Bubbling beyond the barrier: exosomal RNA as a vehicle for soma–germline communication

‘Weismann's barrier’ has restricted theories of heredity to the transmission of genomic variation for the better part of a century. However, the discovery and elucidation of epigenetic mechanisms of gene regulation such as DNA methylation and histone modifications has renewed interest in studies on the inheritance of acquired traits and given them mechanistic plausibility. Although it is now clear that these mechanisms allow many environmentally acquired traits to be transmitted to the offspring, how phenotypic information is communicated from the body to its gametes has remained a mystery. Here, we discuss recent evidence that such communication is mediated by somatic RNAs that travel inside extracellular vesicles to the gametes where they reprogram the offspring epigenome and phenotype. How gametes learn about bodily changes has implications not only for the clinic, but also for evolutionary theory by bringing together intra‐ and intergenerational mechanisms of phenotypic plasticity and adaptation.

Germline transmission of acquired traits indicates that, as well as genetic information, gametes contain environmental cues with the potential to recapitulate acquired physiology in future offspring.Gametic DNA methylation (DNAme) and post-translational histone modifications (PTHMs) have long been considered prime candidates for transmitting environmental signals to developing offspring because of their involvement in regulating transcription (Allis & Jenuwein, 2016) and because their patterning in both somatic and germ cells is responsive to environmental change (Jaenisch & Bird, 2003;Szyf, 2015) and correlated between parent and offspring (Trerotola et al., 2015).Despite previously being dismissed as a result of widespread removal in germ cells and the early zygote, it is now accepted that such 'epimodifications' can be retained in the offspring (Borgel et al., 2010;Brykczynska et al., 2010;Gkountela et al., 2015;Hammoud et al., 2009;Wang et al., 2014), especially at loci known to control embryogenesis and development (Brykczynska et al., 2010;Kobayashi et al., 2012).Accordingly, debate surrounding 'germline epigenetic inheritance' (GEI) of acquired traits has typically focused on these mechanisms.However, as a result of recent data showing sperm also contain diverse populations of RNAs (Ostermeier et al., 2002) that are responsive to lifetime factors (Dias & Ressler, 2014;Rodgers et al., 2013;Sharma et al., 2016) and are passed to the oocyte at fertilisation (Ostermeier et al., 2004), and that injecting zygotes with either total- (Chen, Yan, et al., 2016;Gapp et al., 2014;Ng et al., 2014) or fractions (Rodgers et al., 2013;Sharma et al., 2016) of sperm RNA from manipulated fathers often causes partial (Chen, Yan, et al., 2016;Sharma et al., 2016) or outright (Gapp et al., 2014;Ng et al., 2014;Rodgers et al., 2013) phenocopying in the offspring, attention is now shifting to how gametic RNAs reprogram the offspring epigenome and phenotype.
Of the ∼22,000 transcripts in mature human spermatozoa (Sendler et al., 2013), most (∼60%) are long (>200bp) (Jodar et al., 2013), but it is shorter (∼22 nucleotides) microRNAs (miRs) and (∼16-33 nucleotides) tRNA-derived fragments (tRFs) that are most implicated in GEI as a result of evidence indicating that they modulate embryonic development via gametic de novo DNAme, PTHMs and competitive silencing of maternal mRNA stores in the early zygote (Chen, Liu, et al., 2016;Zhang et al., 2019).For example, when nine miRs over-represented in the sperm of traumatised mice were injected into one-cell zygotes, the resultant offspring developed their fathers stress-sensitive phenotype and exhibited transcriptomic silencing of 288 genes in the paraventricular nucleus, suggesting upstream epigenetic remodelling (Rodgers et al., 2013).However, considering the transcriptional quiescence of sperm cells (Grunewald et al., 2005), the origin of such spermatic non-coding RNAs (ncRNAs) is a separate and somewhat mysterious question.
