Meisosomes, folded membrane platforms, link the epidermis to the cuticle in C. elegans

19 Apical extracellular matrices (aECMs) form a physical barrier to the environment. In C. elegans , the 20 epidermal aECM, the cuticle, is composed mainly of different types of collagen, associated in circumferential 21 ridges separated by furrows. Here, we show that in mutants lacking furrows, the normal intimate connection 22 between the epidermis and the cuticle is lost, specifically at the lateral epidermis, where, in contrast to the 23 dorsal and ventral epidermis, there are no hemidesmosomes. At the ultrastructural level, there is a profound 24 alteration of structures that we term “meisosomes”, in reference to eisosomes in yeast. We show that 25 meisosomes


Introduction
Apical extracellular matrices (aECMs) are associated with all epithelia and are essential for animal life.In C. elegans, a collagen-rich aECM covers the entire surface of the worm, and is called the cuticle.It is a complex multilayer structure that acts as an exoskeleton, to which body-wall muscles are connected via structures called hemidesmosomes that traverse the intervening epidermis (Davies & Curtis, 2011;Johnstone, 2000;Page & Johnstone, 2007).Specific subsets of the more than 170 collagens are enriched in the different layers of the cuticle.Some are involved in the formation of distinct structures, including the circumferential parallel furrows that cover the entire animal (Cox & Hirsh, 1985;Cox et al., 1980;McMahon et al., 2003;Page & Johnstone, 2007;Thein et al., 2003), and the longitudinal lateral alae.The latter have been proposed to be involved in facilitating the traction of C. elegans to its substrate during locomotion, although, notably they are not present from the L2 through the L4 larval stages (Cox et al., 1981;Katz et al., 2022).
The cuticle also constitutes a physical barrier, protecting the underlying epidermal syncytium from biotic and abiotic stresses.When the cuticle is damaged, mechanically or through infection, the epidermis reacts, activating an immune response, reflected by the increased expression of antimicrobial peptide (AMP) genes, including those of the nlp-29 cluster (Belougne et al., 2020;Pujol, Cypowyj, et al., 2008;Taffoni et al., 2020).
If the pathway leading to AMP induction in the epidermis is well described (reviewed in (Martineau et al., 2021)), exactly how the epidermis senses cuticular damage remains obscure.Part of the induction seen in dpy-10 mutants is the consequence of an increase in the levels of hydroxyphenyllactic acid (HPLA).This metabolite, derived from tyrosine by transamination and reduction, activates the G-protein coupled receptor (GPCR) DCAR-1 (Zugasti et al., 2014), switching on a signalling cascade that leads to AMP gene expression (Polanowska et al., 2018;Zugasti et al., 2016).What provokes elevated HPLA levels in dpy-10 mutants is, however, currently unknown.Further, the HPLA/DCAR-1 signalling pathway only accounts for part of the elevated nlp-29 expression seen in furrow collagen mutants (Zugasti et al., 2014).We have therefore proposed that a hypothetical, cuticle-associated, damage sensor exists that would control, in an as yet undefined manner, AMP gene expression.This damage sensor would also function to coordinate antimicrobial responses with the distinct detoxification and hyperosmotic responses that are simultaneously activated in furrow collagen mutants (Dodd et al., 2018;Rohlfing et al., 2010;Wheeler & Thomas, 2006).
In yeast, eisosomes, single invaginations of the plasma membrane underneath the aECM, the cell wall, are responsible for detecting changes in nutrient availability, but also cell wall integrity and membrane tension.
They relay information primarily via BAR domain proteins to induce the appropriate responses to environmental changes (Appadurai et al., 2020;Lanze et al., 2020;Moseley, 2018).While nematodes lack eisosomes, the apical plasma membrane of the epidermis, which is in direct contact with the aECM (the cuticle), is characterised by localised regions of folds that can be observed by electron microscopy (Wood, 1988).Given their superficial similarity, we refer to these structures as meisosomes, for multifold-eisosomes.
In this study, we undertook a detailed ultrastructural analysis of meisosomes in adults, as well as characterising them during development, and in furrowless mutants.This mutant analysis proposed a role for meisosomes in maintaining the structural integrity of the cuticle and the epidermis, and has opened the way to future, more detailed, characterisation of their function.

