Adoption of Exhaust Air Dust Testing in SPF Rodent Facilities

Abstract

Reliable detection of unwanted microbial agents is essential for meaningful health monitoring in laboratory animal facilities. Most rodents at our institution are housed in IVC rack systems to minimize aerogenic transmission of infectious agents between cages. The most commonly used rodent health monitoring systems expose live sentinel rodents to soiled bedding collected from other rodent cages on IVC racks and subsequently test these soiled-bedding sentinels for evidence of infection with excluded agents. However, infectious agents might go undetected when using this health surveillance method, due to inefficient organism shedding or transmission failure. In 2016, our institution switched the health monitoring methodology for the majority of our SPF rodent colonies to real-time PCR testing of environmental samples collected from the exhaust plenums of IVC racks. Here we describe our rationale for this conversion, describe some interesting health monitoring cases that arose soon after the conversion, and discuss a potential problem with the conversion—residual nucleic acids. We compared cost and implementation effort associated with 2 sampling methods, sticky swabs and in-line collection media. We also compared the ability of these 2 sampling methods to detect 2 prevalent microbes in our facilities, Helicobacter and murine norovirus. Our institution-wide switch to health monitoring by real-time PCR assay of exhaust air dust samples thus far has provided a sensitive, simple, and reliable approach for maintenance of SPF conditions in laboratory rodents and has dramatically reduced the use of live sentinel animals.

The use of sentinel animals has traditionally been a key element of the overall health surveillance program for laboratory rodent colonies. For institutions that use microisolation caging systems, the transfer of soiled bedding from colony animals is a typical method for exposing sentinel rodents to excluded infectious agents that may be present in the colony.⁵ Using soiled-bedding sentinels for the detection of excluded agents is widely accepted among research institutions throughout the United States and around the world.^5,8 However, drawbacks include the fact that some agents are not efficiently transmitted through soiled bedding and the process of bedding transfer can be inconsistent.⁵ Most importantly, this method requires the continued use of considerable numbers of live animals. Recent advances in PCR-based surveillance methods present alternatives for use of sentinel monitoring. Highly sensitive, customizable, high-density PCR arrays that detect DNA or RNA from excluded agents can be used on samples collected directly from colony animals¹⁰ as well as environmental samples, such as dust collected from filter material installed inside IVC tops⁷ or exhaust air dust (EAD) collected from exhaust plenums of IVC racks.^17,21 The widespread use of EAD-compatible IVC racks at our institution's animal facilities made testing of environmental EAD samples an attractive monitoring methodology.

A growing body of evidence supports EAD PCR testing as a viable means for the detection of excluded agents and suggests that EAD PCR is a more sensitive method of detection^{10,19,21,24,25} that could replace the use of live soiled-bedding sentinels. EAD sampling may be particularly advantageous for the detection of agents that are not readily transmitted through soiled bedding, such as CAR bacillus,⁴ some Helicobacter species,¹⁰ lymphocytic choriomeningitis virus (LCMV),¹¹ Sendai virus,¹ fur mites,^9,16 and Pasteurella species.^10,20 Multiple studies have demonstrated the sensitivity of EAD PCR for agents excluded from barrier facilities compared with soiled bedding sentinels. In particular, one study¹² demonstrated that EAD PCR testing can detect fur mites with 94.1% accuracy from only a single positive cage on an IVC rack after just 4 wk compared with a 3% detection rate when using soiled-bedding sentinels.¹⁶ In addition, EAD testing detected Pasteurella pneumotropica at a 100% rate of detection, whereas soiled-bedding sentinels failed to detect all positive cages.²⁰ These studies, along with others examining different agents,^13,26 support the use of EAD testing for rodent health monitoring of SPF colonies.

In 2016, the Rodent Health Monitoring Program at our institution implemented the use of EAD PCR testing in the majority of rodent colonies. Here we report our experiences with transitioning to EAD PCR monitoring to maintain our rodent colonies free of excluded agents. Importantly, this transition has reduced the number of animals used for surveillance by more than 2500 mice and rat sentinels annually.

Materials and Methods

Animal care and use protocol.

All procedures involving live animals were approved by the University of Washington IACUC.

Excluded agents.

