Microbial communities in the liver and brain are informative for postmortem submersion interval estimation in the late phase of decomposition: A study in mouse cadavers recovered from freshwater

Abstract

Introduction: Bodies recovered from water, especially in the late phase of decomposition, pose difficulties to the investigating authorities. Various methods have been proposed for postmortem submersion interval (PMSI) estimation and drowning identification, but some limitations remain. Many recent studies have proved the value of microbiota succession in viscera for postmortem interval estimation. Nevertheless, the visceral microbiota succession and its application for PMSI estimation and drowning identification require further investigation.

Methods: In the current study, mouse drowning and CO₂ asphyxia models were developed, and cadavers were immersed in freshwater for 0 to 14 days. Microbial communities in the liver and brain were characterized via 16S rDNA high-throughput sequencing.

Results: Only livers and brains collected from 5 to 14 days postmortem were qualified for sequencing. There was significant variation between microbiota from liver and brain. Differences in microbiota between the cadavers of mice that had drowned and those only subjected to postmortem submersion decreased over the PMSI. Significant successions in microbial communities were observed among the different subgroups within the late phase of the PMSI in livers and brains. Eighteen taxa in the liver which were mainly related to Clostridium_sensu_stricto and Aeromonas, and 26 taxa in the brain which were mainly belonged to Clostridium_sensu_stricto, Acetobacteroides, and Limnochorda, were selected as potential biomarkers for PMSI estimation based on a random forest algorithm. The PMSI estimation models established yielded accurate prediction results with mean absolute errors ± the standard error of 1.282 ± 0.189 d for the liver and 0.989 ± 0.237 d for the brain.

Conclusions: The present study provides novel information on visceral postmortem microbiota succession in corpses submerged in freshwater which sheds new light on PMSI estimation based on the liver and brain in forensic practice.

Keywords: aquatic habitat; decomposition; internal organ; microbial community; postmortem submersion interval.

Figure 1

Composition of microbial communities in internal organs and microbial source tracking. (A–C) Relative abundance of bacterial taxa at different taxonomic levels. Stacked bar charts of the top 6 bacterial phyla (A), top 10 bacterial families (B), and top 10 bacterial genera (C) with the largest mean relative abundance in the liver and brain. Percentage source contributions of water-derived and intestine-derived bacteria to microbial communities in liver (D), and brain (E), over time were determined using FEAST. D, drowning group; PS, postmortem submersion group.

Figure 2

Alpha and beta diversities of microbiota in the liver and brain. Comparisons of the Chao1 (A), and Shannon (B), indices between drowning and postmortem groups at each PMSI. Ordination plot for the first two PCoA axes based on unweighted Unifrac (C), and Bray–Curtis (D), distances. Different colors indicate different PMSIs. Different sample types (liver or brain) are represented by different shapes. D, drowning group; PS, postmortem submersion group.

Figure 3

(A,B) Performance of the RF classification model built on microbiota in the liver. (C,D) Performance of the RF classification model built on microbiota in the brain. (A,C) MDS plot generated by the learning algorithm RF comparing the microbial community between drowning and postmortem submersion groups. (B,D) ROC curves of the RF classification model on data from exploratory and validation experiments. D, drowning group; PS, postmortem submersion group.

Figure 4

Successional dynamics of microbial communities in the liver and brain, and performances of regression models for PMSI estimation. PCoA of bacterial communities in liver (A), and brain (B). Different colors indicate different PMSIs. Predicted PMSI versus actual PMSI for liver (C), and brain (D), samples were plotted with a superimposed one-to-one reference line. Dots represent samples from the validation experiment (n = 4 per PMSI).

Figure 5

Liver biomarker identification and validation for PMSI estimation. (A) Cross-validation results of the initial model established using liver microbial communities. (B) The top 18 ASVs were identified by the RF algorithm. Biomarker taxa were ranked in decreasing order of importance (i.e., %IncMSE). (C) Heatmap demonstrating dynamic changes in abundance of the top 18 PMSI-predictive biomarkers. (D) Predicted PMSI versus actual PMSI for liver samples obtained by the refined regression model plotted with a superimposed one-to-one reference line. The dots represent liver samples from the validation experiment (n = 4 per PMSI).

Figure 6

Brain biomarker identification and validation for PMSI estimation. (A) Cross-validation result of the initial model established using brain microbial communities. (B) The top 26 ASVs were identified by the RF algorithm. Biomarker taxa were ranked in decreasing order of importance (i.e., %IncMSE). (C) Heatmap demonstrating dynamic changes in abundance of the top 26 PMSI-predictive biomarkers. (D) Predicted PMSI versus actual PMSI for brain samples obtained by the refined regression model plotted with a superimposed one-to-one reference line. Dots represent brain samples from the validation experiment (n = 4 per PMSI).