6

Establishing and Evolving Gnotobiotic Facilities

The workshop’s final session focused on how to establish and maintain the facilities and infrastructure needed to house gnotobiotic animals. It also presented ways with which investigators have received support from institutional leaders to ensure that the appropriate resources are available to conduct microbiome-focused research.

BUILDING AND MAINTAINING A GNOTOBIOTIC FACILITY

Timothy Hand, assistant professor of pediatrics and immunology and chair of the committee on gnotobiotics at the University of Pittsburgh, is in the process of establishing a gnotobiotic facility, the first ever at his institution. He identified three critical steps to this process: (1) education and training in his institution, (2) choosing the layout of the facility and the appropriate equipment, and (3) building a gnotobiotic infrastructure.

Regarding education, Hand noted that educating the senior leadership at his institution was critical for obtaining funding to support the facility. The University of Pittsburgh did not have space that could be easily converted into a gnotobiotic facility, so Hand needed to convince senior administration to provide significant funding needed to renovate a space, as well as pay for the trained staff that is essential to the successful operation of a gnotobiotic facility. Part of Hand’s proposal was to demonstrate how such a facility would benefit the university faculty by controlling the costs of conducting gnotobiotic research. Hand explained that having investigators purchase individual gnotobiotic animals from large breeders was not fiscally sustainable, citing an average $500 cost for a single axenic mouse. Breeding mice in house would allow for less expensive long-term research. R. Balfour Sartor, distinguished professor of medicine, microbiology, and immunology and director of the University of North Carolina’s Multidisciplinary Inflammatory Bowel Disease Center, also noted that it is critical to ensure that the facility’s mission and cost structure matches those of the users, and especially that an institution would truly benefit from a gnotobiotic facility. He also underscored the importance of developing sustained, broad-based funding. His 31-year-old facility at the University of North

Carolina, for example, is funded by the National Institutes of Health (NIH) as a national resource facility, similar to the national primate centers, and as a resource center for investigators funded by the Crohn’s & Colitis Foundation of America (CCFA). In addition to securing this stable and external funding, Sartor has also positioned his facility as a resource for local, regional, national, and international investigators who want to explore hypotheses related to the effect of resident microbiota on animal physiology. Chriss Vowles, co-manager of the germ-free research laboratory at the University of Michigan, also stressed the importance of a funding source other than user fees to support ongoing operations of a gnotobiotic facility, particularly in terms of retaining highly trained staff. He noted that the labor alone for preparing materials for the facility often has to begin a full month ahead of the actual experimentation schedule. Increased labor costs, because everything needs to be sterilized, mean that “it is extremely unrealistic for the core to support itself on user fees alone.”

Hand also informed the senior administration that development of an in-house gnotobiotic facility would facilitate the experiments. He noted that collaborating with investigators at other universities or contracting all germ-free work was not sustainable, as such collaborations or contract arrangements would create additional legal and administrative hurdles given the need to establish material transfer agreements for every experiment.

Sartor added that there are additional reasons beyond ease of experimentation that should incline administrators to want an on-campus gnotobiotic facility. Having such a facility at an institution increases faculty competitiveness for NIH and foundation grant applications, and the facility serves as a tool for recruiting faculty who will generate more grant support and will increase an institution’s national visibility and reputation. Betty Theriault, associate professor of surgery and clinical veterinarian in the Animal Resources Center at The University of Chicago, noted that having a gnotobiotic facility creates an environment for collaborative studies with investigators from other institutions who bring ideas and funding to her home institution. The one downside of becoming a center for collaboration, she said, is that it requires developing more memoranda of understanding, material transfer agreements, and training investigators from those outside institutions. While many investigators have some understanding of sterile techniques, for example, working in a gnotobiotic facility requires a level of sterile handling that most researchers are not familiar with and may find challenging.

Hand has found it important to educate the university’s research community about the particular requirements of gnotobiotic research, how gnotobiotic animals are produced, the advantages and limitations of gnotobiotic research, and what can and cannot be done with gnotobiotic animals. Given the level of interest from his colleagues to examine the role microbiota might play in transplant rejection, he has had to explain how difficult such experiments would be. “Without discussing the nuts and bolts and the technical issues associated with a gnotobiotic isolator, you cannot really explain to someone who is a transplant

immunologist what is going to be difficult about transplanting hearts onto the aorta of germ-free mice,” Hand explained.

He added that this educational effort served a purpose beyond informing his colleagues—it also became an ad hoc recruitment effort to utilize the facility. Even though Hand needed to obtain funding from his university to maintain the facility, he had to ensure that his colleagues were actually going to use the facility. To help with recruitment, he and his colleagues at the Center for Medicine and the Microbiome held a full-day symposium in spring 2016 and a “Microbiome Boot Camp” in early 2017.

