from the effects of the cytolytic toxin. In these experiments, it was determined that P. mirabilis utilizes the AipA autotransporter for internalization into bladder cells (46, 55). In addition to internalization, P. mirabilis is capable of lysing bladder epithelial cells using a combination of the Proteus toxic agglutinin (Pta) and hemolysin (55).
In the murine model of ascending UTI, P. mirabilis invades bladder epithelial cells as early as 30 minutes post-inoculation, which may provide transient protection from the immune response and an intracellular niche for initial replication and survival (44, 50, 52, 56–58). However, intracellular bacteria are uncommonly observed at later times post-inoculation, and P. mirabilis appears to instead form large, extracellular clusters within the bladder lumen and adjacent to the urothelium after this initial invasion phase rather than establishing the intracellular communities that are characteristic of uropathogenic E. coli (44) (Fig. 6). Formation of these clusters requires urease activity and the mannose-resistant Proteus-like (MR/P) fimbriae, and provides protection from infiltrating neutrophils (44).
Bladder colonization by P. mirabilis most often results in mild to moderate cystitis in murine models of UTI, including transmural neutrophilic inflammation, epithelial transcytosis of neutrophils, and submucosal edema (25, 55). The pathology of P. mirabilis bladder colonization during ascending UTI is largely due to a combination of urease activity, Pta, and hemolysin (25, 55). In the murine model of CAUTI, infection with P. mirabilis results in more severe cystitis, and this appears to be largely due to urease activity and encrustation of the catheter (25).
Kidney invasion and histopathology
Infection with P. mirabilis tends to result in unique kidney pathology. For instance, P. mirabilis is the only species that causes a high incidence of kidney stone formation in a rat model of pyelonephritis (59), and it causes more kidney stones and greater kidney damage than other urease-positive organisms such as P. stuartii in murine models of ascending UTI and CAUTI (25, 41). Specifically, kidney colonization by P. mirabilis in the ascending model of UTI is most often associated with moderate pyelonephritis, including neutrophilic interstitial nephritis within the peripelvic renal cortex and occasional damage to the surrounding renal parenchyma (25, 55). The severity of pyelonephritis is more strongly influenced by the additive effect of Pta and hemolysin than urease (25, 55). The same is true in the murine model of CAUTI, with P. mirabilis infection similarly resulting in moderate pyelonephritis, largely independent of urease activity (25).
In addition to directly damaging kidney tissue and inducing inflammation, P. mirabilis proliferates within the tubular epithelium of the kidneys in both mice and rats, resulting in necrosis and nephrosis (25, 59). In contrast to pyelonephritis, the severity of nephrosis correlates with urease activity in the murine models of ascending UTI and CAUTI, indicating a role for the urease enzyme and alkaline pH in renal tubule damage (25). These foci of inflammation can also develop into inflammatory lesions, another characteristic feature of P. mirabilis pyelonephritis (25, 55, 59).
Internalization of P. mirabilis by human renal proximal tubular epithelial cells (HRPTECs) has also been directly demonstrated in vitro using a hemolysin mutant to avoid confounding by the cytotoxic effects of the toxin. Interestingly, internalization requires protein synthesis by P. mirabilis, but does not require active phagocytosis or protein synthesis by the HRPTECs (56, 58). Thus, invasion of host cells by P. mirabilis, at least in vitro, is predominantly mediated by bacterial factors. Hemolysin is also thought to be the primary virulence factor responsible for direct lysis of HRPTECs (47, 56).
D. Innate immune responses to P. mirabilis UTI
Although innate immune responses to UTI are an active area of investigation, most of these studies have focused on uropathogenic E. coli (60–62). Because there are several key differences between P. mirabilis UTI and UPEC UTI, including urolithiasis, bacterial metabolism during UTI, and intracellular vs. luminal niches in the bladder, we will focus sharply here on studies that specifically examine P. mirabilis. Likewise, adaptive immune responses to vaccines are described elsewhere in this chapter (link vaccines section).
P. mirabilis induces a pro-inflammatory response during UTI; mice infected with P. mirabilis have elevated levels of CXCL1 at 72 hpi and IL-10 at 6 and 96 hpi in their urine (25). Notably, the cytokine response is further elevated during coinfection with Providencia stuartii, with increased levels of CCL2, CCL5, CXCL1, IL-6, IL-10, IL-17A, TNFα, IFNβ, and IFNγ at 48 hpi. In contrast, P. stuartii monoinfection did not lead to increased levels of any of the measured cytokines (25). Some of the elevated cytokines, such as IL-10, are anti-inflammatory (63), suggesting there are switches between pro- and anti-inflammatory responses as the infection progresses. This is consistent with changes in innate immune responses observed for other uropathogens, including UPEC and Group B streptococcus (60, 62, 64). TNF-α and IL-6 are also elevated in bladder homogenates from mice at 48 hpi (Peng 2016—it’s currently ref 308).
Neutrophils are one of the first innate responses to UTI, and are recruited by host uroepithelial cell signaling in response to conserved bacterial structures (pathogen-associated molecular patterns, or PAMPs) (65). CXCL1, mentioned above, is a potent neutrophil chemoattractant, and leukocytes are frequently detectable in urine during P. mirabilis UTI. Appropriately, there have been several studies investigating interaction of P. mirabilis with these leukocytes. By 10–24 hpi in the mouse model of UTI, P. mirabilis forms large clusters in the bladder lumen that draw a massive infiltration of neutrophils (Fig. 7) (44). Neutrophil extracellular traps (NETs) are webs of decondensed chromatin and antimicrobial proteins that are released in response to infection (66), and neutrophils in regions adjacent to the P. mirabilis bacterial clusters have been observed with extruded DNA that colocalized with extracellular H2A, suggesting development of NETs during P. mirabilis UTI (44). Supporting the mouse model data, proteomic signatures indicative of “NETosis” have been detected in urine samples from patients with P. mirabilis UTI (67, 68). Although neutrophil-mediated phagocytosis and NETosis can be effective bacterial clearance mechanisms, this line of attack may be less effective against P. mirabilis encased within clusters or urinary stones (42, 44). Neutrophils also respond to bacteremic P. mirabilis UTIs, and an increase in band neutrophils (>10% white blood cell count) is associated with an increased risk of P. mirabilis bacteremia (odds ratio 4.18) (69).
Inflammasomes are protein complexes that form in response to a variety of stimuli and act to induce inflammation (70). One particular type, the NLRP3 inflammasome, has been specifically linked to P. mirabilis in the intestine, where P. mirabilis induces a potent IL-1β response (71). In this setting, NLRP3 activation is dependent on HpmA hemolysin-induced K+ efflux (71). NLRP3 inflammasomes are also activated by crystalline material, and thus urease activity during UTI may further boost a potent inflammatory response (44, 70).
Tamm-Horsfall protein (THP, also called uromodulin) is the most abundant protein in normal human urine (72). Mice deficient in THP are susceptible to UTI (73), and P. mirabilis likewise causes more severe infection in THP-deficient mice (74). THP is heavily glycosylated and has been proposed to be a competitive ligand for UCA/NAF fimbriae (link fimbriae) (74).
Flagellin is another conserved structure that is recognized by the innate immune response, specifically toll-like receptor 5 (TLR5) (75), and flagellar motility is a fundamental feature of P. mirabilis (link flagella). In the intestine, Lypd8 protein prevents flagellated bacteria, including P. mirabilis, from invading colonic epithelium and causing inflammation (76). Installation of purified P. mirabilis flagellin into the bladder elicits leukocyte infiltration, histological changes in bladder tissue, and elevated Cxcl1, Cxcl10, and Il6 mRNA (77). Interestingly, pre-treatment with flagellin did not aid clearance of P. mirabilis in a subsequent challenge, and instead led to increased bacterial recovery from the kidneys (77).
Immune evasion
P. mirabilis has several tools for evading the innate immune response. These bacteria readily invade cultured cells (link invasion). However, although invasion has been observed during experimental UTI, P. mirabilis is typically found in the bladder lumen (44). It is possible that invasion plays a larger role during pyelonephritis. Urinary stones may also provide shelter from attack by leukocytes or antimicrobial peptides (3, 19, 42, 78). P. mirabilis is noted for its general resistance to killing by antimicrobial peptides, particularly polymyxins (79). Two components of this resistance are Zap protease, which degrades antimicrobial peptides (link), and LPS modifications to alter surface charge (80). Notably, different types of P. mirabilis LPS O-antigen elicit distinct proinflammatory cytokine IL-8 responses from cultured urothelial and renal cells (81). Furthermore, P. mirabilis produces a lysozyme inhibitor named PliC (82). Finally, at least two major antigenic proteins on the bacterial surface, MR/P fimbriae and flagella, are subject to phase variation. Thus, bacteria producing altered flagellins or phase-off MR/P fimbriae may evade both innate and adaptive immune responses.
P. mirabilis as a possible trigger of autoimmunity
Rheumatoid arthritis (RA) is a chronic autoimmune disorder that primarily attacks the joints. There is an intriguing correlation between P. mirabilis and RA, suggesting that in some populations, P. mirabilis UTI may be a triggering event for autoantibody development (83). Specifically, amino acid sequences IRRET in UreC (a component of urease) and ESRRAL in HpmB (an accessory protein which allows secretion of HpmA hemolysin) may lead to collagen autoantibodies due to cross-reactivity in individuals with particular HLA-DR4 subtypes. A correlation between P. mirabilis and RA has been observed in multiple patient studies (83); however, the urease/hemolysin hypothesis has not yet been directly tested. Importantly, other microbes have also been linked to RA, and it is possible that there are multiple infectious triggers for RA in vulnerable populations (84).
P. mirabilis is also associated with inflammatory bowel disease, which likewise has autoimmune origins. In a TRUC mouse ulcerative colitis model, P. mirabilis and Klebsiella pneumoniae correlated with colitis (85). Similar to the Lypd8 mice mentioned above, P. mirabilis bacteria were observed in the colonic mucus adjacent to epithelial cells in colitic mice, but not in wild-type mice (76, 85). While wild-type mice were negative for P. mirabilis, wild-type pups fostered on TRUC dams developed colitis and were positive for P. mirabilis and K. pneumoniae; furthermore, administration of TRUC-derived P. mirabilis and K. pneumoniae strains to specific pathogen-free wild-type mice elicited colitis (85).
III. Approaches to Studying Virulence
A. Animal models
The first murine model of UTI was described in 1967 (86), and while several different experimental methods have been explored for P. mirabilis infection studies, the most widely used model for investigating P. mirabilis pathogenicity is based on a method developed in the 1980s, referred to as the Hagberg protocol for ascending, unobstructed UTI (87). This protocol was initially developed to study E. coli virulence, particularly bacterial attachment to host bladder cells. Female CBA mice were favored as the model organism in this protocol due to several observations: 1) bacteria were better able to attach to uroepithelial cells from mice than rats, 2) CBA mice exhibited greater susceptibility to colonization and experimental UTI than BALB/c, C57BL/6, or C3H/HeN mice, and 3) the anatomy of male mice was not permissive for transurethral inoculation directly into the bladder. However, many mouse strains are susceptible to bladder colonization by P. mirabilis and ascension to the kidneys. For example, CF-1 mice have been utilized for noninvasive biophotonic imaging studies to monitor progression bacterial progression from the bladder to the kidneys and response to antibiotic treatment (88) (Fig. 8).
Two overviews of the procedures for induction of urinary tract infection in mice have been published in the Journal of Visualized Experiments (JoVE) (89, 90). Briefly, 6–9 week old mice are anesthetized and the peri-urethra and peri-anal areas are sterilized, generally with 10% povidone iodine. A 25 mm segment of gas-sterilized polypropylene tubing (0.28 mm inner diameter, 0.61 mm outer diameter) is threaded over a sterile 30-gauge hypodermic needle in such a way that approximately 15 mm of tubing extend beyond the tip of the needle. For transurethral inoculation, approximately 10 mm of the protruding polypropylene tubing is inserted through the urethra, beneath the pelvic bone, so that the bacterial inoculum is instilled directly into the bladder. This is generally performed using a mechanical syringe pump to slowly infuse a 0.05 ml inoculum over a 30-second interval to reduce the likelihood of reflux into the kidneys, so that the bacteria must actively ascend the ureters to colonize the kidneys and establish an upper urinary tract infection.
Initial studies of P. mirabilis pathogenesis using the murine model of ascending UTI were conducted using an inoculum of 2×1010 CFU/ml (1×109 CFU/mouse) and time points of 2, 7, or 14 days post-inoculation (39, 40, 91). For P. mirabilis HI4320, inoculation with this high dose resulted in significant mortality in mice at later time points (25% after 7 days, and 25% after 14 days), as well as severe pyelonephritis, renal damage, and urolithiasis (39). Inoculation with a lower dose (2×108 CFU/ml, or 1×107 CFU/mouse) was therefore determined to be ideal for studying the contribution of P. mirabilis virulence factors to pathogenicity. The contribution of specific genes to fitness and disease progression are generally assessed by one of two ways: 1) independent challenge, in which one group of mice is inoculated with the wild-type strain and the other is inoculated with the mutant of interest to see if the mutant exhibits significant differences in bacterial burden, or 2) cochallenge with a 1:1 mixture of mutant to wild-type, where the mutant must compete against the wild-type parent in the same mouse. In a cochallenge, wild-type and mutant bacteria are distinguished using a marker, such as antibiotic resistance, and a competitive index is calculated for the mutant by dividing the ratio of mutant/wild-type recovered from the mouse to the ratio of mutant/wild-type in the inoculum (Fig. 9).
With respect to catheter-associated UTI, the presence of a catheter segment within the bladder provides bacteria with a new substrate for attachment and colonization and also elicits numerous changes within the bladder environment, including induction of a potent pro-inflammatory response (25, 88, 92, 93). Thus, as P. mirabilis most commonly causes complicated or catheter-associated UTI, several adaptations to the standard ascending model have been utilized to introduce and maintain a foreign body in the bladders of mice. One such method uses 6 mm segments of polyethylene tubing that have been coiled and sterilized with ethanol, pre-colonized by P. mirabilis for biofilm formation, and stretched back to an uncoiled position for transurethral insertion into the bladder, where they will spontaneously re-coil to prevent loss of the catheter during normal urination (88). Another method uses a 4 mm segment of silicone catheter tubing, sterilized by autoclaving, and introduced into the bladder at the time of transurethral inoculation (25). Both methods require reducing the bacterial density of the inoculum approximately 100-fold over the ascending UTI model to reduce mortality within 7 days of inoculation, and both models result in a greater incidence of cystitis and pyelonephritis than the ascending model. Furthermore, use of the silicone catheter segment was shown to induce a potent pro-inflammatory response during mock infection (25, 92), thereby mimicking the host response to catheterization that is observed in humans. Different catheter segment lengths may be required for different mouse strains and ages. However, for CBA/J mice and inoculation with P. mirabilis HI4320, rapid encrustation of the catheter promotes retention of the catheter segment in the vast majority of mice (25).
The predominant murine models of UTI and CAUTI rely solely on female mice due to the technical challenges of transurethral inoculation that are posed by the anatomy of male mice. A model of ultrasound-guided percutaneous catheter implantation into the bladder was therefore developed to permit studies of CAUTI in both female and male mice (94). Comparison of percutaneous catheter implantation and transurethral catheter placement revealed that both methods resulted in a similar incidence of catheter encrustation during infection with P. mirabilis. Thus, while this model may better approximate suprapubic catheterization than urethral catheterization, it will facilitate investigation of CAUTI and testing of catheter coatings in male mice in addition to female mice.
B. Comparative genomics
P. mirabilis lives in diverse environments, and has long been noted for strain variability in swarming behavior and LPS O-antigen structure (95–97). Given these observations, it might seem that this species would possess great diversity in its genomic content, much like the mosaic pangenome of E. coli (98, 99). However, several lines of evidence point toward a remarkable conservation of the P. mirabilis chromosome.
A proteomic comparison of the P. mirabilis HI4320 reference strain with three other genomes, AOUC-001, CYPM1, and BB2000, shows that the overwhelming majority of predicted proteins are at least 99% identical across all four strains (Fig. 10A). Furthermore, alignment of the first two fully-annotated P. mirabilis genomes, HI4320 and BB2000, shows a high degree of synteny with only a few gaps due to insertions or deletions (Fig. 10B). The major differences that are visible are due to insertion of bacteriophage, a conjugative transposon, and the mobile genetic element ICEPm1, described later.
Virulence factors such as urease, HpmA hemolysin, and MR/P fimbriae, are detected in the vast majority of strains, regardless of isolation source (100, 101). For example, hpmA was found by Southern blot in 63/63 isolates tested, and correlated with hemolytic activity (102); likewise, a survey of 211 isolates using a combination of PCR and dot blot found that all encoded hpmA (103). Zap protease is also widespread, including all P. mirabilis tested from a diverse O-serogroup collection (104). Similarly, ZapA-associated protease activity was detected in the urine of 16/17 P. mirabilis-infected patients (105).
Mannose-resistant hemagglutination is another defining characteristic of P. mirabilis, as both activity and the causative mrp fimbrial genes are nearly ubiquitous in this species (100, 101, 106). In addition, P. mirabilis encodes an abundance of fimbrial operons (link fimbriae section) relative to other bacterial species, and the majority of these of these are conserved across diverse isolates (106) (Fig. 11). The presence of a given fimbrial operon does not correlate with the isolate source. Of the 17 fimbrial operons encoded by P. mirabilis HI4320, only one, fim3, is detected in fewer than half of P. mirabilis isolates. A second operon, uca, is notable because it is present in the majority of isolates, but is located on a mobile element and may be found in different regions of the chromosome (106).
Despite the overall conservation of the P. mirabilis chromosome, there is variability within the species. A few loci in particular are hypervariable and merit further mention (Fig. 10A).
ICEPm1
Comparative genomic hybridization of Providencia stuartii and Morganella morganii with a P. mirabilis HI4320 microarray revealed an integrative and conjugative element (ICE) common to all three species (107) (link ICE section). Further analysis of this 94 kb element, called ICEPm1, revealed genes encoding two previously-identified virulence factors: Proteus toxic agglutinin (Pta) and Nrp siderophore (108–110). Carried within ICEPm1 is the Yersinia spp. high pathogenicity island (HPI), which includes the nrp operon. Consistent with a virulence role, ICEPm1 is found more often in urinary isolates than colonizing isolates (100% vs. 65%) (107). Furthermore, ICEPm1 has been observed to transmit between clinical isolates in a laboratory setting (111). Examination of sequenced genomes reveals that ICEPm1, when present, is not always inserted at the same genetic locus.
