Thanatotranscriptome: genes actively expressed after organismal death

A continuing enigma in the study of biological systems is what happens to highly ordered structures, far from equilibrium, when their regulatory systems suddenly become disabled. In life, genetic and epigenetic networks precisely coordinate the expression of genes -- but in death, it is not known if gene expression diminishes gradually or abruptly stops or if specific genes are involved. We investigated the unwinding of the clock by identifying upregulated genes, assessing their functions, and comparing their transcriptional profiles through postmortem time in two species, mouse and zebrafish. We found transcriptional abundance profiles of 1,063 genes were significantly changed after death of healthy adult animals in a time series spanning from life to 48 or 96 h postmortem. Ordination plots revealed non-random patterns in profiles by time. While most thanatotranscriptome (thanatos-, Greek defn. death) transcript levels increased within 0.5 h postmortem, some increased only at 24 and 48 h. Functional characterization of the most abundant transcripts revealed the following categories: stress, immunity, inflammation, apoptosis, transport, development, epigenetic regulation, and cancer. The increase of transcript abundance was presumably due to thermodynamic and kinetic controls encountered such as the activation of epigenetic modification genes responsible for unraveling the nucleosomes, which enabled transcription of previously silenced genes (e.g., development genes). The fact that new molecules were synthesized at 48 to 96 h postmortem suggests sufficient energy and resources to maintain self-organizing processes. A step-wise shutdown occurs in organismal death that is manifested by the apparent upregulation of genes with various abundance maxima and durations. The results are of significance to transplantology and molecular biology.

dilution factor and determining the isotherm model (e.g., Freundlich and/or Langmuir) 160 that best fit the relationship between signal intensities and gene abundances. 161 Consider zebrafish gene transcripts targeted by A_15_P110618 (which happens to be one were further converted to log10 and are shown in external file Fish_log10_AllProfiles.txt. 173 Details of the calibration protocols to calculate gene expression values, i.e., mRNA 174 relative abundances, are provided in our recent paper where we describe the "Gene 175 Meter" [6]. 176 Statistical analysis. Abundance levels were log-transformed for analysis to stabilize the 177 variance. A one-sided Dunnett's T-statistic was applied to test for increase at one or more 178 postmortem times compared to live control (fish) or time 0 (mouse). A bootstrap 179 procedure with 10 9 simulations was used to determine the critical value for the Dunnett 180 statistics in order to accommodate departures from parametric assumptions and to 181 account for multiplicity of testing. The profiles for each gene were centered by 182 subtracting the mean values at each postmortem time point to create "null" profiles.

