Epigenomic and Other Evidence for Cannabis-Induced Aging Contextualized in a Synthetic Epidemiologic Overview of Cannabinoid-Related Teratogenesis and Cannabinoid-Related Carcinogenesis

Background: Twelve separate streams of empirical data make a strong case for cannabis-induced accelerated aging including hormonal, mitochondriopathic, cardiovascular, hepatotoxic, immunological, genotoxic, epigenotoxic, disruption of chromosomal physiology, congenital anomalies, cancers including inheritable tumorigenesis, telomerase inhibition and elevated mortality. Methods: Results from a recently published longitudinal epigenomic screen were analyzed with regard to the results of recent large epidemiological studies of the causal impacts of cannabis. We also integrate theoretical syntheses with prior studies into these combined epigenomic and epidemiological results. Results: Cannabis dependence not only recapitulates many of the key features of aging, but is characterized by both age-defining and age-generating illnesses including immunomodulation, hepatic inflammation, many psychiatric syndromes with a neuroinflammatory basis, genotoxicity and epigenotoxicity. DNA breaks, chromosomal breakage-fusion-bridge morphologies and likely cycles, and altered intergenerational DNA methylation and disruption of both the histone and tubulin codes in the context of increased clinical congenital anomalies, cancers and heritable tumors imply widespread disruption of the genome and epigenome. Modern epigenomic clocks indicate that, in cannabis-dependent patients, cannabis advances cellular DNA methylation age by 25–30% at age 30 years. Data have implications not only for somatic but also stem cell and germ line tissues including post-fertilization zygotes. This effect is likely increases with the square of chronological age. Conclusion: Recent epigenomic studies of cannabis exposure provide many explanations for the broad spectrum of cannabis-related teratogenicity and carcinogenicity and appear to account for many epidemiologically observed findings. Further research is indicated on the role of cannabinoids in the aging process both developmentally and longitudinally, from stem cell to germ cell to blastocystoids to embryoid bodies and beyond.


Introduction
Aging is the ubiquitous fate of biota and involves progressive loss of function [1]. Whilst the unkempt appearance and often poor physical and/or mental health of the patient chronically dependent on drugs including cannabis is widely appreciated, formal studies of cellular aging following chronic drug exposure are curiously absent from the literature. Major recent advances in various fields including epigenomics, epidemiology, stem-cell physiology and the mechanics of mitotic and meiotic cell division provide a unique opportunity to conduct an investigative review of the interaction of cannabis exposure and aging with a view to stimulating formal investigation of the field with epigenomic and other aging biomarkers.
"Canalization" refers to the process described in Waddington's famous theory of cellular differentiation like a marble rolling down a landscape of hills and valleys and finding its energetically most favorable point, progressively becoming more terminally differentiated [9].
"Epigenomic regulatory pathways" refer to the many mechanisms of gene regulation including: DNA methylation, post-translational histone modifications, micro-RNAs, long noncoding RNAs, involvement in topologically defined domains and adjacency to transcription factories, closeness to the nuclear envelope (which suppresses gene transcription), involvement tin euchromatin or heterochromatin structure (the former promoting and the latter suppressing transcription), various post-transcriptional modifications of RNA including C6-adenosyl methylation, circular DNA structure and microchromsomes, amongst others.

Outline
The plan of this review is as follows. Firstly, twelve independent streams of empirical data for accelerated aging will be presented to make a strong case for accelerated biological aging associated with chronic cannabis exposure to set the context for the following discussion. Secondly, evidence for perturbation of some fundamental cellular machinery by cannabis exposure and withdrawal will be presented including alteration of the epigenomic machinery itself, modulation of various stem-cell factors and epigenomic interference with the chromosomal machinery of cell division. Epigenetic changes in brain and cardiovascular function are briefly considered as changes in these organs not only reflect but drive systemic aging, i.e., they are not only age-defining illnesses but also age-generating disorders. Thirdly, since cancer and congenital anomalies (birth defects) are both agerelated disorders and are clinical reflections of genotoxicity and/or epigenotoxicity and are heightened after cannabis exposure [25,66,[103][104][105][106][107][108][109][110][111][112][113][114][115][116][117][118], contemporary USA and European epidemiological findings are reviewed and form the backdrop for a contextual exploration of the recent powerful longitudinal epigenomic data published by Schrott and Murphy and colleagues on changes in the DNA methylome of human sperm after cannabis exposure and withdrawal annotated for many benign and malignant conditions [27]. These datasets are augmented by other recent organ specific studies highlighting genes of particular interest which are then interrogated in the Schrott data. Consideration is also given to genotoxic effects of cannabinoids more broadly including cannabidiol and ∆8-tetrahydrocannabinol (∆8THC).
These matters are set out in tabular form in Table 1.
It is concluded that these metrics collectively point towards cannabinoid-exposed tissues being of advanced biological age resulting in age related morbidity, and that this process is driven by cannabinoid-disruption of the human epigenome, with increasing global cannabis exposure to a much greater extent than is commonly realized [119], having far-reaching public health implications for the current and future generations

Methods
Literature Review. Evidence was overviewed from the authors prior knowledge of studies examining cannabis effects on mechanisms of ageing. A literature search was conducted of PubMed on 30 November 2022 using the two sets of search terms "cannabis AND aging" and "cannabinoids AND aging". Identified articles were manually searched. In total, 48 and 108 articles were identified from the raw searches. However, these dealt generally with only specific organ systems of aging (such as Alzheimer's disease or pancreatic aging) and not the whole field of the pathobiology of aging itself; or alternatively hypothesized about unproven aging preventative actions. Thus, it was not possible to identify any recent reviews of the impacts of cannabis or cannabinoids on the fundamental pathobiology of aging. This finding formally demonstrates the novelty of the present study.
The 12 streams of evidence referenced flow from cellular systems and mechanisms (Epigenomic Overview) through organ systems (Prefrontal Cortex and Brain), to health disorders including cancer (Carcinogenesis), to population impacts (on birth defects and cancer).
Data. Data on rates of congenital anomalies are taken from published reports in USA [103] and Europe [115,120]. Data on cancer rates are taken from published reports on USA [112][113][114]121] and Europe [121,122]. Epigenomic DNA methylation data were taken from the EWAS (Epigenome Wide Association Study) report of Schrott and colleagues relating to cannabis dependence and withdrawal in human sperm before and 11 weeks Widespread suppression of many key endocrine systems including luteinizing hormone (in males and females), testosterone, prolactin (chronic effect), growth hormone, estradiol and progesterone, Graafian follicle maturation, vasopressin and pregnancy including reduced fertility have been demonstrated in association with chronic cannabis use [67][68][69][70][71][72][73]. It has also been demonstrated in combined opioid-cannabinoid-dependent patients that the reversal of the FSH/LH ratio, a key clinical biomarker of the perimenopause, happened 20 years earlier [197]. Ovarian failure has also been shown to invariably be due to DNA damage [198]. Hormonal signals are rapidly transduced by the epigenome [199]. Hormonal failure and reproductive senescence represent age-defining and age-generating illnesses [1,157,158,200].

Liver Inflammation
Liver inflammation, cirrhosis and cancer have also been linked with cannabinoid exposure [133][134][135][136]. In that hepatic inflammation causes systemic inflammation, insulin resistance and dysmetabolism [201], generally these are also age-defining and age-generating illnesses. Moreover, the complex multi-way interaction between dysmetabolic and immunopathic changes is increasingly being defined and emphasized [202].

Inheritable Cancer
Several cannabis-related cancers occur in the pediatric age group and are therefore evidence of heritable carcinogenesis [204,205] and therefore combine both teratogenesis and malignancy in the one case. This has been found for acute myeloid and lymphoid leukemias and total pediatric cancer [65,66,104,105,206] and for rhabdomyosarcoma and neuroblastoma [207,208]. One recent survey of the cannabis-exposed DNA methylome showed 487 hits for various malignancies [27].