P. mirabilis strains may encode other conjugative elements, but they do not appear to be as prevalent as ICEPm1. For example, HI4320 encodes another conjugative transposon, named as either ICEPmiUSA1 or ICEPmi HI4320, that is a member of the SXT/R391 family of ICEs (112–114). No genes in ICEPmiUSA1 have thus far been identified as virulence-associated. However, other ICEs in the SXT/R391 family carry multiple antibiotic resistance genes, and some of these ICEs are found in P. mirabilis (115–117).
T6SS
Because the type VI secretion system (T6SS) of P. mirabilis is involved in strain self-recognition and competition between strains (link secretion section), it is not surprising that there is a high degree of variability in these genes. Specifically, the genes encoding the T6SS secretion apparatus are highly conserved, but the secreted effectors vary in sequence, type, and number. For example, P. mirabilis HI4320 and BB2000 encode T6SS effector operons adjacent to, and in the reverse orientation from, the T6SS secretion apparatus genes (118, 119). In both cases, the effector operons begin with hcp and vgrG homologs, which are required for T6SS function (120). After that, however, the operons diverge in both sequence and number of genes; for example, BB2000’s idr operon comprises five genes while HI4320’s pef operon has nine (118, 119) (Fig. 12). Furthermore, HI4320 encodes at least two more operons identified as likely T6SS effectors than BB2000 is predicted to encode (112, 121). Examination of other P. mirabilis sequenced genomes indicates that this variability continues throughout the species (Fig. 10A).
O-antigen
Like other gram-negative bacteria, P. mirabilis strains have diverse O-polysaccharides as part of their LPS; 76 O-serogroups have been described thus far for Proteus spp. (122). In fact, the O and H antigen designations for LPS and flagellar typing were originally named by Weil and Felix’s observation that swarming P. mirabilis had a breath-like appearance (German Hauch, breath) while non-swarming colonies did not (ohne Hauch, without breath) (95). In a survey of P. mirabilis strains with different O-serotypes, virulence factors were widespread (urease, Zap protease, swarming), but strains differed in the levels of these activities. Specifically, strains with negatively-charged O-polysaccharides displayed higher urease, protease, and swarming activities compared with positive or neutral O-polysaccharides (104). LPS is abundant on the cell surface of gram-negative bacteria, and its properties can affect bacterial interaction with their environment. Thus, some aspects of strain variability may be determined by LPS biochemistry.
Flagellin
Flagella (link flagella section) are targets of the innate immune system and also energetically expensive to produce (123). Correspondingly, although most flagellar genes are conserved in P. mirabilis, the gene encoding the whip itself is highly variable, as in other bacterial species. P. mirabilis also encodes a second, silent flagellar gene flaB, which can recombine with flaA and yield hybrid FlaAB flagella (124). The properties of FlaAB hybrids have been extensively reviewed elsewhere (19).
Phage
Like other bacteria, P. mirabilis genomes are peppered with integrated phage. HI4320 encodes three apparently complete and three degenerate prophages (112). CRISPR-Cas systems are encoded in approximately one-third of sequenced P. mirabilis genomes as determined by the Bacterial Bioinformatics Resource Center PATRIC (125); notably, CRISPR is not present in two strains commonly used for experimentation: HI4320 and BB2000 (112, 121).
Differential expression of phage genes has been observed in P. mirabilis during swarming (126, 127); specifically, one phage is inserted adjacent to the rcs operon, which encodes a regulator of flagella and swarming. Furthermore, phage have been used to transduce swarming ability into non-swarming strains (128). Thus, it is likely that phage account for some of the observed differences in swarming behavior between strains.
Plasmids
Plasmids are not a defining feature for P. mirabilis. The original P. mirabilis sequenced genome, HI4320, included a 36 kb plasmid that was designated pHI4320 (112), and a signature-tagged mutagenesis study as well as a transposon insertion-site sequencing study indicated that mutations in pHI4320 led to decreased fitness in mice (108, 129). However, several of the targeted genes were involved in plasmid stability, and pHI4320 does not encode any obvious virulence genes (112). In the National Center for Biotechnology Information (NCBI) database, a minority of P. mirabilis strains have been reported to carry distinct plasmids, at least some of which encode antibiotic resistance genes (130).
C. Transcriptomics
P. mirabilis undergoes dramatic shifts in its appearance and surroundings, and it follows that many of these responses result from changes in transcription. Global analysis of gene expression (transcriptomics) has been conducted for P. mirabilis under multiple conditions, initially using microarrays and, more recently, RNA-seq.
Microarrays
The four following studies (109, 126, 131, 132) were conducted using a microarray with 70-mer oligonucleotide probes designed to hybridize with each of the 3,719 predicted open reading frames from an early draft version of the P. mirabilis HI4320 genome (112).
Swarming microarray
Perhaps the most dramatic change P. mirabilis undergoes is the cyclic switch from elongated, hypermotile swarmer cells to short, nonmotile consolidated cells during culture on solid surfaces (19). This transformation was the target of the first P. mirabilis transcriptomic study, where (i) swarm and consolidate transcripts were shown to be more similar to each other than to RNA derived from broth culture, and (ii) while many genes are induced during consolidation, very few are upregulated during active swarming (126). The data were consistent with earlier reports of increased expression of certain virulence genes during swarming (133, 134), and swarmer cells being less metabolically active than consolidates (135). Curiously, although P. mirabilis is largely nonmotile during consolidation, flagellar transcripts were among the most highly-transcribed messages in both swarm and consolidate (126). Based on the accumulated transcription and metabolic data, the consolidation phase has been proposed to be a time of preparation for the next round of swarming, by increasing nutrient uptake systems, central metabolism, and respiration. As P. mirabilis swarming is very robust and occurs under conditions not permissive for other species of swarming bacteria, the preparation that occurs during consolidation phase may contribute to how P. mirabilis is able to swarm under a wider range of conditions that other species, including on solid surfaces such as hard agar and catheters.
Iron limitation microarray
Iron is sequestered by the host to combat bacterial infection, and pathogens activate iron acquisition systems during UTI (link metals section). P. mirabilis was previously thought to lack common iron scavenging systems in earlier studies. However, microarray analysis conducted on iron-restricted P. mirabilis revealed 21 iron-regulated systems, including two new acquisition systems which are detailed later in this chapter (109).
Ascending UTI microarray
A critical point for studying how P. mirabilis causes disease requires knowledge regarding the physiological state of this pathogen during UTI. To answer this question, urine was collected from experimentally infected mice at 1, 3, and 7 days post-inoculation (dpi) for RNA extraction and microarray hybridization (131). Iron acquisition, urease, and peptide transporters were induced during UTI, consistent with an iron-restricted, urea- and peptide-rich urine environment. The genes encoding MR/P fimbriae (link fimbriae section), which were previously shown to contribute to infection, were the most-highly upregulated of the transcriptome compared with broth culture (Fig. 13). However, some known virulence factors were either poorly expressed or downregulated during UTI, including genes encoding PMF fimbriae, Zap metalloprotease, hemolysin, and Proteus toxic agglutinin (Pta). This suggests that voided urine provides an important, but incomplete, snapshot of bacterial gene expression during UTI.
Interestingly, central metabolic pathways that were required during infection differed in several ways compared with a previous study of uropathogenic E. coli (136). Nitrogen assimilation gene expression suggested that P. mirabilis has greater access to nitrogen compared with E. coli, but follow-up experiments indicated this was not simply due to ammonia derived from urease activity (131). In contrast with E. coli, P. mirabilis induced glucose uptake and glycolysis genes, suggesting that P. mirabilis is able to access nutrient sources not available to E. coli (perhaps due to the activity of toxins on bladder epithelium). Gene expression at 1, 3, and 7 days post-inoculation was also compared. Flagellar genes were initially repressed, but this repression was relieved by day 7. In contrast, mrp genes were highly-transcribed throughout the infection but their degree of induction was lower at day 7; this topic is further detailed in the fimbrial section (link fimbriae section).
The transcriptional regulator MrpJ is encoded by the mrp fimbrial operon (link fimbriae section). Originally proposed as a switch between adherent and motile states (137), MrpJ was found to contribute to other phenotypes as well (138). Microarray experiments comparing an mrpJ in vivo mimic, where mrpJ is expressed in trans at levels comparable to those detected in mice during experimental UTI, with an empty vector control showed that MrpJ’s regulatory network extends considerably beyond repression of flagella to include, among others, regulation of additional fimbriae and the type VI secretion system (132). MrpJ and MrpJ paralogs are discussed later in this chapter (link).
RNA-seq
The next three studies (127, 139, 140) compared gene expression for targeted mutants with their isogenic parent strain using RNA-seq. In addition to the aforementioned microarray study, broth-cultured and swarmer cells have also been compared by RNA-seq using the BB2000 lab strain (127). Even though the methods for collecting and analyzing swarmer cells were different in these two studies (126, 127), similar results were obtained for the broth-swarmer comparison regarding the genes encoding flagella, lipoproteins, iron transport, and hemolysin, among others. FliL is a flagellar protein that, when mutated, leads to elongated swarmer-like cells under non-swarming conditions (pseudoswarming) (141). Comparison of a fliL mutant cultured in broth to induce pseudoswarmer cells with the wild-type broth RNA-seq dataset revealed an increase in propanediol utilization in the fliL mutant. This suggests a possible new role for these genes in swarmer cell differentiation (127).
The Rcs phosphorelay is a repressor of swarming and an activator of biofilm formation (139, 140, 142–144). RNA-seq was used to compare a broth-cultured rcsB mutant with its isogenic parent strain PM7002 (139, 140). In this condition, the rcsB mutant undergoes inappropriate elongation that is similar to the fliL phenotype. These studies showed that RcsB controls pathways important for swarming, including cell division (minCDE) and motility (flagella and flagellar regulators). Furthermore, RcsB activates biofilm-associated genes, including genes encoding MR/P, PMF, and UCA fimbriae (link biofilm section). Interestingly, RcsB also represses virulence genes that have previously been shown to be activated during swarming (zapABD, hpmBA) and type VI secretion genes (tss, ids, and idr/pef operons).
D. Mutagenesis
The first targeted mutation to be made in P. mirabilis was a ureC urease mutant (40) (Fig. 14). This mutant was constructed using a single crossover, Campbell-type insertion via an R6K ori-pir suicide vector introduced by conjugation (145), an approach that is still used by multiple groups studying P. mirabilis. This initial paper noted several difficulties with methods for mutant construction in P. mirabilis, as conjugative-incompatible plasmids, phage transduction, and spontaneous curing of multicopy plasmids all resulted in failure to yield chromosomal mutations (40). Later on, sacB counterselection was employed to facilitate identification of double crossover events (146). The ability to directly construct and test mutants in the mouse UTI model was a boon for dissecting P. mirabilis virulence, and allelic exchange via mating with suicide plasmids continues to be used for mutant construction. Even so, mutant construction remained labor-intensive and prone to failure, particularly in some clinical isolates.
Subsequently, sequencing of the HI4320 genome revealed that this strain naturally carries a plasmid that is closely related to R6K and encodes a π protein (112), possibly contributing to the difficulty in constructing mutants in this background using typical suicide vectors. Thus, another method was developed for constructing mutations, using a group II intron mutated to specifically target a gene of interest (147, 148). Advantages over the prior allelic exchange method are the relatively short time required, higher rate of success, and lack of passaging or counterselection that would normally be required to resolve double crossover events. The drawbacks are that the choice of insertion sites is limited, mutations have polar effects on downstream genes, and the method is insertion-based (that is, it is not simple to delete sequences). The latter problem can be managed by use of cre-lox recombination to remove the antibiotic resistance cassette carried within the intron, resulting in an unmarked mutant (126). Additional introns can be inserted, with the potential to remove intervening sequences via an additional round of cre-lox recombination.
Genes that have been specifically mutated and tested in a mouse model of UTI, either independently or by direct cochallenge against a parental strain, are shown in Table 4. It is important to note that the studies presented in Table 4 include a range of P. mirabilis strains, mouse strains, infectious doses, and experiment lengths, but exclude mutants generated by transposon insertion and targeted mutants that were only tested for virulence in an in vitro assay, such as cell culture.
Table 4
Gene | Locusa | Description | Virulence geneb | Reference |
---|---|---|---|---|
ureC | PMI3685 | Urease alpha subunit | Y | (39, 40, 42, 44) |
hpmA | PMI2057 | Hemolysin | N | (45, 48, 55) |
mrpA | PMI0263 | Major mannose-resistant/Proteus-like fimbrial protein | Y | (44, 149–152) |
pmfA | PMI1877 | PMF major fimbrial subunit | Y | (153, 154) (151) |
flaD | PMI1621 | Flagellar hook-associated protein 2 | Y | (45) |
flaAB | PMI1619–20 | Flagellin 1 and flagellin 2 | N | (155) |
mrpG | PMI0269 | MR/P fimbrial subunit | Y | (146) |
zapAc | PMI0279 | Metalloprotease | Y | (134) |
mrpH | PMI0270 | MR/P fimbrial adhesin | Y | (156) |
atfA | PMI2728 | ATF major fimbrial subunit | N | (157) |
mrpJ | PMI0271 | Fimbrial operon regulator | Y | (132, 137) |
mrpI | PMI0262 | Fimbrial recombinase | Y d | (158) |
luxS | PMI0379 | S-ribosylhomocysteine lyase | N | (159) |
ureR | PMI3681 | Urease operon transcriptional activator | Y | (160) |
spa47 | PMI2692 | Type III secretion system ATPase | N | (148) |
PMI0047 | Secreted 5′-nucleotidase | N | (161) | |
rafY | PMI0288 | Putative glycoporin | N | (161) |
PMI0842 | TonB-dependent receptor | Y | (161) | |
fadL | PMI1810 | Long-chain fatty acid transport protein | N | (161) |
PMI2596 | TonB-dependent receptor | Y | (161) | |
pta | PMI2341 | Proteus toxic agglutinin | Y | (55) |
znuC | PMI1151 | High-affinity zinc uptake system ATP-binding protein | Y | (162) |
zur | PMI2748 | Zinc uptake regulation protein | Y | (162) |
dppA | PMI2847 | Dipeptide ABC transporter, substrate-binding protein | Y | (126) |
oppB | PMI1474 | Oligopeptide ABC transporter, permease protein | N | (126) |
cysJ | PMI2250 | Sulfite reductase [NADPH] flavoprotein alpha-component | Y | (126) |
lrhA | PMI0629 | LysR-family transcriptional regulator | N | (126) |
hexA | PMI1764 | LysR-family transcriptional regulator | Y | (126) |
pbtA | PMI0232 | Siderophore biosynthesis protein | N | (109) |
nrpR | PMI2599 | Siderophore synthase | Y | (109) |
aipA | PMI2122 | Adhesion and invasion autotransporter | Y | (46) |
taaP | PMI2575 | trimeric autoagglutinin autotransporter of Proteus | Y | (46) |
gdhA | PMI3008 | NADP-specific glutamate dehydrogenase | Y | (131) |
glnA | PMI2882 | L-glutamine synthetase | Y | (129) |
ucaA | PMI0536 | UCA major fimbrial subunit | Y | (163) |
hfqe | PMI3365 | Hfq (host factor-I protein) | Y | (164) |
speA | PMI2094 | Biosynthetic arginine decarboxylase | Y | (165) |
speBF | PMI2093; PMI0307 | Agmatinase; ornithine decarboxylase, inducible | Y | (165) |
pfkA | PMI3203 | 6-phosphofructokinase | Y | (166) |
pgi | PMI2754 | Glucose-6-phosphate isomerase | Y | (166) |
tpiA | PMI3205 | Triosephosphate isomerase | Y | (166) |
pykA | PMI1155 | Pyruvate kinase II | Y | (166) |
gnd | PMI0655 | 6-phosphogluconate dehydrogenase, decarboxylating | Y | (166) |
talB | PMI0006 | Transaldolase B | Y | (166) |
edd | PMI2760 | 6-phosphogluconate dehydratase | Y | (166) |
sdhB | PMI0568 | Succinate dehydrogenase iron-sulfur protein | Y | (166) |
frdA | PMI3588 | Fumarate reductase flavoprotein subunit | Y | (166) |
fumC | PMI1296 | Fumarate hydratase, class II | Y | (166) |
pckA | PMI3015 | Phosphoenolpyruvate carboxykinase [ATP] | N | (166) |
argG | PMI3239 | Argininosuccinate synthase | N | (166) |
serA | PMI2031 | D-3-phosphoglycerate dehydrogenase | Y | (166) |
rpoE | PMI1894 | RNA polymerase sigma-E factor | Y | (167) |
rseA | PMI1893 | Anti-sigma E protein | N | (167) |
PMI1518 | High-affinity nickel efflux protein | Y | (129) | |
pldA | PMI3344 | Phospholipase A | N | (129) |
ilvD | PMI3302 | Dihydroxy-acid dehydratase (branched chain amino acid biosynthesis protein) | Yf | (129) |
lon | PMI0117 | ATP-dependent proteinase, serine peptidase | Y | (129) |
argR | PMI3399 | Transcriptional regulator, repressor of the arginine biosynthetic pathway | Y | (129) |
E. Genome-wide mutagenesis
Transposon mutagenesis in P. mirabilis was first described in a publication from 1991 using a Tn5 derivative (169, 170). The mini-Tn5 transposon was carried on a suicide plasmid with the π protein-dependent origin of replication from R6K, oriT for conjugal transfer from donor strain to recipient strain, and the IS50R transposase gene tnp. This approach has since been used to generate libraries for isolation of mutants defective for a given phenotype, such as swarming motility (119, 143, 171–175), sensitivity to antimicrobial peptides (80, 176), and biofilm formation (177).
In 1995, this approach was revolutionized through the development of signature-tagged mutagenesis (STM) (178), in which each transposon mutant has a specific DNA sequence within the transposon that acts as a barcode. STM led to the first genome-wide pathogenesis studies by permitting identification of unique mutants within a mixed population and allowing for assessment of mutants lost during the selective process of the infection, and therefore containing transposon insertions in genes that contribute to colonization and pathogenicity. This approach was successfully utilized three times with P. mirabilis strain HI4320 in a murine model of ascending UTI, identifying a combined total of 54 genes that represented important fitness factors for colonization (summarized in Table 5) (108, 179, 180).