Meisosomes: epidermal plasma membrane folds interfacing the apical ECM
The stacked organelles that we refer to as meisosomes were mentioned during early electron microscopy characterisation of C. elegans (Wood, 1988).A survey of the long transverse transmission electron microscopy (TEM) series "N2U" from the MRC archive (White et al., 1986), which is of a 4-5 day old adult hermaphrodite, found hundreds of meisosomes across the 400 odd available transverse prints in the midbody.
As a first step in the detailed investigation of meisosomes, we undertook a focused TEM study to determine their 2D organization and their 3D structure.Meisosomes are repeated folded structures at the interface of the aECM (the cuticle) and the epidermis (Figure 1).They can be found in similar locales at all larval stages, predominantly in the epidermal syncytium hyp7 at the lateral, dorsal and ventral ridges, and in the tail tip epidermal cells.They are not present on the basal side of the epidermis, nor in the seam cells, the rectal epithelia, nor in the pharynx (Figure 1 & Figure 1-figure supplement 1).In adults, meisosomes typically comprised 4 to 10 closely apposed parallel folds of the plasma membrane, although we observed some with up to 30 folds (Figure 1C).Most folds formed an indentation 200-400 nm deep (Figure 1C-G).The folds were regularly spaced.The gap between each cytoplasmic-facing plasma membrane fold was 35 nm, 75 % wider than that between the folds made from cuticle-facing plasma membrane (20 nm) (Figure 1G).The cytoplasmic faces of the folds were free of ribosomes but contained dense material close to the plasma membrane, separated by a thin less electron-dense area (Figures 1G).Meisosomes were frequently found in close proximity to mitochondria (85 %, n=355) (Figures 1C-E).On their apical side, some folds were found close to a furrow (Figure 1C&E).Although very variable in a single worm, meisosomes of similar dimensions were observed in both transverse and longitudinal sections (Table 1), consistent with a random orientation relative to the animal's antero-posterior axis.This random orientation was clearly visible in electron micrographs of freeze-fractured samples (Figure 1-figure supplement 2A-B).It contrasted with a much more regular pattern in moulting larvae, in which meisosomes were in-between the position of circumferential furrows, (Figure 1-figure supplement 2C-D).As described below, a similar organisation could be observed through moulting using in vivo markers.Some much smaller meisosomes, typically with only 2-4 shallow folds were seen in the thin epidermal tissue that lies between body-wall muscles and the cuticle (dorsal and ventral epidermis, see Figure 1A) and that is largely devoid of cytoplasmic content (Figure 1-figure supplement 3).
To understand meisosomes' 3D structure, we undertook a tomographic analysis on serial 350 nm-thick sections.This approach confirmed the existence of groups of parallel folds, all found in continuity with the plasma membrane (Figure 2).The tomographic analysis also revealed variability in the geometry of the folds.
Although most groups of folds were perpendicular to the apical surface, some were tilted.The folds had a relatively uniform depth, but were deeper at the centre of each stack.No break in the plasma membrane was observed on neither the apical nor the basal side of the meisosomes.Despite their close apposition with mitochondria, no membrane continuity was observed between meisosomes and mitochondria (Figure 2A In order to evaluate not just the topology, but also the distribution of these organelles, we developed a fixation protocol for serial block-face scanning electron microscopy (SBF-SEM) of C. elegans samples.Starting with protocols previously described (Deerinck et al., 2010;D.H. Hall et al., 2012;D. H. Hall et al., 2012), we adapted the solvents and temperatures for each contrasting agent, including lead and uranyl acetate, to maximize sample contrast (see Materials and Methods).We acquired series of electron micrographs of the lateral epidermis as transversal views along 12 µm.We produced a voxel-based classification within the Waikato Environment for Knowledge Analysis (WEKA in Fiji) and then used its machine learning algorithms to perform semi-automated organelle recognition.This revealed that the meisosomes were irregularly spaced at the apical surface of the lateral epidermis, with various sizes and orientations and confirmed their frequent apposition to mitochondria (Figure 2-figure supplement 1).

VHA-5 is a marker of meisosomes
VHA-5, one of four α-subunits of the transmembrane V0 complex of the vacuolar ATPase (V-ATPase) (Oka et al., 2001;Pujol et al., 2001), and RAL-1, ortholog of human RALA (RAS like proto-oncogene A) (Frische et al., 2007), are the only known markers of meisosomes, with both proteins being also associated with multivesicular bodies (MVB) that play a role in exosome secretion (Hyenne et al., 2015;Liegeois et al., 2006).The expression pattern of VHA-5 is the better characterised of the two proteins.A TEM/immunogold staining study showed that more than 85 % of the VHA-5 signal can be attributed to meisosomes (Liegeois et al., 2006).
We have used several VHA-5 reporter strains, including one expressing a GFP-tagged version of VHA-5 from a CRISPR/Cas9 engineered genomic locus [KI], a single copy insertion under an epidermis promoter [Si], or classic multi-copy integrated [Is] or extrachromosomal [Ex] transgenic arrays.In all the strains, and in line with previous reports (Liegeois et al., 2006), we observed the same punctate fluorescence at the apical surface of the lateral epidermis, and in the ventral and dorsal ridges, from the head to the tail.As expected, it was almost completely absent from the dorsal and ventral epidermis above the body-wall muscles, and underneath the seam cells (Figure 3A-C).By combining a MUP-4::GFP (Suman et al., 2019) and a VHA-5::RFP reporter (Liegeois et al., 2006), we confirmed the complementary pattern in the epidermis of hemidesmosomes above the muscles, and meisosomes in the lateral epidermis and dorso/ventral ridges (Figure 3D).Higher resolution analysis in the lateral epidermis revealed the VHA-5-marked structures to have an irregular shape in no preferred direction in the lateral epidermis, consistent with the TEM and SBF analyses (Figure 3E).
Interestingly, in worms co-expressing VHA-5::RFP and either one of these membrane probes, we observed a high degree of co-localisation (Figure 3G&H & Figure 3-figure supplement 1D-F).This further reinforces the notion that VHA-5 is primarily a marker of subdomains of the plasma membrane.
The structures labelled with both VHA-5::GFP and CAAX::GFP or PH-PLCδ::GFP in the adult epidermis were similar in size and spatial distribution to the meisosomes reconstituted from the SBF data (Figure 2figure supplement 1D).To confirm that the observable fluorescence signal from VHA-5::GFP indeed primarily originated from meisosomes, we performed Correlation Light and Electron Microscopy (CLEM) using the VHA-5::GFP [Si] strain in which the strong and potentially confounding excretory canal GFP signal is absent, due to the use of an heterologous epidermis-specific promoter.As we used a different fixation technique to preserve the GFP signal (Johnson et al., 2015) and we worked on semi-thin section, meisosomes were revealed by electron tomography.After alignment of the confocal and TEM images, we could show that the fluorescence foci matched meisosomes (Figure 3I-J & Figure 3-figure supplement 2).Together with previous CLEM observations in the excretory duct (Kolotuev et al., 2009), these results indicate that the VHA-5::GFP signal that we observe at the apical membrane in the epidermis corresponds to meisosomes and that VHA-5::GFP can be used in vivo as a bona fide meisosome marker for this study.

