Effect of temperature and time on the thanatomicrobiome of the cecum, ileum, kidney, and lung of domestic rabbits

Abstract

Knowledge of changes in the composition of microbial communities (microbiota) in tissues after death, over time, is critical to correctly interpret results of microbiologic testing from postmortem examinations. Limited information is available about postmortem changes of the microbiota and the associated microbial genes (microbiome) of internal organs in any species. We examined the effect of time and ambient temperature on the postmortem microbiome (thanatomicrobiome) of tissues typically sampled for microbiologic testing during autopsies. Twenty rabbits were euthanized and their bodies stored at 4°C or 20°C for 6 or 48 h. Ileum, cecum, kidney, and lung tissue were sampled. Bacterial DNA abundance was determined by RT-qPCR. Microbiome diversity was determined by 16S rRNA gene sequencing. By relative abundance of the microbiome composition, intestinal tissues were clearly separated from lungs and kidneys, which were similar to each other, over all times and temperatures. Only cecal thanatomicrobiomes had consistently high concentrations and consistent composition in all conditions. In lungs and kidneys, but not intestine, proteobacteria were highly abundant at specific times and temperatures. Thanatomicrobiome variation was not explained by minor subclinical lesions identified upon microscopic examination of tissues. Bacterial communities typically found in the intestine were not identified at extra-intestinal sites in the first 48 h at 4°C and only in small amounts at 20°C. However, changes in tissue-specific microbiomes during the postmortem interval should be considered when interpreting results of microbiologic testing.

Introduction

The microbiome is defined as the combined genetic material of the microorganisms in a particular environment.²⁹ Microbiota are the microbial communities of a reasonably well-defined environment. However, the body is a diverse environment, and the “reasonably well-defined environment” is somewhat ambiguous. In a living animal, microbiota are an important part of many physiologic mechanisms, including contribution to a protective skin barrier and digestion. An example for the latter is hindgut fermentation found in diverse mammalian species. Herbivorous lagomorphs, such as rabbits, have a large cecum that utilizes hindgut fermentation similar to that in equids, such as horses, or other small mammals, such as guinea pigs. Rabbits provide an interesting model to study microbiota involved in hindgut fermentation because of their coprophagous behavior, which plays a vital role in bacterial colonization of the cecum. At age 14 d, the rabbit cecal bacterial community is dominated by phylum Bacteroidetes, family Bacteroidaceae, genus Bacteroides. With increasing age, phylum Firmicutes, families Lachnospiraceae and Ruminococcaceae, become the dominant taxa.¹¹ Impairment of fecal ingestion delays this ecologic succession, with greater relative abundance of Bacteroidaceae and lower relative abundance of Ruminococcaceae at age 35 d.¹¹ The delicate balance of hindgut microbiota can be easily disrupted by administration of various antimicrobials, such as clindamycin, lincomycin, erythromycin, ampicillin, amoxicillin–clavulanic acid, or cephalosporins (https://www.merckvetmanual.com/exotic-and-laboratory-animals/rabbits/management-of-rabbits).²¹ The resultant upset in the balance of microbial populations in the cecum of rabbits can lead to bacterial overgrowth and secondary gastrointestinal disease.¹

The thanatomicrobiome (thanatos, Greek for death) is the microbiome characteristic of the microbiota inhabiting organs, tissues, and body orifices following death.²⁴ The thanatomicrobiome should not be confused with the necrobiome of epinecrotic microbial communities. The necrobiome is defined as the genetic material of the multitude of organisms that begin to colonize a dead body from the environment. Of course, ultimately, the host’s preexisting or resident microbiota will mix and eventually merge with epinecrotic microbial communities. The changes in the composition of microbial communities and their microbiome after death is bound to affect results from microbial testing in the diagnostic setting and may allow extrapolation of time of death in forensic cases.^19,22,26,27 For example, sampling of internal organs of human cadavers at multiple intervals after death ranging from hours to months revealed a shift from aerobic bacteria to anaerobic bacteria in most body sites sampled, and demonstrated variation in the community structure between bodies, between sample sites within a body, and over time within a sample site.²⁵

In addition to changes in microbiota in specific topographic locations within the body, it has been suggested that bacteria translocate from one organ to other organ(s) after death,²⁰ whether locally to adjacent organs or to distant sites via vascular channels or another unknown mechanism(s). Postmortem changes in the intestinal microbiota have been studied in mice²⁰; intestinal bacteria were identified at defined times postmortem in kidney and other extra-intestinal organs using culture in thioglycollate enrichment broth, and the frequencies of culture-positive organs were compared. The authors²⁰ interpreted culture of Lactobacillus spp., Enterococcus spp., E. coli, Bacteroides/Prevotella spp., and Clostridium spp., from extra-intestinal locations including mesenteric lymph nodes and kidney as early as 5 min post-euthanasia, as postmortem translocation of anaerobes from the intestine.

