Cell density-dependent proteolysis by HtrA1 induces translocation of zyxin to the nucleus and increased cell survival

Proteases modulate critical processes in cutaneous tissue repair to orchestrate inflammation, cell proliferation and tissue remodeling. However, the functional consequences and implications in healing impairments of most cleavage events are not understood. Using iTRAQ-based Terminal Amine Isotopic Labeling of Substrates (TAILS) we had characterized proteolytic signatures in a porcine wound healing model and identified two neo-N termini derived from proteolytic cleavage of the focal adhesion protein and mechanotransducer zyxin. Here, we assign these proteolytic events to the activity of either caspase-1 or serine protease HtrA1 and analyze the biological relevance of the resultant zyxin truncations. By cellular expression of full-length and truncated zyxin proteins, we demonstrate nuclear translocation of a C-terminal zyxin fragment that could also be generated in vitro by HtrA1 cleavage and provide evidence for its anti-apoptotic activities, potentially by regulating the expression of modulators of cell proliferation, protein synthesis and genome stability. Targeted degradomics correlated endogenous generation of the same zyxin fragment with increased cell density in human primary dermal fibroblasts. Hence, this newly identified HtrA1-zyxin protease signaling axis might present a novel mechanism to transiently enhance cell survival in environments of increased cell density like in wound granulation tissue.


Introduction
Cutaneous wound healing is a complex tissue response that is initiated upon skin injury to restore the skin's barrier function, which is critical to protect the human body from environmental insults 1 . Proteases play major roles in all phases of wound healing by contributing to formation of the blood clot, initiating innate immune responses through complement activation, facilitating migration of immune cells, keratinocytes and fibroblasts and remodeling of the extracellular matrix during scar formation. Importantly, proteases have also been identified as major contributors to chronic non-healing wounds, which do not progress through the normal healing process, but are stalled in a persistent inflammatory state, with impaired neovascularization and reduced synthesis of collagen [2][3][4] . Thus, identification and mechanistic dissection of proteolytic events in wound repair are critical to better understand this important tissue response and to devise novel strategies for intervention with aberrant protease activities in impaired healing.
In recent studies, we characterized proteolytic signatures during wound healing by iTRAQ-based Terminal Amine Isotopic Labeling of Substrates (TAILS) of wound exudates from a clinically relevant porcine wound healing model and from trauma patients 5,6 . By N-terminomics analysis of liquid biopsies collected from the porcine model, we revealed 1667 neo-N termini from 492 proteins. Among these we identified two neo-N termini in the focal adhesion (FA) protein zyxin indicating proteolytic processing.
Zyxin has three LIM domains localized in the C-terminal region that tether the protein to FAs 7 . LIM domains are cysteine-rich sequences that adopt zinc finger conformations, which allow protein-protein or protein-DNA interactions. LIM domains containing proteins have been associated with important biological functions, such as response to force in the actin cytoskeleton [7][8][9] . More than 20 LIM domain proteins have been identified in FAs, whereby zyxin has received special attention due to its roles in actin dynamics and nuclear-cytoplasm communication 7,10 . The central region of the zyxin protein harbors two leucine-rich nuclear export sequences (NES) 7 . Mutation of these NES sequences or the treatment of cells with leptomycin B (an inhibitor of Crm1dependent nuclear export) lead to accumulation of zyxin in the nucleus, showing a role of zyxin in signal transduction between focal adhesions and the nuclear compartment 11 . However, the mechanism of zyxin translocation to the nucleus is poorly characterized. As zyxin does not have a canonical nuclear localization signal (NLS), shuttling between focal adhesions and the nucleus should be mediated by unconventional mechanisms or through association with NLS containing proteins 12 . Although the role of zyxin in the nucleus is still unclear, evidences obtained from other members of the zyxin family migrating to the nucleus support the idea that these proteins interact with transcription factors or the basal transcription machinery to regulate gene expression [12][13][14] . Studies performed with cells susceptible to high stretch showed translocation of zyxin to the nucleus upon cell stretching, where it influenced the expression of more than 400 genes mostly related with inhibition of apoptosis and proliferation, reinforcement of ECM interactions and proinflammatory response 15,16 . In the present study, we aimed at analyzing the significance of the newly identified proteolytic processing events in zyxin to gain further insight how they might influence its complex functions as a major focal adhesion protein in cutaneous wound repair. Using targeted degradomics, we demonstrate proteolytic processing of zyxin at two sites, which we could assign to either caspase-1 or serine protease HtrA1 activity. The zyxin fragment truncated at the HtrA1 site translocated to the nucleus, protected from etoposide-induced cell death, modulated abundance of wound-related proteins in HeLa cells and was endogenously generated at high cell density in primary fibroblasts and in HtrA1-overproducing colon cancer cells. Hence, our results suggest a novel function of HtrA1 in proteolytic control of subcellular localization of zyxin to enhance survival of cells at increased density.

