The Forensic Relevance of the Thanatotranscriptome

Nathan Routledge -12548189 1

The Forensic

Relevance of the

Thanatotranscriptome

Nathan Routledge

12548189

Masters Forensic Science

Word Count: 8493

Supervisor: Dr Margreet van den Berge

Co-Assessor: Prof.dr.ir Titia Sijen

Nathan Routledge -12548189 2

Table of Contents

3.0 Discussion 20-23

3.1 Animal versus Human based Studies 20 3.1.1 Comparison of the study types 20

3.1.2 How applicable are animal studies on

. humans 20 3.2 Confounding Factors 21 3.2.1 Intrinsic Factors 21 3.2.2 Extrinsic Factors 21 3.3 Genes 22 3.3.1 Apoptosis Genes 22 3.3.2 Tissue Specificity 22 3.4 Forensic Application 22 3.4.1 PMI 22

3.4.2 Determining Cause of Death 23 4.0 Conclusion and Recommendations 24

4.1 Conclusion 24

4.2 Recommendations 24

5.0 References 25

Nathan Routledge -12548189 3

List of Abbreviations

BMI

Body Mass Index

DNA

Deoxyribose Nucleic Acid

DD-PCR

Droplet Digital Polymerase

Chain Reaction

GTEx

Genotype-Tissue Expression

mRNA

Messenger Ribonucleic Acid

MtDNA

Mitochondrial Deoxyribose

Nucleic Acid

PCR

Polymerase Chain Reaction

PMI

Post-Mortem Interval

RNA

Ribonucleic Acid

RNA-seq

Ribonucleic Acid-sequencing

RT-PCR

Reverse

Transcriptase-Polymerase Chain Reaction

RT-qPCR

Reverse

Transcriptase-quantitative Polymerase

Chain Reaction

SDS-page

Sodium Dodecyl

Nathan Routledge -12548189 4

Abstract

The thanatotranscriptome is the collection of genes and transcripts that are in abundance after an individual die. There has been research into understanding the significance of the thanatotranscriptome concerning post-mortem analysis. One potential area is in forensic science, to determine the post-mortem interval (PMI) of a body or determine how they died. Studies have been performed on rodents, zebrafish and humans to examine how the thanatotranscriptome changes over time since death. The results showed that many genes and transcripts from the thanatotranscriptome do change in a time dependant manner, as well as being tissue specific. Animal-based studies were able to give more precise results with wider variety PMIs observed. Some human-based studies also gave detailed PMI prediction models to use based on the genes expressed. However, there is little research on how intrinsic factors e.g., age and weight, and extrinsic factors such as temperature, affect the post-mortem gene expression. As such, the potential forensic use of the

thanatotranscriptome is still a far way off.

Keywords

Nathan Routledge -12548189 5

1.0 Introduction

1.1 Research Question

This literature thesis will investigate the question of how relevant the thanatotranscriptome is in a forensic context. This thesis will examine recent literature of the

thanatotranscriptome and discuss questions such as can a PMI be made using

thanatotranscriptomics. This thesis will also evaluate known literature that show how thanatotranscriptomics could be used in a forensic context. How applicable such methods could be in cases, and the limitations of such methods will be evaluated.

1.2 Background

The thanatotranscriptome describes the abundance of RNA expression after death,

especially in the early hours to days after death, where various changes in RNA expression are seen [1]. This time is also been nicknamed the “twilight of death” where a living body changes to a decomposing corpse [2].

There are many different fields where the investigation into better understanding the thanatotranscriptome would be beneficial. One is the medical field with the better

understanding of what death is. As there are multiple definitions depending on the type of death [2]. For example, “denouement death” where homeostasis can no longer be

maintained, and this is seen as complete death. There is also threshold death, where while the organism may be still be alive, there is no way to prevent the organism from dying. Finally, there is integration death where the organism and its physiological functions have all stopped operating. All three of these definitions have flaws and incidences where a person or organism could be classified as any of the three, and as such understanding more about the process of death via the thanatotranscriptome may help improve the definition of death used in the medical field. There are also practical uses in the medical field to study the thanatotranscriptome, as organ transplantation will often come from an individual who is either dead or “brain dead” [3]. These organs will also often not be of the greatest quality due to inflammatory responses from the donor leading to the organ not fully functioning in the recipient [4]. Understanding the thanatotransciptome of that donor may be a new way to strategize a method of optimizing successful organ transplants. Understanding the thanatotranscriptome may also be advantageous in battling the high rates of cancer that is often seen in patients with organ transplants [5]. Many cancer-associated genes have been shown to express post-mortem and thus understanding how and why they do may lead to new screening or preventative methods in organ transplantation [2].

An additional field that could benefit from studying the thanatotranscriptome is the area of forensic investigation. RNA is frequently used in forensic cases to identify body fluids and specific tissues from mRNA extracted from samples found at a crime scene [6]. With prominent fluids involved in sexual assault cases such as semen, vaginal secretions, and menstrual blood being detectable. This is done via the analysis of mRNA expression of

Nathan Routledge -12548189 6 specific genes associated with the forementioned fluids, to create biomarkers that in

combination can determine a specific fluid. Another use of RNA that is being developed in the forensic field is to estimate the age of wound, via the expression of mRNA of

inflammatory cytokines and wound healing factors [7]. Monitoring the mRNA expression levels of these genes allows for the wound age to be established based on which genes are being expressed and how highly they are expressed. This is still a relatively new principle and many factors can affect the wound estimation. Finally, there have been studies already to determine the post-mortem interval by looking at the degradation of RNA in organs that slowly degrade such as the brain or heart [8]. The mRNA of genes associated with specific tissues are analysed to see how much RNA remains, and how much has been degraded over time. A very similar concept is now being researched with the thanatotranscriptome, by looking at genes that start expressing post-mortem. It is this area of research where most of the thesis will be focused on.

The current techniques used to estimate short PMIs are either based on autopsy, or via the use of body temperature, such as the commonly used Henssge’s nomogram model [9]. The model is based on analysing the difference between the core rectal temperature of the body, and the ambient temperature around it. The method is very subjective and relies on other confounding factors such as the weather, the clothes worn by the body, and the size and weight of the body. A newer method using thermodynamics and a computer-based algorithm has the potential to be more reliable and being accurate to approximately 38 minutes [10]. However, the use of these techniques relies heavily on the availability of a difference in body temperature and ambient temperature. The closer the two temperatures are, the less accurate the prediction. Therefore, there is still an opening to determine PMI without the need of temperature measurement and is why the thanatotranscriptome is now being researched.

Nathan Routledge -12548189 7

2.0 Overview of Literature

A total of 18 research articles and were reviewed in this thesis, as seen in the appendix. The articles chosen were published from 2016 onwards. The main focus of the articles is based on looking at gene expression post-mortem of various tissues in different species. Some also include the added aspect of forming a type of PMI model. The summaries of the literature are split in to two categories: animal-based studies, and human-based studies.

2.1 Animal-based Studies

In total 9 [11-19] papers focused on studying post-mortem gene expression using various animals. These papers were then compared using a variety of criteria as seen in 2.1.1 to 2.1.9

2.1.1 Animal Models and Sample size

The majority (8/9) of the animal studies used mice or rats. However, some used moths or zebrafish (see table 1). The sample size of the studies using mice or rats vary from the smallest size of 4 rats in the Martínez et al. [17], to the largest cohort coming from the Ma

et al. [11] study with 270 rats that included both males and females. Some did not give a

sample size. Hunter et al. [12] used 44 zebrafish and Shafeeq et al. [19] referred to size as a colony. The rodent studies ranged from 18-35 animals, and most of the studies used male or unspecified sex species, see table 1.

