1. Microbial variation in internal organs after death
1.1 Relative abundance profiles of the microbial community during 1 day of decomposition
120 organ samples, including those from the brain, heart, liver, and kidneys, were collected from 30 mouse remains of 5 timepoints over 1 day of decomposition during 1 day of decomposition. In our study, no valid bacterial sequences were detected in the negative controls, indicating that the sequencing data are reliable. To analyze microbial community succession in every organ, the microbial community composition profiles in the four organs were described separately.
In the brain samples, at the genus level (Fig. 1A and Fig. S1A), Ochrobactrum (10.37%~18.54%) and Sediminibacterium (3.23%~17.74%) were dominant and showed a similar decreasing relative abundance profile along PMI progression. Acinetobacter, Cupriavidus, and Agrobacterium showed increasing abundance profiles. In particular, the relative abundance of Agrobacterium significantly increased (11.99% ± 0.41%) at 8 hours compared with 0.5 hours (4.32% ± 0.28%, P = 0.004) and 4 hours (4.02% ± 0.79%, P = 0.004) (P < 0.01, KW-test, Fig. 2A). At the phylum level (Fig. S2A), Proteobacteria (54.06%~87.80%) was the most prevalent in all brain samples. The relative abundance of Bacteroidetes was higher in the H0.5Brain (19.31%±0.81%) and H4Brain (19.63%±3.27%) samples than in the samples from the other three groups (the P values of the six paired comparison groups were all 0.000, P<0.001, LSD t- test). Proteobacteria showed different succession structures during decomposition. In addition, the relative abundances of Cyanobacteria and Thermi illustrated an increase during 12 hours of decomposition. At the order level (Fig. S2A), Rhizobiales (21.97%~31.65%) was dominant in all brain samples. The relative abundances of Saprospirales, Caulobacterales, and Thermales decreased, while those of Burkholderiales and Pseudomonadales showed increasing profiles during 1 day of decomposition. At the species level (Fig. S2A), the relative abundance of Bifidobacterium longum showed a peak value at hour 4 after death. Acinetobacter johnsonii showed a peak value at hour 12 after death. The abundances of Deinococcus geothermalis and Sphingomonas azotifigens abundances increased during decomposition.
In the heart samples, at the genus level (Fig. 1B and Fig. S1B), Thermus (14.52%~25.12%) was more abundant than the other genera in all the heart sample groups. The relative abundances of Enhydrobacter, Caulobacter, and Methyloversatilis gradually decreased during 1 day of decomposition. However, the relative abundance of Pseudomonas increased to 11.38% at 8 hours after death. The relative abundances of Sphingomonas and Cupriavidus increased to peak values of 7.93% and 12.58%, respectively, at 12 hours after death. At the phylum level (Fig. S2), Proteobacteria (47.49%~67.28%) and Thermi (16.70%~27.90%) were dominant in the early postmortem heart samples. The relative abundance of Firmicutes gradually increased during 1 day of decomposition, while that of Actinobacteria decreased. At the order level (Fig. S2B), Pseudomonadales (15.21%~27.57%), Thermales (14.52%~25.12%), and Burkholderiales(15.41%~20.26%) were dominant in all heart samples. The relative abundance of Sphingomonadales increased to a peak value of 8.96% at 12 hours after death. Rhizobiales showed a gradual increasing abundance profile during 1 day of decomposition. Furthermore, the relative abundance of Deinococcales increased to 4.72% at 12 hours after death. However, the relative abundance of Rhodocyclales, Rhodospirillales, and Caulobacterales decreased during 1 day of decomposition. At the species level (Fig. S2B), Pseudomonas viridiflava, Sphingomonas azotifigens, and Deinococcus geothermalis were dominant in all the postmortem heart samples.
