1 INTRODUCTION

The gut microbiota is the collection of microorganisms that inhabit the host's gastrointestinal tract. There is clinical and physiological importance to understanding the gut microbiome based on its interactions with the host in healthy states and its involvement in the etiology of many diseases, including rheumatoid arthritis,¹ colorectal cancer,^{2, 3} cardiovascular disease,⁴ and inflammatory bowel disease.^{5, 6}

Low-stress handling and fear reduction techniques are the new standard of care for veterinary patients.^{7, 8} Using a lubricant can minimize stress, discomfort, and pain and prevent escalating fear in cats. An additional challenge of the dry collection (nonlubricated) approach using a fecal loop is that, in many cases, little to no fecal material is obtained, resulting in variable and insufficient amounts of material for metagenomic analysis, which leads to missing data and an imbalanced experimental design. During sample collection with a fecal loop, lubrication can consistently guarantee sufficient specimens for analysis. In addition to the previously mentioned welfare benefits, the lubrication approach will reduce potential host contamination. Fecal loop use could cause abrasion to the intestinal wall, leading to the shedding of host cells and bleeding, which increases host DNA contamination. Therefore, using lubricant will allow sufficient fecal sample collection with reduced incidence of bleeding, pain, discomfort, and potential infection and less chance for host contamination in the fecal samples. However, using lubricant might alter the microbial composition and abundance in the gut microbiome. Currently, no research has addressed this issue.

Reproducibility and stability of microbial profiles recovered from fecal samples are vital to the reliability of the analytical results. In this study, we evaluated the WGS metagenomic data consistency from cat fecal samples collected using mineral oil lubrication vs no lubrication, to provide information for veterinarians and researchers regarding appropriate methods to evaluate the cat gut microbiome.

2 MATERIALS AND METHODS

3 RESULTS

No significant differences were observed in DNA yield per mg fecal specimen between miOil and noLub groups (75.75 ng [25.80-125.70] vs 60.72 ng [33.49-87.95], P = .95, Wilcoxon signed-rank test; Figure 1A), suggesting a lack of effect on microbial DNA yield using mineral oil lubricant. In total, 1.28 billion 150-bp reads (or 192 Gbp reads) were generated in the WGS metagenomic sequencing of 16 fecal DNA samples. Of these, 1.2% were adapter sequences or low-quality bases, and 12.5% were cat sequences. The yield from the miOil group (13.5 Gbp on average) was not statistically different from the noLub group (10.3 Gbp; P = .2, Wilcoxon signed-rank test; Figure 1B). We did observe an increased variation in sequencing yield in the noLub group (23.1) compared to the miOil group (3.2), which is statistically significant (P = .04, Levene's test of homogeneity of variance). The percentage of host contamination was not significantly different between the miOil group and the noLub group (P = .84, Wilcoxon signed-rank test; Figure 1C,D).

In the WGS metagenomic data from the noLub group, on average, we discovered 84.5 phyla [75.1-93.9], 73.8 classes [69.1-78.4], 157.8 orders [146.4-169.1], 330.5 families [303.7-357.3], 1141.6 genera [1002.5-1280.7], and 4985.9 species [4300.3-5671.5]. Under mineral oil lubrication, there was no significant change in the number of phyla (86.6, [77.2-96]; P = .96), classes (75.1, [71-79.2]; P = .96), orders (162.4, [153-171.8]; P = .83), families (342.8, [323.1-362.4]; P = .83), genera (1196.6, [1076.2-1317]; P = .67), and species (5300.6, [4679.3-5921.9]; P = .67) detected in the rectum microbiome (Figure 2).

Alpha-diversity was measured for the noLub and miOil metagenomes using the Shannon index (Figure 2). No significant differences in α-diversity were detected between the noLub and miOil samples at phylum (1.16 vs 1.18, [1.06-1.27] vs [1.08-1.28]; P = .83), class (1.59 vs 1.54, [1.37-1.82] vs [1.35-1.73]; P = .75), order (1.67 vs 1.62, [1.41-1.93] vs [1.4-1.84]; P = .6), family (2.25 vs 2.15, [1.97-2.52] vs [1.93-2.37]; P = .53), genus (2.48 vs 2.35, [2.16-2.81] vs [2.1-2.6]; P = .4), and species level (4.06 vs 3.88, [3.56-4.55] vs [3.49-4.26]; P = .4; Figure 2). When β-diversity was examined using Bray-Curtis dissimilarity in PCoA analyses, no significant changes were detected either (P > .39 for comparisons at all 6 taxonomical levels; PERMANOVA test; Figure 2). The overlapping confidence intervals indicated the microbial compositions could not be distinguished between the 2 fecal sample collection methods.

In our WGS metagenomic data, none of the 5 major phyla (Bacteroidetes, Firmicutes, Actinobacteria, Proteobacteria, and Fusobacteria) had significant differences in their relative abundance between miOil and noLub groups (P > .05, adjusted P > .85; Wilcoxon signed-rank tests; Figure 3A). In both the miOil and noLub groups, 99.72% of the metagenomic sequences belonged to bacteria reads, 0.11% were viral reads, 0.01% were archaeal reads, and the rest (0.16%) remained unknown. Considering the biological importance, we examined the microbial taxa with high relative abundance (>1.00%) at different taxonomical levels, which included 7 classes, 7 orders, 10 families, 11 genera, and 14 species. No significant changes were detected between miOil and noLub groups in the relative abundance at any taxonomic level (adjusted P > .8, Wilcoxon signed-rank test). The Spearman's rank-order correlation coefficients were extremely high between the noLub and miOil groups at the family (ρ = .99, P < .000001) and the genus levels (ρ = .98, P < .0000001; Spearman's rank-order correlation test; Figure 3B).