Meisosomes align in between furrows before moulting
VHA-5 has been shown to have an essential role in alae formation and secretion of Hedgehog-related peptides through exocytosis via MVBs, but not to be involved in secretion of the collagen DPY-7, nor in meisosome morphology (Liegeois et al., 2006).Indeed, to date, no gene has been assigned a role in determining meisosome structure.As a path to understanding their function, we first observed their morphology during development.Consistent with previous reports (Liegeois et al., 2007), we observed that VHA-5::GFP aligns parallel to the actin fibres in animals entering the L4/adult moult, a stage we refer to here as "late L4".We refined this observation by precisely staging the worms throughout the L4 stage on the basis of vulval morphology and the shape of the lumen as previously described in (Cohen et al., 2020;Mok et al., 2015) (Figure 4A & 5).The parallel circumferential alignment of VHA-5::GFP could be observed at the beginning of the L4 stage, in L4.1 worms, but was then lost at the L4.2 stage.It reappeared progressively starting in the L4.3 stage, culminating between the L4.7 and L4.9 stages, just before the moult.This is consistent with the EM observations of meisosome alignment before moulting (Figure 1-figure supplement 2C-D).
As meisosomes, like the rest of the apical plasma membrane, are in direct contact with the aECM, the cuticle, we investigated the relation of meisosomes to different cuticle components.Different classes of cuticular collagen exist that form either the circumferential constricted furrows, or the cuticle in the regions between the furrows, called the annuli (Cox & Hirsh, 1985;Cox et al., 1980;McMahon et al., 2003;Page & Johnstone, 2007;Thein et al., 2003).As revealed with one marker of a furrow collagen, DPY-7::GFP (Miao et al., 2020), in combination with VHA-5::RFP, meisosomes align in between the furrows at the late L4 stage (Figures 4B).
Notably, during the L4.7 stage, some DPY-7::GFP can be observed in small vesicles on the apical side of the epidermis, that could represent the ongoing secretion of furrow collagen at that stage.Interestingly, these vesicles do not co-localise with VHA-5::RFP (Figure 4B, right panel).We further show that the CAAX and PH-PLC1δ markers that co-localise with VHA-5 in young adult animals (Figure 3G-H) also align during the L4 stage, but other vesicular components, like the one marked by HGRS-1, SNX-1 or LGG-1 do not (Figure 4C-D).Thus, meisosomes together with specific membrane subdomains align in between the furrow before moulting.

Furrow collagens determine the organisation of the cytoskeleton and meisosomes in L4 larvae
Before moulting, there is profound reorganisation of the cytoskeleton in the lateral epidermis.Microtubules and actin fibres align in a series of circumferential bands that are not present in adults (Castiglioni et al., 2020;Costa et al., 1997;Taffoni et al., 2020).Interestingly, meisosomes exhibited the same sequence of dynamic changes in alignment as microtubules and actin.After the moult, there was a concomitant loss of alignment of VHA-5, actin and microtubules, so that in wild-type adult animals, as described above, there was no clear overall pattern to the organisation of meisosomes, actin or microtubules (Figures 5).
We then examined the consequence of knocking down the expression of one of the furrow collagen genes, dpy-7, on the organised VHA-5::GFP pattern of late L4s worms.Compared to the control, RNAi of dpy-7 provoked a loss of meisosome alignment (Figures 5 and Figure 5-figure supplement 1).A similar phenotype was observed in dpy-3 mutant worms; DPY-3 is another furrow collagen (Figure 5-figure supplement 1B).
Strikingly, this loss of expression of furrow collagens was also associated with a disruption of the normal organized pattern of both actin fibres and microtubules from the L4.3 stage (Figure 5).It was previously proposed that the formation of actin fibres and microtubules in apposed circumferential bands plays an instructive role in positioning the furrows (Costa et al., 1997).Our results suggest, on the contrary, that furrow collagens in the cuticle govern the alignment of the underlying cytoskeleton as development progresses into the final moult.Thus, furrow collagens appear to be required to align both meisosomes and the actinmicrotubule cytoskeleton.

Abnormal meisosomes in adult furrow mutant worms
As previously mentioned, different classes of cuticular collagen exist that are expressed and form either furrows or annuli (Figure 6A) (Cox & Hirsh, 1985;Cox et al., 1980;McMahon et al., 2003;Page & Johnstone, 2007;Thein et al., 2003).While mutant in these collagens all have a Dumpy (i.e.short and fat; Dpy) phenotype, only the furrow-less mutants, in contrast to mutants of annuli collagens, exhibit an increased expression of the AMP reporter nlp-29p::GFP (Dodd et al., 2018;Pujol, Zugasti, et al., 2008;Zugasti et al., 2014;Zugasti et al., 2016).This is one reason that furrow collagens have been proposed to be part of a damage sensor that relays information about cuticle integrity to the epidermis (Dodd et al., 2018).
Interestingly, this reporter is also induced at the late L4 stage in the wild type before the last moult when the cuticle is reshaping (Figure 6-figure supplement 1A; (Miao et al., 2020)).
We examined the consequences of knocking down the expression of all these different collagen genes on the pattern of meisosomes in the adult.Collagen inactivation was always confirmed by observing the effect on body size, as well as the change in the expression of nlp-29p::GFP in parallel experiments (Figure 6-figure supplement 1B-C).Compared to control RNAi or to annuli collagen inactivation (dpy-4, dpy-5 and dpy-13), inactivation of all 6 furrow collagen genes (dpy-2, dpy-3, dpy-7, dpy-8, dpy-9 and dpy-10) provoked a marked and specific alteration in the pattern of VHA-5::GFP.The meisosomes' normal reticulated pattern was fragmented, as reflected by a decrease in their average size and Feret's diameter, and a >25 % increase in their density (Figures 6B-C).A similar fragmentation was observed with the different VHA-5 reporter strains, and either following inactivation of the furrow collagen gene's expression with RNAi or in null mutants (Figure 6-Source data file 1).
To test if the fragmentation was associated with a change in VHA-5's relation to other organelles, we inactivated furrow collagen genes in strains combining VHA-5::GFP and different mScarlet-tagged vesicular membrane markers, LGG-1 for autophagosomes, HGRS-1 for multivesicular bodies and SNX-1 for recycling endosomes.In contrast to VHA-5::GFP, we observed no marked alteration in their patterns, and there was still no overlap between the signal from VHA-5::GFP (Figure 6-figure supplement 2A).Further, neither dpy-3 nor dpy-7 inactivation had any effect on the size and density of the vesicular pattern of EEA-1::GFP (Figure 6figure supplement 2B), a marker of early endosomes (Shi et al., 2009).Thus, loss of furrow collagen gene expression leads to a substantial fragmentation of meisosomes, without affecting vesicular organelles in the epidermis.This suggests that furrow collagens play an important and specific role in maintaining meisosome integrity.

