The development of a specific pathogen free (SPF) barrier colony of marmosets (Callithrix jacchus) for aging research

Establishment of SPF marmosets

Marmosets (28 males, 31 females) from Southwest National Primate Research Center (SNPRC) were screened for pathogens to identify potential founders for the SPF colony. The presence of viremia was analyzed for GB virus-A (GBV) and Callitrichine Herpes virus-3 (CHV) in 50 selected individuals. Due to the characteristics of the two viruses processing of blood samples differed prior to assay. CHV is a DNA virus related to the Epstein Barr Virus, a herpes virus. The virus is primarily present in B lymphocytes, so DNA was purified from a total blood cell pellet for PCR. GBV-A is an RNA virus distantly related to hepatitis C virus, a flavivirus, and was detected in serum derived RNA by RT-PCR. The overall prevalence for males and females was approximately the same for both viruses. The small sample size did not allow an estimation of whether the infection rate changed with age. The prevalence for CHV in a group of 50 animals ranging from 1-4 years of age was 60% with 20 negative animals (Table 1A). The same group had a prevalence of 36% for GBV-A with 32 negative animals. Repeat screening of 12 animals resulted in two discordant values for CHV. One detected the conversion of the animal to positive over a 19-month period and continued confirmation of that positive result on subsequent bleeds. Within the group of 50 animals screened, 16 animals were double negative and suitable for selection for the barrier colony. Animals were also screened for the presence of fecal pathogens including Giardia, Cryptosporidium and Clostridium perfringens and the frequency of positive screens are shown in Table 1B. Six females and six males were determined to be negative for all viruses and fecal pathogens of interest and were moved to the Texas Research Park to find the SPF barrier colony. An assessment for viral presence 12 months after the founding of the barrier colony revealed one male to be positive for CHV and he was moved back to SNPRC. No other animals were noted to be positive for these viruses of interest at that time, or when screened 6 years after founding the colony. Animals from the barrier colony were screened for fecal pathogens annually in years three, four and five after forming the barrier colony; the rates of positive screens are in Table 1B. Age and sex matched animals from the conventional colony were screened for fecal pathogens as well and the rates of positive screens are also in Table 1B. The prevalence of Cryptosporidium and Clostridium perfringens were very low in both the conventional and the barrier colony. Giardia was more prevalent than the other fecal pathogens, though prevalence of Giardia remained significantly lower in the barrier than in the conventional colony (F (1,8) =5.921, p=0.041).

Colony growth

The colony began with a single breeding pair that was imported in July 2007, followed by three pairs in July 2008 and four more individuals between August 2009 and April 2010. These founding six breeding pairs produced 36 litters. Breeding pairs of unrelated individuals were formed when the animals were at least two years old and between December 2007 and August 2016 the SPF barrier colony produced 98 litters consisting of 275 infants from 27 breeding females and 23 breeding males. The litters in the colony consisted of four aborted pregnancies of unknown litter size, three singletons, 20 twin, 53 triplet, 17 quadruplet and one quintuplet litter. The reproductive outcomes are summarized in Table 2A, B. Overall the reproduction in the barrier colony was comparable to that reported for other colonies including the conventionally housed SNPRC colony (Table 2A, B). On average the barrier colony produced two surviving infants per female per year while the SNPRC colony produces 1.6-1.8 infants per female per year. The use of rotational hand rearing in the barrier, which consisted of the rotational removal of a single infant from the family group for daily supplementation and return to the family each day, led to higher survival rates amongst triplet litters than those litters in which it wasn’t implemented. For triplet litters in which all infants were born alive, rotational rearing resulted in 83.3% of infants surviving through the first 14 days, while not implementing rotational rearing resulted in 50.6% of infants surviving. If litters that were rotationally reared were removed from the analysis 97.1% of infants that survived to 14 days in the barrier colony survive to six months, 97.1% to 12 months, and 97.1% to 18 months meaning there was no loss of infants between the ages of six months and 18 months in the SPF colony. In comparison to infants at SNPRC that were not rotationally reared, lived to 14 days, and were censored for experimental deaths, 89.2% survived to six months, 81% to 12 months, and 68.4% survived to 18 months.

