1. Introduction
Human forensic taphonomic facilities (HFTFs) are becoming increasingly popular as they allow the comprehensive study of the complex interplay of the intrinsic and extrinsic factors that contribute to cadaveric decomposition. To date, there are eleven HFTFs worldwide including eight in the USA (8), Australia (1), Canada (1) and Europe (1), each operating differently depending on the national or state laws regarding treatment and disposition of human remains. The human skeletal remains associated with HFTFs can provide a valuable resource for conducting biomolecular studies that have significant potential for forensic applications such as PMI estimation [
1,
2,
3,
4,
5,
6]. Amongst the available biomolecular fields of investigations for PMI estimation, metabolomics is acquiring a particular attention in the forensic arena. A recent study from Pesko et al. [
7] recently applied metabolomics on muscle tissue from human individuals decomposed at the Forensic Anthropology Center at Texas State (FACTS). Results showed that over a PMI of maximum 19 days, xanthine, uracil, choline phosphate, n-acetylneuraminate, and 1-methylnicotinamide could be good metabolomic markers for PMI estimation. Similarly, Wood and Shirley [
8] analyzed anterior quadriceps muscle tissue from one donor decomposing at the Forensic Anthropology Center (FAC) at the University of Tennessee and showed that, across 24 days PMI, sterol sulfates, cholines, ethanolamines and phosphatidylglycerols decrease significantly. In contrast, free fatty acids were shown to have a positive relationship with PMI [
8]. At the same institution (University of Tennessee Anthropology Research Facility), a larger test involving the lipidomic analysis of vastus lateralis muscle from a total of 31 individuals over 2000 accumulated degree days showed a robust correlation with PMI based on regression, especially for phosphatidylglycerol (PG) 34:0 and phosphatidylethanolamine (PtdE) 36:4 [
9]. An unpublished work from Dudzik et al. [
10] also proved the applicability of molecular profiling of bone tissue based on lipidomics with mass spectrometry to estimate PMI. They showed that three months are a crucial interval in the decomposition of phosphosphatidylcholine, that could be therefore a useful marker for short PMI estimation.
All the above-mentioned studies clearly show the potential of metabolomics in forensic science, as well as and the importance of HFTFs in providing unvaluable human models for the study of decomposition. However, while there are some commonalities amongst all the facilities, many of the procedures such as chemical and thermal processing skeletal remains to remove soft tissues and fluids are unique to each facility. The lack of common guidelines between HFTFs on the treatment of bones for long-term storage (a procedure known as “soft maceration”) significantly impacts the possibility of conducting biomolecular research on the donated osteological collections across different HFTFs, thereby drastically limiting the potential of obtaining comparable forensic and taphonomic information from the bones when conducting novel research involving state-of-the-art biochemical and molecular methods. As an example, the National Institute of Standards and Technology (NIST) guidelines clearly advise against the use of excessive soaking and the use of heat and chemical agents on human bones [
11] as this can damage or destroy bone, teeth, and DNA. It is not known how soft maceration procedures can affect the survival of bone biomolecules (including DNA, proteins, lipids, and metabolites), and therefore the reliability of the results obtained from processed osteological records must be questioned. Furthermore, inconsistencies on the chemical and physical treatment of the bones could make comparison of data between collections nearly impossible and the use of data in casework unreliable. Several studies have so far demonstrated the importance of developing standardized protocols for the extraction of low molecular weight compounds from different tissues and fluids by controlling extraction buffer, temperature and other factors that can affect the stability of metabolites and lipids [
12,
13,
14,
15,
16]. More specifically, it has been demonstrated that thermal treatment is one of the main factors that can induce the degradation of both compounds [
17,
18,
19]. Due to the increasing interest in the application of small compound profiling for forensic applications, it is essential to evaluate potential sources of biases induced by sample treatment and handling. Moreover, it is fundamental to understand if such metabolomics and lipidomics profiling studies can provide meaningful information when testing macerated bones, to save resources whilst protecting precious human material.