Congenital Anomalies
Clinical genotoxicity is also expressed as congenital anomalies. As a majority of congenital anomalies, particularly those affecting the heart and chromosomal systems, are known to be related to parental age [209,210] the congenital anomaly rate becomes a surrogate or biomarker for biological age. Dozens of congenital anomalies have been described following prenatal or community cannabis exposure in Hawaii, Colorado, Canada, Australia, USA and Europe affecting particularly limbs, central nervous, cardiovascular, gastrointestinal, uronephrological and chromosomal systems [66,103,[108][109][110][111]115,116,118]. Hundreds of positive hits were recorded on a DNA methylome screen for all the organ systems involved including mitochondria, chromosomes, microtubules, body axis and embryonic growth [27].

Telomerase Inhibition
Cannabis inhibits the activity of telomerase one of the key enzymes controlling aging [27,211]. Telomerase reverse transcriptase (TERT) is the key enzyme tasked with maintenance of telomere length and thus chromosomal length maintenance during cell division.
"The sections that follow integrate cannabis ageing theories from eight pathogenetic fields".

Epigenomic Overview
Longitudinal epigenomic data published by Schrott and colleagues on changes in the DNA methylome of human sperm after cannabis dependence and withdrawal [27] provide an explanation for the broad spectrum of cannabis-related teratogenicity and cancerogenicity mentioned above. Table 2 presents a re-formatted extract of the Schrott data looking at the epigenomic modulation of the key epigenomic machinery itself [27]. As shown in the Table, most of these perturbations of DNA methylation occur in introns within genes but some are in upstream presumably promoter regions and some are in downstream enhancer regions.
DNA methyltransferases 1 (DNMT1) and 3A and 3B (DNMT3A, DNMT3B) are the main enzymes which are responsible for laying down the methylation signals on DNA both from conception and in response to many signals thereafter. TET1 (ten-eleven translocase) is the main enzyme responsible for removing the methylation signals. It oxidizes the methylcytosines of CpG dinucleotides and introduces a hydroxyl group which is then oxidized in subsequent steps with the effect of removing the methylation mark. Hence the first lines of this Table show that both writing and erasing the key DNA methylation marks are disturbed by cannabis dependence or withdrawal. Here it is important to note that most habitual cannabis users go through withdrawal daily which is one of the major motivations to repeat use and making withdrawal a major and defining feature of clinical cannabis dependence [227]. UHRF1 (ubiquitin-like containing PHD and RING finger domains 1) is a key enzyme which is involved with both DNA methylation and histone modifications [228]. It recruits both DNMT1 to write DNA methylation marks and histone deacetylases which control access by the transcription machinery [228]. Its tudor-like and PHD-domains recognize and bind histone 3 trimethylated at lysine 9 (H3K9me3) and unmethylated arginine-2 (H3R2me0) and recruits chromatin proteins. Hence this enzyme is regarded as a key epigenomic hub coordinating the activities of the DNA methylation and histone regulatory systems. It regulates both the retinoblastoma gene product and the P53 damage checkpoint. Its expression levels peak in late G1 and it controls the G1/S transition of the cell cycle. It plays a key role in the regulation of pericentric chromatin and thus kinetochore function and chromosomal segregation. It is also involved in DNA repair. It is a known oncogene and has been implicated in liver cancer amongst others [229]. Hence its perturbation can be predicted to have a major effect on epigenomic regulation. DPPA3 (Developmental PluriPotency Associated protein 3) has been shown to protect the epigenome of the oocyte from methylation [230]. Whilst DPPA3 was not identified in the spermatocyte EWAS conducted by Schrott team DPPA2 was identified as indicated. TERT (telomerase reverse transcriptase) is a key enzyme responsible for maintaining the length of telomeres and is key to maintaining pluripotency in stem cells and germ cells and is often highly induced in cancer cells. Telomeres are protective caps on the ends of chromosomes and because some length is lost with each cell replication event they usually shorten with age. Since telomere attrition is one of the key chromosomal hallmarks of aging the regulation of telomere length is a key metric for the cellular aging clock. This important finding of cannabinoid interference with this key cellular enzyme has also been reported by others [211].
The polycomb repressive complex ( Table 2) is one of the main epigenomic complexes which silence heterochromatin long term. Therefore, interference with these activities can be expected to have long-term consequences for cellular health.
SMARCA2 and SMARCA4 are SWI/SNF (SWItch/Sucrose NonFermentable) ATP-dependent modifiers of chromatin which change nucleosome position in an energydependent manner and therefore rearrange the genome and make new sections available for transcription. Modulation of these epigenomic controllers was recently shown to have a very positive effect in advanced castrate resistant prostate cancer which was addicted to their activities [231]. Since the SWI/SNF system is a major rearranger of chromatin perturbation of this system carries major downstream implications for cellular health. SMARCA2 and SMARCA4 (also known as Brahma, BRM and Brahma-related Gene 1, BRG1) were also recently determined to be key determinants of differentiation and canalization of precursor mesodermal cells into a cardiac fate [11].
Not only is DNA methylated but so too are histone proteins. There were 161 hits in the Schrott database for histone methyltransferases which write this mark onto the histone code (some top hits shown in Supplementary Table S1) and 199 hits for the histone demethylases which remove this mark (of which an extract is shown in Supplementary Table S2).
Histone acetylation is a key mark on histone tails. By neutralizing the charge of histone tails histone acetylation opens up chromatin and makes it available for gene transcription. This key acetylation mark is written onto the histone code by histone acetyl transferases and removed by histone deacetylases. Eleven hits in the Schrott data for each of these which were noted in both cannabis dependence and withdrawal are detailed in Supplementary  Tables S3 and S4 respectively. Thus, this brief introductory overview provides good evidence of major changes not only of the DNA methylome but of the central machinery which writes and erases and coordinates the epigenetic code on both DNA and histones. Key chromosomal areas such as the telomeres and centromeres are also impacted which thereby directly impacts processes such as aging (via accelerated telomere loss) and cellular division (via disruptions of centromere/kinetochore function).

Stem-Cell Factors
Takahashi and Yamanaka published their seminal and ground-breaking paper on the use of four defined recombinant stem-cell factors to maintain and induce the pluripotential state of embryonic stem cells in 2006 [12]. Proof of the induced stem-cell concept was provided by their demonstration that they were able to revert mouse fibroblasts to embryonic stem cells by the use of their four defined factors OSKM. These induced embryonic stem (iPS) cells went on to contribute to viable mouse embryos after injection into blastocysts. Intermittent use of the OSKM factors, a technique known as partial reprogramming, was able both to rescue a mouse model of progeria and to dramatically accelerate injury recovery to skeletal muscle and pancreatic islets in aged mice [13] and was able to improve cardiac function after myocardial infarction in a mouse model [232]. Inducible expression of OSK in retinal ganglion cells was able to restore vision in a manner only seen in neonatal mouse pups after glaucoma, optic nerve crush and extreme age in old mice [15].
As shown in Table 3, there were 11 hits in the Schrott database for the Yamanaka stem-cell factors. The name of the Oct3/4 gene has since been changed to POU5F1. SOX2, KLF4 and MYC were positively identified but Nanog was not identified. A modification of the Yamanaka protocol using slightly different stem-cell factors where Klf4 was replaced by Lin28 was also shown to induce iPS induction [14]. Stem-cell factors used by these researchers and also by Yamanaka were further investigated in the Schrott data with results shown in Supplementary Table S5. As there were 146 hits for Ras, 230 hits for Catenin and 185 hits for Kit in this database only a leading selection is shown in the Table. Hits for PAX7, one of the skeletal muscle master transcription factors and Lin28 are also shown at the bottom of the Table. Supplementary Table S6 provides an expanded list of some of the hits for Kit.
There is also a powerful and well documented multi-way link between immune activation, dysmetabolic changes and the aging process. For example, a recent study showed that much of the effect of calorie restriction, which has been well demonstrated to induce life extension in flies, worms and mice, when applied in humans was mediated by PLA2G7 (platelet activating factor acetyl hydrolase/phospholipase A2 group VII) [202]. PLA2G7 is found in cholesterol-rich low density lipoprotein particles and PLA2G7 oxidizes saturated lipids and activates vessel wall macrophages, lymphocytes and endothelial cells. It thereby stands at the intersection of immunity and metabolic processes.
A research group from Stanford developed a biological clock based on immune biomarkers and found that CXCR9 was the key chemokine which accounted for most of the variance they identified [83]. A sizeable literature exists around NAD (nicotinamide adenine dinucleotide) metabolism and the links between its normal dramatic age-dependent decline and the ageing process itself [233][234][235][236][237][238][239][240][241]. The key rate limiting enzyme in the NAD biosynthetic pathway is nicotinamide phosphoribosyl transferase (NAMPT) which acts as the gateway to this pathway [148]. It was therefore of interest to learn if these key immune and metabolic mediators were identified in the Schrott EWAS. The results of this investigation are shown in Supplementary Table S7. Both PLA2G7 and NAMPT were positively identified. CXCR9 was not identified but CXCR13 was found.