Limited information is available about the changes in the postmortem microbiota in domestic animals, and about the microbiome of extra-intestinal body sites in rabbits in health, disease, or postmortem. Furthermore, there is a need for systematic studies that investigate the effect of environmental factors on postmortem changes of the microbiota of internal organs in healthy individuals. We characterized the thanatomicrobiome of internal organs typically collected for postmortem microbiologic testing and determined the effect of time and environmental temperature on the thanatomicrobiome, using the rabbit as a model. We posited that environmental temperature and length of time postmortem affect the qualitative and quantitative composition of the microbiome of internal organs, and that bacteria typically encountered in the large intestine can be identified in extra-intestinal sites postmortem, especially at higher environmental temperatures and at later times.

Materials and methods

Animals

Twenty Dutch rabbits from 10 different litters born and reared at a private farm in the Willamette Valley, OR, were culled for reasons unrelated to the study and donated for this experiment. The procedures were approved by the Institutional Animal Care and Use Committee at Oregon State University. Immediately after euthanasia by carbon dioxide inhalation, the bodies were transferred to the Oregon Veterinary Diagnostic Laboratory (OVDL; Corvallis, OR) either chilled on wet ice (n = 8) in insulated boxes, or in open containers (n = 12); bodies arrived at the laboratory within 10 min of death. Arrival at the laboratory was set as experimental time 0 h. Rabbits had been fed twice a day (Rabbit show formula 15%, Heinold Feed Mill, Kouts, IN). Hay had been available for roughage. The 10 litters were 30–62 d old (mean 45 d; median 40 d). Litters had been weaned at 3–4 wk, depending on litter size, and had begun to eat adult rabbit chow at 3 wk. Rabbits had received neither medications nor vaccinations or prophylactic treatments. After weaning, litters had been kept in their birth cages. Body weight of the rabbits at time of culling was 573–978 g (mean 802 g; median 699 g); 65% were male. Within litters, sexes, and body weights, the bodies were randomly assigned to 5 groups of 4 rabbits each such that no 2 rabbits in each group were from the same litter (Fig. 1). One group (group 1) was sampled at 0 h. Two groups each were held at 4°C and samples collected at 6 h (group 2) and 48 h postmortem (group 3). Two groups were held at 21°C and samples collected at 6 h (group 4) and 48 h (group 5). All rabbits not sampled at 0 h were stored in right lateral recumbency at their respective temperatures in open containers.

Figure 1.

Experimental design; 4 rabbits were assigned to each group.

Light microscopy

Histopathologic findings were recorded to account for the potential contribution of subclinical infections or other conditions to molecular test results. A complete set of tissues was collected for microscopic analysis in 10% formalin at a 10:1 formalin-to-tissue ratio. Tissues were fixed for a minimum of 24 h at room temperature and processed routinely. Sections were stained with hematoxylin and eosin, and analyzed by a board-certified veterinary pathologist (CV Löhr).

Sampling for molecular analyses

Sterile samples were collected in the following order: opening of thoracic cavity and collection of lung; opening of the abdominal cavity and collection of left kidney, ileum, and cecum. Tissue samples were grasped with forceps and, to avoid surface contamination, the grasped end of the tissue was discarded, so that mincing would occur at the opposite non-grasped end of the sample. For samples of intestine, the gut wall and adherent feces were collected. Each organ sample was minced into 8 pieces, ~4 × 4 mm each, using fresh scalpel blades for each tissue. Minced tissues were placed into individual microcentrifuge tubes, flash-frozen in liquid nitrogen, and stored in a −80°C freezer until molecular analyses.