Processing of zyxin by HtrA1
Recombinant human zyxin (Abcam, Cambridge, United Kingdom) was incubated with recombinant human HtrA1 (R&D Systems, Minneapolis, MN) at an enzyme/protein ratio (m/m) of 1/5 for 16 h in 50 mm HEPES (pH 7.8), 100 mM NaCl at 37°C 19 . Reaction was stopped by heat inactivation and the product was analyzed by immunoblot or PRM. To generate the 'spike-in control', an aliquot was digested with GluC and spiked into fibroblast lysates. The experiment was performed in triplicate.

Cloning of zyxin-RFP and zyxin-FLAG
The vector pDsRed-N1 containing full-length zyxin was obtained from Addgene (Plasmid #26720) and used for amplification of N-terminally truncated fragments with primers specific for sequences upstream of the identified cleavage sites. Inserts were cloned into the pDsRed-N1 backbone using HindIII and BamHI cloning sites or into a pFLAG CMV vector (kindly provided by Dr. Tobias Beyer, ETH Zurich, Switzerland) using HindIII and KpnI cloning sites.

Fluorescence microscopy of HeLa cells
HeLa cells were grown overnight in chamber slides with 1.7 cm 2 . Cells were transfected as described above and incubated for 48 h. Next, cell culture medium was removed and cells were fixed with 4% PFA for 10 min at RT. Cells were permeabilized with PBS, 0.5% Triton X-100 and blocked with blocking buffer (2% BSA + 0.05% Triton X-100 in PBS). Actin and nuclei were stained by treatment with Phalloidin-Atto 488 (Sigma Aldrich, Buchs, Switzerland) for 1 h and with Hoechst reagent (Sigma Aldrich, Buchs, Switzerland) for 5 min. Slides were mounted with Mowiol/DABCO solution (Sigma Aldrich, Buchs, Switzerland) and visualized using a Zeiss Axio Imager A1 fluorescence microscope equipped with a HXP 120 Laser for fluorophore excitation. The experiments were performed in triplicate.

Subcellular fractionation of HeLa cells
HeLa cells were grown in culture dishes (ø 6 cm) and lysed in hypotonic lysis buffer (HLB) (10 mM Tris-HCl, pH 7.5, 10 mM NaCl, 3 mM MgCl 2 , 0.3% NP-40) supplemented with cOmplete protease inhibitor cocktail (Roche, Basel, Switzerland), 1 mM NaF, and 1 mM Na 3 VO 4 . Lysates were spun down for 10 s at 4°C and the supernatant was kept as cytoplasmic fraction. The pellet was washed three times with HLB buffer and re-suspended in nuclear lysis buffer (10 mM Tris-HCl (pH 7.5), 0.15 M NaCl, 3 mM MgCl 2 , 0.3% NP-40, 10% glycerol) supplemented with cOmplete protease inhibitor cocktail, 1 mM NaF, and 1 mM Na 3 VO 4 . The nuclear extract was sonicated, spun down for 10 s at 4°C and the supernatant kept as nuclear fraction. The experiments were performed in triplicate.

Influence of zyxin fragments on apoptosis
HeLa cells were transfected with pFLAG CMV vector as described above. After 48 h, 50 mM etoposide (Sigma-Aldrich, Buchs, Switzerland) was added to culture medium for 2 or 4 h. Cells were scraped off the dish and pelleted by centrifugation (retaining the floating cells), washed with PBS, lysed with Triton X-100 lysis buffer (20 mM Tris-HCl (pH 8.0), 137 mM NaCl, 10% Glycerol, 2 mM EDTA, 1% Triton X-100, cOmplete protease inhibitor cocktail) and analyzed by immunoblot. The influence of zyxin on apoptosis was evaluated once in unchallenged cells and once in cells cultured in the presence of etoposide.

