Abstract
The exclusion of opportunistic pathogens is important for protecting animal health and ensuring desired research outcomes in highly immunodeficient mice. Proteus mirabilis has been associated with gastrointestinal tract lesions, septicemia, pyelonephritis, splenomegaly, and hepatitis and can influence select mouse models. To inform health-surveillance practices after we experienced difficulty in excluding P. mirabilis from our mouse colony, we aimed to determine the likelihood of detecting P. mirabilis-positive immunocompromised (SRG), immunovague (Fbn1+/–), and immunocompetent (CD1) colony mice through culture and PCR testing; to evaluate transmission via 2 sentinel-based approaches (direct contact and indirect dirty-bedding transfer); and to further characterize associated pathology. We hypothesized that immunocompromised mice would be better detectors and transmitters of P. mirabilis. Multiple logistic regression models were used for analysis and included PCR copy number, repeated testing, age, sex, and antibiotic-treated (trimethoprim–sulfamethoxazole) diet as covariates. Repeated testing over 10 wk showed that P. mirabilis–colonized immunocompromised colony mice were 95 times more likely than immunocompetent mice to test positive by culture and 30 times more likely by PCR assay. Sentinel mice were 15 times more likely to test positive by PCR assay for P. mirabilis when exposed by direct contact compared with dirty bedding and 18 times more likely to test positive when exposed to positive immunocompromised as compared with immunocompetent colony mice. After 10 wk of exposure, 3.8% of dirty-bedding sentinel PCR tests were positive, as compared with 30.7% of contact sentinels. Only immunocompromised mice on antibiotic diet (37.5%) developed lesions of the urogenital tract and abdominal cavity consistent with known pathology of P. mirabilis. Our findings suggest that PCR testing of dirty-bedding sentinels alone is not sufficient for the detection of P. mirabilis in mouse colonies. Direct-contact sentinels and testing of colony mice—especially if immunocompromised—with adjunct culture may facilitate successful bioexclusion.
Abbreviations: CD1, immunocompetent Crl:CD1(ICR); GEM, genetically engineered Fbn1+/– mice on a 100% C57Bl/6NTac background (C57BL/6NTac–Fbn1tm2.1Regn); SRG, immunocompromised human signal regulatory protein α knock-in mouse on a Rag2−/− Il2rg−/− background (Balb/cAnNTac;129S4/SvJae–Rag2tm1.1Regn Sirpatm1.1(SIRPA)RegnIl2rgtm1.1Regn); TMS, trimethoprim–sulfamethoxazole; OR, odds ratio; MLST, multilocus sequence typing
Bioexclusion and the maintenance of SPF colonies in laboratory animal facilities are common practices to improve animal welfare and limit the effect of spontaneous disease on research outcomes.32,38 Excluded pathogens are often those with known pathogenic potential and established effects on research17,47 and can include opportunistic pathogens and agents that are naturally found in both human and laboratory animal environments. Exclusion of opportunistic pathogens protects susceptible, immunocompromised hosts and has been applied in select facilities, requiring strict biosecurity practices.24,37 Procedures for bioexclusion are specifically needed to refine research outcomes in immunocompromised animals and may improve experimental reproducibility.21,35,49 However, successful exclusion is contingent on detection, for which optimal methods have not been well established.24,37 Many opportunistic pathogens are found in low copy numbers when PCR screening assays are used. When screening enzootically infected colonies, neither culture nor PCR testing has been 100% reliable.43
The opportunistic pathogen Proteus mirabilis is a gram-negative, urease-positive, anaerobic, rod-shaped, swarming bacteria of the order Enterobacteriales.3,5 This pathogen is environmentally ubiquitous and considered commensal gut flora in many vertebrate species, including humans and mice.4,48 The prevalence of P. mirabilis in rodent colonies exceeds 5% and it is not commonly excluded.24 Its route of transmission is unknown; however, both fecal–oral and environmental transmission have been proposed.7,48
Common methods for detecting pathogens such as P. mirabilis in research rodent colonies include the use of direct- and indirect-contact sentinels,34 with direct-contact sentinels used to provide the highest probability of detecting pathogens that are poorly transmitted through feces or air.6,16 The most recent Institute for Laboratory Animal Research survey on colony surveillance found that 100% of institutions used indirect dirty-bedding sentinels, but only 6% of institutions reported using direct-contact sentinels as an adjunct method.10 Instead of using direct-contact sentinels, many institutions sampled colony mice. Similarly, a survey published in 2017 regarding the prevalence of adventitial infections in mice and rats at biomedical research facilities found that 95% of institutions relied on dirty-bedding sentinels. The use of direct-contact sentinels for routine animal monitoring was not reported.37 For P. mirabilis, PCR testing of environmental samples, feces, oral swabs, and body swabs is available and commonly used for colony screening,20 and the culture of P. mirabilis as a predominant or single isolate is the ‘gold standard’ method of diagnosis20
Disease from P. mirabilis has rarely been noted in both immunocompetent and immunocompromised strains of mice and is associated with the hematogenous route of infection.7 P. mirabilis infection is often inapparent; when present, clinical signs include weight loss and diarrhea due to gastrointestinal ulcerations and septicemia.20 The kidney is commonly affected in the septicemic state; specifically, P. mirabilis is a causal organism for suppurative pyelonephritis in male mice with diabetes,46 spontaneous nephritis in C3H/HeJ mice,30,36 and fatal splenomegaly, focal necrotizing hepatitis, peritonitis, and pyelonephritis in SCID mice.42,44 Because of the organism’s propensity to affect the urinary tract, mouse models have emerged to study P. mirabilis catheter-associated urinary tract infections in people,13 which are among the most common health care-associated infections worldwide.25,39 In addition, the presence of P. mirabilis has been found to influence the severity of clinical signs in mouse models of colitis, irritable bowel syndrome,22 and Parkinson disease.12 The use of antibiotics for symptomatic rodents or for bioexclusion purposes is not recommended, because it neither resolves the carrier state nor eliminates bacteria from the environment.11
To best support the production of genetically modified—specifically immunocompromised or immunovague—mice critical for its drug pipeline, our animal facility is considered to be free of both specific and opportunistic pathogens. The large-scale exclusion of P. mirabilis in particular has posed a myriad of transmission and detection challenges related to the design of the animal health surveillance program, testing methodologies, and interpretation of test results. To better understand P. mirabilis detection and transmission and thus inform practical applications for animal health surveillance programs, we first aimed to model the likelihood of detecting P. mirabilis positive colony mice based on immune status in immunocompetent, immunovague, and highly immunocrompromised mice. Second, we sought to model P. mirabilis transmission by assessing 2 sentinel-monitoring methods, direct contact and dirty-bedding transfer, in immunocompetent, immunovague, and highly immunocompromised mice. Third, we further characterized the histopathologic changes in mice colonized with P. mirabilis by evaluating any lesions for evidence of bacterial infection and inflammation. In addition, experiments were designed to reveal the influence of the diet on our aims and on antimicrobial resistance and genetic differences in detectable P. mirabilis strains. We hypothesized the detection and transmission of P. mirabilis would be greater in immunocompromised mice than in immunocompetent strains and that the use of antibiotic-impregnated feed would reduce the likelihood of detection and transmission.
Materials and Methods
Mice.
All mice were housed within an AAALAC-accredited mouse production barrier facility at Regeneron Pharmaceuticals (Tarrytown, NY). All procedures were performed with approval from Regeneron Pharmaceuticals’ IACUC in accordance with the Guide for the Care and Use of Laboratory Animals, eighth edition,26 and the AVMA Guidelines for Euthanasia of Animals, 2013 or 2020 edition,1,2 depending on whether mice were euthanized before or after the updated guidelines were adopted in October 2020. Individual mice with no previous identifiers were ear punched for the purpose of this study. Highly immunocompromised human signal regulatory protein α knock-in Balb/cAnNTac;129S4/SvJae–Rag2tm1.1Regn Sirpatm1.1(SIRPA)RegnIl2rgtm1.1Regn (SRG) mice and immunovague genetically engineered C57BL/6NTac–Fbn1tm2.1Regn (GEM) mice were generated inhouse. We included heterozygous (+/–) Fbn1 mice as representatives of immunovague genetically engineered mice, because they were the only strain engineered inhouse without an overt immunodeficiency that tested positive for P. mirabilis on index animal sampling. All genetically engineered mice can be considered as ‘immunovague’ because their immune system often is not characterized fully, as is true of the Fbn1 mouse. Therefore, all genetically engineered mice can be considered to have potential immunodeficiencies that may affect their susceptibility to opportunistic pathogens.21,47 Immunocompetent Crl:CD1(ICR) mice (CD1) were purchased from Charles River Laboratories (Wilmington, MA) to serve as dirty-bedding transfer sentinels in routine colony health surveillance and for these experiments specifically.
Husbandry.
All SRG, GEM, and CD1 mice on study were housed 1 to 5 per cage in IVC (GM500, Tecniplast, Buguggiate, Italy) on 1/8-in. irradiated corncob bedding with a cotton square and underwent regular cage changes (once every 7 to 14 d). All SRG mice were weaned from dams fed irradiated 0.025% trimethoprim- and 0.124% sulfamethoxazole-containing (TMS) diet (Test Diet, St Louis, MO) ad libitum and were maintained on this diet until the start of experiments. GEM and CD1 mice were fed irradiated standard rodent diet (LabDiet 5053; Purina LabDiet, St Louis, MO). All mice had ad libitum access to reverse-osmosis–purified water that was hyperchlorinated to approximately 2 ppm and administered via automatic watering system valves. Macroenvironmental conditions included a 12:12-h light:dark cycle, temperature maintenance at 68 to 72 °F (20.0 to 22.2 °C), and relative humidity between 30% and 70%. Mouse housing rooms are supplied HEPA-filtered air and have 10 to 15 air changes hourly, and IVC have 75 air changes hourly.