These activities provided Hand with the opportunity to learn what his colleagues’ interests in gnotobiotic research were. Somewhat to his surprise, significant interest was focused on the pulmonary microbiome, the relationship of the microbiome to cancer therapies, and the metagenomics of the human microbiome. This knowledge changed how Hand equipped and staffed his gnotobiotic facility, because these research interests required long-term rather than short-term housing, which affected the type of isolators to purchase. Hand recognized Alexander Chervonsky and his team at The University of Chicago for helping with the design of the University of Pittsburgh’s facility, and segued into his second key step to establishing a gnotobiotic facility—facility layout and equipment. His planned facility will be composed of two independent facilities, each operating autonomously with no crossover between the adjacent spaces. “This provides redundancy,” said Hand. “If one facility gets contaminated, we will still have germ-free mice on the other side that we can repopulate our facility with.” Other notable features of the facility include placing the autoclaves in separate spaces away from the isolators to reduce noise and stress for the animals, security features to prevent unauthorized and untrained personnel from entering, and a heating and ventilation system that will vent positive to the hallway, ensuring that no outside microbes are brought in via air circulation. Sartor noted that, if possible, one should plan a new facility with expansion in mind, so that the facility can be upgraded as technology develops, the needs of users evolve, and the science advances.

Hand explained his rationale for choosing one type of isolator over another, as an example of the many decisions to be made. The first one was a vinyl isolator of the type used at the Centers for Disease Control and Prevention that can house mice for many months, including during breeding. While this isolator has well-established standard operating procedures (SOPs), it is large and not amenable to conducting experiments involving 30 or more microbiomes.

Hand also considered a hermetically sealed, independently ventilated caging system that can house 34 experiments simultaneously in a small space. Turnover time with this type of isolator is fast, but this system cannot be used for breeding because changing the cages is extremely laborious. Vowles reviewed three types of housing used in his facility. The first type is a soft-sided bubble isolator, a double barrier system, that requires very basic personal protective equipment (plastic gowns, gloves). These units can house large experimental populations, but only one microbe or group of microbes may be exam-

ined at a time because cross contamination is unavoidable. The second type is a biosafety level 2 (BSL-2) cabinet that provides easy access and good dexterity, while the animals can be housed in basic, static caging. However, this housing type limits available space and community size significantly, only allowing up to five groups plus controls in each. The third type is the same hermetically sealed system Hand described. Vowles said that decontamination of this system requires extremely toxic chemicals, which are expensive and risk damaging the equipment. It also takes Vowles and his team a great deal of time to prepare the materials needed to work with these systems. Vowles has found that a combination of isolators and rack systems is the safest way to house an axenic colony. “It is easier to rear the mice in the isolators and then transfer them to the individual ventilated units in the rack system,” he explained.

Sartor noted that his facility has developed a customized surgical isolator, complete with warming blankets, stereotactic microscope, and surgical gloves that are only slightly thicker than normal surgical gloves—so not all additions to the design and construction of a facility need to be purchased. When considering how to incorporate various pieces of test equipment into an isolator, Theriault recalls the answer Philip Trexler, the inventor of the first flexible-film isolator, gave when asked that question: instead of getting the equipment into an isolator, wrap the isolator bubble around the test equipment and the animals. She briefly described several approaches to maintaining sterility when transferring animals for imaging or surgery, including the use of a sterilized biological safety cabinet in the gnotobiotic animal facility and training investigators to use this setup (Theriault et al., 2015).

Regarding sourcing mice for the facility, Hand said that the decision to obtain genetically modified strains of gnotobiotic mice from an external source or re-derive them in house would depend on the availability of experienced investigators able to perform these experiments.

In addition to the actual facility, supportive infrastructure is necessary. A germ-free facility, for instance, will not be very effective without the ability to grow organisms to place in the animals. At the University of Pittsburgh, the only anaerobic growth chamber is in Hand’s laboratory, at which he is growing communities for eventual use in the facility. The university has invested heavily in sequencing capabilities and bioinformatics expertise, including hiring two investigators with expertise in assembling full bacterial genomes using shotgun sequences of microbiomes. Hand relies heavily on lessons learned at other facilities. Following SOPs developed in other facilities also helps ensure the welfare and well-being of the mice in his facility. “We are not changing anything from what we did at NIH,” said Hand. “We really are trying to reproduce that facility.”

Maintaining and Operating a Gnotobiotic Facility

Once a facility is operational, additional considerations are necessary to foster sustained success. Sartor, whose facility was established in 1985, said that

stable scientific and technical leadership, as well as staff continuity, is essential, a point with which Hand agreed.