Trimester and annual panels as well as confirmation testing of individual agents of mouse- and rat-specific PCR assays were performed by Charles River Laboratories (Wilmington, MA) or IDEXX Bioanalytics (Columbia, MO). Excluded agents for mice include mouse hepatitis virus, mouse rotavirus, mouse parvovirus, minute virus of mice, Theiler murine encephalomyelitis virus, pneumonia virus of mice, Sendai virus, reovirus types 1 through 4, LCMV, ectromelia (mousepox), pinworms (Aspiculuris, Syphacia), fur mites (Myobia, Myocoptes, Radfordia), Mycoplasma pulmonis, mouse adenovirus types 1 and 2, and mouse cytomegalovirus and, in selected rooms, Helicobacter spp., murine norovirus (MNV), and Tritrichomonas muris. Excluded agents for rats include, rat coronavirus, rat parvoviruses, Sendai virus, pneumonia virus of mice, Mycoplasma pulmonis, fur mites (Myobia, Myocoptes, Radfordia), and pinworms (Aspiculuris, Syphacia).

Sample collection for agent testing.

Trained personnel wearing personal protective equipment collected samples according to established protocols for nucleic acid collection. All sample materials were prepared within a laminar flow hood in the rodent room being sampled (Figure 1 A); all hood surfaces were sprayed with Clidox-S solution (1:18:1, Pharmacal, Naugatuk, CT) prior to use. Gloves and sleeves were changed prior to collecting each rack's samples. Gloves were sprayed with Clidox-S solution (1:18:1, Pharmacal) and wiped dry. Environmental sticky swabs, sterile 5-mL microfuge tubes, 50-mL conical tubes, and in-line EAD collection media strips (SENTINEL, Allentown, Allentown, NJ) were supplied by Charles River Laboratories.

Figure 1.

Racks and exhaust manifolds sampled by using sticky swabs. (A) Exhaust manifolds of racks (Allentown, Allentown, NJ) are easily accessible and face a hood where samples can be manipulated. (B) Open exhaust manifold; horizontal exhaust plena is being swabbed with a sticky swab. (C) Media is placed onto a collar mounted in the exhaust manifold.

For swab sampling, blowers were turned off, the exhaust plenums of the IVC racks (Allentown; Figure 1 A) were opened, and a swab was used to collect dust from the horizontal plenums (Figure 1 B). A total of 3 or 4 swabs were used per rack containing 10 rows of horizontal exhaust plenums. Swabs were pooled into a single submission per rack. The exhaust plenum door was closed after collection, and blowers were turned on. After completion of sample collection, the staff checked all racks to ensure that blowers were functioning correctly.

For media sampling, blowers were turned off, and one collar was placed in each rack at the top of the exhaust manifold. The media was placed in the collar (Figure 1 C). Media was collected every 4 mo and replaced with new media. Dates of media placement and sample collection were recorded on paperwork associated with each rack.

Swab sampling compared with media sampling.

A total of 116 Allentown IVC racks were used to compare swab sampling with media sampling for PCR testing. Media was collected after 4 mo of placement in the rack; rack exhaust plenum swabs were collected at the same time as the media collection. PCR assays for Helicobacter (genus level) and MNV (both agents are enzootic in our rodent colonies) were evaluated as positive controls. Samples that were positive for Helicobacter at the genus level were further evaluated for 6 Helicobacter species; H. bilis, H. ganmani, H. hepaticus, H. mastomyrinus, H. rodentium, and H. typhlonius. Discrepancy rates between the 2 sampling methods were calculated. A discrepancy between sample collection methods was tallied any time the results from swabs did not match those from media for a particular PCR reaction from the same rack. Discrepancy rates were statistically evaluated by using the Fisher Exact test in Prism 5.04 (GraphPad Software, San Diego, CA).

Cost analysis of swab sampling compared with media sampling.