Table 5
PMI Number | Gene Name | Isolate | STM of Ascending UTI | CAUTI | Polymicrobial CAUTI | ||
---|---|---|---|---|---|---|---|
Reference | (180) | (108) | (179) | (129) | (129) | ||
PMI0012 | nhaR | 4B6 | U, K, O | ||||
PMI0020 | carA | D4–8 | U, B, K | B, K | |||
PMI0030 | exbD | 7A1 | U, B, K, O | B | |||
PMI0283 | zapE2 | 9C6 | K | ||||
PMI0565 | sdhC | 10E3 | U, B, K, O | ||||
PMI0589 | aroG | G1–38 | U, B, K | ||||
PMI0711 | serC | G4–12 | U, O | B, K | |||
PMI0842 | D6–33 | U, B, K | K | ||||
PMI1000 | 8H4 | U, O | K | ||||
PMI1045 | arnA | G5–30 | U, B, K | B, K | |||
PMI1184 | 16F1 | K | |||||
PMI1193 | 40A6 | U, B, K, O | |||||
PMI1421 | ppsA | 17A4 | B, K, O | B, K | |||
PMI1448 | 6F6 | U, O | |||||
PMI1546 | guaB | 5F1 | U, B, K, O | B, K | |||
PMI1598 | yidA | 12E4 | U | ||||
PMI1607 | sdaA | 15C3 | U | K | K | ||
PMI1630 | fliF | 8A6 | B, K | B | B, K | ||
PMI1651 | flgE | 21B2 | K | B, K | |||
37D5 | U, B, O | ||||||
PMI1667 | cheW | E6–43 | U, B, K | B, K | |||
PMI1761 | nuoB | 15A6 | U, B, K, O | ||||
PMI1874 | E5–36 | U, B, K | K | ||||
PMI2014 | 32C4 | B | |||||
PMI2046 | aceE | 8B5 | B, K, O | ||||
PMI2259 | metN | 23D5 | U, K, O | B, K | |||
PMI2332 | surA | 11C4 | U, B, K, O | K | K | ||
32E1 | B | ||||||
PMI2342 | sufI | 38E2 | K, O | K | |||
PMI2345 | parE | C6–15 | U, B, K | ||||
PMI2378 | cbbC | 11G3 | U, K, O | B, K | |||
PMI2575 | 36D2 | B, O | |||||
PMI2605 | nrpG | A1–8 | U, K, O | K | |||
PMI2760 | edd | 11D4 | U, K | B, K | |||
PMI2823 | ppiA | 6D1 | U, K, O | ||||
PMI2828 | dsbA | F4–37 | U, B, K | B, K | |||
PMI2893 | pstS | H4–34 | U, B, K | B, K | |||
PMI2894 | pstC | G1–43 | U, B, K | B, K | |||
PMI3001 | 4A3 | U | |||||
11A4 | U, K, O | ||||||
PMI3021 | mrcA | B6–23 | U, B, K | K | |||
PMI3053 | asnC | E6–41 | U, B, K | B, K | |||
PMI3190 | cpsF | 11B2 | B, K, O | K | |||
20C4 | U, B, K, O | ||||||
PMI3295 | hdfR | D5–9 | U, O | K | |||
C1–18 | B, O | ||||||
PMI3329 | hemY | B2–9 | U | B | |||
PMI3333 | cyaA | 20E1 | U, B, K, O | K | |||
PMI3359 | 13D1 | U, O | |||||
PMI3390 | F6–34 | U, B, K | K | ||||
A3–39 | U, B, K | ||||||
PMI3441 | B2 | P | |||||
PMI3451 | nrdD | D6–39 | B, K, O | B, K | |||
PMI3623 | dusBC | F1–29 | U, B, K | B, K | B,K | ||
PMI3645 | rapZ | B5 | U, B, K | K | |||
PMI3687 | ureF | B6–40 | U, B, K | K | |||
PMI3692 | D6–26 | U, B, K | |||||
PMI3700 | D1–30 | U, B, K | B, K | ||||
PMI3705 | 1F6 | U, K, O | B, K | ||||
PMIP09 | pilX4 | C4–41 | B, K, O | K | K | ||
A5–34 | U, K, O | K | |||||
B4–38 | K | K |
P, primary screen; U, urine; B, bladder; K, kidney; O, attenuated overall
In the first P. mirabilis STM study, mice were inoculated with 96 tagged mutants as the input sample, and bacteria recovered from the bladders of 5 mice 2 days post-inoculation were used as the output samples (180). Only two mutants were reproducibly underrepresented in the output samples from all five mice: B2 (determined to be an insertion in a secreted protease, now identified as PMI3441, a U32-family peptidase), and B5 (determined to be an insertion in the rpoN operon, now identified as PMI3645, an RNase adaptor protein RapZ). The fitness defect of each mutant was assessed by direct cochallenge with wild-type P. mirabilis HI4320, and only the PMI3645 mutant remained significantly outcompeted.
The second P. mirabilis STM study used 45 pools of mutants as the input samples (2,088 mutants total), and bacteria recovered from the bladders or kidneys of 5 mice per pool 4 days post-inoculation were used as the output samples (108). 502 mutants were identified as attenuated in the primary screen. A secondary screen was conducted with these 502 mutants divided into 19 pools, and only 114 of the mutants were reproducibly attenuated. 84 of the 114 mutants were re-tested by direct cochallenge against wild-type P. mirabilis HI4320 for a longer experimental challenge of 7 days, and 37/84 (44%) of the mutants exhibited reproducible fitness defects. Nucleotide sequences were obtained from 30 of the mutants, 27 of which map to open reading frames in the current annotation of P. mirabilis HI4320. This approach identified several categories of mutants representing both known and novel fitness factors for UTI, including motility, iron acquisition, plasmid-encoded factors, transcriptional regulation, phosphate transport, urease activity, capsule synthesis, and metabolic pathways, the majority of which would have been difficult to identify as fitness factors by other methods.
The third P. mirabilis STM study screened 40 pools of mutants as the input samples (1,880 mutants total), and bacteria recovered from the bladders or kidneys of 5 mice per pool 4 days post-inoculation were used as the output samples (179). 570 mutants were identified as attenuated in the primary screen, 217 of which had a reproducible defect in the secondary screen. 93 of the most-attenuated mutants that did not exhibit competitive defects in vitro were re-tested by direct cochallenge against wild-type P. mirabilis, and 37 (40%) had reproducible fitness defects. Nucleotide sequences were obtained from 29 of the mutants, 28 of which map to open reading frames in the current annotation of P. mirabilis HI4320. This approach again identified fitness factors pertaining to cellular processes, transport, transcriptional regulation, motility, cell surface structures, and metabolism. However, there was no overlap in the genes or operons identified in this study and the prior two STM studies; as the three combined STM studies only achieved 70% theoretical genome coverage, this is likely due to incomplete saturation and differences between the mutants present in each pool for in vivo competition.
In 2009, the basic technique of STM was further improved upon and integrated with next generation sequencing techniques to allow for massively parallel sequencing and quantitation of the relative abundance of all mutants present in a given setting. Several variations of the method were developed concurrently, including insertion sequencing (INSeq) (181), transposon-directed insertion site sequencing (TraDIS) (182), transposon insertion-site sequencing (Tn-Seq) (183, 184), and high-throughput insertion tracking by deep sequencing (HITS) (185). As Tn-Seq has been applied to the study of P. mirabilis pathogenicity (129), the specifics of this method will be explained in further detail.
The first step of Tn-Seq is to generate a genome-saturating transposon mutant library, generally using a Mariner transposon. Due to the saturating nature of the library, this method can also be used to identify genes that are essential for growth in vitro, as insertion mutants within these genes will be lacking from the starting library. Infection studies are then conducted with the transposon library, and all bacteria in the target site of infection (for instance, the bladder) are collected for extraction of genomic DNA. The frequency of each insertion mutant is then determined, en masse, from the input sample (the saturated library) and the output sample (the bladder), and the fitness contribution of each gene in the genome can be assessed by the change in frequency of insertion mutants for that gene from the output sample compared to the input sample.
This approach was successfully used to generate the first genome-saturating library of transposon insertion mutants in P. mirabilis HI4320, and allowed for identification of 436 genes (12% of the genome) estimated to be essential for growth in Luria broth (129). As would be expected, the essential genes largely pertained to cell cycle control and division, cell wall biogenesis, replication, and ribosomal proteins, and 64% had been identified as essential genes in other bacterial species. The transposon mutant library was then used in a murine model of CAUTI to identify fitness factors for colonization of the catheterized bladder and ascension to the kidneys 4 days post-inoculation (Fig. 15). Massively parallel sequencing and statistical analysis identified 629 genes (17% of the genome) as fitness factors for CAUTI; 93 genes were specifically important for colonization of the catheterized bladder, 209 for colonization of the kidneys, and 286 that were important for colonization of both the bladder and kidneys. Of the 54 genes and operons identified as fitness factors for ascending UTI in the combined STM studies, 31 (57%) were also identified in the CAUTI model by Tn-Seq (summarized in Table 5). Eight mutants were constructed for direct cochallenge with the parental strain to confirm the fitness defects that were identified by Tn-Seq, and fitness defects were verified for 7/8 (88%). Tn-Seq therefore provided confirmation of a role for known P. mirabilis fitness factors during CAUTI, identification of CAUTI-specific fitness factors, and a model for exploring the different metabolic pathways and transport requirements between ascending UTI and CAUTI.
Importantly, this approach was also applied during polymicrobial infection with P. stuartii, in parallel with the above study, to determine how polymicrobial infection influences P. mirabilis fitness requirements (129) (Fig. 15). Of the 629 genes identified as P. mirabilis fitness factors for CAUTI, 217 (35%) were also important during polymicrobial infection. The fitness factors that were important for both infection types include fimbrial genes, components of the flagellar cascade, urease, and factors involved in inorganic ion transport and metabolism. Interestingly, an additional 1,353 candidate fitness factors were identified that appear to specifically contribute to P. mirabilis fitness during polymicrobial infection, including defense mechanisms (such as type III secretion and type VI secretion), signal transduction pathways, at least 10 distinct fimbrial types, iron uptake systems, respiratory nitrate reductase, the oxidative pentose phosphate pathway and the Entner-Doudoroff pathway, arginine import and biosynthesis, and branched chain amino acid (BCAA) biosynthesis. It was further determined that the requirement for BCAA biosynthesis during coinfection was due to high-affinity import of leucine by P. stuartii, indicating the utility of this technique for dissecting polymicrobial interactions during infection (129).
IV. Virulence Factors
A. Urease
P. mirabilis was the model organism for which the genetics of urease were first delineated. Urease is a nickel-metalloenzyme, the synthesis of which is induced by the presence of its substrate, urea (186, 187).
Structure and activity
Apourease is comprised of three structural subunits, UreA, UreB, and UreC, assembled as a homotrimer of individual UreABC heterotrimers (UreABC)3 (Fig. 16). To become catalytically active, the apoenzyme must acquire divalent nickel ions through a process involving four accessory proteins, UreD, UreE, UreF, and UreG. As studied by yeast two-hybrid and immunoprecipitation with monoclonal antibodies to UreC and UreD, it is clear that the accessory proteins interact with the apoenzyme to deliver Ni2+ to the active site (188). Indeed, UreE carries a natural histidine tail that allows it to carry Ni2+ and also to be purified in a single step on a Ni2+-nitrolo-acetic acid column.
In early studies using bacterial cell lysates of four P. mirabilis isolates, the molecular size of urease was originally estimated by molecular sieve chromatography to be 281–338 kDa (189), and the molecular size of the apoenzyme was later determined to be 252,600 Da (190). Including the two Ni2+ in each of three active sites, this brings the predicted molecular size to ~253 kDa with an isoelectric point of 5.9 for P. mirabilis HI4320. The affinity for substrate as estimated by Km is relatively weak, ranging from 22–60 mM urea among the four isolates. However, with urea concentrations of ~400 mM urea in human urine, the enzyme would be fully saturated and working at the Vmax within the urinary tract and therefore extremely potent.
Genetic properties and organization of the ure operon
The urease operon was originally characterized from a DNA fragment of 7.6 kb, derived from a cosmid gene bank of P. mirabilis genomic DNA. Six open reading frames were found within a 4,952 bp region, which were predicted to encode polypeptides of 31.0 (UreD), 11.0 (UreA), 12.2 (UreB), 61.0 (UreC), 17.9 (UreE), and 23.0 (UreF) kDa (190). Reexamination of recombinant clones for urease activity revealed that a seventh gene, ureG, was also required for production of catalytically-active urease enzyme (191) Sequences that preceded these genes later determined to confer urea-inducibility upon the operon (192), resulting in identification of UreR, a 33.4 kDa helix-turn-helix AraC family member responsible for urea-induction of the operon.
Regulation by UreR
UreR is a dimer that binds to two sites within the ureR-ureD intergenic region at a consensus binding site of TA/GT/CA/TT/GC/TTA/TT/AAATTG as predicted from DNASE I protection assays (193). UreR binds to these sites in a urea-dependent manner to activate expression of the urease operon. In the absence of urea, there is no measurable binding or induction of urease expression.
Regulation by H-NS
A poly(A) tract nucleotide sequence of A6TA2CA2TGGTA5GA6TGA5 is present 16 bp upstream of the ureR promoter, which serves as a binding site for the gene repressor histone-like nucleoid structure protein (H-NS) (194, 195). Using a ureR-lacZ reporter plasmid, it was determined that H-NS represses transcription of ureR, and therefore represses urease expression (194, 196). H-NS and UreR compete for binding the ureR-ureD intergenic region, and culture conditions dictate whether the operon will be repressed or activated. H-NS binding is favored at 25°C and in the absence of urea, while UreR binding and derepression of the operon is favored at 37°C when urea is present (196).
Role in virulence
Urease is a critical virulence determinant for P. mirabilis urinary tract infection. Indeed, the first targeted mutation of this species involved the mutation of ureC, the major structural subunit of the apoenzyme (40). The ID50 of the ureC mutant is approximately 3 logs higher than the parental strain, clearly indicating the contribution of urease to disease severity. In an independent challenge with 109 CFU of the wild-type strain or the ureC mutant, the mutant was attenuated by approximately 2 logs in the bladder and kidneys 48-hours post-inoculation and up to 6 logs by 7 days post-inoculation (39, 40). Infection with the ureC mutant also resulted in significantly less pathology in the bladder and kidneys of infected mice. It was later shown that a ureR mutant lacking urea-induction of urease synthesis also resulted in attenuation by >3 logs in the bladder and kidneys (160).
The contribution of urease to pathogenicity has also been assessed in the uncomplicated murine model of UTI compared to a model of catheter-associated UTI (CAUTI), in which a catheter segment is placed in the bladder during inoculation (25). These studies determined that urease significantly contributes to cystitis in both models, but does not play as dramatic a role in pyelonephritis in the CAUTI model. However, urease activity strongly correlated with renal tubule damage and nephrosis during both UTI and CAUTI in this study.
Potentiation of urease activity during polymicrobial infection
Urine colonization in catheterized individuals is frequently polymicrobial, and P. mirabilis is one of the most common organisms present during polymicrobial colonization and CAUTI (4, 8, 9, 17). It is therefore noteworthy that co-culture with other uropathogens can enhance production of active urease enzyme by P. mirabilis (25, 41). In addition to P. mirabilis, the most common CAUTI uropathogens are Enterococcus species, Escherichia coli, Pseudomonas aeruginosa, and Klebsiella pneumoniae, and multiple isolates of all four species are capable of enhancing P. mirabilis urease activity (25). This has important implications for the pathogenesis of polymicrobial CAUTI, as enhanced urease activity resulting from co-culture of P. mirabilis and P. stuartii was determined to significantly impact disease severity, being directly associated with increased urine pH, greater incidence of urolithiasis and bacteremia, and induction of a more potent and destructive inflammatory response (25, 41).
Urease inhibitors
Due to the important role of urease during P. mirabilis pathogenesis in the urinary tract, urease inhibitors have been explored as a potential therapeutic (197). Such inhibitors include hydroxyamic acids, which bind the nickel atoms in the urease active site, and acetohydroxamic acid (AHA), a structural analog of urea. AHA is the only analog to have been tested in clinical trials and be approved by the Food and Drug Administration. However, while it showed efficacy for preventing urolithiasis (198–200), it has potentially severe side effects that limit its clinical use (201). Thus, further research is necessary to identify safe and effective urease inhibitors. It is also worth noting that while the ure operon is urea-inducible, structural analogs of urea such as AHA do not induce urease expression (202).
B. Secretion systems
Proteus mirabilis secretes proteins via Type I, III, IV, V, and VI systems as well as the Sec pathway and Twin Arginine Targeting (Tat) system. These pathways are predicted from the genome sequence (112), as analyzed by KEGG pathways (203) and from the individual studies described below. While examples of secreted proteins can be found for the systems noted, only a few examples of secreted proteins that have been tested for their contribution to virulence have been documented.
Proteases
In an early study, it was found that 90–94% of 48 P. mirabilis strains secreted proteolytic enzymes, detected at pH 8 using gelatin as substrate. A correlation between the ability to swarm and protease secretion was noted. These proteases were all EDTA-sensitive metalloproteinases (204). However, the specific secretion systems involved in secretion of these proteases were not determined in this study.
Type III Secretion
Nucleotide sequencing of the genome of P. mirabilis HI4320 (112) revealed genes appearing to encode a type III secretion system on a low GC-content pathogenicity island (148). A cluster of 24 genes clearly have the potential to encode an intact T3SS with genes for an intact needle complex and at least two effector proteins and their chaperones. The genetic organization of the T3SS is similar to that of Shigella flexneri, there was no evidence of mutation of the genes resulting in inactivation of the system, and RT-PCR analysis demonstrated that these genes were expressed in P. mirabilis HI4320. However, mutation of two genes including a putative ATPase and negative regulator of T3S had no effect on the secreted protein profile when compared to the wild-type strain (148). Furthermore, in the murine model of ascending UTI, there was no difference in CFUs between the mutant and wild-type strain in urine, bladder, or kidneys during either independent infection or cochallenge. Thus, no pathogenicity phenotype for the T3SS was identified for P. mirabilis HI4320 during ascending UTI. However, two components of the T3SS (PMI2688 and PMI2696) were identified as significant fitness factors during polymicrobial CAUTI by Tn-Seq (129), indicating that the T3SS may be involved in mediating host-microbe interactions during polymicrobial infection.
Type V Secretion
There are three subclasses of type V secretion systems (Va, Vb, and Vc), and the nucleotide sequence of P. mirabilis HI4320 predicts members of all three (112, 205, 206).
Classical autotransporters (Va)
Classical autotransporters are a family of virulence proteins in Gram-negative pathogens that contain three domains: an amino-terminal leader peptide for export across the inner membrane via the Sec pathway, a surface-localizing passenger domain, and a carboxy-terminal domain for translocation across the outer membrane (Henderson 2004). One of the three predicted classical autotransporters encoded by P. mirabilis HI4320 has been studied in detail: Proteus toxic agglutinin (pta), encodes a serine protease that was previously identified as an immunogenic outer membrane protein (161) (link Toxins section). The 117-kDa protein has a 58-amino acid long signal peptide, a 75-kDa long N-terminal passenger domain, and a 30-kDa C-terminal translocator (110). The autotransported protease, either cell-associated or secreted, has cytotoxic effects on cultured bladder and kidney epithelial cells. Catalytic residues Ser366, His147, and Asp533, when mutated, abolish protease activity. The protein also has autoaggregation properties not associated with the proteolytic activity. Virulence of a protease mutant, as tested in the murine model of ascending UTI, was significantly reduced.