In the liver samples, at the genus level (Fig. 1C and Fig. S1C), Thermus (16.83%~24.22%) and Cupriavidus (11.40%~14.35) were dominant in all the postmortem liver sample groups. The relative abundance of Microbacterium gradually decreased to zero percent at 24 hours after death. In contrast, the relative abundances of Acinetobacter, Cupriavidus, and Pseudomonas gradually increased during decomposition. The genera Paracoccus and Cryocola were detected only half an hour after death. Prevotella showed a significant increase in relative abundance at 4 hours (1.14% ± 0.57%) compared with that at 0.5 hours (0.44% ± 0.38%) (P=0.037, P < 0.05, KW-test, Fig. 2B). At the phylum level (Fig. S2C), Proteobacteria (48.29%~62.36%) and Thermi (18.79%~25.39%) were dominant in all the liver sample groups. Actinobacteria, Firmicutes, Bacteroidetes, and Cyanobacteria showed relative abundances of more than 1% in all the liver sample groups. Among these phyla, Actinobacteria gradually decreased in relative abundance during 1 day of decomposition. At the order level (Fig. S2C), Burkholderiales (15.25%~20.72%), Pseudomonadales (14.59%~27.66%), and Thermales (16.83%~24.22%) were dominant in all the liver sample groups. The relative abundance of Clostridiales gradually increased during 1 day of decomposition, while that of Actinomycetales decreased during decomposition. Rhodobacterales immediately decreased in relative abundance before 4 hours after death. At the species level (Fig. S2C), Pseudomonas viridiflava, Sphingomonas azotifigens, and Sphingomonas azotifigens were dominant in all the liver sample groups. Paracoccus marcusii decreased in relative abundance before 4 hours after death.
In the kidney samples, at the genus level (Fig. 1D and Fig. S1D), Thermus (7.40%~31.59%) was dominant. The relative abundances of Acinetobacter and Pseudomonas increased during 8 hours of decomposition. The relative abundance of Methyloversatilis decreased during 1 day of decomposition. Bacillus was significantly increased at 4 hours compared with 0.5 hours (P=0.045, P <0.05, KW-test, Fig. 2C). Turicibacter exhibited a significantly higher relative abundance in the H24Kidney group (1.33% ± 0.51%) than in the other kidney sample groups (v.s. H0.5; P = 0.038; v.s. H4, P=0.031; v.s. H8, P=0.013; v.s. H12, P=0.021, P <0.05, KW-test, Fig. 2D). At the phylum level (Fig. S2D), Proteobacteria (41.37%~60.51%), Thermi (8.00%~32.14%), and Firmicutes (7.35%~10.79%) were dominant in all the postmortem kidney sample groups. The relative abundances of Fusobacteria and Cyanobacteria gradually decreased during 1 day of decomposition, while those of Proteobacteria and Actinobacteria gradually increased during decomposition. At the order level (Fig. S2D), Pseudomonadales (11.94%~22.23%) and Thermales (7.40%~31.59%) were dominant orders in all the kidney sample groups. During 1 day of decomposition, the relative abundances of Streptophyta, Clostridiales, and Rhodocyclales gradually decreased. However, the abundances of Burkholderiales, Rhizobiales, Bacteroidales, and Actinomycetales gradually increased during this decomposition period. At the species level (Fig. S2D), Paracoccus marcusii and Lactobacillus reuteri had an increasing abundance profile across the 12 hours after death.
1.2 Comparison of alpha and beta diversity at different timepoints
Alpha diversity was measured by the Shannon index, Pielou’s evenness, Good’s coverage index, observed species index, Faith’s PD index, and Chao1 index in this work. Samples in the H8Brain and H24Brain groups had significantly lower alpha diversity (measured as Shannon diversity, H8 v.s. H0.5, P=0.035; H24 v.s. H0.5, P=0.003 (Fig. 4A), Pielou’s evenness, H8 v.s. H0.5, P=0.009; H24 v.s. H0.5, P=0.046 (Fig. S3A), and Good’ s coverage index H8 v.s. H0.5, P=0.028; H24 v.s. H0.5, P=0.035 (Fig. S3C)) than those in the H0.5Brain group (P<0.05, KW-test). Faith’s PD index, indicating the phylogenetic distances of OTUs in a sample, had a significantly lower value in the H0.5Brain group than in the H8Brain (P=0.0015), H12Brain (P=0.014), and H24Brain (P=0.0024) groups (P<0.05, LSD t-test, Fig. S3B). Faith’s PD index was significantly lower in the H24Liver group than in the H4Liver (P=0.009) and H8Liver (P=0.003) groups (P<0.01, Dunnett T3 test, Fig. S3D). According to a comparison of alpha diversity in the kidney groups, the observed species index demonstrated a significantly lower value in the H12Kidney group than in the H4Kidney (P=0.01) and H8Kidney (P=0.000) groups (P < 0.01, Dunnett T3 test, Fig. S3E). The samples in the H12Kidney group showed a significantly lower value of Chao 1 index than those in the H4Kidney (P=0.01) and H8Kidney (P=0.000) groups (P < 0.01, Dunnett T3 test, Fig. S3F).
