Intermediate filament (IF) proteins, including nuclear lamins and cytoplasmic IF proteins, are essential cytoskeletal components of bilaterian cells. Despite their important role in protecting tissues against mechanical force, no cytoplasmic IF proteins have been convincingly identified in arthropods. Here we show that the ancestral cytoplasmic IF protein gene was lost in the entire panarthropod (onychophoran + tardigrade + arthropod) rather than arthropod lineage and that nuclear, lamin-derived proteins instead acquired new cytoplasmic roles at least three times independently in collembolans, copepods, and tardigrades. Transcriptomic and genomic data revealed three IF protein genes in the tardigrade Hypsibius dujardini, one of which (cytotardin) occurs exclusively in the cytoplasm of epidermal and foregut epithelia, where it forms belt-like filaments around each epithelial cell. These results suggest that a lamin derivative has been co-opted to enhance tissue stability in tardigrades, a function otherwise served by cytoplasmic IF proteins in all other bilaterians.

DOI: http://dx.doi.org/10.7554/eLife.11117.001

eLife digest

Different proteins exist to support the stability of animal cells. The intermediate filament proteins are an important example. One type – called lamins – stabilizes the nucleus (the structure within an animal cell that stores most of its DNA), while another forms scaffold-like structures in the rest of cell. The second type, referred to as “cytoplasmic” intermediate filaments, are not found in many hard-bodied creatures including insects and their closest relatives. This is probably because these animals, which are collectively known as arthropods, are instead supported by their tough external skeleton.

The soft-bodied animals called tardigrades (also known as water bears or moss piglets) are closely related to the arthropods. These microscopic animals can endure extreme environmental conditions such as freezing. The tardigrade’s endurance is likely to require some way to stabilize the animal’s cells. This might involve cytoplasmic intermediate filaments, but nothing was known about these proteins in tardigrades.

Now, Hering, Bouameur, Reichelt et al. have investigated if, and where, intermediate filaments are found in the cells of tardigrades. First, the complete set of active genes was analyzed for a species of tardigrade called Hypsibius dujardini; this revealed that three genes for intermediate filament proteins were active. Staining tissue slices or whole tardigrades with a marker that binds to intermediate filament proteins revealed that two of the three proteins were lamins and located within the nucleus. The third protein, which has been named "cytotardin", was found outside of the nucleus. However, unlike well-known cytoplasmic intermediate filaments, this protein did not form scaffold-like structures throughout the cell. Instead, cytotardin formed belt-like filaments that encircled each cell in the skin of the tardigrades.

Hering, Bouameur, Reichelt et al. then discovered that cytotardin seems to be more closely related to lamins than it is to cytoplasmic intermediate filaments. This suggests that cytotardin actually evolved from a tardigrade lamin and then acquired a new role in building filaments outside of the nucleus.

The fact that cytotardin is only found in the skin of the tardigrade and in those tissues that experience mechanical stress (for example, the mouth and legs) hints that it might help stabilize these cells. This could mean that the protein also helps these animals to resist extreme conditions. Further studies should focus on clarifying cytotardin’s role in stabilizing cells, in particular if it is required for the tardigrades' tolerance to environmental stress.

DOI: http://dx.doi.org/10.7554/eLife.11117.002

Main text


Tardigrades, also known as water bears, are microscopic invertebrates that live in marine, freshwater and semi-aquatic/limno-terrestrial environments (Kinchin, 1994; Nelson, 2002) (Figure 1). Although tardigrades have become renowned for their ability to survive extreme conditions (Møbjerg et al., 2011; Ramløv and Westh, 2001), including exposure to space (Persson et al., 2011; Rebecchi et al., 2011), only little is known about the actual mechanisms that allow their cells to resist severe mechanical stress caused by desiccation and freezing (Møbjerg et al., 2011; Ramløv and Westh, 2001; Wright, 2001; Tanaka et al., 2015; Yamaguchi et al., 2012). The integrity and plasticity of tardigrade tissues might be achieved by specialised cytoskeletal components, such as IF proteins, which are known to be essential for stress resilience of cells (Herrmann et al., 2009; Coulombe and Wong, 2004; Kim and Coulombe, 2007). While lamins, a group of IF proteins found in the nucleus, occur in most eukaryotes, including social amoebae (Krüger et al., 2012) and all metazoans (Dittmer and Misteli, 2011), cytoplasmic IF proteins are thought to have evolved from an ancestral lamin gene by duplication in the bilaterian lineage (Erber et al., 1999; Herrmann and Strelkov, 2011). Genomic and biochemical studies have revealed that the cytoplasmic IF proteins are present in all bilaterian taxa excluding arthropods (Bartnik and Weber, 1989; Goldstein and Gunawardena, 2000; Erber et al., 1998) [but see a contradictory report (Mencarelli et al., 2011) of a putative cytoplasmic IF protein in a collembolan]. The apparent loss of cytoplasmic IF proteins in the arthropod lineage might correlate with the acquisition of an exoskeleton (Herrmann and Strelkov, 2011; Goldstein and Gunawardena, 2000; Erber et al., 1998), which provides mechanical support to the arthropod skin. However, this hypothesis has never been tested, as it is unknown whether or not onychophorans and tardigrades, the soft-bodied relatives of arthropods, possess cytoplasmic IF proteins.

Figure 1.
Download figureOpen in new tabFigure 1. Light micrograph of a specimen of the tardigrade Hypsibius dujardini in dorsal view.

Anterior is left. Scale bar: 20 µm.

