Comparison of mouse models of microbial experience reveals differences in microbial diversity and response to vaccination.

Specific pathogen-free (SPF) laboratory mice dominate preclinical studies for immunology and vaccinology. Unfortunately, SPF mice often fail to accurately model human responses to vaccination and other immunological perturbations. Several groups have taken different approaches to introduce additional microbial experience to SPF mice to better model human immune experience. How these different models compare is unknown. Here, we directly compare three models: housing SPF mice in a microbe-rich barn-like environment (feralizing), adding wild-caught mice to the barn-like environment (fer-cohoused), or cohousing SPF mice with pet store mice in a barrier facility (pet-cohoused); the two latter representing different murine sources of microbial transmission. Pet-cohousing mice resulted in the greatest microbial exposure. Feralizing alone did not result in the transmission of any pathogens tested, while fer-cohousing resulted in the transmission of several picornaviruses. Murine astrovirus 2, the most common pathogen from pet store mice, was absent from the other two model systems. Previously, we had shown that pet-cohousing reduced the antibody response to vaccination compared with SPF mice. This was not recapitulated in either the feralized or fer-cohoused mice. These data indicate that not all dirty mouse models are equivalent in either microbial experience or immune responses to vaccination. These disparities suggest that more cross model comparisons are needed but also represent opportunities to uncover microbe combination-specific phenotypes and develop more refined experimental models. Given the breadth of microbes encountered by humans across the globe, multiple model systems may be needed to accurately recapitulate heterogenous human immune responses.IMPORTANCEAnimal models are an essential tool for evaluating clinical interventions. Unfortunately, they can often fail to accurately predict outcomes when translated into humans. This failure is due in part to a lack of natural infections experienced by most laboratory animals. To improve the mouse model, we and others have exposed laboratory mice to microbes they would experience in the wild. Although these models have been growing in popularity, these different models have not been specifically compared. Here, we directly compare how three different models of microbial experience impact the immune response to influenza vaccination. We find that these models are not the same and that the degree of microbial exposure affects the magnitude of the response to vaccination. These results provide an opportunity for the field to continue comparing and contrasting these systems to determine which models best recapitulate different aspects of the human condition.

Natural rodent model of viral transmission reveals biological features of virus population dynamics.

Emerging viruses threaten global health, but few experimental models can characterize the virus and host factors necessary for within- and cross-species transmission. Here, we leverage a model whereby pet store mice or rats-which harbor natural rodent pathogens-are cohoused with laboratory mice. This "dirty" mouse model offers a platform for studying acute transmission of viruses between and within hosts via natural mechanisms. We identified numerous viruses and other microbial species that transmit to cohoused mice, including prospective new members of the Coronaviridae, Astroviridae, Picornaviridae, and Narnaviridae families, and uncovered pathogen interactions that promote or prevent virus transmission. We also evaluated transmission dynamics of murine astroviruses during transmission and spread within a new host. Finally, by cohousing our laboratory mice with the bedding of pet store rats, we identified cross-species transmission of a rat astrovirus. Overall, this model system allows for the analysis of transmission of natural rodent viruses and is a platform to further characterize barriers to zoonosis.

The Placental Response to Guinea Pig Cytomegalovirus Depends Upon the Timing of Maternal Infection.

Human cytomegalovirus (HCMV) infects the placenta, and these placental infections can cause fetal injury and/or demise. The timing of maternal HCMV infection during pregnancy is a determinant of fetal outcomes, but how development affects the placenta's susceptibility to infection, the likelihood of placental injury post-infection, and the frequency of transplacental HCMV transmission remains unclear. In this study, guinea pig cytomegalovirus (GPCMV) was used to model primary maternal infection and compare the effects of infection at two different times on the placenta. When guinea pigs were infected with GPCMV at either 21- or 35-days gestation (dGA), maternal and placental viral loads, as determined by droplet digital PCR, were not significantly affected by the timing of maternal infection. However, when the transcriptomes of gestational age-matched GPCMV-infected and control placentas were compared, significant infection-associated changes in gene expression were only observed after maternal infection at 35 dGA. Notably, transcripts associated with immune activation (e.g. Cxcl10, Ido1, Tgtp1, and Tlr8) were upregulated in the infected placenta. A GPCMV-specific in situ hybridization assay detected rare infected cells in the main placenta after maternal infection at either time, and maternal infection at 35 dGA also caused large areas of GPCMV-infected cells in the junctional zone. As GPCMV infection after mid-gestation is known to cause high rates of stillbirth and/or fetal growth restriction, our results suggest that the placenta becomes sensitized to infection-associated injury late in gestation, conferring an increased risk of adverse pregnancy outcomes after cytomegalovirus infection.

Cytomegalovirus infection elicits a conserved chemokine response from human and guinea pig amnion cells.

Human cytomegalovirus (HCMV) infects the chorioamnion, but whether these infections cause fetal membrane dysfunction remains poorly understood. We sought to assess whether guinea pig cytomegalovirus (GPCMV) infects amnion-derived cells in vitro, compare the inflammatory response of amnion cells to GPCMV and HCMV, and determine if GPCMV infects the amnion in vivo. We found that GPCMV replicates in primary guinea pig amnion derived cells and HPV16 E6/E7-transduced amniotic epithelial cells (AEC[E6/E7]s). HCMV and GPCMV infection of amnion cells increased the transcription of the chemokines CCL5/Ccl5, CXCL8/Cxcl8, and CXCL10/Cxcl10. Myd88-knockdown decreased Ccl5 and Cxc8 transcription in GPCMV-infected AEC[E6/E7]s. GPCMV was detected in the guinea pig amnion after primary maternal infection, revealing that guinea pigs are an appropriate model to study fetal membrane physiology after cytomegalovirus infection. As inflammation is known to cause fetal membrane weakening, the amnion's response to cytomegalovirus infection may cause preterm birth and other adverse pregnancy outcomes.

Assessing Zika virus replication and the development of Zika-specific antibodies after a mid-gestation viral challenge in guinea pigs.

Primary Zika virus (ZIKV) infections that occur during pregnancy can cause spontaneous abortion and profoundly disrupt fetal development. While the full range of developmental abnormalities associated with congenital Zika syndrome is not yet known, severe cases of the syndrome can present with microcephaly, extensive neurologic and ocular damage, and pronounced joint malformations. Animal models that accurately recapitulate congenital Zika syndrome are urgently needed for vaccine development and for the study of ZIKV pathogenesis. As guinea pigs have successfully been used to model transplacental infections by cytomegalovirus, syphilis, and Listeria monocytogenes, we sought to test whether ZIKV could productively infect guinea pigs and whether viral transmission with attendant fetal pathology would occur after a mid-gestation viral challenge. We found that guinea pig cells supported ZIKV replication in vitro. Experimental infection of non-pregnant animals did not result in overt disease but low-level, detectable viremia was observed. When pregnant guinea pigs were challenged with ZIKV at between 18 and 21 days gestational age, ZIKV was not detected in maternal or pup blood, plasma, or tissues and no significant differences in maternal weight gain or pup size were observed following challenge. Nonetheless, a robust antibody response against ZIKV was detected in both the pups and dams. These results suggest that, while guinea pigs can model aspects of the immune response to ZIKV infection during pregnancy, naturally circulating ZIKV strains are not pathogenic during the pregnancy of immunocompetent guinea pigs and do not interfere with normal pup development.