Cell density-dependent proteolysis of zyxin
Human dermal fibroblasts were cultured as above until cellular confluence reached approximately 50%, 80% and for 7 days after full confluence. Growth medium was changed every day and cells were lysed with Triton X-100 lysis buffer before analysis by immunoblot or PRM. The experiments were performed in triplicate.

Processing of cell extracts for MS analysis
Human dermal fibroblasts and HeLa cells were lysed with Triton X-100 lysis buffer and cell debris was removed by centrifugation. Next, cell extracts were reduced with 3.5 mM TCEP, alkylated with 5 mM chloroacetamide, cleaned by acetone/methanol precipitation and digested for 16 h with trypsin gold (Promega, Madison, WI) or GluC endopeptidase (Roche, Basel, Switzerland).

Parallel reaction monitoring (PRM)
For PRM analysis of recombinant proteins, digested peptide samples were desalted with custom made C18 stage tips and analyzed with a Q Exactive mass spectrometer (Ther-moFisher Scientific, Bremen, Germany) using a 50 cm column at 45°C. Data was recorded in unscheduled PRM mode over a 70 min gradient with 35000 resolution using 1e6 ions with isolation windows of 1.2 m/z and 500 ms injection time. For PRM analysis of fibroblast cellular extracts, samples were analyzed with a Q Exactive HF-X instrument (ThermoFisher Scientific, Bremen, Germany) using a 50 cm column at 45°C. Data was recorded in unscheduled PRM mode over a 70 min gradient with 30000 resolution using 3e6 ions with isolation windows of 1.2 m/z. Maximum injection times of 700 ms and 100 ms were set for analysis of GluC generated zyxin cleavage and GAPDH peptides, respectively. For detection of tryptic peptides, 5e5 ions were used with injection times of 200 ms.

Tandem mass tags (TMT)-based proteomics
HeLa cells were transfected in triplicates with pFLAG CMV vector encoding full length zyxin and zyxin 277-572 and cultured for 48 h before lysis with Triton X-100 lysis buffer as described above. Peptides were labeled with TMT reagents (ThermoFisher Scientific, Rochford, IL) according to the manufacturer's instructions. Briefly, 50 µg digested peptides per condition were labeled at a 4/1 (w/w) TMT/protein ratio for 1 h at room temperature. Labeling reaction was quenched with 100 mM ammonium bicarbonate for 30 min. Next, peptide samples were mixed, desalted with sep-pak C18 columns (Waters, Milford, MA) and analyzed on a Q Exactive (Thermo-Fisher Scientific, Bremen, Germany) instrument using a 50 cm column at 45°C. Data was recorded in datadependent acquisition mode over a 140 min gradient using full-MS scans recorded with 70000 resolution using 3e6 ions for MS1, and resolution 35000, top 10 precursors with isolation windows of 1.6 m/z and 120 ms transient time for MS2.

Normalization, annotation and statistical analysis
For normalization of TMT-channels, the function "normalize to peptide amount" was selected in Proteo-meDiscoverer 2.2. Next, the value of each channel was normalized to the pooled reference channel. Subsequently pairwise comparisons with two-sided t-tests were performed using Microsoft Excel. Raw p < 0.05 was considered significant. Heatmaps of protein abundances were created using MeV Viewer (www.tm4.org).