Murine pathogen exclusion.
The following murine pathogens are excluded at the mouse production facility: Sendai virus, mouse coronavirus, murine norovirus, mouse parvoviruses, murine rotavirus, ectromelia virus, pneumonia virus of mice, mouse theilovirus, reovirus types 1 through 4, lymphocytic choriomeningitis virus, mouse adenovirus types 1 and 2, Helicobacter spp., Citrobacter rodentium, Streptobacillus moniliformis, Mycoplasma pulmonis, cilia-associated respiratory bacillus, Rodentibacter heylii, Clostridium piliforme, Pseudomonas aeruginosa, Salmonella, Campylobacter, Bordetella bronchiseptica, B. pseudohinzii, Corynebacterium kutscheri, C. bovis, Streptococcus pneumoniae, Pneumocystis spp., external and internal parasites as well as the opportunistic pathogens Klebsiella pneumoniae, K. oxytoca, Pseudomonas aeruginosa, Proteus mirabilis, Rodentibacter pneumotropicus, Staphylococcus aureus, and β-hemolytic Streptococcus groups A, B, C, and G. Prior to enrollment in the study and as is standard procedure at the facility, SPF status was monitored through a dirty-bedding sentinel program consisting of testing sentinels quarterly via PCR assay and serology and annually through whole-animal assessment. Samples were submitted to Charles River Laboratories according to FELASA guidelines.35,40 Sentinel mice were purchased free of excluded agents and were pair-housed CD1 female mice that were 5 to 8 wk old at first bedding exposure and that were maintained for a maximum of 6 mo. One sentinel cage is located on each side of each IVC rack, and each side houses 88 cages of colony mice. At scheduled cage change, 1 tablespoon of dirty bedding is transferred from each colony cage to the sentinel cage; this amount is equivalent to 16 oz of bedding per side.
Experiment inclusion criteria.
All P. mirabilis–positive mice used in the experiments were spontaneously infected and derived from P. mirabilis–positive mouse production racks within the barrier facility. The SRG and GEM mice were former colony mice that had been housed on P. mirabilis–positive racks, and the CD1 mice were prior rack sentinels that tested PCR-positive for P. mirabilis during health surveillance. All cages with a P. mirabilis–positive result were transferred to an isolation room outside of the barrier for confirmatory testing. Cages were selected after at least 2 consecutive P. mirabilis–positive PCR test results from a combined analysis of samples of feces, oral swabs, and body swabs. No minimal PCR copy number was required for inclusion purposes. Negative mice were either transferred from a known negative rack within the barrier facility (SRG and GEM) or purchased from Charles River Laboratories at the start of the study (CD1). All negative mice received at least 1 confirmatory negative PCR test prior to the start of experiments.
Aim 1: Modeling the likelihood of P. mirabilis detection.
A total of 64 mice (24 males and 40 females; age, 27 to 89 wk at the time of euthanasia; Table 1) were used to model the likelihood of detecting P. mirabilis–positive colony mice based on immune status. Specifically, 32 P. mirabilis–positive SRG mice were assigned randomly to either TMS or 5053 diet; mice assigned to 5053 diet were switched from TMS diet to 5053 feed on the first day of the experiment (n = 16/group). In addition, 16 P. mirabilis–positive GEM mice were maintained on 5053 for the duration of the experiment (n = 16/group). Finally, 32 P. mirabilis–positive CD1 mice were assigned randomly to receive either TMS or 5053 diet; mice assigned to TMS diet were switched from 5053 feed to TMS diet on the first day of the experiment (n = 8/group). Feces, oral swabs, and body swabs were collected and pooled by cage (n = 1 to 4 mice per cage) for PCR testing at week 0. Individual (nonpooled) samples were collected at weeks 1 and 2. Samples from weeks 0, 1, and 2 were evaluated together with regard to the consistency of positive PCR results (that is, as always positive or inconsistent). Live mice were then shipped to CRL (n = 4/group for SRG and GEM; n = 2/group for CD1 mice) for final P. mirabilis PCR, cecal culture, antimicrobial susceptibility testing, gross necropsy, and histopathology at weeks 4, 6, 8, and 10. Diagnostic lab personnel who performed PCR and culture were blind to all experimental groups.
Table 1.
Demographic profile of immunocompromised (SRG), immunovague (GEM), and immunocompetent (CD1) P. mirabilis–colonized mice
| n | Sex (M/F) | Age (wk) | PCR-positive (final) | PCR copy number (final) | Consistently PCR-positive (weeks 0, 1, and 2) | Culture-positive (final) | Trimethoprim–sulfamethoxazole susceptibility testing | |||
|---|---|---|---|---|---|---|---|---|---|---|
| Resistant | Intermediate | Sensitive | ||||||||
| Overall | 64 | 37%/63% | 53 (27–89) | 81% | 100 (0–53367) | 86% | 80% | 76% (39/51) | 2% (1/51) | 22% (11/51) |
| Diet | ||||||||||
| TMS | 24 | 25%/75% | 54 (27–89) | 88% | 404 (0–26561) | 83% | 88% | 100% (21/21) | 0 | 0 |
| 5053 | 40 | 45%/55% | 53 (36–63) | 78% | 50 (0–53367) | 88% | 75% | 60% (18/30) | 3% (1/30) | 37% (11/30) |
| SRG | ||||||||||
| All | 32 | 44%/56% | 53 (42–71) | 97% | 608 (0–53367) | 94% | 97% | 100% (31/31) | 0 | 0 |
| TMS | 16 | 38%/62% | 54 (42–71) | 100% | 404 (50–26561) | 94% | 100% | 100% (16/16) | 0 | 0 |
| 5053a | 16 | 50%/50% | 50 (43–61) | 94% | 1630 (0–53367) | 93% | 94% | 100% (15/15) | 0 | 0 |
| GEM | ||||||||||
| All (5053 only) | 16 | 38%/62% | 50 (36–62) | 69% | 25 (0–404) | 67% | 75% | 0 | 8% (1/12) | 92% (11/12) |
| CD1 | ||||||||||
| All | 16 | –/100% | 58 (27–89) | 63% | 19 (0–1630) | 81% | 50% | 100% (8/8) | 0 | 0 |
| TMSb | 8 | –/100% | 55 (27–89) | 63% | 63 (0–1630) | 63% | 63% | 100% (5/5) | 0 | 0 |
| 5053 | 8 | –/100% | 58 (53–63) | 63% | 9 (0–100) | 100% | 38% | 100% (3/3) | 0 | 0 |
Data are given as median (range), percentage, or number.
All SRG mice previously on TMS diet and transitioned to 5053 diet at week 0.
All CD1 mice were previously on 5053 chow and transitioned to TMS diet at week 0 (one had prior exposure to TMS as a dirty-bedding sentinel).
Aim 2: Modeling the likelihood of P. mirabilis transmission.
To model P. mirabilis transmission, contact sentinel cages were set up to cohouse a single P. mirabilis–positive colony mouse (either CD1 or SRG) with 4 negative contact sentinel mice (2 SRG mice and 2 CD1 mice; Figure 1). The 4 groups (n = 4 cages/group) comprised sentinels cohoused with 1) a positive SRG mouse on TMS; 2) a positive SRG mouse on 5053; 3) a positive CD1 mouse on TMS; and 4) a positive CD1 mouse on 5053. In total, 80 mice were used for the direct contact transmission study, all of which were female (age, 11 to 84 wk; Table 2). Mice were not age-matched, given that Aim 1 found that age was not a predictor for PCR-positive results (See Table 4). Male mice were not used as contact sentinels due to animal welfare concerns regarding cohousing unfamiliar adult males and practical utility for contact sentinel programs for female embryo transfer recipients used in the production of genetically modified mice.33 We collected 3 PCR samples per cage on weeks 0, 1, 3, 5, 7, and 9 to monitor for positive P. mirabilis results in the originally negative mice: 1 from the known positive mouse, a pooled sample from the 2 SRG contact sentinels, and a pooled sample from the 2 CD1 contact sentinels.
Figure 1.
Aim 2 experimental design: modeling the likelihood of P. mirabilis transmission through direct contact and via transfer of dirty bedding (indirect contact).
Table 2.
Demographic profile of direct- and indirect (dirty-bedding)-contact sentinels exposed to known P. mirabilis–positive colony mice
| SRG mice on TMS diet | SRG mice on 5053 diet | CD1 mice on TMS diet | CD1 mice on 5053 diet | |
|---|---|---|---|---|
| Known-positive colony mice | ||||
| n | 4 | 4 | 4 | 3a |
| Age (wk) | 56 | 63 (61–68) | 44 | 68 (68–92) |
| Sex (M/F) | –/100% | –/100% | –/100% | –/100% |
| Copy number of screening PCR assay | 302 (25–1630) | 1630 (201-6579) | 63 (3–404) | 19 (6–201) |
| Direct-contact sentinel mice | ||||
| SRG (n = 2/group) | 8 | 8 | 8 | 6a |
| Age (wk) | 50 (22–57) | 38 (37–43) | 22 | 22 |
| Sex (M/F) | –/100% | –/100% | –/100% | –/100% |
| PCR-positive results | 35% | 65% | 5% | 0 |
| CD1 (n = 2/group) | 8 | 8 | 8 | 6 |
| Age (wk) | 38 | 38 | 38 (19–52) | 38 |
| Sex (M/F) | –/100% | –/100% | –/100% | –/100% |
| PCR-positive results | 55% | 60% | 10% | 0 |
| Dirty-bedding sentinel mice | ||||
| SRG (n = 2–5/group) | 4 | 8 | 8 | 10b |
| Age (wk) | 41 (22–61) | 34 (33–35) | 34 | 38 (34–42) |
| Sex (M/F) | 100%/– | 100%/– | –/100% | 100%/– |
| PCR-positive results | 0 | 20% | 0 | 0 |
| CD1 (n = 3/group) | 6 | 6 | 6 | 6 |
| Age (wk) | 38 | 38 | 38 | 38 |
| Sex (M/F) | –/100% | –/100% | –/100% | –/100% |
| PCR-positive results | 0 | 10% | 0 | 0 |
Data are given as median (range) or percentage (no. of positive tests / total no. of tests performed) of mice positive for P. mirabilis during 10 wk of exposure.