Longevity in barrier

Maintaining marmosets in barrier conditions significantly increased the survival of animals, and adult survivorship plateaued in both males and females when compared to marmosets maintained in conventional housing protocols at SNPRC (Figure 1). We compared survival in a population of 247 barrier animals (115 females and 132 males) versus a population of 370 conventionally housed animals (173 females and 197 males) over a 10-year period. Barrier females had a median lifespan of 8.9 y (95% CI 5.9+ y, upper confidence limit could not be determined due to censoring) compared to a median lifespan of 4.86 y (95% CI 4.06 to 6.38 y) at the primate center with a hazard ratio relative to barrier females of 2.8 (p = 0.001). The female barrier survivorship curve displays a predicted precipitous decline at 9 y which we believe was due to the fact that the longest lived females died, while many middle-aged females were still alive and censored. We remain confident that a difference between the colonies exists based on the Cox proportional hazard analysis and the differences in the median lifespan and their confidence limits. Further, barrier males had a median lifespan of 8.64 y (95% CI 7.6+ y, upper limit could not be determined due to censoring) compared to the 5.08 y (95% CI 4.33 to 7.61 y) at SNPRC with a hazard ratio relative to barrier of 3.44 (p =0.001). Because the populations were tracked for no more than 10 years, there is not a valid lifespan comparison of the SPF colony data with that from the CLEA colony in Japan (Nishijima et al.,2012). However, the early to mid-adulthood survival percentages suggest that the barrier and CLEA populations are similar (Table 3).

Figure 1. (A) Probability of survival for female marmosets in the Barshop SPF barrier colony and the conventionally housed SNPRC colony. (B) Probability of survival for males.

Health outcome

There were significant differences in the blood cell counts of barrier and SNPRC animals, including white blood cell (F (1,42) = 14.586, p<0.001), monocyte (F (1,42) = 14.808, p<0.001) and neutrophils (F (1,42) = 17.208, p<0.001) (Table 4). Matched comparison of blood chemistry values also showed significant differences between colonies for albumin (F1,26) = 8.032, p=0.009), globulin (F1,26) = 8.019, p = 0.009), and A/G ratio (F (1,26) =11.075, p=0.003). Regardless of which colony the animal was from, age was significantly correlated with lymphocyte count (r = -0.311, p=0.04), alkaline (r= 0.551, p=0.002), globulin (r= 0.57, p=0.001), BUN (r= -0.434, p=0.17), A/G (r= -0.516, p=0.003), and BC (r= -0.425, p=0.019). There were no significant changes in blood parameters for individuals over the 18 months of sampling regardless of location. There were also no significant differences in any measured urinary parameters between the locations.

There were very few adult deaths during the tenure of the barrier colony. There were only 12 female deaths and 7 male deaths from the original founders and the 141 animals that survived past the age of 14 days. From these deaths there were only a few for which conclusive pathology and cause of death were determined. A male age 4.5 y died from congestive heart failure and epilepsy. A male 2 y old appears to have died due to complications associated with diabetes. A male juvenile age 6 months died from congenital pancreatic failure. Siblings age 3.5 y died unexpectedly, the male from congenital gastrointestinal defect and the female from anemia. A female 4.5 y died from postpartum complications. A female’s death was associated with cardiac and renal failure at 4.3 y. For several deaths in the colony we do not have conclusive pathology or cause of death, we present here anecdotal symptomology noted around time of death. A male age 7.6 y was found with a rectal prolapse of unknown etiology immediately prior to death. A female age 9 y exhibited signs of dehydration following infant delivery shortly prior to her death. Her mate age 8.6 y was noted to have enlarged hardened kidneys during a physical prior to his death of unknown cause. One infant male age 22 days was being supplemented with formula and may have aspirated during feeding. For the other 11 deaths there were no reported clinical symptoms or signs prior to death. There is no evidence from the pathology reports that any animals in the barrier died due to inflammatory gastrointestinal or infectious disease.