We compared the costs of swab sampling and media sampling over a single wash cycle for a rack (8 mo). On 3 to 9 occasions, a technician was timed while performing the steps associated with each type of sample collection (Figure 2). After a rack goes into service (that is, occupied with rodents or cages), swabbing involves 2 time points when samples must be handled: 1) at 4 mo and 2) at 8 mo, before the rack is sent to the washroom. The media sampling method has 3 time points for sample handling: 1) as the rack goes into service, the technician installs a steel collar into the top of the open IVC plenum chamber and inserts media into the collar; 2) at 4 mo, dirty media is collected and replaced with clean media; and 3) at 8 mo, dirty media is collected, and the collar is removed before the rack is sent to the washroom. We performed timed tests for swabbing for both single-sided and double-sided racks, because access to sampling locations on single-sided racks is typically more difficult due to rack configuration, such that the plenum doors are usually against a wall. Times for media collection were assumed to be similar between double and single racks. For the 8-mo time point, time to remove media was calculated separately from the time to remove the collar, and the averages were added. Mean and standard deviation were calculated for each step; then the averages were summed and combined standard deviations calculated through propagation of error.

Figure 2.

Cost analysis of swabs and media; time points requiring handling of sample or associated equipment over the course of an 8-mo wash cycle. On 3 to 9 occasions, technicians were timed regarding the specified tasks, to estimate the time requirements for each EAD sampling method.

Results

Converting from sentinels to EAD PCR sampling.

In 2015, the sentinel program at our institution required the use of 2352 mice and 180 rats. To reduce the number of live animals required for health monitoring of rodents in our SPF facility, we began using EAD PCR monitoring. The majority (90%) of the racks used for housing rodents at our institution were Allentown IVC racks, making our facilities ideal for evaluating SPF status by using EAD PCR testing. In 2016, the use of soiled-bedding sentinel rodents was discontinued and EAD PCR monitoring was implemented on IVC racks in all of our institution's animal facilities. EAD sampling was performed every trimester by using either swabs to sample the horizontal exhaust manifolds or by using specialized media; samples were submitted to Charles River Laboratories for PCR testing. The media is designed for and compatible with Allentown IVC racks only. In general, EAD PCR testing results were negative across facilities for excluded agents, except for the occasional detection of fur mites (see later sections). Helicobacter spp. and MNV were detected frequently; these pathogens are known to be endemic in our colonies. From the implementation of the EAD testing program in 2016 through February 2019, our institution has sent 319 shipments of laboratory rodents to more than 135 other institutions spread throughout 14 countries; none of the receiving institutions has notified us of the detection of any excluded agents in these shipments.

Detection of fur mite infestations by using EAD testing.

Soon after the program-wide transition from soiled-bedding sentinels to EAD PCR testing, we detected Radfordia affinis in 2 established colonies, both of which historically tested negative for fur mites by soiled-bedding surveillance methods. The positive EAD PCR results were confirmed by using PCR testing of cage perimeter swabs (IDEXX BioAnalytics, Columbia, MO) and in-house tape tests, with tape tests providing positive visual confirmation of fur mite infestation.

False-positive results for LCMV.

LCMV is an enveloped RNA virus that is rarely detected in contemporary laboratory mouse populations. However, LCMV is often included in diagnostic serology and PCR panels because of its pathogenic potential for rodents and humans^22,23 and its potential to complicate research studies. The virus has been shown to alter research results of in vitro and in vivo rodent test systems as well as experimental studies of virus–host interactions.¹⁵ In addition, LCMV may be approved by institutional biosafety committees for research into areas such as mechanisms of viral persistence and autoimmunity. This approval may include research using mice transgenic for LCMV proteins, such as LCMV glycoprotein, or mice experimentally infected with LCMV. At our institution, LCMV infection is permitted only under ABSL2 containment, whereas breeding colonies of transgenic mice expressing LCMV proteins are held in ABSL1 rooms. When we began the transition to EAD PCR testing in 2016, a few of our earliest positive PCR results in ABSL1 rooms were for LCMV, despite many preceding years of negative sentinel serology.

We suspected that our LCMV-positive results were due to the detection of genomic DNA shed by mice at our institution that are transgenic for LCMV glycoprotein, given that real-time PCR testing for LCMV recognizes a conserved region of the glycoprotein, as others have shown.¹⁴ Therefore, we challenged the diagnostic laboratory to find a way to differentiate between genuine LCMV infection and artifact. Taking advantage of the fact that LCMV is an RNA virus, the laboratory repeated the PCR reaction but omitted the reverse transcriptase step used to create cDNA from RNA. Because the resulting copy numbers remained unchanged from the original assay, the laboratory concluded that the genetic material detected was genomic DNA and not viral RNA. In agreement with some of our colleagues,¹⁴ we recommend that institutions implementing EAD PCR testing should obtain confirmation of a positive result before embarking on an eradication program.

