An ambient temperature collection and stabilization strategy for canine microbiota studies

Similar to humans, the fecal microbiome of dogs may be useful in diagnosing diseases or assessing dietary interventions. The accuracy and reproducibility of microbiome data depend on sample integrity, which can be affected by storage methods. Here, we evaluated the ability of a stabilization device to preserve canine fecal samples under various storage conditions simulating shipping in hot or cold climates. Microbiota data from unstabilized samples stored at room temperature (RT) and samples placed in PERFORMAbiome·GUT collection devices (PB-200) (DNA Genotek, Inc. Ottawa, Canada) and stored at RT, 37 °C, 50 °C, or undergoing repeated freeze–thaw cycles, were compared with freshly extracted samples. Alpha- and beta diversity indices were not affected in stabilized samples, regardless of storage temperature. Unstabilized samples stored at RT, however, had higher alpha diversity. Moreover, the relative abundance of dominant bacterial phyla (Firmicutes, Fusobacteria, Bacteriodetes, and Actinobacteria) and 24 genera were altered in unstabilized samples stored at RT, while microbiota abundance was not significantly changed in stabilized samples stored at RT. Our results suggest that storage method is important in microbiota studies and that the stabilization device may be useful in maintaining microbial profile integrity, especially for samples collected off-site and/or those undergoing temperature changes during shipment or storage.

The canine gastrointestinal (GI) tract harbors a highly dense and diverse population of microorganisms, including bacteria, fungi, archaea, and viruses, collectively termed microbiota. In recent years, DNA-based, highthroughput technologies have allowed a better characterization of the GI microbiome and its role in host health and disease. A large number of studies have demonstrated that altered gut microbiota composition and functions are associated with several metabolic disorders, such as obesity 1,2 , diabetes 3,4 , inflammatory bowel diseases [5][6][7] , autoimmune diseases 8,9 and cancers 10,11 in humans and/or dogs. Because microbial shifts are often noted with disease, microbiome profiles may be used as biomarkers for diagnosis and/or monitoring in the future [12][13][14] .
The microbiome field has much to offer, but appropriate sample collection and storage methods are needed for sample integrity and accurate and reproducible data. It has been shown that microbiome profiles of human fecal samples are changed after 48 hours of storage at ambient temperatures 15,16 . Many studies also demonstrated that proper stabilization and storage methods are key to having accurate and reproducible microbiome data 15,[17][18][19][20][21] . To date, the most appropriate and widely used collection protocol is to rapidly freeze and store samples at − 80° or − 20 °C 17,21,22 . However, these methods may not be practical in many cases. For example, very low-temperature storage might not always be accessible when collecting fecal samples from study subjects at home or in remote locations. Although studies have reported that storing samples at 4 °C slows bacterial growth and maintains sample integrity, they must be processed or stored at lower temperatures within 24 hours 21,23 . In addition, maintaining samples at low temperatures during shipment may be challenging and costly. Therefore, development of strategies able to stabilize samples at ambient temperature is of great interest.
Previous studies used a commercially available ambient temperature stabilization device OMNIgene•GUT (DNA Genotek, Inc. Ottawa, Canada) to collect and store human fecal samples. Results suggested that this device maintained the integrity of microbiome data and allowed for a higher recovery of nucleic acids at room temperature when compared with freezing at − 20 °C [23][24][25] . Similar to humans, researchers are interested in identifying the relationship between the GI microbiota, nutritional status, and health of dogs. The number of studies www.nature.com/scientificreports/ of the dog microbiome has grown in recent years [26][27][28] . There is, however, limited information about canine fecal sample collection and storage methods. The microbiota communities of humans and dogs are distinct. The human gut is dominated by Bacteroidetes and Firmicutes 29,30 while the canine gut is co-dominated by Bacteroidetes, Firmicutes and Fusobacterium 6,31 . Here, we used a similar commercial stabilization device designed specifically for animals, PERFORMAbiome•GUT (DNA Genotek, Inc. Ottawa, Canada), to study canine fecal sample collection and storage. In addition to what was conducted in previous human studies [23][24][25] , the current study investigated the effectiveness of this device in preserving samples during different storage temperatures that was meant to mimic shipment conditions where temperatures often change. The effects of a longer storage time (60 days in this study; up to 28 days in human studies) at room temperature (RT) were also studied. To assess the ability of this device to preserve fecal samples in different conditions, stabilized samples were stored at (1) RT (~ 23 °C), (2) 37 °C, (3) 50 °C, or (4) underwent repeated freeze-thaw (-20 °C/30 °C) cycles. We compared samples extracted on the collection day (baseline) to stabilized samples incubated for 1, 3, 14 or 60 days. Unstabilized samples were also collected and stored at RT to compare with stabilized samples stored at RT (Fig. 1). Finally, in order to determine intra-sampling variation, we compared samples collected from three locations of each fecal sample.