One originally P. mirabilis–positive CD1 mouse on 5053 diet was found dead during week 2 of the study, so the fourth group of contact sentinels was removed from analysis.
3 of the 5 mice in one of the 2 groups were euthanized during week 6 due to fight wounds.
Table 4.
Logistic regression model of the relationship between mouse strain (SRG, GEM, CD1) and P. mirabilis PCR results
| OR | 95% CI | P | |
|---|---|---|---|
| Full model (n = 56) | 0.0001 | ||
| Strain | |||
| SRG | 33.9 (31.00) | 0.85–1350.99 (4.23–227.08) | 0.061 (0.001) |
| GEM | 7.87 (2.20) | 0.16–381.39 (0.76–6.33) | 0.298 (0.144) |
| CD1 | Reference group | ||
| Diet | |||
| TMS | 11.38 (2.03) | 0.35–369.39 (0.49–8.40) | 0.171 (0.327) |
| 5053 | Reference group | ||
| Consistency of PCR results | |||
| Always positive | 100.50 (33.00) | 2.59–2914.08 (4.94–220.56) | 0.014 (<0.0001) |
| Inconsistent | Reference group | ||
| Culture (final) | |||
| Positive | 0.99 (18.80) | 0.08–12.34 (4.14–85.40) | 0.994 (<0.0001) |
| Negative | Reference group | ||
| Age (wk) at final | 0.98 (0.94) | 0.87–1.11 (0.88–1.01) | 0.761 (0.084) |
| Reduced model (n = 56) | <0.0001 | ||
| Strain | |||
| SRG compared with CD1 | 30.40 | 1.63–566.56 | 0.022 |
| Consistency of PCR resultsa | |||
| Always positive compared with inconsistent | 35.94 | 3.29–392.90 | 0.003 |
For the multivariate full model, odds ratios (OR), 95% CI, and P values are reported as adjusted (crude). Sex and final PCR copy number were removed due to perfect predictability. Interaction terms as a set could not be evaluated due to perfect predictability, collinearity, or lack of observations; therefore, interactions were not included in the model. For example, interaction between diet and strain could not be evaluated, because 100% of SRG mice on TMS diet were P. mirabilis–positive on PCR assay (see Table 1). The reduced model excludes variables found nonsignificant by likelihood-ratio χ2 test (age, diet, and culture result).
Confounding variable.
The dirty-bedding sentinels were exposed to dirty bedding generated from the previously mentioned contact sentinel experimental cages (Figure 1). A total of 16 oz. of dirty bedding (approximately 50% from each cage) was transferred weekly from each of the 16 contact sentinel cages with at least 1 known P. mirabilis positive mouse to each of the 16 cages of known negative dirty-bedding sentinel mice. The amount of bedding transferred was approximately equal to the amount transferred to a dirty-bedding sentinel cage after collection from a large rack (88 cages) and included an equal amount of fresh bedding. A total of 4 groups received dirty bedding (n = 4 cages/group): 1) SRG sentinels on TMS diet, 2) SRG sentinels on 5053 diet, 3) CD1 sentinels on TMS diet, and 4) CD1 sentinels on 5053 diet. Negative dirty-bedding sentinel CD1 females were housed at n = 3 per cage, and negative SRG mice were housed n = 2 to 5 per cage and by sex (26 male, 4 female). Previous cage configurations were not disturbed to prevent hierarchical aggression between unfamiliar male mice,33 and male mice were not excluded because, to our knowledge, no evidence indicates that sex plays a role in pathogen detection in soiled-bedding sentinels.8 In total, 54 mice were used as dirty-bedding sentinels aged 22 to 61 wk (Table 2). To prevent contamination between cages, the cage change station, bedding transfer scoop, and outer gloves were cleaned with germicidal wipes between cages (Super Sani-Cloth, Germicidal Disposable Wipe, CovCare, Staten Island, NY). Pooled PCR samples were collected by cage on weeks 0, 2, 4, 6, 8, and 10 after start of bedding transfer.
Aim 3: Experimental design for pathologic characterization of P. mirabilis-associated lesions.
We selected 24 P. mirabilis–positive mice (from Aim 1) to further characterize histopathologic changes in mice colonized with P. mirabilis: 8 SRG mice on TMS diet, 8 SRG mice on 5053 diet, 4 GEM on 5053 diet, 2 CD1 mice on TMS diet, and 2 CD1 mice on 5053 diet. Another 8 mice were submitted as P. mirabilis–negative controls: 2 SRG mice on each diet, 2 GEM on 5053 only, and 1 CD1 mouse on each diet. A representative section from each fixed tissue was trimmed, embedded in paraffin, and cut at approximately 4 µm onto glass microscope slides. Fixed tissue from all mice included heart, lung, liver, spleen, kidneys, bladder, stomach, duodenum, ileum, cecum, jejunum, and colon. When gross lesions were noted in any other tissues, they were collected also and included pancreas, mesentery, gallbladder, lymph nodes, preputial gland, and thymus. All slides were stained with hematoxylin and eosin. At the discretion of the pathologist (TA), additional sections were cut and stained with a modified Gram stain to visualize bacteria. Slides were examined under a light microscope and findings were reported by tissue. Pathology was reported as either negative (not present) or positive (present) for lesions. The severity and distribution of lesions were described. Signalment was provided to the pathologist.
PCR analysis.
Mice were handled by the tail base for perianal and dorsal body swab sampling and then scruffed for ventral body swab sampling. Ventral and dorsal swabs were taken from cranial to caudal and caudal to cranial. Perianal swabbing was performed from ventrum to the perianal area to the tail base. The oral swab was then collected by inserting the cotton-tipped applicator into the mouth and swabbing in both cheeks. A naturally voided fresh fecal sample was collected at any point during the sampling process directly from the mouse. Feces, oral swabs, and body swabs were pooled per sample type and then sent to Charles River Laboratories for Taqman qPCR analysis targeting the ZapA gene of P. mirabilis in a combined analysis of all 3 sample types.24 Based on an alignment of available DNA sequences using National Center for Biotechnology Information BLAST® tool at the time of assay development, targeting ZapA will not amplify other species of Proteus. PCR copy numbers were calculated by comparing the count for the test sample with the score for our plasmid-based positive control that represents 100 copies of the target sequence. The score represents the PCR cycle at which the amplification of the test sample crosses the threshold to become interpreted as positive, and all count scores are rounded to the next whole number by the thermal cycler software. There is an inverse correlation between count and estimated copy number.
Aerobic culture and antimicrobial susceptibility testing.
At the final time point for Aim 1, live mice were sent to CRL for euthanasia via CO2 overdose in accordance with the AVMA Guidelines for the Euthanasia of Animals: 2013 and 2020 editions.1,2 Full necropsy was performed and gross lesions recorded as noted in aim 3. Samples were aseptically collected postmortem from the cecum for P. mirabilis culture and isolation. Isolates were grown on Hektoen enteric agar, MacConkey agar, Pseudomonas isolation agar, and trypticase soy agar containing 5% sheep blood (Hardy Diagnostics, Springboro, OH). P. mirabilis was identified via MALDI-TOF spectrophotometry. Subcultures of P. mirabilis were grown on trypticase soy agar containing 5% sheep blood for determining the antimicrobial susceptibility to TMS by using the Kirby–Bauer disk diffusion assay according to standard methods.
Multilocus sequence typing.
A P. mirabilis isolate resistant to TMS and an isolate sensitive to TMS were submitted for multilocus sequence typing (MLST) through the Accugenix Microbial Identification Portal (Newark, DE). The resistant strain was derived from the cecal contents of a 62-wk-old male SRG mouse on TMS diet. The sensitive strain was derived from the cecal contents of a 54-wk-old male GEM on 5053 diet. Four housekeeping genes— recA, rpoB, gyrB, and helD—were analyzed for comparison of sequence types.
Statistical analysis.
Three logistic regressions were performed to model the likelihood of detecting P. mirabilis in colonized mice (aim 1). The first 2 regressions model the likelihood of detecting P. mirabilis in immunocompromised, immunovague, and immunocompetent mice by either culture (Table 3) or PCR (Table 4) (positive or negative; binary variables). The third regression models the likelihood of detecting PCR copy numbers (Table 5) greater than or equal to 100 (<100 or ≥100; binary variable). Copy number as a continuous variable could not be evaluated by logistic regression, because linearity between the logit of the outcome and copy number was not established (Box–Tidwell, P < 0.0001); in addition, these data could not be transformed to meet necessary assumptions. The binary variable was developed by using 5 PCR reactions or duplications as a cut-off. The major exposure for each regression was the strain of mice (SRG, GEM, CD1; tertiary variable). Additional variables controlled for in each model include diet (TMS or 5053, binary variable), consistency of PCR positivity over previous time points (all positive or not; binary variable; 8 values not reported), and age at the final time point (continuous variable).
Table 3.