DOI: http://dx.doi.org/10.7554/eLife.11117.003

Results and discussion

To clarify whether or not onychophorans and tardigrades possess cytoplasmic IF proteins, we analysed Illumina-sequenced transcriptomes of five distantly related onychophoran species and the freshwater tardigrade H. dujardini. Although this model tardigrade (Gabriel et al., 2007; Gross et al., 2015) shows only a limited ability to tolerate desiccation (anhydrobiosis), it clearly survives immediate freezing (cryobiosis; see Video 1). Our analyses revealed only one putative IF protein transcript in each of the five onychophoran species but three in H. dujardini. Additional screening of the recently sequenced genome (Boothby et al., 2015; Koutsovoulos et al., 2015) of H. dujardini confirms that the identified transcripts correspond to the three potential IF protein-coding genes of this species. According to sequence comparisons with well-characterized IF proteins from humans and the nematode Caenorhabditis elegans, all corresponding proteins of the three identified genes share a similar α-helical rod domain organization with three coiled coil-forming segments (coil 1A, coil 1B, coil 2; Figure 2A,B and Figure 2—figure supplements 1 and 2) (review Chernyatina et al., 2015).This, in conjunction with the highly conserved intermediate filament consensus motifs (review Herrmann and Aebi, 2004) at the beginning and end of the rod domain of all three tardigrade proteins, classify them as intermediate filament proteins. The three tardigrade IF proteins further possess 42 residues in the coil 1B (Figure 2B and Figure 2—figure supplement 1) — a feature that is shared between all eukaryote lamins and protostome cytoplasmic IF proteins but that must have been deleted from the ancestral cytoplasmic IF protein gene in chordates (Herrmann et al., 2009; Peter and Stick, 2015). In contrast to the similar organization of the rod domain, the flanking sequences vary between all three tardigrade IF proteins. One of them, which we named lamin-2, possesses all major regions known from other eukaryote lamins (Herrmann et al., 2009; Dittmer and Misteli, 2011; Burke and Stewart, 2013), including the nuclear localization signal (NLS) — which generally mediates the import of proteins into the nucleus (Mical et al., 2004) — the lamin-tail domain (LTD), and the carboxy-terminal CaaX motif (Figure 2A, B). In contrast, the second tardigrade IF protein (named lamin-1) lacks the CaaX box, whereas the third IF protein from H. dujardini is missing all three motifs, including the NLS, indicating this protein may instead localize in the cytoplasm rather than the nuclear lamina of H. dujardini cells. Since the structure of the latter resembles the domain composition of known bilaterian cytoplasmic IF proteins (Figure 2—figure supplements 1 and 2) we consequently named it cytotardin.

Video 1. The tardigrade Hypsibius dujardini survives freezing.

This time-lapse video shows thawing specimens of the tardigrade H. dujardini after 4 days frozen in ice. One specimen starts with minuscule movements of one leg after 20 min of thawing and fully recovers locomotion within 120 min.

DOI: http://dx.doi.org/10.7554/eLife.11117.004

To firmly place the onychophoran and tardigrade IF homologs on the evolutionary tree, we reconstructed the phylogeny of broadly sampled metazoan lamin and cytoplasmic IF protein genes. In our phylogenetic analyses, all identified tardigrade and onychophoran sequences cluster within the bilaterian lamin clade, whereas none of them groups with cytoplasmic IF protein-coding genes (Figure 3 and Figure 3—figure supplements 1 and 2). Surprisingly, the three identified IF sequences of H. dujardini, together with two sequences from Milnesium tardigradum, form a strongly supported monophyletic clade of tardigrade lamins (GTR+G: Bootstrap support BS=77, LG+G: BS=80) in our analyses (Figure 3 and Figure 3—figure supplements 1 and 2). This implies that there were at least two duplication events in the tardigrade lineage that gave rise to lamin-1, lamin-2 and cytotardin genes — consequently characterizing the tardigrade IFs (including cytotardin), for example, as co-orthologous to nematode lamins rather than orthologous to nematode cytoplasmic IFs (Figure 3 and Figure 3—figure supplements 1 and 2). Our results further show that the isomin sequence of the collembolan Isotomurus maculatus does in fact cluster with other identified collembolan transcripts (GTR+G: BS=74, LG+G: BS=70) within the clade of arthropod lamins (Figure 3 and Figure 3—figure supplements 1 and 2); it had previously been interpreted (Mencarelli et al., 2011) as closely related to cytoplasmic IF proteins of nematodes and therefore as an ortholog of the bilaterian cytoplasmic IF proteins, likely due to the narrower dataset used for their phylogenetic analysis. These results clearly challenge the identity of isomin as a member of the bilaterian cytoplasmic IF protein clade (Mencarelli et al., 2011) and suggest that orthologs of genes encoding these proteins are entirely missing in arthropods, at least in those with known genomic sequences. In fact, besides the putative IF proteins from chelicerates, crustaceans and hexapods obtained from publicly available databases (e.g. GenBank), our transcriptomic and genomic analyses, which included screening of the genome of the centipede Strigamia maritima (see Chipman et al., 2014), the water flea Daphnia pulex (see Colbourne et al., 2011), and more than 70 transcriptomes from hexapod species sequenced as part of the 1KITE project (Misof et al., 2014), strongly suggest that these genes were already lost in the panarthropod lineage, since all of these panarthropod IF proteins cluster within a well-supported monophyletic clade of bilaterian lamins (GTR+G: BS=84, LG+G: BS=76; Figure 3—figure supplements 1 and 2). In this respect, even if the metazoan lamins are polyphyletic, as recently proposed by Kollmar, 2015 based on the finding of putative nematocilin homologs in Bilateria, our results favour the tardigrade, copepod, and collembolan IF proteins as members of the bilaterian lamins rather than the bilaterian cytoplasmic IFs or nematocilins (Figure 3—figure supplements 1 and 2).