These facilities have now been in use from several decades and keep offering invaluable samples for anthropological research. However, there is a lack of standardized guidelines for the treatment of human remains. For the first time, to the best of our knowledge, we are highlighting an issue that could compromise all future metabolomics studies based on such material. The current study aims to investigate the effect of three different maceration procedures developed on routinely performed protocols at Forensic Anthropology Center at Texas State University (FACTS) facility (n = 1), employing a bath of deionized water in laundry detergent at 87 °C, and from the Southeast Texas Applied Forensic Science (STAFS) facility (n = 2), consisting of a milder bath of detergent and deionized water at 55 °C, on metabolites and lipids extracted from bovine bone tissue and analyzed by LC-MS/MS. Results of the present study have the objective of clarifying whether these protocols are suitable or not to preserve the biomolecular integrity of the bone matrix. In particular, we aim at providing a preliminary understanding on how different protocols could differentially affect the bone metabolomic and lipidomic profile in order to ensure the comparability of studies conducted on material sourced at different HFTFs, to increase the awareness on this issue and to eventually stimulate debates on future change in policies for more appropriate and standardized treatment of human remains.
3. Results
PCA for metabolomics assays, both in positive (ESI+) and negative (ESI−) ionization modes, explained 71% of the total variance. No clear inter-individual differences can be observed between the three individuals, as the non-macerated bones form a unique cluster. However, it is possible to see a clear planar separation on the first dimension between fresh and macerated samples (
Figure 2).
Protocols #1 and #2 (differing uniquely by the duration of the procedure) cluster together in the upper right region of the plot, whereas samples processed with protocol #3, which involves a higher water temperature (87 °C), seem to behave differently and clusters in a different quadrant in comparison with the other two groups, suggesting that the second component may retain variation between the different protocols. The hierarchical clustering shows that similar metabolomic profiles are shared by samples subjected to the same maceration treatments. Samples cluster according to their condition (e.g., macerated vs. non-macerated) but also to the specific protocol used, suggesting that all variables (i.e., duration and temperature of the protocol and cleaning agent) play a role in the final metabolomic profile. As can be seen in the heatmap of metabolites analyzed in ESI+ (
Figure 2B), the bulk of the metabolites profiled are more abundant in the fresh samples than in the macerated ones. The heatmap also suggests the presence of a cluster of compounds on the top right of the plot that are more abundant in the samples treated with protocols #1 and #2 (55 °C), as well as of another cluster of compounds at the bottom of the plot found more abundant in the bones processed specifically with protocol #3 (87 °C). Very similar same trends can also be identified in ESI−, where clear clusters can be observed for the two main groups (e.g., macerated vs. non-macerated) as well as for the three different maceration protocols (
Figure 2D). Several identified molecules are more abundant in the non-treated samples, despite a group of compounds being more abundant in samples treated with protocols #1 and #2 and another group abundant in those treated with protocol #3.
Compounds identified and selected with ESI+ after the filtration steps reported in the Methods were then evaluated in detail for their abundance in non-macerated vs. macerated samples and their significance level (
Figure 3 and
Table S1). As previously mentioned, most of the compounds are confirmed to be more highly expressed in the fresh specimens. However, seven compounds are found to be more abundant in the macerated group: 4-guanidinobutyric acid, a degradation product of arginine, Bis (2-ethylhexyl) amine, cannabidiol and dimethyldecylamine N-oxide, not commonly identified as part of endogenous metabolism. N,N-dimethyldecylamine, N-oxide and 1-tetradecylamine, both known to be present in washing products and surfactants, and dodecyltrimethylammonium, another surfactant commonly used as oil-dissolving agent. In ESI− (
Figure S1), eight compounds like these are found to be significantly more abundant in the macerated group compared to the fresh counterpart. They include (+/−)9,10-dihydroxy-12-octadecenoic acid, a derivative of linoleic acid diol, that has been reported to be toxic in human’s tissue preparations, myristyl sulfate, an exogenous compound used in cosmetics, dodecyl sulfate, an anionic organic compound used as surfactant, (16R,17S)-8-(4-morpholinyl) and xylene sulfonate, respectively an anionic surfactant and a coupling agent in commercial dishwashing products, 13-hydroxyoctadecadienoic acid (13-HODE), a linoleic acid oxidation product that has vasoactive properties, neopterin, a catabolic product of guanosine triphosphate, and 3-(cyclohexylamino)-2-hydroxy-1-propanesulfonic acid (CAPSO), a chemical used as buffering agent in biochemistry. In contrast, the metabolites more abundant in the fresh samples, in both ionisation modes, are mostly endogenous compounds normally involved in tissue metabolism.