2.1.2 Sample Tissues

The breakdown of the tissue types used can be seen in table 1. The most common types were brain and liver, which were both used in 4 studies. Other tissue types such as skin, spleen and gastrocnemius muscle were also used.

2.1.3 Method of Death

Most of the rodents (6/8) were killed by cervical dislocation, but intraperitoneal injection of ketamine was also used (see table 1). The zebrafish were killed via ice water [12], and the moths died naturally [19].

2.1.4 Sample Time Points

Table 1 shows the various time points that were used in the studies. The time ranges varied from 0-6 hours in the Halawa et al. studies [13,15] to 144 hours in the Ma et al. study [11]. Shafeeq et al [19] was unique as it was the only study to take time points in days rather than hours, with the last time point being after 15 days. Intervals of the time points varied with the smaller ranged studies keeping smaller intervals such as 2 hours Martínez et al. [17], or 3 hours Halawa et al. [13,15]. The studies that went on for 24 hours or more would increase the intervals as the time increased [11,12,14,16,17,18], except for Ibrahim et al. [16] which kept its intervals to 24 hours.

Nathan Routledge -12548189 8

Table 1: Overview of the Animal-Based Literature Criteria

*Indicates Housekeeping Gene

Study Year Animal models Sample

size Tissue Method of death Time Points Gene/RNA Marker Type Specific names mentioned Statistical Analysis

(A-Comparison. B-Tests. C-Software.) Ma et al. 2015

Sprague-Dawley Rats 270 males and females Brain Cervical Dislocation 0, 1, 3, 6, 12, 24, 36, 48, 96, 120, 144 hours

miRNA, SrRNA miR-125b, miR-9, U6, 5SrRNA, RPS29, 18SrRNA, β-actin*, GAPDH* A- 2-ΔΔct B-GraphPad C-Matlab Hunter et al. 2017 Mice (C57BL/6JRj) and Zebrafish 20 males (Mice) 44 (Zebrafish)

Brain and Liver (Mice), Whole organism (Zebrafish) Cervical Dislocation (Mice) Ice water (Zebrafish) 0, 15min, 30min, 1, 4, 8, 12, 48, 96 hours Whole

Transcriptome A-Signal Intensity of dilution. B-One-sided Dunnett’s T-test. C-C++ Halawa et

al. 2018 Albino Rats 21 males Liver Cervical Dislocation 0, 1, 3, 6 hours Inflammation and Apoptosis IL-1β and TNFα, Bcl-2 and Casp3, c-fos, β-actin* A-2-ΔΔct B-One-way ANOVA, Pearson’s correlation C-SAS Sawaguchi

and Taki 2018 Mice (C57BL6) N/A Brain, Liver, Spleen N/A 0,4, 8, 12, 24, 48, 60, 72 hours

RNA abundance 100alfa5, Cxcl2, Ap3m2, Adph, Lrgl, Pomcl, Cga, Prl A-N/A. B-N/A. C-N/A. Ibrahim et

al. 2019 Albino Rats 18 females Skin of the thigh Cervical Dislocation 0, 24, 48 hours miRNA miRNA-205, miRNA-21, U6* A-2-ΔΔct

B-Mann-Whitney U

C-SPSS

Martínez et

al. 2019 Wistar Rats 4 adults Gastrocnemius muscle (calf) Intraperitoneal injection of ketamine

0, 2, 4, 6,

8 hours Autophagy (and Melatonin) LC3, Beclin-1, ATG7, ATG12, MT2, GAPDH* A-2-ΔΔct B-Shapiro-Wilk and Levine test, Mann-Whitney U C-SPSS, Sigma Plot

Noshy 2020 Mice (BALB/c) 35 males Liver Cervical

Dislocation 0, 3, 6, 9, 12, 18, 24 hours

Apoptosis Trp53, Bcl2, Bax,

Casp3, and miR-122* A-2-ΔΔct

B-Mann-Whitney U C-SPSS

Shafeeq et

al. 2020 Moths (Plodia interpunctella) A Colony Whole species Natural 7, 8, 9, 10, 11, 13, 15 days

Development, stress, and host defence mechanisms hsp25, hsp70, grp78, hsp90, EcR, USP, βgrp, ProPO, and Tpx, - βactin* and rpS7* A-2-ΔΔct B-DMRT C-SAS

Nathan Routledge -12548189 9

2.1.5 Extraction and Quantification of RNA

In all the studies RNA was extracted from the samples using Trizol reagent and chloroform. In the Hunter et al. study [12] RNA is labelled by a One-Color Microarray-based Gene Expression Analysis. All the other studies [11, 14-19] state that RNA purification is checked using a nano-spectrophotometer. Hunter et al. [12] hybridised its labelled RNA to a DNA microarray. The other studies used RT-qPCR using 40 cycles (5/7) [11,14-19], only the Halawa et al. studies [14,15] used 35 cycles.

2.1.6 Genes and RNA Markers

Table 1 shows the types of genes/ RNA markers that the studies examined, and any specific genes/ RNA markers that were directly mentioned in the study. There were various focuses between the studies, but the most prominent was on genes that are involved in cellular death such as autophagy and apoptosis related genes. This is seen 4 of the 9 studies [13,15,17,18]. The other common focus was inflammation and host defence related genes [13,15,19]. Hunter et al. [12] was the only study not to select specific genes and RNA markers, but instead did whole transcriptome wide analysis of rats and zebrafish. β-actin and GAPDH were commonly used as housekeeping genes in the studies (5/9),

[11,13,15,17,18,19]. The two most common genes to appear across the studies were Caspase 3 and Bcl2 appearing in 3 of the studies, all of which focused on apoptosis related genes [13,15,18].

2.1.7 Experimental Design and Conditions

The Ma et al. study [11] was based around differing temperatures, so once the rats were killed, they were split into 4 groups at the different temperatures of 15°C, 25°C and 35°C. An additional 36 rats were also placed in temperatures at 10°C, 20°C, and 30°C for validation of the PMI model. Samples were taken from each group at the time intervals seen in table 1. The Hunter et al. study [12] was focusing on identifying transcripts from mice and zebrafish that could be used to determine a PMI. The mice were maintained at room temperature and fed ad libitum till they were killed, after which they were placed in a plastic bag at room temperature. Samples were taken at the time periods stated in table 1, and flash frozen at -80°C to later be analysed. The zebrafish were kept in an aquarium of 28°C. At 0 hours 4 zebrafish were flash frozen at -80°C. The rest of the zebrafish were put in a container that had a mesh at the bottom and placed in an 8-litre tank of ice water for 5 minutes. The container was then placed back in the aquarium and 4 fish were taken at every time interval and flash frozen.

The Halawa et al. studies [13,15] were focused on examining how heat stress affects the expression of inflammation and apoptosis genes after death. For each study after the rats were killed 12 rats were placed at room temperature, and 12 rats were put in a 41°C water bath. The other 6 rats were the control group and the samples were taken straight away. From each of the groups 4 rats would have their samples taken at the time intervals shown in table 1.

Nathan Routledge -12548189 10 The Sawaguchi and Taki study’s [14] main focus was to detect gene expression during the supra vital time after death. The killed mice were kept at 4°C and had their brain, liver, and spleen removed for sampling and analysis. DD-PCR was performed and visualised using gel electrophoresis, as well a DNA microarray of the brain samples from after 12 hours. Finally, RT-PCR was performed on certain genes chosen from a dendrogram. The expression was compared to wild type and mice who had the genes knocked out.