Zyxin is proteolytically processed in cutaneous wound exudates
In a recent study, we systematically assessed limited proteolysis in the healing skin wound in vivo by quantitative degradomics analysis of wound exudates from a porcine wound model 6 . From this dataset we identified a novel proteolytic cleavage of zyxin C-terminal to Val276 as well as a known processing site C-terminal to Asp149 that had been previously associated with caspase-1 activity (Fig. 1a) 20 . Importantly, both cleavage site sequences are conserved between pig and human (Fig. 1a), suggesting that the proteolytic processing of zyxin might influence its biological function. Concomitant identification of multiple internal tryptic peptides C-terminal to cleavage sites ( Supplementary Fig. S1) indicated generation of stable N-terminally truncated zyxin fragments comprising its LIM domains, and C-terminal nuclear export sequence ( Fig. 1b). High coverage of this part of the protein has also been observed by others in pig and human tissues as well as in human plasma 21,22 .
Caspase-1 and serine protease HtrA1 are responsible for zyxin cleavage Identification of the protease responsible for a particular proteolytic event is essential for understanding the influence of protein processing in physiological contexts. Therefore, we treated cellular extracts from differentiated THP1 monocytes with recombinant caspase-1 and by immunoblot analysis observed a fragment indicating cleavage C-terminal to Asp149 and confirming previous results from Agard et al. 20 (Fig. 2a). To identify the protease responsible for processing of zyxin C-terminal to Val276, we focused on proteases concomitantly identified with the respective zyxin processing in wound exudates 6 and mined the literature and bioinformatics resources (MEROPS 23 , TopFind 24 ) for proteases with preference for amino acid sequences that resemble the identified cleavage site (Fig. 1a). We selected serine protease HtrA1 (HtrA1) as a candidate protease, since HtrA1 was also highly abundant in wound exudates 6 and recently described to prefer small hydrophobic residues (such as valine) in the P1 position 19 . To assess the capacity of HtrA1 to process zyxin, we incubated recombinant human zyxin with recombinant human HtrA1 and analyzed the cleavage product by immunoblot with an antibody directed against the Cterminal region of zyxin (Fig. 2b). Thereby, we observed generation of a fragment with expected molecular weight (~37 kDa), confirming the ability of HtrA1 to cleave zyxin. Next, since immunoblot analysis does not provide sequence information of the cleavage site, we developed a parallel reaction monitoring (PRM)-based targeted degradomics assay 25 to evaluate the generation of the human homolog of the respective zyxin neo-N terminus identified in vivo. Applying this PRM assay to recombinant human zyxin incubated with recombinant human HtrA1, we identified the expected neo-N-terminal peptide with an intensity approximately one order of magnitude higher in the HtrA1 treated than the control sample (Fig. 2b). Hence, our data identified HtrA1 as a candidate protease cleaving zyxin C-terminal to Val276 during cutaneous wound repair.
Processing of zyxin at the HtrA1 cleavage site generates a fragment that translocates to the nucleus Since zyxin can shuttle between focal adhesions and the nucleus 11 , we evaluated the influence of N-terminal truncation of zyxin on its subcellular distribution. For this purpose, zyxin-RFP and zyxin-FLAG fusion constructs corresponding to full length zyxin and the two Nterminally truncated fragments were expressed in cultured HeLa cells. As expected 26 , visualization of RFP signal revealed localization of full-length zyxin to focal adhesions. The fragment zyxin 277-572 readily translocated to the nucleus, whereas cells expressing zyxin 150-572 presented a globally low RFP signal and few RFP positive cells (Fig. 3a). Moreover, the few RFP positive cells showed morphological signs of dying cells. Immunoblot analysis of cytosolic and nuclear subcellular fractions of HeLa cells expressing zyxin-FLAG validated localization of full-length zyxin in the cytoplasm, presence of zyxin 277-572 in the nucleus and lower expression levels of zyxin 150-572 , confirming the observations made by fluorescence microscopy (Fig. 3b). Additionally, the immunoblot showed fragmentation of full-length zyxin with Fig. 2 Processing of zyxin by caspase-1 and serine protease HtrA1. a Proteolytic processing of zyxin by caspase-1. Cellular lysates from differentiated THP1 monocytes were incubated with recombinant caspase-1 for 2 or 4 h and analyzed by immunoblot with an antibody raised against the C-terminal region of zyxin (B71). b Proteolytic processing of zyxin by HtrA1. Recombinant human zyxin was incubated with recombinant human HtrA1 for 16 h and analyzed by immunoblot (as in a) and by parallel reaction monitoring (PRM). PRM assay was based on a semi-GluC peptide released from human zyxin (Fig. 1A). Cellular lysates from HeLa cells expressing the fragment zyxin 277-572 -FLAG were used in the immunoblot analysis for molecular weight comparison.
formation of two fragments of molecular weights comparable with the truncated versions, supporting intracellular and specific proteolysis, potentially by caspase-1 and HtrA1.
Cell density determines proteolytic processing of endogenous zyxin at the HtrA1 cleavage site Interestingly, we observed that the abundance of the zyxin fragment with a molecular weight of~37 kDa was dependent on the number of days cells had been cultured between transfection with the full-length construct and lysis. Hence, to test the hypothesis that high cell density induces proteolytic processing of endogenous zyxin, we cultured human dermal fibroblasts for seven days after cells reached 100% confluency and assessed the fragmentation of zyxin by immunoblot. Indeed, endogenous zyxin was increasingly processed in cells cultured for longer periods (Fig. 4a). To correlate the observed fragment with the truncation generated by HtrA1, we used PRM to quantify the neo-N terminus starting at A277 and the abundance of zyxin in cells cultured for distinct periods. We observed an increase in abundance of the neo-N terminus potentially generated by HtrA1 from day 4 to day 7, while the zyxin protein only increased in abundance from 50% confluency to day 2 but showed no change from day 4 to day 7 (Fig. 4b, c; Supplementary Figs. S2-S10), indicating regulated limited proteolysis of zyxin during increase in cell density. In addition, a fragment of the same size was observed in SW480 cells genetically engineered for overproduction of HtrA1 but not in controls, indicating that indeed HtrA1 activity is responsible for this cleavage and not another protease with closely related specificity (Fig. 4d).

