Cohousing with Dirty Mice Increases the Frequency of Memory T Cells and Has Variable Effects on Intracellular Bacterial Infection

The immunological landscape of nonlymphoid tissues from mice housed in specific pathogen–free (SPF) conditions resembles that of a neonate human (1). These naive tissues are devoid of most lymphocytes, including memory T cells (1), reflecting the fact that naive lymphocytes largely restrict their movement to lymphoid tissues and blood before activation (2). In marked contrast, nonlymphoid tissue samples from pet shop mice, previously infected SPF mice, or adult humans are usually replete with memory lymphocytes generated during prior immunological events, such as past infections (1, 3–5). These memory lymphocyte populations are capable of accelerating immune protection against pathogens in an Ag nonspecific manner (1). A mouse model has been established that allows investigation of this nonspecific protection where genetically identical SPF mice share housing for a period of time with pet shop or feral mice, allowing laboratory mice to be exposed to a “dirty” environment and generate a more “experienced” immune system (1, 6). Previous studies show that cohousing of laboratory mice with dirty mice caused an increase in memory lymphocyte populations and enhanced control of systemic infection with Listeria monocytogenes and Plasmodium berghei (1, 6). However, it is not yet clear whether similar protection extends to other infections, especially intracellular bacteria that require clearance by CD4 T cells (7, 8). Our laboratory has previously shown that non-cognate activation of Ag-experienced CD4 T cells enhanced clearance of intracellular bacteria (9, 10), suggesting that effector memory CD4 T cells circulating through blood and tissues might potentially contribute to early nonspecific defense. Although the mechanism of this non-cognate activation of CD4 T cell remains under investigation, it is clear that a variety of inflammatory cytokines provoke the production of IFN-γ to restrict growth of intracellular pathogens (11). Thus, the increase in memory CD4 T cells observed in an experienced immune system might potentially provide substantial nonspecific protection against intramacrophage pathogens.

Chlamydia trachomatis is an obligate intracellular Gram-negative bacterium that causes a local infection of the upper reproductive tract in humans and mice (12). The adult human female reproductive tract (FRT) is replete with lymphocytes and contains memory lymphoid structures that might enhance clearance of genital tract infections such as C. trachomatis (1, 13, 14). In the current study, we sought to determine whether increasing the immunological experience of laboratory mice (SPF) would generate a mouse model that was more relevant to human Chlamydia infection. To answer this question, we examined infection with mouse-adapted Chlamydia muridarum, a model that is commonly used to study Chlamydia infection of the genital tract (15–17). Previous work from our laboratory has demonstrated that increased numbers of nonspecific memory T cells in CCR7-deficient mice correlated with enhanced resistance to C. muridarum infection (18). Therefore, we hypothesized that increasing nonspecific memory lymphocyte populations in SPF mice using a cohousing strategy would similarly enhance clearance of C. muridarum. In this study, we report that cohousing of dirty mice with SPF laboratory mice increased the frequency of effector memory T cells (T_EMs) in laboratory mice and enhanced protection against systemic Listeria infection. However, we did not detect any alteration in the course of genital Chlamydia infection or systemic Salmonella infection, demonstrating that this cohousing approach has variable effects on the clearance of intracellular pathogens.

C57BL/6 6-wk-old mice were purchased from The Jackson Laboratory and housed in SPF or conventional conditions. Dirty mice were purchased from Kasch’s Kritters (Citrus Heights, CA), Petco (Davis, CA), Pet Supplies Plus (Woodland, CA), or were caught in traps at the University of California Davis. Laboratory mice were conventionally housed with dirty mice at a 3:1 ratio for ∼2 mo before analysis. All experimental time points mentioned therefore relate to the beginning of this cohousing experience. Cohoused C57BL/6 mice were age-matched with regular SPF mice in each experiment. All animal procedures were approved by the Institutional Animal Care and Use Committee at the University of California Davis.

Salmonella enterica serovar Typhimurium (S. Typhimurium; BRD509 aroA mutant) was injected i.v. to induce systemic bacterial infection in mice, as previously described (19). S. Typhimurium was grown in Luria-Bertani broth at 37°C overnight and 5 × 10⁵ CFU were injected i.v. in a 200-μl vol of PBS. Listeria monocytogenes strain LM10403s was injected i.v. to induce systemic bacterial infection in mice. L. monocytogenes was grown in brain-heart infusion media at 37°C overnight to an OD₆₀₀ of 0.8 and 4.5 × 10⁴ CFU were injected in 200 μl of PBS. C. muridarum was purchased from American Type Culture Collection and propagated in HeLa cells as previously described (20). Chlamydia infections were delivered at a dose of 10⁵ inclusion-forming units (IFU) in 5 μl of SPG (sucrose–phosphate–glutamic acid) buffer intravaginally to induce an upper reproductive tract infection.