Despite in vitro evidence that gametic RNAs relevant to transmitting lifetime experiences may be transcribed endogenously in response to ligands (e.g.cocaine) (Meizel, 2004;Yazigi et al., 1991) directly binding gametic membrane-receptors, this direct route to influencing the gametic RNA profile is probably not the case in vivo, where the blood-testis and blood-follicle barriers (BTB and BFB, respectively) impede all non-essential somatic factors from reaching the gametes (Mruk & Cheng, 2015;Ohno et al., 2016).Alternatively, a recent hypothesis suggests that these gametic RNAs instead arrive from the soma.Here, we review data pointing to mobile somatic RNAs as vehicles for animal soma-germline communication.We focus first on the extracellular vesicle-mediated release of RNA from somatic cells and then introduce the contribution of extracellular RNAs (exRNAs) to intercellular signalling, before discussing evidence that EV exRNA-signalling occurs over short and long somatic distances to transmit information about acquired phenotypes to the animal germline.
Extracellular vesicles: a minute and numerous cellular messaging platform?Extracellular vesicles (EVs) are tiny (40 nm to 1.5 μm) lipid bilayer-enclosed bubbles released by almost all cells and all recorded taxa (Andaloussi et al., 2013) (Fig. 1A).They are present in most biofluids (blood, urine, breast-milk, spinal-fluid, saliva and both amniotic and seminal fluids), as well as various fluids inhabited by cellular life (e.g.cell culture media, beer and the ocean) (Biller et al., 2017;Février & Raposo, 2004;Stensballe & Bennike, 2014).EVs exhibit a morphological diversity partly explained by different modes of biogenesis.Exosomes (30-100 nm) are the smallest EV and are formed intracellularly in multivesicular bodies (MVBs) and actively released by exocytosis, whereas microvesicles (50 nm to 1.5 μm) and apoptotic bodies (50 nm to 5 μm) are released by direct budding from the cell membrane (Andaloussi et al., 2013) (Fig. 1B).Biofluids contain mixed populations of EVs arising within a variety of tissues, but it is what is on the inside that counts.EVs contain an assortment of proteins, lipids and nucleotides that they shield from extracellular catabolic enzymes (Raposo & Stoorvogel, 2013).Moreover, rather than being 'snapshots' of the parental cell cytosol, evidence suggests that EVs, especially exosomes, and particularly their RNAs, are 'packaged with purpose' .For example, some RNAs prominent in cells (such as rRNAs) are largely absent from their exosomes, whereas many different mRNAs and ncRNAs (primarily miRs and tRFs) are often relatively over-represented (Colombo et al., 2014;Nolte-'t Hoen et al., 2012;Skog et al., 2008).Some comparatively abundant EV RNA species are practically undetectable in the cytosol of their parent cell (Li et al., 2013).For example, Valadi et al. (2007) found 1300 mRNAs and >120 miRs in mast-cell derived exosomes, 270 of which were not detected in the source cell population.
The mechanisms that begin to explain such selective cargo sorting into exosomes are slowly being uncovered.These mechanisms largely revolve around pathways that direct exosome biogenesis machinery to initiate at sites on the endosome membrane pre-loaded with site-specific molecular cargo of various kinds.Classical exosome biogenesis depends on endosomal complexes required for transport (ESCRT), whereby four ESCRT protein complexes and their accessory proteins recruited into the maturing endosome recognise and bind to mono-and poly-ubiquitylated lysine residues in endosomal transmembrane proteins to initiate local exosome formation via controlled membrane invagination and scission.Alternative pathways are the syndecan-syntenin-ALIX pathway and 'wedge-driven curvature' mechanisms of exosome biogenesis, featuring cone-shaped membrane lipids (e.g.ceramide) and proteins (e.g.tetraspanins); for a review, see Dixson et al. (2023).Although different exosome biogenesis pathways themselves influence the molecular composition of exosome cargos (Shukla et al., 2023), another layer of regulation is introduced by the interaction of the exosome biogenesis machinery with myriad sorting 'effector' molecules on the endosome membrane that function to bundle their own unique subset of varied cargo into the MVB.For RNA, these effector molecules are RNA-binding proteins (RBPs) that selectively bind to and concentrate RNAs with specific sequence motifs, structures or modifications to sites of exosome biogenesis (Fabbiano et al., 2020), with simple miR binding motifs (EXOmotifs) of many cargo effector molecules (e.g.hnRNPA2B1 -GAGG/AGG/UAG, hnRNPK -AsUGnA, SYNCRIP -GGCU, FMRP -AAUGC) already being characterised (Dixson et al., 2023).For example, hnRNPA2B1 incorporates various miRs (e.g.miR-198, miR-93 and miR-17) and long ncRNAs (lncRNAs) (e.g.AFAP1-AS1, AGAP2-AS1, H19 and LNMAT2) into MVBs (Dixson et al., 2023), with LNMAT2 (Chen et al., 2020) and H19 (Lei et al., 2018) depending on cannonical GAGG-binding, whereas YBX1 and NSUN2 uniquely bind to motif sequences characterising exosomal mRNAs in HEK293 cell lines, especially at specific locations (3 untranslated region) within the exosomal mRNA primary structures (Kossinova et al., 2017), thus highlighting the contribution of both sequence-and context-specific cues in the sorting of exosomal RNA.