Nuclear HtrA1-generated zyxin fragment has antiapoptotic properties
Upon transient transfection, HeLa cells expressed significantly higher levels of zyxin 277-572 than of full length zyxin or zyxin 150-572 (Fig. 3b). Previous studies described pro-or anti-apoptotic properties of zyxin, depending on tissue, cell type and stimulus [27][28][29][30] . To test the hypothesis that the differences in expression of full-length zyxin and the two truncated fragments resulted from pro-or antiapoptotic functions of the ectopically expressed fragments, we transfected HeLa cells with zyxin-FLAG fusion constructs and monitored levels of cleaved PARP 31 48 h after transfection. Interestingly, we did not observe differences in the levels of cleaved PARP between control, full-length zyxin and zyxin 150-572 , whereas cells transfected with zyxin 277-572 showed lower intensity of the PARP cleavage fragment (Fig. 5a). To further test the antiapoptotic properties of HtrA1-truncated zyxin, we treated HeLa cells expressing zyxin 277-572 with etoposide and monitored activation of caspase-3 and PARP cleavage by immunoblot. Thereby, we observed a lower caspase-3 activation and PARP processing in cells expressing zyxin 277-572 than in control cells (Fig. 5b), confirming the pro-survival properties of zyxin 277-572 . Hence, our data suggest that the lower expression of zyxin 150-572 is most likely due to degradation of the fragment, while the higher expression of zyxin 277-572 is most likely associated with pro-survival properties of the fragment.

Ectopic expression of zyxin 277-572 modulates abundance of wound-related proteins
To elucidate the molecular mechanism underlying the pro-survival phenotype observed in cells expressing b Immunoblot analysis of subcellular fractions of HeLa protein extracts. HeLa cells were transfected with FLAG-tagged full-length zyxin and truncated fragments. Forty-eight hours after transfection cells were fractionated into cytosolic and nuclear fractions and analyzed by immunoblot using an antibody directed against FLAG epitope. GAPDH (cytosolic protein) and matrin-3 (nuclear protein) were used as controls for efficiency of subcellular fractionation.
zyxin 277-572 , we transiently expressed full-length zyxin-FLAG and zyxin 277-572 -FLAG fusion constructs in cultured HeLa cells and analyzed the quantitative changes in cellular proteomes by tandem mass tag (TMT)-based quantitative proteomics (Fig. 6a; Supplementary Table S1). For statistically robust comparison of distinct conditions, we exploited the multiplexing capabilities of TMT isobaric tags and analyzed three replicates per condition within the same experiment, using a pooled reference for scaling. By student t-tests we performed pairwise comparisons of protein abundance in cells expressing zyxin 277-572 with full-length zyxin and mock transfection (control). Thereby, we identified a subset of four proteins (alkaline phosphatase (ALP), protein S100-P, heterochromatin protein 1 homolog gamma (HP1γ) and αactinin-4 (ACTN4)) with significantly higher abundance in cells expressing zyxin 277-572 and a subset of three proteins (basic transcription factor 3 (BTF3), nucleolar and coiled-body phosphoprotein 1 (NOLC1) and DNA mismatch repair protein MSH6) with lower abundance in the same condition ( Fig. 6b; Supplementary Table S2) (raw p-value<0.05). These results suggest that nuclear zyxin 277-572 may regulate expression of proteins related to cell proliferation and migration (S100P, ACTN4 and BTF3), protein synthesis (CBX3, BTF3, NOLC1) and genome stability (CBX3, BTF3, NOLC1 and MSH6), processes which are all related to wound healing progression.

