Correlation of pathogenic factors with antimicrobial resistance of clinical Proteus mirabilis strains

Proteus mirabilis is the third most common etiological factor of the urinary tract infection (UTI). It produces urease, which contributes to the formation of crystalline biofilm, considered to be one of the most important virulence factors of P. mirabilis strains, along with their ability to swarm on a solid surface. The aim of this study was to analyze the pathogenic properties of two selected groups of clinical P. mirabilis isolates, antimicrobial-susceptible and multidrug-resistant (MDR), collected from hospitals in different regions in Poland. The strains were examined based on virulence gene profiles, urease and hemolysin production, biofilm formation, and swarming properties. Additionally, the strains were differentiated based on the Dienes test and antibiotic susceptibility patterns. It turned out that the MDR strains exhibited kinship more often than susceptible ones. The strains which were able to form stronger biofilm had broader antimicrobial resistance profiles. It was also found that the strongest swarming motility correlated with susceptibility to most antibiotics. The correlations described in this work encourage further investigation of the mechanisms of pathogenicity of P. mirabilis. Author summary Proteus mirabilis is widely widespread in environment but also it is responsible for most Proteus infections, especially in human urinary tracts. They cause complicated, persistent infections especially due to the ability to form urinary stones. The clinical importance of P. mirabilis have been described in the literature many times. However, the role of pathogenic features with correlation to drug resistance require further investigation. In this research we analyzed thee virulence factors in relation to drug resistance of clinical P. mirabilis strains isolated from urine. The virulence genes, ureolytic and hemolytic activity, biofilm formation, swarming growth and strains kindship were analyzed. The most important observation was that the strains exhibited a stronger territorialism were kindred to a lower number of other strains, formed weaker biofilm and exhibited a lower resistance to antibiotics. Furthermore, we proved that the strains which were more likely to mutual growth, they were also less similar in the drug resistance profile but exhibited a higher resistance to antibiotics, which can be beneficial for different bacteria living together. We believe that P. mirabilis with strong territorialism can represent a wild group of strains with poor experience of antibiotic pressure. The environmental influence (toxins, antibiotics, bacterial neighbors) stimulates the development of a less dispersed community with stronger biofilm, exchange of genes and increase of resistance to antibiotics.


Introduction
lowest number of kinship was expressed by four strains that exhibited confluent growth with only three other strains, and the highest number was observed in case of one strain, which was kindred with 31 others.   This correlation was more clearly noticed in the analysis of the two groups of susceptible and MDR 145 isolates - (Fig 3). It turned out that kinship with a higher number of strains was observed among MDR strains 146 (p=0.0148). The kinship also correlated with the swarming motility of the P. mirabilis strains. The strains with 147 weaker swarming growth were kindred with a higher number of strains compared to those with stronger swarming 148 growth (Fig 4).

149
The MDR and susceptible groups of strains were compared based on expansiveness of swarming and 150 biofilm strength (Fig 5). Eighty-four percent of susceptible strains exhibited the highest swarming motility rate 151 and these did not include any isolate of the weakest swarming growth. In contrast, 68% of MDR strains exhibited 152 the weakest swarming and only 8% of them showed the strongest motility rate. Additionally, 56% of susceptible 153 strains formed the weakest biofilm, whereas 50% of MDR strains formed the strongest biofilm. Otherwise, the 154 strongest biofilm was formed by only 20% of susceptible strains and the weakest biofilm was found in only 22% 155 of the MDR group. Similar observations were made in the case of the biofilm formation on the polyurethane: the 156 susceptible isolates revealed significantly weaker biofilm when compared to the MDR strains (unpaired Mann-Differentiation of the isolates confirmed the correlations described above but here we proved their interdependence. Cluster 1 represents the 164 strains kindred with fewer strains compared to cluster 2 (two tailed T-test, unpaired, p=0.017), i.e. the cluster 2 165 strains exhibited kinship to each other more often than those in cluster 1. Moreover, cluster 1 included strains 166 forming weaker biofilm, both on the polyurethane (two tailed T-test, unpaired, p=0.055) and on the glass (two 167 tailed T-test, unpaired, p=0.015), and exhibiting statistically significant more expansive swarming motility (two 168 tailed T-test, unpaired, p<0.001) compared to cluster 2. There was no difference either in the level or rate of urease 169 and hemolysin production between the two clusters.

172
The study was focused primarily on pathogenicity factors of P. mirabilis isolates, recovered from hospital

178
Owing to specific virulence factors, P. mirabilis is particularly troublesome for catheterized patients and is responsible for complicated UTI with the development of urinary stones [24][25][26]. In our study we analyzed the presence of virulence factor genes, starting with fliL, representative for the class II flagellar operon (fliLMNOPQR)

181
[27], which is involved in the swarming motility. Other genes included those determining the mannose-resistant crucial role in the catheter-associated biofilm formation, and the bladder and kidney colonization, respectively 184 [10,26,28]. The remaining tested genes zapA, hpmA, hpmB and ureC are engaged in the immune system evasion 185 and/or iron acquisition [24], with zapA also involved in the swarmer cell differentiation and swarming behavior 188 virulence genes. The P. mirabilis chromosome is strongly conservative and the location of the analyzed genes 189 seems to be more stable [1] compared to E. coli [31].