Logistic regression model of the relationship between mouse strain (SRG, GEM, and CD1) and P. mirabilis culture results
| OR | 95% CI | P | |
|---|---|---|---|
| Full model (n = 56) | <0.0001 | ||
| Strain | |||
| SRG | 140.07 (31.00) | 2.27–8643.74 (3.37–285.26) | 0.019 (0.002) |
| GEM | 35.16 (3.00) | 0.74–1674.59 (0.67–13.40) | 0.071 (0.150) |
| CD1 | Reference group | ||
| Diet | |||
| TMS | 35.34 (2.33) | 0.88–1416.30 (0.57–9.52) | 0.058 (0.237) |
| 5053 | Reference group | ||
| Consistency of PCR results | |||
| Always positive | 149.50 (21.00) | 2.68–8353.80 (3.42–128.91) | 0.015 (0.001) |
| Inconsistent | Reference group | ||
| PCR result (final) | |||
| Positive | 1.12 (18.80) | 0.07–17.28 (4.14–85.40) | 0.934 (<0.0001) |
| Negative | Reference group | ||
| Final PCR copy number | |||
| 100 or greater | 0.71 (12.03) | 0.04–13.97 (2.38–60.71) | 0.820 (0.003) |
| Less than 100 | Reference group | ||
| Age (wk) at final | 1.02 (0.95) | 0.89–1.16 (0.88–1.01) | 0.811 (0.096) |
| Reduced model (n = 56) | <0.0001 | ||
| Strain | |||
| SRG compared with CD1 | 95.25 | 3.72–2437.84 | 0.006 |
| GEM compared with CD1 | 26.85 | 1.18–613.68 | 0.039 |
| Dieta | |||
| TMS compared with 5053 | 27.64 | 1.21–627.88 | 0.037 |
| Consistency of PCR resultsa | |||
| Always positive compared with inconsistent | 111.54 | 5.86–2124.56 | 0.002 |
In the multivariate full model, odds ratios (OR), 95% CI, and P values are reported as adjusted (crude). Sex and final PCR copy number were removed due to perfect predictability. Interaction terms as a set could not be evaluated due to perfect predictability, collinearity, or lack of observations; therefore, interactions were not included in the model. For example, interaction between diet and strain could not be evaluated, because 100% of SRG mice on TMS diet were P. mirabilis–positive on culture (see Table 1). The reduced model excludes variables found nonsignificant by likelihood-ratio χ2 test (age, PCR result, and final PCR copy number). The data interpretation is as follows: SRG mice were 95 times (OR) more likely to test positive by culture than were CD1 mice (reference group; P = 0.006); GEM were 27 times (OR) more likely to test positive by culture than were CD1 mice (reference group; P = 0.039); mice on a TMS diet were 28 times (OR) more likely to test positive by culture than were mice on 5053 diet (reference group; P = 0.037); and mice with consistently positive PCR results were 112 times (OR) more likely to test positive by culture than were mice with inconsistent PCR results (reference group; P = 0.002).
Confounding variable
Table 5.
Logistic regression model of the relationship between mouse strain (SRG, GEM, CD1) and P. mirabilis PCR copy number ≥100
| OR | 95% CI | P | |
|---|---|---|---|
| Full model (n = 56) | <0.0001 | ||
| Strain | |||
| SRG | 15.21 (21.27) | 1.36–170.04 (4.33–104.36) | 0.027 (<0.0001) |
| GEM | 0.04 (0.51) | 0.00–2.64 (0.10–2.62) | 0.130 (0.418) |
| CD1 | Reference group | ||
| Diet | |||
| TMS | 2.62 (3.32) | 0.32–21.67 (1.09–10.09) | 0.370 (0.035) |
| 5053 | Reference group | ||
| Consistency of PCR results | |||
| Always positive | 20.25 (15.40) | 0.84–488.81 (1.74–136.56) | 0.064 (0.014) |
| Inconsistent | Reference group | ||
| Culture result | |||
| Positive | 1.99 (12.03) | 0.14–28.32 (2.38–60.71) | 0.610 (0.003) |
| Negative | Reference group | ||
| Age (wk) at final) | 0.88 (0.96) | 0.76–1.02 (0.91–1.01) | 0.086 (0.138) |
| Reduced model (n = 56) | <0.0001 | ||
| Strain | |||
| SRG compared with CD1 | 25.89 | 3.51–190.86 | 0.001 |
| Consistency of PCR results | |||
| Always positive compared with inconsistent | 24.66 | 0.94–644.02 | 0.054 |
| Age (wk) at final | 0.89 | 0.79–1.00 | 0.056 |
For the multivariate full model, odds ratios (OR), 95% CI, and P values are reported as adjusted (crude). Sex and Final PCR result were removed due to perfect predictability. Interaction terms as a set could not be evaluated due to perfect predictability, collinearity, or lack of observations; therefore, interactions were not included in the model. For example, interaction between Diet and Strain could not be evaluated, because all but one SRG mouse on TMS diet had a PCR copy number of 100 or greater (see Table 1). The reduced model excludes variables found nonsignificant by likelihood-ratio χ2 test (Diet and Culture result).
One logistic regression was performed (Table 6) to model the likelihood of detecting P. mirabilis in sentinel mice after 10 wk of exposure (at least 1 positive result or all negative, binary variable) controlled for method of exposure (direct contact or indirect [dirty-bedding] sentinels; binary variable), strain of P. mirabilis–positive colony mice (SRG or CD1, binary variable), screening PCR copy number of P. mirabilis–positive colony mice (less than 100 or greater than or equal to 100, binary variable), diet (TMS or 5053, binary variable), strain of sentinel mice (SRG or CD1, binary variable), sex of sentinel mice (male or female, binary variable), and age of sentinel mice (continuous variable). This regression was further stratified according to method of detection and strain of positive colony mice.
Table 6.
Logistic regression model of the relationship between method of exposing sentinel mice (direct contact and indirect contact [dirty-bedding transfer]) and P. mirabilis–positive PCR screening results
| OR | 95% CI | P | |
|---|---|---|---|
| Full model (n = 46) | 0.0001 | ||
| Method of exposure | |||
| Direct contact | 73.67 (8.00) | 3.64–1390.63 (1.54–41.49) | 0.005 (0.013) |
| Indirect contact (dirty bedding transfer) | Reference group | ||
| Strain of P. mirabilis–positive colony mice | |||
| SRG | 37.19 (10.56) | 2.33–593.40 (2.42–45.98) | 0.010 (0.002) |
| CD1 | Reference group | ||
| PCR copy number of colony mice | |||
| ≥100 | 0.50 (2.63) | 0.04–5.78 (0.69–10.02) | 0.580 (0.158) |
| <100 | Reference group | ||
| Diet | |||
| TMS | 1.05 (0.87) | 0.16–6.80 (0.27–2.84) | 0.959 (0.813) |
| 5053 | Reference group | ||
| Strain of sentinels | |||
| SRG | 0.14 (0.48) | 0.01–1.60 (0.14–1.60) | 0.114 (0.230) |
| CD1 | Reference group | ||
| Sex of sentinels | |||
| Female | 0.03 (3.70) | 0.00–2.09 (0.40–34.60) | 0.104 (0.252) |
| Male | Reference group | ||
| Age (wk) of sentinels at final | 1.08 (1.09) | 0.96–1.21 (1.01–1.19) | 0.171 (0.033) |
| Reduced model (n = 46) | <0.0001 | ||
| Method of exposure | |||
| Direct contact compared with indirect | 15.07 | 2.15–105.50 | 0.006 |
| Strain of P. mirabilis–positive colony micea | |||
| SRG compared with CD1 | 17.81 | 3.15–100.79 | 0.001 |
| Reduced model stratified according to contact sentinels (n = 30) | 0.0007 | ||
| Strain of P. mirabilis–positive colony mice | |||
| SRG compared with CD1 | 15.89 | 2.65–95.21 | 0.002 |
| Reduced model stratified according to P. mirabilis–positive SRG colony mice (n = 24) | 0.0077 | ||
| Method of exposure | |||
| Direct contact compared with indirect | 19.53 | 1.73–222.40 | 0.016 |
| Dieta | |||
| TMS compared with 5053 | 0.18 | 0.02–1.95 | 0.158 |
For the multivariate full model, odds ratios (OR), 95% CI, and P values are reported as adjusted (crude). Age and sex of originally positive mice were removed due to failed convergence, because all mice exposed to a positive mouse (age 61 or 63 wk) tested positive and all contact sentinels were female (see Table 2). PCR copy number data of the positive colony mice was derived from screening results. The reduced model excludes variables found to be nonsignificant according to likelihood-ratio χ2 test (PCR copy number of colony mice, Diet, Strain of sentinels, Sex of sentinels, and Age of sentinels).
For stratification according to Method of exposure, Sex was again removed as all contact sentinels were female. The model could not be stratified by indirect contact dirty-bedding sentinels, because the probability of the χ2 test could not be computed due to limited observations of positive results (n = 4) by this method.
For stratification by Strain of positive colony mice, Sex was removed from the model due to collinearity. We were unable to stratify according to CD1 mice due to perfect predictability by Method of exposure and Sex.
Confounding variable
All logistic regressions were reported as adjusted odds ratios (OR; crude OR), 95% CI, and P values of the overall test of the model and each parameter estimate. The favored model for each logistic regression was the reduced model that included covariates that met the following inclusion criteria and were thus found to contribute to the predictability of the model: 1) P value less than 0.1 (α) of covariate in full model and then 2) P value less than 0.05 (α) of likelihood-ratio χ2 test. The likelihood-ratio χ2 test for parameter estimates was used to compare the full logistic model with a model excluding the covariate of interest. All possible interactions in the reduced model were evaluated as a set to determine significance by using a χ2 test to compare the reduced logistic models, with or without the set of interaction variables. Confounders were defined as covariates that, when added to the favored model, resulted in a 10% change or greater in the slope of the major exposure: strain (aim 1) or method of exposure (aim 2) and were controlled for by inclusion in the reduced model.
