The cardiac syndecan-4 interactome reveals a role for syndecan-4 in nuclear translocation of muscle LIM protein (MLP)

Costameres are signaling hubs at the sarcolemma and important contact points between the extracellular matrix and cell interior, sensing and transducing biomechanical signals into a cellular response. The transmembrane proteoglycan syndecan-4 localizes to these attachment points and has been shown to be important in the initial stages of cardiac remodeling, but its mechanistic function in the heart remains insufficiently understood. Here, we sought to map the cardiac interactome of syndecan-4 to better understand its function and downstream signaling mechanisms. By combining two different affinity purification methods with MS analysis, we found that the cardiac syndecan-4 interactome consists of 21 novel and 29 previously described interaction partners. Nine of the novel partners were further validated to bind syndecan-4 in HEK293 cells (i.e. CAVIN1/PTRF, CCT5, CDK9, EIF2S1, EIF4B, MPP7, PARVB, PFKM, and RASIP). We also found that 19 of the 50 interactome partners bind differently to syndecan-4 in the left ventricle lysate from aortic-banded heart failure (ABHF) rats compared with SHAM-operated animals. One of these partners was the well-known mechanotransducer muscle LIM protein (MLP), which showed direct and increased binding to syndecan-4 in ABHF. Nuclear translocation is important in MLP-mediated signaling, and we found less MLP in the nuclear-enriched fractions from syndecan-4−/− mouse left ventricles but increased nuclear MLP when syndecan-4 was overexpressed in a cardiomyocyte cell line. In the presence of a cell-permeable syndecan-4–MLP disruptor peptide, the nuclear MLP level was reduced. These findings suggest that syndecan-4 mediates nuclear translocation of MLP in the heart.

In response to biomechanical stress and extracellular matrix cues, the heart is able to remodel its form and function to cope with increased demands. Major sites of biomechanical stresssensing and transduction are the costamere and z-discs, which are the merging points between the sarcomere and sarcolemma in the cardiomyocytes, and focal adhesions, which are the linkage sites between the extracellular matrix (ECM) 4 and cytoskeleton in fibroblasts (1)(2)(3). Molecules able to merge and coordinate signaling inward at these sites are likely key players in the biomechanical stress-sensing apparatus in the heart.
Syndecan-4 is a ubiquitously expressed transmembrane proteoglycan with a large ectodomain decorated with ECM-binding glycosaminoglycan chains, a transmembrane domain, and a short cytoplasmic tail that connects to intracellular-binding partners and the actin cytoskeleton (4). Because syndecan-4 holds no intrinsic enzymatic activity, direct signal transduction must occur through interactions with its cytoplasmic tail or indirectly through co-receptors like the integrins (4). Because of its membrane-spanning quality and ideal position at z-discs (5) and focal adhesions (6), syndecan-4 has attracted much attention as a potential mechanosensing and transducing molecule in the heart. When syndecan-4 Ϫ/Ϫ mice are subjected to mechanical stress by inducing pressure overload of the left ventricle (LV), a reduction of the expected hypertrophy and ECM stiffness is observed, placing syndecan-4 as a player in the biomechanical stress-sensing apparatus (7,8). Several direct or indirect syndecan-4 partners are implicated in cardiac remodeling and pathology, including PKC␣ (9), calcineurin (7), and the calcineurin-dependent transcription factor nuclear factor of activated T-cells (NFAT) (1,2).
To better understand how syndecan-4 exerts its cellular function in the heart, we set out to map the cardiac syndecan-4 interactome. By combining two different large-scale affinity purification methods with MS (AP-MS), we identified 21 novel syndecan-4 interaction partners together with 29 partners previously described in the literature, which together make up the cardiac interactome. 38% of these 50 interactions were altered during LV pressure overload, including MLP, a known mechanotransducer directly involved in human cardiomyopathy (10,11). We found a direct interaction between MLP and syndecan-4. Nuclear translocation of MLP, which is important for its ability to initiate hypertrophic gene expression (12), was reduced in syndecan-4 Ϫ/Ϫ mice and in H9c2 rat cardiomyoblast cells treated with a cell-permeable syndecan-4 -MLP disruptor peptide. In contrast, the nuclear MLP level increased when syndecan-4 was overexpressed in H9c2. Altogether, our data suggest syndecan-4 mediates nuclear translocation of MLP.

