First‐time characterization of viable but non‐culturable Proteus mirabilis: Induction and resuscitation

Abstract Pathogenic bacteria can enter into a viable but non‐culturable (VBNC) state under unfavourable conditions. Proteus mirabilis is responsible for dire clinical consequences including septicaemia, urinary tract infections and pneumonia, but is not a species previously known to enter VBNC state. We suggested that stress‐induced P. mirabilis can enter a VBNC state in which it retains virulence. P. mirabilis isolates were incubated in extreme osmotic pressure, starvation, low temperature and low pH to induce a VBNC state. Resuscitation was induced by temperature upshift and inoculation in tryptone soy broth with Tween 20 and brain heart infusion broth. Cellular ultrastructure and gene expression were examined using transmission electron microscopy (TEM) and quantitative real‐time polymerase chain reaction (qPCR), respectively. High osmotic pressure and low acidity caused rapid entry into VBNC state. Temperature upshift caused the highest percentage of resuscitation (93%) under different induction conditions. In the VBNC state, cells showed aberrant and dwarf morphology, virulence genes and stress response genes (envZ and rpoS) were expressed (levels varied depending on strain and inducing factors). This is the first‐time characterization of VBNC P. mirabilis. The ability of P. mirabilis pathogenic strains to enter a stress‐induced VBNC state can be a serious public health threat.

phase upon release from the human body to the environment can run undetectable or underestimated and can thus threaten public health. 10 Some bacteria in the VBNC state were found to display manifestations of their virulence attributes while others can lose part of their virulence properties. 11 Bacteria can also display virulence traits upon resuscitation. 12 Furthermore, cells in the VBNC state were found to show additional virulence traits, such as antibiotic resistance which were not shown in their normal state. 13 This might explain recurrent infections by antibiotic-resistant bacteria that enter VBNC state and then retain their full activity upon condition improvement. 7  Proteus mirabilis bacteria are Gram-negative, facultative anaerobes that can be responsible for dire clinical consequences including septicaemia, urinary tract infections and pneumonia. 15 Proteus sp. possess several virulence factors including fimbriae, flagella, urease enzyme, lipopolysaccharide, capsular polysaccharide and metalloproteinase (serralysin). 16 Proteus sp. are normal flora in the intestinal tract. 17 Their presence in soil or water is an indicator of faecal pollution and can consequently cause infection to human consuming contaminated water or contaminated seafood. 18 Some sea animals are reported to absorb and accumulate Proteus spp.
such as sponges (Spongia offinialis), 19 oysters 20 and turtles. 21 Proteus sp. is the causative agent for many urinary tract and wound infections as well as nosocomial infections. 22 Studies have reported similarities between Proteus sp. isolated from human intestine and those causing urinary tract infection (autoinfection) 23 or food poisoning. 21 We suggested that P. mirabilis can enter a VBNC state and that this VBNC state impacts gene expression and cellular ultrastructure of the VBNC cells as compared to control cells. We also studied different P. mirabilis VBNC-inducing factors and resuscitating conditions. To the best of our knowledge, this is the first-time characterization of VBNC P. mirabilis.

| Bacterial strains
Three P. mirabilis clinical isolates were recovered from patients in Kasr Al-Aini Hospital, Cairo, Egypt. Approvals from the research ethics committees of the hospital and the faculty of Pharmacy, October University for Modern Sciences and Arts, Giza, Egypt were obtained prior to conducting the study. Two isolates, P6 and P7, were recovered from urinary tract infections, and the isolate, coded P3, was recovered from wound infection. Written consents were obtained from all participants in the study.