The main aim of the Ibrahim et al. study [16] was to observe the change of miRNA expression in incisional wound of rats’ post-mortem. Before death the rats were

anesthetized using thiopental sodium injection, and an incisional wound of 4cm was made to the thigh. The rats were then killed and placed into 3 groups (0hrs, 24hrs, 48hrs) and left in cages at 4°C with their wounds undressed. Skin samples, from around the wound, were taken histologically examined and expression of miRNA was recorded.

The Martínez et al. study [17] focused on apoptosis genes post-mortem in rats, and the oxidative stress levels of the rats post-mortem. The rats were kept alive for 15 days at a cycle of 12 hours sunlight/daylight. They were kept at room temperature and 55-60% humidity, they also had ad libitum food. Once killed they were kept at room temperature and at every time point (see table 1) 20mg of tissue was taken for analysis, of which 10g was used to measure oxidative stress.

The Noshy study [18] also focused on apoptosis gene expression after death. Mice were kept alive for 10 days then killed and kept at 30°C. Five mice were taken per time period seen in table 1, samples from their liver were then flash frozen and kept at -20°C. Finally, the Shafeeq et al. study’s [19] main aim was to understand what genes are expressed after Plodia interpunctella die. A colony of Plodia interpunctella were kept in a plastic box at 21°C and a humidity of 40% +/- 5%. They were kept on a cycle of 16 hours of light and 8 hours of darkness, while being fed wheat bran and pollen. After each day seen in table 1 a batch of 5 moths was taken and either killed if still alive (before day 9), or already dead and the RNA was extracted for analysis.

2.1.8 Statistical Analysis

The main way expression was calculated was by using the 2-ΔΔct formula. Signal intensity of probes in RNA was also used (see table 1). The most common statistical test used was the Mann-Whitney U test, used in 3/9 of the studies [15-17]. The most used statistical software was SPSS [16-19], and SAS [12,13, 19], although both Matlab and C++ were also used (see table 1).

2.1.9 PMI Models Used

Only 3 of the studies mentioned using a mathematical model to form a PMI.

The Ma et al. study [11] used a bivariate cubic curve fit with the analysis of Δct, temperature and PMI. Equations of t PMI x t and PMI2 _{were calculated and then validated with the 36}

Nathan Routledge -12548189 11 mice. The known temperature and Δct were added into the equation to predict PMI. The study used R to calculate the best PMI from the given values.

The Hunter et al. study [12] used linear regression on data from a training and test sets. With the training set building the regression equations and the test set used to validate the model. Three designs were made: the simple design based of one gene, the over-defined design based of the top 3 preforming genes form the simple design, and finally the perfectly defined design based of transcripts randomly selected from the transcriptome.

The final PMI model was in the Noshy study [18], which used four different equations linear, cubic, quadratic, and exponential. The models with the highest R2 _{for each gene were used}

to predict PMI.

2.1.10 Results and Conclusions

The results and conclusions will be broken down paper by paper. Ma et al. 2015

The study found that the lower the temperature the slower and less likely that the RNA markers would degrade, and that as the PMI increased, more RNA had degraded. With the higher temperatures causing the RNA to degrade the most over increasing PMIs. The only exception to this being U6 which showed increased expression as the PMI increased at high temperatures. When creating the PMI model based on Δct and temperature, they found that degradation of β-actin correlates with PMI across all the temperatures. It was concluded that β-actin was the most reliable marker to estimate PMI especially at high temperatures.

Hunter et al. 2017

The study of the whole zebrafish transcriptome found 548 genes that significantly increased after death. In mice 515 genes were found to significantly express after death, of which 36 were from the liver and 478 were from the brain. From their PMI models the top 3

transcripts identified were A_15_P121158, A_15_P295031, and A_15_P407295 from zebrafish (all non-coding). From the mice brains the top 3 transcripts were A_66_P130916, A_55_P2127959, and A_55_P2216536 (H2-Ob, Zfp36l3, and Rbx1 respectively). From the mice livers the top 3 transcripts were A_51_P318381, A_30_P01018537, and

A_55_P2006861 (Trio, Prok2, and placenta growth factor isoform 1 precursor respectively). They found that using the over-defined design compared to the simple design there was no improvement in the PMI prediction. Using the perfectly defined design three panels were shown for each species, with the mice having two brain panels and one liver panel. The results showed that in mice the liver was significantly better at estimating the PMI than brain, but in both cases the estimation could be affected by the abundance of the

transcripts. They concluded that the study showed their model could accurately predict the PMI of zebrafish and mice, from their liver transcripts, and that the model could be used on human cadavers in the future.

Nathan Routledge -12548189 12 Halawa et al 2018

The study looked at the inflammation genes in the liver: TNF-α expression decreased after 0 hours initially and in room temperature the expression not changing at 3 and 6 hours, while the heat stressed group saw expression increase after 3 hours but decrease again after 6 hours. With IL-1β there was significant decrease in expression compared to 0 hours in both groups, with the expression showing a slight but insignificant increase in expression in 3 and 6 hours in both groups.

For the apoptosis genes in the liver: For the room temperature group Bcl-2 showed an initial decrease in expression after 1 hr compared to 0 hours, but then increased in 3 and 6 hours. While with the heat stress group there was a similar initial decrease, but only an increase at 3 hours, as it decreased after 6 hours. Casp3 in the room temperature group showed a decrease of expression at 1 hour compared to 0 hours, and then saw increased expression from 3 hours and more at 6 hours. In the heat stress group, there was increased expression after 1 hour, but then a decrease at 3 hours, followed by an increase again at 6 hours. Finally, with c-fos at there were only significant increases of expression compared to 0 hours at 3 and 6 hours, where the expression was the same for heat stress group, and an increase at 6 hours compared to 3 hours with the room temperature group.

Sawaguchi and Taki 2018

In the brain and spleen there was a decrease in total RNA only after 72 hours, while liver saw a decrease after 4 hours. There was an increase in RNA seen after 4 hours in brain. No increase in RNA was seen in the spleen between 0 and 12 hours. By 12 hours almost all of the RNA in the Liver had been degraded.

Halawa et al. 2019

The study looked at the inflammation genes in the brain: TNFα showed no significant changes in expression in the room temperature group compared to 0 hours. With the heat stress group there was significant increase of expression at 1 hour and 6 hours. IL-1β

showed no significant changes in expression in the room temperature group compared to 0 hours. In the heat stress group, there was significant increase in expression compared to 0 hours with all the time groups. With the expression increasing as the time increased. the apoptosis genes in the brain: In the room temperature groups Bcl-2 showed a decrease in expression as time went on compared to 0 hours. While in the heat stress group there was decreased expression at 1 and 3 hours, but a significant increase at 6 hours. Casp3 showed a decrease in expression compared to 0 hours in the room temperature group, with 3 hours showing little to no expression, but 6 hours showing some but not more than at 0 hours. In the heat stress group, there was no expression after 1 hours, and then some expression at 3 and 6 hours, but all were less than at 0 hours.

Finally, in the room temperature group c-fos showed a decrease in expression across all the time periods compared to 0 hours, while in the heat stress group there was an increase in expression compared to 0 hours, with the expression at 1 hour being higher than at 3 hours. Ibrahim et al. 2019

The study showed that inflammatory cells were observed immediately after death, but by the degenerative stage from 24 hours these cells were no longer visible, and instead dermal fibrosis was visible. The expression of miRNA-205 and miRNA-21 increased in expression from 0 hours, with maximum expression in both seen at 24 hours. The study concluded that

Nathan Routledge -12548189 13 there is potential to determine perimortem wounds and wound vitality based on acute inflammation. Also, the possibility of studying the changes of miRNA in 24 hours to determine a PMI.