Spleens and livers were harvested from S. Typhimurium–infected mice and homogenized in a known quantity of PBS before serial dilutions were plated on MacConkey agar plates and incubated overnight at 37°C. The following day, bacteria were enumerated on plates and the total CFU/organ were calculated. For Listeria-infected mice, bacteria were similarly enumerated in the liver of infected mice using brain-heart infusion plates. For Chlamydia infection experiments, mice were swabbed at regular intervals every 3–4 d and calcium alginate swabs were agitated in 500 μl of SPG buffer with two glass beads. These supernatants were frozen at −80°C until the end of experiment for analysis. Frozen swabs from all time points were thawed, serially diluted, and plated on a HeLa cell monolayer, centrifuged for 1 h at 37°C, and incubated for ∼16–20 h in media and cycloheximide. After this incubation period, cells were washed and fixed with 100% methanol for 15 min. After fixation, HeLa cells were stained with Chlamydia hyperimmune serum for 45–60 min, washed, and then incubated with secondary FITC-conjugated goat anti-mouse IgG Ab and washed again. Chlamydia inclusions were enumerated under a fluorescence microscope and the total IFU/swab were calculated, as previously described (21).

Blood was collected by cheek bleed 1 and 2 mo after the start of cohousing, ACK (ammonium-chloride-potassium) lysed, and washed with FACS buffer. Spleens were harvested and homogenized through a filter into a single-cell suspension, ACK lysed, and washed with FACs buffer. Cell suspensions were incubated in Fc block for 10–30 min and then further washed before staining. Cells were stained using Abs resuspended in Fc block for 30 min at room temperature. Abs used in the experiments included: CD11b PE, CD11c PE, B220 PE, F4/80 PE, CD8 PerCP eFluor 710, CD4 FITC, CD44 PE-Cy7, CD4 allophycocyanin, and CD62L FITC. All stained cells were subsequently fixed and samples analyzed using a BD LSRFortessa flow cytometer.

All statistical analyses were conducted using GraphPad Prism version 8.

Dirty mice for cohousing were obtained from a local pet shop vendor and assessed for pathogen exposure. Serological, culture, parasitology, and PCR analyses of up to nine pet shop mice uncovered a wide variety of pathogen exposures in this cohort of animals (Supplemental Table I). Previous work had noted an increased frequency of memory T cells in mice previously exposed to a non-SPF environment (1). Thus, we initially examined the frequency of CD44⁺ memory T cells in “pet shop mice,” compared with inbred C57BL/6 mice housed in our SPF facilities. In the blood of pet shop mice, ∼10% of CD4 T cells expressed high levels of CD44, compared with 5% of CD4 T cells in the blood of SPF C57BL/6 mice (Fig. 1). Similarly, the spleen of pet shop mice contained ∼30% CD4⁺CD44⁺ cells whereas SPF mouse spleens had <10% CD4⁺CD44⁺ cells (Fig. 1). Surprisingly, there were no significant differences noted in the frequency of CD8⁺CD44⁺ T cells in the blood and spleen of this particular batch of pet shop mice when compared with SPF mice (Fig. 1). Thus, consistent with a previous report, mice obtained from a pet shop displayed a higher frequency of memory T cells in circulation, and this correlates with prior pathogen exposure.

Next, we examined whether close contact between these two groups of mice would be sufficient to modify the memory compartment of SPF C57BL/6 mice. We cohoused female SPF mice with female pet shop mice in a 3:1 ratio for 2 mo to provide sufficient time to normalize environmental factors. Blood was sampled from C57BL/6 mice at 1- and 2-mo time points after cohousing and analyzed by flow cytometry to determine CD4 and CD8 T_EM (CD44^hiCD62L⁻) frequencies. After 1 mo of cohousing, a substantial increase in CD4 and CD8 T_EMs was observed in the blood of cohoused SPF mice, although this was still substantially lower than for pet shop mice (Fig. 2, left panel). After 2 mo of cohousing the frequency of CD4 and CD8 T_EMs in cohoused C57BL/6 mice remained higher than in SPF C57BL/6 mice (Fig. 2, right panel). Thus, cohousing with locally sourced pet shop mice is sufficient to increase the percentage of T_EMs in the blood of SPF laboratory mice, as previously reported (1).