These data suggest that, despite being historically dismissed as 'cellular debris' , EV and exRNA production is regulated across many levels and functions as a novel medium for intercellular signalling.This is supported by data showing that EVs can induce receptor-mediated intracellular signalling cascades in somatic cells (Raposo et al., 1996) or fuse with their cellular membrane before unloading their cargo inside (Mulcahy et al., 2014), thus modifying the recipient cell (e.g. by the catalysis of EV enzymes) (Golub, 2011), translation of EV mRNAs (Aliotta et al., 2010) or miRNA-and other ncRNA-dependent transcriptional silencing and epigenomic remodelling (Quesenberry et al., 2015).Heusermann et al. (2016) provided structural evidence for a functional role of these tiny bubbles by showing that cells have adaptations for actively 'pulling' and 'grabbing' EVs out of their extracellular environment before internalising them (Fig. 2).Indeed, there are now conclusive examples of EVs modifying the behaviour of target cells by various means.B-lymphocytes infected with Ebstein-Barr Virus secrete exosomes containing miRNAs that down-regulate two immunoregulatory genes in neighbouring B-cells to promote infection (Pegtel et al., 2010).Haematopoietic stem cells incubated with leukemia cell-derived EVs exhibited hyper-methylation of the tumour-suppressor genes P53 and RIZ1 and a leukemia-like malignant phenotype caused by uptake of DNA-methyltransferase and BCR-ABL1 mRNAs (Zhu et al., 2014).Human tumour cells incubated with vemurafenib (a BRAF-inhibiting anti-cancer drug) were demonstrated to release exosomes containing miR-211-5p, which was not found in exosomes of control tumours, and which slowed cancer growth by inducing the TRPM1-dependent pathway in neighbouring cells (Lunavat et al., 2017).
Intriguingly, lifetime habits and experiences relevant to GEI have also been found to modify the RNA content of EVs.Exosomes of human monocytes exposed to alcohol both in vitro and in the blood of alcoholics contain a higher proportion of miR-27a (that polarises monocytes into M2 macrophages) compared to non-exposed control EVs (Saha et al., 2016).Cultured vascular endothelial cells release exosomes with widespread changes in their mRNA profiles after exposure to experimental hypoxia (de Jong et al., 2012).Circulating exosomes of male rats exposed to acute stress contain significantly more of the heat shock protein 72 (HSP 72) and less of two miRs (miR-142-5p and miR-203) compared to non-stressed control rats (Beninson et al., 2014) and white adipocyte-derived exosomes of obese human subjects were found to contain 55 differentially expressed miRNAs compared to EVs from lean individuals (Ferrante et al., 2015).