Two-Partner Secretion (Vb)
P. mirabilis hemolysin, encoded by hpmA, is exported by the protein product of hmpB via the type Vb two-partner secretion pathway. This requires sequential unfolding of the HpmA protein (207), for which a partial crystal structure has been solved (208). Two polypeptides, HpmB and HpmA, synthesized in that transcriptional order, are responsible for hemolysin activity of P. mirabilis (link Toxins section). A Fur-binding site upstream of hmpB overlapping the −35 site of the promoter indicates that expression is governed by iron concentration. HpmB and HpmA have 17- and 29-amino acid leader peptides, confirmed for purified HpmA. HpmB is necessary for secretion of HpmA. HpmB and HpmA are most similar with respect to amino acid sequence to the Serratia marcescens hemolysin proteins ShlB and ShlA, respectively (209).
Trimeric autotransporters (Vc)
P. mirabilis HI4320 encodes three putative trimeric autotransporter proteins (Pearson 2008). Two of these, AipA and TaaP, are annotated as “adhesin-like” and “agglutinating adhesin-like”. Based on their homology with other trimeric autotransporters, the two autotransporters would contain four antiparallel β-sheets and form homotrimers. Recombinant AipA and TaaP bind extracellular matrix proteins, produce polypeptides of 28-kDa and 78-kDa, respectively, and form high molecular weight homotrimers. A 51-amino acid invasin-like motif of AipA is necessary for function. Gly-247 in AipA and Gly708 in TaaP are required for trimerization and activity. AipA and TaaP confer an advantage at 7 dpi during cochallenge in the murine model of ascending UTI.
Type VI secretion
P. mirabilis has long been observed to distinguish between disparate strains when swarming on agar surfaces, and there is now substantial evidence that this phenomenon is mediated by type VI secretion systems (T6SS). The discovery of T6SSs in P. mirabilis stemmed from an unusual observation about this species. The Dienes phenomenon, first described in 1946 by Louis Dienes, refers to the fascinating ability of two P. mirabilis swarming colonies of a single strain to merge with each other, while swarms of different strains form a line of demarcation where they meet (210) (Fig. 17). Formation of the Dienes line requires direct cell-cell contact by living bacteria, and involves killing of at least one strain at the boundary (211) (Fig. 18).
The structural components of the T6SS are Hcp and VgrG, which comprise the hollow tube and puncturing tip of the apparatus, VipA and VipB (or TssBC) that form the tube-like structures or sheath of the apparatus, and ClpV to mediate ATP hydrolysis for polysheath disassembly (120). Work conducted in P. mirabilis strain BB2000 identified an operon that contributed to Dienes line formation, named Ids for identification of self (175), which was found to encode Hcp and VgrG (175) and therefore provided the first link to T6S and Dienes line formation. Further investigations in P. mirabilis strains BB2000 and HI4320 determined that formation of the Dienes line represents either T6S-mediated killing of one strain by the other, or both strains killing each other using T6SSs (118, 119).
Using a transposon screen for P. mirabilis HI4320 mutants that failed to kill opposing strains, a 33.5-kb region of chromosome, different from the Ids locus, was identified as harboring the T6SS apparatus (PMI0749-PMI0733) and the primary effector operon (pefABCDEFG) (PMI0750-PMI0758) of this strain (119). This operon was found to encode a single immunity protein, PefE, which was responsible for providing immunity from killing by the Pef effectors. However, strain HI4320 was also found to encode at least four other potential T6S effector operons that could use the T6S apparatus, including idsABCDEF (175, 212) and three additional operons (PMI0207-PMI0212, PMI1117-PMI1121, and PMI1332-PMI1324) (112, 119) (Fig. 19). Differential use of these systems or possibly even recombination between them may therefore contribute to the wide range of strain-killing activities and Dienes types exhibited by P. mirabilis isolates.
A direct role for any individual P. mirabilis T6S operon during infection has yet to be elucidated. For instance, none of the T6S operons encoded by P. mirabilis HI4320 exhibited fitness defects in the murine model of CAUTI by transposon insertion-site sequencing (Tn-Seq) of monospecies infection. However, all of the known P. mirabilis T6S operons were strikingly overrepresented as fitness factors during coinfection with Providencia. stuartii (129), indicating a potential role for T6S in mediating competitive and cooperative interactions during polymicrobial infection. It remains to be determined if this system is important for mediating microbe-microbe interactions or microbe-host interactions in this infection model.
C. Toxins
While the complete genome sequence of P. mirabilis strain HI4320 revealed numerous potential toxins, only three have been well-characterized for a role in virulence. These are hemolysin, Proteus toxic agglutinin, and the ZapA metalloprotease.
Hemolysin
Secreted pore-forming toxins are a common feature of pathogenic bacteria. In particular, hemolysins are secreted pore-forming toxins that insert into eukaryotic cell membranes, causing efflux of sodium ions and cell damage (213). Two hemolysins have been described for members of the Proteus genus (214), one that is calcium-dependent and similar to the α-hemolysin of E. coli (hlyA), and another that is calcium-independent. The calcium-independent hemolysin is encoded by two genes comprising a two-partner secretion system: hpmA, which produces a 166 kDa secreted exoprotein, and hpmB, which produces a 63 kDa translocase protein that is required for both the activation and secretion of HpmA through cleavage of an N-terminal peptide (209). HpmA is the predominant hemolysin in Proteus species and appears to be the only hemolysin encoded by P. mirabilis, present in 268/274 (98%) of P. mirabilis clinical and fecal isolates from Brazil and the United States while none of these isolates encoded hlyA (102, 103). Similarly, hpmA is present in all 7 complete P. mirabilis genome sequences available through NCBI as of May 2017. An N-terminal fragment of HpmA, called HpmA265, has been crystallized and its structure solved (Fig. 20) (208). This fragment lacks the C-terminal pore-forming domain, but can still activate full-length HpmA in the absence of HpmB when the proteins are mixed with erythrocytes. The formation of HpmA265 dimers suggests that full-length HpmA may also function as a dimer.
HpmA mediates lysis of a broad range of cell types from numerous host species and appears to be the primary P. mirabilis virulence factor responsible for cytotoxicity to human renal proximal tubular epithelial cells (HRPTECs) (47, 48, 50, 56). Deletion of hpmA dramatically decreases cytotoxicity, and allows for internalization of P. mirabilis by HRPTECs in vitro (47, 56). As HRPTECs form a protective barrier for the kidney parenchyma, it was hypothesized the hemolysin may be a critical virulence factor that mediates spread of P. mirabilis into the kidneys and development of pyelonephritis (28). However, deletion of hpmA does not appear to impact tissue colonization or damage during independent challenge in the murine model of ascending UTI (45, 48, 55), indicating that either hemolysin has less of an impact during experimental infection than the in vitro cell culture studies suggest, or the activity of other virulence factors mask its contribution in vivo. In agreement with this finding, hpmBA was not identified as a fitness factor for UTI or CAUTI in any of the genome-wide transposon mutagenesis studies (108, 129, 179, 180), likely due to a combination of complementation in trans by the other hemolysin-producing transposon mutants present during infection and possibly production of other cytolysins with similar functions.
Proteus toxic agglutinin
The genome of P. mirabilis strain HI4320 encodes six putative autotransporters (112) (link secretion section), of which only the Proteus toxic agglutinin (Pta) has been fully characterized. Pta was initially identified as an outer membrane surface-expressed protein recognized by the murine immune system (161), and determined to be encoded within the integrative and conjugative element ICEPm1 in P. mirabilis strain HI4320 (107) (link to ICEPm1).
Pta is a 120 kDa protein that is catalytically processed to become a ~110 kDa active protein within the outer membrane (110). Expression of this protein in E. coli promotes autoaggregation of the bacteria, as well as lysis of bladder epithelial cells in vitro. Both properties of Pta require translocation to the outer membrane and cytotoxicity requires serine protease activity from the passenger domain, but protease activity is not required for autoaggregation. Pta is also classified as a subtilisin-like alkaline protease owing to the observation that expression of pta is induced by alkaline pH and protease activity is maximal at pH 8.5–9.0, conditions that would be encountered by P. mirabilis during urinary tract infection due to its urease activity.
Pta contributes to colonization of the bladder and kidneys in a murine cochallenge model of ascending UTI, as well as dissemination to the spleen (110). Interestingly, HpmA and Pta have an additive effect on cytotoxicity both in vitro and during experimental UTI, particularly with respect to cystitis and possibly interstitial nephritis (55). However, Pta appears to be the more potent toxin during experimental infection, as disruption of pta has a much greater impact on infectivity than loss of hpmA. Similar to hpmA, pta is present in all seven of the complete P. mirabilis genome sequences in NCBI as of May 2017, and was also encoded in eight P. mirabilis infection and fecal isolates (55).
ZapA Metalloprotease
P. mirabilis produces a metalloprotease with broad specificity, originally thought to be an IgA protease but later determined to be capable of cleaving IgA, IgG, secretory component (the heavily glycosylated protein that complexes with dimeric and polymeric IgA), antimicrobial peptides hBD1 and LL-37, complement protein C1q and C3, fibronectin, actin, collagen, laminin, casein, and gelatin (215–219). In all cases, cleavage was sensitive to the metal chelator EDTA, indicating that the enzyme is a metalloprotease. This protease was identified as a 55 kDa metalloprotease of the serralysin family, encoded by zapA (220, 221).
The contribution of ZapA to pathogenicity has been explored both in vitro and in vivo. Intact IgG and IgA1 can interact with Fc receptors on neutrophils to stimulate a respiratory burst, and this process is significantly reduced when IgG has been cleaved into Fab and Fc fragments in vitro by the P. mirabilis metalloprotease (218). Similarly, cleavage of hBD1 and LL-37 by ZapA reduces their antimicrobial activity (219). Thus, ZapA may contribute to evading the innate immune response during infection. A ZapA-deficient mutant, specifically constructed by insertional inactivation of zapA in the chromosome of P. mirabilis BB2000, was used to assess virulence in the murine model of ascending UTI (134). CBA mice were transurethrally inoculated into the bladder with 106 CFU P. mirabilis BB2000 or the zapA mutant. After 7 days, quantitative cultures indicated that the zapA mutant was dramatically attenuated (by 3 logs in the urine and 5 logs in the bladder or kidney), indicating that the protease contributed strongly to virulence of P. mirabilis BB2000. The same wild-type and mutant pair were tested in a rat model of prostatitis (168). Unlike the wild-type, inoculation of the zapA mutant resulted in reduced levels of acute prostatitis as determined by lower levels of tissue damage, bacterial colonization and inflammation (Fig. 21).
D. Integrative and conjugative element ICEPm1
Pathogenicity islands (PAIs) are a specific group of genomic islands that contribute to genomic variability and virulence of bacterial pathogens. PAIs carry virulence determinants and are typically present in pathogenic strains but absent in nonpathogenic strains of the same species. These islands consist of large genomic regions ranging from 10–200 kb, and often have different G+C contents as compared to the host. They also frequently contain mobile genetic elements, are flanked by direct repeat sequences, have mosaic-like structure, insert at the site of tRNA genes, and have likely been acquired by horizontal gene transfer (222, 223).
A subset of PAIs has been identified that can excise from the bacterial chromosome, following a recombination event at the site of direct repeats flanking the island, and actively transfer via a type IV secretion system to another bacterium. These are called Integrative and Conjugative Elements (ICE) and many P. mirabilis strains carry an ICE designated ICEPm1 (107) (link comparative genomics).
ICEPm1 is a 94-kb region with a G+C content of 44.84%, which differs substantially form that of the P. mirabilis genome (38.88%) (107, 112), and contains core modules and a syntenic structure consistent with prototypical ICEs (113, 224) (Fig. 22). Direct repeats flank ICEPm1, and at the left-most end the 52-bp direct repeat is located with the 5′ coding end of tRNAPhe gene (attL). The direct repeat at the right-most end (attR) is part of the 3′ end of a truncated tRNAPhe gene. In the genome of P. mirabilis HI4320, ICEPm1 is annotated as being integrated into the phenylalanine tRNA gene pheV. However, ICEPm1 can integrate into either pheV or pheU sequences (111).
ICEPm1 contains 91 ORFs, with the core genes constituting putative integration, replication and conjugative modules, including an integrase, genes for a putative tyrosine-like recombinase, putative helicase that could act as a relaxase, six transposases, and five plasmid-transfer related proteins (107). A 26-gene region encodes eight putative exported proteins and nine putative membrane proteins comprising the T4SS.
ICEPm1 also encodes known “cargo” genes that contribute to virulence, interspersed between the core modules. These cargo genes include the nrp operon (PMI2596-PMI2604), which encodes genes for the synthesis, transport, and uptake of an iron siderophore (link metals section), and Proteus toxic agglutinin (Pta), which contributes to aggregation of P. mirabilis and represents a potent autotransported protease toxin (46, 109, 110, 161).
Highly conserved homologs of ICEPm1 were present in all of 39 P. mirabilis urinary isolates tested and were also found in Providencia stuartii (6 of 10 isolates) and Morganella morganii (11 of 38 isolates) from cases of catheter-associated bacteriuria (107). ICEPm1 was also found at a reduced frequency in colonizing isolates of P. mirabilis (15 of 23 isolates) that were cultured from the oropharynx, nasopharynx, wound, groin, or perianal area.
When the ICE-encoded integrase is activated, the ICE can excise from the chromosome, form a circular intermediate, and subsequently transfer to a recipient cell via a mating pore formed by the ICE-encoded T4SS (113, 225). Transfers occurs at a frequency of 1.35 × 10−5 transconjugants/donor to ICEPm1-deficient P. mirabilis (111). Insertional inactivation of a putative integrase gene (PMI2549), a specific recombinase of the tyrosine-like family, on ICEPm1 decreases transfer frequencies to below the limit of detection. Mutation of the relaxase of ICEPm1 also eliminates transfer (111).
E. Fimbriae and adhesins
Fimbriae (pili) are hair-like protein structures that extend from the bacterial surface and usually mediate adherence to surfaces. Gram-negative bacteria produce a subset, called chaperone-usher fimbriae for their method of secretion and assembly (226, 227). P. mirabilis HI4320 encodes 17 chaperone-usher fimbrial operons (112). Although the function of most of these is not yet defined, transcription has been detected from all 17 operons, and most of these fimbriae are encoded by P. mirabilis isolated from diverse sites (Fig. 11) (106). For example, a PCR screen for the predicted major structural subunit gene for each fimbria showed that 85% of clinical isolates encode at least 14 unique fimbriae (106). Of seven clades defined for classical chaperone-usher fimbriae, three are represented in P. mirabilis (γ1, γ2, and π) (Table 6) (106, 228).
Table 6
Fimbria | Genes | Classa | Implicated in virulenceb, c | Protein | MrpJ paralog | Ref | ||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|
IC | CO | STM | SS CAUTI | CO CAUTI | Exp | In vitro | ||||||
MR/P′ | PMI0254-PMI0261 | π | PMI0261 | (129) | ||||||||
MR/P | PMI0262-PMI0271 | π | ↑ | * | PMI0271 | (44, 100, 106, 129, 131, 146, 149–152, 158, 229, 240–245) | ||||||
Fimbria 3 | PMI0296-PMI0304 | π | PMI0296 | (129) | ||||||||
UCA | PMI0532-PMI0536 | γ1 | ↓ | PMI0532 | (106, 129, 131, 163, 246–251) | |||||||
Fimbria 5 | PMI1060-PMI1067 | π | ↓ | PMI1060 | (129, 131) | |||||||
Fimbria 6 | PMI1185-PMI1190 | γ1 | None | |||||||||
Fimbria 7 | PMI1193-PMI1197 | γ1 | None | |||||||||
Fimbria 8 | PMI1464-PMI1470 | γ1 | ↓ | PMI1470 | (106, 131) | |||||||
PMF | PMI1877-PMI1881 | π | ↓ | * | None | (106, 129, 131, 151, 153, 154, 243, 252) | ||||||
Fimbria 10 | PMI2207-PMI2214 | γ1 | PMI2209, PMI2207 | (129) | ||||||||
PMP | PMI2216-PMI2224 | π | PMI2224 | (129, 253) | ||||||||
Fimbria 12 | PMI2533-PMI2539 | γ2 | None | (129) | ||||||||
ATF | PMI2728-PMI2733 | γ1 | PMI2733 | (106, 129, 157) | ||||||||
Fimbria 14 | PMI2997-PMI3003 | ND | ↑ | PMI3003 | (106, 129, 131, 179) | |||||||
Fimbria 15 | PMI3086-PMI3093 | π | None | (129, 131) | ||||||||
Fimbria 16 | PMI3348-PMI3352 | γ1 | None | (129) | ||||||||
Fimbria 17 | PMI3435-PMI3440 | γ1 | None | (129) |
Table adapted from (19)
P. mirabilis may produce multiple types of fimbriae at once. For example, at least two fimbrial morphologies have been seen on a single bacterium by transmission electron microscopy (229, 230). In a mass spectrometry study, proteins from six of the 17 fimbriae were detected in a sheared surface protein preparation from a single broth culture (MR/P, UCA, PMF, ATF, Fim8, and Fim14) (106). This ability sets P. mirabilis apart from other bacteria such as E. coli that tend to produce one type of fimbria at a time (231). Likewise, P. mirabilis devotes a larger share of its chromosome to encoding fimbriae than most other fimbriated pathogens; for example, uropathogenic E. coli typically encode 9–12 fimbrial operons (98).
Because fimbriae are often found to contribute to virulence, there are multiple studies examining the roles of P. mirabilis-encoded fimbriae to UTI (Table 6) (19, 20, 232). P. mirabilis obtained from healthy stool or diarrhea, urine, or blood can adhere to voided human uroepithelial cells regardless of the bacterial source (233), which is consistent with the conservation of fimbrial genes in this species (106). In general, mutations in individual fimbrial genes result in bacteria that are still able to colonize the urinary tract, but are less fit in direct competition cochallenge experiments (151, 153, 234). This suggests that the 17 fimbriae have overlapping or redundant functions. Furthermore, as surface-localized, abundant, immunogenic proteins, P. mirabilis fimbriae have been tested as vaccine candidates with some success (link vaccines section) (235–239).