Discussion
In this study, we characterized the biological relevance of two proteolytic cleavages of the focal adhesion protein zyxin, which we had identified in vivo by iTRAQ-TAILS analysis of porcine wound exudates 6 . A cleavage event Cterminal to Asp149 was previously associated with caspase-1 activity 20 , yet the biological relevance has not been assessed. Here, we provide evidence that a Cterminal zyxin fragment related to processing by caspase-1 can only be expressed with low efficiency in mammalian cells, suggesting instability of the truncated protein. Fig. 4 Endogenous and cell density-dependent proteolysis of zyxin. Human dermal fibroblasts were cultured until reaching a cellular density of approximately 50%, 80% and for 7 days after reaching confluency. a Immunoblot analysis of zyxin fragmentation at distinct cell densities with an antibody directed against the C-terminus of zyxin (B71). b Analysis of zyxin semi-GluC neo-N terminus and fully tryptic peptides (see also Supplementary Fig. S2) in cellular extracts of fibroblasts cultured at distinct cellular densities. Fully tryptic and GluC peptides of GAPDH were monitored in each MS run for loading control (see also Supplementary Figs. S3 and S4). 'Spike-in control' was generated by incubating recombinant human zyxin with HtrA1 and GluC and spiking into fibroblast extracts. c Relative abundances of zyxin 277 neo-N terminus and total protein in lysates from fibroblasts grown to increasing cell densities measured by PRM. Data were recorded in three independent experiments, normalized to GAPDH and scaled to the maximum (for PRM traces, see Supplementary Figs. S5 to S10). Error bars: SD. d Immunoblot analysis of zyxin fragmentation in HtrA1-overproducing (pHtrA1) and HtrA1-silenced (shHtrA1) SW480 cells and vector-transfected controls (pEV, shEV) 17,18 . Black arrow points to band at~37 kDa only present in samples from HtrA1-overproducing cells. Cells were analyzed in duplicates.
Moreover, we describe for the first time a proteolytic mechanism regulating zyxin's subcellular localization by inducing its migration to the nucleus. We propose that this nuclear translocation is induced by HtrA1-mediated proteolysis of zyxin C-terminal to Val276 in response to cues triggered by cell density and that nuclear zyxin 277-572 affects cell survival and homeostasis in part by increasing expression of ALPP, CBX3, ACTN4 and S100P and reducing expression of BTF3, NOLC1 and MSH6. These observations provide novel insight into limited proteolysis as a signaling mechanism transmitting information from the cell periphery to the nucleus under conditions of altered cellular environments in response to environmental stress (Fig. 7).
Agard et al. identified caspase-1-dependent processing of zyxin C-terminal to Asp149 both in vitro by incubation of lysates with the active protease and in cells in response to several inflammatory stimuli 20 . Having identified this cleavage in liquid biopsies from wounds and thus samples from a complex inflammatory tissue response 6 indicates for the first time its in vivo relevance. In vitro efficiency of zyxin cleavage by caspase-1 was very low 20 , which would be in agreement with our findings that forced expression of the related C-terminal fragment generates an unstable truncated protein with detrimental effects on cell viability. This warrants further investigations to understand the fine-tuned regulation of this cleavage event and how it contributes to cellular responses under inflammatory stress conditions in various cell types.
In an effort to elucidate the importance of different regions of zyxin for its subcellular localization, Nix et al.   a Experimental design for quantitative proteomics analysis of HeLa cells expressing zyxin-FLAG fusion constructs. HeLa cells were transfected with full length zyxin, zyxin 277-572 or mock transfected in triplicates. Forty-height hours post transfection cells were lysed, tryptic protein digests labeled with TMT isobaric tags and analyzed by MS using a pooled reference for experimental normalization. b Subset of proteins with significant differences in abundance between cells expressing zyxin 277-572 and full length zyxin and mock control group (raw p-value<0.05). The proteins alkaline phosphatase (ALP), protein S100-P, heterochromatin protein 1 homolog gamma (HP1γ) and αactinin-4 (ACTN4) showed higher abundance in cells expressing zyxin 277-572 , while the proteins basic transcription factor 3 (BTF3), nucleolar and coiled-body phosphoprotein 1 (NOLC1) and DNA mismatch repair protein MSH6 showed lower abundance in cells expressing zyxin 277-572 . Heatmap representation of protein abundances was produced in MeV Viewer (www.tm4.org). Values were mean-centered and color scale limits are indicated. zyxin 13 , but chicken zyxin lacks the NES consensus sequence closest to the N terminus 32 . Interestingly, based on NetNES prediction 33 pig and mouse zyxin also harbor both NES sequences homologs to the human sequence. This might indicate evolution towards multiple NESs to ensure increased flexibility in control of subcellular localization. Importantly, to our knowledge, our study provides the first evidence of a proteolytic mechanism that induces translocation of zyxin to the nucleus.