Perhaps the most important aspect of establishing a gnotobiotic facility, said Hand, is hiring dedicated, detail-oriented staff who understand and observe SOPs and who can be trained in an existing gnotobiotic facility, something Theriault and Vowles also stressed. Working in a gnotobiotic facility can be very physically demanding, as most operations need to be done manually. Therefore, staff should possess a commitment to cleanliness and attention to details, said Theriault. All interested parties should be aware of this issue. Vowles similarly noted the intensity and high level of fatigue. “You cannot make any mistakes. You could lose months and months of work contaminating an isolator by just touching something that might be dirty.” Vowles noted that training could take up to six months before new staff could work unsupervised. A gnotobiotic technician is a highly skilled staff member deserving a higher salary than a standard animal facility technician, which Sartor also emphasized.

Another element of success is having a committed core user base. Sartor’s facility, for example, has a core group of 15 or so investigators and another 20 to 25 additional users each year. Given that all 40 researchers cannot use the facility simultaneously, it is important to have transparent and equitable prioritization, he said. His facility utilizes a web-based scoring system that takes into account NIH or CCFA funding, being a local center member or young investigator, how long someone has been on the waiting list, and if someone has a grant or manuscript pending. The higher an investigator’s score, the sooner that investigator will be able to use the facility.

Theriault noted that prior to 2005 very few people engaged in gnotobiotic research and very few germ-free facilities existed. Between 2005 and 2010, several institutions, including hers, began to develop such facilities, and today it appears that everyone would like to develop some type of gnotobiotics program. She cautioned that “nothing with this technology is as easy as it appears.”

Vowles believes that a lack of national standards for gnotobiotic facilities increases the challenges of operating such facilities. “I think as a gnotobiotic community, we should come together, have an open dialog and create some of these standards in the next few years,” said Vowles. He also advocated for Institutional Animal Care and Use Committee (IACUC) standards for gnotobiotic facilities and animals.

Typically, most rodents coming into Theriault’s facility are from approved vendors with well-characterized specific pathogen-free (SPF) status. Animals from non-approved sources are quarantined to keep the general animal facility free of specified pathogens. Vowles commented that veterinarians and technicians must also be aware of potential contamination to the facility even while treating the animals. Medications and associated materials brought into the facility must be sterile. Furthermore, noted Theriault, veterinarians need to be cognizant not only of the health of the animals within the facility but also of the potential for contamination when the veterinarian interacts with the animals.

The University of Chicago first built its gnotobiotic animal facility in a remodeled old storage building. Increased demand led at first to an expansion into a neighboring storage closet and then to an entirely new, dedicated facility. Referring to Sartor’s prior comments, Theriault repeated that this technology is ergonomically challenging, labor intensive, and expensive. The current standard approach to monitoring for adventitious pathogens in SPF colonies is labor intensive and requires euthanizing animals. However, polymerase chain reaction–based assays to evaluate the microbial status of the colonies are in development, which would help reduce the number of animals to be euthanized for monitoring purposes.

Housing these animals and keeping them germ-free is only a means to the end of conducting research with them. Protecting these animals from pathogens and other microbes starts, she said, with education, training, and communication, both within and outside of her university. “We have to educate new members of our research community,” said Theriault. “We have to educate our IACUCs, our biosafety committees and program visitors, or maybe even our USDA [U.S. Department of Agriculture] inspectors.”

Maintaining the health of germ-free animals entails more than just protecting them from infection by adventitious pathogens, said Vowles. As they age, for example, germ-free animals develop an abnormally large cecum, which occasionally causes volvulus. Providing proper nutrition is challenging as well, because key nutrients, such as thiamine and vitamin K, degrade during the chow autoclaving process. Even irradiating food or using supplements can be unreliable.

Theriault explained that many experiments might require animals to be anesthetized, immobilized, X-rayed, and even subjected to radiotherapy—on top of routine administration of test agents and sample collection—all of which require special procedures in the germ-free and gnotobiotic context. This should not be surprising given that the composition and diversity of the microbiome significantly influence homeostatic and metabolic processes. These differences, she added, can create additional challenges regarding protocol review, because many institutions’ guidelines apply to conventional animals. As a result, she often has to educate IACUCs about the possibility of straying outside of these guidelines. In fact, she advocates that researchers include pilot trials to assess anesthetic regimens in germ-free and gnotobiotic animals as part of their study design.

Given the challenges of operating these types of facilities, and the fact that an increasing number of institutions want to establish microbiome research programs, Theriault wondered if it was time to consider establishing regional gnotobiotic and microbiome centers of expertise and excellence.