To determine their subcellular localization and organization, we generated antisera against the three IF proteins of H. dujardini and confirmed their specificity (see Figure 4—figure supplement 1 and Figure 5—figure supplement 1). Immunolocalization of lamin-1, lamin-2 and cytotardin proteins in whole-mount preparations and cryosectioned specimens, in conjunction with confocal laser-scanning microscopy, revealed a highly specific subcellular distribution and tissue-restricted expression of these proteins in H. dujardini. As predicted, lamin-1 and lamin-2 proteins both display a typical, lamin-like distribution within all nuclei (Figure 4A–D and Figure 4—figure supplement 2). While lamin-1 is localised throughout the nucleoplasm (Figure 4A), lamin-2 is restricted to the nuclear periphery (Figure 4B), although there is a small overlap region between these two proteins (Figure 4C). The intranuclear distribution of lamin-1 corresponds to its lack of the CaaX motif, which is responsible for the association of lamins with the nuclear envelope (Kitten and Nigg, 1991). In contrast to the two lamins, cytotardin of H. dujardini is not localised within the nucleus, but in the peripheral cytoplasm of all epidermal and foregut cells, where it appears to be closely associated or aligned with the plasma membrane (Figure 4E–G and Figure 4—figure supplement 3 and Figure 6—figure supplement 1). In this setting, it encircles the apical regions of the cells in a belt-like, filamentous array (Figure 4E–G and Figure 4—figure supplement 3).

In order to investigate the filament-forming capacity of cytotardin, we transiently transfected human mammary epithelial MCF-7 cells with a corresponding cDNA (Figure 5A–F and Figure 5—figure supplement 1). Immunofluorescence analyses using cytotardin antibody revealed that this protein is located exclusively in the cytoplasm of transfected MCF-7 cells, where it forms both short filaments and extensive cytoskeletal networks which most likely are homopolymeric (Figure 5A–F). Notably, some of the cytotardin arrays display cage-like perinuclear structures, while others are located in the periphery close to the cell membrane (Figure 5E,F). Double labelling for cytotardin and desmoplakin, a desmosomal protein mediating membrane attachment of mammalian IF proteins (Simpson et al., 2011), shows that cytotardin occurs close to desmosomes but is not co-localised with desmoplakin (Figure 5C,D). To examine whether this arrangement was mediated by interactions between cytotardin and keratins endogenously expressed in MCF-7 cells, we double-labelled the transfectants for cytotardin and keratin-8. Our data show that these two proteins are not co-localised and that the endogenous keratin networks are displaced in cells with dense cytotardin arrays (Figure 5E,F). These findings strongly support an intrinsic ability of the cytotardin protein of H. dujardini to both form homopolymeric filaments and cytoplasmic networks — both properties that are functionally analogous to mammalian cytoplasmic IFs (Bohnekamp et al., 2015).

Our data on tissue-specific distribution of cytotardin in H. dujardini further show that its belt-like arrays are confined to the ectodermal epithelia, including the epidermis, buccal tube, pharynx, and oesophagus (Figures 4E–G and 6A,B and Figure 4—figure supplement 3 and Figure 6—figure supplement 1). Thus, the entire tardigrade body is ensheathed by a grid of belt-like filaments formed by the cytotardin protein, which retain their integrity even in contracted specimens (Figure 4E–G and Figure 4—figure supplement 3 and Figure 6—figure supplement 1). The most prominent anti-cytotardin immunoreactivity is found in areas exposed to considerable physical stress in living specimens, including the bases of claws and the stylet apparatus (Figures 4F,G and 6B and Figure 4—figure supplement 3 and Figure 6—figure supplement 1). We therefore anticipate that much of the resistance of the tardigrade body to extreme conditions, such as cryobiosis (Wright, 2001; Hengherr et al., 2009), might be attributable to the dense, fibre-like cytotardin meshwork. The belt-like structures encircling each epidermal cell might help to resist the shearing forces that arise during freezing and thawing cycles, whereas the dense meshwork at the basis of each claw and around the stylets might provide the tissue stability necessary for locomotion and feeding.

Together, our data demonstrate that cytotardin of H. dujardini is a cytoplasmic IF protein, predominantly expressed in ectodermal epithelia, that has evolved independently from the cytoplasmic IF proteins of other bilaterians by gene duplication and subsequent neofunctionalization (Figure 6C). This process was most likely triggered by an initial loss of the CaaX motif and nuclear localization signal from the ancestral lamin gene in the tardigrade lineage. Expression of the cytotardin protein in tissues that are subject to mechanical stress might have served as a pre-adaptation for the ability of tardigrades to survive extreme environmental conditions. Similar duplication and neofunctionalization events might have occurred in the collembolan and copepod lineages, as our findings demonstrate that these two taxa also show duplicated lamins that have lost their nuclear localization signals (Figures 3 and 6C and Figure 3—figure supplements 1 and 2). While isomin might stabilize the intestinal epithelial cells in collembolans (Mencarelli et al., 2011), the localization and function of the putative lamin-derived cytoplasmic IF protein in copepods has yet to be clarified.