Demographic data were described by using medians and range. The Shapiro–Wilk test was used to determine whether copy number data were normally distributed. The nonparametric Mann–Whitney or Kruskal–Wallis test was used to compare copy number data outside of logistic regression. Pairwise and group comparisons were corrected by using the Dunn multiple-comparison test. A χ2 test was performed to compare percentage of positive test results outside of logistic regression analysis. A P value of less than 0.05 was considered to indicate statistical significance.
Statistical analysis for logistic regressions was performed by using STATA/IC 16.0 for Mac (StataCorp, College Station, TX). Prism version 9.0 for Mac (GraphPad Software, La Jolla, CA) was used for all other statistical analyses.
Results
Modeling the likelihood of P. mirabilis detection.
As compared with their immunocompetent counterparts, P. mirabilis-colonized immunocompromised mice were more likely to test positive by culture and PCR assay and to have higher PCR copy numbers.
Mice colonized with P. mirabilis were more likely to test culture positive when they were immunocompromised (SRG mice) or immunovague (GEM), were fed a TMS diet, or had consistently positive PCR results (Table 3). According to logistic regression, the odds of culturing positive for P. mirabilis for SRG mice was 95 times those of CD1 mice (P = 0.006), whereas the odds for GEM was 27 times those of CD1 mice (P = 0.039), controlled for diet and consistency of PCR results. The odds of culturing positive for P. mirabilis for mice on the TMS diet were 28 times those on 5053 chow (P = 0.037). The odds of culturing positive for P. mirabilis for mice with 3 consecutively positive PCR results over a 3-wk time frame was 112 times those without 3 consecutively positive PCR results (P = 0.002).
Mice colonized with P. mirabilis were more likely to test PCR positive when they were immunocompromised or had consistently positive PCR results (Table 4). By using logistic regression, the odds of SRG mice testing positive for P. mirabilis on PCR assay were 30 times those of CD1 mice, when controlled for diet and consistency of PCR results (P = 0.022). The odds of testing positive for P. mirabilis on PCR assay at the final time point for mice with 3 prior consecutively positive PCR results over a 3-wk time frame was 36 times those without 3 consecutively positive PCR results (P = 0.003).
Mice colonized with P. mirabilis were more likely to have higher PCR copy numbers (≥100) if they were immunocompromised, young, or had consistently positive PCR results (Table 5). According to logistic regression, the odds of higher copy numbers were 26 times higher in SRG mice than in CD1 mice, controlled for age and consistency of PCR results (P = 0.001). Every 1 wk increase in age results in 11% lower odds of having copy numbers greater than or equal to 100 (P = 0.056). The odds of higher copy numbers for mice with 3 consecutively positive PCR results over a 3-wk time frame were 25 times those without 3 consecutively positive PCR results (P = 0.054).
Compared with other immunophenotypes, immunocompromised mice more consistently maintained P. mirabilis–positive status: 93.8% of SRG mice on 5053 (n = 15) and 100% of SRG mice (n = 16) on TMS tested positive for P. mirabilis at every time point, 37.5% of GEM (n = 6), 50% of CD1 mice on 5053 (n = 4) and 62.5% of CD1 mice on TMS (n = 5). Many mice that tested negative by both PCR and culture at necropsy (final time point) varied in test status at earlier time points. For example, the one SRG mouse that did not maintain positive status was positive on PCR during prescreening and at week zero, negative at week one, positive on week 2, and negative at necropsy by both PCR and culture. Of mice with positive test results at necropsy, 84% (n = 47) agreed by both PCR and culture as positive, 7% (n = 4) tested positive only on culture, and 9% (n = 5) tested positive only by PCR (Figure 2). When analyzed according to strain, all of the positive SRG mice (100%; n = 31) and the majority of GEM (91%; 10 of 11) tested positive through both methodologies, whereas less than half of the CD1 mice that tested positive at necropsy tested positive by both methodologies (39%; 5 of 13).
Figure 2.
Agreement between culture and PCR test results for P. mirabilis–
positive mice at the final experimental time point (aim 1).
TMS resistance and MLST results revealed the presence of at least 2 different P. mirabilis strains. All culture-positive SRG and CD1 mice were colonized with P. mirabilis organisms that were resistant on disk diffusion to TMS antibiotics, regardless of diet provided on study (Table 1); note that all SRG mice had prior exposure to the TMS diet, and only 1 CD1 mouse experimentally on 5053 had a known previous indirect exposure to TMS as a dirty-bedding sentinel. All GEM that cultured positive at the final time point were colonized with P. mirabilis organisms that were either sensitive (92%, 11 of 12) or intermediate (8%, 1 of 12) to TMS antibiotics; all GEM had no prior exposure to TMS diet. One TMS-resistant and 1 TMS-sensitive isolate were selected randomly to undergo MLST. The resistant isolate was derived from the cecal contents of a 62-wk-old male SRG mouse on the TMS diet, with a PCR copy number of 26,561 at this time point. The sensitive isolate was derived from the cecal contents of a 54-wk-old-male GEM on 5053 diet, with a PCR copy number of 100 at this time point. The submitted resistant and sensitive isolates differed at 3 of the 4 housekeeping genes sequenced—0% at recA, 0.58% at rpoB, 0.18% at gyrB, and 0.85% at helD— and were determined to be different sequence types.
Modeling the likelihood of P. mirabilis transmission.
Sentinel mice exposed to known P. mirabilis–positive mice were more likely to test PCR positive when exposed to immunocompromised mice and through direct contact.
By using logistic regression, the odds of sentinel mice testing P. mirabilis–positive at least 1 time by PCR assay was 15 times greater (P = 0.006) for those exposed through direct contact compared with those exposed through indirect contact (dirty-bedding transfer) during 10 wk of exposure to a positive colony mouse (Table 6). The odds of sentinel mice that were exposed to positive SRG mice testing P. mirabilis–positive through PCR analysis at least once during 10 wk of exposure were 18 times those of sentinel mice exposed to positive CD1 mice (P = 0.001).
When stratified according to method of detection, the odds of direct-contact sentinels exposed to positive SRG mice testing P. mirabilis–positive by PCR assay were 16 times (P = 0.002) those of direct-contact sentinels exposed to positive CD1 mice (Table 6). When stratified according to strain of positive colony mice, the odds of contact sentinels exposed to positive SRG mice testing P. mirabilis–positive by PCR analysis were 20 times (P = 0.018) those of dirty-bedding sentinels exposed to positive SRG mice (Table 6). The odds that sentinel mice exposed to positive SRG mice on TMS diet would test positive by PCR assay were 82% lower than those of SRG mice on 5053 diet (P = 0.158).
Despite the limited predictability of PCR copy number in this model, positive SRG colony mice on 5053 diet had the highest copy numbers at screening (Kruskal–Wallis, P = 0.031; Figure 3 A), and sentinels, both direct-contact and dirty-bedding, exposed to this group of positive colony mice had the greatest percentage of positive sentinel test results (Table 2, Figure 3 B). The incidence of positive PCR results in bedding sentinels was only 3 of 80 total tests (3.8%) performed over the 10 wk of exposure, with the first positive result occurring at week 8. In contrast, a total of 46 of 150 (30.7%) PCR test results were positive in the contact sentinel group (P < 0.0001). Direct-contact sentinels began to test positive for P. mirabilis at 1 wk after exposure (Table 7). The positive results in bedding sentinels occurred only in mice exposed to P. mirabilis–positive SRG mice on 5053 diet. Two of the positive results in bedding sentinels occurred at weeks 8 and 10 from the same cage of SRG sentinels, and the third positive result occurred during week 8 from a cage of CD1 sentinels (Table 8).
Figure 3.
(A) Median and range of copy numbers after PCR screening of P. mirabilis–positive colony mice (aim 2). Median PCR copy numbers for colony positive SRG mice on TMS diet was 302.5 (25-1630) and 1630 (201-6579) for mice on 5053 diet. Median PCR copy numbers for colony positive CD1 mice on TMS diet was 62.5 (3-404) and 18.5 (6-201) for mice on 5053 diet. When considering all 4 groups, screening copy numbers were significantly different (Kruskal–Wallis; *, P = 0.031; pairwise comparisons were not significantly different after adjustment by using the Dunn multiple-comparisons test). (B) Percentage of PCR-positive results (no. of positive tests / total no. of tests performed × 100%) of sentinels (direct contact and dirty bedding [indirect contact]) exposed to P. mirabilis–positive colony mice. These data also were evaluated separately through regression analysis.
Table 7.
Weekly P. mirabilis–positive test results of direct-contact sentinels
| Week | ||||||
|---|---|---|---|---|---|---|
| 0 | 1 | 3 | 5 | 7 | 9 | |
| SRG + / TMS diet | ||||||
| Cage 1A | ||||||
| SRG | − | − | − | − | − | − |
| CD1 | − | + | + | + | + | + |
| Cage 1B | ||||||
| SRG | − | + | + | + | + | + |
| CD1 | − | + | − | + | − | + |
| Cage 1C | ||||||
| SRG | − | − | − | − | + | + |
| CD1 | − | − | − | − | − | + |
| Cage 1D | ||||||
| SRG | − | − | − | − | − | − |
| CD1 | − | − | − | − | + | + |
| SRG + / 5053 diet | ||||||
| Cage 2A | ||||||
| SRG | − | − | − | − | − | − |
| CD1 | − | − | + | − | − | − |
| Cage 2B | ||||||
| SRG | − | + | + | + | + | + |
| CD1 | − | + | + | + | + | + |
| Cage 2C | ||||||
| SRG | − | + | − | + | + | + |
| CD1 | − | + | + | + | + | + |
| Cage 2D | ||||||
| SRG | − | − | − | + | + | + |
| CD1 | − | + | − | − | − | − |
| CD1 + / TMS diet | ||||||
| Cage 3A | ||||||
| SRG | − | − | − | − | − | − |
| CD1 | − | − | − | + | − | − |
| Cage 3B | ||||||
| SRG | − | − | − | − | − | − |
| CD1 | − | − | − | − | − | + |
| Cage 3C | ||||||
| SRG | − | − | − | − | − | − |
| CD1 | − | − | − | − | − | − |
| Cage 3D | ||||||
| SRG | − | − | + | − | − | − |
| CD1 | − | − | − | − | − | − |
| CD1 + / 5053 diet | ||||||
| Cage 4A: Excluded from analysis due to CD1+ mouse found dead | ||||||
| Cage 4B | ||||||
| SRG | − | − | − | − | − | − |
| CD1 | − | − | − | − | − | − |
| Cage 4C | ||||||
| SRG | − | − | − | − | − | − |
| CD1 | − | − | − | − | − | − |
| Cage 4D | ||||||
| SRG | − | − | − | − | − | − |
| CD1 | − | − | − | − | − | − |
Table 8.