To visualize the similarities and dissimilarities in community composition between samples, as a measure of beta diversity, PCoA plots were calculated based on the weighted UniFrac dissimilarity index. Overall, PCoA of brain samples revealed two distinct clusters (the PCoA explained 42.8% of the variation): one including samples of the H0.5Brain and H4Brain groups and another including samples from the H8Brain, H12Brain, and H24Brain groups (Fig. 3E). PCoA of kidney samples showed three clusters: a cluster of samples from the H0.5Kidney, H4Kidney, and H8Kidney groups; a cluster of samples from the H12Kidney group; and a cluster of samples from the H24Kidney group (Fig. 3H). Liver and heart samples did not show significant clusters according to beta diversity analysis (Fig. 3).
To identify the taxonomic bacteria with significantly different abundances among different PMIs for internal organs, the LEfSe method was used for biomarker analysis (Fig. 4). As shown in Fig. 4A, Sediminibacterium and Ochrobactrum were representative genera in the H0.5Brain group. Lactobacillales, and Mycoplana were significant microbes in the H4Brain group. Agrobacterium was the significant genus in the H8Brain group. Deinococcus and Acinetobacter, were representative microbes in the H12Brain group. Cupriavidus and Cryocola were representative microbes in the H24Brain group.
When the taxonomic composition was compared within different decomposition stages in the heart (Fig. 4B), Phascolarctobacterium, Xanthomonadaceae, Sphingobium, and Ellin6529 were significant microbes in the H0.5Heart group; Enterobacteriaceae, Lysinibacillus, and Steroidobacter were detected as representatives in the H4Heart group; Proteobacteria was significant in the H8Heart group; Cupriavidus was regarded as a significant microbe in the H12Heart group; and Lactococcus showed significance in the H24Heart group. In regard to the liver groups, Microbacteriaceae,Comamonadaceae, and Caulobacter were significant microbes in the H0.5Liver group (Fig. 4C); Actinomycetales, PRR_12, and Solibacterales were representative microbes in the H4Liver group; Myroides, Kocuria, Citrobacter, and Streptomycetaceae were significant microbes in the H8Liver group; Gemmatimonadetes and Raphanus were representatives in the H12Liver group; and mitochondria were regarded as significant microbes in the H24Liver group. For the kidney groups (Fig. 4D), Enterobacteriaceae and Streptophyta were significant microbes in the H0.5Kidney group; Limnobacter, Sphingomonadaceae, Clostridiales, Sphingomonadaceae, Hyphomonadaceae, and Planococcaceae were significant in the H4Kidney group; Thermus, Bacteriovoracaceae, and Luteimonas were significant in the H8Kidney group; Rhodobacterales, Perlucidibaca, Bacillales, Flavobacterium, Anoxybacillus, Comamonas, and Rhodospirillaceaewere significant in the H12Kidney group; and S24_7, Oxalobacteraceae, Turicibacter, Bacteroides, Akkermansia, Cupriavidus, and Deltaproteobacteria were significant microbes in the H24Kidney group.
2. Differential postmortem microbial community structure and diversity between different organs
Alpha diversity was significantly different between the brain and the other organs (heart, kidney, and liver) (P < 0.05, Student’s t-test (LSD), KW-test, Fig. 5). The brain groups exhibited lower observed amplicon sequence variant (ASV) richness values than the other organs at half an hour, 4 hours, 8 hours, and 24 hours after death. Kidney samples showed the lowest observed ASV richness values among the other organ groups at 12 hours after death (Fig. 5). Beta diversity based on weighted UniFrac distance showed obvious clustering at the early decomposition stage according to the organ, while the distinction decreased with the decomposition process (Fig. 5). In general, the brain samples were dominated by the bacterial genera Acinetobacter, Cupriavidus, Ochrobactrum, and Sediminibacterium, while the other organs were dominated by Thermus, Enhydrobacter, and Pseudomonas (Fig. 1).