Discussion
As the relationship between the gut microbiota and human health has become clearer, interest in the microbiome field has increased dramatically. A similar trend has been observed in companion animals, as pet owners are now considering their pets to be family members and prioritize health and longevity. One critical factor that contributes to microbiome data integrity and accuracy is sample collection and storage methods. Currently, rapid freezing and storage of fecal samples at − 80 °C is widely considered to be the gold standard. This method, however, is often challenging and costly, which limits the ability of collection at remote sites or collection of www.nature.com/scientificreports/ a large sample number. Therefore, an ambient collection and stabilization strategy would be of great value for such microbiome studies.
In this study, we collected and stored canine fecal samples using a commercially available device, PERFORMAbiome•GUT (DNA Genotek, Inc. Ottawa, Canada). Stabilized samples were incubated at different conditions to evaluate the ability of the stabilization device to preserve fecal samples for microbiota analysis. In accordance with previous studies using a similar device for humans 23,24,32 , our results revealed that alpha-(species richness and evenness within samples) and beta diversity (species richness among samples) were not altered in stabilized samples after storage at 37 °C and 50 °C for up to 3 days, at RT for up to 14 days, or undergoing 6 freeze-thaw cycles. In contrast, substantial changes were observed in the unstabilized samples, where alpha diversity increased, and the abundance of several taxa were altered after 14 days at RT. In support of our findings, shifted microbiota profiles were reported in unstabilized human samples after 48 or 72 hours at RT 16,23,33 .
For stabilized samples stored at RT for 60 days, the Chao1 index for alpha diversity measures was decreased compared with baseline samples and samples stored for 14 days. The Chao1 index is a metric highly favoring rare OTUs. The longer storage with the PERFORMAbiome•GUT kit may reduce the diversity by decreasing www.nature.com/scientificreports/ numbers of rare OTUs. However, the relative abundance of the main phyla and genera were not significantly different among stabilized samples stored at RT at any time points. Additionally, correlations of relative abundance of genera after 60 days against baseline were high (r = 0.961). This was also observed in a previous study where human and canine fecal samples were stored at RT for 56 days using the OMNIgene•GUT, and the correlations of OTU abundance against baseline samples were r = 0.96 32 . Previous studies using OMNIgene•GUT kits for human fecal samples have reported changes in the relative abundance of a few genera. For example, a decreased abundance of Sutterella and Faecalibacterium were observed after 72 hours at RT 23 . Hill et al. noted the decreased abundance of Clostridium (XIVa and XVII) and Sorobacter as well as an increased abundance of Faecalibacterium after 1 week at RT 34 . Here, we found changes in Faecalibacterium and Clostridium, but not other genera reported in previous studies. After fecal samples were stabilized in the PERFORMAbiome•GUT device, an immediate increase in Faecalibacterium and decrease in Clostridium were noted when compared to the unstabilized samples. Additionally, the abundance of Faecalibacterium was decreased after six freeze-thaw cycles. This stabilization device and a similar product may not have the best ability to preserve Faecalibacterium as shown here and in two previous studies 23,34 . Therefore, this should be considered when testing samples of animals with gastrointestinal diseases, whereby Faecalibacterium abundance has been shown to be altered in both humans and dogs 13,35,36 .
Significant changes in the abundance of a few other taxa were observed in stabilized samples at high temperatures or freeze-thaw cycles. These changes, however, are considered relatively small when samples are stored at extreme conditions. In addition, correlations of genera abundance between baseline and after storage were high (r > 0.96). Similar to the results reported by Song et al. (2016) who used OMNIgene•GUT kits for human and dog fecal samples, small changes were noted in stabilized samples stored under two freeze-thaw (− 20 °C/ RT) and two heat (4 °C/40 °C) cycles, as correlations between OTU abundance at baseline and after 8 weeks were high (r = 0.93 and 0.77, respectively). The unstabilized samples, on the other hand, showed large shifts after freeze-thaw or heat cycles (r = 0.15 and 0.41, respectively) 32 .
Finally, we showed the intra-sampling variation of the microbiota profile in this study. The relative abundance of nine genera were different among collection sites within a fecal sample. Gorzelak et al. also found high variation in microbial abundance when human fecal samples were not homogenized before DNA extraction 37 . Therefore, these data suggest that homogenization of fecal samples should be done so that a consistent microbial composition reflecting the entire fecal sample is reported.