Residual nucleic acids and EAD testing.

Ineffective washing of IVC racks is a potential concern regarding EAD PCR because residual nucleic acids from a previous animal colony could lead to false positives. To evaluate residual nucleic acids on our facility racks, we tested 67 washed Allentown racks that were either recently washed or that were washed and placed in a clean storage area. Table 1 shows that a single rack wash left residual nucleic acids for Helicobacter and MNV in a low percentage of cases, despite the high prevalence of these agents in our facilities. For these particular agents which are not excluded throughout our entire facility, residual nucleic acids from Helicobacter or MNV on cleaned racks is likely to affect the monitoring of Helicobacter- or MNV-free rooms only. Our current standard operating procedure after the detection of a positive Helicobacter or MNV test in one of these rooms is to follow up by: 1) testing animals directly through fecal collection and PCR testing and 2) replacing the suspect rack with one that was washed and had a negative swab PCR test for Helicobacter or MNV, as appropriate. An outbreak would be declared only if positive test results were obtained from animals or a rack that had a negative test result prior to placement in the room. In practice, detection of residual nucleic acid in racks housing rodents has been rare (n = 1) at our institution despite multiple rounds of testing. Currently, racks are either dedicated to Helicobacter- or MNV-free use or are washed twice before housing Helicobacter- or MNV-free rodents.

Table 1.

2016 testing of dirty racks compared with racks washed once (swab sampling)

	Dirty racks No. positive (%) (total n = 212)	Washed racks No. positive (%) (total n = 67)
Helicobacter spp.	198 (93%)	5 (7%)
Murine norovirus	164 (77%)	2 (3%)

3 swabs per rack

With regard to a positive result for an agent that is excluded from our entire facility, the possibility of the result being from residual nucleic acids would need further consideration if subsequent direct testing of animals and cages from the affected room were negative. For example, racks that tested positive for Radfordia affinis in 2016 were washed twice, swabbed, and determined by PCR analysis to be negative for fur mite DNA before their return to service. No fur mite DNA has subsequently been detected by PCR testing of swabs or media collected from these racks.

Comparison of sampling media with swabs.

To determine whether using one type of EAD sampling (swab compared with media) conferred any advantage over the other type, we performed a field test under our normal sampling regimen for routine rodent health monitoring and compared the 2 methods by using test results from Charles River Laboratories. Because Helicobacter and MNV are enzootic in rodent housing facilities at our institution, we compared Helicobacter and MNV detection by using both EAD sampling methods and evaluated the discrepancy rate of the test results. A discrepancy between sample collection methods was tallied any time the results from a swab did not match that from media for a particular PCR reaction. In addition, we evaluated the direction of the discrepancy by categorizing the discrepancies according to the type of result (that is, positive or negative) obtained by using a particular sampling type (that is, percentage of swabs giving negative results:media samples giving positive results compared with percentage of swabs positive:media negative; Table 2). In general, few discrepancies were detected (Table 2). The results for individual assays are described in later sections.

Table 2.

Discrepancy in detection of Helicobacter and MNV depending on sampling method

PCR test type	Total discrepancy (%)	Swab sample negative but media sample positive (%)	Swab sample positive but media sample negative (%)
Helicobacter genus	2.6	2.6	0.0
MNV	4.6	2.8	1.8
H. bilis	1.9	1.0	1.0
H. ganmani	2.9	1.9	1.0
H. hepaticus	3.8	0.0	3.8
H. mastomyrinus	5.7	1.9	3.8
H. rodentium	1.9	0.0	1.9
H. typhlonius	1.9	1.0	1.0

Helicobacter spp.

A genus-level Helicobacter PCR test was used to evaluate 109 racks housing mice and 7 rat racks. For samples that were positive for Helicobacter at the genus level, 6 single-species tests (H. bilis, H. ganmani, H. hepaticus, H. mastomyrinus, H. rodentium, and H. typhlonius) were used to evaluate a second DNA extraction of the originally submitted EAD sample.