Furrow mutants, with small meisosomes, display a detached cuticle
The cuticle is connected, through the epidermis, to the underlying body-wall muscles via hemidesmosomes.
The maintenance of hemidesmosome integrity is vital; their complete loss causes a fully penetrant lethality.
On the other hand, partial loss of the hemidesmosome component MUA-3 causes the cuticle to detach above the muscles (Bercher et al., 2001).We have shown above that hemidesmosomes and meisosomes are present in complementary and non-overlapping regions of the epidermis (Figure 3B-D).This suggest that hemidesmosomes cannot play a role in the attachment of the lateral epidermis and the dorso/ventral ridge to the cuticle.As the meisosomes, containing numerous folds of the plasma membrane, increase the surface contact between the epidermis and the cuticle, we asked whether meisosome fragmentation could impact the attachment of the epidermis to the cuticle.Examination by TEM first confirmed that the meisosomes are significantly smaller in furrow collagen mutants compared with another Dpy mutant (dpy-13) or the wildtype, irrespective of the direction of the section, as both longitudinal and sagittal sections show the same phenotype (Figure 7A-D, Table 1 and Figure 1 for the wild type).In all furrow collagen mutants examined, there was a frequent disruption of the contact between the epidermal plasma membrane and the cuticle, either in the lateral epidermis or the dorso or ventral ridge, but not above the muscle quadrants (Figures 7B&G and Figure 7-figure supplement 1).This detachment is clearly distinct from what is observed in so-called Blister mutants where, due to the absence of the connective struts, the detachment happens between the 2 main layers of the cuticle (Page & Johnstone, 2007).To confirm the phenotype, we analysed by SBF entire transversal worm sections over a length of 21.5 and 34.4 µm, for a wild-type for a dpy-2 young adult mutant worm, respectively.This confirmed that the detachment between the cuticle and the epidermis was always found in the furrow collagen mutant outside the region of the body-wall muscles (Figures 7H).In furrow collagen mutants, the space between the cuticle and the underlying epidermal cell was often filled with a diverse range of cytoplasmic content, including membrane-bound vesicles with the appearance of endosomes, lysosomes, mitochondria, as well as electron-dense particles the size of ribosomes (Figures 8A).
To exclude the remote possibility that this detachment was an artefact linked to the different fixation protocols used for electron microscopy, we carried out live imaging on two independent strains in which the cuticle was labelled with a collagen tagged with mScarlet (ROL-6::mScarlet [KI]) and the epidermal plasma membrane was labelled with GFP::CAAX or GFP::PH-PLC1δ.Compared to the wild-type, where the GFP signal is restricted to heterogeneous macrodomains in the plasma membrane (Taffoni et al., 2020), in a dpy-3 furrow collagen mutant, the GFP was seen in numerous brightly-stained vesicular structures that accumulated outside the epidermis at the level of the mScarlet cuticular signal (Figure 8B).Together, these phenotypes suggest that the meisosomes may play an important role in attaching the cuticle to the underlying epidermal cell and that loss of this intimate connection causes a profound alteration of epidermal integrity.

Furrow mutants have abnormal biomechanical properties
We predicted that the changes in cuticle attachment seen in the furrow mutants would impact the biomechanical properties of worms.It was previously shown that furrows are stiffer than the rest of the cuticle in wild-type worms (Essmann et al., 2016).We therefore used atomic force microscopy to measure the resistance to force in wild-type and mutant worms, as previously described (Essmann et al., 2016;Essmann et al., 2020).While topographic AFM imaging (Figures 9A) provided further corroboration of the fact that in the absence of furrow collagens the cuticle has a disorganised aspect with irregular folds, lacking the usual repeated linear pattern of annuli and furrows, force spectroscopy AFM revealed differences in stiffness.In contrast to the non-furrow dpy-13 mutant that had a rigidity similar to wild-type, the different dpy furrow mutants (dpy-2, dpy-3, dpy-7 and dpy-8) exhibited markedly less steep force-indentation curves (Figure 9B), and hence lower stiffness or Young's Moduli (Figure 9C).This suggests that furrow collagens are required for normal stiffness.While lack of certain collagens in the cuticle could directly affect cuticle stiffness, we hypothesise that the effect on stiffness is a consequence of the fact that furrow collagens are essential for the presence of normal meisosomes.