Mannose-resistant Proteus-like fimbriae (MR/P)
These are the most-extensively studied of the fimbriae encoded by P. mirabilis. MR/P fimbriae were originally named for their ability to agglutinate untreated erythrocytes in the presence of mannose, a trait which was frequently found in P. mirabilis isolates (229) (Fig. 23). This was in contrast to mannose-sensitive agglutination, e.g., type 1 fimbriae from E. coli, or mannose-resistant Klebsiella-like (MR/K) agglutination, which occurred in the presence of tannic acid-treated bovine erythrocytes (254). MrpA is the major structural subunit of the fimbria, and MrpH is a two-domain adhesin which is located at the tip of MR/P fimbriae. Extensive details on MR/P structure and assembly have been recently reviewed elsewhere (19).
Expression
The mrp gene cluster consists of two transcripts. Genes required for fimbrial structure and assembly are encoded by mrpABCDEFGHJ (156). The last gene, mrpJ, encodes a transcriptional regulator and is discussed later. The second transcript comprises a single gene, mrpI, which is divergently transcribed from the rest of the mrp genes (241). MrpI is a recombinase that flips a 252 bp invertible element containing the mrpABCDEFGHJ promoter (241). The transcriptional start site lies within the invertible element (132), and thus the orientation of this element dictates whether or not mrp fimbrial genes are transcribed. MrpI is the sole recombinase for the invertible element, and therefore mutations in mrpI result in bacteria that either constitutively produce MR/P fimbriae (“locked on”) or are devoid of MR/P fimbriae (“locked off”) (158). The orientation of the element may be easily detected using PCR-based assays (Fig. 24A), and phase variation of MR/P fimbriae has been observed at the single cell level (Fig. 24B).
The nine genes in the mrp operon are the most highly induced genes by bacteria in urine collected from experimentally-infected mice compared with in vitro broth culture (131). Correspondingly, even when mice are infected with P. mirabilis cultured so the population is almost completely phase off, after seven days, the majority of the bacteria have the invertible element in the on orientation (241). MR/P fimbriae are produced at a relatively low level in vitro, with the highest amounts produced during 48 h static culture in broth or culture under 5% oxygenation, followed by aerated broth, and finally, growth on an agar surface (152, 249, 255). Expression of mrp genes is both induced and subject to positive selection in a 5% oxygen atmosphere, which is logical for a virulence factor given reduced oxygen availability in the bladder (152). That is, an mrp locked on mutant outcompetes locked off in broth, and, at the same time, mrpA expression increases during oxygen limitation (152). Transcription of the recombinase mrpI may also be regulated by oxygen levels (152). Furthermore, orientation of the invertible element is influenced by the bacterium’s ability to assemble fimbriae. Assembly requires disulfide bond formation, and mutation of the gene encoding thiol:disulfide interchange protein DsbA leads to an increase in element-off bacteria (152).
In vitro function
In addition to hemagglutination, MR/P fimbriae mediate in vitro phenotypes that give clues to their function during infection. Biofilm formation (link biofilm section), measured in plastic multi-well plates, on glass coverslips, as pellicles in culture tubes, or in static cultures in glass tubing, is dependent on MR/P fimbriae (156, 244, 256). Interestingly, phase variation contributes to biofilm development. MR/P locked on bacteria establish biofilms more rapidly than wild-type, but over time the wild-type biofilm is much thicker (65 μm vs 12 μm) and has a classic structured channel appearance (244). In contrast, the locked on biofilm is short, very dense, and lacks channels. MR/P locked on bacteria autoaggregate (152, 244), which is reminiscent of the bacterial clustering that happens prior to stone formation in the bladder (44) (Fig. 25A–D). Furthermore, MR/P fimbriae allow P. mirabilis to adhere to some cultured cell lines, including T24 and Vero (150, 245). There are conflicting reports about MR/P-mediated binding to shed bladder epithelial cells, where purified MR/P and MR/P-positive P. mirabilis bound cells (240), but there was no difference in binding between a different wild-type strain and its isogenic mrpA mutant (149). It is possible that these two studies were in fact not examining the same fimbria (19).
Contributions to infection
MR/P fimbriae contribute to UTI as described in multiple studies (Table 6). The magnitude of deficiency of an mrp mutant in mouse model independent challenges varies from study to study, from the initial mrpA experiment, where a 6–18-fold decrease was observed in urine, bladder, and kidneys, to a four log defect in the bladder for an mrpG mutant (146, 149, 151). A mutation in mrpH, which encodes the fimbrial adhesin, does not decrease fitness in independent challenge, although the mutant was unrecoverable following a one-week cochallenge experiment (156). Subsequent cochallenge infections with an mrpA or an mrpA-D mutant likewise resulted in the mutant being significantly out-competed in both the bladder and kidneys (150, 151).
Newer evidence may explain the observed variability in the requirement for MR/P expression. Two studies have reported equal numbers of recovered bacteria from the bladders of mice infected with wild-type compared with an mrpA mutant (44) or mrp locked-off (244). However, in both studies, the mrp mutant was not found in the same location as wild-type, suggesting that MR/P fimbriae not only facilitate adherence to urinary tract tissue but also dictate the localization of P. mirabilis within the bladder. Notably, both of these studies used a relatively early endpoint (24 hpi) (44, 244) compared with other mouse independent challenges lasting 7 days (146, 149, 151, 156). Thus, the location of P. mirabilis in the urinary tract, whether extracellular, intracellular, or bound to regions where uroepithelial cells have sloughed off, may be important for its long-term survival (Fig. 25E and F).
Although prior studies suggested that MR/P fimbriae bind a receptor in renal cells and contribute to pyelonephritis (149, 240), recent data point to their importance in cystitis. First, when bacteria are isolated from the bladder or urine of infected mice, the invertible element of the mrp promoter is overwhelmingly (95–100%) in the on orientation, allowing transcription of mrp genes. In contrast, the invertible element orientation is highly variable in the kidneys, ranging from mostly on to 85% off (158, 241). Second, the percentage of bacteria with the mrp invertible element in the on orientation directly correlates with the amount of bladder colonization (Fig. 24C) (152). Third, when a locked-on mrpI mutant is competed against wild-type bacteria in a 7 d mouse cochallenge experiment, locked-on bacteria outcompete wild-type in the bladder but not the kidneys (158). Fourth, P. mirabilis locked-on mrpI mutants adhere to the luminal surface of the bladder when examined 24 hpi, while locked-off bacteria adhere to regions where umbrella cells have been shed, revealing the lamina propria (244). Fifth, wild-type P. mirabilis forms large clusters in the bladder lumen within 24 hpi, but an mrpA mutant does not form clusters and is primarily intracellular (44) (Fig. 25A–D). The precise function of MR/P fimbriae and their contributions to cystitis and/or pyelonephritis will be difficult to address until their binding target(s) are identified.
Uroepithelial cell adhesin (UCA)
A fimbrial preparation from P. mirabilis cultured to stationary phase in a minimal medium at 37 °C was used to isolate a long, 4 nm-thin protein that adhered to human uroepithelial cells obtained from voided urine (246). The importance of this protein, UCA, in uroepithelial cell adhesion was subsequently confirmed when a mutation of the gene encoding the major structural subunit of UCA (ucaA) was tested with desquamated cells (163).
A second group working on UCA renamed the protein “non-agglutinating fimbriae” (NAF) and found the protein in shear preparations obtained from diverse P. mirabilis isolates (248). UCA/NAF are readily expressed under several laboratory conditions at 37°C, including Luria broth or agar; production may be enhanced by addition of serum to Luria broth or during culture on urine agar (248, 249).
Expressing cloned uca genes in E. coli conferred the ability to bind uroepithelial cells (247). However, no fimbriae were observed on the recombinant E. coli; only ucaA was sequenced and it is not clear whether all genes in the uca operon were present in the expression construct (247). A UCA/NAF preparation purified from P. mirabilis binds to cultured cells, including EJ/28 (urinary carcinoma), HEp-2 (laryngeal carcinoma), and MDCK (canine kidney epithelium) (249, 251), although these results have not been confirmed using other methods, such as by comparing adhesion of a uca mutant to that of wild-type.
Although UCA fimbriae are encoded by most P. mirabilis isolates, their sequence is more variable compared with other fimbriae (106). Antiserum raised against purified fimbriae further indicated that UCA is widely present but variable in size (248). It is possible that variations in UCA may explain distinct phenotypes reported by different groups, who are typically using locally-obtained isolates in their experiments. Notably, the uca operon is part of a mobile element and flanked by phage genes; in the first two fully-sequenced P. mirabilis genomes, HI4320 and BB2000, the uca operon is not in the same chromosomal locus (106).
In vitro receptor experiments
A possible receptor for UCA/NAF has been identified using fimbrial protein purified from a shear preparation using differential centrifugation (248, 250). Purified UCA/NAF bound in vitro to glycolipids, including asialo-GM1, asialo-GM2, and lactosyl ceramide (250). A follow-up study suggested another molecule, galectin-3, was also a receptor for purified UCA/NAF on MDCK cells (251). The possible receptors have not been identified in healthy bladder tissue, although galectin-3 is a marker for bladder cancer (257). Thus, it is not apparent whether these molecules are true in vivo targets for UCA.
Infection studies
Given that bladder cell adherence is a defining feature of UCA, it is perhaps paradoxical that a ucaA mutant is significantly attenuated in the kidneys but not the bladder during either independent or cochallenge of the murine urinary tract (163). When this ucaA mutant was injected into the tail vein of mice, bacteria were only recoverable from 7/16 mice 7 days post-inoculation, while the wild-type parent was recovered from 15/16 mice (163). Transcription of ucaA has been detected in urine from infected mice, although the level was 2.5-fold decreased relative to in vitro broth culture (131). UCA also contribute to CAUTI, where ucaA and ucaB were both detected in a Tn-Seq study (129).
Proteus mirabilis fimbriae (PMF)
PMF fimbriae were initially identified when a P. mirabilis cosmid gene bank was screened using antiserum generated against UCA (255). These fimbriae are encoded by the pmfACDEF operon (226, 258). Despite initial speculation that PMF was the unidentified MR/K agglutinin (255), a pmfA mutant had no effect on either MR/P or MR/K-type agglutination (153). Nevertheless, PMF fimbriae contribute to UTI in mouse models, with an 83-fold decrease in recoverable CFU for a pmfA mutant during independent challenge in the bladder (153). The defect was more profound in cochallenge, where the mutant was significantly attenuated in both bladder and kidneys (154). In a hematogenous model, where P. mirabilis was introduced through the tail vein of mice, the pmfA mutant was unrecoverable from the kidneys of 55% of mice 7 days post-inoculation, while the wild-type parent was still recovered from 94% (154). There are conflicting data on whether PMF fimbriae mediate adherence to voided human uroepithelial cells (153, 154); however, a pmfA mutant had diminished adherence to cultured T24/83 bladder carcinoma cells (154). In summary, evidence exists for PMF to contribute to infection in the bladder, kidneys, or both, and the receptor(s) have not yet been identified. PMF also contribute to fitness during CAUTI (129).
Interestingly, although PMF are expressed during UTI and mice infected with P. mirabilis generate an immune response against these fimbriae, pmf genes are repressed in urine relative to in vitro culture (131, 243). This suggests that PMF fimbriae may be transiently produced during infection, or that PMF-producing bacteria are not readily voided in urine. A pmfA/mrpA-D double mutant is more attenuated during a triple challenge in mice than either single mutation or wild-type, but interestingly, the double mutant is still infective in independent challenge (albeit less than the wild-type parent strain) (151). This is consistent with overlapping roles for the P. mirabilis “pan-fimbriome” and suggests that other fimbriae also contribute to adherence in the urinary tract (Fig. 26).
Ambient temperature fimbriae (ATF)
ATF fimbriae were first identified in a preparation of sheared surface proteins derived from an mrpA mutant (234). The atf operon was originally reported to consist of three genes, atfABC (259) as isolated from a cosmid clone, but subsequent genome sequencing revealed three additional genes (atfDEF) including an mrpJ paralog (112). Although ATF are produced at 37°C (106, 234), they are optimally expressed following static culture at 23°C (234). In a similar vein, an atfA mutant is recovered at the same rate as wild-type from either independent or cochallenge mouse studies (157). However, when an mrpA mutant was examined by immunofluorescence microscopy in infected mice, AtfA was readily detected (Fig. 27) (158, 244). This suggests that ATF may play a role in infection that is masked by other fimbriae, such as MR/P, and also there may be a fimbrial hierarchy that determines gene expression. ATF may contribute to biofilm formation under some conditions (256).
P. mirabilis P-like fimbriae (PMP)
This fimbria was first discovered in a sheared protein preparation from a canine UTI isolate of P. mirabilis, where PMP was found along with UCA (253). PMP fimbriae have also been implicated in both single-species and polymicrobial CAUTI, where pmpG was identified by Tn-Seq (129), and a pmpA mutant was found to be attenuated in a diabetic mouse UTI model (260). The pmp operon was also reported to be directly regulated by cyclic AMP receptor protein (Crp) (260).
Fimbria 14
Fimbria 14, so named because it is encoded by the fourteenth of the 17 chaperone-usher fimbrial operons in HI4320, seems like it should be non-functional because the fim14 operon completely lacks a chaperone and has a frameshift mutation in usher gene fim14C (112). However, putative minor fimbrial subunit fim14B was detected in an STM screen for genes that contribute to UTI (179), and putative major structural subunit Fim14A was identified in a preparation of sheared surface proteins (106). The last gene in the operon, fim14D, is induced during UTI (131), and genes from the fim14 operon were found to contribute to polymicrobial CAUTI (129). Although chaperone-usher secretion apparatuses are usually specific for a single fimbrial type, it is likely that Fimbria 14 is assembled using one or more of the other 16 chaperone-usher systems encoded by P. mirabilis.
Other fimbriae
The remaining fimbriae encoded by P. mirabilis are poorly characterized (Table 6). P. mirabilis have been reported to employ an unknown fimbria to agglutinate tannic acid-treated erythrocytes, a pattern that is called mannose-resistant Klebsiella-like (MR/K) (100, 229, 230, 261). This ability is variable for P. mirabilis, with some studies suggesting MR/K agglutination by P. mirabilis is widespread (229, 230) but others suggesting MR/K is more often seen for Proteus vulgaris or Proteus penneri (100, 252, 261). MR/K fimbriae produced by Providencia stuartii and P. penneri mediate attachment to catheter surfaces, and therefore may contribute to catheter colonization by P. mirabilis strains that possess MR/K fimbriae (261, 262). They have also been shown to mediate adherence to Bowman’s capsules in kidney glomeruli (240). Some of the 17 fimbriae are more variable in sequence than others. For example, one operon, fim3, is present in less than half the isolates in a diverse collection (106). Likewise, UCA is widespread in P. mirabilis, but its size and sequence are variable, so UCA binding properties might not be constant from strain to strain (106, 248). Thus, MR/K agglutination could be mediated by one or more of these fimbriae.
Transcriptomic and high throughput sequencing methods are beginning to shed light on the complex collection of the 17 fimbriae. Transcripts from three additional fimbriae, fim5, fim8, and fim15, were detected by microarray analysis of urine from infected mice, although both fim5 and fim8 were decreased compared with in vitro culture (131). In addition, Fim8A, E, and F protein were detected by mass spectrometry from sheared surface proteins produced during in vitro culture (106). Southern blot analysis suggested P. mirabilis strains encoded two copies of mrp genes (150). Subsequent genome sequencing revealed the mrp′ operon, which is an apparent duplication of the mrp operon; the two operons are encoded next to each other in the chromosome (112). Tn-Seq analysis of genes that contribute to single-species CAUTI suggest a role for mrp′ but not mrp, which is unexpected given the well-documented contributions of MR/P fimbriae to uncomplicated UTI (129). Most of the remaining fimbriae were also detected in polymicrobial CAUTI Tn-Seq with P. stuartii but not single-species CAUTI (mrp, atf, fim3, fim5, fim12, fim14, fim15, fim16, fim17, plus orphan fimbrial genes PMI1812, PMI1920, and PMI3023), indicating once again that the contributions of these multiple fimbriae may be subtle and much work remains to discern their functions (129).
Non-fimbrial adhesins
P. mirabilis produces other, non-fimbrial adhesins that are described elsewhere in this chapter, including autotransporter proteins Pta, AipA, and TaaP (link toxins and autotransporter sections). P. mirabilis also encodes two putative type IV pili (Pearson 2008).
Transcriptional regulation by MrpJ
MrpJ is a transcriptional regulator encoded at the end of the mrp fimbrial operon (137). Like other mrp genes, it is not expressed well in vitro, but is among the most highly-induced genes in urine collected from infected mice (131). MrpJ was originally proposed to serve as a balance between adherent and motile states (reciprocal regulation); when MR/P fimbriae are produced (adherent state), MrpJ is also produced and switches off flagella (motility) (137). That is, expression of mrpJ on a plasmid represses FlaA flagellin, in turn inhibiting both swimming and swarming motility (137) (Fig. 28). Similarly, an mrpI locked-on mutant constitutively produces MR/P fimbriae but not flagella, while disruption of mrpJ in the locked-on background results in a bacterium that produces both fimbriae and flagella (137).
MrpJ is predicted to be a 110 amino acid protein consisting of a helix-turn-helix (HTH) DNA-binding domain and a unique C-terminal tail (137, 138). It is a member of the xenobiotic response element (XRE) superfamily of transcriptional regulators. Consistent with other HTH proteins, mutation of residues predicted to form DNA contacts interferes with MrpJ’s ability to repress swimming motility (138). However, deletion of the C-terminal domain has almost no effect on motility repression, suggesting that this region is disposable for reciprocal regulation (138). Consistent with its role as a repressor of motility, MrpJ directly binds the promoter of the master flagellar regulatory genes flhDC (132, 138).
In addition to influencing motility, MrpJ also appears to be involved in autoregulation of the mrp operon. Expression of mrpJ on a plasmid results in elevated mrpA expression (132), and an mrpJ mutant produces less MrpA protein (137). The mrpJ mutant also has the mrp invertible element overwhelmingly in the off orientation, even under mrp inducing conditions (152).
Deletion of mrpJ has little effect in vitro, which may be due to either low expression of mrp genes in most laboratory conditions or masking of MrpJ function by redundancy (see MrpJ paralogs below) (137). However, an mrpJ mutant is significantly outcompeted during mouse independent and cochallenge experiments (132, 137), indication a strong contribution to fitness and pathogenesis within the urinary tract. The exact role of MrpJ during infection is complicated considering that it regulates other virulence factors, such as production of MR/P fimbriae and flagella; MrpJ could be acting by altering either of these proteins, or eliminating coordination of motility and adherence, or by acting on other gene targets.