Chromosomal Mechanics
During the process of cell division at the beginning of prometaphase, the nuclear membrane breaks down and what has very properly been called the "mammoth" supramolecular mitotic and meiotic machine involving the mitotic spindle begins to form [242]. The process takes place on the large scale of the whole cell cytoplasm and each of its innumerable steps are tightly regulated, carefully choreographed and finely coordinated by elegant and sophisticated mechanisms. The implication of this vastness, complexity and sophistication is that the delicate process of cell division is open to perturbation and disruption at numerous steps.
Whilst the process of cell division is well known to students of biology the world over from watching time lapsed video micrographs, it is less well known that in the human oocyte the process is highly error prone with error rates of 60-90% being reported even when young oocytes are used [243][244][245][246][247][248][249] and this error rate is known to rise sharply with age [243][244][245][246]248,249]. The bipolar alignment of the mitotic spindle with two spindle poles is critical to directing the cell to divide into two daughter cells during the subsequent anaphase separation. Whilst most species have a pair of centrioles and pericentriolar material (called centrosomes) which direct this process this is absent from higher (non-rodent) mammalian species including humans. Such species organize their spindle poles using acentriolar microtubule organizing centers (aMTOC) organized by NUMA (nuclear mitotic apparatus protein) and the kinesin motor protein KIFCI (kinesin family member C1) to draw the microtubules together [248]. Supplementation of human oocytes with KIFC1 largely rescued the high mitotic error rate [248] and in mice its knockdown via degron mediated destruction increased the error rate of bovine and modified aMTOC-free mouse oocytes to be highly similar to that of the human oocyte [248]. In actual fact, the number of poles in human oocytes mitotic spindles oscillates dynamically during oocyte maturation over several hours from several poles to just one pole and most frequently settles at just two spindle poles [245]. This implies that NUMA and KIFC1 are key to the integrity and reliability of the inherently error-prone oogenesis mitotic process in humans [248].
The anaphase-promoting complex/cyclosome (APC/C) is known to be a key organizer of the mitotic spindle and to determine when all the paired chromosomes are aligned satisfactorily on the metaphase plate and thus licences and controls the chromosomal separation of anaphase [249,251]. In human-derived HEK293 cell lines it was shown that APC/C also localizes to the centrosome where its activity is controlled by Cep152 (Centrosomal Protein 152) in complex with Cep 57 and Cep 63 [249].
Tubulin is also subject to numerous post-translational modifications particularly acetylation, polyglutaminylation and tyrosinylation [252]. Acetylation is key to the formation of the tubulin polymers of the mitotic spindle and this is controlled by lysine (K) acetyltransferase and histone deacetylases (HDAC) particularly HDAC3, HDAC6 and HDAC11 and the sirtuin (SIRT) HDAC's SIRT2 in meiosis I and SIRT1 in meiosis II [245]. The process is also sensitive to oxidative stress and ROS (reactive oxygen species) are known to play important roles in both folliculogenesis and oocyte maturation but excessive ROS levels have been linked to shrinkage of the width and length of the mitotic spindle, disruption of the spindle asters, chromosomal misalignment in metaphase II, chromosomal disassembly in meiosis I and II and increased aneuploidy rates [245]. Adducts of ROS including 4-hydroxynonenal form and co-localize with α-, βand γ-tubulins [245]. Ovarian ROS production also rises with age [245].
Importantly, acetylation of lysine-40 on polymerized α-tubulin by α-tubulin acetyl transferase 1 (ATAT1) occurs on the inner surface of the microtubule and allows for running repairs to be undertaken on the polymer when the microtubules is stressed or bent thereby adding greatly to the structural strength and flexibility of the structure [253]. Without K-40 alpha-tubulin acetylation, the microtubules remain brittle and bending leads to microtubule fracture and chromosomal derailment, isolation, aneuploidy and micronucleus development during the anaphase disjunction. Unlike female meiosis, cell division in the fertilized zygote is organized around centriole-containing centrosomes which are derived from the paternal gamete as those associated with the female pronucleus are rudimentary [243,244]. It is therefore clear that interference with any of these structural, binding, signaling or motor proteins will lead to an elevated error rate of human female gametogenesis [248].
Supplementary Table S8 therefore presents the hits identified in the Schrott database for NUMA, CEP and kinesin-and dynein-dynactin motor proteins. It is noted that KIF14 is an alternate nomenclature for KIFC3 which was noted to be critical [248]. Hits in intron, exon and enhancer regions are noted. There were 218 hits for kinesin motors and these hits were some of the strongest hits identified in cannabis dependency in both Schrott's Tables S1 and S4 [27]. Some of the top-scoring kinesin motor protein hits are detailed in Supplementary Table S9. It is noted that these results for the DNA methylome come from sperm so it remains to be determined how the detailed results from oocytes might compare.
When one considers tubulins in the database of Schrott and colleagues, 106 hits are obtained. Some of those for tubulin (not including the pseudogenes) and ATAT1 are shown in Supplementary Table S10. This Table also shows epigenomic hits identified for some of the key enzymes which write and modify the tubulin code including acetylation, tyrosinylation/detyrosinylation and acetylation. In total, 86 of the hits observed for tubulin are for TUBB6 (β-tubulin 6 class V) and these appear as the most significant of all of the functional annotations in the Schrott Table S4 for cannabis dependence as partially extracted in Supplementary Table S11. TUBB6 epimutations are also linked with many cancers [27].