DNA extraction and microbiome sequencing

DNA from lung, kidney, ileum, and cecum tissue was isolated (QIAamp DNA stool mini kit; Qiagen, Germantown, MD) following the manufacturer’s protocol with addition of a 10-min incubation at 90°C for better lysis of bacteria. In brief, frozen tissues were weighed and lysed in 1.4 mL of buffer ASL from the QIAamp kit and homogenized with 2.8-mm ceramic beads, followed by 0.5-mm glass beads (OMNI bead ruptor; OMNI International, Kennesaw, GA). Total DNA was quantified (Bioanalyzer 2100 nano chip; Agilent Technologies, Santa Clara, CA). Quantification of bacterial DNA was established by quantitative PCR (Fast SYBR mix, Qiagen) and universal 16S ribosomal RNA (rRNA) bacterial primers UniF1047-1067 (5’-GTGSTGCAYGGYTGTCGTCA) and UniR1174-1194 (ACGTCRTCCMCACCTTCCTC), with nucleotide numbers referring to the location in the 16S rRNA gene of Escherichia coli.²⁸ Cycle parameters were as follows: melting 94°C for 30 s, annealing/extension 60°C for 30 s, for 40 cycles. Standard curves derived from dilutions of bacterial DNA from a 24-h bacterial culture of a fecal sample from a healthy mouse were used to calculate DNA amounts in each sample.¹⁸ Bacterial DNA abundance over time and temperature, as measured by reverse-transcription quantitative PCR (RT-qPCR), were analyzed using 2-way ANOVA with 0 h time point not included in the analysis because it could not be assigned to either the 4°C or to the 20°C groups. However, the 0 h time was included in the analysis of overall bacterial composition.

For sequencing (MiSeq; Illumina, San Diego, CA), total genomic DNA was subjected to PCR amplification targeting the 16S rRNA variable region 4 (V4) using the modified bacterial primers 515F and 806R (http://www.earthmicrobiome.org/emp-standard-protocols/16s/).⁷ Primer sequences without barcode or adapter were forward primer GTGCCAGCMGCCGCGGTAA and reverse primer GGACTACHVGGGTWTCTAAT.⁸ PCR reactions contained PCR-grade water (MilliporeSigma, St. Louis, MO; or MoBio; Qiagen), and master mix (GoTaq green master mix; Promega, Madison, WI), with a final master mix concentration of 1× and final primer concentration of 0.4 µM. Amplification was carried out in 96-well plates at melting 94°C for 45 s, annealing 57°C for 60 s, and extension 72°C for 90 s, for 32 cycles.^7,17

PCR reactions were performed in triplicate for samples of ileum (plus digesta), kidney, and lung tissue, and in duplicate for samples of cecal tissue (and feces). Each set of duplicate and triplicate PCR reactions was pooled for each tissue separately. PCR amplification products were purified for ileum, kidney, and lung samples (MinElute PCR purification kit; Qiagen), and for cecum samples (QIAquick PCR purification kit; Qiagen).

The pools were checked for expected band size of 382 bp (according to reference gene of E. coli)⁸ by gel electrophoresis (2% agarose pre-cast E-gels; Invitrogen, Thermo Fisher Scientific, Waltham, MA) and were visualized under ultraviolet light. Negative controls consisted of water samples without template for DNA extraction and PCR amplification. The pools were then quantified (Qubit dsDNA BR assay kit; Qiagen).

Amplicons were barcoded at the reverse primer with pre-designed 12-bp barcodes as previously described and pooled at equal volumes and concentrations (2 µL of 10 ng/µL DNA).^7,8 The total pool was sequenced (MiSeq; Illumina) at the Center for Genome Research and Biocomputing at Oregon State University (Corvallis, OR) to generate pair-ended 250 nt reads.

Bioinformatics

Raw forward-end FASTQ reads from the Illumina sequencing output were quality-filtered, de-multiplexed, and analyzed using quantitative insights into microbial ecology (QIIME).⁶ Reads were quality-filtered using default QIIME parameters; reads were discarded in the case of a Phred quality score of <20, ambiguous base calls, or <187 nucleotides (75% of 250 nucleotides) of consecutive high-quality base calls. Additionally, truncation occurred on reads with 3 consecutive low-quality bases. The samples were de-multiplexed using 12-bp barcodes, allowing for a maximum of 1.5 errors in the barcode. Reads were clustered using UCLUST at 97% similarity into operational taxonomic units (OTUs) at QIIME default parameters.¹⁴ A representative set of sequences from each OTU was used for taxonomic identification by selecting the cluster seeds (first read assigned to that OTU). Representative sequences for each OTU were aligned using BLAST (e-value < 0.001) to Greengenes (v.13.8, http://greengenes.secondgenome.com/downloads) OTU reference sequences (97% similarity) to obtain taxonomy assignments.¹² OTUs were filtered for singletons, which were only found in one sample, and relative abundance was quantified by dividing raw read counts by total number of reads for each sample. Principal coordinate analyses and summary bar charts were created using QIIME. Absolute abundance quantification was performed by multiplying relative abundances of OTUs by the total amount of bacterial DNA in each sample quantified by qPCR.