That the RNA cargo of EVs depends on a cell's physiological context is now generally accepted, but how such context-specific cargo is sorted into EVs remains an outstanding and important question.Given their  central role in sorting exosomal RNA, the differential expression, degradation and localisation of RBPs will probably explain a large portion of the variability in exosomal RNA cargoes, with experience-dependent modifications to effector RBP(s) or the RNA cargo itself also playing a significant role.For example, oxidative stress and hyperoxia following H 2 O 2 exposure leads to phosphorylation of caveolin 1, a component of membrane lipid rafts, which then binds to and stabilises the exosome RNA sorting effector RBP hnRNPA2B1.Following O-GlcNAc-modification and sumoylation of respective RNA-binding domains, hnRNPA2B1 recruits many stress-specific miRs to the MVB, thus supporting a hyperoxia-regulated exosome RNA sorting axis (Lee et al., 2019).Additionally, the role of RBPs in forming and interacting with liquid-liquid phase separated membraneless granules [e.g.ribonucleoprotein (RNP)/Processing (P) and stress granules] (Lin et al., 2015) positions them at a promising regulatory intersection between cytoplasmic signalling and dynamic cell state-dependent exosomal RNA loading (Suresh et al., 2018;Liu et al., 2021); however, more work needs to be conducted.Conversely, following EV release into the extracellular environment, it is now relatively well understood that EVs target specific cell-types via the complementarity of extracellular factors on their cell surface and target cell membrane-receptors (O'Brien, Breyne, et al., 2020).Such factors include electrical charge and the protein (integrin and tetraspanin profiles, fibronectin), lipid and glycan composition of the EV external membrane (Murphy et al., 2019;Rana et al., 2012).There is substantial effort to exploit these mechanism so that EVs can be co-opted to deliver therapeutic drugs to specific cells implicated in various disease-states and even through the formidable blood-brain barrier (Bunggulawa et al., 2018).For example, mice exosomes engineered to express a neuron-specific CRM and loaded with anti-BACE1 small interfering (siRNA) knocked-down expression of BACE1 only in the targeted cortical cells when injected i.v.back in to the mice (Alvarez-Erviti et al., 2011).
Bubble-bound soma-germline signalling.Above, we introduced the newly uncovered role and widespread permissibility of EVs throughout the body.Here, we show how this permissibility also extends to the germline.It is increasingly hypothesised that the gametic ncRNAs found to mediate GEI travel to the gametes from the soma inside exosomes.Proteome analyses from the EV-cargo database 'EVpedia' supported this hypothesis by showing that exosomes from many tissues, including the central nervous system, feature extracellular proteins predicted to facilitate interactions with spermatozoa (Kim et al., 2015).Indeed, EVs released by testicular Sertoli cells were discovered in the interstatial space and seminiferous tubules after being injected into the rat efferent duct, consistent with their crossing the BTB (Ma et al., 2022; see also Choy et al., 2022).Additionally, ncRNA-rich prostate-(prostasomes) and epididymis-derived EVs (epididymosomes) interact with and assist sperm during spermatogenesis (Frenette & Sullivan, 2001;Oh et al., 2009;Stegmayr & Ronquist, 1982;Suryawanshi et al., 2012).Epididymosomes even have their own 'docking bay' at the post-acrosomal domain of the sperm head where receptor-bound EVs are actively transported prior to cargo delivery (Zhou, Stanger, et al., 2019) (Fig. 3).Similarly, oocytes interact with EVs from nearby cells in the follicular antrum (da Silveira et al., 2012;Sang et al., 2013) and when travelling through the fallopian tubes (Sohel et al., 2013;Tannetta et al., 2014).Interactions with some of these 'nurse cells' are confirmed to modify sperm RNA profiles.Sharma et al. (2018) reported a global increase in the small ncRNA content of sperm co-incubated with epididymosomes in vitro and, by metabolically labelling RNA in live mice, conclusively showed that RNAs in their mature sperm were transcribed in the epididymis.Furthermore, during their post-testicular maturation moving through the epididymis, the RNA profile of sperm cells changes in a manner reflected by the RNA cargoes of epididymosomes derived from the corresponding segments (Hutcheon et al., 2017;Reilly et al., 2016;Sharma et al., 2018), providing evidence that epididymosomes modulated the RNA-profile of sperm during their 2 week passage from the testes to the vas deferens.For example, RNA-sequencing of mouse epididymosomes isolated from different segments of the mouse epididymis identified >350 different miRs, most of which (>60%) were also present in sperm cells isolated from the same epididymal locations (Reilly et al., 2016).