Centromeres and Kinetochores
In addition to the poles, organization and microtubular rays of the mitotic and meiotic spindles the points of attachment of the chromosomes to the microtubules also form a key locus of control for the whole mitotic process and a key point of vulnerability at which xenotoxins may impact. Somewhat confusingly the combination of the central repetitive non-coding DNA at the center of the chromosome (the centromere) together with its accompanying histones and proteins is (also) referred to as the centrosome. The key marker for the development of the centromere is the substitution of histone 3 (H3) for its derivative CENPA (Centrosomal protein A) and the formation of neocentromeres can be induced by the forced expression of CENPA along chromosomal arms [254]. A multiprotein complex of 16 other centrosomal proteins called the kinetochore is then assembled on the centrosome at CENPA to form a large multimolecular complex which binds to the growing plus ends of 25-30 microtubules for each chromosome.
Detailed descriptions of the protein composition of the kinetochore have appeared [254][255][256]. When these proteins are run through the Schrott database 109 hits are obtained for the 19 proteins listed in Table 4. Interestingly, 86 of these hits are for CENPN which is the equal second protein to assemble alongside CENPA at the very commencement of kinetochore assembly. Some of the most significant hits for CENPN are shown in Supplementary Table S12 and are extracted from the Table S4 in Schrott's dataset for cannabis dependence. They are notable for their very high levels of statistical significance along with their association with uniformly malignant disorders. With the exception of SPC24, all the hits identified were in cannabis dependence rather than cannabis withdrawal.  Table 4 also includes details on the addition of the Small Ubiquitin-like MOdifier (SUMO) protein to histones. Sumoylation is a key post-translational modification (PTM) of many proteins which has been shown to be critically involved in many key genomic functions such as DSB repair, DNA transcription and replication and chromosomal segregation and synapsis [257,258]. Sumoylation is a foundational post-translational modification on many proteins including RNA polymerase II which forms the basis for the addition of sometimes lengthy chains of PTM's which control these key genomic activities [258]. ∆9THC acting via CB1Rs has been shown to directly modulate P53 (the "guardian of the genome") and Mdm2 (murine double minute, one of its key controlling proteins) [259]. As documented in the lower segment of Table 4, it was demonstrated in the Schrott EWAS that SUMO1 itself, one of the key E3 SUMO ligases which attaches the PTM to proteins, ZNF451 (zinc finger 451) and two of the SUMO endopeptidase proteins (SENP6 and SENP7) which cleave the SUMO PTM's are affected epigenomically by cannabis dependence and withdrawal.
Since centromeres form the site of attachment of the chromosomes to the mitotic spindle, it follows that centromeric stability is key to maintenance of genomic stability [2]. In fact, centromeres are intrinsically "stiffer" and more fragile than the rest of the chromosome and represent "hot spots" for double stranded break (DSB) occurrence and chromosomal rearrangements [2]. Accurate repair of these breaks by homologous recombination is therefore essential to genome stability. Homologous recombination is normally understood to be suppressed in the G1 phase of the cell cycle. However, it has recently been reported that CENPA together with its chaperone HJURP (Holliday Junction Recognition Protein) and dimethylation of H3 (H3K4me2) permit invasion of the double stranded DNA by the DNA-RNA hybrids (R-loops) and licences the assembly of the RAD51 (RAD51 Recombinase)-BRCA1 (BRCA1 DNA Repair Associated 1)-BRCA2 complex which is the core complex of the main high fidelity homologous recombination (HR) pathway. Inhibition of HR necessarily leads to activation of much lower fidelity pathways such as microhomology-mediated end joining mediated by RAD52 and compromises genomic stability [2]. These investigators were able to demonstrate that RAD51 inhibition greatly increased centromeric breaks and centromeric translocations in NIH3T3 cells (as immortalized embryonic fibroblast cell line). Inhibition of both RAD51 and RAD52 together, inhibited both major repair pathways and blocked the formation of chromosomal translocations [2].

Prefrontal Cortex and Brain
It is of interest to consider the representation in the Schrott EWAS of some of the key genes and pathways which are believed to be central to brain development. DSCAM (Down syndrome cell adhesion molecule) is most highly expressed in the fetal brain and retina where it is involved in neuronal self-avoidance, axon growth cone guidance, amacrine and retinal ganglion cell dendrite arborisation, commissural midline crossing in the spinal cord, homophilic synapse development and congenital heart disease [263,264]. It is overexpressed in Down syndrome and this has been implicated in some of the development of intellectual impairment in that disorder [264]. Supplementary Table S14 sets out the 14 EWAS hits in the Schrott database for DSCAM. DLGAP2 (DLG associated protein) is an autism associated candidate gene also implicated in schizophrenia which has previously been linked with paternal cannabis exposure in sperm EWAS Studies [27]. It was thus of interest to see if the present study confirmed these earlier results. Supplementary Table S15 shows that indeed these results were strongly confirmed by the present EWAS series.
It was shown in the last decade that one of the main reasons for the relatively very enlarged frontal lobes of the human brain is the increased activity of Robo (Roundabout) signaling in the frontal cortex which leads to a greatly expanded neurogenesis in the frontal lobes and hyperproliferation of dedicated neural progenitor cells which feed into the exuberant frontal lobar growth [265][266][267]. Slits 1-3 form the natural ligand for robo receptors. The system is involved in both nervous system development and patterning and axonal guidance and also in arterial pathfinding and steering [209]. It has also been shown that this activity is blocked by cannabinoids [268]. It was therefore fascinating to observe that SRGAP2C (SLIT-ROBO Rho GTPase Activating Protein 2C) was identified by genomic screens and comparative genetics across many species to be the gene responsible for the exuberant outgrowth of the human forebrain neocortex [269]. Indeed, inducible expression of the forebrain of mice increased the cortical neuronal density and the synaptic short and long range corticocortical and bidirectional thalamocortical connectivity of layer 2/3 pyramidal cortical cells, enhancing their computational power and the rodents' ability to quickly learn complex sensory-discriminant tasks [269].
For these reasons, it was of interest to observe how this system performed in the Schrott EWAS. Supplementary Table S16 sets out five results for Slits, Supplementary Table S17 sets out 26 results for Robo and Supplementary Table S18 sets out the eight results for SRGAP2C and its natural antagonist and controller SRGAP2B.
Another system which has also been shown to induce the relative overgrowth of the enlarged human forebrain is retinoic acid (RA). It was recently shown that high concentrations of RA at the frontal pole decline to lower and more normal levels at the posterior of the prefrontal neocortex in the premotor cortex [270]. The enzyme at the anterior pole which is chiefly responsible for synthesizing the high levels of RA is ALDH1A1 (aldehyde dehydrogenase 1 family member 1), the RA signal is transduced by the retinoid receptors RXRG and RARB, and RA is catabolized near the premotor cortex by CYP26B1 which is part of the cytochrome P450 system [270].
It was thus of interest to examine how these systems were affected in the Schrott EWAS. Supplementary Table S19 lists 11 hits for ALDH1 including two hits for ALDH1A1 and cadherin and protocadherin (PCDH17) which also function in this pathway. Indeed there were 156 hits for protocadherin 17 from the very lowest p-vale of 7.73 × 10 −20 [27]. The nine hits for retinoid receptors are disclosed in Supplementary Table S20. Although CYP26B was not identified in the Schrott screen there were twelve hits for CYP2 series cytochromes including CYP20A1, CYP27A1, CYP27C1 and CYP27C2; and CYP2B7P, CYP2C1, CYP2C18, CYP2C61P and CYP2W1.