Materials and methods

Culturing of specimens, transcriptomics, genomics and phylogenetic analyses

Specimens of Hypsibius dujardini (Doyère, 1840) were purchased from Sciento (Manchester, UK) and cultured as described by Mayer et al., 2015. The Illumina-sequenced transcriptomes of H. dujardini and five onychophoran species from Hering et al., 2012 and Hering and Mayer, 2014 were screened for expressed intermediate filament genes, including lamins, by BLAST searches (Altschul et al., 1997) with known metazoan lamins and cytoplasmic IF genes as bait sequences. Three putative IF protein genes from H. dujardini and one from each of the onychophoran species studied were verified afterwards by reciprocal BLAST searches against the nr database of GenBank (http://blast.ncbi.nlm.nih.gov/Blast.cgi). In addition, BLAST searches of the putative tardigrade intermediate filament genes in the genome of H. dujardini (http://badger.bio.ed.ac.uk/H_dujardini/; Koutsovoulos et al., 2015) yielded nearly identical sequences as obtained in our transcriptomic data, although the automatically predicted transcripts from the genome seem to be erroneous as previously reported by Hering and Mayer, 2014 and had to be corrected manually. Furthermore, publicly available resources and databases [nr, TSA and EST databases of GenBank, Compagen (Hemmrich and Bosch, 2008)] as well as the genome of the centipede Strigamia maritima (see Chipman et al., 2014), the water flea Daphnia pulex (see Colbourne et al., 2011), and more than 70 transcriptomes from hexapod species sequenced as part of the 1KITE project (Misof et al., 2014) were comprehensively screened for putative eukaryotic intermediate filament genes. In total, 447 eukaryotic lamin and cytoplasmic IF genes were manually curated and selected for further analyses. Each of the sequences was verified by screening for the presence of a coiled-coil-forming domain (Pfam PF00038), a typical feature of intermediate filament proteins, and the presence or absence of a lamin tail domain (LTD; Pfam PF00932, SCOP d1ifra_, PDB 2lll/2kpw/1ufg/1ivt) using the webserver of Pfam 27.0 (Finn et al., 2014) and SMART (Schultz et al., 1998; Letunic et al., 2015), respectively. In addition, the occurrence of a nuclear localization signal (NLS motif) was predicted for all sequences by using the webserver of NucPred (Brameier et al., 2007) and cNLS Mapper (Kosugi et al., 2009) (cut-off score = 2.0). The protein structures (α-helices and β-sheets) were predicted for the tardigrade lamin-1, lamin-2 and cytotardin using Jpred3 and JNetPRED (Cole et al., 2008). For phylogenetic analyses, the rod domains of all sequences were aligned using the Mafft online version v7.245 (Katoh and Standley, 2013) with the most accurate option L-INS-i and default parameters. To remove homoplastic and random-like positions, the alignment was afterwards masked with the software Noisy rel. 1.15.12 (Dress et al., 2008) (-cutoff=0.8, -seqtype=P, -shuffles=20,000). Two Maximum likelihood analyses were conducted with the Pthreads version of RAxML v8.1.15 (Stamatakis, 2014). For each run, the best tree was obtained from 10 independent inferences and GAMMA correction of the final tree under either the empirical LG substitution model or a dataset-specific GTR substitution matrix. The LG model was automatically selected by RAxML (PROTGAMMAAUTO option) as best-fitting substitution model, which is in line with the best-fitting model obtained with ProtTest v3.4.1 (Darriba et al., 2011) (LG+G) according to the Akaike information criterion (Akaike, 1974) (AIC), Bayesian Information Criterion (Schwarz, 1978) (BIC), corrected AIC (Sugiura, 1978; Hurvich and Tsai, 1989), and Decision Theory Criterion (Minin et al., 2003) (DT). Notably, the analysis using the dataset-specific GTR+G model yielded a better log likelihood score (−174,216.74) for the best tree than using the best obtained empirical model LG+G (−178,092.26). Bootstrap support values for both trees were calculated using the rapid bootstrapping algorithm implemented in RAxML from 1,000 pseudoreplicates. The protein domain structures of the sequences analysed were mapped on the trees using iTol v2 (Letunic and Bork, 2011).


Total RNA from several hundred specimens of H. dujardini was extracted and purified using TRIzol Reagent (Life Technologies, Carlsbad, CA) and RNeasy MinElute Cleanup Kit (Qiagen, Hilden, Germany) according to the manufacturers’ protocols. First strand cDNA synthesis was performed using random hexamer primer and SuperScript II Reverse Transcriptase (Life Technologies) and afterwards used as template for amplification of the whole coding sequence (CDS) of lamin-1, lamin-2 and cytotardin using gene specific primers. The cytotardin specific primers contained restriction sites that were required for subsequent cloning into bacterial or mammalian expression vectors (see Figure 5—figure supplement 2). The amplicons of lamin-1 and lamin-2 were cloned into the pGEM-T Vector System (Promega, Madison, WI) to generate the plasmids pGEM-T-lamin-1 and pGEM-T-lamin-2. Cytotardin CDS was cloned into expression vectors pET15b (Novagen Merck Millipore, Darmstadt, Germany), pcDNA3.1/Zeo (+) (Life Technologies) and pEGFP-C3 (Clontech Laboratories, Mountain View, CA) to generate the following plasmids: pET15b-cytotardin, pcDNA3-cytotardin, pcDNA3-HA-cytotardin and pEGFP-cytotardin. All PCR-amplified constructs were verified by Sanger sequencing and have been deposited in GenBank (http://www.ncbi.nlm.nih.gov/genbank) under accession numbers KU295460–KU295467.

Antibody generation

Polyclonal antibodies against HPLC-purified synthetic peptides (C-terminal) of lamin-1, lamin-2 and cytotardin of H. dujardini were newly generated, following coupling to KLH (key limpet hemocyanin) (Peptide Specialty Laboratories GmbH, Heidelberg, Germany). Anti-lamin-1 (antigen: SNLDIHNDSVRDSPRSAG-C) and anti-cytotardin (antigen: EQKITETFKASGRVGPRTDW-C) were purified from sera of immunised guinea pigs and anti-lamin-2 (antigen: REMTQSSTRDDSYLGPSGLPKR-C) from the serum of immunised rabbits (Peptide Specialty Laboratories GmbH).