Martínez et al. 2019

The study showed that as the time since death increased so did the levels of oxidative stress. Breaking down the autophagy genes LC3 increased in expression compared to at 0 hour, with the highest peak being at 2 hours, after which the expression decreased till a small increase at 8 hours. Beclin-1, ATG7, and ATG12 all increased till a highest peak at 4 hours, and then expression continued to decrease.

Finally, MT2 only started expressing after 2 hours with the highest peak being at 4 hours. There was no expression after 6 hours, and a small amount of expression after 8 hours. They concluded that the results showed proof that there is cross talk between autophagy and apoptosis genes during hours 6-8 that can be used to determine a PMI. They also concluded that MT2 could also be used as indictor to determine PMI.

Noshy 2020

The study showed Casp3 increased in expression as the time since death increased. Bax only start to increase in expression from 3 hours till a peak at 18 hours, after which the

expression started decreasing. Bcl2 started to decrease in expression as the time since death increased with the lowest expression seen between 18 and 24 hours. Finally, Trp53 showed increased expression between the hour of 3 and 6, before it continued to decrease in expression as the time since death increased. The cubic equations gave the best PMI results for Bax and Trp53, while the quadratic equations gave the best PMI results for Casp3 and Blc2. They concluded that the apoptosis genes in the liver of rats have different

expression patterns after death, possibly due to their different roles in apoptosis. Shafeeq et al. 2020

The study showed that hsp25, hsp70, grp78, hsp90, EcR, ProPO, and Tpx all increased in expression before death, and then all but hsp70 then decreased in expression after death. The expression levels of hsp70 increased after death with the max at 15 days. USP

expression levels did not change before or after death, while βgrp expression levels began declining even before death at 8 days, and after death maintained a low level of expression. They concluded the gene expression profiles after death differed enough to potentially be used as biomarkers to determine PMI.

Nathan Routledge -12548189 14

2.2 Human-based Studies

In total 9 [20-28] papers focused on studying post-mortem gene expression in human tissue. These papers were then compared using a variety of criteria as seen in 2.2

Table 2: Overview of the Human-based studies criteria

*Indicates Housekeeping Gene

Study Year Sample Tissue Sample Size PMIs Recorded Gene

Type/Role Specific Genes Statistical Analysis (A-Comparison. B-Tests. C-Software)

Javan et al 2015 Liver 4 6 (control), 16,

16, 48 hours Apoptosis Related Array of 84 genes A-N/A. B-N/A. C-R

Javan et al 2016 Cardiac and

Cerebral 4 6, 16, 36.5, 58 hours Not genes (Autophagy Proteins)

LC3 I, LC3 II, P62, BNIP3, Beclin-1, Atg 7, and GAPDH* A-N/A. B-N/A. C-N/A. Zhu et al 2017 Multiple 573 transcripts (900 donors) 1 – 27 hours Genome

wide N/A A-N/A.

B-Pearson’s correlation, spearman’s rank, Levene test.

C-Matlab.

Fais et al 2018 Gingival 10 Short (1-3 days), Medium (4-5 days), Late (8-9 days)

Hypoxia

related HIF-1a A-2-ΔΔct. B-Student T test. C-N/A.

Ferreira et al 2018 Ischemic (36

types) 20 per tissue (540 donors) 17 – 1739 minutes Genome wide N/A A-Multidimensional scaling. B-Pearson’s correlation, spearman’s rank. C-R.

Tolbert et al 2018 Prostate 5 24, 38, 77, 96,

120 hours Apoptosis Related Array of 84 genes including: BCL2, BFAR, and BIRC2. CL2A1, BCL2L1, and BCL2L10. CASP-2, DIABLO

and APAF-1. XIAP, BIRC3, and BCL10

A-2-ΔΔct. B-Student T test. C-RT2 _Profiler.

González-Herrera et al 2019 Myocardial, Pericardial Fluid, Blood

30 N/A (Not

studying PMI) Myocardial Injury and repair

TNNI3, MYL3, MMP9,

TGFB1 and VEGFA. GAPDH* A-2-ΔΔct.

B-Spearman’s rank, Mann-Whitney U.

C-SPSS

Uerlings et al 2019 Brain, heart, spleen, adrenal gland, skeletal muscle, and blood 21 Short (2-4 days), Medium (5-7 days), Late (10 days) “PMI

Markers” HIF1A, PER3, PHF1, USF2, KIAA0586, ZNF74, RNU24

and hsa-miR-191-5p

A-ΔCq.

B-N/A.

C-GenEx, LineRegPCR.

Javan et al 2020 Liver 30 3.5 hours – 37

days Tissue specific AMBP, F2, SPP2, CFHR2, F9, MBL2, AHSG, C9 (Liver) A-N/A.

B-Banded Smith-Waterman algorithm. C-Miseq.

Nathan Routledge -12548189 15

2.2.1 Tissue Type and Sample Size

The most common tissue type to be analysed was Liver, Brain, and Heart all appearing at least in 2/9 studies. The majority of the studies involved multiple tissue types, with Ferreira

et al. [24] and Zhu et al. [22] using over 15 types of tissue in their studies. The sample sizes

ranged from 4 cadavers to 573 transcripts and 20 samples per tissue from 540 donors (see table 2).

2.2.2 Breakdown of the Subject Characteristics

Table 3 shows a full breakdown of the age and weight range of the subjects. Including sex, ethnicity, and cause of death. The age range was from 15-90 years and a weight range from 50-177 kilos. Most of the subjects were male, and Caucasian, with only Uerlings et al. [27] having more female subjects than male. Only Javan et al 2015 [20] and 2020 [28] also mentioned ethnicities directly. Both Zhu et al. [22] and Ferreira et al. [24] stated they did take into account age, weight, sex, and ethnicity as part of their model, but the actual data was not available. Most of the studies used subjects that had differing causes of death, see table 3. Only Javan et al. [21] and Fais et al. [23] had a single cause of death in their studies.

Table 3: Breakdown of the subjects’ characteristics in the Human-Based Studies

Study Year Age Range

(Yrs)

Weight (Kg) Sex Ethnicity Cause of Death

Javan et al 2015 22-55 66.4-112.5 All M C=2, A=2 Gunshot wound, Coronary Heart Disease

Javan et al 2016 17-55 88-154 M=3, F=1 N/A Coronary Heart Disease

Zhu et al 2017 N/A* N/A* N/A* N/A* N/A

Fais et al 2018 18-60 N/A Mixed N/A Trauma-related

Ferreira et al 2018 N/A* N/A* N/A* N/A* Mixed (most

Coronary Heart Disease) Tolbert et al 2018 30-73 70-90 All M All C Accidental,

Homicide, Suicide

González-Herrera et al 2019 30-90 N/A M=25, F=5 N/A Sudden cardiac arrest, Multiple Trauma, Mechanical asphyxia, other Uerlings et al 2019 15-91 N/A (BMI) M=10, F=11 N/A N/A

Javan et al 2020 19-85 50-177 M=22, F=8 C=24, H=2, A=4

Accidental, Natural, Homicide, Suicide

M-Male, F-Female, A-African American, C-Caucasian, H-Hispanic

2.2.3 PMIs of the Subjects

The overall range was from 17 minutes in the Ferreira et al. study [24] to 37 days in the Javan et al. study [28]. None of the PMIs went up in consistent intervals but were based on the subjects they had available, see table 2. Both Fais et al. [23] and Uerlings et al. [27]

Nathan Routledge -12548189 16 grouped their subjects into types of PMI rather than giving specific PMIs. González-Herrera

et al. [26] did not give any PMIs, but focused on identifying cause of death, not PMI. 2.2.4 Genes and RNA Markers of Interest

Apoptosis related genes were the most prominent types of genes analysed, being the main focuses in 2/9 of the studies, see table 2. Zhu et al. [22] and Ferreira et al. [24] did not focus on a specific type gene, but instead carried out a genome wide analysis of various tissues. Javan et al. 2015 [20] and Tolbert et al. [25] used the same array set of genes in their studies.