Transcriptomic analysis confirms that MrpJ has other virulence-associated targets, in addition to MR/P fimbriae and flagella (132). Because mrp genes are transcribed at a low level in vitro, an in vivo mimic has been used to study MrpJ targets, where mrpJ is expressed in trans at levels comparable to those detected in mice during experimental UTI (132). In a comparison of the in vivo mimic with a vector control, 217 genes were differentially regulated. Classes of MrpJ target genes include flagella, MR/P and other fimbriae, type VI secretion, metabolism, LPS modifications, virulence factors such as Zap and Pta, and transporters (132). Because many MrpJ-regulated genes have known or predicted roles in disease, MrpJ has been proposed as a master regulator of virulence.
Chromatin immunoprecipitation followed by PCR (ChIP-PCR) indicated that MrpJ binds within the invertible element of the mrp operon, in agreement with detection of an MrpJ-responsive element 156–256 bp upstream of the transcriptional start site of the mrp promoter (132). This is unusual, given that bacterial regulators typically bind in the vicinity of the −35 (activator) or +0 (repressor) promoter sequences so they can interact or interfere with RNA polymerase, respectively (263). However, a distant site of action for MrpJ, relative to other transcriptional regulators, is consistent with the flhDC promoter binding data mentioned above (132, 138).
MrpJ paralogs
Genomic sequencing of P. mirabilis revealed that this species encodes 14 additional mrpJ-type genes (112, 138). All but four are part of fimbrial operons, and they are always located at the beginning or the end of the operon (Table 6). Expression of all but two in trans causes repression of motility (138). Apparently, the paralogs do not all repress motility by the same mechanism, because overexpression of individual paralogs leads to flagellin levels ranging from wild-type to near-complete repression (138) (Fig. 29A). Different paralogs also induce distinctive swarming phenotypes and uniquely aberrant differentiation into swarmer cells (138) (Fig. 29B). This suggests that although most MrpJ paralogs repress motility, they likely have other, non-overlapping functions. However, at least one paralog, UcaJ, has been shown to bind the same fragment of the flhDC promoter as MrpJ (138).
Comparison of MrpJ and its paralogs revealed a conserved SQQQFSRYE motif within the helix-turn-helix (HTH) DNA-binding domain, plus a highly-variable C-terminal tail and short, highly-variable N-terminal sequence (138). Notably, MrpJ paralogs are not restricted to P. mirabilis as they have been detected in fimbrial operons from related genera (264–266). Because fimbriae allow bacteria to adhere or form biofilms in specific environments, and thus respond to niche-specific signals, MrpJ paralogs have been proposed to orchestrate genetic programs that are beneficial to those environments (132, 267).
Like MrpJ itself, the MrpJ paralogs regulate other fimbrial operons in addition to repressing motility (267). Screening of MrpJ, UcaJ, AtfJ, and Fim8J showed that some paralogs positively auto-regulate their operons. Strikingly, AtfA and Fim8A strongly induced their respective operons while having modest or no effects on other fimbrial genes (mrpA, ucaA, or pmfA); in contrast, UcaJ has no effect on the uca operon (267). Examination of the atf promoter using deletion analysis indicated that AtfJ interacts with a region 486 to 655 bp upstream of the transcriptional start site (267). Thus, other MrpJ paralogs may share a common feature of binding DNA far away from transcriptional machinery binding sites. Deletion of the AtfJ C-terminal tail had little effect on motility repression, but was required for atf autoregulation (267). Protein modeling of the AtfJ C-terminus suggests this region is involved in protein-protein interactions, and it has been proposed that MrpJ paralogs exert their unique properties via their divergent C-termini (138, 267). These potential, unidentified, binding partners may explain how MrpJ paralogs exert transcriptional control from distant DNA binding sites.
F. Biofilms
Biofilms, which are adherent microbial communities, are a notorious problem on catheter surfaces, including urinary catheters. The ability of P. mirabilis to form biofilms on catheter surfaces is well-established (5, 268). Less well-understood is the potential establishment of biofilms within the urinary tract, and to what extent these biofilms contribute to disease. Because catheterization is a major risk factor for P. mirabilis UTI, biofilms within both catheters and urinary tissue will be considered here.
P. mirabilis readily adheres to a wide variety of surfaces, including clinically-relevant materials such as silicone, latex, glass, and polystyrene. In the presence of urine, struvite and apatite minerals are deposited among the developing P. mirabilis-colonized surface, leading to a crystalline biofilm. In laboratory models, crystalline biofilms form within 6 h after inoculation with P. mirabilis (269); similarly, mineralization has been detected in experimentally-infected mice within 24 hrs (44), and stones are detectable by 4 days after inoculation (42). This process occurs as a result of urease activity (link urease section), which causes locally-increased pH and subsequent mineral precipitation. This trait is a major reason why P. mirabilis is particularly problematic in patients with indwelling urinary catheters: crystalline biofilms can completely obstruct catheters (5, 27, 270) (Fig. 30).
Random mutagenesis to find biofilm mutants indicates that, as in other bacterial species, the biofilm lifestyle for P. mirabilis is complex, heterogeneous, and involves the coordination of many genes (177, 271). For example, a non-saturating transposon screen identified altered biofilm formation in 575/3850 mutants (15%) (177). When a subset of these mutants were further examined, most were also deficient in swimming and swarming motility. Notably, swimming and swarming may be involved in both establishment of biofilms, as a means of rapidly colonizing a surface, and dispersal into new environments. P. mirabilis swarms on catheters, and differentiated swarmer cells have been reported on P. mirabilis biofilms (272, 273); thus, swarming may help distribute bacteria from the catheter to the urinary tract. However, motility may also interfere with establishing adherent communities (274), and it is likely that balance is required for fully-developed biofilm formation and swarming behavior.
P. mirabilis encodes an arsenal of adhesive proteins, several of which have been shown to contribute to biofilm formation (Table 7). In particular, MR/P fimbriae have been linked to biofilm formation using multiple methods, including crystal violet-stained multi-well plates (256, 275), confocal microscopy on glass coverslips (244), glass tubing (256), bacterial clustering within the bladder lumen (44), and aggregative adherence to cultured cells (275). MR/P fimbriae allow complex biofilm architecture to develop, but only if the bacteria are allowed to modulate their fimbrial production (link fimbrial section). If MR/P expression is genetically shut off, biofilm formation is greatly diminished. However, bacteria engineered to constitutively produce MR/P fimbriae produce relatively thin (12 μm compared with 65 μm for wild-type), albeit very dense, biofilms (244).
Table 7
Biofilm Factor | Proposed Role | Reference |
---|---|---|
Urease | Nickel metalloenzyme, local increase in pH causes mineral deposition which facilitates crystalline biofilms | (5, 30, 306) |
MR/P fimbriae | Adherence and auto-aggregation; mutants are defective in biofilm formation | (44, 177, 244, 256, 275) |
Fimbria 10 (PMI2210) | Adherence; mutant has increased biofilm (perhaps due to fimbrial cross-talk) | (177) |
UCA, PMF, and ATF fimbriae | Mutations have varying effects on biofilms (CFU, matrix) depending on culture conditions | (256) |
RcsBCD (RsbA) phosphorelay | Phosphorelay system that enhances extracellular polysaccharide production; may control biofilm formation via regulation of fimbriae | (140, 177, 279) |
Pst transporter | High-affinity phosphate transporter, mutants are defective in biofilm formation | (297) |
Capsule or extracellular polysaccharides | Facilitates mineral aggregation into crystalline biofilms and bacterial colonization | (282–284) |
LPS | P. mirabilis with different LPS charges vary in biofilm formation; pmrI LPS modification mutant and waaE inner-core LPS biosynthetic protein mutant are biofilm-defective | (53, 280, 281) |
RsmA | RNA binding protein; expression of P. mirabilis rsmA in Escherichia coli suppresses biofilm | (307) |
Ppk | Polyphosphate kinase; may act by regulating MR/P fimbriae | (308) |
Hfq | RNA chaperone; may act by regulating MR/P fimbriae or motility | (164) |
GlnE | Glutamate-ammonia ligase adenylyltransferase; mutant has increased biofilm | (177) |
Lrp | Leucine-responsive regulator | (177) |
NirB | Nitrate reductase; biofilm-deficient in crystal violet assay and takes longer to block catheters | (177) |
Bcr | Multidrug efflux pump; mutant has increased biofilm but takes longer to block catheters | (177) |
GltS | Sodium/glutamate symport carrier protein | (177) |
PMI1551 | Putative lipoprotein | (177) |
PMI1608 | Unknown; biofilm-deficient in crystal violet assay but blocks catheters more rapidly | (177) |
PMI2861 | Putative membrane protein | (177) |
PMI3402 | MuA-like DNA binding protein; mutant has increased biofilm in crystal violet assay and blocks catheters more rapidly | (177) |
Crystalline biofilm formation is thought to increase risk of bladder stones, as one study found that 62% of patients with encrusted catheters also had bladder stones (276). Notably, P. mirabilis is recovered from most (65–79%) patients with obstructed catheters (5, 276). Infection stones and crystalline biofilms are both dense, complex bacterial communities, and at least two P. mirabilis proteins contribute to both processes: urease and MR/P fimbriae. During experimental UTI in mice, P. mirabilis assembles into large clusters in the bladder lumen, where mineralization starts to occur within 24 hours. Mutation of either urease or MR/P genes results in loss of both clustering and mineralization (44). P. mirabilis bacteria become embedded within stones, and likewise, may be embedded in extracellular matrix produced during biofilm growth (42, 277). In both cases, the bacteria may be protected from immune responses (44, 278).
Other processes also contribute to P. mirabilis biofilm formation (Table 7). Logically, regulators of the above biofilm factors, such as the RcsBCD phosphorelay (RcsD is also known as RsbA) (140, 177, 279) and RNA chaperone Hfq (164), control biofilm formation. Surface hydrophobicity, mediated by lipopolysaccharide (LPS) (53, 280, 281) and capsular polysaccharides (282–284), influence both initial surface colonization and biofilm maturation.
Prevention of catheter blockage
Currently, the typical approaches to avoid urinary catheter biofilms are to 1) limit the duration of catheterization and 2) replace blocked/encrusted catheters. Multiple groups are developing new approaches to prevent biofilm establishment, including alternative catheter materials, antimicrobial coatings or instillation of anti-biofilm chemicals, mechanical or electrical biofilm dispersal, bacteriophage, and control of urinary pH (278, 285–287). Because P. mirabilis is the major agent of crystalline biofilm formation and catheter blockage during CAUTI, many of these efforts focus on this species. Thus far, no catheter type has been immune to P. mirabilis biofilm (270), although some materials and coatings delay blockage. In some cases, P. mirabilis colonizes antimicrobial surfaces after crystalline material has coated the catheter (288). Filling the retention balloon of urinary catheters with 10 g/L triclosan is able to inhibit P. mirabilis biofilm in an in vitro model (289, 290), although exposure of P. mirabilis to triclosan may rapidly select for triclosan resistance (291). Ultimately, a combination of these strategies may be necessary to successfully combat P. mirabilis crystalline biofilm formation in a clinical setting.
Methods for studying P. mirabilis biofilms
An important caveat for biofilm studies in general is that biofilms can be generated using many techniques, and the chosen method has a strong influence on the outcome of the study (292). For example, P. mirabilis mutants with altered biofilm in a high-throughput crystal violet screen had no correlation with time to blockage in a catheter/artificial bladder model (177). Thus, there is a need to use a combination of simple and/or high-throughput in vitro methods, and more complex in vitro models and in vivo methods to study the contributions of P. mirabilis biofilms to infection.
A common, simple method for surveying biofilm formation involves culturing bacteria in multi-well plates, and visualizing biofilm by staining with crystal violet. The major advantage is that this assay allows for simultaneous screening of many samples; it is also easily quantifiable by dissolving the crystal violet and measuring the dye with a spectrophotometer. For this reason, the assay has been used to screen P. mirabilis clinical isolates and transposon mutants, as well as screening of targeted mutants (53, 177, 256, 275, 280, 281, 293, 294). Other variations on this assay include the addition of short catheter segments to multiwall plates and culture in static, closed tubes (283, 295). It is, however, important to note that these methods have their limitations. While urinary catheter biofilms are subject to the flow of urine, bacteria in multi-well plates are in a closed system where nutrients are depleted. Furthermore, if urine (artificial or real) is used as the medium, P. mirabilis urease activity will cause a rapid increase in alkalinity to approximately pH9. This lethal effect in static systems can be seen in a comparison of P. mirabilis biofilms cultured in the presence or absence of urea (296). Therefore, while the crystal violet method is useful, results may be less likely to lead to translational applications.
Continuous flow systems, where fresh medium is supplied at a constant rate, allow study of P. mirabilis biofilm development in catheters (30, 273, 297). The flow can be adjusted to match physiological urine rates in humans, and a flow cell can be inserted to facilitate microscopy. Urease activity still leads to crystalline biofilm formation if urine is used as the medium, but the pH increase is not as extreme as in the multi-well plate biofilm assay and may be maintained within a physiological range. A modification of continuous flow is an artificial bladder (298). Flow is maintained into a glass chamber that serves as a bladder analog, which is maintained at 37°C via a water jacket. A catheter is inserted into the base of the bladder chamber, which may be further connected to tubing and a drainage bag. The advantage of this bladder model is that it captures several aspects of the urinary tract: continuous introduction of urine from above, insertion of a catheter from below (including inflation of the retention balloon), and retention of a residual volume of urine similar to what occurs in a catheterized patient. However, this model requires specialized equipment and is limited in how many samples can be analyzed at one time.
P. mirabilis adheres to cultured cells in an aggregative pattern that has been likened to biofilms (275). This method may be used to assess aspects of biofilm development that are host-influenced, or require a biotic substrate for the developing biofilm. A drawback is that P. mirabilis produces toxins, such as hemolysin, which lead to cell death within a few hours and pose a challenge for studying longer-term biofilm development.
Animal models are also used to study biofilm development. Stones and clusters form in the mouse ascending UTI model (40, 42, 44), and P. mirabilis biofilms may be studied in vivo using the CAUTI mouse model (link animal models section) (25, 129). These models are essential for investigating the interaction of P. mirabilis biofilms with urinary tissues, the innate immune response, and the catheter-bladder interface.
Culture medium also greatly affects biofilm development. Using continuous flow coupled with confocal scanning laser microscopy, P. mirabilis biofilms were found to produce towers of biomass separated by channels in a rich medium (LB). However, in artificial urine (299), biofilms were flat, lacked channels, and had swarmer cells protruding from the surface (273). Crystalline biofilms will only develop if urine or artificial urine is used. Catheters in patients will be coated with cellular material and other material not present in artificial urine, and these coatings may further facilitate biofilm formation (268, 270).
Polymicrobial catheter-associated biofilms
P. mirabilis is frequently part of a mixed-species infection (4, 5, 9, 300), and catheter biofilms are a useful model for studying polymicrobial UTI. Catheters taken directly from patients may be studied to explore polymicrobial biofilms (300–302), or continuous-flow systems (300, 303, 304) or static cultures (305) may be inoculated in the lab. Urease activity can give P. mirabilis a competitive advantage in catheter biofilms over Pseudomonas aeruginosa (303); similarly, in the presence of artificial urine, P. mirabilis inhibits Klebsiella pneumoniae biofilm formation (305). The CAUTI mouse model, in which a catheter segment is pushed into and retained in the bladder during the course of the experiment, is also shedding light on P. mirabilis as part of a mixed-species infection (25, 129) (link to urease and genome-wide mutagenesis sections). Other aspects of mixed-species bacteriuria and infection have been recently reviewed elsewhere (3, 20).
G. Flagella and motility
P. mirabilis was named for the Greek god Proteus due to its distinctive dimorphic nature, which is directly related to its motility: in the vegetative state, these bacteria exist as short motile rods (1–2 microns in length) with 6–10 peritrichous flagella, but under permissive conditions they differentiate into swarm cells by elongating 20- to 50-fold (20–80 microns in length) and expressing hundreds to thousands of flagella (309, 310) (Fig. 31). The vegetative swimmer cells exhibit normal chemotactic behavior in liquid medium, being attracted by nutrients and repelled by toxic substances or unfavorable conditions (310), while swarm cells allow for migration across solid surfaces, forming a characteristic bull’s eye pattern through sequential rounds of the differentiation process (95). Swarm cell differentiation and the mechanics and regulation of swimming and swarming motility in P. mirabilis have been extensively reviewed elsewhere (3, 18, 19, 311). We will therefore sharply focus on the direct contribution of flagella and motility to pathogenesis.
Flagella
In contrast to other motile bacteria, all of the flagellar components and chemotaxis proteins of P. mirabilis are encoded within a single locus in the chromosome that spans approximately 54 kb (112). Within this region, P. mirabilis encodes two flagellin genes, flaA and flaB (312). FlaA appears to be the predominant flagellin produced by P. mirabilis, but recombination can occur between flaA and flaB, resulting in occasional hybrids (124, 313, 314). As the flagella produced by P. mirabilis are recognized by the host immune response and can elicit a pro-inflammatory response, antigenic variation and flaAB hybrids may play a role in immune evasion during infection (77, 161).
Flagella are not an absolute requirement for establishment of UTI, as naturally-occurring nonmotile strains and flagellar mutants of P. mirabilis are still capable of colonizing the urinary tract in experimental models (155, 315). However, there is clear evidence for the contribution of flagella to pathogenesis. The importance of flagella to host cell invasion and ascending UTI was directly assessed by disrupting the gene encoding the flagellar filament (flaD) in a P. mirabilis mutant lacking hemolysin activity to avoid confounding by expression of the toxin. Production of the flagellar filament was found to facilitate invasion of human renal proximal tubular epithelial cells in vitro, partly by allowing the bacterial cells to come into close proximity to the host cells (45). Flagella may also promote internalization through other mechanisms, as the flaD mutant still exhibited reduced internalization compared to the parental strain when the bacteria were centrifuged onto the cell monolayer. This mutant also exhibited reduced bladder and kidney colonization in the murine model of ascending UTI, while loss of hemolysin alone did not significantly impact pathogenesis.
A role for chemotaxis and components of the flagellar apparatus in ascending UTI was also directly indicated by signature-tagged mutagenesis (STM) studies as a positive regulator of chemotaxis (cheW), the flagellar M-ring (fliF) and flagellar hook protein (flgE) were all identified as fitness factors for ascending UTI (108, 179). Transposon insertion in any of these genes resulted in a defect in kidney colonization, while fliF and cheW also affected bladder colonization. Transposon insertion-site sequencing (Tn-Seq) similarly indicated a role for chemotaxis and flagella during catheter-associated CAUTI (129). Specifically, a regulator of chemotaxis (cheR), the flagellar master regulator (flhDC), and a gene involved in regulation of flhDC (umoA) were identified as fitness factors for kidney colonization, while another gene involved in regulation of flhDC (umoB) and the flagellar basal-body rod protein (flgC) were fitness factors for colonization of the kidneys and the catheterized bladder, and the flagellar M-ring (fliF) and ATP synthase (fliI) were fitness factors for bladder colonization (129). Thus, the flagellar M-ring has been identified as having a significant contribution to fitness through two separate mutagenesis screens in two distinct models of infection. It is also noteworthy that flagellar genes are temporally regulated during ascending UTI, being poorly expressed in vivo relative to growth in rich medium at early time points post-inoculation but increasing in expression by 7 dpi, further indicating a role for flagella in pathogenicity (131).