212
Similar quantities of total mRNA were extracted from zebrafish samples for the first 12 h 213 postmortem (avg. 1551 ng per µl tissue extract) then the quantities abruptly decreased 214 with postmortem time (Table S1). The quantities of total mRNA extracted from the 215 mouse liver samples were about the same for the first 12 h postmortem (avg. of 553 ng 216 per µl tissue extract) then they increased with time (Table S2). The quantities of total 217 mRNA extracted from the mouse brain samples were similar (avg. of 287 ng per µl tissue 218 extract) for all postmortem times (Table S2). Hence, the amount of total mRNA 219 extracted depended of the organism/organ/tissue and time.  Figure 1 shows the sum of all gene abundances calculated from the calibrated probes with 226 postmortem time. In general, the sum of all abundances decreased with time, which 227 means that less targets hybridized to the microarray probes. In the zebrafish, mRNA 228 decreased abruptly at 12 h postmortem (Fig 1A), while for the mouse brain Fig 1B), total 229 mRNA increased in the first hour and then gradually decreased. For the mouse liver, approach to another molecular approach (i.e., Agilent Bioanalyzer). Hence, total mRNA 234 abundances depended on the organism (zebrafish, mouse), organ (brain, liver), and 235 postmortem time, which are aligned with previous studies [8,9,10,11,12,13].  243 The abundance of a gene transcript is determined by its rate of synthesis and its rate of 244 degradation [14]. We focused on genes that show a significant increase in RNA 245 abundance --relative to live controls --because these genes are likely to be actively 246 transcribed in organismal death despite an overall decrease in total mRNA with time. An 247 upregulated transcription profile was defined as one having at least one time point where 248 the abundance was statistically higher than that of the control (Fig 2 A to 2C). It is 249 important to understand that the entire profiles, i.e., 22 data points for the zebrafish and 250 20 points for the mouse, were subjected to a statistical test to determine significance (see 251 Materials and Methods). We found 548 zebrafish transcriptional profiles and 515 mouse 252 profiles were significantly upregulated. The fact that there are upregulated genes is 253 consistent with the notion that there is still sufficient energy and functional cellular 254 machinery for transcription to occur --long after organismal death. Based on GenBank gene annotations, we found that among the upregulated genes for the 269 zebrafish, 291 were protein-coding genes (53%) and 257 non-coding mRNA (47%) and, 270 for the mouse, 324 known protein-coding genes (63%), 190 non-coding mRNA (37%), 271 and one control sequence of unknown composition. Hence, about 58% of the total 272 upregulated genes in the zebrafish and mouse are known and the rest (42%) are non-273 coding RNA. 274 Examples of genes yielding transcripts that significantly increased in abundance with  Non-random patterns in transcript profiles 284 Ordination plots of the significantly upregulated transcript profiles revealed prominent 285 differences with postmortem time (Fig 2D and 2E), suggesting the expression of genes 286 followed a discernible (non-random) pattern in both organisms. The biplots showed that 287 203 zebrafish transcript profiles and 226 mouse profiles significantly contributed to the 288 ordinations. To identify patterns in the transcript profiles, we assigned them to groups 289 based on their position in the biplots. Six profile groups were assigned for the zebrafish 290 (A to F) and five groups (G to K) were assigned for the mouse. Determination of the 291 average gene transcript abundances by group revealed differences in the shapes of the 292 averaged profiles, particularly the timing and magnitude of peak transcript abundances, 293 which accounted for the positioning of data points in the ordinations. 294 Genes coding for global regulatory functions were examined separately from others (i.e., 295 response genes). Combined results show that about 33% of the upregulated genes in the 296 ordination plots were involved in global regulation with 14% of these encoding 297 transcription factors/transcriptional regulators and 19% encoding cell signaling proteins 298 such as enzymes, messengers, and receptors (Table S3). The response genes accounted 299 for 67% of the upregulated transcripts. 300 The genes were assigned to 22 categories (File S8) with some genes having multiple 301 categorizations. For example, the Eukaryotic Translation Initiation Factor 3 Subunit J-B 302 (Eif3j2) gene was assigned to protein synthesis and cancer categories [15]. 303 Genes in the following functional categories were investigated: stress, immunity, 304 inflammation, apoptosis, solute/ion/protein transport, embryonic development, epigenetic 305 regulation and cancer. We focused on these categories because they were common to 306 both organisms and contained multiple genes, and they might provide explanations for 307 postmortem upregulation of genes (e.g., epigenetic gene regulation, embryonic 308 development, cancer). The transcriptional profiles of the genes were plotted by category 309 and each profile was ordered by the timing of the upregulation and peak transcript  Stress response 314 In organismal death, we anticipated the upregulation of stress response genes because 315 these genes are activated in life to cope with perturbations and to recover homeostasis 316 [16]. The stress response genes were assigned to three groups: heat shock protein (Hsp), 317 hypoxia-related, and 'other' responses such as oxidative stress. 318 Hsp In the zebrafish, upregulated Hsp genes included: 'Translocated promoter region' 319 (Tpr), Hsp70.3, and Hsp90 (Fig 3) The timing and duration of Hsp upregulation varied by organism. In general, activation 329 of Hsp genes occurred much later in the zebrafish than the mouse (4 h vs. 0.5 h 330 postmortem, respectively). There were also differences in transcript abundance maxima 331 since, in the zebrafish, maxima were reached at 9 to 24 h, while in the mouse maxima   Other stress responses 365 In the zebrafish, upregulated response genes included: Alkaline ceramidase 3 (Acer3), 366 Peroxirodoxin 2 (Prdx2), Immediate early (Ier2), Growth arrest and DNA-damage-367 inducible protein (Gadd45a), Zinc finger CCH domain containing 12 (Zcchc12), 368 Corticotropin releasing hormone receptor 1 (Crhr1), and Zinc finger AN1-type domain 4 369 (Zfand4) (Fig 3). The Acer3 gene encodes a stress sensor protein that mediates cell-  Ras association domain family 6 (Rassf6) (Fig 6). The Jdp2 gene encodes a protein that 539 represses the activity of the transcription factor activator protein 1 (AP-1) [92]. The zebrafish [142]. The Rgs4 gene encodes a protein involved in brain development [143]. 705 The Prrt4 gene encodes a protein that is predominantly expressed in the brain and spinal   Serine/threonine-protein kinase (Sbk1), and Tyrosine-protein kinase transmembrane 785 receptor (Ror1) (Fig 9).  (Fig 9). 807 These genes were classified as "cancer genes" in a Cancer Gene Database [7] (Fig 9). 808 The timing, duration and peak transcript abundances differed within and between 809 organisms. Note that some transcripts had two abundance maxima. In the zebrafish, this Probable JmjC domain-containing histone demethylation protein 2C (Jmjd1c) (Fig 10). 851 The Ttll10 gene encodes a polyglycylase involved in modifying nucleosome assembly genes stopped (Fig 11, "All genes"). It should be noted that the same pattern was found 880 in stress, transport and development categories for both organisms. However, in the 881 zebrafish, the immunity, inflammation, apoptosis and cancer categories differed from the 882 mouse. Specifically, the genes in the immunity, inflammation, and cancer categories 883 were upregulated much later (1 to 4 h) in the zebrafish than the mouse, and the duration 884 of upregulation was much shorter. For example, while 90% of the genes in the immunity 885 and inflammation categories were upregulated in the mouse within 1 h postmortem, less 886 than 30% of the genes were upregulated in the zebrafish (Fig 11), indicating a slower 887 initial response. It should be noted that while the number of upregulated immunity genes Active gene expression or residual transcription levels? 905 One could argue that our study identifies only residual transcription levels of pre-906 synthesized mRNA (rather than newly-synthesized mRNA) in dead tissues that happen to 907 be enriched with postmortem time. In other words, the observed upregulation of genes 908 may be viewed as an artifact merely reflecting the "enrichment of specific mRNA 909 transcripts" (e.g. stable mRNA) with time. The data, however, does not support this idea 910 because if it were true, one would expect stable transcripts to monotonically increase with 911 time, as they become more enriched (higher abundances) with postmortem time. The 912 data show that the transcripts of most upregulated genes did not display monotonic 913 behavior; rather, the transcripts reached abundance maximum (peak) or maxima (peaks) 914 at various postmortem times (Fig 2D and 2E). This finding should not be a surprise 915 because a statistical procedure was implemented to detect genes that were significantly 916 upregulated -which is essentially selecting gene transcriptional profiles that had peaks. 917 The statistics for the procedure was calibrated with more than a billion simulations. The 918 simulation process corrected for multiple comparisons. The residual transcription level 919 and enrichment idea is also not supported by transcriptional profiles displaying an up-, 920 down-, and up-regulation pattern, which putatively indicates feedback loops (Fig S3). complex systems (e.g., societies [191], government [192], electrical black outs [193]). 957 Yet, to our knowledge, no study has examined long-term postmortem gene expression of 958 vertebrates kept in their native conditions. The secondary motivation for the study was to 959 demonstrate the precision of Gene Meter technology for gene expression studies to 960 biologists who believe that high throughput DNA sequencing is the optimal approach. 961 Thermodynamic sinks 962 We initially thought that sudden death of a vertebrate would be analogous to a car driving 963 down a highway and running out of gas. For a short time, engine pistons will move up 964 and down and spark plugs will spark --but eventually the car will grind to a halt and 965 "die". Yet, in our study we find hundreds of genes are upregulated many hours 966 postmortem, with some (e.g., Kcnv2, Pafr, Degs2, Ogfod1, Ppp2rla, Ror1, and Iftm1) 967 upregulated days after organismal death. This finding is surprising because in our car 968 analogy, one would not expect window wipers to suddenly turn on and the horn to honk 969 several days after running out of gas. 970 Since the postmortem upregulation of genes occurred in both the zebrafish and the mouse 971 in our study, it is reasonable to suggest that other multicellular eukaryotes will display a 972 similar phenomenon. What does this phenomenon mean in the context of organismal 973 life? We conjecture that the highly ordered structure of an organism -evolved and 974 refined through natural selection and self-organizing processes -undergoes a 975 thermodynamically driven process of spontaneous disintegration through complex 976 pathways, which apparently involve the upregulation of genes and feedback loops. While 977 evolution played a role in pre-patterning of these pathways, it does not play any role in its 978 disintegration fate. One could argue that some of these pathways have evolved to favor 979 healing or "resuscitation" after severe injury. For example, the upregulation of 980 inflammation response genes indicate that a signal of infection or injury is sensed by the 981 still alive cells after death of the body. Alternatively, the upregulation may be due to fast 982 decay of some repressors of genes or whole pathways (see below). Hence, it will be of 983 interest to study this in more detail, since this could, for example, provide insights into 984 how to better preserve organs retrieved for transplantation.