Use of two different AP-MS approaches to capture syndecan-4 interaction partners
To increase confidence in our results, two different affinity purification (AP) approaches were used to identify syndecan-4 protein partners. In the first approach, we used biotinylated peptides covering the cytoplasmic part of syndecan-4 as bait, with or without phosphorylation at serine 179 (SDC4 cyt and pS179 -SDC4 cyt , respectively) ( Fig. 1A, panel i). A biotinylated scrambled peptide was used as a negative control (peptide sequences are given in Fig. 1B). The integrity of the peptides was first confirmed by immunoblotting. Both biotin-SDC4 cyt and biotin-pS179 -SDC4 cyt were recognized by anti-syndecan-4 cyt (custom-made with an internal epitope, Genscript) ( Fig.  S1A, upper panel), and biotin-pS179 -SDC4 was only recognized by anti-pS179 -syndecan-4 (custom-made, Genscript) (Fig. S1A, middle panel), whereas all three peptides were recognized by anti-biotin-HRP (Fig. S1A, lower panel). To confirm the binding capacity of the bait, biotin-SDC4 cyt was shown to bind to the syndecan-4 C1-domain partner ezrin (13) in an ELISA-based interaction assay, whereas the scrambled control peptide did not (Fig. 1C). For the mass spectrometry (MS) analyses, equivalent amounts of the three peptides were separately incubated in rat LV lysate, and putative syndecan-4 -interacting proteins were pulled out with antibiotin-agarose beads. Nonspecific binders (light blue proteins in Fig. 1A) were removed by several wash steps, and the remaining bound proteins (light brown proteins) were eluted with 25 mM biotin before precipitation with ice-cold acetone, trypsin digestion, and identification by MS.
We also performed an extensive literature search and a summary of the 71 direct or indirect literature syndecan-4 partners reported in different species, tissues, and cells is given in Table  S3. Interestingly, 29 of the literature partners were also significant in our IP-syndecan-4 approach (Fig. 2, Venn diagram, green circle, and Table S3, p Ͻ 0.05). The remaining literature partners were either not significant or showed no interaction with syndecan-4 in LV lysates.

Confirmation of syndecan-4 interaction partners in HEK293
Novel syndecan-4 protein partners were validated for syndecan-4 binding in HEK293 cells. MPP7, PARVB, CDK9, EIF2S1, PFKM, CCT5, RASIP1, EIF4B, and CAVIN1 were cloned with either a 3ϫFLAG tag at the N terminus or a single FLAG tag at the C terminus and co-expressed with HA-syndecan-4 in HEK293 cells. The HEK293 lysates were further subjected to immunoprecipitation using anti-FLAG, thereby using the putative partner as bait and reversing the setup used for the MS analysis. Immunoblotting with anti-HA or anti-syndecan-4 revealed that syndecan-4 co-precipitated strongly with MPP7 ( . We detected binding between CAVIN1 and the 25-kDa syndecan-4 when probing with anti-syndecan-4 and not anti-HA, indicating that this band could represent the endogenous syndecan-4 core protein. Only nonspecific binding was observed in the negative controls 3ϫFLAG-ADD3 (␥-adducin) and 3ϫFLAG-STK39 (STE20/ SPS1-related proline-alanine-rich protein kinase) (Fig. 3, J and K, respectively) (n ϭ 3). Thus, we conclude that the syndecan-4 binding in Fig. 3, A-I, was not due to simple overexpression. Previous studies have shown unsuccessful translocation of Precipitation of syndecan-4 -interacting proteins was performed in panel i by fishing with a biotinylated peptide covering the syndecan-4 cytoplasmic part (biotin-SDC4cyt), or the syndecan-4 cytoplasmic part phosphorylated at serine 179 (biotin-pS179 -SDC4cyt), or by precipitation of endogenous syndecan-4 using syndecan-4 -specific antibodies in panel ii WT LV lysate or in panel iii ABHF LV lysate. A scrambled syndecan-4 control peptide, nonrelevant IgG, beads, and lysate from SHAM LVs were used as respective negative controls. Illustrations of the full-length transmembrane syndecan-4 and aortic banding are given at left. B, sequences of the three biotinylated peptides used in the pulldown approach in A. C, binding of biotin-SDC4cyt to ezrin was confirmed by ELISA (unpaired t test, ****, p Ͻ 0.0001). D, anti-SDC4 (KY/8.2) was overlaid arrays of immobilized overlapping 20-mer peptides covering the full-length syndecan-4 rat protein sequence. The core epitope is underlined. Immunoblotting without any primary antibody was used as negative control (lower panel) (n ϭ 2, two independent peptide arrays syntheses). E, antibody core epitope is underlined in the alignment of rat syndecan 1-4 (DNA Star, Madison, WI). Black boxes indicate identical amino acids. F, anti-SDC4 (KY/8.2) (with an external epitope) precipitated endogenous syndecan-4 in lysates from both rat neonatal cardiomyocytes (CM) and cardiac fibroblasts (CFB). Nonrelevant rat IgG was used as negative control.