2.2.5 RNA Extraction and Quantification

Most of the studies used the RNAeasy isolation kit to extract RNA followed by a nanodrop analysis.

2.2.6 Statistical Analysis

The type of statistical analysis varied between the studies, with either Spearman’s rank test and/or Pearson’s correlation being favoured, see table 2. Javan et al. 2015 [20] and 2016 [21] did not mention their statistical analysis. The most commonly used software was R [24,25], but others included Matlab and SPSS, see table 2.

2.2.7 Experimental Design

Javan et al. 2015 [20] took 390cm3 _{of liver tissue from autopsies of the cadavers and stored}

them at -80°C. A PCR array of 84 apoptosis genes was performed from the samples and their RNA abundance levels were measured. Tolbert et al. [25] also used the same PCR array in its methods, except the tissue samples were taken from the prostate, and some needed to be shipped from Italy to the USA. The Javan et al. 2020 study [28] also took samples during an autopsy and kept them at -80°C. In their study a 46 gene multiplex was analysed using TruSeq.

Javan et al. 2016 [21] similarly took tissue samples from autopsies and stored them at -80°C. The samples in the 2016 study were homogenised with a glass tissue grinder and western blotting was performed and shown on an SDS-Page to identify protein expression.

Zhu et al. [22] and Ferreira et al. [24] used tissue samples from the GTEx project. Gene expression in various tissues were analysed using RNA-seq. Both studies formulated PMI prediction models based on the gene expression and known variables of the tissue donors. Fais et al. [23] focused on the specific gene HIF-1a, while also doing an

immunohistochemical analysis of the gingival tissue. 0.5-1.0cm2 _{of gingival tissue was}

sampled. Part of the tissue was immediately fixed for immunohistochemical analysis, and the other part was frozen at -80°C for later analysis. Both the RNA expression levels of HIF-1a and the immunohistochemical changes were analysed.

Nathan Routledge -12548189 17 The focus of the González-Herrera et al. study [26] was to identify cause of death based on RNA expression. In this study the samples were collected during an autopsy and the cause of death was noted via various combinations of data. The samples were then split up according to cause of death and the RNA was analysed using RT-PCR.

Uerlings et al. [27] focused on determining the best type of PMI marker, endogenous or exogenous. The samples were taken during an autopsy and both types of markers were analysed using RT-qPCR.

2.2.8 PMI Models

Only the Zhu et al. and Ferreira et al. studies [22,24] put forward a model to predict PMI. Both used multiple linear regression that adjusted for covariates such as age, weight, height, BMI, gender. The Zhu et al. study [22] used a PEER algorithm as a two-step approach to create the model. With the PEER algorithm identifying the common variation between covariates and then being applied in the multiple linear regression model. Similarly, Ferreira

et al. [24] used machine learning with training set and testing sets. A linear regression

model was created using the training set with the covariates, and the test set used to validate the model. The study also went one step further and produced a protocol that could be used in cases to determine the PMI, as seen in figure 1.

Figure 1: Protocol to determine PMI based on the PMI model from Ferreira et al. [24]

2.2.9 Results and Conclusions

The results of each study are as follows. Javan et al. 2015 [20]

The anti-apoptotic genes BAG3, BAK1, BAX, BIRC5, IL10, NAIP, NFKB1, and RIPK2 all showed continued decreased expression from 6 hours to 48 hours, while XIAP showed increased expression over time. The negative regulators BCL10, BCL2L2, BCL2, CD40LG, and CIDEA also showed decreased expression however to a lesser degree than the anti-apoptosis genes. The death domain regulators TNFRSF10A, TNFRSF11B, YNFRSF25, TNFRSF9, and TNFRSAF showed similar decreased expression patterns over time as the anti-apoptosis genes, while the positive apoptosis regulators ABL1, AIFM1, CIDEB, PYCARD, and TNFRSF10B all showed

1. Note start time of forensic process (START_TIME)

2. Note the time at which the samples of choice are stabilized E.g (SKNS_STAB) For Skin sun exposed (lower leg) sample 3. Prepare sample and extract RNA. Perform RNA sequencing

4. Process RNA-seq data, and gather gene expression quantifications for tissue sample

5. Input the gene expression data, START_TIME and STAB into the PMI Prediction Software

6. Software gives PMI of the tissue type and the individual, as well as the coefficient variation of the tissue PMIs.

Nathan Routledge -12548189 18 increased in expression at the longer PMI (48hrs). Likewise, the caspases CASP3, CASP4, and

CASP9 showed increased expression over time. They concluded that gene expression in the

liver is stable for up to 48 hours. Javan et al. 2016 [21]

In cardiac tissue LC3 I, LC3 II and Beclin-1 all showed a time dependant increase in

expression with LC3 I and LC3 II peaking at 58 hours, and Beclin-1 peaking at 36.5 hours and staying the same at 58 hours. BNIP3 however showed a time dependant decrease in

expression, while Atg7 decreased from 6 to 16 hours, then showed an increase peaking at 58 hours.

In the cerebral tissue again LC3 I and LC3 II also showed a time dependant increase in expression, peaking at 58 hours. However, BNIP3 showed a similar pattern, but at 58 hours it showed no expression. Beclin-1 decreased in expression as time increased. Atg7 did not show any expression till 16 hours, did not vary much at 36.5 hours, but then had a very large increase in expression at 58 hours. They concluded that thantophagy occurs in a time

dependant manner, that can potentially be used to determine a PMI. Zhu et al. 2017 [22]

A total of 7546 genes were identified as being linked with PMI. The number of genes per tissue varied from 0 to 2763 genes. 1992 genes were shown to be expressed in two tissues, 482 in three, 96 in four, 14 in five, and 1 gene was expressed in 6 tissue types. For the genes that were expressed in multiple tissue types the expression direction (increase or decrease) would be the same. Translation and translational initiation genes and structural constituent of the ribosome genes showed the most increased expression. Finally, the genotype of the donor also had an effect on expression. For example, the SNP rs1521177 in SPIN3 showed increased expression over time with the genotype GG but decreased expression with the phenotype TT. They concluded that the degradation of mRNA is tissue and gene specific, as well as dependant on a person’s genotype.

Fais et al. 2018 [23]

Immunohistochemical analysis at the short PMI showed high levels of HIF-1a in the oral epithelium compared to the sub-oral connective tissue. At medium PMI HIF-1a levels were lower. The sub-oral connective tissue did show signs of HIF-1a but to a lower level than the oral epithelium. In late PMI there was no sign of HIF-1a.

With the mRNA analysis there was a slight increase in expression compared to the living control, which was larger in the medium PMI, but late PMI showed no sign of HIF-1a expression. They concluded that HIF-1a could be used as a biomarker in forensic investigations.

Ferreira et al. 2018 [24]

The number of genes that show expression after death vary from tissue to tissue. These changes are also time specific. For example, muscle tissue showed genes mostly change expression within the first few hours, while some other tissues took at least 6 hours before any change was noticeable. RNASE2 was the gene that showed the highest expression in the most different types of tissue. While the genes HBA1 and HBA2 showed expression in tissue

Nathan Routledge -12548189 19 such as the lungs, but in the tissue where they are normally the most expressed (blood) they were not showing expression. In blood most increased gene expression occurred during 7-14 hours and by 24 hours the expression levels had stabilised.