Weekly P. mirabilis–positive test results of indirect (dirty-bedding transfer) sentinels
| Week | ||||||
|---|---|---|---|---|---|---|
| 0 | 2 | 4 | 6 | 8 | 10 | |
| Bedding from SRG + / TMS diet | ||||||
| Cage 1: SRG | − | − | − | − | − | − |
| Cage 2: SRG | − | − | − | − | − | − |
| Cage 3: CD1 | − | − | − | − | − | − |
| Cage 4: CD1 | − | − | − | − | − | − |
| Bedding from SRG + / 5053 diet | ||||||
| Cage 5: SRG | − | − | − | − | + | + |
| Cage 6: SRG | − | − | − | − | − | − |
| Cage 7: CD1 | − | − | − | − | − | − |
| Cage 8: CD1 | − | − | − | − | + | − |
| Bedding from CD1 + / TMS diet | ||||||
| Cage 9: SRG | − | − | − | − | − | − |
| Cage 10: SRG | − | − | − | − | − | − |
| Cage 11: CD1 | − | − | − | − | − | − |
| Cage 12: CD1 | − | − | − | − | − | − |
| Bedding from CD1 + / 5053 diet | ||||||
| Cage 13: SRG | − | − | − | − | − | − |
| Cage 14: SRG | − | − | − | − | − | − |
| Cage 15: CD1 | − | − | − | − | − | − |
| Cage 16: CD1 | − | − | − | − | − | − |
Pathologic characterization of P. mirabilis-associated lesions.
Some P. mirabilis–colonized immunocompromised mice on TMS diet had lesions consistent with known pathology of P. mirabilis. All mice submitted for pathology appeared clinically healthy on daily animal health and wellbeing assessments throughout experimentation. Of the 8 sampled P. mirabilis–positive SRG mice on TMS diet, 3 mice (37.5%; nos. 5, 6, and 8; 3 of 16 SRG in total [18.8%]) had inflammatory lesions associated with gram-negative bacilli. These 3 mice also had multiple gross findings that correlated with the microscopic lesions. All 3 mice had nephritis (Figure 4 A and B). Mice 5 and 8 had intralesional gram-negative bacilli in the kidney lesions. Mouse 5 had other foci of inflammation affecting the abdominal wall, pancreas, and cecal mesentery, suggesting peritonitis; gram stain of the abdominal wall was negative for bacteria. Mouse 6 had inflammation of the abdominal mesentery with intralesional gram-negative bacilli, suggesting peritonitis (Figure 4 C). Mouse 8 had inflammatory lesions in connective tissue near the heart base, urinary bladder (Figure 4 D), and accessory sex glands, with gram-negative bacilli in the urinary bladder and accessory sex gland lesions; gram stain of the heart lesion was negative for bacteria. Despite the absence of bacteria in some of the inflammatory lesions, they were considered to be part of the constellation of lesions caused by bacterial infection in these 3 mice. In addition, all 3 mice had splenic or hepatic extramedullary hematopoiesis of mild to moderate severity, consistent with an ongoing inflammatory process.
Figure 4.
Representative histologic bacterial induced inflammatory lesions in P. mirabilis–colonized SRG mice on TMS diet. (A) Kidney from mouse 6, showing moderate neutrophilic pyelitis and interstitial nephritis. Hematoxylin and eosin stain; magnification, 40×. (B) Intralesional gram-negative bacilli (arrows) in mouse 5. Gram stain; magnification, 100×. (C) Focal inflammation of the mesenteric fat with gram-negative bacilli (arrows), indicative of peritonitis in mouse 6. Gram stain; magnification, 100×. (D) Mild neutrophilic debris and gram-negative bacilli (arrows) within the urinary bladder, indicative of cystitis in mouse 8. Gram stain; magnification, 40×.
PCR analysis was performed on fixed sections with gram-negative rods and 2 negative controls. P. mirabilis was identified by PCR assay of fixed tissue in 2 blocks from mouse 8: block 1 included the kidneys (PCR copy number, 3), and block 2 included all other lesions including the urinary bladder, accessory sex glands, and connective tissues of the heart base (PCR copy number, 12). The grossly visible abscess of the seminal vesicle in mouse 8 was also cultured prior to paraffin fixation. Both P. mirabilis and Enterococcus faecalis were isolated from the lesion. The P. mirabilis strain isolated was resistant to TMS. No cocci were visualized microscopically.
All 16 SRG mice had decreased splenic lymphoid tissue, which was consistent with their highly immunodeficient status (strain-related). Excluding this change, 6 of the 16 SRG mice (3 on 5053 diet, 3 on TMS diet) and 2 of the 4 GEM on 5053 diet had no microscopic changes. All other mice, including CD1 mice, had mild noninflammatory and inflammatory microscopic changes, which were considered incidental and unlikely to be related to or caused by bacterial infection.
Discussion
This study documented that the detection and transmission of P. mirabilis is increased and thus improved in highly immunocompromised mice. Specifically, as compared with immunocompetent mice, P. mirabilis–colonized immunocompromised mice were more likely to test positive by culture and PCR analysis and to have higher PCR copy numbers. In addition, sentinel mice exposed to known P. mirabilis–positive mice were more likely to test PCR positive when exposed to the bedding of immunocompromised mice and by direct contact. Incorporating contact sentinels as a component of animal health surveillance, especially within immunocompromised mouse colonies, could improve diagnostic potential as compared with dirty-bedding sentinels. Additional considerations for program design for the exclusion of P. mirabilis are discussed later and include diagnostic methodology (culture and PCR analysis, including copy numbers), repeated testing, direct testing of colony mice, use of antibiotic-containing diet, and the immune status of colony and sentinel mice at exposure.
Our results suggest that culture with bacterial isolation remains the gold-standard diagnostic method for P. mirabilis. SRG mice were 95 times more likely than CD1 mice to culture positive for P. mirabilis, compared with 30 times more likely to test positive on PCR testing. Diet and immunovague status were additional predictors for testing positive by culture that were not predictive by PCR analysis; specifically, GEM were 27 times more likely than CD1 mice to culture positive for P. mirabilis, but PCR testing had no predictive value. Although PCR analysis is considered to be a more sensitive diagnostic method than culture, PCR testing has limitations as a diagnostic method. For example, in the context of P. mirabilis–positive environmental samples, the limit to PCR sensitivity was defined by samples that grew at least 1 × 103 cfu/mL.50 This threshold may explain some of the inconsistencies in PCR-positive results at different time points in the current study. PCR analysis as a sole testing method may not be optimal for detecting opportunistic organisms with low bacterial burden and should be used in conjunction with culture.
As a screening method, repeated PCR testing strengthens the interpretation of a positive PCR result. This study showed that 3 prior consecutively positive PCR results predicted both final culture and final PCR positive results. In known P. mirabilis– positive mice, those with consistently positive PCR results were 112 times more likely to culture positive and 36 times more likely to test positive on PCR assay as compared with mice that had inconsistent PCR results. Culture samples were not collected at each time point but only at the final time point, which prevents consistent comparison of culture and PCR results. In addition, high PCR copy numbers reinforce PCR interpretation, given that mice with consistently positive PCR results were 25 times more likely to have high copy numbers than were mice with inconsistent PCR results. However, high copy numbers (≥100) were not indicative of culture positivity, thus complicating the development of a copy number cut-off for a true PCR-positive result for P. mirabilis,9 which is yet to be established and validated. This drawback could be especially problematic for facilities using antibiotic-treated diet for immunocompromised rodent colonies; such diets could lower bacterial load and result in persistent bacterial infection and antibiotic tolerance as described for other facultative anaerobes.23 Furthermore, a potential decrease in pathogenicity and distribution may hinder detection.28 PCR copy numbers also decreased with age; every 1 wk increase in age resulted in 11% lower odds of developing high copy numbers. This association could be indicative of latent infection27,41 in older mice or the development of colonization resistance.18,31
When applying PCR as a colony surveillance tool, our study found that sentinel mice exposed to known P. mirabilis–positive mice were more likely to test PCR-positive when exposed to highly immunocompromised mice and direct contact. Overall, contact sentinels were 15 times more likely than dirty-bedding sentinels to test positive and were 20 times more likely to be positive when exposed to immunocompromised P. mirabilis–positive mice. Among the total of 80 PCR tests performed on dirty-bedding sentinels, only 3 (3.8%) were positive for P. mirabilis throughout the 10 wk of exposure, as compared with 46 of the 150 tests (30.7%) on direct-contact sentinels. This finding conveys that P. mirabilis is much more effectively transmitted via direct contact than indirectly in dirty bedding. When considering the use of direct-contact sentinels, a minimum of 4 to 6 wk of contact time has previously been recommended to allow for adequate exposure;14,15 in our study, direct-contact sentinels began to test positive for P. mirabilis at 1 wk after exposure. In addition, female mice have been recommended to reduce incidence of fight wounds; however, another choice is to use castrated males to avoid unwanted pregnancy. A disadvantage of this sentinel strategy is the large number of mice necessary for direct-contact monitoring.14,15 Our study also showed that the strain of sentinel mice exposed to P. mirabilis–colonized mice by contact or bedding was not predictive of a positive result. This outcome suggests that immunocompromised mice may not be more likely than immunocompromised mice to become colonized via both sentinel strategies and thus, in this regard, are not more efficient as sentinels. However, once known to be colonized, immunocompromised mice do have increased likelihood of testing positive by culture and PCR assay and of having higher copy numbers (as noted earlier); therefore, the detection of infection or colonization in sentinels might be easier if the sentinels are immunocompromised.