Moving to the lipidomic profiling, for both ESI+ (
Figure 4A) and ESI− (
Figure 4B), the two groups (fresh vs. macerated) are less clustered than in the metabolomics assays, despite the cumulative variance explained is, respectively, 55% and 67%. In ESI+, there is a separation on the first PCA component especially between the samples macerated with the protocol #3 and the fresh ones, whereas the other macerated ones are located closer to the fresh ones on the first component, despite their separation on the second component. Additionally, in this case, consistently with what observed for the metabolomic results, the bone processed with protocol #3 shows a much clearer separation from the remaining bones treated with protocols #1 and #2 and from the fresh ones, with a clear distinctive pattern also in the respective heatmap (
Figure 4C). In ESI−, PCA shows that bones treated with protocol #1 cluster fresh bones, whereas bones macerated with the other protocols cluster in other quadrants, overall complicating the interpretation of the obtained results. When considering the hierarchical clustering (
Figure 4D), two fresh bones and the two specimens processed with protocols #1 and #2 form a cluster, whereas the bone treated with protocol #3 are in a separate cluster together with one of the three untreated bones. According to the heatmap, protocols #1 and #2 share similar profiles, that appear to be distinct from the fresh bones despite their positioning in the tree. These results suggest that differences in the lipidic content are not related exclusively to the maceration protocols tested, and that additional variables may play a role in the post-maceration lipidomic signature. No lipid compounds are seen to behave consistently according to the three different maceration protocols, despite it is noticeable a class of lipids more abundant only in the samples processed with protocol #3 in both ionization modes (
Figure 3C,D). Considering only the differences between fresh and macerated bones in ESI+, also due to the larger sample size,
t-test reveals that acetyl-CoA (AcCa), ceramides (Cer), phosphatidylethanolamines (PE), and phosphatidylserines (PS) are the classes of lipids significantly more abundant in the fresh samples (
Table S1. In the macerated sample there is higher abundance of several triglycerides (TG) of variable chain length, sphingomyelins (SM), phosphatidylinositols (PI), phosphatidylcholines (PC), two diglycerides (DG) and one compound for lysophosphatidylcholine (LPC), lysophosphatidylethanolamine (LPE), and lysophosphatidylserine (LPS). Boxplots and
t-test results are reported, respectively, in
Figure S2 and
Table S1. Regarding lipidomics ESI− (
Figure S3 and
Table S1), significant differences were seen for ceramides with higher abundances for the fresh samples. The same trend was seen for PSs and one LPS. In contrast, LPCs, LPEs, PCs and PIs show higher abundances in the macerated samples confirming what also shown by lipids profiled in ESI+ mode.
4. Discussion
This pilot study demonstrates how procedures normally adopted by some HFTFs to clean and prepare human bones for long-term storage may impact the ability to perform biomolecular analyses aimed at obtaining reliable and unbiased results from bones for PMI estimation. The increased interest in metabolomics and lipidomics applied to the forensic field, in addition to the technological developments in mass spectrometry that have characterized the recent years, have started to pose questions on the reliability of the biomolecular results obtainable from the skeletal collections at the HFTFs. Due to the small sample size, results of the present study are considered to be preliminary, and the overall conclusions of the study are uniquely intended to be a proof-of-concept to start building awareness around this topic, which has never been considered a major issue until now. Therefore, statistical analyses were used more as a guide in the interpretation of the results rather than a tool to draw conclusive evidence, and further studies on a larger sample size, ideally human, will be beneficial to expand on these findings. We used an animal model to test three different maceration protocols used at HFTFs and then applied a high-sensitivity LC-MS/MS approach for small compounds and lipids profiling. As expected, our results show a general decrease in the normalized intensities of all the endogenous metabolites (e.g., amino acids and nucleobases) as well as the presence of new compound associated with the chemical detergents used during the cleaning process (e.g., surfactants). Changes in the lipidomic profile resulted in being the most complex to interpret, with certain classes being more abundant in the macerated samples and with others being more intense in the fresh ones. Despite the use of animal proxies has been largely employed in forensic sciences, there are several factors that could not be controlled and could have an impact on the outcome of the study. Amongst those there are nutrition, health status and age of the animal. However, due to the source of the bones employed in the experiment (butchered animals to be used for human consumption) it is reasonable to assume that all animals were similarly fed, treated, shared good health conditions and were slaughtered at the same age. This is supported by the data shown in
Figure 2 and
Figure 4, where the fresh individuals closely clustered together, suggesting overall a minimal interindividual variability.