Cardiac syndecan-4 interactome is enriched in cell communication/signal transduction, cardiovascular disease, stress-response proteins, and PDZ domains
To get an overview of the cardiac syndecan-4 interactome, the 21 novel partners together with the 29 literature partners (Venn diagram in Fig. 2) were clustered according to the biological process using the GO annotation data from the Human Protein Reference Database (HPRD) (Fig. 4). Most of the syndecan-4 partners were involved in cell communication/signal transduction, protein metabolism, and cell growth/maintenance. Within "cell communication/signal transduction" (right in Fig. 4), "enzymatic activity," which includes serine/threonine/tyrosine kinases and phosphatases, formed the largest cluster, whereas "receptor activity" and "GTPase activity" were the next biggest clusters. An in silico interaction analysis using STRING 10.5 (http://string-db.org 5 (15)) showed that the cardiac syndecan-4 interactome contained more interactions than expected from a random set of proteins, indicating that these proteins indeed form complexes (nodes, 50; edges, 166 (expected, 69), average node degree, 6.64; average local clustering coefficient; 0.621, PPI enrichment p value, Ͻ1.0e-16) (Fig.   S2). According to STRING, 42% of the proteins in the cardiac syndecan-4 interactome were coupled to stress response (Fig.  S2, labeled in yellow).

Changes in the syndecan-4 interactome during development of heart failure
To identify changes within the syndecan-4 interactome during pressure overload, large scale IP-syndecan-4 experiments were performed in LV lysate isolated from ABHF and SHAM rats (Fig. 1A, panel iii). In total, 19 syndecan-4 interactions, involving novel and literature partners, were found to be either up-or down-regulated in ABHF compared with SHAM ( Fig. S4 and Tables S1 and S3, right columns, n ϭ 3, p Ͻ 0.05). These 19 proteins clustered into 11 different syndecan-4 networks (Fig.  5, novel partners in pink) according to STRING 10.5 and literature. The largest cluster contained the novel partners MLP and UACA and the literature partners PRKCD, PRKCA, ARF6, RHOA, RHOG, and ACTN2. The second largest cluster contained collagen ␣-1 (COL1A1) and other extracellular literature partners and is consistent with a role for syndecan-4 in matrix remodeling and fibrosis in heart (recently reviewed in Ref. 56).

Syndecan-4 mediates nuclear translocation of MLP
As MLP is a well-known mechanotransducer, we wanted to analyze its interaction with syndecan-4 in more detail. Consistent with the MS dataset (Figs. S3K and S4E), we showed that MLP co-precipitated stronger with anti-syndecan-4 (KY/8.2) than nonrelevant rabbit antibodies (rIgG) (Fig. 6A) and that the interaction was increased in ABHF compared with SHAM ( Fig.  6B). Consistently, the total MLP protein level was increased in ABHF compared with SHAM ( Fig. 6C), whereas the syndecan-4 expression was unchanged (Fig. 6D). An ELISA-based assay using biotin-SDC4 cyt against recombinant MLP identified the syndecan-4 -MLP interaction to be direct (Fig. 6E). To assess whether syndecan-4 influenced nuclear translocation of MLP, MLP was analyzed in nuclear-enriched fractions isolated from wildtype (WT) and syndecan-4 knockout (SDC4 Ϫ/Ϫ ) mouse hearts. The nuclear MLP level was reduced when syndecan-4  Table 2). SDCBP, CALM1, and COL1A1 were also detected in the pulldown approach, but only COL1A1 was significant.