Results

Histopathology

Histopathologic findings consistent with subclinical infections by Encephalitozoon cuniculi were identified in 5 of 20 rabbits and characterized by mild multifocal lymphohistiocytic encephalitis with astrogliosis, mild multifocal lymphoplasmacytic-to-histiocytic interstitial nephritis, and mild interstitial lymphoplasmacytic pneumonia. Mild intestinal coccidiosis was observed in 4 of 20 rabbits and was limited to the small intestines. Rare nematodes diagnosed as juvenile stages of Passalurus ambiguus were present in the mucosa of the small intestine and cecum of 4 of 20 rabbits.

Bacterial DNA in tissues

As expected, bacterial DNA was most abundant in the cecum, followed by ileum, and low in kidney and lung (Fig. 2). In a few kidney and lung samples, bacterial DNA was too low to be quantified. Bacterial DNA abundance was generally higher at later times compared to 0 h in all tissues except in the lungs; it was not possible to quantify bacterial DNA abundance in 3 of 4 lung samples at 0 and 48 h (20°C; Fig. 2). In the cecum (Fig. 2A, Supplementary Table 1), there was no statistically significant difference for bacterial DNA abundance in samples held at different environmental temperatures for different lengths of time. In the ileum (Fig. 2B, Supplementary Table 2), the difference between times was not significant, likely because of high variability within the groups. There was, however, higher bacterial DNA abundance at 4°C compared to 20°C (p < 0.03). In the kidney (Fig. 2C, Supplementary Table 3), there was an interaction between the effect of time and temperature (p < 0.0004), with higher bacterial DNA amounts at the later time at 4°C. Samples held at 20°C had consistently lower bacterial DNA at both the 6 and 48 h times. Interestingly, in the lungs (Fig. 2D, Supplementary Table 4), bacterial DNA abundance was higher at earlier times, at both temperatures (p < 0.02).

Figure 2.

Abundance of bacterial DNA, as determined by qPCR, over time (0 h, 6 h, or 48 h) and temperature (rabbits at experimental time 0 h = 10 min after euthanasia); or held at 4°C or 20°C postmortem. A. cecum; B. ileum; C. kidney; D. lung. Bars represent median values; circles, squares, and triangles are individual samples. p ≤ 0.05 in comparisons by 2-way ANOVA is indicated by asterisks; exact numbers are given in the text.

Microbiome composition

Principal coordinate analysis of OTUs using binary Pearson distances showed distinct clustering of lung and kidney versus cecum and ileum (Fig. 3). The separation between these 2 groups of organs on principal coordinate 1 represented 12.8% of the variation in microbial relative abundances in these samples. This separation was independent of time. Lung and kidney samples had more reads that were not matched to an OTU when compared to cecum and ileum samples. The relative abundance of bacterial phyla showed a large presence of proteobacteria in several lung and kidney samples that was not observed in cecum and ileum samples (Fig. 4).

Figure 3.

Principal coordinate analysis plot using binary Pearson distances demonstrates a clear separation by tissue type at all time points (i.e., intestinal tissues [cecum, ileum] are located together and apart from extra-intestinal organs [lung, kidney]). PC1, PC2 = principal coordinate 1 and 2, respectively.

Figure 4.

Phylum-level relative abundance in a side-by-side comparison. A large presence of proteobacteria was identified in several lung and kidney samples and not observed in cecum and ileum samples. Pale blue wedges indicate increasing postmortem time interval for each organ.

Absolute abundances of bacterial phyla and genera were calculated by multiplying bacterial DNA amounts by relative abundances of OTUs (Fig. 5, Supplementary Figs. 1–3). The cecal microbiome, which was most abundant, was dominated by the phyla Firmicutes followed by Bacteroidetes, and at the order rank by Clostridiales across all times and temperatures (Fig. 5). These taxa also dominated the ileal microbiome, although low abundance of some reads made the results quite variable across examined rabbits and environmental conditions (Supplementary Fig. 1). The same taxa were also present in kidney and lung samples, albeit in much lower abundance (Supplementary Figs. 2, 3). The extra-intestinal thanatomicrobiome mostly contained different genera compared to those in the gut. The thanatomicrobiome abundance and composition differed across environmental temperatures and times with the exception of the cecum. The differences were most striking in the kidney (Supplementary Fig. 2) and lung (Supplementary Fig. 3). Proteus spp. were present in all lung samples at 6 h at 4°C. A high abundance of Proteus spp. was also seen in kidney, however, at the higher environmental temperature (6 h, 20°C) or at a later time (48 h, 4°C). No significant differences were noted in cecal microbiome composition over time or across temperatures (Fig. 5). Overall abundance of bacteria was too variable in the ileum to identify a distinct pattern in the composition across time and temperature.