In the case of genus-level Helicobacter PCR testing, only 3 discrepant comparisons (2.6%) were detected among the 116 racks tested. All 3 discrepancies of the Helicobacter genus PCR test were negative by EAD swab and positive by EAD media and occurred when PCR values were near the cutoff level for the assay. To increase the number of comparisons between EAD sampling types, we evaluated discrepancy rates between sampling methods by using 6 different Helicobacter single-species PCR tests bringing the total number of comparisons to 630. Racks that were positive according to the Helicobacter genus-level PCR testing of both EAD swabs and media (105 of 116 racks tested) were included. Overall, the discrepancy rate was low (1.9% to 5.7%). Discrepancies were not skewed with the positive result always being obtained by using a particular sampling type; that is, the percent swab negative:media positive was not significantly different from the percent swab positive:media negative (Table 2). Even with the inclusion of 630 additional comparisons, the discrepancies between sampling methods was low, making statistical comparisons underpowered.

MNV.

Although previous studies have reported the inability to detect MNV by using EAD PCR methods,² we detected MNV in 82% of racks sampled by swab (data not shown). When we compared the sampling methods, there were 5 discrepancies (4.6%) among 109 comparisons (including 3 rooms that were maintained MNV-free). Similar to that observed in comparisons with Helicobacter tests, the discrepancies were not skewed toward a particular sampling method type, that is, 3 of the 5 discrepancies were positive for the media sample, and the remaining 2 were positive for the swab sample (Table 2).

The Helicobacter species and MNV data suggest that both sampling methods likely have a similar margin of error and do not indicate strong advantages of using one sampling method over the other for agent detection.

Cost and labor analysis.

Because the cost of testing is important when screening large numbers of animals, we performed a cost analysis of the supplies and effort necessary to use the different methods of EAD PCR testing over an 8-mo, single-wash cycle for a rack. We determined that the time to communicate with facilities management, coordinate sample collection and shipment, and compile results were the same between sampling methods. In addition, disposables for both types of sampling are now offered complimentary by testing laboratories, as a part of testing. Therefore, the key difference was in the time spent on actual sample collection. Overall, the average effort (and thus cost) involved in sample handling is approximately 1.4 times greater for the media compared with swabs when the time to sample double-sided racks (mean ± 1 SD; media, 8.9 ± 0.85 min per rack; swab, 6.5 ± 0.34 min per rack) was considered and was only slightly more time-intensive compared with swabbing for single-sided racks (mean ± 1 SD; media, 8.9 ± 0.85 min per rack [same time value as for double-sided racks]; swab, 8.4 ± 0.39 min per rack). However, the steel collar that is required for placement of the media is potentially a high up-front cost per rack. This cost is difficult to assess, because the cost of outfitting racks with the collars can be negotiated depending on the circumstances of the facility.

Discussion

Multiple factors favored our department's decision to transition to EAD testing for surveillance of excluded agents to maintain SPF rodent facilities. A major consideration was to reduce or eliminate the numbers of live animals used as sentinels. Prior to our institution's transition to EAD monitoring, we used more than 2500 mice per year for monitoring excluded agents and outbreaks. Additional considerations that are supported by previous research (presented in the Introduction) included improved detection of various excluded infectious agents that are not readily transferred by soiled bedding sentinels, elimination of the risk of receiving contaminated sentinel animals from vendors, and cost savings.

Prior to and during our implementation of an EAD PCR monitoring program, we needed to consider and troubleshoot multiple issues. Standard operating procedures needed to be developed regarding communication with and education of animal technicians and coordinating scheduling of EAD sample collection with rack washing schedules. In addition, we considered the compatibility of the method with rack design, because in some institutions, the rack design precludes the use of EAD sampling.⁷ Our institution's use of ventilated Allentown racks in the majority of rooms allowed for easy access to the exhaust plenum and sampling of exhaust air by using either swabs or media. Therefore, the decision was made to convert to EAD sampling in all of the animal rooms that had compatible ventilated racks. Currently, our institution has only a few animal rooms with static or other racks that are incompatible with EAD PCR testing; these are monitored by using live sentinels or direct testing (or both). Additional considerations included communication with and education of investigators, both at our institution itself as well as at those that receive animals from our institution. Receiving institutions were informed of the change in monitoring in advance of shipping, and, with only a few exceptions, EAD PCR results were accepted for receipt of SPF animals; in a few cases, the receiving institution requested supplemental direct testing of animals.