Partly as a result of sperm lacking machinery for endocytic uptake and vesicular degradation (Jones et al., 2013), it has been hypothesised that epididymosomes form transitory membrane fusion pores and detach after cargo release (Trigg et al., 2019).This would explain why many in vitro studies report epididymosomes still present in the seminal fluid after shuttling their cargo into sperm.It is also consistent with the identification by immunoelectron microscopy of stalk-like projections formed between epididymosomes and sperm cells during their interactions (Nixon et al., 2019) (Fig. 4A), as well as between sperm and oviduct-derived vesicles (oviductosomes) (Al-Dossary et al., 2015) (Fig. 4B), which, taken together, suggests an evolutionary conserved mode of sperm EV-reception spanning maturation and fertilisation.Similarly, in vivo fluorescence imaging found that cumulus cell vesicles shuttle mRNAs and lncRNAs through the zona pellucida region of bovine oocytes when travelling inside tunnelling nanotubule-like projections (Macaulay et al., 2014) (Fig. 4C and D).
Relevant to the working hypothesis that parental EV-derived gametic ncRNAs shape offspring phenotypes, ncRNAs delivered to the sperm cells by epididymosomes are argued to be essential for embryonic implantation and development.Conine et al. (2018) reported that embryos produced by intracytoplasmic sperm injection (ICSI) of immature caput sperm into oocytes exhibited overexpression of many epigenetic regulators and failed to implant murine uteri, whereas mature cauda sperm-derived ICSI embryos implanted and developed normally.Significantly, this group discovered that the ICSI implantation capacity of caput sperm-derived embryos could be rescued by injecting them with small ncRNAs (sncRNA) cargo of cauda epididymosomes, providing strong evidence for a causative role of epididymosome RNAs in regulating essential post-fertilisation developmental processes.
However, it should be noted that these data seemingly contradict other studies reporting successful implantation and development of ICSI-generated embryos derived from sperm taken from the mouse caput epididymis (Suganuma et al., 2005;Zhou, Suzuki, et al., 2019).Conine et al. (2018) suggested different mice strains and ages, as well as epididymal dissection margins and pre-transfer embryo-stage, as potential reasons for these conflicting findings.They reinforced their previous work by rescuing the fertilisation capacity of caput sperm by injecting them with synthetic copies of 27 of the most abundant miRs in mature sperm that are absent in caput sperm (Conine et al., 2019) and also detailed a method for trafficking sncRNAs into sperm in vitro using engineered epididymosomes to improve in vitro fertilisation efficacy and clinically attenuate detrimental GEI-linked sperm sRNAs (Rando et al., 2022).Short-range environment-germline messaging.Above, we described how EVs, primarily from the male epididymis, interact with and shuttle ncRNA into gametes, as well as discussing evidence that they have a functional role in development.However, there is now emerging evidence that epididymosomes also transmit environmental information to the gametes via their altered RNA cargoes.For example, Sharma et al. (2016) reported that male mice fed a low-protein diet acquired metabolic alterations (i.e.upregulation of hepatic lipid and cholesterol biosynthesis genes) that were inherited by in vitro fertilisation-generated offspring, and that both the father's epididymosomes and sperm contained significantly more Gly-tRFs and Let-7 miRNAs than control mice.Similarly, Rompala et al. (2018) showed that, in an existing model of alcohol-related GEI, male mice exposed to chronic intermittent ethanol consumption had epididymosomes carrying altered tRF-cargoes compared to controls, and that these changes were also observed in their sperm tRF-profile.Although correlative, these studies showing a concurrent experience-dependent modification of the ncRNA content of gametes and the EVs known to contribute to their RNA-profiles provide the best evidence to date that the gametic RNAs causally mediating GEI are delivered there by somatic EVs.This pathway for an environmental exposure to alter the sperm transcriptome has been shown most contiguously by Trigg et al. (2021).Here, mice exposed to the reproductive toxicant acrylamide exhibited altered gene expression in the endothelial lining of the caput epididymal, namely differential expression of transcription factors linked to miR biogenesis.At the time of exposure, caput epididymosomes recapitulated altered miR profiles seen in the caput endothelial, and so too did the sperm cells that interacted with them, once again indicating shuttling of environmentally-responsive somatic RNAs into sperm via exosomes.These acrylamide miR-altered spermatazoa then produced embryos with transcriptional differences during development that can conceivably account for their developmental abnormalities (Exon, 2006;Trigg et al., 2021) and demonstrate the existence of a bubble-bound communications platform linking parental environment to offspring phenotype.Furthermore, as a result of sperm exhibiting adaptations that make them receptive to such communication (Figs 2 and 3), it has been argued that sperm actively regulate the inheritance of acquired traits, rather than being passive conduits for their transmission (Sciamanna et al., 2019).This viewpoint is further supported by active mechanisms for the internalisation and incorporation of exogenous nucleic acids into the sperm nucleus (Sciamanna et al., 2003).