985
Chemical automator -on the way down to equilibrium 986 As one would expect, a living system is a collection of chemical reactions linked together 987 by the chemicals participating in them. Having these reactions to depend on one another 988 to a certain extent, we conjecture that the observed upregulation of genes is due to 989 thermodynamic and kinetic controls that are encountered during organismal death. For 990 example, epigenetic regulatory genes that were upregulated included histone modification 991 genes (e.g., Histh1l) and genes interacting with chromatin (e.g., Grwd1, Chd3, Yeats, 992 Jmjd1c) (Fig 10). It is possible that the activation of these genes was responsible for the 993 unraveling of the nucleosomes, which enabled transcription factors and RNA 994 polymerases to transcribe developmental genes that have been previously silenced since 995 embryogenesis. The energy barrier in this example is the tightly wrapped nucleosomes 996 that previously did not allow access to developmental genes. Other energy or entropy 997 barriers could be nucleopores that allow the exchange of mRNA and other molecules 998 between the mitochondria and the cytosol (e.g., Tpr, Tnpo1, Lrrc59), or the ion/solute 999 protein channels (e.g., Aralar2, Slc38a4) that control intracellular ions that regulate 1000 apoptotic pathways [194,195]. 1001 The upregulation of genes indicates new molecules were synthesized. Hence, there was 1002 sufficient energy and resources (e.g., RNA polymerase, dNTPs) in dead organisms to 1003 maintain gene transcription to 96 h (e.g., Zfand4, Tox2, and Slc14a2) in the zebrafish and 1004 to 48 h (e.g., Deg2, Ogfod1, and Ifitm1) in the mouse. Gene transcription was apparently 1005 not prevented due to a lack of energy or resources. Several genes exhibited apparent 1006 regulation by feedback loops in their transcriptional profiles (e.g., Rbm45 and Cdc42 1007 genes in the mouse (Fig 8 and 9, respectively; Fig S3)). Hence, an underlying regulatory 1008 network appears to be still turning "on" and "off" genes in organismal death.