By contrast, in the comparative SNPRC population there were five deaths of animals under the age of one year for which the cause of death was determined to be necrotizing colitis, two that were noted as failure to thrive infants with extensive amyloidosis, two with Pseudomonas septicemia, and three with diarrhea associated with colitis. Between the ages of one year and two years four animals died due to necrotizing colitis, eight deaths were colitis associated, three deaths were associated with pneumonia, one death was due to cholangitis, two deaths were a result of trauma, and two were associated with amyloidosis. For animals aged 2-5 y there was one death associated with necrotizing colitis, six due to colitis, one due to pneumonia, one due to cholangitis, five associated with amyloidosis, two individuals with lymphoma, three deaths due to nephritis, one animal died due to cardiomegaly, and one female died during delivery due to dystocia. For animals between the ages of 5 -8 y there were seven deaths associated with colitis, one death due to cholangitis, one death due to lymphoma, one associated with nephritis and one female death due to postpartum complications. For animals over the age of eight there were two deaths associated with colitis, two deaths due to amyloidosis, three deaths due to nephritis, and five deaths due to cardiomyopathy.

Microbial communities

The lack of gastrointestinal problems in the SPF barrier colony and the litany of gastrointestinal problems observed in the SNPRC animals might be linked to differences in the composition and function of microbial communities in their gastrointestinal tracts. Sequences of the 16S ribosomal RNA gene were used to characterize microbial community composition for each of 38 SPF marmosets and 16 SNPRC marmosets. A subset of 19 microbial taxa were used in the comparison because these taxa occurred in at least one sample at a frequency of 5%. A hierarchical cluster analysis of the microbial communities based upon the frequencies of these taxa was performed (Figure 2). This analysis showed that the gastrointestinal microbiomes of barrier and SNPRC animals differed in terms of the kinds and abundances of various bacterial taxa (Figure 2B). There were three major clusters that grouped together microbiomes with similar bacterial compositions. The frequencies of the clusters in the two colonies were significantly different (Table 5; χ²= 12.8, simulations=5,000, p-value< 0.002) [27].

Figure 2. (A) Hierarchical clustering of marmoset gastrointestinal microbiomes from Barshop SPF barrier colony (Barrier) and the conventionally housed SNPRC colony (Center). (B) Frequency of taxa used in the cluster analysis, ordered by position in (A).

The frequencies of Bifidobacterium and Fusobacterium B (a group that is mischaracterized as Clostridium XIX) largely determined membership within a given cluster, and they were negatively correlated with each other (Figure 2). Thus, communities in Cluster 1, which was the predominant type in the SNPRC colony, had comparatively high frequencies of Fusobacterium B (>6%, median=27%) and low frequencies of Bifidobacterium (<8%, median=4%). In contrast, communities in Cluster 3, which was predominant in the barrier colony, had higher frequencies of Bifidobacterium (median=17%) and lower frequencies of Fusobacterium B (median=1%) with only a single exception, which had high frequencies of both. Communities of Cluster 2 had the highest frequencies of Bifidobacterium, >50% (median=70%) and the lowest frequencies of Fusobacterium B (median=0.0%). Animals with these communities were infrequent in both colonies.

Specific pathogen free marmosets

The SPF barrier marmoset colony was begun at the Barshop Institute for Longevity and Aging Studies, University of Texas Health Science Center at San Antonio in 2006. In order to populate this colony, animals from the Southwest National Primate Research Center were screened for several identified agents associated with pathogenic outcomes. Blood from animals of interest was screened for the presence of the viruses GB virus-A (GBV) and Callitrichine Herpes virus-3 (CHV) by the Lanford viral core at the Texas Biomedical Research Institute [39,40]. Marmoset blood was obtained with EDTA anticoagulant. Plasma was removed after centrifugation. The cell pellet was suspended with PBS to 1 ml. DNA was purified from the 200 µl of the cell suspension using a Qiagen mini-DNA column purification kit. DNA was eluted in a volume of 25 µl of water and the concentration was determined by using a Nanodrop ND-1000. TaqMan real time PCR amplification for CHV DNA was performed with 200 ng of cell DNA using an ABI 7500 TaqMan machine with the ABI Fast Advance PCR solution. The TaqMan primers and probe were designed against the major internal repeat of CHV. The primers and probe were: forward primer (TGGGCCTAGTCTCCCCATAGA), reverse primer (GTGAGGGAGTCCATAAGGAAACTTT), and a Fam-Tamra labeled probe (CGCCTGTATGTCTTACTGGGACCCCTG). The RT-PCR assay for GBV-A RNA was a gel based assay. Multiple attempts to develop a quantitative TaqMan RT-PCR were not successful. The sequence of GBV-A was not available from multiple isolates. The target region selected for this assay was based on available sequences and was variable (Buhk). Sequence of the target region from animals within the SNPRC colony, confirmed the sequence variation confounding the ability to develop a probe for the detection of all isolates. The RT-PCR used for this project employed a single round of 40 cycles of amplification followed by detection of the product by agarose gel electrophoresis and ethidium bromide staining. RNA was purified from 50 µl of plasma using RNA-Bee following the manufacturer’s protocol. The final RNA pellet was suspended in 50 µl of water and 10 µl was amplified using the Invitrogen RT-PCR Superscript Kit for Long Templates. The primers detected a 400 nucleotide region in the 5’untranslated region (UTR) of marmoset GBV-A. The primers were: forward primer (AGGGTTCGTAGGTGGTAAATCCC) and reverse primer (TGCCACCAGGGGTCACCCGAAG).