Once the PMI model had been adjusted for the covariates only 54 genes per tissue were still seen to correlate with the PMI. MtDNA was also investigated, with there being less mtDNA the longer PMI. They concluded many transcriptional changes that occur due to death, and that a model based on these changes could be created to determine the PMI.

Tolbert et al. 2018 [25]

The anti-apoptosis genes BCL2, BFAR, and BIRC2 all increased in expression as time increased. BCL2 peaked at 96 hours, while BFAR reached maximum at 120 hours. CL2A1,

BCL2L1, and BCL2L10 all also presented an increase in expression as time increased.

Equally the negative regulators XIAP, BIRC3, and BCL10 all showed significant increase in a time dependant manner.

The Apoptosis genes CASP-2, DIABLO and APAF-1 showed significant expression at 96 and 120 hours. They concluded that overexpression of apoptosis related genes occurs as the prostate decomposes in a time dependant manner.

González-Herrera et al. 2019 [26]

The expression of TNNI3 in blood was higher in subjects that had died from mechanical asphyxia. The expression of MMP9 was higher in the pericardial fluid of subjects that had died from mechanical asphyxia than in sudden cardiac arrest. No difference in expression could be seen in myocardium tissue. They concluded that there is potential in using gene expression analysis to determine cause of death.

Uerlings et al. 2019 [27]

The exogenous marker TATAA Universal RNA Spike I was not suitable for normalisation, beacuase its expression fluctuated too much between samples and tissue types. The endogenous markers were more stable but could not recognise specific expression levels within PMI groups. They concluded the TATAA Universal RNA Spike I could not be used as a normalisation for PMI biomarkers.

Javan et al. 2020 [28]

The liver markers were the highly specific with all but AHSG showing expression in the samples AHSG did not show any expression in the American subjects. 98% of the reads were liver markers. However, the skeletal biomarkers TNNI2 and ATP2A1 showed expression in 2 of the Italian subjects, and 3 Italian subjects expressed the adipose biomarker PLIN1. One American subject also expressed the intestine biomarkers DEFA6 and LCT while another expressed the stomach biomarker PGA3. They concluded that RNA was stable enough in the liver samples up to 7 days, and that the RNA markers can be effectively used.

Nathan Routledge -12548189 20

3.0 Discussion

3.1 Animal versus Human based Studies

3.1.1 Comparison of the study types

When comparing the two types of studies the most striking difference is how large the sample size and sampling times are in the animal studies. The studies were generally able to use at least 18 mice or rats in their studies and have several time intervals to study

expression over time. When compared to the human studies and in particular the studies with human cadavers, the sample size is far more limited due to having to require extra permission to study human bodies. Also, there is the reliance on having to wait till the body becomes available. The main exception to this is with Zhu et al. [22] and Ferreira et al. [24] where due to using the human tissue from the GTEx project and not cadavers, they could use the largest sample size and include a large amount of PMI times. Due to the animal studies having larger sampling sizes and sampling times, the results in the animal studies were more precise. However, all of the animal studies besides the Ibrahim et al. study [16] had the animals in good health when they died, and in all bar the Martínez et al. study [17] the mice/rats were killed the same way. In forensic cases it is very unlikely that if the death is natural the people were in a healthy condition before death, and in non-natural deaths it is unknown if the manner of death may have an impact on gene expression. No animal study looked into how the type of death may change expression, which could leave the results skewed as most of the subjects were killed via cervical dislocation. The only human study that specifically viewed how the manner of death could affect expression was the González-Herrera et al. study [26]. Where the main aim of the study was to examine if it was possible to determine manner of death based on gene expression, with their results

showing in the case of mechanical asphyxia that it may be possible. One advantage the animal studies did have, was to be able to use controls of expression before the animals died. Therefore, having a better understanding of the initial changes in expression from the transcriptome to the thanatotranscriptome, and this can lead to a better understanding of why such changes are occurring. A final observation is that both studies often had time intervals that were taken from different animals or subjects, rather than the same animal or subject over the same time period. For example, in Tolbert et al. [25] the sample taken at 24 hours, was taken from a different individual than the sample taken at 38 hours. This meant especially in the human studies that there was interpersonal variation between the results, and this could affect the reliability of the observed trends.

3.1.2 How applicable are animal studies on humans.

The effectiveness of animal-based models in a human context is debatable [29,30].

However, when it comes to apoptosis and inflammation related genes, research has shown that these genes are highly conserved in mammals such as mice and rats [31], as well as zebrafish [32]. For example, while caspase 4 and 5 are not found in mice, but are in humans, they share a total of 10 of the caspase gene family [31]. So, genes such as Casp3 and Bcl2

Nathan Routledge -12548189 21 that were studied in many of the animal-based reviews [13,15,17] could be applicable in human studies, and were in fact used in some of the human based-studies. Even more conserved is the BAG family of genes where all 6 genes are seen in both mice and humans [31]. Similarly, the mechanism of apoptosis in zebrafish has been seen to show similar genes expressed to those in mammals [32]. Other RNA related markers such as the ones used in Ma et al. [11] are also observed in humans. Any research based on animals that wants to be applied to humans in the future should focus on genes, and gene families that are highly conserved and detected in both.

3.2 Confounding Factors

3.2.1 Intrinsic Factors

It is known that the age of an individual can have an effect on gene expression [33,34]. The human studies kept a note of the ages of their subjects, but the studies lacked any analysis into how the age of a subject may affect results. However, in Zhu et al. [22] and Ferreira et

al. [24] they both include age as one of their covariates to create their PMI models so that

the age of the individual is taken into account when calculating a PMI. This seems to be a constant across the studies, intrinsic factors such as age, weight, height are all mentioned in the studies, but the actual knowledge of how these factors affect expression is not explored. This is an area where the investigation into how the thanatotranscriptome changes based on intrinsic factors can be further explored, especially when it comes to age, and its relation to the increase chances of cancer. As many of the genes studied are also heavily linked with various cancers, using them in a forensic context could bring extra difficulties.

3.2.2 Extrinsic Factors

Similarly, extrinsic factors such as the temperature and humidity of the subjects was not touched on in much detail. The animal studies mention the extrinsic factors that the animals were kept at both before and after death, but it is only Halawa et al. studies [13,15] that analysed how the ambient temperature around the dead animal effected gene expression. Their results showed that indeed the temperature can affect not only the rate of expression but also whether there is any expression at all. For example, Bcl-2 at room temperature showing no expression, but expression after 6 hours at 41°C. The human studies main extrinsic focus was manner of death, again only being noted, not analysed. Temperature is not mentioned in the human studies, mainly due to the fact that as the research is reliant on the information given to them by the cases, or samples. If the temperature of the body and/or the area in which they were found are not given, the studies cannot analyse what their effect is on gene expression. Even the extensive studies by Zhu et al. [22] and Ferreira

et al. [24] did not take into account temperature when forming their PMI models. There

needs to be more specific research into how extrinsic factors affect the

thanatotranscriptome if it is to be used in a forensic context, where there are often various different factors involved.