Immunolocalization in whole-mount preparations and cryosections

For cryosectioning, specimens of H. dujardini were concentrated by filtering the culture medium through a polyamide mesh (pore size: 30 µm). The concentrated specimens were transferred into an embedding medium for cryosectioning (Tissue-Tek O.C.T Compound; Sakura Finetek, Staufen, Germany). Single drops of the medium containing the tardigrades were immediately frozen in dry ice-cooled 2-methylbutane. The frozen drops were cryosectioned into 5 µm thick sections. The sections were attached to SuperFrost Plus slides (Menzel, Braunschweig, Germany), then dried at room temperature and stored at −80°C. The ice-cooled cryosections were fixed on slides with a 4% solution of formalin (FA) freshly prepared from paraformaldehyde in phosphate-buffered saline (PBS; 0.1 M, pH 7.4) for 15 min. The sections were then shortly washed with an ammonium chloride solution (50 mM in PBS) and rinsed in Tris-buffered saline (TBS; 0.01 M, pH 7.6) two times for 3 min each. The anti-lamin-1 serum was applied to cryosections at a concentration of 1.9 µg/mL in TBS containing 1% bovine serum albumin (BSA). Equally applied were the anti-lamin-2 serum with a concentration of 1.8 µg/mL and the anti-cytotardin serum with a concentration of 2.1 µg/mL. The incubation with the primary antibody solution was performed at room temperature for one hour. After two 3-min washing steps in TBS, a secondary antibody, either donkey anti-guinea pig Alexa Fluor 488 (3 µg/mL in TBS with 1% BSA; Jackson ImmunoResearch Laboratories, Hamburg, Germany) for anti-lamin-1 and anti-cytotardin or goat anti-rabbit Alexa Fluor 568 (4 µg/mL in TBS with 1% BSA; Molecular Probes, Darmstadt, Germany) for anti-lamin-2, was applied at room temperature for one hour. The sections were then washed in TBS two times for 3 min and shortly rinsed with distilled water and afterwards with ethanol (100%).

For whole-mount immunohistochemistry, specimens were first concentrated as described above and then pipetted into a 1 mL Eppendorf tube. After carefully removing excess water, the specimens were flash-frozen by placing each tube on dry ice. Frozen specimens were fixed immediately by applying a 4% solution of formaldehyde (FA) in PBS with 1% dimethyl sulfoxide (DMSO). The fixative was applied at room temperature overnight. The whole-mounts were then washed in PBS two times for 10 min, two times for 30 min and two times for an hour. The specimens were cleared by dehydration in an ascending ethanol series (70%, 90%, 95%, 100%, 100% ), applying xylene as a clearing agent two times for 3 min and rehydration in a descending ethanol series (100%, 100%, 90%, 70%, 50% ). The whole-mount preparations were then washed at 37°C in PBS two times for 10 min. A mixture of collagenase/dispase (each 1 mg/mL; Roche Diagnostics, Mannheim, Germany) and hyaluronidase (1 mg/mL; Sigma-Aldrich, Munich, Germany) diluted in PBS was applied at 37°C for 10 min. After the incubation with enzymes, a 15-min post-fixation with 4% FA in PBS followed at room temperature. The specimens were washed in an ammonium chloride solution (50 mM) two times for 15 min, afterwards in PBS with 1% Triton X-100 (Sigma-Aldrich) two times for 10 min, two times for 30 min, once for an hour, overnight and then two times for 1 min. The whole-mount preparations were incubated in a blocking solution containing 10% normal goat serum (NGS; Sigma-Aldrich), 1% Triton X-100 and 1% DMSO in PBS at room temperature for one hour. The anti-cytotardin serum with a concentration of 4.2 µg/mL in PBS with 1% NGS, 1% DMSO and 0.02% sodium azide was applied at room temperature for three days. After washing in PBS with 1% Triton X-100 and 1% DMSO three times for 5 min, two times for 15 min, four times for two hours, overnight and two times for 15 min a secondary antibody (donkey anti-guinea pig Alexa Fluor 488; Jackson ImmunoResearch Laboratories) with a concentration of 3 µg/mL in PBS with 1% NGS, 1% DMSO and 0.02% sodium azide was applied at room temperature for three days. Finally the specimens were washed in PBS with 1% Triton X-100 and 1% DMSO three times for 5 min, three times for 15 min and two times for one hour and afterwards in PBS (without Triton X-100 and DMSO) four times for 15 min.

Either DAPI (1 µg/mL; Carl Roth, Karlsruhe, Germany), SYBRGreen I (diluted 1:10,000; Molecular Probes), propidium iodide (0.5 µg/mL; Carl Roth) or TO-PRO-3 iodide (diluted 1:1,000; Molecular Probes) were used as nuclear counterstain markers. Each of these fluorescent dyes was simply added to the solution containing the secondary antibody. Cryosections and whole-mount preparations were mounted in Prolong Gold antifade reagent (Molecular Probes). The whole-mount preparations were mounted between two cover slips using petrolatum at all corners as spacer. All slides and cover slips were sealed at the edges with transparent nail polish for long time storage. Specimens were analysed with the confocal laser-scanning microscopes Zeiss LSM 780 (Carl Zeiss Microscopy, Göttingen, Germany) and Leica TCS STED (Leica Microsystems CMS, Wetzlar, Germany).

Recombinant protein expression of cytotardin in bacteria

Escherichia coli BL21 (DE3) were transformed with pET15b or pET15b-cytotardin vectors. Expression of recombinant protein was induced with 0.1 mM IPTG for three hours in Luria Bertani (LB) medium. Bacterial suspensions were centrifuged at 3,000 g for 10 min at 4°C. Bacterial pellets were then washed twice in PBS and lysed in 5x Laemmli buffer (Volume in µL = OD600/10xVolume of culture in mL) containing protease inhibitor (#78439; Thermo Scientific, Waltham, MA), heated at 95°C for 10 min, briefly sonicated and heated again. Lysate of cells transformed with pET15b-cytotardin was diluted at 1:50. Recombinant cytotardin protein expression was first assessed by Coomassie blue stained SDS-PAGE and further by Western blotting.