Long-range environment-germline messaging.Here, research is introduced which extends the hypothesis that EVs inject gametes with GEI-mediating RNAs to EVs travelling from far away animal tissues.Indeed, that 'deeply' somatic ncRNAs reach the germline has been known since Craig Mello's Nobel Prize winning work showing that exogenous short double-stranded RNA (sdsRNA) injected into the nervous system of Caenorhabditis elegans blocked translation of a target transcript in both injected organisms and their progeny via the now extensively studied RNA inteference (RNAi) pathway (Fire et al., 1998).Indeed, the idea that sdsRNAs injected into an organism reach the germline and subsequently silence targeted genes in the zygote (termed 'parental RNAi') is now relatively well studied because of its conceivable application to controlling pests, and it has been demonstrated for populations of parasitic wasps, crickets and milkweed bugs (Aronstein et al., 2011).Similarly, Devanapally et al. (2015) more recently showed that a synthetic siRNA transcribed only in the nervous system of C. elegans silenced a germline-specific reporter construct for >25 generations, and Posner et al. ( 2019) identified an endogenous neuron-specific siRNA that is amplified in the germline where it silences offspring genes that inhibit stress-sensitive chemotaxis, namely saeg-2, for at least three generations.However, only in the last few years has evidence emerged indicating that such long-distance RNA movements may be mediated by EVs.Cossetti et al. (2014) reported that xenografting green fluorescent protein (GFP)-expressing tumours onto mice resulted in GFP mRNA being detected in both the mices' circulating exosomes and sperm, suggesting circulatory EV-shuttling of exRNAs to the mammalian germline.Additional evidence is now provided by O'Brien et al. (2020).They describe how a virally-inserted gene encoding human miR-941 expressed only in mice neurons resulted in the presence of miR-941 in the lymph nodes, vas deferens, epididymis and a very small fraction of circulating-EVs.The synthetic viral vectors were not found outside of the injected neurons, and so cannot account for the spread of miR-941, consistent with transportation within the circulatory system.Crucially, miR-941 was also detected in embryos derived from pairings between miR-941-expressing males and wild-type females.Rinaldi et al. (2023) most recently demonstrated that transgenic neural-and adipose-specific Cre recombinase mice, as well as AAV-mediated neural Cre expression, resulted in clear off-target recombination in the epididymis, and that this distal recombination involved transportation via the circulatory system using both parabiosis and either serum or serum-EV injections into reporter mice.Together, these data provide the first direct evidence that ncRNA transcribed in the soma can be shuttled inside EVs into animal gametes and even into the next generation (Fig. 5).An objective for future work is to demonstrate an endogenous somatic exRNA being shuttled into the germline via EVs and subsequently modifying the developmental trajectory of the offspring in a manner that phenocopies the parent's acquired phenotype.