Nathan Routledge -12548189 22

3.3 Genes

3.3.1 Apoptosis Genes

The most studied types of genes so far, are related to apoptosis. This is understandable as they are directly involved in the either inducing or inhibiting cellular death. A common array that included 84 apoptosis genes was used in two of the human based studies. Javan et al. 2015 [20] and Tolbert et al [25] monitored different time periods, with Javan et al. [20] only observing till 48 hours, while Tolbert et al. [25] continued up to 120 hours but lacked any information before 24 hours. The two studies could be combined to get a better overall pattern of how these apoptosis genes are expressed. For example, the anti-apoptosis genes in the Javan et al. study [20] showed decreased expression in the first two days, but in the Tolbert et al. study [25] they were showing a peak expression at 120 hours.

3.3.2 Tissue Specificity

The studies demonstrate that the thanatotranscriptome is highly tissue specific, with the type of tissues affecting which genes are being expressed and to what degree. Any type of forensic tool will be reliant on the sample source that is used, and the genes analysed would need to have known expression for that tissue. The studies have shown that this is easily achievable with the likes of liver and brain. The Zhu et al. and Ferreira et al. studies [22,24] have also shown for various tissue samples which genes are expressed post-mortem in which tissue and which could form the basis of further research when using tissue specific genes.

3.4 Forensic Application

3.4.1 PMI

The main forensic application that is the potential to use the thanatotranscriptome as a new innovative way to determine the PMI of an individual. A majority of the studies focused on observing how the expression of genes and RNA markers changed as the PMI changed. With most concluding that there were time dependant expression changes. However, the

application of this concept in a forensic context is generally only briefly mentioned, with a only 5 studies describing how the PMI could be estimated. Of those 5 only the Ferreira et al. study [24] created a protocol that could actually be used. The protocol took into account the fact that while expression might be the most diverse and abundant in areas such as the brain, liver or prostate, they would not be the best suited tissue samples to use, due to the impracticality of having to perform a full autopsy to get such samples taking time, or the samples might not be available.

The PMI model suggested by Ferreira et al. [24] is the most developed and from their results showed that even when not fully knowing the reasons why and how the confounding

factors affect the thanatotranscriptome, they could still predict PMI by including them as corrections. The PMI models work by genome wide analysis of the thanatotranscriptome

Nathan Routledge -12548189 23 and then by collecting all of the transcripts together forming formulas that best fit the PMIs based on the expression over time of the combined transcripts. The major problem with this is the time it takes to perform RNA-seq, which can vary from 1.5 to 12 days. However, using Miseq that could be limited to 55 hours [35]. This could be a major drawback as knowing the PMI is vital for a case, so the earlier the information the better. Using a PCR microarray as seen in Javan et al. and Tolbert et al. studies [20,25] maybe a quicker and cheaper

alternative, however there is then a reliance on knowing the specific genes needed. If this method were to be used then there has to be far more research into how factors affect the expression of the specific genes, for the array to be made. While the studies shown in this review have all shown proof of concept that specific genes do alter over time after death, the studies that focused on specific genes were not large enough to form the proof that they can now be used in a forensic context. The other issue with the Ferreira et al. [24] and also the Zhu et al. [22] models is that they are only valid till at most 27 hours, after which they do not have the data to predict PMI.

For a PMI model based on the thanatotranscriptome to be forensically applicable, there cannot be a degree of randomness about which transcripts are used. As such the most effective strategy would be to create an array of genes, such as the 84 apoptosis genes used in Javan et al. [20] and Tolbert et al. [25]. Newer candidates for such arrays could be drawn from the Ferreira et al. study [24]. While they did not mention specific genes to focus on, they did list in their supplementary data all the genes analysed, as well as in which tissues, they were expressed. They also noted the time intervals in which these changes occurred. These could be used as a starting point to investigate new arrays that could be used for specific tissue. For example, an array specifically based on genes expressed in a time dependant manner in skeletal muscle.

3.4.2 Determining Cause of Death

The other potential use of the thanatotranscriptome in a forensic context is to determine the cause of death, but there has been far less research into determining if it is possible. However, the study by González-Herrera et al. [26] did offer a proof of concept that could lead to progress in this area. The study showed that differences in expression after death in subjects that had died from mechanical asphyxiation, compared to other forms of deaths. The identified genes were specific to myocardial damage and repair. This type of approach could improve determining a natural or unnatural death.

Nathan Routledge -12548189 24

4.0 Conclusion and Recommendations

4.1 Conclusion

The main aim of this review is to analyse the current research of the thanatotranscriptome and its application in the forensic context. Both animal and human based studies have been analysed to understand the present state of research. Animal-based studies have been able to be more precise in the intervals at which the thanatotranscriptome can be analysed. Most of the research has been proof of concept, and little more. It has been demonstrated that there are a large number of genes that express after death and do so in a

time-dependant manner. The expression is also very tissue specific, which provides a potential use in taking a tissue sample and identifying a PMI based on the gene expression from the tissue. However, there has been little to no research into how the gene expression may alter depending on intrinsic factors e.g., age and weight, and extrinsic factors such as

temperature. There have been models using human tissue that take into account these factors to predict PMI, however they have so far only focused on the initial hours till just over 1 day after death. While not as greatly researched there appears to be potential in using the thanatotranscriptome to determine causes of death.

4.2 Recommendations

The main recommendation is to do more specific research, primarily using animals, into how the thanatotranscriptome changes based on intrinsic and extrinsic factors. As the

knowledge of what happens is still sparse. Furthermore, human research needs larger sample sizes and varied PMIs. Also, more research needs to be done on tissue types that would be more accessible in a forensic context. Tissue such as skin or muscle are far easier to analyse quickly compared to waiting for a full autopsy to get heart and brain tissue. Research should be focussed on those tissues to identify which genes are being expressed post-mortem and how different factors affect them. In the future if RNA-seq is cheap and rapid it may be able to perform genome-wide sequencing of the thanatotranscriptome to reliably predict PMI based on machine learning algorithms. If not, then an array of genes that all vary in expression patterns may have to be made in order to predict PMI based on overlapping expression patterns.

A starting point for this could be to take genes identified in the Ferreira et al. [24] study and analyse how factors like temperature or different manner of death impact the expression patterns of the genes, in animals and, if possible, humans.

Nathan Routledge -12548189 25

5.0 References

1. Pozhitkov AE, Neme R, Domazet-Losˇo T, Leroux BG, Soni S, Tautz D, Noble PA. 2017 Tracing the dynamics of gene transcripts after organismal death. Open Biol. 7: 160267. http://dx.doi.org/10.1098/rsob.160267

2. Pozhitkov A, Noble P. Gene expression in the twilight of death. BioEssays. 2017;39(9):1700066.

3. Westphal G, Garcia V, Souza R, Franke C, Vieira K, Birckholz V et al. Guidelines for the assessment and acceptance of potential brain-dead organ donors. Revista Brasileira de Terapia Intensiva. 2016;28(3).

4. Auråen H, Mollnes T, Bjørtuft Ø, Bakkan P, Geiran O, Kongerud J et al. Multiorgan procurement increases systemic inflammation in brain dead donors. Clinical Transplantation. 2013;27(4):613-618.

5. Hall E, Pfeiffer R, Segev D, Engels E. Cumulative incidence of cancer after solid organ transplantation. Cancer. 2013;119(12):2300-2308.

6. Lindenbergh A, Maaskant P, Sijen T. Implementation of RNA profiling in forensic casework. Forensic Science

International: Genetics. 2013;7(1):159-166.

7. Li N, Du Q, Bai R, Sun J. Vitality and wound-age estimation in forensic pathology: review and future prospects. Forensic

Sciences Research. 2018;5(1):15-24.

8. van den Berge M, Wiskerke D, Gerretsen R, Tabak J, Sijen T. DNA and RNA profiling of excavated human remains with varying postmortem intervals. International Journal of Legal Medicine. 2016;130(6):1471-1480.