Our findings demonstrate that the collection and storage strategy tested in this study minimized changes in microbiota profiles of canine fecal samples stored at ambient temperatures or undergoing significant temperature changes. This device should enable sample collection and storage when ultra-low temperature storage and transport methods are not feasible, allowing for collection at remote locations. This strategy also allows for shipment at low or high temperatures and long-term storage at ambient temperatures. Together, this device appears to be a robust approach for canine fecal sample collection and storage. Methods fecal sample collection. All animal care and experimental procedures were approved by the University of Illinois Institutional Animal Care and Use Committee before experimentation (protocol #17008, #17135, and #17180). All methods were performed in accordance with the United States Public Health Service Policy on Humane Care and Use of Laboratory Animals. Fresh fecal samples were collected from 30 healthy female adult beagles (average age: 4.1 ± 1.3 years) within 15 minutes of defecation. All fecal samples were scored according to the following scale that has been used by our research group for decades 38,39 : 1 = hard, dry pellets, small hard mass; 2 = hard, well formed, dry stool; 3 = soft, formed, and moist stool, retains shape; 4 = soft, unformed stool, assumes shape of container; and 5 = watery, liquid that can be poured. All fecal samples had scores ranged from 2.5-3, which are considered to be normal, healthy stools. Three subsamples (2 ml; from each end and middle of a fecal sample) were collected and placed into cryovials and immediately frozen and stored at − 80 °C. The remaining fecal samples were aliquoted either into a conical tube stored at room temperature or into three PERFORMAbiome•GUT collection devices (PB-200) (DNA Genotek, Inc. Ottawa, Canada). After baseline microbiota data were obtained from each collection device, one stabilized sample was aliquoted and incubated at 37 °C and 50 °C. The other two stabilized samples were stored at RT (~ 23 °C) or undergoing six freeze-thaw (− 20 °C/30 °C) cycles (Fig. 1).

Figure 5.
Fecal microbiota communities of stabilized and unstabilized samples at baseline (D0-RT, D0-UNST) and stored at room temperature (RT) for 14 (D14-RT, D14-UNST) and 60 days (D60-RT). (A) Alpha diversity measures, including phylogenetic diversity (PD) whole tree, Chao1, and observed operational taxonomic units (OTUs), suggested that species richness and diversity were greater in unstabilized samples stored at RT for 14 days (Tukey's HSD, p < 0.001). The Chao1 metric showed a lower diversity in D60-RT than those of D0-RT (Tukey's HSD, p = 0.0193), D14-RT (Tukey's HSD, p = 0.0431), and D0-UNST (Tukey's HSD, p = 0.013). Principal coordinates analysis (PCoA) plots of unweighted (B) UniFrac distances of fecal microbial communities revealed that beta diversity was not altered by storage condition. PCoA plots of weighted (C) UniFrac distance showed that D14-UNST samples clustered together (circled area) and away from other samples. Unweighted UniFrac distance from baseline (D) revealed that D14-UNST and D60-RT had greater distance than 14-RT. Weighted UniFrac distance (E) showed that D14-UNST had a greater distance than stabilized samples. a,b mean values with unlike letters were significantly different (Tukey's HSD, p < 0.05). Scatterplots show the relative abundance of each genus in baseline samples against stabilized samples stored at RT for 14 (F) and 60 days (G), and unstabilized samples stored at RT for 14 days (H).  www.nature.com/scientificreports/ Bioinformatics and statistical analysis. Trimmomatic was used to remove sequencing adaptors and low-quality reads 40 . The FLASH algorithm was used for read merging and automated rejection of low quality sequences 41 . Quality screening for length and ambiguous bases was performed with proprietary scripts. Quantitative Insights Into Microbial Ecology (QIIME 1.9.1) 42 was used to process the sequence data. A closed-reference taxonomic classification was performed, where each sequence was aligned to the curated SILVA version 123 reference database 43 . Sequences were aligned at 97% sequence identity using the NINJA-OPS algorithm, version 1.5.1 44 . At 97% sequence identity, each OTU represents a genetically unique group of biological organisms. These OTUs were then assigned a curated taxonomic label based on the seven-level SILVA taxonomy. Alpha-and beta diversity measures were assessed at an even sampling depth sequences per sample. Data for each specific comparison (i.e., research aim) were analyzed separately. Alpha diversity was estimated using phylogenetic diversity whole tree, Chao1, and observed OTU matrices. Beta diversity was calculated using weighted and unweighted UniFrac 45 distance measures, and presented as PCoA plots. UniFrac distances between baseline and samples stored at 37 °C and 50 °C as well as baseline and samples stored at RT were calculated and compared. Correlation of relative abundance of genera between baseline and samples after incubations were analyzed using Pearson's correlation. Statistical analysis was conducted via Statistical Analyses of Metagenomic Profiles (STAMP) software 2.1.3 46 and SAS 9.4 using ANOVA and Tukey-Kramer multiple comparison tests. All tests were corrected for multiple inferences using the Benjamini-Hochberg method to control for false discovery rate 47 . Statistical significance was set at p < 0.05.
Received: 12 November 2019; Accepted: 27 July 2020 Table 2. Relative abundance of predominant bacterial phyla and genera in samples collected from different sites (Rep1 and Rep1: two ends; Rep2: middle) of a single fecal sample (% of total sequences). a,b,c Mean values within a row with unlike superscript letters differ (Tukey's HSD, p < 0.05). *Genera with statistically different relative abundances among collection sites are presented.