Although concerned explicitly with neither exRNA nor EVs, Zeybel et al. (2012) provide evidence for an endogenous circulatory factor involved in transgenerational inheritance of an environmentally acquired trait.They reported that rats with induced liver fibrosis (scarring caused by liver damage), when bred with healthy control rats, had two generations of offspring each more resistant to liver fibrogenesis than their parents.Similar to these offspring, rats injected with serum from liver-injured rats became significantly more resistant to liver fibrosis than controls, and exhibited the same epigenetic mark (enriched H3K27me3 at the PPARGγ gene) in their sperm as observed in both the sperm and liver of fibrotic fathers.These data suggest that soluble factors in the father's blood were what communicated the epigenetic 'memory' of liver-specific wound healing to their gametes.Although the identity of the soluble factor(s) was not discovered, it was shown that they were released by cultured fibroblasts in vitro, and so it is not improbable that they were EVs containing protective epigenome-remodelling RNAs expressed during liver damage.Similarly, Benito et al. (2018) reported that mice exposed to exercise-based enrichment benefited from improved cognitive performance, and that this benefit was transmitted to F1 offspring by two sperm miRs: miR212 and miR132.Interestingly, De Miguel et al. ( 2021) provide indirect evidence that the GEI-mediating sperm RNAs reported by Benito et al. (2018) could have arrived from the bloodstream by showing that analogous benefits of exercise could be transferred from exercised to sedentary mice by way of repeated serum injections.Although proteomics suggested it was higher levels of complement-cascade inhibitors in the serum of exercised-mice that contained the 'memory' of the transferred trait (Schroer et al., 2023), the serum transcriptomes were not analysed, meaning that exRNAs such as miR212 and miR132 could have also played a role.These inter-animal transfers of acquired characters are strikingly reminiscent of James McConnell's highly controversial 'memory transfer' experiments in the 1960s, where learned behavioural responses were transferred from trained flatworms into naive ones by injecting them with RNAs extracted from the brains of the trained organisms (McConnell et al., 1968).Those studies of 'chemical memories' and the subsequent ones on other organisms, including rodents, were largely discredited, but are now gaining renewed interest coinciding with the discovery of exRNAs and rigorously controlled replications by many notable physiologists (Bédécarrats et al., 2018;Shomrat & Levin, 2013).Now that circulating ncRNA may indeed be a legitimate, non-chromosomal heritable factor with the potential to transmit acquired traits across both bodies and generations, a view is emerging of RNA as a quasi-universal 'memory molecule' , residing at a previously hidden intersection between intraand intergenerational learning and phenotypic plasticity.
Similarly, apparently adaptive cases of epigenetic inheritance (Katzmarski et al., 2021;Vallaster et al., 2017;Zeybel et al., 2012) and the aforementioned evidence for exosomal RNAs as vehicles for (particularly long-range) soma-germline signalling have revived debate regarding the status of 'Lamarckism' in modern evolutionary theory.Most specifically, Charles Darwin's particular mechanism for how 'Lamarckian' inheritance of acquired traits would work, his theory of Pangenesis (1868), is receiving renewed attention as a result of seemingly pre-empting the discoveries surrounding soma-germline exRNA signalling.Darwin reasoned that somatic information travelled between every part of the body, including to the gametes, in the form of tiny particles called 'gemmules' , and that these particles were released by all cells and were as 'inconceivably minute and numerous as the stars in heaven' (Darwin, 1868;p. 404).Darwin hypothesised that his imagined gemmules probably functioned by '[penetrating] other nascent cells and [modifying] their subsequent development' (Seward, 2010;p. 130), and that, because their nature would be modified in accordance with the state of the body, they could effectively communicate their experiences to the gametes contained within it.Although an assessment of if and how such a 'pangenic' mode of acquired inheritance might contribute to evolution and speciation is beyond the scope of the present study (for a discussion, see Spadafora, 2023), one conceivable and now testable hypothesis involving 'physiological selection' was proposed almost 150 years ago by Darwin's successor Romanes (1887) and is discussed in an article by Noble and Phillips (2023).Summary 'Weismann's barrier' has had a formative impact on biology for over a century by reducing inheritance to the transmission of gametic DNA and the inheritance of acquired traits to an impossibility.However, although the developmental and physiological isolation of soma and germline is consistent with this view, the growing and unignorable evidence for germline epigenetic inheritance flatly repudiates it.The rapidly developing body of evidence in support of germline epigenetic inheritance nonetheless raises its own new and important questions, the most fundamental being 'how are lifetime experiences encoded in an animal's still tightly guarded gametes?' .Recent studies have identified gametic RNAs as encoding environmental information using a methodology that involves injecting them into zygotes, and the recently uncovered and apparently boundless realm of mobile exRNA-bubbles suggests they are of an often deeply somatic origin.Long-range soma-germline signalling is especially significant because it drastically enlarges the scope of which acquired traits, at least in principle, can be transmitted to the next generation and, in doing so, adds greater resolution to the 'image of the body' constructed in the gametes during an animal's lifetime.The growing recognition of EVs is illuminating previously hidden complexities of multicellular homeostasis and disease aetiology, and provides a promising platform for their future clinical detection and control, but the theoretical significance is greater still.It is only a matter of time before the discovery of the adaptations (or more probable exaptations) enabling far away cells to send EVs targetted directly to the gametes is reported, and this will provide one of the clearest examples to date for the evolution of evolvability itself.Moreover, as well as radically deviating from orthodox views of the flow of heredity information, the idea that so many cases of GEI are expected to change an animal's evolutionary fitness is re-igniting the historical debate between Lamarckian and neo-Darwinian evolution.By the same token, because GEI may use a soma-germline vehicle so close to Darwin's 'gemmule' , this ought to remind us that, in science, 'progress [is] often achieved by a 'criticism from the past' … Theories are abandoned and superseded by more fashionable accounts long before they have had an opportunity to show their virtues' (Feyerabend, 1993;p.59).