Discussion
In this study, we undertook the characterisation of structures that link the nematode epidermis to the cuticle.
Across species, interfaces exist between flexible and dynamic cell membranes and more rigid extracellular matrices.Because of requirements for growth, signal transduction, and repair, the microstructures of the ECM need to be tightly linked to the plasma membrane and cytoskeleton of the underlying cell (Chebli et al., 2021).
In yeast, eisosomes are single membrane invaginations located under the cell wall that bridge this boundary and fulfil this function.They can disassemble in minutes to buffer changes in membrane tension, protecting cells from osmotic shock (Lemiere et al., 2021).Eisosomes are specific to yeast; there are no orthologues for core components, such as LSP-1, in animals.Conversely, the meisosomes that we describe here in C. elegans, with their multiple membrane invaginations that individually are similar in appearance to eisosomes, are, to the best of our knowledge, distinct from interfacial structures in non-nematode species.Interestingly, we show here that they are enriched in a PH-PLCδ marker, which is known to bind phosphatidylinositol 4,5bisphosphate (PIP 2 ) (Lemmon et al., 1995).PIP 2 has a major role in signal transduction and in regulating cellular processes including actin cytoskeleton and membrane dynamics (Katan & Cockcroft, 2020).Moreover, we have previously shown that the same PH-PLCδ marker rapidly reorganises upon wounding of the lateral epidermis (Taffoni et al., 2020).So, it is tempting to propose that analogous to eisosomes, meisosomes could have a role as a signalling platform in response to stress.
While the presence of meisosomes had been noted in earlier studies (Hyenne et al., 2015;Liegeois et al., 2006), we have been able to go beyond their previous characterisation, in part because of improvements in electron microscopy techniques.Specifically, we adapted the fixation protocol after high pressure freezing to have a better membrane contrast in serial block scanning electron microscopy, allowing semi-automated in silico segmentation of meisosomes.Moreover, adapting a CLEM protocol, we were able to match the VHA-5::GFP observed by fluorescence microscopy to meisosomes revealed by tomography.VHA-5, together with RAL-1, are currently the only known meisosome components.In contrast to the well-defined roles of these two proteins in alae formation and exosomes biogenesis (Hyenne et al., 2015;Liegeois et al., 2006), their function in meisosomes remains to be characterised.Notably, inactivation of ral-1 did not eliminate VHA-5::GFP fluorescence in the epidermis (Hyenne et al., 2015), and knocking down the expression of vha-5 did not affect the secretion of DPY-7 (Liegeois et al., 2006).This suggests that the V-ATPase on meisosomes is not involved in cuticle synthesis.Further study will be required to determine the catalogue of proteins that are needed for meisosome formation and maintenance.
Notably, a recent study reported the isolation of mutants with an abnormal pattern of VHA-5::RFP in the epidermis but attributed this to a change in MVBs (Shi et al., 2022), despite a lack of substantial colocalisation with HGRS-1, a well characterised MVBs marker, part of the ESCRT-0 complex that sorts endosomes to MVBs (Babst, 2011).Since previous studies (Liegeois et al., 2006), and the results presented here, show that VHA-5 is predominantly a marker of meisosomes, more so than of MVBs, we hypothesise that the one gene that Shi et al. characterised in detail, fln-2, which encodes the F-actin cross-linking protein filamin (Zhao et al., 2019), could actually be involved in the formation and/or maintenance of meisosomes.
Interestingly, a fln-2 loss of function mutation has been serendipitously found in several C. elegans strains originating from a different wild-type stocks (Zhao et al., 2019), so careful attention to genotypes will be needed in future work.Regardless, fln-2 may represent an important tool to investigate meisosome function.
Setting this issue aside, by taking an ultrastructural approach, we were able to build up a detailed picture of the organisation of meisosomes.One of their defining features is the constant 25 nm spacing of their constituent plasma membrane folds.This raises the question of how the membrane folds with such precision.
One possibility is that the striking electron-dense material that is apposed to each side of the membrane on the cytoplasm-facing folds, spaced less than 10 nm apart, will contain specific structural protein that maintain the uniform width of each meisosome fold, and influence their mechanical properties.These structures will require more precise characterisation.We equally have yet to establish whether the frequent proximity of meisosomes to mitochondria, with a close apposition of membranes, has a functional significance.
Contrary to the cuticle of many adult insects, the nematode cuticle is flexible enough to allow bending during locomotion.It is also thought to stretch to accommodate growth between moults.When the old cuticle is shed, it leaves in its place the new cuticle that had been moulded underneath it.The circumferential furrows of the new cuticle thus appear exactly in register with the position of old furrows.Before moulting, the cytoskeleton aligns in the apical epidermis, underneath and parallel to each furrow.Although this had been proposed to be important for positioning the furrows of the new cuticle (Costa et al., 1997;McMahon et al., 2003;Page & Johnstone, 2007), a recent study found unexpectedly, that actin is dispensable for the alignment of furrows (Katz et al., 2018).On the other hand, we found that the furrows are required for the alignment of actin fibres before the last moult.We propose therefore that only the old furrows are required to pattern the new furrows.
Consistent with such a model, the LPR-3 protein that is part of the transient pre-cuticle that is formed between the old and the new cuticles before each moult is absent from the region of the furrows (Forman-Rubinsky et al., 2017).We have shown that furrow determines the regular parallel and circumferential positioning of meisosomes.It is not yet clear whether this alignment of meisosomes is functionally important.It could result from steric constraints during moulting, in the limited space between nascent furrows of the new cuticle and the closely apposed circumferential actin fibres.It should, however, be noted that this alignment is not seen for vesicular organelles like MVBs, endosome or autophagosomes.
As adults, furrowless collagen mutants have fragmented meisosomes and a detached cuticle.We hypothesise that this fragmentation causes the detachment, and that the multiple folds of plasma membrane normally increase its contact surface with the cuticle thus ensuring a more robust connection of the aECM to the lateral epidermis.While the lateral epidermis is rich in meisosomes, it is devoid of hemidesmosomes.Conversely, in the dorsal and ventral quadrants, there are essentially no meisosomes, but abundant hemidesmosomes.These latter structures secure the muscles to the cuticle through epidermis and are indispensable for worm development and viability.Above the muscles, the epidermis is extremely thin, with the apical and basal plasma membranes juxtaposed, linked via intermediate filaments that bridge apical and basal hemidesmosome protein complexes (Zhang & Labouesse, 2010).MUA-3 is an hemidesmosome transmembrane protein in direct contact with the cuticle.In hypomorphic mua-3 mutants, large gaps form between the apical epidermal surface and the cuticle in the dorso-ventral quadrants, reflecting a loss of attachment of apical hemidesmosomes to the cuticle.Unlike the cytoplasm-filled gaps we observed in furrowless mutants, in mua-3(rh195) worms, these spaces appear devoid of contents, and the apical epidermal membrane is intact (Bercher et al., 2001).So, in contrast to the loss of hemidesmosomes, fragmentation of meisosomes in furrowless mutants affects the integrity of the apical epidermal membrane in the lateral epidermis, potentially explaining the permeability phenotype of furrowless mutants (Sandhu et al., 2021).Despite these differences, both meisosomes in the lateral epidermis, and hemidesmosomes in the dorso-ventral quadrants, do appear to have an analogous function, ensuring the attachment of the apical plasma membrane to the cuticle.
In animals, ECMs provide mechanical support for tissue assembly and organ shape.During embryogenesis in C. elegans, the aECM is essential during elongation as it not only maintains embryonic integrity, but also relays the mechanical stress produced by the actomyosin cytoskeleton and the muscles (T.T. Vuong-Brender et al., 2017; T. T. K. Vuong-Brender et al., 2017).In the adult, the mechanical properties of the aECM have only recently started to be explored.Atomic force microscopy revealed that the furrows have a higher stiffness than the annuli (Essmann et al., 2016).Here, we show that loss of specific furrow collagens, but not of non-furrow collagens, decreases stiffness.Part or all of this could be a direct consequence of the altered cuticle morphology, an analogy being the increased stiffness that corrugation provides.Furrow Dpy mutants are known to have a higher internal concentration of glycerol (Wheeler & Thomas, 2006), which will decrease their internal hydrostatic pressure.We propose that this decreased hydrostatic pressure is a consequence of the decrease in the stiffness of the cuticle.It would ensure the necessary balance of inward and outward pressures required for body integrity.Since we used a 10 micrometer diameter AFM probe to indent the worm, and the indentation depth was greater than the thickness of the cuticle (ca.800 nm compared to 500 nm for the cuticle), our measurements did not directly assess the cuticle stiffness, so further investigations will be needed to confirm our hypothesis.It is interesting to note, however, that a decrease in stiffness and an increase in the activity of innate immune signalling pathways in the epidermis are signatures of ageing in C. elegans (E et al., 2018;Essmann et al., 2020).How physiological and pathological modifications of the biomechanical properties of the aECM are surveyed by the epidermis remains an open question for future studies.