Cardiovascular System
Aging of the cardiovascular system is known to be a critical determinant and driver of systemic aging [271][272][273][274][275][276]. Indeed, it is said that one is as "old as one's arteries" [157,158,[277][278][279]. This is true at both the macrovascular level, with myocardial infarction being a major cause of death in developed nations, and at the microvascular levels where capillaries and sinusoids often form critical elements of many stem-cell niches [157,278,279]. Moreover, a two-way crosstalk has recently been defined between major cardiovascular disorders (myocardial infarction, hypertension and atherosclerosis) and the bone marrow haemopoietic stem-cell niche where endothelial inflammation in one compartment directly signals to the stemcell compartment of the other system [280,281]. For these reasons, consideration of the epigenomic findings in the Schrott cannabis exposure and withdrawal data of relevance to arterial health are central to any consideration of cannabinoid-related aging processes. A detailed consideration of the cardiovascular hits in the Schrott study is deferred until the later section on teratology (see Supplementary Table S25).
It is of interest to consider the genomic processes controlling arterial health. The key genes involved in generating arteries from embryonic angioblasts are listed as sonic hedgehog (shh), vascular endothelial growth factor (VEGF), notch and ephrin B2 [209]. These genes and pathways were therefore screened through the Schrott dataset and the hits identified in Table 5A,B were identified. PTCH1 is the main shh receptor. Gli3 (GLI family zinc finger 3) is one of the key transcription factors which mediates shh signaling in the nucleus [282]. Gli3 scored 185 hits in the Schrott EWAS data of which only a selection has been extracted for illustration. PSENEN (Presenilin enhancer, gamma secretase subunit) is a key plasmalemma bound enzyme which processes the shh ligand after receptor binding. SUFU (SUFU negative regulator of hedgehog signaling) inhibits shh [283].
Supplementary Table S21A,B list genes involved in the notch signaling pathway identified in the Schrott screen. JAG1 is a canonical notch ligand. Notch 1-3 are notch receptors. RBPJ (Recombination Signal Binding Protein for Immunoglobulin Kappa J Region) is an important transcriptional regulator of notch signaling. PSENEN also processes the notch ligand at the cell membrane [284].
Supplementary Table S22A,B list the six hits in the Schrott database relating to VEGF and EphrinB2 signaling. Both VEGF and EphrinB2 are key signaling and transduction factors involved in mediating numerous major morphogenic decisions and pathways [209,251].
In this regard, fascinating recent detailed studies have appeared on the profound impact of prenatal cannabinoid (as ∆9THC) exposure on cardiac development. Robinson and colleagues showed that prenatal exposure to ∆9THC led to cardiac wall thickening in three week old mice and thickening and hypertrophy of the semilunar valves and increased ventricular septal defects [285]. Myocardial cell proliferation was increased and cardiac function was reduced with lower ejection fraction, fractional shortening and cardiac output.
Lee and co-workers demonstrated rat fetal growth restriction following in utero exposure to ∆9THC, smaller hearts and reduced a heart to body weight ratio at birth [286]. By three weeks of post-natal life this has been reversed by post-natal catchup growth which resulted in larger but stiffer ventricular wall thickness and a corresponding reduction in cardiac output. This was linked with increased expression of collagens I and III, reduced matrix metalloproteinase 2 and increased glycogen synthase kinase 3β signaling all of which are linked with cardiac remodeling. This study is highly significant as it relates the smaller hearts at birth to subsequent cardiac stiffness and reduced cardiac output, all of which are age related changes [277]. These changes in early postnatal life are known to be causally related to increased incidence of adult heart disease in later life which is the leading cause of death globally [285][286][287].
Many congenital anomalies and cancers in USA and European epidemiological datasets have been shown to be heightened after cannabis exposure. The following sections on these cannabinoid-related teratogenic and carcinogenic findings are respectively reviewed using the epigenomic data on changes in the DNA methylome of human sperm after cannabis exposure and withdrawal with a focus on genotoxicity and/or epigenotoxicity.

Cannabinoid-Related Teratogenesis
The consistent association between congenital anomalies and cannabis exposure provides functional examples of how cannabis ageing mechanisms contribute to intergenerational disability. Table 6 directly compares the congenital anomalies which were found to be cannabis-associated in USA [103] with those identified in recent reports in the larger European dataset [115]. In total, 45/62 congenital anomalies were found to be cannabis-associated in the US dataset compared to 89/95 in the larger European dataset [103,115]. These concerning findings are noted to be highly concordant with those of other investigators in recent large population-based series [66,[107][108][109][110][111]116,118,288,289]. These data are presented to introduce and contextualize the system-based narrative discussion undertaken in the following sections.      The p-values which relate to these various anomalies may be extracted from the Schrott EWAS database as indicated in Supplementary Table S23. This Table provides a list of  245 systems, targets and annotations ordered by their system for all of the above EWAS hits. The above table demonstrates cross-nationally consistent associations between cannabis exposure and varied congenital abnormalities. The sections that follow evaluates evidence associating cannabis exposure with epigenomic mechanisms for congenital abnormalities.
Supplementary Tables S24-S32 present the systems-based interrogation of the Schrott database for the cardiovascular, central nervous, face, general, limb, gastrointestinal, chromosomal, uronephrological and body wall systems respectively. Examination of these Supplementary Tables demonstrates that they offer profound insights into the possible pathogenesis of the congenital anomalies described in Table 6.
Supplementary Table S24 describes 73 central nervous system EWAS hits and lists features such as brain size, brain formation, forebrain patterning, development of many kinds of synapses, head development, head size, movement and viability of cerebral cortex cells, neurite growth, neuronal growth, neuronogenesis and neuronal outgrowth and proliferation, brain cell migration, axonogenesis and outgrowth which would be consistent not only with defects such as brain growth and size (microcephalus and anencephalus) but also defects of brain function such as epileptiform disorders, autism [117,174,288,290,291], intellectual disability (mental retardation) and many mental illnesses in childhood and later life [290][291][292][293][294][295][296][297][298][299][300]. Many disorders of eye development are also noted which is consistent with the finding of microphthalmia in both the USA and European series. Many disorders of inner ear development are noted consistent with the findings of microtia and anotia in the USA and European datasets. Associations are also reported with some malignant brain conditions which is consistent with earlier reports [59].
Supplementary Table S25 shows the 29 EWAS hits which are linked with the 23 cardiovascular anomalies in Europe and the eleven cardiovascular anomalies in USA. Supplementary Table S25 and reported for both left and right ventricles in the congenital anomaly (CA) list of Table 6. Septal defects are reported in the EWAS list and in the CA list for both atria and ventricles. Anomalies of the atrioventricular valves/endocardial cushions are mentioned in the EWAS hit list and mitral and tricuspid valvular anomalies including Fallot's teratology are mentioned in the CA teratological list. Many defects of vasculogenesis, angiogenesis, pulmonary venogenesis and vascular breakdown are mentioned on the EWAS list and the cardiovascular anomalies of transposition of the great arteries, total anomalous pulmonary venous return, vascular disruptions, VACTERL (vertebral, anal, cardiac, tracheoesophageal atresia, renal and limb) syndrome, aortic arch anomalies, coarctation of the aorta, severe cardiac congenital anomalies, double outlet right ventricle, tetralogy of Fallot and others were identified on the CA list.