190
Despite the common presence of fliL and zapA, we observed a poor or even lack of the typical swarming 191 growth in approximately 20% of the isolates that might be due to their low expression and/or correspond to the 192 significant role of other genes in this type of motility [32][33][34][35][36]. This mechanism is also associated with phylogenetic 193 relationships among P. mirabilis strains [29,37]. We noticed that the strains with more expansive swarming were 194 related to less of the other strains in the Dienes test. The demarcation line is still rather poorly understood, and 195 probably regulated by multiple mechanisms, like the type VI secretion system, operons idsABCDEF and 196 idrABCDE, and the hpc-vgrG effector [38][39][40]. However, factors responsible for the disability of strains to 197 extensively swarm in place of the mutual cohabitation with other strains have not been unambiguously identified.

198
Budding et al. [41] suggested that this characteristic might reflect environmental competition between P. mirabilis 199 strains which might also have implications in hospital settings.

200
No correlation between the urease and hemolysin expression levels and the other pathogenicity properties 201 or susceptibility profiles was observed. The ureolytic activity peaked quickly during culture incubation and was 202 independent on the type and rate of the bacterial growth, i.e. in broth and on agar, being so similar in both swimmer 203 and swarmer cells. This activity is critical in the formation of bacteria-induced stones, particularly dangerous for 204 long-term catheterized patients [1,42]. Urease is also involved in forming crystalline biofilms on the catheter [24]; 205 however, we did not observe any correlation between the enhanced urease expression and stronger biofilm.

206
The ability to form biofilm varies remarkably, even among strains of the same species [43][44][45]. Our 207 results (Fig 7) indicate that the P. mirabilis biofilm may be stronger than that of E. coli [46]. That may mean that 208 P. mirabilis grows faster, providing greater biofilm yield, which might be important during the host invasion. The 209 decrease of the P. mirabilis growth rate might inhibit the biofilm formation, unlike in E. coli. We also observed 210 that the biofilm formation on the glass correlated proportionally with that on the polyurethane, which was again in 211 contrast to E. coli [46]. Czerwonka et al. [11] reported that high hydrophobicity of the P. mirabilis cell surface 212 correlated with low biofilm amount, which is important for hydrophobic surfaces like glass. It was proved that 213 hydrophilic catheters may prevent catheter-associated UTIs [47], so it would be advisable to evaluate 214 hydrophobicity rates that are characteristic for the majority of P. mirabilis clinical strains.

215
The statistical analysis allowed us to observe specific correlations between virulence-associated 216 properties and antimicrobial susceptibility profiles. We found that stronger swarming was significant among 217 susceptible strains and vice versa. It was not confirmed in case of tazobactam, the effect of which is independent 218 of the swarming motility [48]. Similar tests were conducted by Auer et al. [49] who tested swarmer cells for sensitivity to wall-modifying antibiotics. They found that thickness of the peptidoglycan makes swarmer P.

233
Susceptibility to antibiotics also correlated with biofilm formation level. In general, we observed that 234 MDR strains formed stronger biofilm, both on the polyurethane and on the glass, compared to susceptible isolates.

235
It has been well-evidenced that biofilm increases resistance to antimicrobials, mainly due to extracellular matrix The Ward's agglomeration test applied for the analyzed MICs allowed for deeper differentiation of the 251 strains. As expected, the isolates were separated into two clusters: cluster 1, grouping the susceptible isolates, and 252 cluster 2, comprising the MDR ones (Fig 6). Moreover, the correlations described above were clearly demonstrated 253 in these two groups. The antibiotic resistance correlated with pathogenic properties of P. mirabilis differently than 254 in case of E. coli [58]. A similar study of pathogenic properties of P. mirabilis was reported by Stańkowska et al.

255
[59] who differentiated the swarming growth rate, and ureotylic, proteolytic and hemolytic activities, calculating 256 the relative virulence index based on cumulative scores for these activities. However, our study showed that 257 virulence properties may be expressed alternatively by P. mirabilis strains. Biofilm and swarming growth seemed 258 to be antagonistic to each other, ureolytic activity was similar in all of the isolates, and hemolysin was rarely 259 detected, suggesting lower relevance of the relative virulence index.

260
To sum up, we excluded the potential rule, that kindred strains which not form a demarcation line will   Table) Table 3. Cycling conditions were as follows: denaturation at 94°C for 2 min, followed 301 by 30 cycles of 1 min at 94°C, 1 min at varying annealing temperature, and 1 min at 72°C, followed by 5 min at 302 72°C.

308
The ureolytic activity in the Christensen broth was analyzed according to the method described previously regarded as related to each other. Additionally, the expansiveness / rate of swarming motility was evaluated for 325 each strain by its proportional coverage of the plate after the overnight incubation, compared to other strains 326 cultured on the same plate. The expansiveness was classified into three categories: category 1, weak swarming 327 (coverage <5%); category 2, medium swarming (5% -≤ 25%); and category 3, intensive swarming (25% -≤ 50%).

328
The bacterial behavior in each combination of strains was tested in triplicates.

Biofilm formation 330
The ability of P. mirabilis strains to form biofilm was analyzed on glass and polyurethane surfaces Five of the most representative images were photographed. All strains were classified into three groups: group 1, 335 lack of biofilm, single cells observed; group 2, single microcolonies of biofilm; and group 3, biofilm covering the 336 entire coverslip. The biofilm formation on the polyurethane was measured spectrophotometrically at 531nm after 337 overnight incubation of bacterial strains in Luria Bertani broth and crystal violet staining. The strains unable to 338 form biofilm were classified based on the crystal violet adsorption on the polyurethane measured as A 531 <0.08.

551
The unpaired two-tailed T-test was used for a statistically significant difference (p<0.05) (Graph Pad Prism v. 6).

576
S2 Table. The raw data of the research results.