Transmission Electron Microscopy (TEM)
Day 1 adult worms were frozen in NaCl 50 mM medium containing 5 % of BSA and E. coli bacteria using Leica EM Pact 2 high pressure freezer.After freezing, samples were freeze-substituted at -90 °C in acetone containing 2 % OsO4 for 96 hours.The temperature was gradually increased to -60 °C and maintained for 8 hours.The temperature was then raised to -30 °C and maintained for 8 hours, before to be raised again to RT.
Samples were finally washed in acetone and embedded in Epoxy resin.Resin was polymerised at 60 °C for 48 hours.70 nm ultrathin and 350 nm semithin sections were performed using a Leica UC7 ultramicrotome and post-stained with 2 % uranyl acetate and Reynolds' lead citrate.Images were taken with a Tecnai G2 microscope (FEI) at 200 kV.For tomography acquisitions, tilted images (+60°/-60° according to a Saxton scheme) were acquired using Xplorer 3D (FEI) with a Veleta camera (Olympus, Japan).Tilted series alignment and tomography reconstruction was performed using IMOD (Mastronarde, 1997).

Freeze fracture
Wild-type adults were fixed in buffered 2.5 % glutaraldehyde, then cryoprotected in 30 % glycerol overnight prior to freezing.Fixed animals were positioned between two gold discs, and plunge frozen in liquid nitrogenchilled isopentane.Frozen worms were placed into a double replica holder for a Balzer's 301 freeze etch device.Samples were cleaved within the freeze etch device by free breaks, then shadowed with Pt/C to form a metal replica.Replicas were washed in bleach to remove all tissue prior to mounting on slot grids for examination by TEM.

Scanning electron microscopy by Serial Block Face (SBF)
After freezing in the aforementioned conditions, samples were incubated at -90 °C in acetone containing 2 % OsO4 for 106 hours.The temperature was gradually increased to 0 °C and samples were washed over 1h in acetone at RT.Samples were then incubated in acetone containing 0.1 % TCH for 60 min, washed over 1h in acetone, and incubated in acetone containing 2 % OsO4 for 1 hour at RT.After rehydration in ethanol decreasing graded series, samples were incubated ON in 1 % aqueous uranyl acetate at 4 °C and in 30 nM lead aspartate for 30 min at 60 °C.Samples were finally dehydrated in graded series of ethanol baths and pure acetone and embedded in Durcupan resin.Resin was polymerised at 60 °C for 48 hours.For the segmentation of meisosomes, regions of the lateral epidermis were acquired over a length of 12 µm, with a resolution of 10 nm par pixels.For scanning the cuticle detachment, entire transversal sections were acquired over a length of 21.5 and 34.4 µm, for a wild-type and a dpy-2 mutant worm, respectively, with a resolution of 10 nm par pixels.