Although cohousing with pet shop mice increased the frequency of memory T cells in laboratory mice, the magnitude of this effect appeared somewhat lower than that found a previous report (1). When considering possible variables, we first examined whether the type of mouse used for cohousing was a factor in this process. The capacity of feeder (snake food) mice, non-feeder (fancy) mice, or feral mice (captured locally) on memory T cell transition after cohousing was examined. Female feeder, fancy, and feral mice were individually cohoused with female SPF laboratory mice for 2 mo and assessed for memory T cell frequency. Interestingly, cohousing C57BL/6 mice with feral mice only caused a modest increase in CD4 T_EMs in the blood, perhaps indicating that this particular cohort of feral mice had fewer transmissible agents. In contrast, both feeder and fancy mice induced greater increases in T_EM frequency across CD4 and CD8 T cell populations (Fig. 3). The CD4 T_EM frequency was comparable in laboratory mice cohoused with either source of pet shop mice, resulting in a memory frequency of ∼20–25% (Figs. 2, 3). However, note that none of the dirty mice used in our study was able to induce T_EM population changes of the magnitude that was reported previously (1), perhaps reflecting geographical differences in the animal sources or other uncontrolled variables. Magnitude aside, the basic pattern of changes observed in our study is consistent with the idea that cohousing of SPF mice with dirty mice causes expansion of memory T cell populations in laboratory mice.

It was previously shown that cohousing SPF mice with pet shop mice enhanced nonspecific immunity against systemic L. monocytogenes and P. berghei infection, two infections that predominantly depend on CD8 T cells for clearance (22–24). To confirm that our cohousing model was similarly protective, after 2 mo of cohousing, we challenged cohoused laboratory mice i.v. with 4.5 × 10⁴ CFU of L. monocytogenes and examined bacterial burdens. C57BL/6 mice that had been cohoused with pet store mice displayed a significantly lower burden of Listeria in the liver, compared with age-matched non-cohoused mice (Fig. 4A). Again, these data are consistent with a previous study showing that cohousing with dirty mice enhances immunity to Listeria challenge (1).

We next tested whether this observation of nonspecific protection extended to an intramacrophage pathogen, S. Typhimurium, that requires CD4 T cells for clearance (9, 25). Cohoused and non-cohoused C57BL/6 mice were infected with 5 × 10⁵ CFU of S. Typhimurium and euthanized at 1-wk intervals during the course of infection. Bacterial burdens were assessed in the spleen and liver of infected mice because these are the primary sites of systemic Salmonella replication (26). However, no difference in bacterial loads was detected in the spleen or liver of cohoused or non-cohoused mice at any of the time points assessed (Fig. 4B). Thus, an increase in CD4 T_EMs through the process of cohousing with dirty mice did not substantially affect the outcome of systemic S. Typhimurium infection in laboratory mice.

Although cohousing with dirty mice did not affect the replication of systemic CD4-dependent pathogen, it was of interest to determine whether there would be a greater effect on Chlamydia infection, as pathogen replication is largely localized to the FRT (27, 28). Laboratory mice were cohoused with feral or pet shop mice for 2 mo before intravaginal infection with 10⁵ IFU of C. muridarum. Infected mice were vaginally swabbed every 3–4 d over the month-long infection period, and no significant differences were detected in bacterial shedding between these mice and non-cohoused age-matched cohorts (Fig. 5). We reasoned that this lack of effect on an FRT infection could be due to the inability to transfer sexually transmitted infections during the cohousing period, as all cohoused mice were female (Fig. 6). To examine this particular variable, we cohoused female laboratory mice with male or female pet shop mice for 2 mo and all mice were rested for an additional month to ensure that none of the animals was pregnant before challenge infection. However, no significant differences were detected in the course of Chlamydia infection between these differently cohoused mice. Thus, prior cohousing with male mice did not substantially alter the course of Chlamydia infection in female laboratory mice. Taken together, the cohousing of laboratory mice with pet shop or feral mice expands the CD4 memory T cell pool but does not affect the replication of Chlamydia in the FRT.

Given the large number of uncontrolled variables during mouse studies, experimental approaches to limit their influence on complex mechanistic questions are often implemented, including age and sex matching, inbreeding, diet and environmental standardization, and SPF housing. Although these operating practices are designed to limit the effect of uncontrolled variables on experimental studies, they also have the potential to introduce new variables that complicate data interpretation. SPF housing controls were developed to limit animal exposure to unknown infectious agents and thus enhance reproducibility, but it has only recently been appreciated that SPF mice do not develop an immune system that is comparable to an adult human or wild mouse (1, 3). This is evident by a low number of memory T cells and lack of ectopic lymphoid structures, when compared with wild mice and adult humans (1, 13, 14, 18, 29, 30). Diminished memory lymphocyte frequencies in SPF laboratory mice are likely due to minimal exposure to the typical pathogens or microbiota that would commonly activate and develop a normal lymphocyte repertoire (1). Because the study of outbred wild mice is challenging using current reagents, a new animal model has been developed that provides genetically controlled inbred mice with appropriate maturation of the immune system (1, 6, 31). This model involves cohousing of SPF laboratory mice with feral or pet store mice that typically have murine pathogens, and are often referred to as dirty mice (1, 32). Cohousing with dirty mice therefore provides laboratory mice with limited exposure to a range of typical pathogens, which promotes the development of an Ag-experienced immune system that is arguably more reflective of an adult human (1, 32). Indeed, this exposure period has been shown to increase the overall frequency of memory lymphocytes in mice and confer nonspecific protection against systemic pathogens (1, 6, 31).