Cardiac syndecan-4 interactome reveals MLP connection
was absent (Fig. 6F, right panel), even though the total amount of MLP was equal in RIPA extracts of WT and SDC4 Ϫ/Ϫ mouse hearts (Fig. 6G). On the other side, adenovirus-mediated overexpression of full-length syndecan-4 in an H9c2 rat cardiomyoblast cell line resulted in an increased nuclear MLP level compared with null virus (control) (Fig. 6H). The amount of syndecan-4 overexpressed and the distribution in cytoplasmic, nuclear, and membrane fractions is given in Fig. S5B. Because we found the syndecan-4 -MLP interaction to be direct, breaking this bond should disrupt the biological function of this interaction. We generated a cell-permeable syndecan-4 -MLP disruptor peptide by adding a TAT tag to the N terminus of the cytoplasmic tail of syndecan-4 (TAT-SDC4 cyt ) (upper panel in Fig. 6I), and we tested its effect on isoprenaline (ISO)-stimulated H9c2 cells. H9c2 cells show a similar hypertrophic response to ISO in primary neonatal rat cardiomyocytes in vitro (17). The time point of increased MLP nuclear entry in response to ISO was determined to be 6 -12 h post-stimulation (Fig. S5C). In the presence of the disruptor peptide (TAT-SDC4 cyt ), the nuclear fraction of MLP was reduced compared with the TAT-tagged scrambled control peptide (Fig. 6I, lower panel). As a physiological readout, we also measured the protein content. Consistent with a role of MLP in gene transcription and hypertrophic growth (12), we observed less protein in the cytoplasmic fractions of cells treated with TAT-SDC4 cyt compared with cells treated with TAT-scram (Fig. 6J). The principle of this disruptor peptide experiment is illustrated in Fig. 6K. Altogether, our data suggest syndecan-4 to be a mediator for nuclear translocation of MLP.

Cardiac syndecan-4 interactome
An important step in understanding how syndecan-4 mediates its cardiac function is to determine its molecular interaction partners. In this study, we have identified the cardiac syndecan-4 protein interaction network of 21 novel partners and 29 described previously in the literature. This is the first experimental syndecan-4 interactome, and it is enriched with proteins involved in cell communication/signal transduction, stress response, metabolic, and cardiovascular disease.
In our analysis, proteins had to be detected by two different AP-MS approaches to be regarded as a novel syndecan-4 partner. This strategy might have favored more stable syndecan-4 interactions, and transient and weak interactions might have been lost. As input material, we used LVs lysed in a buffer containing 1% Triton and thus included all cellular compartments, except the nucleus, from all cardiac cell types. The complexity of the input material might have introduced false proteinprotein interactions not relevant in vivo. However, false-positives seemed not to be a problem because all protein partners we tested also revealed syndecan-4 binding in HEK293, whereas nonrelevant proteins did not. It has also been noted by others that the cytoplasmic domain of syndecan-4 has a high number of interaction partners compared with size (18). The large number of partners can be explained by different types of interactions (direct/indirect) in different cardiac cell types and subcellular locations. Syndecan-4 locates to both costamere and z-disc (5), but it is also reported to localize to Golgi structures, intracellular vesicle membranes, and perinucleus during muscle differentiation (19).

Changes in the syndecan-4 interactome during development of heart failure
Interestingly, 19 proteins in the interactome (38%) showed an altered interaction with syndecan-4 in ABHF versus SHAM and clustered into 11 smaller syndecan-4 networks (Fig. 5). Altered interactions could be due to several factors, e.g. changes in availability, post-translational modifications, syndecan-4 oligomerization level, and/or competition between the protein partners for syndecan-4 binding.
One network contained SDCBP, which binds directly through its PDZ2 domain to the C-terminal FYA motif in syn-   decan-4 (20). SDCBP is involved in recycling of the syndecan to the plasma membrane (21) and exosome formation through an SDCBP-dependent recruitment of PDCD6IP (programmed cell death 6-interacting protein), another partner in the syndecan-4 interactome (Fig. 4), and the ESCRT-III complex (endosomalsorting complex required for transport III) (22). Our findings indicate that these syndecan-4 -dependent trafficking processes might change during pressure overload. SDCBP also negatively regulates syndecan-4 function in focal adhesion formation by reducing the syndecan-4 -PKC␣ interaction and thus PKC␣ activation (20).