Transfection and immunolabelling of cytotardin in MCF-7 cell culture

Cells from the human cell line MCF-7 (ATCC HTB-22) were cultured in Dulbecco’s modified Eagle’s medium (DMEM, GE Healthcare, Little Chalfont, UK) supplemented with 10% foetal bovine serum (FBS, GE Healthcare), 100 U/mL penicillin, and 100 μg/mL streptomycin. Cells were transfected with pcDNA-cytotardin, pcDNA-HA-cytotardin, pEGFP-cytotardin and the corresponding empty vectors using Xfect Transfection reagent (Clontech Laboratories) according to the manufacturer’s protocol. For immunofluorescence microscopy, cells were grown on glass slides and fixed 24 hr after transfection. MCF-7 cells were washed twice with PBS and fixed either in methanol or in 4% FA freshly prepared from paraformaldehyde in PBS. For methanol fixation, cells grown on a coverslip were put in ice-cold methanol for 10 min, acetone for 1 min and dried at room temperature for 30 min. For FA fixation, cells were put in a 4% FA solution in PBS for 15 min at room temperature. After fixation, cells were washed in PBS and used for immunolabelling. FA-fixed cells were first permeabilised in PBS 0.1% Triton X-100 for 5 min and blocked in PBS 5% bovine serum albumin for 30 min. Methanol- and FA-fixed cells were then incubated overnight at 4°C with primary antibody solutions, washed three times, incubated in secondary antibodies + DAPI (Sigma-Aldrich) solution for one hour, washed two times in PBS, once in water, once in 100% ethanol, dried for 30 min and mounted in mounting medium (Dianova, Hamburg, Germany). Primary antibodies were used at the following dilutions: guinea pig anti-cytotardin, 1:1,000; mouse anti-desmoplakin (11-5F), 1:150; mouse anti-keratin-8 (Ks 8.7), 1:100; mouse anti-HA-tag (MMS-101P; Covance, Princeton, NJ), 1:100. Images were taken using a confocal laser-scanning microscope (LSM 780; Carl Zeiss Microscopy) with 63×/1.46 NA oil immersion objective and Z-stack images were assembled by 'Maximum Intensity Projection' of the ZEN 2012 software (Carl Zeiss Microscopy).

Western blotting and specificity tests

Western blots were performed either on lysates of transfected MCF-7 cells or multiple specimens of H. dujardini. Transfected MCF-7 cells grown in 10 cm2 dishes were lysed in 200 µL 5x Laemmli sample buffer, boiled for 10 min at 98°C and briefly centrifuged prior to use. Concentrated specimens of H. dujardini were lysed in 800 µL 5x Laemmli buffer containing protease inhibitor, boiled for 10 min, sonicated and boiled again for 10 min. SDS-PAGE and Western blotting were performed as described previously (Löffek et al., 2010). Primary antibodies, diluted in Tris-buffered saline containing 0.05% Tween 20 (A4974; AppliChem, Darmstadt, Germany), were used at the following concentrations: anti-lamin-1, 1:10,000; anti-lamin-2, 1:5,000; anti-cytotardin, 1:10,000, anti-HA tag (MMS-101P; Covance), 1:1,000 and anti-GFP (sc-5385; Santa Cruz Biotechnology, Dallas, TX), 1:1,000. Anti-lamin-1 and anti-lamin-2 antibodies were only tested in Western blots based on lysates of tardigrade specimens.



We thank Sandra Treffkorn and Vladimir Gross for help with tardigrade culture, Fanny Loschke for assistance with confocal microscopy, Susann Kauschke for the light micrograph in Figure 1, Miriam Richter and Gabi Baumbach for technical assistance with cloning and cell culture, and Vladimir Gross, Jens Bohnekamp and Matthias Behr for critical discussions. We are grateful to members of Luca Borradori's laboratory (Bern, Switzerland) for providing the pet15b vector. This work was partially supported by the German Research Foundation (DFG; INF 268/230-1; MA1316/21-1) to TMM and the Emmy Noether Programme of the DFG (Ma 4147/3-1) to GM. The authors declare no conflicts of interests.

Decision letter

Harald Herrmann, Reviewing editor, Deutsches Krebsforschungszentrum, Germany

In the interests of transparency, eLife includes the editorial decision letter and accompanying author responses. A lightly edited version of the letter sent to the authors after peer review is shown, indicating the most substantive concerns; minor comments are not usually included.

Thank you for submitting your work entitled "Novel origin of lamin-derived cytoplasmic intermediate filaments in tardigrades" for consideration by eLife. Your article has been reviewed by four peer reviewers, and the evaluation has been overseen by a Guest Reviewing Editor (who also served as one of the reviewers) and Randy Schekman as the Senior Editor.

The following individuals involved in review of your submission have agreed to reveal their identity: Michael Schaffeld (peer reviewer 2) and Reimer Stick (peer reviewer 3).

The reviewers have discussed the reviews with one another and the Reviewing editor has drafted this decision to help you prepare a revised submission.

General assessment:

All four reviewers found the article very interesting, i.e. intermediate filament (IF) proteins in evolution. Here is a comprehensive phylogenetic analysis about their appearance in the water bear (Tardigrada) as an example of a very simple and fascinating "micro-animal" and how their sequences fit into the complex evolutionary tree of IF proteins.

Central conclusions:

1) The water bear Hypsibius dujardini genome exhibits three coding units for potential intermediate filament (IF) proteins, two with a nuclear localization signal sequence, one without.

2) A potential structural fold is introduced and the proteins are grouped – according to their primary sequence – into an evolutionary tree.

3) The proteins are called, because of sequence homology with lamins, lamin-1 to -3. Lamin-1 and lamin-2 exhibit nuclear localization signals and an Ig fold-like domain. Lamin-2 does not have a CAAX box for isoprenylation and hence resembles a lamin C of vertebrates.

4) Lamin-1 and lamin-2 are localized at the nuclear envelope in authentic Hypsibius dujardini tissue sections, lamin-3 is found in epithelial cells. It is essential to change the name lamin-3, since naming a cytoplasmic protein "lamin" will confuse readers tremendously. For instance, Hd-IF-1 would make sense – provided the protein is demonstrated to be really IF-like by sequence criteria.