Hypoplasia of the cardiac chambers is mentioned both in
Supplementary Table S26 lists 22 EWAS hits of interest for facial development. Development of the face has been shown to impact brain development embryologically as the organizers for both regions interact during gestation and both are controlled by strong anterior gradients of sonic hedgehog and retinoic acid [209]. Supplementary Table S26 lists anomalies of the head, palate, nose, lens, iris and ear which relate to listed CAs of microcephaly, cleft lip and palate (which may involve the nasolabial groove), congenital cataract (in both USA and Europe) and anotia/microtia (in both USA and Europe). Importantly the severe CA holoprosencephaly which is strongly associated with abnormal brain development was identified as a strong association of cannabis teratogenesis in Europe and a weak association in USA [103,115].
Supplementary Table S27 lists 60 hits from the Schrott EWAS dataset relating to "general" issues which do not readily classify under other systems. In total, 36 (60%) of these hits relate to cannabis dependence and 24 (40%) to cannabis withdrawal. The EWAS list provides fascinating and powerful insights to the observed teratological profile documented in Table 6. Defects of cell growth, embryonic growth, organismal growth and embryonic morphogenesis head up the Table. Defects of most major DNA activities are comprehended including synthesis, binding, recombination, transcription, translation, repair, recombination, replication, and synapsis (crossing over) are shown. Defective RNA translation is indicated. Defects of chromosomal synapsis, homologous pairing, assembly and synapsis are shown.
Mitochondrial defects are listed. This is important as mitochondria supply both the energy for genomic and epigenomic reactions and the underlying substrates for the epigenomic machinery. Two hits for microtubular impairment are shown, one each in cannabis dependence and withdrawal. This may relate to anomalous chromosomal missegregation disorders for chromosomal trisomies and monosomies affecting chromosomes 13, 18, 21 and X (Supplementary Table S27). Reproductive defects are indicated with diminished ovarian reserve-a hallmark of ovarian ageing-and three hits for breast cells which potentially relate to recently reported elevated rates of breast malignancy in USA in relation to cannabis consumption [66,[112][113][114]121].
Anomalies of body trunk and body axis development are shown. In total, 22 anomalies of bone development are listed consistent with very elevated rates of VACTERL syndrome reported from Europe.
Supplementary Table S28 reports six hits for limb anomaly development consistent with major limb anomalies including limb reductions reported from both Europe and USA. These studies may be extended further as indicated in Table 7. It is known that morphogens such as retinoic acid, fibroblast growth factors (FGFs) and Wnts play pivotal roles in the three dimension temporally sequenced complex choreography of limb development [209]. Genes such as Meis1/2 (Meis homeobox), FGF4, RXRA (retinoid X receptor) and RARB (Retinoic Acid Receptor B), TBX4/5 (T-box transcription factor), Wnt's, shh, GREM1/2 (Gremlin), CHD7 (Chromodomain Helicase DNA binding protein 7), TMEM107 (Transmembrane Protein 107), MEGF8 (Multiple EGF-like domains 8), BMP4, and GLI3 play key roles [27,209,301].     Key: The first entry in each type of gene is in bold. This signifies the gene class. Its initial entry signifies the number of entries for that gene in the data set.
Some of the hits from the Schrott EWAS data are extracted and illustrated in Table 7. Numbers shown in bold on the right-hand side of the second column on the right are the total hits for that gene. The other numbers listed in the "Numbers of genes column" are the numbers of genes identified with the particular DNA methylation pattern identified and listed in the Schrott dataset. Hence Meis1 had 37 hits in the EWAS, Meis2 97 hits, FGFs 175 hits, FGF4 7 hits, RXR/RARs 10 hits, CHD7 124 hits, MEGF8 105 hits, TMEM107 232 hits, BMP4 166 hits and Gli3 183 hits. Together, this accounts for 1129 hits in these major morphogens and gene pathways which is a very substantial number of perturbations compromising limb morphogenesis.
In total, 37 gastrointestinal EWAS hits are listed in Supplementary Table S29 which relate to the many gastrointestinal congenital anomalies reported in Table 6 which affect most of the major gastrointestinal organs. 27/37 (73%) relate to cannabis dependence and 10 (27%) are in withdrawal. Supplementary Table S29 also lists most of the gastrointestinal organs. Cancer and carcinoma are prominently identified.
Supplementary Table S30 lists four Schrott EWAS hits for chromosomal disorders. Given that trisomies 13, 18 and 21, Turners, Klinefelters and genomic deletions along with all chromosomal disorders are all listed in Table 6 this is highly important. As discussed in earlier sections on the underlying subcellular pathoaetiology, it is not clear if these chromosomal disorders relate to epigenomic, microtubular, kinetochore, centrosome or related problems or possibly some combination of these aberrations.
The eight identified EWAS hits for renal disorders are shown in Supplementary  Table S31. These clearly cover most aspects of uronephrological development. These relate to the many uronephrological CAs identified in Table 6 including overall urinary anomalies, multicystic renal disease, obstructive genitourinary disorder, congenital posterior urethral valve, renal agenesis, bladder extrophy and hydronephrosis. Importantly, renal agenesis was a strong association of cannabis teratogenesis in both USA and Europe. This fits with the above pathophysiological narrative as sonic hedgehog and retinoic acid are major morphogens in renal and urinary development [209].
Supplementary Table S32 lists 15 EWAS hits for body wall development. In total, 7/15 (46.7%) are in cannabis dependence and 8 (53.3%) are in cannabis withdrawal. Body trunk and body axis development are prominent as is development of the abdomen. Growth and differentiation of embryonic tissues is clearly predominant in the lower part of the Table. These various Tables may be combined by body system as shown in Supplementary  Table S33. This Table does not include the extended studies listed above for congenital limb anomalies. Supplementary Figure S1 presents the summary of the p-values as the negative log of the p-value as boxplots. Non-overlapping notches indicate statistically significant differences. Gastrointestinal, chromosomal and neurological defects appear towards the right end of the graph. Table 8 provides the mean and median p-value for each system. A significantly rising trend by body system is noted (β-est. = 1.21, Student's t = 7.65, p = 4.69 × 10 −13 ; Adj R Squ. = 0.1908, F = 58.53, df = 1, 243, p = 4.69 × 10 −13 ). If one considers 39 of the (arguably) most significant target organs of interest the results for mean and median p-value shown in Supplementary Table S34 are revealed which are plotted graphically in Supplementary Figure S2. Gastrointestinal, liver, brain, atrioventricular valves, head and chromosomes appear towards the right-hand side of this Figure as most severely affected. Again, the trend across this graph is highly statistically significant (β-est. = 0.31, Student's t = 9.23, p = 6.82 × 10 −18 ; Adj R Squ. = 0.2565, F = 85.16, df = 1, 243, p = 6.82 × 10 −18 ).
These data may be summarized by target organ as shown in Table 9. The number of annotations cited in the Schrott EWAS data by target for cannabis dependence and withdrawal is shown in Figure 1. The gene numbers identified in each condition by target are shown in Supplementary Figure S4.

Cannabinoid-Related Carcinogenesis
The consistent association between varied cancers and cannabis exposure provides further examples of how cannabis ageing mechanisms contribute to disease. Table 10 sets out the most significant associations of various cancers with cannabis or cannabinoids in USA and Europe [112][113][114]121]. The Table lists the minimum p-value, the model type and the primary correlate of the various cancers listed. Two of the main features of this Table  are the number of cancers listed and the commonality between the USA and European experience which are the two largest datasets on this issue available internationally.  Table S36 summarizes the  data of the preceding Table for minimum, mean and median significance levels by tumor type and is ordered by descending minimum p-value. The cumulative gene number includes duplicate mentions for some genes. Thyroid, melanoma and urinary cancers head this list. When the list is ordered by median p-value thyroid, testis, stomach, liver and oropharyngeal tumors head the list (Supplementary Table S37). Some of these key data are shown in Supplementary Figure S5 which lists the number of annotations, the cumulative gene number and the negative log of the p-value for each tumor type.
Because the Schrott dataset is elegantly organized into both cannabis dependence and withdrawal it may be categorized for 19 tumors in cannabis dependence as shown in Supplementary Table S38, which is listed in descending order of minimum p-value. The cumulative gene number again includes duplicate mentions for some genes. This list is headed by thyroid, melanoma and urinary cancers. When the same list is ordered by median p-value the order of significance is thyroid, melanoma, stomach, colorectal urinary and testis cancer as indicated in Supplementary Table S39. Some of these key data are illustrated graphically in Figure 3 which lists the number of cancers, the cumulative gene number from the Schrott EWAS dataset, and the negative log of the p-value for each tumor type.
Supplementary Table S40 lists the applicable p-values for cannabis withdrawal for 18 tumor types and is ordered by minimum p-value. The list is headed by melanoma, brain, oropharynx and esophageal cancers. These significance levels are noted to be lower than those in the preceding Tables. When the list is sorted by median p-value oropharynx, melanoma, brain, urinary, acute myeloid leukemia and testicular cancer head the list (Supplementary Table S41). These results are illustrated graphically in Figure 4 which shows, respectively, the number of gene annotations, the cumulative gene number and the negative log of the significance levels by tumor type.  Supplementary Table S40 lists the applicable p-values for cannabis withdrawal for 18 tumor types and is ordered by minimum p-value. The list is headed by melanoma, brain, oropharynx and esophageal cancers. These significance levels are noted to be lower than those in the preceding Tables. When the list is sorted by median p-value oropharynx, melanoma, brain, urinary, acute myeloid leukemia and testicular cancer head the list (Supplementary Table S41). These results are illustrated graphically in Figure 4 which shows, respectively, the number of gene annotations, the cumulative gene number and the negative log of the significance levels by tumor type.  Supplementary Figure S6 directly compares the significance levels of the tumors by cannabis dependency status. It is observed that the tumors are in a very different order and that the level of significance is generally much lower in cannabis withdrawal than in cannabis dependence. Table 11 directly compares the significance levels and gene numbers for the various tumors types in dependence and withdrawal. Whilst the overall pattern is clearly that there are more genes implicated and at higher levels of statistical significance by cannabis dependence than cannabis withdrawal, there are a few notable exceptions to this pattern. Both acute myeloid leukemia (AML) and acute lymphoid leukemia (ALL) have a lower gene number dependence/withdrawal ratio than unity. AML also has a lower minimum (and median and mean) p-value dependence/withdrawal ratio. Data are listed by the gene number ratio in Supplementary Table S42 and ovarian, Non-Hodgkins, pancreas thyroid and testicular cancers are noted to head up the list. Some of these data are shown graphically in Figure 5 which lists the log of the ratio of the minimum p-values, the log of the gene number for dependence/withdrawal and the log the gene number for the withdrawal/dependence ratio. In this way, the distinctly higher withdrawal/dependence ratios in the pediatric AML and ALL cancers are highlighted. Figure 5. Log plots of significance levels for (A) ratio of p-values between cannabis dependence and withdrawal, (B) log of the dependence/withdrawal ratio of total gene numbers affected between cannabis dependence and withdrawal and (C) log of the withdrawal/dependence ratio of total gene numbers affected between cannabis dependence and withdrawal; each by tumor type from the Schrott EWAS data.