Figure 5.

Absolute abundances of bacterial microbiota, as determined by 16S rRNA sequencing, in cecum postmortem over time (0 h, 6 h, or 48 h) and temperature (4°C or 20°C). The top 10 genera are shown according to average abundance. The top graph shows phyla and corresponding legend to the right of the graph, and the bottom graph shows genera and corresponding legend below the graph.

Differences in microbiome composition could not be explained by the identified subclinical microscopic lesions. For example, the rabbit lung sample (0 h) with larger amounts of bacterial DNA, in particular of Proteobacteria, Proteus spp., than the other lung samples at this time, did not show any noticeable pneumonia (Supplementary Fig. 3). Similarly, there was no histologic evidence of bacterial overgrowth or dysbacteriosis in the ileum samples that had a higher abundance of phyla Firmicutes, Bacteroidetes, and Verrucomicrobia than the other samples collected at 6 h (both at 4°C and 20°C; Supplementary Fig. 1).

Discussion

Our study provides a systematic examination of the thanatomicrobiome under environmental conditions, relevant to veterinary postmortem testing in a model species, seen both in the diagnostic and experimental pathology setting. We focused on a diagnostically relevant postmortem interval (PMI) of up to 48 h, similar to the autopsy delay over a weekend. Our results also provide insight into the microbiome of the cecum, ileum, kidney, and lung of clinically healthy juvenile rabbits shortly after weaning. Our study was designed such that we sampled different individuals at each time and temperature rather than repeatedly sampling the same bodies, with the explicit intent to avoid exposure of internal organs to environmental oxygen and potentially associated shift in microbiota.

As posited, microbiota were more similar between cecum and ileum versus kidney and lung. In the digestive tract of the rabbits in our study, the majority (83.3–99.7%) of the identifiable sequences were classified as either phylum Bacteroidetes or Firmicutes. Within Firmicutes, the highest proportions were families Lachnospiraceae and Ruminococcaceae. Our findings are consistent with most published reports on the cecal microbiome of juvenile and adult rabbits,^2,11 and are similar to those observed in the large intestine of healthy humans.³² Coprophagous behavior of rabbits results in successive cecal implantation by Bacteroidetes, predominantly by family Bacteroidaceae, at 2 wk, and by Firmicutes, largely Lachnospiraceae and Ruminoccocaceae, at 2.5 mo of age.¹¹ This presents a shift from simple but unstable microbiota to complex and stable microbiota.¹⁰ Interestingly, the microbiota in the rabbits in our study, at the ages of 30–70 d, was more similar to that reported for 2.5-mo-old rabbits than that of 2-wk-old rabbits.¹¹ The predominance of Firmicutes and Bacteroidetes in the cecum and ileum continued in the PMI, independent of time or temperature. Similar observations were made in decomposing human bodies, wherein these phyla predominated in the large intestinal postmortem microbiome.^5,19 In the cecum of deceased and decomposing human subjects, Bacteroidetes decline and Clostridiales increase at later times.¹³ Neither the relative abundance of microbiome populations nor bacterial DNA amounts in the rabbit cecum postmortem significantly changed with the examined times and temperatures. This may be a reflection of the characteristics of hindgut-fermenting herbivores, in that ingesta in the cecum will continue to undergo bacterial fermentation after death, with formation of heat, gas, lactate, and short fatty acids.^18,23 Perhaps the physiologically oxygen-poor environment allows for fermentation to continue postmortem, and helps to maintain the same relative abundance in the microbiome profile. The similarities in bacterial populations between the ileum and cecum are likely a result of their close association anatomically, and underline the need for samples from more orad or proximal segments of the small intestine for accurate microbial testing of small intestinal diseases not specific to the ileum. Higher microbial abundance, especially Firmicutes, in the ileum of some rabbits kept at 4°C cannot be readily explained and may have been the result of differences in nutrient availability.