Figure 1 .
Figure 1.Imaging the intercellular and extracellular environmental of extracellular vesicles A, left: a transmission electron micrograph showing internal multivesicular bodies (MVBs) and a lysosome of a B cell releasing exosomes.Reproduced from Edgar (2016).B, right: a scanning electron micrograph of an OVCA-432 ovarian cancer cell shedding many microvesicles.Reproduced from Giusti et al. (2013).

Figure 2 .
Figure 2. Somatic cells feature adaptations for actively catching extracellular vesicles A, showing (green) exosomes 'surfing' the length of filopodia to the cell body of a human primary fibroblast, as recorded by live cell digital image correlation confocal laser scanning microscopy (DIC/CLSM).Trajectory and speed (0-5 μm s -1 ) of exosome travel shown by lines of different colours.B, DIC/CLSM images showing 'pulling' extracellular vesicles and their movement to the cell surface by filopdia.C, 'grabbing' of (green) extracellular vesicles towards the cell surface by filopodia captured by total internal reflection flourescence (TIRF) microscopy.Coloured arrows indicate direction of filopodia movement and (purple) lines indicate trajectory of exosome movement.All images taken from Heusermann et al. (2016).

J
Physiol 602.11

Figure 3 .
Figure 3. Mammalian sperm cell feature adaptations for capturing and internalising extracellular vesicle cargo Left to right: duel-labelled fluorophores associated with lipid raft component GM1 gangliosideson (red) and biotin-labelled epididymal protein (green) distributed across a caputderived spermatozoa.Physical interactions between GM1+ lipid rafts and biotin+ exosomes (yellow) occur following their spatial concentration at the subacrosomal ring and post-acrosomal domains prior to cargo delivery.Reproduced from Zhou, Stanger, et al. (2019).

Figure 4 .
Figure 4. Continuous interaction between extracellular vesicles an germ cells from maturation to conception A, transmission immunoelectron micrograph showing a gold-particle conjugated epididymosome (ES) forming a bridge-like connection with a sperm membrane Reproduced from Nixon et al. (2019).B, transmission electron micrograph showing an oviductosome ( * * ) forming a fusion stalk with a sperm cell membrane.Reproduced from Al-Dossary et al. (2015).C and D, showing (green) RNAs from cumulus cells travel inside (2 mm) trans-zonal projections (TZPs) into the zona pellucida region of bovine oocytes.Reproduced from Macaulay et al. (2014).

J
Physiol 602.11

Figure 5 .
Figure 5. Schematic of the emerging role of extracellular vesicles (EVs) RNA in short-and long-range soma-germline communication Lifetime factors including nutrition, stress, and disease induce the production and release of EVs containing variousRNAs into the circulatory system, including from the brain after surpassing the blood brain-barrier (BBB).Tissue EVs migrate through the blood-follicle barrier (BFB) and blood-testis barrier (BTB) and enter the antral follicle (left) and seminiferous tubules (right), respectively.In the antral follicle, EVs travel inside trans-zonal projections extending from the cumulus granulosa cells through the zona pellucida and into the maturing ooctyte.In the testis, EVs interacting with developing sperm cells in the seminiferous tubules.EVs are also produced by different segments of the epididymis and interact with sperm prior to ejaculation.Modified gametes produce a fertilised zygote with influenced by lifetime factors of one or both parents.