Cardiac syndecan-4 interactome reveals MLP connection
A second syndecan-4 network consisted of the two novel partners CAVIN1 and TRIM72, where the latter interaction was only identified in ABHF (Fig. 5). CAVIN1 plays an important role in caveolae formation and organization (23) and recruits TRIM72 in an emergency membrane repair response to protect against stress-induced loss of cardiomyocytes (24,25). Our findings of a novel syndecan-4 -CAVIN1-TRIM72 axis might suggest a compensatory role for syndecan-4 in cell membrane repair during pressure overload. Mutations in CAVIN1 are reported to cause skeletal muscle hypertrophy, lipodystrophy, and cardiac dysfunctions, including arrhythmias (25,26).
A third syndecan-4 network contained the novel partner AP3D1 and the literature partner CLTC (27), a major protein of the polyhedral coat of coated pits. AP3D1, which showed an increased interaction with syndecan-4 (Fig. 5), is a component in the ubiquitous heterotetrameric AP-3 (adaptor protein complex 3) involved in intracellular trafficking and generation of vesicles with different cargo or protein compositions. The lit-

Cardiac syndecan-4 interactome reveals MLP connection
erature partner CLTC is suggested to play a cargo-selective role in this process (28). We speculate whether syndecan-4 is subjected to increased trafficking through an AP-3-dependent pathway during pressure overload and perhaps also plays a cargo-selective role. AP3D1 is reported to associate with myocardial infarction and atherosclerosis in a Japanese population (29). A fourth syndecan-4 network contained the novel partner RRAGC and the literature partner serine/threonine-protein kinase mTOR (mammalian target of rapamycin) (Fig. 5). mTOR activity is regulated by assembly into two functionally distinct complexes, mTORC1 and mTORC2. mTORC1 is a major driver of cell growth and protein synthesis, triggered by growth factors, stress signals, energy, and nutrients (30). In response to amino acids, RRGAC GDP targets mTORC1 to lysosomes where it is activated by Rheb (31,32). Our findings of a novel syndecan-4 -RRGAC interaction only in ABHF hearts (Fig. 5) could suggest that syndecan-4 is more involved in mTORC1 signaling in heart failure development. In contrast, activity of the mTORC2-Akt-eNOS axis depends on a syndecan-4 -PKC␣-dependent recruitment of mTORC2 to rafts in endothelial cells (33).
A fifth syndecan-4 network contained the two novel partners TJP2 and EPB41 (Fig. 5), where the latter partner links cellsurface membrane proteins to the actin cytoskeleton. EPB41 and TJP2 are reported to form a complex at cell-cell junctions, which are structures known to be important in maintaining mechanical integrity of cardiomyocytes and resulting in arrhythmias when absent (34,35). Our findings of a decreased EPB41-syndecan-4 interaction (Fig. 5) indicate a reduction in the linkage between syndecan-4 and the actin cytoskeleton during pressure overload. Whether this interaction is lost at cellcell junctions will have to be investigated in future studies. However, syndecan-4 showed an increased binding to the novel partner tight junction adaptor protein MMP7 (36), supporting a role for syndecan-4 at cell-cell junctions during pressure overload.
A sixth syndecan-4 network contained the two novel partners EIF4B and EIF2S1 (Fig. 5). EIF4B is involved in the initiation of translation and recruitment of the 40S ribosomal subunit to the mRNA, a process that is regulated by phosphorylation of EIF4B according to the cell's need, through the mTOR pathway, for example (37). EIF2S1 is also involved in translation initiation and can be regulated through phosphorylation. In response to cell stress, phosphorylation of EIF2S1 has been shown to inhibit protein synthesis except for selected stress-response mRNAs (38). More EIF2S1 and less EIF4B precipitated with syndecan-4 in ABHF, which could indicate competition between the two proteins and perhaps a role for syndecan-4 in directing translation in heart failure development.
The largest syndecan-4 network contained partners involved in actin remodeling, signal-and mechanotransduction (Fig. 5). PKC␦ (PRKCD), PKC␣ (PRKCA), and UACA interacted only with syndecan-4 during pressure overload. PKC␣ is a wellknown syndecan-4 partner and is involved in adhesion formation and contractility (9,39). Upon dephosphorylation of pSer-179, syndecan-4 oligomerizes and activates PKC␣ and can thereby regulate the time and place for PKC␣ activity (40). PKC activation is recently found to be up-regulated at the costameres in heart disease (41), and we speculate whether this is due to an increased interaction with syndecan-4. In line with this, we have previously found reduced pSer-179 -syndecan-4 levels in aortic stenosis patients and in mice after aortic banding (7). In contrast, the literature partner PKC␦ phosphorylates syndecan-4 at serine 179, thus reducing PKC␣ activity and impairing endothelial cell function (42). The literature partners RHOA, RHOG, and ARF6 are involved in actin cytoskeletal remodeling (4). Syndecan-4 controls both suppression and activation of these GTPases, for example, it activates RHOG by first activating PKC␣ thereby triggering RHOG/caveolin-dependent endocytosis of integrins (4, 18). ARF6, which is also implicated in syndecan and integrin recycling (18, 21), interacts with Arrows denote whether the syndecan-4 interaction was increased (1) or decreased (2). Novel partners identified are in pink. *, interaction significantly changed in ABHF (p Ͻ 0.05, n ϭ 3); **, interaction only detected in ABHF (p Ͻ 0.05, n ϭ 3). The STRING database 10.5 (https://string-db.org/) 5 was used to generate small string maps. Figure 6. Syndecan-4 binds directly to MLP and mediates its nuclear translocation. LV lysates from WT rat (A) and ABHF or SHAM rats (six individual rat hearts) (B) were subjected to immunoprecipitation using anti-syndecan-4 (KY/8.2, which has an extracellular epitope). Levels of immunoprecipitated MLP and syndecan-4 were analyzed by immunoblotting with anti-MLP and anti-syndecan-4cyt (which has an intracellular epitope). Nonrelevant rat IgG was used as a negative control. C, total MLP protein level in SHAM and ABHF lysates was analyzed through semiquantified densitometry analysis of immunoblots shown in