The FRT is a mucosal tissue exposed to the outside environment that frequently comes into contact with pathogens. Indeed, sexually transmitted infections such as Chlamydia trachomatis are common among young women (33), but there is limited understanding of why some women control this infection better than others (34). Human studies of Chlamydia infection in women show the existence of memory lymphocyte clusters in the genital tract that are thought to play a role in fighting Chlamydia infections (30). This hypothesis is supported by a recent study demonstrating that CCR7-deficient mice display increased nonspecific memory CD4 T cells and display increased capacity to control Chlamydia infections, compared with wild-type mice (18). Given this observation and the fact that CD4 T cells are required for primary clearance of Chlamydia infection (27, 28), we hypothesized that boosting of memory CD4 T cell frequencies in laboratory mice via cohousing could enhance protection against Chlamydia infection.

Indeed, we have been able to confirm previous results that cohousing induces enhanced frequencies of circulating CD4 and CD8 memory T cells in laboratory mice. This cohousing approach allowed enhanced control of L. monocytogenes infection in laboratory mice, as previously reported (1). Thus, our study confirms that cohousing laboratory mice with dirty mice allows maturation of T cell memory responses and correlates with natural resistance to Listeria infection. However, the magnitude of memory cell generation in our cohousing experiments is somewhat lower than previously reported (1), and similarly Listeria clearance was much less robust. Although there could be many reasons for these differences, it would seem most likely that the magnitude of change is driven by environmental differences in the source of dirty mice used for cohousing. Although it may be tempting to speculate that a pet shop mouse from Sacramento has fewer pathogens, it is also possible that the opposite is true. Indeed, during our study we noted that many cohoused C57BL/6 mice displayed lack of grooming and sluggishness, suggesting overt pathogen exposure. Although a previous study also documented similar illness, mice in this study fully recovered after 6 wk (1, 32), whereas these effects persisted in our cohorts. Analysis of common pathogens present in pet shop mice from Sacramento did not uncover obvious differences to those previously reported from Minnesota. Furthermore, the use of feeder or fancy mice was not a primary driver of obvious differences in our study, although it was notable that feral mice were less able to drive enhanced memory development. It would clearly be of interest to directly compare pet shop mice derived from these different geographical locations head to head; however, restrictions in shipping these animals between facilities precludes such investigation. However, it would seem likely that the quantity or quality of pathogens or the microbiota between these two different animal sources is the driver of differential effects. Although we are currently unable to identify the source of this variability, it would be important to define this in more detail before embarking on long-term studies of environmental effects on memory responses.

The main thrust of this particular investigation was to determine whether pet shop cohousing could enhance protective responses against a bacterial infection that primarily requires CD4 T cells for clearance. However, we found no effect of prior cohousing on the course of systemic S. Typhimurium or localized C. muridarum infection. The absence of effects on Chlamydia FRT infection was independent of whether female laboratory mice were previously housed with female or male pet shop mice. Overall, we did not detect any major influence of prior cohousing on the protective immune response to two CD4 T cell–dependent pathogens. Although it remains possible that such effects would emerge if there had been a greater effect on memory cell formation during our cohousing, the fact that we could replicate protective effects on Listeria infection makes it likely that CD4-dependent pathogens are simply less affected by nonspecific memory expansion.

Overall, it is hard to deny that using a pet shop cohousing model is a useful way to model human disease because the maturation of the immune system more accurately reflects the condition of an adult human. However, our experience is that this model displays considerable heterogeneity in driving memory formation, most likely due to the source of dirty mice causing “normalization.” Furthermore, this limited study also suggests that the pre-exposure memory pool may have differential effects on pathogens that require CD8 versus CD4 T cells for clearance. Efforts to standardize systems across different laboratories and examine a wider variety of potential pathogens would be a prudent strategy.

We thank the Bales and Foley laboratories at the University of California Davis for assistance in acquiring feral mice. We also acknowledge the openness, invaluable assistance, and helpful advice of the members of the Masopust and Jameson laboratories at the University of Minnesota when we initiated these studies at the University of California Davis.

The authors have no financial conflicts of interest.