1009
Interrupt the shutdown? 1010 A living biological system is a product of natural selection and self-organizing processes 1011 [196]. Genes are transcribed and proteins translated in response to genetic and epigenetic 1012 regulatory networks that sustain life. In organismal death, we assumed most of the 1013 genetic and epigenetic regulatory networks operating in life would become disengaged 1014 from the rest of the organism. However, we found that "dead" organisms turn genes on 1015 and off in a non-random manner (Fig 2D and 2E). There is a range of times in which  Methodological validity 1035 The Gene Meter approach is pertinent to the quality of the microarray output obtained in 1036 this study because conventional DNA microarrays yield noisy data [197,198]. The Gene 1037 Meter approach determines the behavior of every microarray probe by calibration -which is analogous to calibrating a pH meter with buffers. Without calibration, the 1039 precision and accuracy of a meter is not known, nor can one know how well the 1040 experimental data fits to the calibration (i.e., R 2 ). In the Gene Meter approach, the 1041 response of a probe (i.e., its behavior in a dilution series) is fitted to either Freundlich or 1042 Langmuir adsorption model, probe-specific parameters are calculated. The "noisy" or 1043 "insensitive" probes are identified and removed from further analyses. Probes that 1044 sufficiently fit the model are retained and later used to calculate the abundance of a 1045 specific gene in a biological sample. The models take into consideration the non-linearity 1046 of the microarray signal and the calibrated probes do not require normalization 1047 procedures to compare biological samples. In contrast, conventional DNA microarray 1048 approaches are biased because different normalizations can yield up to 20 to 30% 1049 differences in the up-or down-regulation depending on the procedure selected [199][200][201][202]. 1050 We recognize that next-generation-sequencing (NGS) approaches could have been used 1051 to monitor gene expression in this study. However, the same problems of normalization 1052 and reproducibility (mentioned above) are pertinent to NGS technology [203]. Hence, 1053 the Gene Meter approach is currently the most advantageous to study postmortem gene 1054 expression in a high throughput manner.

1055
Practical implications 1056 The postmortem upregulation of genes in the mouse has relevance to transplantation 1057 research. We observed clear qualitative and quantitative differences between two organs 1058 (liver and brain) in the mouse in their degradation profiles (Fig 1). We also showed the 1059 upregulation of immunity, inflammation and cancer genes within 1 h of death (Fig 11). It This is the first study to demonstrate active, long-term expression of genes in organismal 1068 death that raises interesting questions relative to transplantology, inflammation, cancer, 1069 evolution, and molecular biology.

Supplementary Text
Figs S1, S2 and S3 Table S1 Other Additional Files for this manuscript include the following: Data files S1 to S8 as zipped archives: File S1.      Transcriptional profiles in the zebrafish (Acer3 gene) and mouse (Cdc42 and Rbm45 genes) by postmortem time. Red arrows, up-regulation; blue arrows, down-regulation. One-way T-tests show significant differences between means. Results suggest that the differences in upand down regulation by postmortem time are due to changes in regulation rather than changes in "residual transcription levels".