Cardiac syndecan-4 interactome reveals MLP connection
the novel syndecan-4 partner UACA, another cytoskeletal remodeler (43). The network also revealed a novel syndecan-4 -MLP-ACTN2 axis. The novel partner MLP is a well-known stress-and mechanotransducer in muscle cells and associates directly with human myopathies (11). The literature partner ACTN2 is an F-actin cross-linking scaffolding protein at z-disc and cell-cell junctions, where it couples receptors, like syndecan-4, to the actin cytoskeleton (35). A human hypertrophic cardiomyopathy mutation in MLP has been shown to destabilize the interaction between MLP and ACTN2 and a third component N-rap leading to myocyte disarray (44). The increased interaction between ACTN2 and MLP with syndecan-4 during pressure overload supports a role for syndecan-4 in mechanosignaling complexes.
Altogether, our data indicate that syndecan-4 associates with important cellular structures and functions, including focal adhesions, tight junctions, endocytosis, trafficking, mechanotransduction, and cytoskeleton remodeling during development of heart failure induced by pressure overload. These wide-ranging functions are probably explained by its many intracellular binding partners.

Syndecan-4 mediates nuclear translocation of MLP
The large multiprotein structures of z-discs have been suggested as major sites for sensing and transducing mechanical stimuli (1), and of particular interest are LIM domain proteins, which are able to shuttle to the nucleus, thus connecting mechanical stimuli with gene expression (45,46). MLP, which we identified as a novel and direct syndecan-4 partner, contains two independent LIM domains that facilitate formation of multiprotein complexes (47). MLP interacts with structural and signaling proteins at the z-discs and translocates to the nucleus to initiate hypertrophic gene expression in response to biomechanical stress (11,12,14). Nuclear translocation of MLP is crucial for its mechanotransducing probabilities, and its nuclear accumulation associates with pathology (14). Interestingly, we found the syndecan-4 -MLP interaction to be increased during pressure overload, suggesting a function for this complex in cardiac pathology. Our findings of a direct syndecan-4 -MLP binding led us to develop a specific cell-permeable disruptor peptide to assess the biological function of this interaction. The syndecan-4 -MLP disruptor peptide inhibited nuclear translocation of MLP and attenuated the increase in protein content following ISO stimulation of H9c2 cells, suggesting the nuclear translocation of MLP and the hypertrophic gene expression to be syndecan-4 -dependent. The syndecan-4 -MLP interaction also appears to be relevant in vivo, because less nuclear MLP was found in the syndecan-4 Ϫ/Ϫ mouse LVs, an observation consistent with loss of concentric myocardial hypertrophy and rather LV dilatation and dysfunction during pressure overload (7,12,48).
The increased syndecan-4 -MLP interaction can be due to changes in availability, structures, or post-translational modifications during pressure overload. We have previously shown reduced pSer-179 -syndecan-4 levels in human aortic stenosis patients and in mice after pressure overload, enabling calcineurin binding and activation of the hypertrophic NFAT pathway (7). Perhaps the pSer-179 also works as a molecular switch for the MLP interaction and its nuclear translocation. Syndecan-4 has previously been found in the nuclear membrane and perinuclear compartment (19), and syndecan-1 has been suggested to translocate into the nucleus (49). Future studies are needed to investigate how syndecan-4 directs the translocation of MLP and whether syndecan-4 also goes into the nucleus. It will also be interesting to investigate whether the syndecan-4 binding overlaps with any of the MLP mutations associated with human hypertrophic cardiomyopathy (10,50) or dilated cardiomyopathy (11). Collectively, these data indicate that syndecan-4 act upstream of MLP and mediates its nuclear translocation upon stress.