| Induction of the VBNC state
The VBNC state induction was done according to Pinto et al 24 with some modifications. P. mirabilis strains were grown overnight in tryptone soy broth (TSB) at 37°C with shaking. The culture was centrifuged at 5000 rpm for 10 minutes at 5°C, and pellets were washed twice by sterile saline solution. The resulting pellets were resuspended in sterile deionized water, and the cell concentration was adjusted to be equivalent to McFarland 0.5 (1.5 × 10 8 CFU/ mL). A volume of 400 µL was used to inoculate 40 mL of the induction medium, in a 50-mL falcon tube, to reach a final density of approximately 10 6 CFU/mL. Different induction media were used to create different stressful conditions for isolates: (a) deionized water, for starvation and low osmotic pressure stress; (b) deionized water + NaCl (0.9%), for starvation stress; (c) deionized water + NaCl (0.9%) at pH 5, for starvation and acidic condition stresses; (d) deionized water + NaCl (4%), for starvation and high osmotic pressure stresses, and (e) deionized water + NaCl (7%), for starvation and high osmotic pressure. Each induction condition was done in duplicate, and tubes were incubated at 4°C, without shaking and in dark. Culturability on TSA was assessed every week in duplicate plates, and time required for loss of culturability was recorded. Test of culturability was done by testing the growth of 1 mL of culture, and then, when no growth is detected, 10 mL of culture was centrifuged and resuspended in 1 mL and repeated subculture was done. Heat killed bacteria at 100°C for 15 minutes were used as control cells. 30 mL of each VBNC sample was centrifuged at 10 000 g for 10 minutes at 4°C. Total RNA was extracted from cultures using the SV Total RNA Isolation ® System (Promega). Extraction was done according to the manufacturer's instructions. RNA was stored in RNase-free water and then stored at −80°C until use. RNA quantity and quality were measured at wavelengths 260 and 280 nm using ND-1000 spectrophotometer (NanoDrop Technology). RNA quality was also tested by agarose gel electrophoresis. The RNA samples were treated with RNase-free DNase I (New England Biolab).

| RNA isolation and cDNA preparation
Samples were confirmed to be free of genomic DNA by 40 cycles of conventional PCR using 16S rRNA primers (long amplicon, ~1.5 kb).
After confirming the absence of genomic DNA by agarose gel analysis, ~l µg DNase-treated RNA was subjected to reverse transcription using the SensiFastTM cDNA synthesis kit (Bioline) following the supplier's directions (Table 1). Control bacteria for qRT-PCR were prepared by inoculation in BHI medium and incubation at 37°C for the late exponential phase. This was followed by total RNA extraction using the same methodology that was implemented for the VBNC samples.

| Validation of the VBNC state
The RNA expression was used as tool indicating viability of cells. 25 After loss of culturability on routine media by VBNC-induced cells, the continuous expression of the housekeeping genes 16S rRNA (short amplicon, 189 bp) and Rpos was used as confirmation for cell viability under different stressful conditions using specific primers ( Table 1). The amplicon was detected using gel electrophoresis for 16S rRNA gene and qRT-PCR for RpoS gene.  Inoculated media were incubated at 37°C for 72 hours, and then, the appearance of turbidity indicated resuscitation of VBNC cells and regaining of the culturability of VBNC cells. 27

| Quantitative PCR conditions
We studied the change in gene expression of the following genes: stress response genes (rpoS and envZ), virulence genes (fliL, zapA and luxS) and genes for cell division (ftsZ). Expression in VBNC cells was compared to control cells. The primers were designed using the complete genome sequence of P. mirabilis (HI4320) obtained from the NCBI database (NCBI accession no. NC_010554). Primers for the qPCR used in the current study (Table 1)   TA B L E 1 List of oligonucleotide sequences and the corresponding annealing temperatures RT-PCR cycling conditions were as follows: one cycle with 95°C for 2 min; then 40 cycles of denaturation at 95°C for 5 seconds, annealing at 55°C for 10 seconds, and extension and fluorescent data collection at 72°C for 20 seconds. A dissociation curve was generated at the end of each reaction. In all qPCR runs, negative controls without template were run in parallel. The 16s rRNA gene (housekeeping gene) was selected as the internal control based on the results of BestKeeper ® software tool. 28 The relative mRNA levels of genes of interest were determined and normalized to the expression of the housekeeping gene using the ∆∆ CT value analysis. 29 The qPCR data were expressed as the fold change in expression levels of genes in P. mirabilis potential VBNC as compared to their levels in the control cells unexposed to stress (calibrators) grown in BHI medium and incubated for the late exponential phase. Only genes with a relative 2 −∆∆C value above 1.0 or below 1.0 were considered significant. [Correction added on 17 February 2020: 2 ΔΔC has been changed to 2 −ΔΔC in the preceding sentence.]