Implications of Findings
From such a very broad array of objective reported results, basic cellular mechanisms and highly concordant epidemiological findings in both addiction medicine and aging science, it is necessary in discussing these results to highlight just a few key findings which are of particular importance to the overall flow of the main themes of this review and the major concepts presented. More detailed discussions have been presented in the references cited and other exhaustive and encyclopaedic sources [302][303][304][305][306][307]. The study is the Figure 5. Log plots of significance levels for (A) ratio of p-values between cannabis dependence and withdrawal, (B) log of the dependence/withdrawal ratio of total gene numbers affected between cannabis dependence and withdrawal and (C) log of the withdrawal/dependence ratio of total gene numbers affected between cannabis dependence and withdrawal; each by tumor type from the Schrott EWAS data.

Implications of Findings
From such a very broad array of objective reported results, basic cellular mechanisms and highly concordant epidemiological findings in both addiction medicine and aging science, it is necessary in discussing these results to highlight just a few key findings which are of particular importance to the overall flow of the main themes of this review and the major concepts presented. More detailed discussions have been presented in the references cited and other exhaustive and encyclopaedic sources [302][303][304][305][306][307]. The study is the first to combine and connect data from a broad range of genotoxic areas. Perhaps the most striking finding is the extraordinarily accurate predictive power of the epigenomic results to apparently explain the epidemiologically observed mutagenic and teratological phenomenology. This accuracy provides confirmation of the validity of the cannabis ageing mechanisms outlined in this paper. The epigenomic results of the Schrott group [27] not only predict with great accuracy such disparate findings as the high rates of atrial septal defect widely observed in Canada, Australia, Colorado, Hawaii, USA and Europe [103,107,108,111,[114][115][116]118] and elevated rates of ventricular septal defect noted by the American Academy of Pediatrics and the American Heart Association and elsewhere [103,109,111,115,308], but also the mechanistically closely related pattern of congenital cardiac and renal anomalies which both share critical sensitivity to inhibition of the notch, sonic hedgehog and retinoic acid morphogenic pathways. Both atrial septal and ventricular septal defects feature prominently in the spectrum of cannabis teratological defects, and also in the multisyndromic VACTERL syndrome which formally relates renal, cardiac and limb anomalies (amongst others) and was the most strongly significantly cannabis-associated of all the European birth defects studied [114].
The likely foundational importance of cannabis-induced epigenotoxicity implies that not only has the salience of epigenomic disturbances emerged as being pre-eminent from the mechanistic confusion surrounding the aging process itself [16] but in a similar way it appears that with time and further research the epigenomic perturbations induced by cannabis dependence and withdrawal are likely to be shown to be foundational in understanding the plethoric and protean manifestations of cannabinoid-induced mutagenesis, teratogenesis, carcinogenesis and indeed aging [2].
It was noted that the ratios of the most significant p-values were inverted for the pediatric tumor ALL, and for AML of which some cases occur early in life. This suggests the intriguing possibility that it is the cannabis withdrawal state following birth which triggers and launches the leukemogenic gene cassettes of childhood.
Many other features of the above series of results stand out prominently. The high numbers and wide ranges of both congenital anomalies-45/62 in USA and 89/95 in Europe (Table 6)-and cancers-33/40 in Europe and 25/28 in USA (Table 10)-are striking both in their own right and by virtue of the range of tissues and organ systems affected. As these observations have been made previously [66,103,[112][113][114][115][116][120][121][122], they do not form the primary focus of the present investigation. What is more important for the present discussion of cannabis-related aging and its mechanisms is the relationship of oncogenicity and teratogenesis to aging related processes.
The North Carolina group reported that the significance of the DMR's in cannabis dependence was higher than cannabis withdrawal [27]. Hence most of the ratios for the gene numbers affected in Table 11 were expected. However, the higher gene numbers affected in ALL (primarily a pediatric cancer) and AML (occasionally a pediatric cancer) and the higher significance level in AML found in withdrawal were unexpected and raise the intriguing possibility that the cannabis withdrawal state following birth may trigger leukemogenic gene activation. Whether this holds true for the other pediatric cancers previously related to cannabis remains to be studied. Moreover, this topic was shown to be of much greater significance beyond the field of pediatric cancer by the recent finding that many adult haemopoietic tumors have been shown to commence in fetal life [335], a finding which these latter investigators note may also apply more widely to the field of solid organ tumorigenesis.
One of the prominent findings to emerge from the above epidemiological overview was the salience of chromosomal disorders in both the congenital anomaly and the cancer datasets. Trisomies or monosomies of chromosomes 13, 18, 21 and X (including syndromes described by Turner and Klinefelter) were observed directly [66,103,115,120]. Moreover, strong signals were detected for acute lymphoid leukemia (which has been shown to often involve translocations between chromosomes 4, 9, 10, 11 and 22) [105,336] and testicular cancer [105,[112][113][114]121,122] (which has been shown to implicate chromosomes 1, 7, 8, 11, 12, 13, 18, 21, X and Y) [337]. The total length of these chromosomes together comprehends 1754 megabases of the 3000 megabases, or 59%, of the whole human genome directly impacted by cannabinoid-related genotoxicity/epigenotoxicity. Deletions of chromosome 22 in USA and microdeletions in Europe were also significantly cannabis-associated [66,103,115]. These data make the issue of chromosomal non-segregation, non-disjunction, aneuploidy, chromosomal breaks and translocations and subsequent teratogenic malignancy a very prominent feature of cannabis related genotoxicity. As described in considerable detail in the pathophysiological review, multiple direct and epigenomic pathways exist which comfortably explain and may account for these prominent and important clinical findings of hundred megabase scale epi/genotoxic activities.
Given so much powerful evidence for cannabinoid-related epigenotoxicity, the possibility that these epigenomic changes are potentially reflected as pro-ageing effects effectively accelerating natural aging warrants particularly careful consideration. On this issue, three tissues are of particular and pivotal importance namely: spermatocytes, oocytes and zygotes.