Animal handling
All animal handling was preapproved by the Norwegian National Animal Research Committee (FOTS ID 3820 for rats, and FOTS ID 6989 for SDC4 Ϫ/Ϫ mice) and conformed to the Guide for the Care and Use of Laboratory Animals (National Institutes of Health Publication 85-23, revised 2011). Animals were stored with 12:12-h light/dark cycles and were given access to food and water ad libitum. For neonatal cell isolation, the permit of approval number IV1-17U was obtained.

Aorta banding in rats
Approximately 170-g male Wistar-Hannover rats (Taconic, Denmark) were anesthetized and ventilated with a mixture of 68% N 2 O, 29% O 2 and 2-3% isoflurane through endotracheal intubation. The chest was opened, and aorta banding was induced by 3.0 silk suture tightened around the ascending aorta proximal to the brachiocephalic trunk. SHAM animals were used as controls and went through the same surgical procedure except tightening of the suture around the ascending aorta and buprenorphine was used as analgesic. Echocardiography was performed with a Vevo 2100 (VisualSonics, Canada) 6 weeks after surgery, before myocardial tissue samples were harvested in deep surgical anesthesia. The left and right ventricles were rapidly dissected and subsequently put into liquid nitrogen and stored at Ϫ70°C for later molecular analysis. The criteria for congestive heart failure (ABHF) were increased lung weight (Ͼ2.0 g), increased posterior wall diameter (Ͼ1.9 mm), and left atrial diameter (Ͼ5.0 mm). Characteristics of the animals used in the large-scale AP-MS experiments are given in Table S4. For the qPCR and Western blot analysis, 23 additional animals were included using the same criteria as in Table S4.

Left ventricular lysates
Frozen rat or mouse LVs were pulverized in liquid nitrogen before homogenization with a Polytron 1200 homogenizer in three series of 1 min in lysis buffer (150 mM NaCl, 20 mM HEPES, pH 7.5, 1 mM EDTA, 0.5% Triton X-100 (Sigma)) supplemented with cOmplete protease inhibitor mixture (05050489001, Roche Applied Science, Switzerland) or in RIPA buffer (catalog no. R0278, Sigma) on ice. Following centrifugation at 100,000 ϫ g for 60 min at 4°C, supernatants were stored at Ϫ80°C until analysis.

ELISA
The ELISA plate was first coated with 0.46 g/well recombinant MLP or 10 g/ml recombinant ezrin in 1ϫ PBS and rotated ON at 4°C. The next day, the wells were washed once in 1ϫ PBS-T (0.1%) before blocking in PBS with 1% gelatin (G-1890, Sigma) for 1 h while rotating at room temperature. After blocking, the coated wells were incubated with either biotin-syndecan-4 cyt or biotin-scram (control peptide) in 1ϫ PBS-T (0.1%) for 2 h at 37°C while rotating. Thereafter, the wells were washed five times in 1ϫ PBS-T (0.1%) before incubating with anti-biotin-HRP (1:5000) for 30 min while rotating at room temperature. Then wells were washed five times in 1ϫ PBS-T (0.1%) before adding Ultra TMB/ELISA solution (34028, Thermo Fisher Scientific) and incubated for 15-30 min while rotating at room temperature. Reaction was stopped by adding 2 N HCl, and the signal was measured by a plate reader at 450 nm (Hidex Sense multimodal microplate reader, Finland).