ALTERNATIVES TO GNOTOBIOTICS: NORMALIZING THE ENVIRONMENT

Stephen Jameson, professor of laboratory medicine and pathology at the University of Minnesota, investigates whether the characteristics of a mouse’s

immune system would change if the mice had a more natural, physiological, and immunological experience. His concern with raising mice in a gnotobiotic facility is that animals are separated from natural threats in the environment that enable their immune system to develop as it might in the wild. To explore how a broader physiological infectious history would alter the immune system and immunological responses in inbred mice, Jameson and his collaborators are generating and working with what they call “dirty mice.” They house wild mice or mice from pet stores with genetically modified ones to produce dirty mice. “We know that these wild and pet store animals have been exposed to many pathogens, some of which we can define, and have their own particular blend of commensal microbes,” said Jameson. “Some of these pathogens and commensals will be acquired by the co-housed mice, but we do not try to control this.” He believes the dirty mice are a more authentic reflection of the human microbial experience and suggested that these models may better reflect a human immunological response. “[Dirty mice] are essentially the opposite of gnotobiotic mice.”

There are many logistical challenges to this approach, said Jameson. Dirty mice carry pathogens excluded from most animal facilities, and perhaps others that are equally dangerous, so working with them requires installing an isolation barrier. His group houses its animals under BSL-3¹ conditions, at great expense, even though none of the pathogens involved are above BSL-2 status. He noted that deliberate sequential infection is a valuable alternative to working with wild-caught or pet store mice because it allows controlling an animal’s infection history.

Jameson commented on the tremendous frustration among mouse immunologists that predictions regarding the human immune system have not always proved to be true (Mestas and Hughes, 2004; Payne and Crooks, 2007; Rivera and Tessarollo, 2008; von Herrath and Nepom, 2005). Data show, for example, that immune cell populations from SPF mice correspond to those found in human umbilical cord blood (naïve CD8 T cells in both cases), which he said is fine if the goal is to model a newborn, but not if the goal is to model the adult human immune system.

Investigators use mice maintained in barrier facilities under SPF conditions to normalize the immune system prior to study. Humans, however, are exposed to a wide range of pathogens and commensals that shape the immune

___________________

¹ “Animal Biosafety Level 3 involves practices suitable for work with laboratory animals infected with indigenous or exotic agents … and requires that: 1) access to the animal facility is restricted; 2) personnel must have specific training in animal facility procedures… 3) personnel must be supervised by individuals with adequate knowledge of potential hazards… and 4) procedures involving the manipulation of infectious materials, or where aerosols or splashes may be created, must be conducted in BSC’s (biological safety cabinets) or by use of other physical containment equipment.” See https://www.cdc.gov/biosafety/publications/bmbl5/bmbl5_sect_v.pdf (accessed March 2, 2018).

system, which is the reason his research group began a collaboration with Dave Masopust’s group at the University of Minnesota to explore what happens in dirty mice. Large populations of CD8 T cells from wild mice had an effective memory phenotype, whereas the vast majority of CD8 T cells from SPF mice had a naïve phenotype (Beura et al., 2016). The problem with wild-caught mice is that there are many variables, such as their pre-capture diet and the season in which they are caught, that could affect the characteristics of these animals.

Mice from pet stores are easier to obtain and work with, plus they also present with the phenotypic conversion of CD8 T cells seen in wild-caught mice. Their main limitation is that they are not inbred. One solution is to convert inbred SPF mice to the pet store mouse phenotype by housing the SPF mice with pet store mice for two months. Jameson noted that crossover is close to 100 percent except for pinworms and mites, and that in some cases pathogen transfer can be fatal for the previously unexposed mice. However, after two months of co-habitation, the CD8 T cell phenotype of the SPF mice stabilizes, and activated T cells and other characteristics of a maturing immune system appear. Though these studies are in early stages, the investigators have found that the microbiota of co-housed mice becomes more similar to that of pet store mice; that is, it is changing. Jameson noted that gene expression analysis showed that co-housing pet store animals with standard SPF mice produced a profile that compared well with that seen in adult human blood, including the activation of type 1 interferon-inducing genes.

Regarding animal welfare, 20 percent mortality among B6 mice following exposure to pet store animals is a concern, though the mortality was much higher in BALB/c mice. Jameson noted that SPF-designated pathogens are the most likely cause of death, based on data from necropsies and pathology. Using contaminated bedding instead of co-housing leads to some reduction in mortality. Another way to avoid this problem, Jameson added, would be to establish sequential infection models to recapitulate the dirty mouse phenotype. Doing so could eliminate the need for costly BSL-3 equipment.