LEfSe analyses were applied to detect representative microbes among the different sample sites during decomposition at different taxonomic levels (Fig. 6). The number of significant taxa in each organ increased before 8 hours and decreased from then on, leading to no significantly differentially abundant taxa in any organ at 24 hours after death, with a linear discriminant analysis (LDA) threshold of 4 (Fig. 6). Half an hour after death (Fig. 6A), Agrobacterium, Sediminibacterium, and Ochrobactrum were detected as significantly differentially abundant genera in the H0.5Brain group; Comamonadaceae, Sphingomonas, and Sphingomonas azotifigens were detected in the H0.5Heart group at multiple levels; Pseudomonas, Cupriavidus, Deinococcus, and Cryocola were significantly differentially abundant genera in the H0.5Liver group; Pseudomonas viridiflava and Deinococcus geothermalis were detected as significantly differentially abundant species in the H0.5Liver group; and Enterobacteriaceae, Streptophyta and Methyloversatilis were significantly differentially abundant bacteria in the H0.5Kidney group. After 4 hours of decomposition (Fig. 6B), Bradyrhizobiaceae and Agrobacterium were significantly differentially abundant microbes in the H4Brain group; Pseudomonas, Acinetobacter, and Sphingomonas were detected as representative genera in the H4Heart group. In addition, Sphingomonas azotifigens, Pseudomonas viridiflava, and Deinococcus geothermalis were significantly differentially abundant species in the H4Heart group. Thermus, Limnobacter, Perlucidibaca, Methyloversatilis, and Flavobacterium were significantly differentially abundant genera in the H4Kidney group. Moraxellaceae and Cupriavidus were detected as significantly differentially abundant microbes in the H4Liver group. After 8 hours of decomposition (Fig. 6C), Cupriavidus, Ochrobactrum, Agrobacterium, Sediminibacterium, and Acinetobacter were significantly differentially abundant genera in the H8Brain group; Sphingomonas, Pseudomonas, and Deinococcus were significantly differentially abundant genera in the H8Heart group. Pseudomonas viridiflava, Sphingomonas azotifigens, and Deinococcus geothermalis were significantly differentially abundant species in the H8Heart group. In the H8Kidney group, Thermus and Methylobacterium were significantly differentially abundant genera. After 12 hours of decomposition (Fig. 6D), Acinetobacter, Cupriavidus, Acinetobacter, and Agrobacterium were detected as significantly differentially abundant genera in the H12Brain group; Thermus, Pseudomonas, Sphingomonas, and Pseudomonas were representative genera in the H12Heart group. Sphingomonas azotifigens and Pseudomonas viridiflava were detected as significantly differentially abundant species in the H12Heart group. Perlucidibaca and Flavobacterium were significantly differentially abundant genera in the H12Kidney group. We also counted the genera with continuous difference in the comparison of four organs based on the LDA effect size (LEfSe) analysis results (Table 1).
3. Metabolically active members in postmortem organs during decomposition
The metabolic pathways based on the KEGG database were used to link microbial genomic information by the PICRUSt2 algorithm with higher-order functions that were significantly altered during 1-day decomposition. To facilitate writing, the organ groups harvested before 4 hours after death were called “before 4 h”, while those harvested after 4 hours of decomposition were called as “after 4 h”. Broad classes of metabolic pathways within the samples are presented in Fig. 7. The enriched pathways of the current study were amino acid metabolism, carbohydrate metabolism, cofactor and vitamin metabolism, xenobiotic biodegradation and metabolism, other amino acid metabolism, terpenoid and polyketide metabolism, lipid metabolism, and energy metabolism, which are summarized in Fig. 7A. The top 5 relative abundance pathways were similar in the four organs. Certain pathways, including ketone body synthesis and degradation; ansamycin biosynthesis; valine, leucine, and isoleucine biosynthesis; C5-branched dibasic acid metabolism; and fatty acid biosynthesis, were dramatically enriched in “after 4 h” samples, compared to “before 4 h” samples of the brains and hearts (Fig. S4, P < 0.05, KW-test). A heatmap shows particular pathways that were differentially abundant according to the organ (Fig. 7B). According to comparisons of different organs at 0.5 hours, selenocompound metabolism, histidine metabolism, and several amino acid metabolism pathways were obviously more enriched in the brain than in the other organs. The bacterial chemotaxis and flagellar assembly pathways were depleted in kidney samples. After 4 hours of decomposition, pantothenate and CoA biosynthesis, selenocompound metabolism, and amino acid metabolism were enriched in the H4Brain group compared with in the other organ groups. After 8 hours of decomposition, the biotin, lipoic acid metabolism, carbon fixation in prokaryotes and terpenoid backbone biosynthesis pathways were depleted in brain samples compared to those in samples from the other organs. After 12 hours of decomposition, the pyruvate metabolism pathway was enriched in the kidney compared with in the other organs.