Motility
Despite numerous studies, the exact contribution of flagella-mediated motility to P. mirabilis pathogenicity remains inconclusive. P. mirabilis has been observed to differentiate into swarm cells on catheter surfaces, which is thought to allow for migration and entry into the catheterized urinary tract (7, 272, 298, 316) (Fig. 32). As non-swarming mutants are only capable of migrating across hydrogel-coated catheters, swarming is likely an important aspect of initial bladder colonization in catheterized individuals. Five cues that can induce swarming in P. mirabilis are present in human urine (arginine, glutamine, histidine, malate, and ornithine), and may contribute to swarm cell differentiation and migration across urine-bathed catheters as they promote swarming on urine agar plates (165, 317, 318) (Fig. 33).
Once within the urinary tract, in vitro experiments indicate that swimming motility may facilitate contact with uroepithelial cells, thereby promoting internalization and cytotoxicity (45). Swarm cells in particular have been postulated to contribute to host cell invasion, as these differentiated cells may be capable of invading urothelial cells faster and to a greater extent than vegetative cells (52). Numerous virulence factors are coordinately expressed during swarming, including Zap protease and hemolysin, which may cause swarm cells to be more cytotoxic to the host urothelium (52, 133) (Fig. 34).
Swimming motility is also thought to contribute to dissemination within the urinary tract, particularly ascension from the bladder to the kidneys and spread between kidneys. For instance, immunization to stimulate production of antibodies that immobilize P. mirabilis can prevent spread to the contralateral kidney following direct inoculation into the right renal medulla (319). Swarm cells have also been implicated for contributing to kidney colonization and development of pyelonephritis, particularly during long-term infection, as swarm cells have been visualized within the kidney parenchyma (52, 320). However, elongated swarm cells were rarely observed during four-day infection studies in the (uncatheterized) murine model of ascending UTI (321). Thus, while flagella clearly contribute to P. mirabilis pathogenesis, the importance of swimming motility and swarm cell differentiation to disease progression and severity remain to be fully elucidated.
H. Metal acquisition
Bacterial pathogens compete with the host for micronutrients. One method the host uses to combat pathogens is sequestration of these nutrients so that the bacteria become starved for these ions. Bacteria that have evolved to colonize humans, therefore, have strategies to scavenge these elements, including iron and zinc (322, 323).
Iron
The most highly studied of the nutrient competition systems is iron, and much like the rest of the human body, the urinary tract is iron-restricted (324). Similar to other pathogens, P. mirabilis encodes iron scavenging and uptake systems that are induced during infection (109, 112, 131).
Bacteria typically produce iron-scavenging molecules called siderophores, which bind iron with higher affinity than host iron sequestration systems (322). Early evidence suggested that P. mirabilis lacked siderophores because it fares poorly relative to other bacteria in iron chelation experiments and tested negative in siderophore detection assays (325, 326). In addition, P. mirabilis is inhibited in its ability to establish infection in an iron-deficient rat pyelonephritis model (327). However, we now know that P. mirabilis has at least three ways to scavenge iron: proteobactin (Pbt), a non-ribosomal peptide synthesis system (Nrp), and α-keto acids. It can also use enterobactin, a siderophore produced by other enteric bacteria, although this species is not able to synthesize enterobactin on its own (109).
The nrp system was initially identified as a possible iron-regulated swarming signaling system (328). Genome annotation, microarray analysis, and a more sensitive siderophore detection method revealed that nrp and a second system, proteobactin (pbt), are siderophore biosynthesis and transport operons (109, 112). While the nrp genes are homologous to yersiniabactin genes and are encoded within a high pathogenicity island derived from Yersinia spp. (107), the products are not interchangeable (109). Although both siderophore systems must be mutated to produce a negative iron scavenging result in the laboratory (chrome azurol S, or CAS assay), only the nrp system has a significant contribution to infection (109).
P. mirabilis produces α-keto acids from amino acids, likely via two amino acid deaminases (PMI2834 aad and PMI2149) (326, 329, 330). These α-keto acids act as non-canonical siderophores and can restore growth of P. mirabilis on iron-chelated medium when supplied externally (326, 329). Amino acid deaminase activity is restricted to genera closely related to P. mirabilis (that is, Proteus, Providencia, and Morganella), as most enteric bacteria do not have Aad activity and cannot use externally-provided α-keto or α-hydroxy acids to overcome iron chelation (326, 329).
P. mirabilis encodes ferric, ferrous, and heme uptake systems that have been discovered by analysis of genes and proteins differentially regulated during iron restriction, genome annotation, and transposon mutagenesis (109, 112, 331, 332). In vivo analyses have demonstrated that iron uptake proteins are immunogenic (161, 331) and produced in a rat intraperitoneal chamber model (333); likewise, iron-related genes are induced during UTI (131), and several of these have directly been shown to contribute to UTI (108, 161, 179, 332). Iron acquisition and storage systems are even more strongly required during polymicrobial CAUTI, suggesting that competition with other species increases pressure on P. mirabilis iron-scavenging capabilities (129). Details on iron-associated genes in P. mirabilis and their contributions to infection are listed in Table 8.
Table 8
Gene Designation(s) | Proposed function | PMI number (s) | Upregulated in iron limitation | Upregulated in vivo | Implicated in vivo | Antigenic in vivo | CAUTI | Citations | |
---|---|---|---|---|---|---|---|---|---|
Single | Poly | ||||||||
Heme uptake | |||||||||
TonB-dependent receptor | PMI0409 | (336) (335) (129) | |||||||
hemin uptake protein | PMI1424 | (335) (337) | |||||||
hmuR1R2STUV | hemin uptake system | PMI1425–1430 | (336) (335) (337) (338) (331) (129) | ||||||
Ferrous Iron Uptake | |||||||||
sitDCBA | Iron ABC transporter | PMI1024–1027 | (335) (337) (129) | ||||||
feoAB | ferrus iron transport | PMI2920–2921 | (337) (129) | ||||||
Ferric citrate transport | |||||||||
exported protease | PMI3704 | (335) (129) | |||||||
iron-related ABC transporter | PMI3705 | (335) (339) (129) | |||||||
iron receptor | PMI3706–3707 | (335) (129) | |||||||
extracytoplasmic function family σ factor | PMI3708 | (335) (129) | |||||||
TonB-like protein | PMI3709 | (335) (129) | |||||||
Siderophore production | |||||||||
TonB-dependent receptor | PMI2596 | (336) (335) (333) (129) | |||||||
nrpXYRSUTABG | nrp siderophore | PMI2597–2605 | (335) (337) (340) (129) | ||||||
pbtIABCDEFGH | proteobact in | PMI0231–0239 | (339) (129) | ||||||
Other TonB-dependent receptors | |||||||||
irgA | TonB-dependent enterobact in receptor | PMI0842 | (336) (335) (337) (333) (129) | ||||||
ireA | ferric siderophore receptor | PMI1945 | (336) (335) (337) (333) (129) | ||||||
TonB-dependent receptor | PMI0363 | (335) (337) (129) | |||||||
TonB-dependent receptor | PMI1548–1551 | (335) (337) (129) | |||||||
hasR | TonB-dependent receptor | PMI3120–3121 | (340) (129) | ||||||
ABC-transport system | |||||||||
iron-related ABC transporter | PMI0331 | (335) (337) (129) | |||||||
iron-related ABC transporter | PMI2957–2960 | (335) (337) (340) (129) | |||||||
Iron-related | PMI0229 | (335) (129) | |||||||
ABC transporter | −0230 | ||||||||
Iron Metabolism | |||||||||
iron utilization protein | PMI1437 | (335) (337) (129) | |||||||
Iron Sulfur Cluster Formation/Uptake | |||||||||
iron sulfur cluster | PMI1411–1416 | (335) (337) (129) | |||||||
iron sulfur cluster | PMI3253 | (335) (337) (129) | |||||||
iron sulfur cluster | PMI0176–0172 | (335) (337) (129) |
Zinc
The high-affinity zinc importer genes znuACB are induced during UTI compared with laboratory culture (131), and ZnuB is recognized by the humoral immune response (161). Furthermore, a znuC mutant is outcompeted in mouse cochallenge experiments (162), and znuACB contributes to fitness in a CAUTI model (129). However, these systems are not an absolute requirement for infection, as the znuC mutant remains able to colonize during independent infection (162), and the requirement for znuACB in single-species CAUTI disappears during polymicrobial CAUTI (129). Zinc export may also contribute to infection, as exporter ppaA contributes to single-species CAUTI (129).
At least two virulence factors are linked to zinc: flagella and ZapA. ZapA is a member of the serralysin metalloprotease family of zinc metalloproteases, which have a conserved zinc-binding motif (219, 341). While this clearly suggests a zinc requirement for ZapA, it has not been directly tested. Work in E. coli showed that zinc is bound by the flagellar master regulator FlhC (342), so it is not surprising that a znuC mutant exhibits defects in flagella-mediated motility (both swimming and swarming) (162). Swarming is also inhibited in the presence of the transitional metal chelator TPEN, further supporting a role for zinc in motility (162). Interestingly, a zinc exporter ppaA mutant also exhibits aberrant swarming (343).
Nickel
Nickel is an essential component of catalytically active urease, in addition to other bacterial enzymes. However, there are few studies where P. mirabilis nickel homeostasis has been examined. Likewise, nickel sequestration by the host as a pathogen defense strategy has been proposed, but not yet conclusively demonstrated (344). Yet, there are some indications that nickel balance contributes to P. mirabilis fitness in the urinary tract. Genes from two predicted nickel import systems are induced in experimentally-infected mice compared with laboratory culture: nikAB and yntABCD (131). Curiously, in a CAUTI model, these import systems were not found to be essential, but nickel export (PMI1518) was (129). Thus, nickel homeostasis appears to be important during infection, but the balance between enzymatic requirements and metal toxicity may be affected by the presence of a catheter and/or competing nutritional requirements of other bacteria.
Phosphate
Phosphorus, another essential element for life, is most often incorporated as inorganic phosphate. Phosphate sensing and uptake has been linked to virulence in many bacterial species (345), and P. mirabilis is no exception. Several studies have pointed to the importance of the pstSCAB phosphate uptake system to P. mirabilis virulence. Transposon mutations in pstS and pstA were both identified in a signature-tagged mutagenesis study (108), and complemented in the mouse model (346). All four genes in this operon were also required in a CAUTI model (129). One way the pst system contributes to virulence may be via regulation of biofilm formation (297). Interestingly, phosphate mutants outcompete wild-type bacteria during culture in human urine (346) and also during polymicrobial UTI (129), suggesting that the energy cost of operating this system may actually confer a fitness defect if the bacteria are growing under conditions in which phosphate sensing and uptake are not required.
I. Metabolism in urine
Unlike most E. coli strains, P. mirabilis isolates generally possess two important metabolic features that make them ideally suited to growth in human urine: the ability to utilize citrate as a sole carbon source, and the ability to hydrolyze urea to produce an abundant nitrogen source. Urea is the most abundant organic metabolite in human urine, at concentrations of ~400 mM (347–349). Filtration of citrate occurs in the kidney glomeruli, with approximately 10–35% of filtered citrate excreted in urine, making it the most abundant organic anion in human urine (350, 351). Thus, the combined use of citrate as a carbon source and ammonia as a nitrogen source provides P. mirabilis with a significant metabolic advantage over other uropathogens, and has implications for the carbon/nitrogen ratio that is sensed by P. mirabilis during growth in urine and the metabolic pathways favored by this organism.
Ascending UTI
Much has been learned regarding the metabolic pathways favored by P. mirabilis in the urinary tract through a combination of transcriptomics and genome-wide mutagenesis studies (Fig. 35). P. mirabilis appears to utilize glucose uptake and glycolysis during ascending UTI, as genes in these pathways are upregulated in vivo (131). The importance of glycolysis to P. mirabilis fitness was further verified in the murine model of ascending UTI using mutants in glucose-6-phopsphate isomerase (pgi), 6-phosphofructokinase transferase (pfkA), triosephosphate isomerase (tpiA), and pyruvate kinase (pykA) (166). Disruption of any single gene resulted in a dramatic fitness defect in bladder colonization, and all but pykA caused a fitness defect in kidney colonization.
Both the oxidative and non-oxidative branches of the pentose phosphate pathway for NADPH production have also been shown to contribute to P. mirabilis fitness within the urinary tract. Transaldolase (talB) catalyzes a reversible step in the non-oxidative branch of the pentose phosphate pathway. While this gene was not identified as differentially regulated by in vivo transcriptomics and has not been detected as a fitness factor by signature-tagged mutagenesis (STM), it has been shown experimentally to contribute to P. mirabilis fitness in the bladder and kidneys during ascending UTI (166). Similarly, 6-phosphogluconate dehydrogenase (gnd) catalyzes a reaction in the oxidative branch of pentose phosphate pathway, and this gene has been shown to contribute to P. mirabilis fitness within the bladder and kidneys during ascending UTI (166). Notably, the D-glyceraldehyde 3-phosphate generated by the combined branches of the pentose phosphate pathway can feed directly into glycolysis.
P. mirabilis also appears to utilize the Entner-Doudoroff pathway during ascending UTI, as phosphogluconate dehydrogenase (edd) was upregulated during ascending UTI, identified as a fitness factor by STM, and disruption of this gene results in a fitness defect in bladder and kidney colonization (131, 166, 179). The other member of the Entner-Doudoroff pathway, a bifunctional 4-hydroxy-2-oxoglutarate aldolase/2-dehydro-3-deoxy-phosphogluconate aldolase encoded by eda, was also upregulated during ascending UTI, underscoring the importance of this metabolic pathway to P. mirabilis fitness (131). Products of the Entner-Doudoroff pathway are D-glyceraldehyde 3-phosphate, which can feed into glycolysis, and pyruvate, which can be directly catabolized to generate acetyl-CoA.
Regarding the aerobic tricarboxylic acid (TCA) cycle, subunit C of succinate dehydrogenase (sdhC) was identified as a fitness factor by STM, and subunit B (sdhB) was shown to directly contribute to fitness during kidney colonization (166, 179). Fumarate dehydratase (fumC) also contributes to P. mirabilis fitness in the bladder and kidneys (166). These results indicate a role for the aerobic TCA cycle during infection, which is further supported by the finding that cytochrome bo3 oxidase (cyoABCD), a component of the aerobic respiratory chain, is upregulated during ascending UTI (131). However, fumarate reductase (frdA), which catalyzes the reduction of fumarate to succinate during anaerobic respiration, also contributes to P. mirabilis fitness in the bladder and kidneys, which may indicate use of a branched TCA cycle during infection (166). Other portions of the TCA cycle found to be upregulated during ascending UTI include citrate synthase (gltA), citrate hydro-lyase 2 (acnB), and isocitrate dehydrogenase (icd) (131).
Additionally, transcriptomics and STM data indicate that P. mirabilis utilizes pyruvate catabolism via pyruvate dehydrogenase and dihydrolipoamide acetyltransferase (encoded by poxB and aceEF) to generate acetyl-CoA, which would feed into the TCA cycle through the action of citrate synthase (gltA) (131, 179). An abundance of factors pertaining to amino acid transport and metabolism are also upregulated during ascending UTI, including D-serine dehydratase (dsdA). D-serine is present in a relatively high concentration in human urine (352, 353). Thus, P. mirabilis may utilize DsdA as an additional pathway for generating pyruvate and ammonia.
The transcriptome of P. mirabilis in the murine urinary tract also revealed a preferential role for glutamate dehydrogenase (gdhA) over glutamine synthetase (glnA) for nitrogen assimilation during ascending infection, which was recapitulated in vitro only when citrate was supplied as the sole carbon source (131). It was further determined that gdhA provides P. mirabilis with a competitive advantage for bladder and kidney colonization (131). Glutamate dehydrogenase is generally favored by bacteria for nitrogen assimilation when energy and carbon are limited but there is an excess of ammonia and phosphate (354). Thus, the combined pattern of differential gene regulation coupled with mutagenesis studies indicates that P. mirabilis primarily utilizes pathways that feed into glycolysis, pyruvate catabolism, and citrate metabolism to fuel the TCA cycle through to production of α-ketoglutarate, which intersects with the production of L-glutamate by glutamate dehydrogenase using the excess of ammonia produced by urease (131, 179).
Catheter-Associated UTI (CAUTI)
The recent use of a genome-saturating transposon library and transposon insertion-site sequencing (Tn-Seq) has provided insight into the metabolic pathways favored by P. mirabilis in the catheterized urinary tract compared to during ascending UTI (129). A list of fitness factors in common between CAUTI and other UTI studies is found in Table 9. However, this study indicated that the pro-inflammatory environment induced by the presence of a urinary catheter dramatically impacts P. mirabilis metabolism. For instance, while glycolysis, the pentose phosphate pathway, and the Entner-Doudoroff pathway contributed to P. mirabilis fitness in ascending UTI, the only gene pertaining to these pathways that was a fitness factor in the CAUTI model was pyruvate kinase (pykF), which catalyzes the transfer of a phosphate group from phosphoenolpyruvate to yield ATP and pyruvate. For the TCA cycle enzymes, only subunit A of succinate dehydrogenase (sdhA) and malate dehydrogenase (mdh) represented fitness factors for CAUTI. However, cytochrome bo3 quinol oxidase (cyoABCDE) remained an important fitness factor for P. mirabilis in the CAUTI model, indicating a continued need for aerobic respiration.