5) Transfection of lamin-3 cDNA into human epithelial MCF-7 cells revealed that these proteins were able to form filaments that organized into extensive cytoplasmic networks.

6) The cytoplasmic IF protein of Hypsibius dujardini is most probably, like the cytoplasmic IF protein of Caenorhabditis elegans, derived from the nuclear version by loss of the nuclear localization signal.

Required revisions:

The reviewers raise a number of concerns that should be appropriately considered before the paper can be accepted. Those comments concerning the sequences used to construct the "tree" may cause some extra work, but the remainder of the comments concern the improvement of the text, in particular the background of the research.

1) The amino acid sequence of the α-helical rod provides a good starting point to characterize potential intermediate filament (IF) proteins by comparing them to authentic vertebrate ones. As the conserved end domains are evolutionary signatures with functional roles, those segments should be looked at in particular. Hence, an amino acid comparison of lamin-1 to lamin-2 to lamin-3 with each other and with e.g. human, C. elegans and Hydra or a protist lamin would be a meaningful way to analyze these new proteins before drawing domain schemes.

2) With respect to the last comment, the work by Martin Kollmar (Scientific Reports 5:10652) has to be discussed appropriately.

3) The fact that cytoplasmic IF proteins (which are essentially not lamins) are derived from lamins is not so new as one would learn from the work over the last two decades of the Weber group in Göttingen. How can one leave out this worm in light of its close relatedness to water bears?

4) A figure with an evolutionary "tree" that can be printed on one page would be good – as it would only allow giving main relationships. The Weber group did such overviews in their papers.

5) The issue of cytoplasmic IFs – as well as isomin – may be discussed with a broader perspective. In the light of the million insect species, Drosophila is surely not "the" insect.

6) The sequences used to build the tree should be "cleaned".

7) IF proteins form – with very rare exceptions (Keratin 18) – parallel coiled-coil dimers. Meaningful molecular modeling should consider that fact.

Full Reviews

Reviewer #1:

Lars Hering and colleagues report on the occurrence of both nuclear and cytoplasmic intermediate filament (IF) proteins in Tardigrades, the water bear. Based on in silico work done for Hypsibius dujardini they derive two amino acid sequences that exhibit similarity to lamins and one that is cytoplasmic. As tardigrades are very special organisms, tolerating various extreme stresses, they are of high interest for every life scientist. These data they use to investigate evolutionary relationships. The evolutionary trees provided are interesting, but the definition of the proteins is poor (see below). But immuno-cytochemical methods the authors demonstrate that the two nuclear IF proteins localize to the nuclear envelope. Moreover, these authors use cDNAs to transfect human epithelial cells and demonstrate that the cytoplasmic lamin-3 (what a misnomer) assembles into some type of regular network. All the cell biological data are very interesting and of high quality.

For a reader following the DNA-RNA-protein dogma for the last decades, a protein is still defined by its sequence, and not its position in a phylogenetic tree. Old fashioned as it may be, it would be good to discuss such issues as domain size, coiled-coil forming ability, i.e. the occurrence of heptad or hendcad patterns. The sequences are, however, not shown neither are they directly compared with authentic lamins such as those from Hydra, man or C. elegans. Strange enough, the latter organism is not centrally discussed, as for this worm Klaus Weber and his group in Göttingen demonstrated how lamins and invertebrate cytoplasmic intermediate filament proteins are related, i.e. by loss of the nuclear localization signal (NLS) the cytoplasmic IF proteins emerged (Dodemont et al. 1990 & 1994 EMBO J., Andreas Erber et al. 1998 & J. Mol. Evol., and several more papers). Obviously a fruitful strategy.

A selected paradigmatic-type comparison would be very instructive. I took the CAAX-box-free version and compared its sequence with human lamin A: 1. Coil 1B is missing 4 amino acids and coil 2 carries a 3-amino acid insertion. The size of these helical segments including coil 1A is highly conserved, with the rare exception of coil 2 of the single lamin of C. elegans, which is missing 2 heptads or 14 amino acids. Notably, the IF consensus motif in coil 1A exhibits high conservation as does the IF consensus motif at the end of the rod in coil 2. More surprisingly, the linker segment L1 is entirely kept both to the very amino acid number and with respect to the quality of amino acids, which all are good for α-helix formation; the same holds true for linker L12 from what one can conclude that structural principals in the α-helical rod domain are highly conserved. Figure 1 does not give any clue to how related these domains are. In Figure 1C the subdivision in 2A and 2B is completely illusory, it is wrong. The authors should refer to a recent structural review of Sergei Strelkov (Leuven) in Current Opinion in Cell Biology (2015). "Head" and "tail" sequences vary very much in IF proteins – notably here: the Ig fold sequence in the carboxyl-terminal domain exhibits highest identity with that of human lamin A, when the whole amino acid sequence is considered. The NLS shows an interesting variation from the standard, both CDK1 sequences flanking the rod are there but slightly shifted compared with the position of lamin A. In summary, Figure 1C gives the impression that everything is like in every IF protein and that lamin-1 to -3 are identical: to what extent are they? The tree in Figure 1A does not tell anything in detail. So how different is the tardigrade cytoplasmic IF protein from its lamins, and how closely is it related to human keratins K8 and K18? Does it have features that are anything like "keratin-related"? Along these lines: the structural models shown as Figure 1—figure supplement 3 are pretty meaningless – all three lamin-1 to -3 should form coiled coils.

Additional data files and statistical comments:

The amino acid sequences and the kind how coiled-coil forming regions are present and can/do define a domain should definitively be provided.