Regarding exposure to antibiotic-containing diet, sentinel mice exposed to P. mirabilis–positive SRG mice on TMS diet were 82% less likely to test positive by PCR assay than were SRG mice on 5053 diet. This difference may be due to antibiotic-induced suppression of bacterial burden or shedding (as mentioned earlier) and thus impaired transmission to sentinels, as we hypothesized. Contrary to this explanation, colony mice on TMS diet were 28 times more likely than mice on 5053 diet to culture positive, thereby perhaps suggesting intestinal dysbiosis or antibiotic resistance and resulting in increased bacterial burden or shedding, at least of resistant organisms. The aim of this study to determine the presence of antibiotic resistance and genetic differences of detectable P. mirabilis strains in colonized mice showed that 2 different P. mirabilis strains were present in spontaneously infected colony mice at our institution. The TMS-resistant strain isolated from the SRG mouse on TMS diet was indistinguishable from 3 other P. mirabilis isolates previously identified by MLST at our institution, whereas the TMS-sensitive strain isolated from the GEM on 5053 diet had not previously been identified. However, all immunocompromised and immunocompetent mice—regardless of diet—had isolates that were resistant to TMS. The P. mirabilis strain evaluated in this study was isolated from an immunovague mouse that had no resistance to TMS, suggesting that these inhouse-generated genetically modified mice may have a strain less prevalent or newer than resistant endemic strain(s) circulating in the barrier after generational exposure to TMS. This aspect of the study had notable limitations, in that at our institution, all immunocompromised mice are weaned and maintained on TMS diet and were only moved to 5053 diet for experimental purposes, whereas immunocompetent mice on this study evaluated experimentally by TMS diet were previously on 5053 diet. This study design limitation prevents an accurate assessment of the role of antibiotic diet exposure on the development of antibiotic resistance in P. mirabilis over time, and a follow-up study is imperative. Despite MLST analysis confirming 2 distinct P. mirabilis strains, specific and official sequence-type identification for these strains is unavailable; their allelic profiles could not be compared with those in a large central database29 for precise characterization and potential useful community information, because such data are not available for P. mirabilis. Future studies could include MLST of additional isolates, sequencing strains further to analyze genetic variants associated with P. mirabilis virulence,19 and further characterizing selective pressure on antimicrobial resistance.
This study has shown that immune function is an important factor in both the detection and transmission of P. mirabilis. Other strain differences in genetic background might also contribute. In addition to other supportive results mentioned earlier, immunocompromised SRG mice were 26 times more likely than CD1 mice to have high PCR copy numbers. Overall, both direct-contact and dirty-bedding sentinels exposed to SRG P. mirabilis–positive mice were 18 times more likely to test positive than were sentinels exposed to CD1 P. mirabilis–positive mice; for direct contact specifically, sentinels exposed to SRG mice were 16 times more likely to be positive than were sentinels exposed to CD1 mice. These differences in diagnostics suggest that a difference in colonization dynamics might also influence the likelihood of developing disease. In addition, P. mirabilis–colonized immunocompromised mice on TMS diet developed inflammatory lesions associated with gram negative bacilli. The constellation of pathology seen in this population of mice was consistent with known pathology of P. mirabilis in laboratory mice.30,36,42,44 PCR analysis after fixation showed the presence of P. mirabilis in lesions of one mouse, providing a presumptive correlation with pathology. Future studies should be performed to satisfy Koch’s Postulates and specifically to demonstrate P. mirabilis causes these lesions after pure culture and direct inoculation of immunocompromised and immunocompetent mice.45
Further work to define optimal methods for the detection for opportunistic rodent pathogens is increasingly necessary as the laboratory animal field moves to adopt higher levels of biosecurity and refinement of animal models. Failure to detect such organisms in immunocompromised rodent colonies, including failure of transmission through indirect means, could confound outcomes in research involving the urinary and gastrointestinal tracts and contribute to poor reproducibility of results. Additional studies should assess colony-wide detection and transmission of other opportunistic rodent pathogens that could be endemic and below the threshold for detection in barrier colonies by using standard animal health screening tools. For now, our recommendations regarding animal health-surveillance programs include the use of direct-contact sentinels or direct testing of colony mice, especially in immunocompromised or immunovague genetically modified populations, to best screen for P. mirabilis and the inclusion of culture, PCR, and repeated testing for confidence in exclusion status. Furthermore, additional studies are needed to evaluate alternative testing methods, such as culture of dirty-bedding sentinels and exhaust air dust monitoring, which may also improve detection.
In conclusion, improving the detection of opportunistic agents in rodent colony health monitoring and transmission to sentinels will in turn improve the quality of animal care and research. P. mirabilis–colonized highly immunocompromised mice are more likely than immunocompetent mice to test positive for P. mirabilis by culture and PCR analysis and are therefore better for the detection of this organism in colonies. In addition, as compared with their immunocompetent counterparts, immunocompromised mice transmit P. mirabilis more efficiently via direct contact. Finally, immunocompromised mice on antibiotic-containing diets can develop inflammatory lesions of the urogenital tract and abdominal cavity consistent with known pathology of P. mirabilis, thus potentially confounding research outcomes.
Acknowledgment
We thank Arturo Contreras and Deborah Dimke for assistance with sample collection.
References
- 1.AVMA. [Internet]. 2013. AVMA guidelines for the euthanasia of animals: 2013 ed. [Cited 13 Aug 2021]. Available at: https://www.avma.org/sites/default/files/resources/euthanasia.pdf.
- 2.AVMA. [Internet]. 2020. AVMA guidelines for the euthanasia of animals: 2020 ed. [Cited 13 Aug 2021]. Available at: https://www.avma.org/kb/policies/documents/euthanasia.pdf.
- 3.Adeolu M, Alnajar S, Naushad S, Gupta RS. 2016. Genome-based phylogeny and taxonomy of the ‘Enterobacteriales’: proposal for Enterobacterales ord. nov. divided into the families Enterobacteriaceae, Erwiniaceae fam. nov., Pectobacteriaceae fam. nov., Yersiniaceae fam. nov., Hafniaceae fam. nov., Morganellaceae fam. nov., and Budviciaceae fam. nov. Int J Syst Evol Microbiol 66:5575–5599. 10.1099/ijsem.0.001485. [DOI] [PubMed] [Google Scholar]
- 4.Armbruster CE, Mobley HLT. 2012. Merging mythology and morphology: the multifaceted lifestyle of Proteus mirabilis. Nat Rev Microbiol 10:743–754. 10.1038/nrmicro2890. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Armbruster CE, Mobley HLT, Pearson MM. 2018. Pathogenesis of Proteus mirabilis infection. Ecosal Plus 8:ecosalplus.ESP-0009-2017. 10.1128/ecosalplus.ESP-0009-2017. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Artwohl JE, Cera LM, Wright MF, Medina LV, Kim LJ. 1994. The efficacy of a dirty-bedding sentinel system for detecting Sendai virus infection in mice: a comparison of clinical signs and seroconversion. Lab Anim Sci 44:73–75. [PubMed] [Google Scholar]
- 7.Barthold SW, Griffey SM, Percy DH. 2016. Pathology of laboratory rodents and rabbits, 4th ed. Hoboken (NJ): Wiley-Blackwell. 10.1002/9781118924051 [DOI] [Google Scholar]
- 8.de Bruin WC, van de Ven EE, Hooijans CR. 2016. Efficacy of soiled bedding transfer for transmission of mouse and rat infections to sentinels: a systematic review. PLoS One 11:e0158410. 10.1371/journal.pone.0158410. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Caraguel CGB, Stryhn H, Gagné N, Dohoo IR, Hammell KL. 2011. Selection of a cutoff value for real-time polymerase chain reaction results to fit a diagnostic purpose: analytical and epidemiologic approaches. J Vet Diagn Invest 23:2–15. 10.1177/104063871102300102. [DOI] [PubMed] [Google Scholar]
- 10.Carty AJ. 2008. Opportunistic infections of mice and rats: Jacoby and Lindsey revisited. ILAR J 49:272–276. 10.1093/ilar.49.3.272. [DOI] [PubMed] [Google Scholar]
- 11.Charles River Laboratories International. [Internet]. 2009. Proteus mirabilis technical sheet. [Cited 13 Aug 2021]. Available at: https://www.criver.com/sites/default/files/resources/ProteusmirabilisTechnicalSheet.pdf.