Fecal samples from candidate animals were evaluated for the presence of Giardia spp., Cryptosporidium spp., Clostridiumperfringens, helminths, parasitic ova, enteropathic E. coli, Campylobacter spp., and Trichimonas spp. through IDEXX commercial veterinary screening services. In initial screenings of the SNPRC colony no marmoset tested positive for the presence of Campylobacter spp. or Trichimonas spp. and these assays were dropped from further assessments. Screening for all other pathogens was done prior to colony formation and then periodically after the formation of the SPF colony. The first SPF male-female pair of marmosets was moved to the barrier facility at the Barshop Institute July 2007. The initial formation of the colony consisted of a total of 12 founder animals, with animals entering between July 2007 and April 2010. In order to assess the maintenance of the colony as specific pathogen free, viral screening occurred for all animals six months after entry to the barrier and sentry animals (one from each family group) were six years after formation of the colony. Fecal collections to assess the presence of Giardia, Cryptosporidium, and Clostridium perfringens were done annually following the formation of the colony.

Barrier maintenance

The first barrier rooms of the colony were two 10’x 29’ rooms within the animal care facility at the Barshop Institute located at the UT Health San Antonio Texas Research Park. Rooms were equipped with bioBubble units (Fort Collins CO, www.biobubble.com) consisting of two HEPA filtered airshower power units providing 200-300 air changes per hour in the 10’ x 18’ animal vivarium space (Figure 3A). The front entry of the room contained clean PPE storage for entry into the screened vivarium. Each room maintained a positive air pressure to the outside hall. All items within the bioBubble and entering the bioBubble area were considered clean and rigorous SOP’s to ensure cleanliness were used. The vivarium housed six family units consisting of at most two adults, four independent offspring and two-three dependent offspring in marmoset breeding cages measuring (1.3 x 1.5 x 0.6m) made of stainless steel specifically designed for this colony (LGL) (Figure 3B). All items entering the barrier were sanitized prior to entry or autoclaved. Water bottles were autoclaved before use, and all water used in the room was autoclaved. The paper placed under the cage to capture waste was DACB neomycin coated cage liner (Shepherd Specialty Paper, Watertown, TN) to decrease the potential for aerosolized urinary product and contamination between cages. Personal protective equipment for the facility included standard shoe covers, hair bonnet, face mask and gloves as well as a disposable outer gown. Staff were required to enter the barrier rooms prior to work with any other animals or entry into other animal facilities. Clean scrubs were required to enter the barrier spaces, and if a staff member had already entered another animal room they were required to shower before entering the marmoset barrier.

Figure 3. (A) Barrier space at the Barshop Institute with bioBubble airshower units. (B) Stainless steel marmoset breeding cage (LG). (C) Layout of converted BSL3 Barrier space for marmoset colony (* from - Nishijima et al., 2012).

SPF marmosets in the barrier were fed a purified diet to prevent food borne pathogens from entering the colony. The Harlan Teklad purified marmoset diet (TD99468) used by SNPRC [41] was modified to be irradiated at the production factory and double bagged in 1kg portions. This purified diet was prepared under a hood using gelatin as the setting agent and autoclaved water. The food containers were sealed under the hood and transported to a refrigerator in the animal room. Due to the unusual nature of the facility food enrichment was restricted within the barrier. An irradiated primate enrichment mix consisting of nuts, seeds, and dried fruit from Harlan Teklad was given daily. In cases of ill animals, or animals that were deemed to need supplements, items were chosen that could be purchased in sealed single use containers such as baby food or high calorie nutrition drinks, and unused portions were disposed of daily.