When considering the metabolites, the PCA score plot (
Figure 2A) revealed that specimens treated using protocol #1 and #2 differ significantly from protocol #3. It is reasonable to think that the high temperature employed (87 °C compared to 55 °C), combined with the different type of detergent used, drives the variations observed, and that the length of the procedure (two days vs. one week) affects the results only marginally. There is evidence in literature supporting the fact that high temperatures may induce significant changes in the chromatographic pattern of the Metabolome. Fang et al. [
19] tested human serum by means of LC-MS/MS after exposing the material to temperatures ranging between 60 °C and 250 °C for an exposure time up to 300 s and proved that the harsher treatment induced alterations in up to 40% of the chromatographic peaks. Nucleosides and nucleotides were transformed into pure derivates, while tri- and di-organophosphates into their mono-components [
19]. Moreover, it must be noted that water, in general and regardless of its temperature, poses issues for metabolomic analyses when samples are submerged prior to analysis. In fact, water is the main polar solvent used for metabolite extraction, including the Folch extraction protocol [
20] employed in the present study, and it is not surprising therefore that submerging bones in water as a pre-treatment, can severely affect their metabolomic profiles. The potential release of small molecules and compounds in solution during the maceration of the bones can explain why, in general, most of the endogenous metabolites are observed only in the non-macerated samples and not in their treated counterparts. In addition to the intrinsic metabolites, cannabidiol, 1-tetradecylamine bis(2-ethylhexyl) amine, N,N-dimethyldecylamine, N-oxide, dodecyltrimethylammonium, xylenesulfonate, (16R,17S)-8-(4-morpholinyl), dodecyl sulfate and myristyl sulfate were observed specifically in the samples treated with protocols #1 and #2, that share the same detergent, showing that traces of the detergent used during the maceration process can be identified in the cortical portion of the bones, altering the original metabolomic fingerprint of the specimen. Similarly, for protocol #3 (where a different cleaning agent was used), other exogenous contaminants such as 4-Guanidinobutyric acid, (+/−)13-HODE, (+/−)9,10-dihydroxy-12Z-octadecenoic acid, neopterin, and CAPSO were identified.
t-tests run on the single metabolites when comparing fresh and macerated samples further support the hypothesis that maceration can induce not only the loss of specific metabolites, but also introduce specific contaminants depending on the chemicals used. The investigation of the changes associated with the metabolome is popular in food science and can be related to this work. For example, an experiment run to estimate the ideal combination of cooking temperature and duration for Thai chicken bone soup showed an increase in the abundance of free amino acid (AAs) and essential AAs (EAAs) in the bone broth following boiling, with an associated and expected decrease in their concentration in the bone tissue [
22]. This fully supports our findings, confirming a general decrease in the endogenous metabolites following maceration, as shown in
Figure 3.
Previous works conducted on several soft tissues in forensic science highlighted the great potential for metabolomics to reveal information on PMI [
7,
23,
24]. Despite bones may offer a similar success rate when analyzed via metabolomics for PMI, the changes introduced by these protocols can significantly affect their metabolomic signature, consequently posing severe limitations on the use of treated osteological collection for this purpose. As an example, hypoxanthine, choline, creatine, and betaine have been previously found to be good candidates for PMI estimation in vitreous humor [
24,
25]. Similarly, xanthine, uracil, choline phosphate, and 1-methylnicotinamide have been identified as potential bio markers for PMI estimation in muscle tissue [
7]. Our study showed that these metabolites are also present in the non-treated bone (
Figure 3), and that their abundance is severely affected by the maceration treatment. Therefore, these molecules may lose their predictive power for PMI when bones are treated prior long-term storage. As a result, it is of primary importance to ensure the best possible post-mortem preservation of metabolites in bones, reducing the invasiveness of any procedure related to bone tissue handling and cleaning to allow an optimal performance of the subsequent biochemical studies.