- 12.Choi JG, Kim N, Ju IG, Eo H, Lim S-M, Jang S-E, Kim D-H, Oh MS. 2018. Oral administration of Proteus mirabilis damages dopaminergic neurons and motor functions in mice. 19; 8(1):1275. Sci Rep. 10.1038/s41598-018-19646-x. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Coker C, Poore CA, Li X, Mobley HLT. 2000. Pathogenesis of Proteus mirabilisurinary tract infection. Microbes Infect 2:1497–1505. 10.1016/S1286-4579(00)01304-6. [DOI] [PubMed] [Google Scholar]
- 14.Compton SR, Homberger FR, MacArthur Clark J. 2004. Microbiological monitoring in individually ventilated cage systems. Lab Anim (NY) 33:36–41. 10.1038/laban1104-36. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Compton SR, Homberger FR, Paturzo FX, Clark JM. 2004. Efficacy of three microbiological monitoring methods in a ventilated cage rack. Comp Med 54:382–392. [PubMed] [Google Scholar]
- 16.Cundiff DD, Riley LK, Franklin CL, Hook RR, Jr, Besch-Williford C. 1995. Failure of a soiled bedding sentinel system to detect cilia-associated respiratory bacillus infection in rats. Lab Anim Sci 45:219–221. [PubMed] [Google Scholar]
- 17.Dobson GP, Letson HL, Biros E, Morris J. 2019. Specific pathogen-free (SPF) animal status as a variable in biomedical research: have we come full circle? EBioMedicine 41:42–43. 10.1016/j.ebiom.2019.02.038. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Fehervari Z. 2019. Mechanisms of colonization resistance. Nature Research.
- 19.Filipiak A, Chrapek M, Literacka E, Wawszczak M, Głuszek S, Majchrzak M, Wróbel G, Gniadkowski M, Adamus-Białek W. 2020. Correlation of pathogenic factors with antimicrobial resistance of clinical Proteus mirabilis strains. Cold Spring Harbor Laboratory. 10.1101/2020.02.24.962514 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Fox JG, Taylor Bennett B. 2015. Laboratory animal medicine, 3rd ed. Cambridge (MA): Academic Press. 10.1016/B978-0-12-409527-4.00001-8 [DOI] [Google Scholar]
- 21.Franklin CL. 2006. Microbial considerations in genetically engineered mouse research. ILAR J 47:141–155. 10.1093/ilar.47.2.141. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Garrett WS, Gallini CA, Yatsunenko T, Michaud M, DuBois A, Delaney ML, Punit S, Karlsson M, Bry L, Glickman JN, Gordon JI, Onderdonk AB, Glimcher LH. 2010. Enterobacteriaceae act in concert with the gut microbiota to induce spontaneous and maternally transmitted colitis. Cell Host Microbe 8:292–300. 10.1016/j.chom.2010.08.004. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Grant SS, Hung DT. 2014. Persistent bacterial infections, antibiotic tolerance, and the oxidative stress response. Virulence 4:273–283. 10.4161/viru.23987. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Henderson KS, Clifford CB. [Internet]. 2013. A guide to integrating PCR rodent infectious agent (PRIA) panels in rodent quality control programs. [Cited 13 Aug 2021]. Available at: https://www.cmol.it/wp-content/uploads/2018/03/PCR-Rodent-Infectious-Agent-PRIA.pdf
- 25.Hooton TM, Bradley SF, Cardenas DD, Colgan R, Geerlings SE, Rice JC, Saint S, Schaeffer AJ, Tambayh PA, Tenke P, Nicolle LE. 2010. Diagnosis, prevention, and treatment of catheter-associated urinary tract infection in adults: 2009 international clinical practice guidelines from the Infectious Diseases Society of America. Clin Infect Dis 50:625–663. 10.1086/650482. [DOI] [PubMed] [Google Scholar]
- 26.Institute for Laboratory Animal R. 2010. Guide for the care and use of laboratory animals, 8th ed. Washington (DC): National Academies Press. [Google Scholar]
- 27.Jacoby RO, Fox JG, Davisson M. 2002. Biology and diseases of mice, 35-120. In Fox JG, Loew FM, Anderson LC, Quimby FW, editors. Laboratory animal medicine. Cambridge (MA): Academic Press. [Google Scholar]
- 28.Jakobsen TH, Xu Y, Bay L, Schønheyder HC, Jakobsen T, Bjarnsholt T, Thomsen TR. 2021. Sampling challenges in diagnosis of chronic bacterial infections. J Med Microbiol 70. 10.1099/jmm.0.001302. [DOI] [PubMed] [Google Scholar]
- 29.Jolley KA, Bray JE, Maiden MCJ. 2018. Open-access bacterial population genomics: BIGSdb software, the PubMLST.org website and their applications. Wellcome Open Res 3:124. Available at https://wellcomeopenresearch.org/articles/3-124/v1. https://doi.org/10.12688/wellcomeopenres.14826.1. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Jones JB, Estes PC, Jordan AE. 1972. Proteus mirabilis infection in a mouse colony. J Am Vet Med Assoc 161:661–664. [PubMed] [Google Scholar]
- 31.Kim S, Covington A, Pamer EG. 2017. The intestinal microbiota: antibiotics, colonization resistance, and enteric pathogens. Immunol Rev 279:90–105. 10.1111/imr.12563. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Lane-Petter W. 1962. The provision and use of pathogen-free laboratory animals. Proc R Soc Med 55:253–256. 10.1177/003591576205500402. [DOI] [PubMed] [Google Scholar]
- 33.Lidster K, Owen K, Browne WJ, Prescott MJ. 2019. Cage aggression in group-housed laboratory male mice: an international data crowdsourcing project. Sci Rep 9:15211. 10.1038/s41598-019-51674-z. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.Lipman NS, Homberger FR. 2003. Rodent quality assurance testing: use of sentinel animal systems. Lab Anim (NY) 32:36–43. 10.1038/laban0503-36. [DOI] [PubMed] [Google Scholar]
- 35.Mähler Convenor M, Berard M, Feinstein R, Gallagher A, Illgen-Wilcke B, Pritchett-Corning K, Raspa M. 2014. FELASA recommendations for the health monitoring of mouse, rat, hamster, guinea pig and rabbit colonies in breeding and experimental units. Lab Anim 48:178–192. 10.1177/0023677213516312. [DOI] [PubMed] [Google Scholar]
- 36.Maronpot RR, Peterson LG. 1981. Spontaneous proteus nephritis among male C3H/HeJ mice. Lab Anim Sci 31:697–700. [PubMed] [Google Scholar]
- 37.Marx JO, Gaertner DJ, Smith AL. 2017. Results of survey regarding prevalence of adventitial infections in mice and rats at biomedical research facilities. J Am Assoc Lab Anim Sci 56:527–533. [PMC free article] [PubMed] [Google Scholar]
- 38.Masopust D, Sivula CP, Jameson SC. 2017. Of mice, dirty mice, and men: using mice to understand human immunology. J Immunol 199:383–388. 10.4049/jimmunol.1700453. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39.Mathur S, Sabbuba NA, Suller MTE, Stickler DJ, Feneley RCL. 2005. Genotyping of urinary and fecal Proteus mirabilis isolates from individuals with long-term urinary catheters. Eur J Clin Microbiol Infect Dis 24:643–644. 10.1007/s10096-005-0003-0. [DOI] [PubMed] [Google Scholar]
- 40.Nicklas W, Baneux P, Boot R, Decelle T, Deeny AA, Fumanelli M, Illgen-Wilcke B. 2002. Recommendations for the health monitoring of rodent and rabbit colonies in breeding and experimental units. Lab Anim 36:20–42. 10.1258/0023677021911740. [DOI] [PubMed] [Google Scholar]
- 41.Percy DH. 2007. Pathology of laboratory rodents and rabbits. Ames (IA): Blackwell Publishing. 10.1002/9780470344613 [DOI] [Google Scholar]
- 42.Percy DH, Barta JR. 1993. Spontaneous and experimental infections in scid and scid/beige mice. Lab Anim Sci 43:127–132. [PubMed] [Google Scholar]
- 43.Perkins CLMP, Parkinson C, Weagle J, Morin A, Henderson KS. 2017. Evaluation of PCR and culture methods for the detection of opportunistic bacteria in barrier-reared colonies. Abstracts Presented in the 68th AALAS National Meeting, Austin, Texas. 15–19 October 2017. J Am Assoc Lab Anim Sci 56:639. [Google Scholar]
- 44.Scott RA, Croy BA, Percy DH. 1991. Diagnostic exercise: hepatitis in SCID-beige mice. Lab Anim Sci 41:166–168. [PubMed] [Google Scholar]
- 45.Segre JA. 2013. What does it take to satisfy Koch’s postulates two centuries later? Microbial genomics and Propionibacteria acnes. J Invest Dermatol 133:2141–2142. 10.1038/jid.2013.260. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 46.Taylor DM. 1988. A shift from acute to chronic spontaneous pyelonephritis in male MM mice associated with a change in the causal micro-organisms. Lab Anim 22:27–34. 10.1258/002367788780746557. [DOI] [PubMed] [Google Scholar]
- 47.Treuting PM, Clifford CB, Sellers RS, Brayton CF. 2011. Of mice and microflora: considerations for genetically engineered mice. Vet Pathol 49:44–63. 10.1177/0300985811431446. [DOI] [PubMed] [Google Scholar]
- 48.Whary MT, Baumgarth N, Fox JG, Barthold SW. 2015. Biology and diseases of mice, p 43–149. In: Fox JG, editor. Laboratory animal medicine, third ed. Cambridge (MA): Academic Press. [Google Scholar]
- 49.White WJ, Anderson LC, Geistfeld J, Martin DG. 1998. Current strategies for controlling/eliminating opportunistic microorganisms. ILAR J 39:291–305. 10.1093/ilar.39.4.291. [DOI] [PubMed] [Google Scholar]
- 50.Zhang W, Niu Z, Yin K, Liu P, Chen L. 2012. Quick identification and quantification of Proteus mirabilis by polymerase chain reaction (PCR) assays. Ann Microbiol 63:683–689. 10.1007/s13213-012-0520-x. [DOI] [Google Scholar]