With HtrA1, we succeeded to identify a protease that is capable of directly processing zyxin at the observed cleavage site C-terminal to Val276. HtrA1 is predominantly secreted, but at least 20% of the HtrA1 pool is localized in the cytoplasm 34 . Intracellularly, it can be associated with microtubules and proteolytically process components of the cytoskeleton to regulate cell motility and alter mechanical cell phenotypes 35,36 . Moreover, HtrA1 degrades unfolded and misfolded proteins, thereby reducing the burden of unfolded proteins, contributing to proteostasis and cell survival 34,37,38 . Thus, there is clear evidence for intracellular HtrA1-mediated proteolysis, which is further supported by this study but needs more exploration to better understand the pleiotropic activities of this evolutionary conserved protease.
Translocation to the nucleus has been observed for zyxin in response to stretch-induction in endothelial cells as it appears in vasoconstriction 15,16 . Wound fibroblasts also experience mechanical tension 39 but an overall distinct extracellular environment, which might elicit different cellular responses. Here, limited proteolysis rather than cytoskeletal rearrangement might be used to disrupt zyxin's attachment to focal adhesions, facilitate nuclear translocation and regulation of gene transcription. Zyin 277-572 , the zyxin fragment corresponding to HtrA1 cleavage, was endogenously generated in fibroblasts at high cell density (Fig. 4). Increased fibroblast density is also seen in wound dermis 40 and might explain presence of zyxin 277-572 in wound exudate sampled at the onset of granulation tissue formation 6 , but the definite environmental cue to trigger zyxin cleavage has still to be discovered.
After generation by HtrA1 zyxin 277-572 retains its LIM domains, which have been shown to be able to either directly interact with DNA or other proteins to mediate zyxin's activities in regulating gene expression 41 . Overexpression of zyxin 277-572 in HeLa cells moderately but significantly regulated abundances of proteins that have been associated with maintaining cellular homeostasis in environments of rapid proliferation and migration (Fig. 6) [42][43][44][45][46][47][48] . This is in agreement with observations that zyxin influences cell migration, adhesion and proliferation as well as its potential involvement in cancer progression [49][50][51][52][53] . Zyxin's role in regulating cell survival is controversial [27][28][29][30] , but nuclear activities have generally been related to antiapoptotic effects 15,16 . This is also supported by our own findings, which directly correlated high expression of nuclear zyxin 277-572 with reduced abundance of apoptosis markers. Thus, induced nuclear translocation of zyxin appears as protective response in conditions requiring extended cell viability like in tissue repair.
In the healing skin wound, keratinocytes and fibroblasts undergo a series of complex changes leading to a hyperproliferative and migratory phenotype transiently resembling cancer cells 54 . Our newly identified HtrA1-zyxin proteolytic signaling pathway presents a mechanism to support this phenotype by contributing to enhanced cell survival under conditions of high cell density. Hence, regulation of expression and activity of the HtrA1 protease might be exploited to advance granulation tissue formation in non-healing wounds 4 . Fig. 7 Proposed model for the biological consequences of zyxin proteolysis. Proteolysis of zyxin by caspase-1 C-terminal to Asp149 generates a C-terminal fragment that is most likely degraded (a). High cellular density activates HtrA1 (b) which processes zyxin C-terminal to Val276 (c) and generates a C-terminal fragment that migrates to the nucleus (d). Nuclear zyxin 277-572 induces cell survival potentially by increasing the expression of ALP, HP1γ, ACTN4 and S100-P and reducing the expression of BTF3, NOLC1 and MSH6 (e).