Despite single lipid compounds do not seem to change consistently according to the different protocols used, certain lipid classes show a more regular pattern when considering either fresh or macerated samples. The separation between pre- and post-treatment samples, as well as those existing between different protocols, are better appreciated in ESI+ mode. In fact, when investigated in ESI−, the modifications induced by the maceration processes do not show clear distinctions between any of the groups, including fresh vs. macerated. In a previous study, Nawar et al. [
18] showed that lipids in glyceride mixtures are unaltered at 60 °C, showing that the thermal hydrolysis of triglycerides (TG) results in the preferential release of shorter-chains TG and of unsaturated fatty acids [
18]. More recently Charuwat et al. [
17] tested the effect thermal hydrolysis between 90–150 °C and 160–250 °C for a maximum of eight hours on long chain fatty acids in sludges by means of gas chromatography. Results showed that long chain fatty acids are resistant to thermal hydrolysis treatments, with approximately only 1% of saturated fatty acids becoming shortened at 160 °C [
17].
It is difficult to determine if a shortening of the lipid chains was directly linked with the maceration processes due to the experimental set up. However, the observed modification in the lipidome could be potentially linked to the fact that the bones used in the experiment were fresh and had some remaining adhering soft tissues and blood on their surface. When macerating them at temperatures above 50 °C, some lipid classes originated from these remaining soft tissues may have been released in the water bath and may have eventually adhered onto the periosteal surface, increasing therefore the abundance of these compounds specifically in the macerated bones, particularly at higher temperatures (87 °C,
Figure 4).
Table 2 summarizes the most striking outcome of the study highlighting those certain classes of lipids specifically increase or decrease after the treatment. Generally, AcCA, Cer, PE, and PS are less abundant in the macerated samples, while the opposite trend is true for the remaining classes in both ionization modes.
Despite the limited body of research in the application of lipidomics analyses to forensics, several studies have shown the potential of lipidomics for PMI estimation from several tissues. Wood and Shirley [
8] sampled the anterior quadriceps muscle from one donated cadaver 1, 9, and 24 days post-mortem, and showed the decline of sterol sulphates, choline plasmalogens, ethanolamine plasmalogens, and phosphatidylglycerols and the contemporaneous increase in free fatty acids with increasing PMIs, indicating that the decrease in membrane glycerophospholipids was the best indicator for PMI estimation. To our knowledge, only one other study has conducted analyses on lipids extracted from bone material for forensic applications. Dudzik et al. [
26] analyzed dry trabecular bone marrow cores in their native state at the time of death and specific timepoints for up to two years post-mortem from vertebral bodies, tibiae, and calcanei of human donated cadavers. With direct infusion MS they identified stable lipids belonging to 56 lipid subclasses. after approximately seven years retrieving 76 potential N-acyl amino acids that could serve for PMI estimation. The same authors following-up the previous work [
10] analyzed the trabecular bone of calcaneus, proximal tibia, and 4th lumbar vertebra from 20 fresh donors with a maximum PMI of 24 months collecting samples every six months. Additionally, they analyzed a cross-sectional sample by coring bones belonging to 130 individuals curated in the William Bass Skeletal collection, with PMIs ranging from 3–37 years. Results from the calcanei showed an increase in PC 34:1 and decrease in PC 36:1 and PC 36:2 up to 12 months PMI. Once again, PC is one of the lipid classes affected by all the maceration procedures employed in the present study. More generally, they showed that the first three months are crucial for the degradation of phosphatidylcholines in bones, after which PCs reach a plateau, making them less effective for prolonged PMI estimations [
10]. Despite the non-treated bones in the study (e.g., those sampled from the decomposing cadavers) may have given reliable observations, it is difficult to exclude that the relatively stable abundances of PCs observed by Dudzik et el. [
10] in the samples originated from the curated collection were not linked to the maceration treatments used at the facility, as specifically observed in this study for PCs. Therefore, additional studies on non-treated remains for prolonged PMIs (e.g., >2 years) may be crucial to shine a light on this topic and to clarify which compounds may be the best indicators for PMI estimation in bones. Despite metabolomics is not yet at a stage in which could be applied to forensic investigations, the development of new methods profoundly depends on the analysis of taphonomically “controlled” human material, that can therefore only be sourced from HFTFs. This undoubtedly demonstrates the potential that lies in the skeletal collections provided by these facilities, and the consequent necessity to standardize the bone cleaning procedures aiming for the creation of long-term skeletal collections of which biomolecular and biochemical profiles are not, or are minimally (and in a controlled way), affected by the treatment protocols.