As expected, bacterial DNA abundance was low in kidney and lung and higher in cecum than ileum of the clinically healthy rabbits in our study. Although long considered sterile, bacterial DNA has been identified in lungs in previous studies. Low bacterial loads calculated at 10–100 bacterial cells per 1,000 host cells,³² whether continually introduced (from the upper respiratory tract, oral cavity, or the environment) or resident, are considered physiologic in the lower respiratory tract of people.^3,9 Bacterial communities identified in upper airways of healthy humans largely overlap with those of the upper digestive tract.⁹ Unsurprisingly, lung had the highest proportion of sequences that either belonged to very low abundance phyla or could not be matched to published sequences, primarily because of low bacterial DNA yields. Low bacterial DNA yields increased the likelihood of obtaining nonspecific amplifications of host DNA during 16S rRNA PCR amplification, yielding sequences that could not be aligned to bacterial genomes.

In the rabbits in our study, the relative abundance of phyla was similar between the lung and kidney at early times, with the exception of Bacteroidetes and Proteobacteria, which were more abundant in kidney and lung, respectively. One might speculate that this is, in part, because of direct exposure of the respiratory tract to atmospheric oxygen, air pollutants, and potentially more rapid cooling of these organs after death.

Bacterial overgrowth in tissues and possible translocation from intestine to extra-intestinal sites after death is a concern both in routine and forensic postmortem examinations. From an experimental standpoint, it proves challenging to reliably differentiate postmortem from in vivo bacterial translocation, such as reported in human patients with cirrhotic livers.³³ There are few studies that have systematically examined this phenomenon in an experimental model.^4,20 Our study does not provide evidence that microbiota common in the intestine are present in detectable quantities in the extra-intestinal site in the first 48 h postmortem. We observed more Proteus spp. in lung at 6 h at 4°C than at 0 or 48 h. Proteus spp. were also more abundant in kidney, albeit at the higher temperature (6 h, 20°C) or later times (48 h, 4°C). The gastrointestinal tract is known to harbor Proteus spp. as part of the normal microbiota,³¹ even though Proteus spp. sequences were not detected in the examined samples from the ileum or cecum of the rabbits in our study. Certain Proteus spp. are known to have swarming characteristics,³⁰ which could facilitate bacterial migration (translocation), especially along anatomic conduits such as ducts and vessels. However, one would expect translocation of bacteria from the intestine to extra-intestinal tissue to affect tissues in close proximity (adjacent kidney) earlier than more distant sites (lung). The observed pattern in our study is more consistent with presence of Proteus spp. locally in the lung and kidney rather than bacterial translocation from the gastrointestinal tract to the lung or kidney, or from the lung to the kidney. The lesions identified on microscopic examination of tissues were well within the expected range of incidental findings in nonspecific pathogen–free rabbits and could not explain the observed differences in microbial communities.

Limitations of our study include the small population size, the few times analyzed, non-serial sampling, and the low bacterial DNA yield of certain tissues (i.e., kidney and lung). These factors may have impacted the characterization of tissue-specific microbiota.^15,34 Genus-level identification of OTUs was not always possible given a variety of reasons, such as too few high-quality reads, or highly similar 16S rRNA sequences in the phylogenetic group.¹⁶ Therefore, some of the analyses had to be performed at higher phylogenetic levels. Another possible limitation could be that sampling was performed on different individuals at each time and temperature rather than repeat sampling of the same bodies to avoid exposure of internal organs to room air. This experimental design likely contributed to the high variability in bacterial DNA abundance, most notably in the ileum at 4°C, and may have obscured statistical significance of observed differences.

There was clear separation of the 4 examined tissues over all times and temperatures based on relative abundance of the respective microbiomes. Based on absolute bacterial DNA abundance, only cecal bacteria had sustained postmortem growth in all conditions. Interestingly, several bacterial phyla were much more abundant at 4°C than at 20°C in the 2 extra-intestinal, internal organs (kidney and lung), whereas abundance of bacterial phyla was similar at both temperatures in the intestines. Our results show that the microbiome not only changes after death, as previously reported,^20,22 but, more importantly, that these alterations are organ- and temperature-specific. Altogether, our data suggest that the analysis of microbiomes by16S rRNA gene sequencing provides a true snapshot of the status quo at time of death only when samples or bodies are frozen in the immediate PMI to arrest local microbiota. Our findings provide evidence that refrigeration does not prevent bacterial outgrowth over time and, more importantly, may lead, in some tissues, to temperature-specific changes in the microbiome and presumably microbiota in the first 48 h postmortem. Our results provide insights into the thanatomicrobiome of a model species in the diagnostically relevant PMI, and may inform the interpretation of results from diagnostic and forensic microbiology testing.