| Transmission electron microscopy (TEM)
The structures of control, VBNC and resuscitated cells were investigated using transmission electron microscope (TEM). Cells were harvested by centrifugation, and the bacterial pellets were washed three times with PBS. Cell fixation and processing steps were performed according to the method described in Henriques et al. 30 Micrographs for each sample were captured at 10 000g magnification power, and cells were observed for structural alterations using

| Statistics
Results of three replicates were analysed for statistical significance with PRISM 5 (GraphPad). A one-way analysis of variance (ANOVA) was performed. Data comparisons were performed using Bonferroni's correction for multiple testing.

| RE SULTS
The samples were exposed to different VBNC induction conditions which include a combination of either acidic or osmotic pressure in addition to both cold and starvation stresses. The number of weeks taken by each of the three samples to enter in the VBNC state varied from 14 to 26 weeks according to the inducing condition ( Figure 1).
Isolates incubated under isotonic pressure (deionized water + 0.9% NaCl) lost their ability for culturing after a time ranging from 18 to 26 weeks (Average 22 weeks). High osmotic pressure (deionized water + NaCl 4% or 7%) and acidic conditions caused rapid entry into VBNC state compared to isotonic water microcosm. The time range required for entry of low osmotic pressure (deionized water only) was between 16 and 21 weeks (average 18.5 weeks).
Samples that lost their ability for culturing were confirmed to be viable by subjecting them to total RNA extraction followed by reverse transcription to cDNA. After amplification by conventional PCR, the expression of the 16S rRNA and RpoS genes was detected by gel electrophoresis (Figure 2) and qRT-PCR (Figure 3), respectively. The continuous expression of these genes was detected in all samples under different VBNC induction conditions, and this confirmed the entry into the VBNC state. No detectable 16s rRNA or Rpos in killed control cells after 48 hours of heat treatment.
As shown in Table 2

| D ISCUSS I ON
The VBNC state is a potential public health hazard in the environment, 7 because bacterial cells under different environmental stresses such as low or high osmotic pressure, 31 starvation 32 and acidic pH 33 can shift to the VBNC state which could run undetected.
During the transient existence of pathogenic bacteria in the environment before infecting another person, they might be exposed to stressful conditions that cause the entry of these cells into the VBNC state. 34 There are no previous data in the literature that characterizes P. mirabilis VBNC state.
The entry of P. mirabilis (starting count 1.5 × 10 6 CFU/mL) into the VBNC state under cold conditions (4°C) was more rapid than at room temperature (25°C), and the maximum time required for loss of culturability was 26 weeks at 4°C and over than 65 weeks at 25°C (data not shown). The rapid entry into VBNC state at 4°C could be attributed to the inability of the cells to detoxify lethal peroxides produced under stressful conditions, because cold incubation prevents catalase activity and its de novo synthesis 35 This can lead to an increase in toxins and failure of the antitoxin effect which is an inducer of the VBNC state. 7 Proteus mirabilis showed more rapid entry (20 weeks) into the VBNC state compared to the E coli O157:H7 in a study 8 (34 weeks),

F I G U R E 1 The time required for
Proteus mirabilis cells to lose culturability under various VBNC induction condition. All samples were incubated at 4°C in water microcosm, hence exposed to cold and starvation stresses. Induction media include the following: (A) Water (W) + 0.9% NaCl. (B) Water (W) + 0.9% NaCl, pH 5. (C) Water (W). (D) Water (W) + 4% NaCl. (E) Water (W) + 7% NaCl F I G U R E 2 Gel electrophoresis of PCR product amplified with the 16s rRNA primer set (189 bp) of the cDNA produced by reverse transcription of RNA extracted from P. mirabilis samples 3 (lanes 1-5) and 6 (lanes 6-10) exposed to different VBNC conditions at 4°C. Media used include the following: W, deionized water (lanes 1,6); S, deionized water + NaCl 0.9% (lanes 2,7); deionized water + NaCl 0.9% at pH 5 (lanes 3,8); 4%, deionized water + NaCl 4% (lanes 4,9); deionized water + 7% NaCl (lanes 5,10). Lane Lad: represents the molecular size ladder 100-bp DNA ladder that used similar starting bacterial concentration and similar VBNCinducing conditions. Thus, the speed of entry depends on the microbial species. In our study, it was noticed that exposure of P. mirabilis to combined stressful factors lead to more rapid entry (P < .01) into the VBNC state than exposure to single factor as detected in E. coli by Muela et al 36 Putting the cells under starved conditions makes them more resistant to stresses such as acid and oxidizing agents than growing cells. 37 Moreover, Proteus sp was found to survive in marine environments due to their halotolerance trait attributed to the polyphosphate kinase encoded by the ppk gene, 16,38 and ppk mutant strains of Campylobacter jejuni were found to exhibit decreased ability to enter the VBNC state, 39 which could explain their ability to retain viability while losing culturability under high salt concentrations.
The entry of the P. mirabilis isolates into the VBNC state was confirmed by detecting the continuous expression of the 16S rRNA and rpoS genes in tested strains after losing their culturability. RNA detection was used because the half-life of RNA is 3-5 minutes and its continuous expression is an indication of continuous viability of F I G U R E 3 Transcription levels of selected genes (ftsZ, zapA, fliL, luxS, rpoS and envZ) of each P. mirabilis strain [3(A), 6 (B), 7 (C)] in the VBNC state, incubated in various environmental conditions. The conditions used include: water microcosm, water + 0.9% NaCl, water + 0.9% NaCl, pH5, water + 4% NaCl and water + 7% NaCl. Standard deviations are indicated as vertical bars. Asterisks indicate statistically significant differences in the expression of each gene between VBNC samples and control, as analysed using the one-way ANOVA with Bonferroni's correction for multiple testing (*P ≤ .05; **P ≤ .01; ***P ≤ .001; ns, no significant difference) bacteria that have lost their culturability. 40 Note: (+) indicates growth, (-) indicates no growth.