Reviewer #2:

By whole-transcriptome analysis (using NGS), database mining and subsequent thorough phylogenetic analysis Hering and colleagues revealed that the ancestral genes encoding cytoplasmic IFs in bilaterians have once been lost in the entire panarthropod lineage, including the soft skinned velvet worms (Onychophora) and water bears (Tardigrada). This important finding suggests that the development of an exoskeleton may not have been the main driving force for the loss of cytoplasmic IFs in euarthropods. Moreover, the presented phylogenetic analysis strongly supports that several members of this taxon, notably tardigrades, collembolans and copepods each independently evolved novel cytoplasmic IFs, again originating from gene duplications of a nuclear lamin gene, long after the first bilaterian cytoplasmic IF arose. Furthermore, the authors demonstrated the capacity of the novel cytoplasmic IF protein lamin-3 to form IF networks within transfected cultured cells and traced its expression in the tardigrade H. dujardini, showing a unique, belt-like distribution within the peripheral cytoplasm of ectodermally derived epithelial cells, such as the epidermis.

In the study presented here, the authors do not only provide a sound and comprehensive phylogenetic analysis, but also combine their results with a broad survey of expression data as well as functional analyses. This is a well and carefully reasoned scientific approach, providing important insights into the early evolutionary steps that ultimately led to the high diversity of metazoan IF proteins as well as to the role of selective pressure that may have triggered those evolutionary events. The structure and function of a given protein or protein family is a result of a long-term evolutionary process and therefore cannot be completely understood without such phylogenetic considerations. Overall the results and conclusions presented in the submitted manuscript are convincing, making a valuable contribution to both, a better understanding of the evolutionary history of the IF protein family and how the novel cytoplasmic IF may account for the tardigrades ability to withstand the extraordinary mechanical stress enforced by extreme environmental conditions such as desiccation and cryobiosis.

Therefore, this referee strongly recommends the publication of the submitted manuscript in eLife.

Reviewer #3:

Hering et al. report on the evolution of cytoplasmic intermediate filament proteins (cIF) in the panarthropod lineage. They have characterized three tardigrade IF proteins in more detail: two lamins and one cytoplasmic IF protein. They convincingly show by immunofluorescence microscopy that the two lamins are nuclear and that the third protein is cytoplasmic. It forms filamentous structures in the cytoplasm of epithelial cells. This part of the manuscript is clear and convincing.

From their data they conclude that the genes for cytoplasmic IF proteins have been lost at the base of the panarthropod lineage and that cIF genes have been acquired at least three times independently in collembolans, copepods, and tardigrades. This is a very interesting point. This statement is based solely on evolutionary tree construction of a broad range of IF rod sequences. However, it is not fully convincing. For the collembolan lineage it is based solely on the fact that the isomin rod sequence groups with bilaterian lamin rod sequences. At none of the nodes of the two trees bootstrap values exceed 50%. Given that collembolans are at the base of the arthropod lineage this conclusion requires more solid evidence.

In general, the tree construction needs a more careful handling of the sequence data. A closer look at some of the sequences revealed that the presence of cyclin and UDG motifs in the putative IF sequences is probably an artifact due to errors in the automatic gene predictions; e. g.: XP 003386754_1 and GAA48725_1 predict a lamin-cyclin chimera and a lamin-uracil−DNA glycosylase chimera, respectively. The existence of such chimeric proteins is not supported by expression data. Other sequences are incomplete precluding an assignment as either lamin or cytoplasmic IF. The sequence data used for the tree building should be carefully checked to avoid unnecessary confusion.

Additional data files and statistical comments:

1) The sequences of the proteins described in this study should be given in the supplementary information together with the accession numbers of the corresponding genomic contigs (at least for the three tardigrade proteins).

2) A sequence alignment with a selection of other lamin/cytoplasmic IF proteins would be desirable.

Reviewer #4:

This is an interesting paper which is based on two pieces of evidence.

First, the authors perform a detailed phylogenetic analysis of IF proteins and their relatives, including isomin. They conclude that isomin clusters with lamins rather than with cytoplasmic IFs.

Second, the authors experimentally explore the localization of three lamins (1,2 and 3) present in a tardigrade. They show that while 1 and 2 are located in the nucleus as expected, lamin-3 is observed in the cytoplasm of epithelial cells.

The main conclusion of the paper is that the true cytoplasmic IFs were lost in panarthropods, but lamin-like IFs can take over their functions in the cytoplasm. While this theory certainly needs additional proof (beyond the reported data on tardigrade lamin-3 and in fact on isomin as well), I think that this is an important advance.

The new findings also suggest that the current terminology may need some adjustment, as the name 'cytoplasmic IFs' becomes ambiguous as one reads through the paper. Should one speak of 'non-lamin' IFs vs. lamins to reflect the phylogeny, which would then be uncoupled from the cytoplasmic vs. nuclear localization?

The phylogenetic analysis of Mencarelli et al. was less rigorous than the one presented here. As far as I can tell from that paper, Mencarelli et al. simply claimed that isomin was an IF relative, found in the cytoplasm (just like the tardigrade lam3 discussed here). The new phylogenetic analysis is much more systematic, relating isomin to lamins.

Thus there does not seem to be a real contradiction between Mencarelli et al. and the current paper (as suggested in the first paragraph of the Results and Discussion section). I think that the authors should adjust this part of the text accordingly.

In any case, the new paper is an important step forward.

1) While it is clear that the full phylogenetic tree can only fit in the supplement, the authors should present some simplified but yet scientifically rigorous view of such a tree. The current drawing shown in Figure 1B is oversimplified. (In fact, the tree shown in Mencarelli et al. would be a good example).

2) Regarding Figure 1—figure supplement, it has been known for decades how coiled coils fold. Trying to model a structure of a monomer without taking into account the interaction between the chains (usually two) of the coiled coil clearly yields a nonsense result here. The authors should either re-do the modelling properly or remove the figure (which I believe is not important here).

DOI: http://dx.doi.org/10.7554/eLife.11117.054

Author response


If your username is different from your full name, we require you to identify yourself within the comment itself. Comments are checked by a moderator (and/or an eLife editor) before they appear. Comments should be constructive, relevant to the article, conform to our terms and conditions, and include any pertinent competing interests.