One concern regarding the use of EAD PCR for rodent health monitoring is the effect of residual DNA on cleaned racks that potentially could lead to false positives. One group has addressed this issue and has tested the efficacy of various rack washing methods in removing residual DNA from a previously housed rodent colony.⁶ Although more stringent rack washing methods and use of additional decontamination, such as a hydrogen peroxide wash, may be useful for some monitoring programs, we have found that they are unnecessary in our monitoring program and may be useful only in the case of a true outbreak. In our facilities, we have observed potential false positives only twice, involving genetically modified mice containing LCMV genomic DNA (described earlier) and maintaining Helicobacter-free rooms where the use of a rack that previously housed Helicobacter-positive mice remained contaminated with Helicobacter DNA after cleaning. Since 2015, the rodent health monitoring staff at our institution has worked with animal husbandry staff to develop, with these considerations in mind, standard operating procedures for rack washing and agent monitoring with EAD PCR testing, a summary of which can be provided on request to institutions that receive our animals. Our current EAD media monitoring practices are as follows. First, racks and collars are washed every 8 mo—3 wk after removal of the dirty media strips—and clean racks, collars and media strips are placed in animal rooms only after results are returned as negative. After 4 mo, dirty media strips are again removed for testing and replaced with fresh media. In case of positive results, dirty racks are swabbed and samples are submitted to a second laboratory for comparison and confirmation before racks and collars are washed and placed in animal rooms if postwash PCR tests are negative. In addition, samples are collected directly from involved rodents, and cage perimeter swabs may be performed as well. As in the case of Helicobacter-free rooms, an outbreak would be declared only when positive results were obtained from animals or from a rack that was tested to be negative prior to placement in the room.

With the increased use of EAD PCR testing for SPF monitoring, different sampling methods are being tested to aid in refinement of EAD sampling.²⁰ However, the literature lacks comparisons of EAD sampling methods. Our limited comparison of environmental surveillance by using EAD swab and media sampling revealed no evidence that suggests that either the swab or the media is more effective at the detection of MNV or Helicobacter. However, due to the low discrepancy rate between sampling methods, our power to detect statistically significant differences was low. Although the cost of using media appears to be higher due to the amount of time required for sample handling, our institution now uses this sampling method in almost all rooms across facilities based on several lines of reasoning. First, we surmised that in-line media might be less likely to detect residual DNA contamination remaining after rack washing procedures and thus yield a lower number of false-positive results. Second, the use of media could improve detection of intermittent infection or shedding of excluded agents due to the consistent presence of media in the airflow over a period of time. This situation, combined with reduced sampling variability due to technicians performing swabbing of exhaust plena, may lead to more robust detection of intermittent infection or shedding. To our knowledge, no study to date has sufficiently addressed whether these lines of reasoning result in meaningful differences in the quality of monitoring.

In conclusion, our experience with EAD surveillance over the past 3 y has been favorable, dramatically decreasing the use of live animals needed as sentinels. The absence of outbreaks or detection of excluded agents at other institutions that received shipments from our facility supports EAD monitoring as a reliable method of surveillance. Several institutions in the United States and internationally have transitioned from the use of live sentinels to this alternative for routine monitoring of their animal facilities as well as for detection and elimination of specific pathogens, such as Corynebacterium bovis.¹⁸ Other institutions are awaiting additional validation of this methodology before considering it as a viable method for rodent colony surveillance. Some studies have acknowledged limitations of EAD PCR monitoring methods in some cases,^2,3 although similar limitations exist in the use of live sentinel monitoring. Some institutions use a hybrid approach by performing live sentinel surveillance concurrently or alternately with EAD PCR testing. We encourage institutions that have adopted this or other forms of environmental surveillance to continue to present and publish their findings, so that we can improve the detection of excluded agents, reduce the chances of outbreaks, and reduce the reliance of our research institutions on live sentinel animals for rodent colony surveillance.