Table 9
Pathway | PMI Number | Gene | Function | UTI Transcriptome | UTI STM | CAUTI Tn-Seq |
---|---|---|---|---|---|---|
Amino acid transport and metabolism | PMI0145 | hypothetical protein | up | B | ||
PMI0187 | dsdA | D-serine ammonia-lyase | up | B, K | ||
PMI0347 | lysA | diaminopimelate decarboxylase | up | B, K | ||
PMI0682 | artP | L-arginine ABC transporter ATP-binding protein | down | B | ||
PMI0841 | lysP | lysine:proton symporter, AAT family | up | K | ||
PMI1607 | sdaA | L-serine ammonia-lyase | U | K | ||
PMI1827 | cysK | cysteine synthase | up | K | ||
PMI2093 | speB | agmatinase | down | K | ||
PMI2094 | speA | arginine decarboxylase | down | B, K | ||
PMI2259 | metN | D-methionine ABC transporter, ATP-binding protein | up | U, K | B, K | |
PMI2736 | potB | spermidine/putrescine ABC transporter, permease protein | up | K | ||
PMI2822 | pabA | aminodeoxychorismate synthase, glutamine amidotransferase subunit | up | B | ||
PMI2882 | glnA | L-glutamine synthetase | down | B, K | ||
PMI3052 | asnA | aspartate-ammonia ligase | up | B, K | ||
Carbohydrate transport and metabolism | PMI0453 | nagA | N-acetylglucosamine 6-phosphate deacetylase | up | K | |
PMI0454 | nagB | glucosamine-6-phosphate deaminase | up | B, K | ||
PMI0873 | nagZ | beta-N-acetylhexosaminidase | up | B | ||
PMI1046 | polysaccharide deacetylase | U, B, K | B, K | |||
PMI1776 | PTS system IIB component, L-Asc family | up | B | |||
PMI1828 | hpr/ptsH | PTS system phosphocarrier protein | up | B, K | ||
PMI1829 | ptsI | Phosphoenolpyruvate–protein phosphotransferase | up | B, K | ||
PMI1830 | crr | PTS system D-glucose-specific IIA component, Glc family | up | B, K | ||
PMI1943 | MFS-family transporter | down | B, K | |||
PMI2134 | agaA | N-acetylgalactosamine 6-phosphate deacetylase | up | B | ||
PMI2141 | agaR | transcriptional regulator, DeoR family | up | B, K | ||
PMI2414 | deoB | phosphopentomutase | up | B, K | ||
PMI3095 | N-acetylglucosamine kinase | up | K | |||
PMI3716 | gpmB | probable phosphoglycerate mutase | up | K | ||
Coenzyme transport and metabolism | PMI0105 | cyoE | protoheme IX farnesyltransferase | up | B, K | |
PMI2022 | visC | 2-octaprenyl-3-methyl-6-methoxy-1,4-benzoquinol hydroxylase | up | K | ||
PMI2353 | ribB | 3,4-dihydroxy-2-butanone 4-phosphate synthase | down | K | ||
PMI3547 | fre | NAD(P)H-flavin reductase | up | B, K | ||
Energy production and conversion | PMI0107 | cyoB | cytochrome bo3 quinol oxidase subunit 1 apoprotein | up | B, K | |
PMI0108 | cyoA | cytochrome bo3 quinol oxidase subunit 2 | up | B, K | ||
PMI0354 | nqrC | Na+-translocating NADH-quinone reductase subunit C | up | B, K | ||
PMI0355 | nqrD | Na+-translocating NADH-quinone reductase subunit D | up | B, K | ||
PMI0705 | pflB | formate acetyltransferase | up | K | ||
PMI1200 | pntA | NAD(P) transhydrogenase subunit alpha (pyridine nucleotide transhydrogenase subunit alpha) | up | B | ||
PMI1753 | nuoK | NADH dehydrogenase subunit K | up | B | ||
PMI1754 | nuoJ | NADH dehydrogenase subunit J | up | K | ||
PMI1760 | nuoC/nuoD | NADH dehydrogenase subunit C (EC 1.6.5.3)/NADH dehydrogenase subunit D | up | B, K | ||
PMI1771 | ackA | acetate kinase | up | K | ||
PMI1772 | pta | phosphotransacetylase | up | K | ||
PMI2045 | aceF | dihydrolipoyllysine-residue acetyltransferase component of pyruvate dehydrogenase complex | up | B, K | ||
PMI2046 | aceE | pyruvate dehydrogenase E1 component | up | B, K | B, K | |
PMI2740 | qor | quinone oxidoreductase | up | K | ||
PMI2930 | glpD/glyD | homodimeric glycerol 3-phosphate dehydrogenase (quinone) | down | B, K | ||
PMI3108 | fdoI | formate dehydrogenase (quinone-dependent) cytochrome b subunit | up | K | ||
PMI3300 | ilvM | acetolactate synthase, small subunit | up | K |
The CAUTI Tn-Seq results predominantly support a role for energy production through the catabolism of peptides and amino acids, and again implicate a role for D-serine dehydratase (dsdA) (129). Pyruvate catabolism remained important for both infection models, as pyruvate dehydrogenase (aceEF) represented an important fitness factor during CAUTI. Several genes involved in chorismate biosynthesis were also identified as CAUTI fitness factors, including aroF, aroK, and aroL. Chorismate is a branch point for biosynthesis of tryptophan, phenylalanine, and tyrosine. None of the tryptophan biosynthesis genes after the step catalyzed by ubiC (trpEDCBA) were identified as fitness factors for CAUTI, but pheA was a candidate fitness factor for bladder colonization and tyrB for both bladder and kidney colonization, indicating the potential importance of phenylalanine and tyrosine during CAUTI. It is also worth noting that pyruvate can be generated from chorismate in two ways: 1) ubiC, a fitness factor for bladder and kidney colonization, produces pyruvate from chorismate during synthesis of the ubiquinone precursor 4-hydroxybenzoate, and 2) pyruvate is generated from chorismate and glutamine during synthesis of 4-aminobenzoate by pabB, a fitness factors for bladder colonization. Thus, interplay between chorismate biosynthesis and pyruvate catabolism may be important for P. mirabilis fitness during CAUTI.
Most striking, however, was the shift in nitrogen assimilation requirements between ascending UTI and CAUTI. In contrast to the ascending UTI studies, glutamate dehydrogenase (gdhA) was not identified as a fitness factor for CAUTI, while glutamine synthetase (glnA) was identified as a fitness factor and verified to significantly contribute to urine and kidney colonization in the CAUTI model (129). In E. coli, glutamate dehydrogenase is favored when energy is limited, and glutamine synthetase is favored when energy limitation is no longer an issue (354). Thus, the preference for glutamine synthetase in the CAUTI model may reflect an excess of carbon sources in the catheterized urinary tract.
Polymicrobial Infection
P. mirabilis is one of the most common organisms present during polymicrobial urine colonization and CAUTI (4, 8, 9, 17). Considering the unique features of P. mirabilis metabolism in urine, it is likely that the presence of this bacterium could have a dramatic impact on the metabolic pathways favored by other uropathogens during infection, and possibly vice versa. For instance, in the ascending model of UTI, P. mirabilis and E. coli appear to require different but complementary central metabolism pathways as P. mirabilis utilizes glycolysis, while E. coli favors gluconeogenesis (166). P. mirabilis also appears to utilize the pentose-phosphate pathway, Entner-Doudoroff, and the aerobic TCA cycle during ascending UTI as mutations affecting any of these pathways resulted in a fitness defect (166). Interestingly, coinfection of P. mirabilis with E. coli shifted the metabolic requirements for both species by alleviating the pentose-phosphate pathway requirement for P. mirabilis but making this pathway required for E. coli fitness during the coinfection.
Considering that a urease-negative isolate of E. coli was used for these experiments, the shift in metabolic requirements could be due in part to the impact of P. mirabilis urease on the carbon/nitrogen ratio available to E. coli. However, a genome-wide exploration of P. mirabilis fitness factors conducted during coinfection with Providencia stuartii, a urease-positive organism, similarly identified a dramatic impact on metabolic pathways in a murine model of CAUTI (129). Similar to coinfection with E. coli, coinfection with P. stuartii shifted the requirements for the pentose-phosphate pathway and also the Entner-Doudoroff pathway. Coinfection also resulted in a greater need for the import and synthesis of arginine and branched-chain amino acids. The requirement for branch-chain amino acid biosynthesis was determined to be the result of high-affinity import by P. stuartii, as mutation of a gene involved in branched-chain amino acid import in P. stuartii alleviated the requirement for synthesis by P. mirabilis. Regarding energy production and respiration, coinfection appeared to alleviate the requirement for cytochrome bo3, but imposed a requirement for the respiratory nitrate reductase system (narGHJI, narXL, narP,and narK) and an Rnf redox-driven ion pump (rnfABCDGE).
V. Vaccines
Infections associated with urolithiasis (stone formation) can be difficult to treat. Indeed, prevention of infection by vaccination may be preferable in high-risk groups such as those who are chronically catheterized in the urinary tract. To that end, there have been a number of relevant studies in murine models that suggest that vaccination is feasible, an example of which is given in Fig. 36.
Multiple approaches have been tested that include formalin-killed whole bacteria (237) and numerous surface-exposed or secreted proteins (55, 236–238, 355–359). In an early study (237), CBA/J mice were vaccinated with formalin-killed P. mirabilis bacteria or purified MR/P fimbriae or MrpH, the adhesin of the fimbria. For systemic immunizations, complete Freund’s adjuvant was used as adjuvant; for intranasal administration, cholera toxin was used as adjuvant. Four routes of immunization were examined: subcutaneous, intranasal, transurethral, and oral. Antibodies to the antigens were measured in serum, urine, bladder, vaginal washes, and bile. For whole-killed bacteria, vaccination by the subcutaneous route resulted in protection for 60% of mice (defined as having fewer than 100 CFU per gram of tissue 7 days post-inoculation), and intranasal vaccination provided protection in 67% of mice, but vaccination by the oral or transurethral routes failed to provide protection. Protection correlated with high serum antibody titers at day 21. For the intranasal route, measurable antibody was also found in the urine and bladder, albeit at a much reduced titer. When purified MR/P fimbriae were used, 60–63% protection, also at 2.5–3-log reduction, was noted by the intranasal or transurethral route. Protection also correlated with high serum antibody titer to the purified MR/P fimbriae. The tip adhesin was also examined as an antigen. Maltose-binding protein (MBP) fusions with MrpH (residues 23–157) and a truncated MrpH (residues 23–157) representing the receptor-binding domain were also used for intranasal immunization and yielded significant protection, especially for the MrpH receptor-binding domain, which provided up to 5-logs of protection in the bladder.
In a follow-up study, the MrpH receptor binding domain (residues 23–157) was translationally fused with the cholera toxin A2 domain, which was co-expressed with the cholera toxin B subunit (a holotoxin-like chimera) (238). Mice immunized with MrpH receptor-binding domain-CT complex were colonized by significantly fewer bacteria in the bladder (median log10 CFU per gram of tissue, 6.43 for the naive mice versus 2.00 for the immunized mice; P = 0.02) and the kidneys (median log10 CFU per gram of tissue, 5.28 for the naive mice versus 3.43 for the immunized mice; P= 0.046).
As well, two additional studies demonstrated that vaccination, either intranasally or transurethrally, with fusion proteins that contain both MrpH of P. mirabilis and FimH, the adhesin of type 1 fimbriae of E. coli, conferred protection against P. mirabilis UTI (359, 360).
In another fimbrial vaccination study (236), the main structural subunits (MrpA, UcaA, and PmfA) of three P. mirabilis fimbriae (MR/P, UCA, and PMF) were administered as vaccine antigens by the subcutaneous route. Mice were challenged with P. mirabilis either transurethrally or intravenously. MrpA protected against UTI and hematogenous challenge. UcaA protected only against hematogenous challenge. PmfA did not protect by either route of P. mirabilis challenge. In studies where flagellin was included with MrpA as antigens in intranasal vaccination, neither flagellin alone nor flagellin with MrpA led to protection from P. mirabilis challenge in the urinary tract (358). Administration of adjuvant cholera toxin in the urinary tract enhanced humoral and cytokine response but did not influence the degree of protection against UTI provided by MrpA (356).
MrpA expressed in heterologous hosts has also been tested as a vaccine antigen. MrpA was expressed as either a cell wall-anchored or secreted protein in Lactococcus lactis (355). Protection assays against challenge by P. mirabilis were tested in the mouse model, and significant induction of specific serum IgG and IgA was found in mice immunized with L. lactis expressing the MrpA antigen. A significant reduction of renal bacterial colonization was observed using both constructs. MrpA was also expressed as a MrpA-TetC (tetanus toxin fragment) in Salmonella Typhimurium (357) and used for intranasal vaccination. After two immunization doses, intranasally vaccinated mice showed a significant increase in specific serum IgG against MrpA and against the LPS of the Salmonella strain. Significant decreases in both kidney and bladder colonization by P. mirabilis after transurethral challenge were noted.
A non-fimbrial secreted protein, Proteus toxic agglutinin, proved to be a potent protective antigen for the upper urinary tract (55). Mice intranasally vaccinated with a site-directed mutant (S366A) of Proteus toxic agglutinin, which inactivated the protease active site, conjugated to cholera toxin, had significantly lower bacterial counts in their kidneys (P=.001) and spleens (P=.002) than mice that received cholera toxin alone. Bladders were not protected. Serum IgG levels correlated with protection.
Lastly, it is interesting to note that a vaccine used in Europe, Solco-Urovac, contains heat-killed enteric bacteria including P. mirabilis (361, 362). The vaccine has been shown to have demonstrable, though limited, efficacy against recurrent UTI (363). Further research into P. mirabilis vaccines is necessary, but heat-killed bacteria and surface-exposed or secreted proteins may have efficacy as vaccine antigens, and an intranasal route may be effective for vaccination. One issue is the identification of the target population for vaccination again P. mirabilis. For example, the population that might benefit most from vaccination against this species is elderly catheterized patients in chronic care facilities, but they may have diminished capacity to respond immunologically to vaccination. This poses an additional hurdle for development of an effective vaccine.
VI. Summary/Conclusions
When surveying the extensive literature on P. mirabilis infection and considering the relatively few labs that have undertaken studying this pathogen, there has been remarkable progress made in identifying virulence factors and mechanisms of pathogenesis for this uropathogen. Clearly, the catheterized urinary tract is a preferred niche for P. mirabilis and often the bacterium coexists with several other members of a polymicrobial community. However, monoinfection has been used for the most part to identify and characterize the virulence arsenal that includes urease, 17 different fimbriae, secreted cytolethal and cytolytic toxins and proteases, extensive networks for iron and other metal ion acquisition, flagella, and numerous secretion systems. We are just beginning to exploit relatively new tools and screens such as Tn-Seq and RNA-Seq to investigate mechanisms of pathogenesis in models of polymicrobial bacteriuria. This added layer of complexity is daunting, but ultimately manageable and will open the door to the development of treatment strategies.
In the early days, investigators worked with simple molecular tools to identify single genes or operons involved in pathogenesis. Frankly, victories were hard fought although targets of discovery were plentiful. With the advent of facile sequencing technologies, PCR, microarrays, fruitful screens including signature-tagged mutagenesis and transposon-directed insertion site sequencing and RNA-Seq, numerous new candidate genes that contribute to colonization, establishment of infection, and damage to the host have been highlighted for P. mirabilis.
While often tedious, such screens are straightforward and yield a staggering number of genes that represent virulence or fitness factors. What lies ahead is the hard work of probing the biology of these new-found genes. Thus, with the groundwork presented here, we can renew our efforts to refine a view of pathogenesis that includes the interactions and temporal expression of the battery of virulence factors and metabolic cycles discussed in this review.
Acknowledgments
This work was supported by Public Health Service grants from the National Institutes of Health (K99 DK105205 to C.E.A.; AI059722 and DK094777 to H.L.T.M.).
We are grateful to the many Proteus mirabilis researchers whose work contributed to this review.
References
Full text links
Read article at publisher's site: https://doi.org/10.1128/ecosalplus.esp-0009-2017
Read article for free, from open access legal sources, via Unpaywall: https://www.asmscience.org/deliver/fulltext/ecosalplus/8/1/ESP-0009-2017.pdf?itemId=/content/journal/ecosalplus/10.1128/ecosalplus.ESP-0009-2017&mimeType=pdf&isFastTrackArticle=
Citations & impact
Impact metrics
Article citations
Emergence and genomic chion of Proteus mirabilis harboring blaNDM-1 in Korean companion dogs.
Vet Res, 55(1):50, 09 Apr 2024
Cited by: 0 articles | PMID: 38594755 | PMCID: PMC11005143
Comprehensive Review of Urinary Tract Infections in Renal Transplant Recipients: Clinical Insights and Management Strategies.
Cureus, 16(2):e53882, 08 Feb 2024
Cited by: 0 articles | PMID: 38465031 | PMCID: PMC10924982
Review Free full text in Europe PMC
Multidrug-Resistant Proteus mirabilis Infections and Clinical Outcome at Tertiary Hospital in Riyadh, Saudi Arabia.
Infect Drug Resist, 17:571-581, 14 Feb 2024
Cited by: 0 articles | PMID: 38375102 | PMCID: PMC10875173
Determining the cytotoxicity of the Minimum Inhibitory Concentration (MIC) of silver and zinc oxide nanoparticles in ESBL and carbapenemase producing Proteus mirabilis isolated from clinical samples in Shiraz, Southwest Iran.
BMC Res Notes, 17(1):40, 29 Jan 2024
Cited by: 0 articles | PMID: 38287416 | PMCID: PMC10826267
Differential Contribution of Hydrogen Metabolism to Proteus mirabilis Fitness during Single-Species and Polymicrobial Catheterized Urinary Tract Infection.
Pathogens, 12(12):1377, 22 Nov 2023
Cited by: 0 articles | PMID: 38133262 | PMCID: PMC10745698
Go to all (121) article citations
Other citations
Lay summaries
Plain language description
Similar Articles
To arrive at the top five similar articles we use a word-weighted algorithm to compare words from the Title and Abstract of each citation.
MrpJ Directly Regulates Proteus mirabilis Virulence Factors, Including Fimbriae and Type VI Secretion, during Urinary Tract Infection.
Infect Immun, 86(10):e00388-18, 21 Sep 2018
Cited by: 4 articles | PMID: 30082479 | PMCID: PMC6204732
Proteus mirabilis and Urinary Tract Infections.
Microbiol Spectr, 3(5), 01 Oct 2015
Cited by: 129 articles | PMID: 26542036 | PMCID: PMC4638163
Review Free full text in Europe PMC
Requirement of MrpH for mannose-resistant Proteus-like fimbria-mediated hemagglutination by Proteus mirabilis.
Infect Immun, 67(6):2822-2833, 01 Jun 1999
Cited by: 34 articles | PMID: 10338487 | PMCID: PMC96588
Proteus mirabilis Overview.
Methods Mol Biol, 2021:1-4, 01 Jan 2019
Cited by: 5 articles | PMID: 31309490
Merging mythology and morphology: the multifaceted lifestyle of Proteus mirabilis.
Nat Rev Microbiol, 10(11):743-754, 08 Oct 2012
Cited by: 126 articles | PMID: 23042564 | PMCID: PMC3621030
Review Free full text in Europe PMC
Funding
Funders who supported this work.
NIAID NIH HHS (1)
Grant ID: R01 AI059722
NIDDK NIH HHS (2)
Grant ID: K99 DK105205
Grant ID: R01 DK094777