A total of 834 014 nonredundant microbial genes were identified in the 16 metagenomes. Of these, 798 430 genes were identified in 8 miOil metagenomes, and 769 476 genes were identified in 8 noLub metagenomes (Figure 4A). There was no significant difference in the number of observed genes between fecal samples collected with mineral oil and samples collected without lubrication (P = .31, Wilcoxon signed-rank test; Figure 4A). The alpha diversity based on the Shannon index of the observed genes from the 2 groups showed no significant difference either (P = .38, Wilcoxon signed-rank test; Figure 4B). Based on the Bray-Curtis distance matrix, no significant dissimilarity was detected between miOil and noLub groups in the PCoA analysis, and 95% confidence interval ellipses were overlapped (P = .94, PERMANOVA test; Figure 4C).

4 DISCUSSION

Fecal specimen collection is a commonly used approach in veterinary clinics and research to detect zoonotic parasites and diagnose pathogenic infections. Common methods for fecal sample collection in veterinary medicine include collection from the litter box and collection from the rectum using a fecal loop. Studies using cat fecal samples from litter boxes have 2 potential shortcomings. First, it is difficult to determine the freshness of the specimen, and excessive time at room temperature will shift the microbial composition. Second, it is likely the sample is contaminated by the environment. For projects designed to accurately represent the gut microbiome, fecal samples are collected from the rectum using a fecal loop, which can cause increased distress compared to litterbox collections. Lubrication is necessary to increase sample quality and to improve animal welfare during sample collection. In this group of cats, our results suggest that if adequately lubricated, the mineral oil applied will not affect the fecal microbial DNA extraction or gut microbiome analysis. The benefits of collection with lubrication include improved animal welfare and reduced variability in sequencing yield.

Fecal sample collection using a fecal loop in cats is challenging. It can cause discomfort, pain, and bleeding in the cat and result in little to no samples if it is done incorrectly. In an earlier attempt to collect fecal samples from the 8 cats enrolled in this research using a dry collection approach, we were only able to obtain a sufficient quantity of feces from 6/8 cats, and bleeding was observed in 5/8 cats. This failure led us to explore modifications to the fecal sample collection methods. Lubrication can ease the fecal sample collection process and ensure a sufficient amount of samples, but the lubricant could introduce materials that could prevent DNA extraction and sequencing or alter the microbiome composition. To determine if using lubricants during fecal sample collection has potential effects on the gut microbiome, we performed fecal collections with lubrication and without lubrication on the same 8 cats. The sample size was determined from the rarefaction plots from a previous study, in which a sample size of n = 6 will detect >90% of the bacterial species and microbial genes in the cat reference microbiome.¹⁵ In our metagenomic data analyses, we did not observe any significant changes in alpha-diversity, beta-diversity, the number of taxa discovered at each taxonomy level, and the relative abundances of taxonomic units are also highly correlated, indicating that the microbial composition was not affected by the use of lubricant.

We expected to see less host sequence contamination in the samples collected with mineral oil lubrication because the use of the fecal loop without lubrication is more likely to damage the intestinal wall, resulting in a higher proportion of host contamination. However, we did not observe any significant differences in host contamination, which could be because we sampled the same cat on the same day, and the more readily sloughed host cells had already exfoliated after the fecal loop collection without lubrication. The level of host DNA contamination (12.5%) is acceptable in both groups, given that intestinal cells are constantly sloughed off into the gut lumen.

Lubrication did not cause any issues in microbial DNA extraction, and a similar yield was observed in both groups, which is sufficient for subsequent research purposes. Interestingly, we observed a lower variability of the metagenomic sequencing yield, and this homogenous yield across the samples is beneficial for achieving an even level of metagenome coverage.

Whole-genome shotgun (WGS) metagenomic sequencing was performed in this study to assess the feline microbiome, as this method is rapidly becoming the new standard for assessing microbiomes across all species. We anticipate that the results will be applicable to 16S rDNA ampliconic sequencing, because a high correlation in taxonomic composition was observed in the feline microbiome using these 2 approaches.¹⁵

One limitation of this study is that our results only applied to fecal samples stored immediately in an ultracold freezer after collection. The same results might not hold in other storage conditions (room temperature, 4°C refrigeration, DNA stabilizing solution, etc.).

CONFLICT OF INTEREST DECLARATION

Authors declare no conflict of interest.

OFF-LABEL ANTIMICROBIAL DECLARATION

Authors declare no off-label use of antimicrobials.

INSTITUTIONAL ANIMAL CARE AND USE COMMITTEE (IACUC) OR OTHER APPROVAL DECLARATION

The study was reviewed and approved by the Auburn University IACUC, project number 2019-3482.

HUMAN ETHICS APPROVAL DECLARATION

Authors declare human ethics approval was not needed for this study.