Correlative Light Electron Microscopy (CLEM)
Sample for CLEM experiments were treated as in (Johnson et al., 2015).Briefly, the worms were high pressure frozen (EMPACT2, Leica) and then freeze-substituted (AFS2, Leica) for 20 hours from -130 °C to -45 °C in an acetone-based cocktail containing 0.2 % uranyl acetate, 0.1 % tannic acid and 5 % H 2 O.After 2 h of acetone washes at -45 °C, the samples were infiltrated with gradients of HM20 resin over 9 h, with pure resin for 18 h at -45 °C and the resin was polymerized under UV for 24 h at -45 °C and for 12 h at 0 °C.350 nm semithin sections were processed as described for TEM tomography above.TEM grids were first analysed at by confocal imaging where a bright field image is overlaid with the fluorescent image, then analysed in TEM at low magnification.Brightfield, GFP confocal and TEM images were aligned using AMIRA.Several positions with 2 or 3 GFP spots were chosen to do a high magnification tomography, as described above, to reveal the meisosomes.

Segmentations and 3D Image analysis
For electron tomography datasets, a binned version of the reconstructed tomogram was segmented using the Weka 3D segmentation plugin in Fiji/ImageJ to visualize the mitochondria and the meisosomes.The cuticle was visualized by the Amira-embedded Volume Rendering plugin from a manually segmented mask.A cropped area of interest of the full resolution electron tomogram was segmented in iLastik to visualize a representative portion of the organelle.For Serial Block-face datasets, the segmentation of the meisosome and mitochondria was generated using the Weka 3D segmentation plugin in Fiji/ImageJ.Animations and snapshots were generated in Amira.

RNA interference
RNAi bacterial clones were obtained from the Ahringer or Vidal libraries and verified by sequencing (Kamath et al., 2003;Rual et al., 2004).RNAi bacteria were seeded on NGM plates supplemented with 100 μg/ml ampicillin and 1 mM Isopropyl-β-D-thiogalactopyranoside (IPTG).Worms were transferred onto RNAi plates as L1 larvae and cultured at 20 °C or 25 °C until L4 or young adult stage.In all our experiments, we are using sta-1 as our control, as we have shown over the last decade that it does not affect the development nor any stress or innate response in the epidermis (Dierking et al., 2011;Lee et al., 2018;Taffoni et al., 2020;Zhang et al., 2021;Zugasti et al., 2014;Zugasti et al., 2016).

Fluorescent Image acquisition
Live young adult worms were placed on a 2 % agarose pad containing 0.25 mM levamisole in NaCl to immobilize the worms.Images were acquired using a confocal laser scanning microscopy: Zeiss LSM780 and its acquisition software Zen with a Plan-Apochromat 40 X/1.4Oil DIC M27 objective with a zoom 2 to 4, a Plan-Apochromat 63 X/1.40 Oil DIC M27 with a zoom 1. Spectral imaging combined with linear unmixing was used in most confocal images to separate the autofluorescence of the cuticle.

Airyscan super-resolution microscopy
Airyscan imaging were performed using a commercial Zeiss confocal microscope LSM 880 equipped with an Airyscan module (Carl Zeiss AG, Jena, Germany) and images were taken with a 63x/1.40NAM27 Plan Apochromat oil objective.In this mode, emission light was projected onto an array of 32 sensitive GaAsP detectors, arranged in a compound eye fashion.The Airyscan processing was done with Zen Black 2.3 software by performing filtering, deconvolution and pixel reassignment to improve SNR.The Airyscan filtering (Wiener filter associated with deconvolution) was set to the default filter setting of 6.1 in 2D.

Fluorescent Image analysis
To extract the morphological properties of meisosomes, we devised an automatic Fiji segmentation procedure (GitHub https://github.com/centuri-engineering/BD_BlobsSeg).We first restricted the analysis to manually drawn ROIs and isolated organelles (foreground image) from the background by using the "remove outliers" function of Fiji (radius = 30 pixels and threshold = 30).We next applied a Gaussian blur (sigma = 1 pixel) on the foreground image and automatically defined a threshold value to binarize the newly blurred image.This threshold was determined automatically by multiplying the background value (retrieved by averaging the fluorescent levels of the background image) by a constant coefficient.This allowed us to normalize the segmentation, since the expression levels of fluorescent proteins may vary from one animal to another.
Finally, after filtering out smaller objects (less than ~ 0.15 µm²), we measured the averaged organelles area, Feret's diameter (longest axis) and density in the different conditions.Unpaired t test was used to compare the samples which passed the normality test (Shapiro-Wilk normality test) and with homogeneity variances (Fisher test) and unpaired nonparametric Mann-Whitney test for the others.For co-localisation analysis, we counted the percentage of segmented objects in a given channel, green (G) or red (R), whose centroid is located in an object of the other channel.We then averaged these percentages across images, each representing a different worm (n=10 for each strain analysed).