Spermatocytes
Classic photomicrographs of cannabis exposed sperm featuring multiple (up to four) heads, multiple tails, obviously deformed heads and tails on a background of proteinaceous and inflamed tissue [24] along with gross chromosomal translocations and ring and chain formation [23,260] give an obviously degenerate genotoxic appearance. Multiple cannabinoids are known to induce adverse mitochondrial effects, reduced energy charge and increased free radial flux [33,140] which are all changes that are well established as being age related. It has been shown that cannabinoid signaling via CB1R has a deleterious effect on sperm chromatin which increases along the epididymis, altered histone-protamine substitution via inhibition of transition protein 2 (TNP2) and leads to genome DNA fragmentation with compromise of male fertility [34]. Moreover, the above demonstration of cannabinoid-related gross changes to the tubulin code and meiotic apparatus (Supplementary Tables S9-S13) implies that not only are the microtubules of the sperm flagellum disrupted but so also are those comprising the sperm centrioles and first and second meiotic spindles. Since all of these various changes are age-defining and age-causing disorders, this implies that the age of cannabinoid-exposed sperm is advanced.

Oocytes
Diminished ovarian reserve was noted in the epigenomic dataset of Schrott (Supplementary Table S27; Schrott [27] Page 349) and is both a defining feature of female aging [1] and an important cause thereof [200]. Gross and severe morphological changes were noted in cannabis exposed oocytes induced to divide including chromosomal nucleoplasmic bridges, non-disjunctions, tripolar, quadripolar and pentapolar cell divisions along with an extremely high (20%) rate of oocyte death with just a single cell division. Moreover, oocyte depletion has been attributed primarily to failure of DNA damage repair [230]. As noted above, cannabis has been shown to suppress pituitary FSH secretion thereby interfering with the normal female hormonal cycle. All of these are clearly age-related and age-inducing changes.

Zygotes
Since both sperm and oocytes bear many chromosomal, genetic and epigenetic features of aging, it seems clear that these changes would persist in the pronuclei of the fertilized zygote and carry important influences into the first few rounds of zygotic cell division which are epigenetically controlled from the time of fertilization. These deleterious changes would be compounded by aberrant histone and protamine changes in sperm and by the disrupted tubulin code known to be borne by sperm. Together, these changes indicate that not only are the gametes themselves aged, but so too must the fertilized zygote be aged from-and actually even prior to-conception. It is noted again that the fragile process of human female meiosis is highly error prone ordinarily [248] which suggests that the tolerance for error under the influence of external xenobiotic genotoxic agents is very narrow indeed. These considerations raise the intriguing and very concerning possibility that the zygote itself may manifest advanced epigenomic age from even before fertilization and conception. It is noted that the newly described method of analysis of blastocystoid bodies derived from induced human pluripotential embryonic stem cells (iPS) might provide an ideal and ethical laboratory method to formally assess these issues [338].
It was recently shown that a key part in sperm maturation is played by the addition of mRNA exosomes (as epididymosomes) in the tail of the epididymis during sperm maturation. These extracellular packages of mRNA play a key part in early embryonic development during the initial divisions of the fertilized zygote and are under close control at several points by CB1R-mediated cannabinoid control [339]. Interference with this normal mechanism led to profound perturbation of sperm maturation, fertility and function. In this regard, the human system closely mirrors that seen in mice.

Cannabidiol and ∆8THC
At the time of writing, cannabidiol and ∆8THC have been allowed to freely penetrate culture without restriction in many places and have been made available in cookies, sauces, lollies, candies, crackers and in solid translucent blocks often being marketed as "legal weed".
In such a context, it is important to note that it was found long ago that the genotoxic moiety of cannabinoids lies primarily in their central olevitol nucleus, an activity which is little modified by their various side chains [340,341]. This important finding implicates most cannabinoids in genotoxic effects.
A recent very concerning paper demonstrated not only that many cannabinoids (including ∆9THC, ∆8THC and cannabidiol) could freely pass into the milk of dairy cattle fed legal hemp (with nominally less that 0.3% THC content) but that the cannabinoid concentration in milk could rise to a level where the total recommended daily dose of ∆9THC was exceeded [119]. Moreover, the cows themselves became obviously ataxic and "stoned" and stood motionless for extended period, not moving and not eating, apparently "stoned". They were also ataxic and had difficulty walking. After cessation of the hemp/cannabinoid feed, these changes abruptly declined. Most concerningly, the levels of cannabinoid found in the feed when analyzed by state-of-the-art tandem liquid chromatography/gas chromatography-mass spectrometry (LCGC-MS) techniques were more than ten times those found with the standard legally prescribed tests for cannabinoid, a finding which necessarily impugns and indicts so called legally safe "low-THC" hemp products and imperils public heath and safety.
Moreover, such findings dramatically and eloquently illustrate the florid manner in which such grossly affected animals in the food chain might pass on the severe genotoxic cannabinoid-mediated damage (which includes limblessness) as has been chronicled in recent reports from France and Germany [378][379][380][381].

Conclusions
Many metrics, including hormonal, mitochondriopathic, cardiovascular, hepatotoxic, immunological, genotoxic, epigenotoxic, disruption of chromosomal physiology, congenital anomalies, cancers including inheritable tumorigenesis, telomerase inhibition and elevated mortality point towards cannabinoid-exposed tissues being of advanced biological age. Evidence from many studies indicates extensive perturbation of the human epigenome by exposure to many cannabinoids. Since the epigenome has emerged as the key and central mediator of the panorganismal aging process [13][14][15][16]202,382], it becomes of primary importance to investigate its likely implication in aging processes directly by the application of late-generation epigenomic clocks [383][384][385][386][387][388][389]. The likely involvement of spermatogonia, oocyte and fertilized zygote in this accelerated aging process increases the importance of this enquiry for the health of subsequent generations, an enquiry which is heightened and intensified by the transgenerational transmission of cannabinoid-related epigenotoxicity in human sperm [26,27], to subsequent rodent generations [28][29][30][31][32]390], for pediatric brain function and development including autistic-like disorders [117,174,288,290,292,[295][296][297][298][299]391] and through the heritable passage of many birth defects [103,[108][109][110][111]115,118,120] including several pediatric cancers [63][64][65][66]105,121]. Inversion of the ratio of the minimum p-values between dependence and withdrawal for ALL and AML may imply that it is the activation of leukemogenic gene cassettes by the withdrawal state occasioned by birth which gives rise to these pediatric cancers. The genotoxic, epigenotoxic, mutagenic and teratological issues raised are clearly very serious and have been shown several times to greatly outweigh those attributable to tobacco and alcohol [103,[112][113][114][115]. These changes carry such far-reaching public health implications that they are worthy of investigation by the most advanced multiomics techniques including multichannel single cell epigenomic and 3D chromosomal topological techniques with appropriate resourcing to exhaustively perform these investigations in a translational multigenerational context.  Figure S1: Overall results-negative logarithm of p-Values for congenital anomalies from Schrott Database by organ system. Figure S2: Overall results-negative logarithm of p-Values for congenital anomalies from Schrott Database by organ target. Figure S3: Overall results-Boxplot of negative logarithm of grouped p-Values for congenital anomalies from Schrott Database comparing cannabis dependency with cannabis withdrawal. Figure S4: Number of genes annotated in the Schrott database for target organs by dependency status in (A) cannabis dependence and (B) withdrawal. Figure S5: (A) Numbers of gene annotations, (B) numbers of genes affected and (C) negative logarithm of p-value by cancer type-overall Schrott data. Figure S6: Direct comparison between p-values for cannabis cancer relationships between (A) cannabis dependence and (B) cannabis withdrawal, Schrott data.
Author Contributions: A.S.R. assembled the data, designed and conducted the analyses, and wrote the first manuscript draft. G.K.H. provided technical and logistic support, co-wrote the paper, assisted with gaining ethical approval, provided advice on manuscript preparation and general guidance to study conduct. A.S.R. had the idea for the article, performed the literature search, wrote the first draft and is the guarantor for the article. All authors have read and agreed to the published version of the manuscript.
Funding: This research received no external funding. No funding organization played any role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.