9. Henßge C, Madea B. Estimation of the time since death in the early post-mortem period. Forensic Science International. 2004;144(2-3):167-175.

10. Wilk L, Hoveling R, Edelman G, Hardy H, van Schouwen S, van Venrooij H et al. Reconstructing the time since death using noninvasive thermometry and numerical analysis. Science Advances. 2020;6(22):4243.

11. Ma J, Pan H, Zeng Y, Lv Y, Zhang H, Xue A et al. Exploration of the R code-based mathematical model for PMI estimation using profiling of RNA degradation in rat brain tissue at different temperatures. Forensic Science, Medicine, and Pathology. 2015;11(4):530-537.

12. Hunter M, Pozhitkov A, Noble P. Accurate predictions of postmortem interval using linear regression analyses of gene meter expression data. Forensic Science International. 2017;275(2017):90–101.

13. Halawa A, El-Adl M, Marghani B. Thanatotranscriptome Study on Particular Hepatic Genes and Their Correlation With Postmortem Interval in the Presence or Absence of Postmortem Heat Stress. Alexandria Journal of Veterinary Sciences 2018;57(2):13-20.

14. Sawaguchi T, Taki T. Possibility of gene expression in post-mortem supra-vital time. International Medical Journal. 2018;25:186-188.

15. Halawa A, El-Adl M, Marghani B. Postmortem Heat Stress upregulates Thanatotranscriptome of Genes encode Inflammation, Apoptosis and Neuronal Stress in Brain of Rats at Short Postmortem Intervals. Australian Journal of Forensic

Sciences. 2019

16. Ibrahim S, Ali M, Basyouni H, Rashed L, Amer E, Abd El-Kareem D. Histological and miRNAs postmortem changes in incisional wound. Egyptian Journal of Forensic Sciences. 2019;9(1).

17. Martínez P, Menéndez S, Villaronga M, Ubelaker D, García-Pedrero J, C. Zapico S. “The big sleep: Elucidating the sequence of events in the first hours of death to determine the postmortem interval”. Science & Justice. 2019;59(4):418-424.

18. Noshy P. Postmortem expression of apoptosis-related genes in the liver of mice and their use for estimation of the time of death. International Journal of Legal Medicine. 2020

Nathan Routledge -12548189 26

19. Shafeeq T, Malik A, Zaheer A, Lee K. Differentiated transcriptional regulation before and after death in Indianmeal moth Plodia interpunctella. Journal of Asia-Pacific Entomology. 2020;23(4):1096-1100.

20. Javan G, Can I, Finley S, Soni S. The apoptotic thanatotranscriptome associated with the liver of cadavers. Forensic

Science, Medicine, and Pathology. 2015;11(4):509-516.

21. Javan G, Kwon I, Finley S, Lee Y. Progression of thanatophagy in cadaver brain and heart tissues. Biochemistry and

Biophysics Reports. 2016;5(2016):152–159.

22. Zhu Y, Wang L, Yin Y, Yang E. Systematic analysis of gene expression patterns associated with postmortem interval in human tissues. Scientific Reports. 2017;7(1).

23. Fais P, Mazzotti M, Teti G, Boscolo-Berto R, Pelotti S, Falconi M. HIF1a protein and mRNA expression as a new marker for post mortem interval estimation in human gingival tissue. Journal of Anatomy. 2018;2018(232):1031-1037.

24. Ferreira, P.G., Muñoz-Aguirre, M., Reverter, F. et al. The effects of death and post-mortem cold ischemia on human tissue transcriptomes. Nature Communications 2018;9(490)

25. Tolbert M, Finley S, Visonà S, Soni S, Osculati A, Javan G. The thanatotranscriptome: Gene expression of male reproductive organs after death. Gene. 2018;675:191-196.

26. González-Herrera L, Márquez-Ruiz A.B, Serrano M.J, Ramos V, Lorente J.A, Valenzuela A. mRNA expression patterns in human myocardial tissue, pericardial fluid and blood, and its contribution to the diagnosis of cause of death. Forensic

Science International. 2019;302(2019):109876

27. Uerlings S, Madea B, Grabmüller M. A comparison of endogenous and exogenous RNA reference marker as T relevant for accurate Post-Mortem Interval estimation. Forensic Science International: Genetics Supplement Series.

2019;7(2019):129–131.

28. Javan G, Hanson E, Finley S, Visonà S, Osculati A, Ballantyne J. Identification of cadaveric liver tissues using thanatotranscriptome biomarkers. Scientific Reports. 2020;10(1).

29. Bracken M. Why animal studies are often poor predictors of human reactions to exposure. Journal of the Royal Society

of Medicine. 2009;102(3):120-122.

30. Bryda E. The Mighty Mouse: The Impact of Rodents on Advances in Biomedical Research. The Journal of the Missouri

State Medical Association. 2013;110(3):207-211.

31. Reed J, Doctor K, Rojas A, Zapata J, Stehlik C, Fiorentino L et al. Comparative Analysis of Apoptosis and Inflammation Genes of Mice and Humans. Genome Research. 2003;13(6b):1376–1388.

32. Eimon, P.M., Ashkenazi, A. The zebrafish as a model organism for the study of apoptosis. Apoptosis. 2010;15:331–349. 33. Harris SE, Riggio V, Evenden L, Gilchrist T, McCafferty S, Murphy L, et al. Age-related gene expression changes, and transcriptome wide association study of physical and cognitive aging traits, in the Lothian Birth Cohort 1936. Aging. 2017;9(12):2489-2503.

34. Glass D, Viñuela A, Davies M, Ramasamy A, Parts L, Knowles D et al. Gene expression changes with age in skin, adipose tissue, blood and brain. Genome Biology. 2013;14(7):75.

35. Kukurba K, Montgomery S. RNA Sequencing and Analysis. HHS Author Manuscripts. 2015;11:951-969.

Nathan Routledge -12548189 27

Appendix

The appendix contains the search strategy for obtaining the papers looked at during this review

A- Animal H-Human R-Review

• Search website: lib.uva.nl

o Search term: “Thanatotranscriptome”

§ Search parameters: publication date ≥2016 § Results: 21

§ Papers based on the title and abstract that were deemed of interest:8 • Tolbert et al., 2018 (H)

• Javan et al., 2020 (H) • Halawa et al., 2019 (A) • Halawa et. Al., 2018 (A) • Scott et al., 2020. (H/R) • Shafeeq et al., 2020 (A) • Hunter et al., 2017 (A) • Javan et al., 2016 (H)

o Search term: “post-mortem genes forensics”

§ Search parameters: publication date ≥2016, articles only § Results: 840 (too many to check all)

§ New papers based on the title and abstract that were deemed of interest: 3 • Fais et al., 2018 (H)

• Ferreira et al., 2018 (H) • Zhu et al., 2017 (H)

o Search term: “post-mortem RNA forensics”

§ Search parameters: publication date ≥2016, articles only § Results: 386

§ New papers based on the title and abstract that were deemed of interest: 2 • Uerlings et al., 2019 (H)

• Ibrahim et al., 2019 (A) • Search website: google scholar

o Search term: “Thanatotranscriptome”

§ Search parameters: publication date ≥2016 § Results: 27

§ New papers based on the title and abstract that were deemed of interest: 5 • Noshy, 2020 (A)

• Bordonaro, 2019 (Hypothetical Paper) • Pozhitkov et al., 2016. (A, same as hunter) • Sawaguchi and Taki, 2018 (H)

• Martínez et al., 2019 (A) • Papers of interest based on common references

o Javan et al., 2015 (H) o Ma et al., 2015 (A)