Atomic Force Microscopy (AFM)
Worms were prepared as described before (Essmann et al., 2016).Briefly, staged 1-day-old young adult worms were paralysed in 15 mg/ml 2, 3-butanedione monoxime (Sigma) for 2 h at room temperature and transferred to a ~2 mm thick 4 % agarose bed in a petri dish (30 mm).Heads and tails were fixed with tissue glue (Dermabond, Ethicon) and the dish filled with a 2.5 ml M9 buffer.AFM data of worms were obtained using a NanoWizard3 (JPK) under aqueous conditions.Type qp-CONT-10 (0.1 N/m; nanosensors) cantilevers were used for imaging in contact-mode at setpoint 0.3 V and 0.5 Hz scanning speed, and NSC12 tipless cantilevers (7.5 N/m; MikroMash) with a 10 μm borosilicate bead attached (produced by sQUBE www.sQUBE.de)were used in force spectroscopy mode to obtain force-indentation measures at 450 nN force setpoint and 0.5 μm/s indentation speed.Cantilever sensitivity and stiffness (k) were calibrated using the JPK system calibration tool before each experiment.AFM raw data were analysed using the JPK analysis software.All force curves were processed to zero the baseline, to determine the tip-sample contact point and to subtract cantilever bending.The Young's Modulus was calculated within the software by fitting the Hertz/Sneddon model respecting the indenter shape (10 μm bead) to each curve.All topographical images are flattened using the plane fitting option of the JPK software at 1-2 degree to correct for sample tilt and natural curvature of the worm.Adjacent meisosomes in two serial thick (300 nm) sections were analysed with by electron tomography.(A) Selected virtual image from the serial reconstruction.(B-E) Segmentation of membranes and mitochondria reveal their 3D topology.Meisosomes (in yellow) are in close apposition to, but not in continuity with, mitochondria (orange) and are formed by epidermal plasma membrane folds, as observed in an en face view after removing the cuticle in silico (E).(F) Two folds were extracted and manually filled for a schematic view.Cuticle (cut), epidermis (epi), furrow (f), mitochondria (mit), meisosomes (m); scale bar, 200 nm.and red (R) objects (left panels), the G objects whose centroid is located in a R object are represented in white and counted as 1, the G objects whose centroid is located outside a R object are represented in green and counted as 0, the R objects whose centroid is located outside a G object are represented in magenta and counted as 0. The average % of all objects is represented, for each reverse situation, with each dot being an analysed ROI in one worm, n=10 for each strain.Confocal images of worms expressing VHA-5::GFP [Ex] (upper paired panels), TBB-2::GFP (middle paired panels), LIFEACT::GFP (lower paired panels) from early L4 to young adult (YA) stage, treated with the control (sta-1) or furrow Dpy (dpy-7) RNAi clones, n>4.Scale bar, 5 µm.To define the precise L4 stage, the vulva was observed and worms classified according to (Cohen et al., 2020).A representative example of the vulva at each stage is shown on the top row in worms expressing the marker TBB-2::GFP.Two serial tomograms were stitched and segmentations of the plasma membrane, mitochondria and cuticle were made (see material and methods for details).The continuity of the plasma membrane could be revealed by semi-automated segmentation, except under certain orientation due the high membrane curvature of the meisosomes and the missing wedge artefact, which is inherent to electron tomography.At the end of the video, the segmentation of two folds was completed manually to present a schematic view.

Functional
Figure 2S1 were performed at the New York Structural Biology Center, with help from KD Derr and William Rice.We thank John White and Jonathan Hodgkin for sharing the MRC/LMB archive of nematode micrographs.Some C. elegans strains were provided by the CGC, which is funded by NIH Office of Research Infrastructure Programs (P40 OD010440).We acknowledge the PICsL-FBI photonic microscopy facility of the CIML (ImagImm) and the PICsL-FBI electron microscopy facility of the IBDM, members of the national infrastructure France-BioImaging supported by the French National Research Agency (ANR-10-INBS-04).The project leading to this publication has received funding from France 2030, the French Government program managed by the French National Research Agency (ANR-16-CONV-0001) and from Excellence Initiative of Aix-Marseille University -A*MIDEX.

Figure legends Figure 1 :
Figure legends

Figure 1 -
Figure 1-figure supplement 1: Meisosomes are present in epidermis at all development stages.

Figure 1 -
Figure 1-figure supplement 3: Smaller meisosomes can be found over the muscle quadrants.

Figure 2 :
Figure 2: Meisosomes are formed by epidermal plasma membrane facing the apical ECM.

Figure 3 :
Figure 3: VHA-5 is a marker of plasma membrane containing meisosomes

Figure 5 :
Figure 5: Furrow collagen inactivation provokes a loss of meisosome and cytoskeleton alignment during the L4 stage.

Figure 5 -
Figure 5-figure supplement 1: Furrow collagen inactivation provokes a loss of meisosomes alignment during the L4 stage.

Figure 6 -
Figure 6-figure supplement 1: Worms increase AMP gene expression at the late L4 stage and in furrow collagen mutants.

Figure 6 -
Figure 6-figure supplement 2: Inactivation of furrow Dpy do not change the VHA-5 relationship to endosomes and MVB, nor affect early endosomes.

Figure 7 :
Figure 7: Furrow collagen inactivation leads to smaller and abnormal meisosomes and detachment of the cuticle.

Figure 7 -
Figure 7-figure supplement 1: Furrow collagen inactivation leads to detachment of the cuticle in lateral and ventral/dorsal ridges.

Figure 8 :
Figure 8: Furrow collagen inactivation provokes extrusion of membrane and cytoplasmic contents into the cuticle.

Figure 9 :
Figure 9: Furrow collagen inactivation provokes a reduction in stiffness of the cuticle.

Figure 2 -
Figure 2 -Video 1: Visualisation of the electron tomography and the 3D segmentation of meisosomes.

Figure 6 -
Figure 6 -Source data file 1: Quantification of the fragmentation of the meisosomes.Using different VHA-5 reporter strains in wild type and different collagen RNA inactivation or mutants, VHA-5 positive objects were segmented and three parameters were quantified, average size, Feret's diameter and density; the number of worms, total surface analysed per condition are presented, together with statistical analysis (see material and methods for details).

Table 1 : Quantification of the length of the meisosomes on TEM images in young adult wild type and different collagen mutants.
), dpy-7(e88) and dpy-8(e130) mutant worms, respectively.The orientation of the section, transversal or longitudinal, the number of different worms observed and the number of meisosomes analysed are reported.
Supplementary File 1: C. elegans strains used in this study.