F I G U R E 4 Representative TEM
micrographs showing the structural changes in P. mirabilis (strain 6), in (A) control, (B) resuscitated and (C) VBNC state (after starvation for 182 days in water microcosm + 0.9% NaCl). Samples A and B were examined at magnification 15 000× while VBNC cells were examined at 50 000×

A B
C upshift with a resuscitation success rate of 93.3% (14/15), which could be attributed to reaching the optimum temperature for catalase enzyme activity which can catalyse the breakdown of toxic peroxides. 35 The lowest percentage of resuscitation was observed using BHI medium (1/8 concentration). By comparing the effect of low-nutrient (BHI medium of 1/8 concentration) and high-nutrient medium (TSB with additive Tween 20) on resuscitation, it was observed that the high-nutrient medium was more efficient in resuscitating VBNC P. mirabilis strains than the low-nutrient medium. This result is in contrast to what was reported by Bloomfield et al 45 who noted that low-nutrient environment was more efficient in resuscitation because of the toxic effect of high nutrients on cells in the VBNC state as a result of production of free radicals. Adding Tween 20 to TSB medium was assumed to reduce surface tension and improve the interaction between VBNC cells and the culture medium, 46 in addition to acting as carbon source that supplies energy to VBNC cells. 47 The virulence factors known to be involved in pathogenesis of P.
mirabilis include adhesins, motility, biofilm formation, immunoavoidance, toxins and nutrient acquisition. 48  In our study, the envZ gene, which is part of osmoregulating system Stress δ s factor, coded by rpoS gene, controls the production of trehalose, which acts as a protective osmolyte, and is associated with persistence in the VBNC state. 59  which was shown to cleave secretory immunoglobulins and antimicrobial peptides and hence establishing urinary tract infections. 65 Although zapA is expressed at low levels in vivo, a P. mirabilis zapA mutant is less efficient than wild type at colonizing the urine, bladder and kidneys in murine infections. 66 In our study, the expression of zapA gene was variable depending on the strain and the VBNC induction conditions but it persisted under unfavourable conditions indicating that P. mirabilis retained its metalloproteinase virulence traits during the VBNC state.
Proteus mirabilis possesses a luxS homologue and produces AI-2, which can mediate interactions between species. 67 Continuous expression of luxS gene was observed in all tested strains under VBNC conditions. However, AI-2 does not likely contribute to pathogenicity, 67 in spite of studies that suggest the involvement of luxS expression in biofilm formation. 68 The down-regulation of the luxS gene under different VBNC-inducing conditions in our study might indicate a reduced ability of VBNC cells for biofilm formation.
In conclusion, P. mirabilis can enter the VBNC state under several conditions of starvation, extreme osmotic pressure, acidity and low temperature. The time required for entry into a VBNC state is dependent on the inducing condition and the bacterial strain.

CO N FLI C T O F I NTE R E S T
The authors declare no competing financial interests.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data used to support the findings of this study is available from the corresponding author upon reasonable request.