In May 2013 the SPF marmoset colony was moved to a new Barrier space at the Barshop Institute (Figure 3C). This new barrier space was 2850 ft² decommissioned and reconfigured BSL3 space in the animal vivarium. This new barrier facility consisted of a pass-through double door entry, a food preparation room, laboratory procedure room, four animal holding rooms (~180ft² each), a surgical suite, and 2 autoclave entry areas. The isolated air filtration system in this space was modified such that instead of HEPA filtration preventing contaminants from leaving the space the filtration prevented contaminants from entering the space. The entire space once entering through the pass through double door entry was considered clean barrier space. The air exchange in this facility was also maintained at the higher rate of 200 changes per hour. Humidifiers (Humidifirst) were added to each animal holding room that use double filtered RO water to maintain room humidity between 35 and 40%. Barrier protocols were maintained similarly to the bioBubble space and all other general maintenance and husbandry protocols were adopted from SNPRC’s marmoset colony [42]. The research adhered to ethical guidelines for primatological research as outlined by the American Society of Primatologists and the Institutional Animal Care and Use Committee at the University of Texas Health Science Center San Antonio and SNPRC.

Health monitoring

To compare the health outcomes for SPF marmosets maintained in a barrier to conventionally housed marmosets we evaluated blood CBC and blood chemistry of all animals in the barrier to age and sex matched animals maintained in the SNPRC marmoset breeding colony. All of the blood counts were done in the Tardif lab by a single technician, and the blood chemistry samples were sent to IDEXX for evaluation. Additionally, urine was collected for these age- and sex-matched individuals and basic urinary values were assessed using a strip test to establish values of protein, glucose, ketone, specific gravity and presence of blood in the urine. Results from the two colonies were compared statistically using SPSS 23.0 repeated measures ANOVA.

To examine longevity in the SPF colony, age at death for individuals in the barrier were compared to the SNPRC colony records for age of death during the same date frame. We compared survival in a population of 247 barrier animals (115 females and 132 males) versus a population of 370 conventionally housed animals (173 females and 197 males) over a 10-year period. The final date of comparison was August 2016 and animals were censored from the data if they were still alive at the end of date frame, if they had been transferred to another primate facility, or if death was associated with an experimental protocol. For marmosets, the first two weeks of life present a high risk of neonatal death and are typically removed from longevity analyses, as we did here. Survivorship curves were estimated using Kaplan-Meier estimation, Cox proportional hazard model and Fisher’s exact test with males and females evaluated separately.

Microbial community analysis

To evaluate the microbial diversity of the marmoset gastrointestinal tract, 38 marmosets living in the barrier colony and 16 marmosets from the conventionally housed SNPRC colony were sampled between October, 2011 and April, 2012. A fecal sample was taken from each marmoset by inserting a mini e-swab (Copan Diagnostics) into the rectum. E-swabs were stored in Amies transport medium and immediately frozen at -80° C. Samples were mailed to the University of Idaho on dry ice and stored at -80° C. Using methodology that is standard at the University of Idaho [43], DNA was extracted from the fecal samples, and the V1-V3 region of the 16S ribosomal RNA gene (universal 16S rRNA primers 27F and 534R) was amplified and then sequenced by an Illumina MiSeq. Sequence reads were processed for quality and assigned to best taxonomic level by the IBEST Genomics Resources Core (http://www.ibest.uidaho.edu/cores/genomic-resources-core/). The depth of coverage for each community was sufficient to detect taxa that constituted ≈0.01% of a community. Taxa that occurred at low frequency were removed from further consideration (ie: if a taxon had no single sample with a frequency greater than 5%).

The R statistical package, vegan, was used to calculate alternative Gower distances among samples based upon taxon frequencies, and these distances were clustered by a hierarchical algorithm, hclust, using complete linkage. The number of clusters was determined by eye and confirmed using silhouette plots. The number of samples from barrier and conventional marmosets were enumerated for each cluster, and Fisher’s exact test using simulated p-values was used to test for random associations between cluster and colony type. A stepwise discriminant analysis was used to determine whether specific bacteria contribute to the placement of samples within clusters.