Epitope mapping
Rat syndecan-1-4 was synthesized as 20-mer with three amino acid offsets onto cellulose membranes by a MultiPep automated peptide synthesizer (INTAVIS Bioanalytical instruments AG, Germany) as described previously (51). After blocking, the membranes were incubated with anti-syndecan-4 ON at 4°C. After washing three times in TBS-T, membranes were incubated with secondary antibodies for 1 h at room temperature, before washing another three times in TBS-T and developed as an immunoblot.

Small-scale immunoprecipitation and immunoblotting
For all IPs, lysates were incubated with 2 g of antibody and protein A/G-agarose beads (sc-2003, Santa Cruz Biotechnology) ON at 4°C and unless otherwise stated washed three times in IP buffer (150 mM NaCl, 20 mM HEPES, pH 7.5, 1 mM EDTA, 1% Triton X-100 with cOmplete protease inhibitor mixture (Roche Applied Science)) before elution by boiling in 2ϫ SDS

Cardiac syndecan-4 interactome reveals MLP connection
loading buffer. In Fig. 6A, 200 l (6.1 g/l) of rat LV lysate was used per IP, and 5 l of lysate was used as input (2.5% of IP lysate). After precipitation, this IP was washed three times in PBS buffer. In Fig. 6B, 200 l (2.0 g/l) of rat LV lysate was used per IP.
Lysates, immunoprecipitations, and peptides were analyzed on a 4 -15 or 15% Criterion TM Tris-HCl precast gel (3450028 and 3450021, Bio-Rad) and blotted onto a polyvinylidene difluoride membrane (catalog no. 1704157, Bio-Rad). The membranes were blocked in 5% nonfat dried milk or 1% casein in TBS-T (TBS with 1% Tween 20 (161-0781, Bio-Rad)) for 1 h at room temperature. Membranes were thereafter incubated with primary antibodies for 1 h at room temperature or ON at 4°C, followed by three 5-min wash in TBS-T before incubation with horseradish peroxidase-conjugated secondary antibody for 1 h at room temperature. Blots were developed using ECL Prime (RPN 2232, GE Healthcare). When the signal was too high, blots incubated with anti-MLP were washed in TBS-T before repeating development. For reprobing, the membranes were stripped in a Restore TM Western blotting Stripping Buffer (21059, Thermo Fisher Scientific). Densitometric analysis was achieved with ImageJ (National Institutes of Health). Total protein concentration of the various samples was measured using the Micro BCA protein assay kit (23235, Thermo Fisher Scientific). Equal protein amounts were loaded per lane: 20 g of nuclear fraction isolated from mouse hearts; 35.5 g of RIPA extracts isolated from mouse hearts; 25 g of nuclear fractions isolated from virus-transduced cells, and 11 g of nuclear fractions isolated from ISO-stimulated cells.

Large-scale pulldown experiments and immunoprecipitations
For the large-scale pulldown experiments, LV lysates were pooled and mixed with 0.01 mM biotinylated peptides before rotation at 4°C ON. Streptavidin-coated Dynabeads (Dynabeads TM M-270 streptavidin, catalog no. 65305, Life Technologies, Inc.) were washed three times in PBS before adding the peptide containing LV lysate and rotated for 40 min at room temperature. The beads were washed five times in 1 ml of PBS, and protein complexes were eluted in 250 l of 25 mM biotin for 3 h at 60°C. Proteins were precipitated in 1 ml of 4ϫ ice-cold acetone and added glycoblue ON at Ϫ20°C. After centrifugation, the pellets were air-dried before analysis by MS.
For the large-scale immunoprecipitation experiments, 10 g/mg beads of anti-syndecan-4 or anti-rat IgG were coupled to magnetic dynabeads (Dynabeads TM antibody coupling kit, catalog no. 14311D, Thermo Fisher Scientific) according to the manufacturer's protocol. The antibody-coupled beads were incubated with LV lysates from WT, ABHF, or SHAM rat hearts and rotated ON at 4°C. After washing three times in ice-cold PBS and two times in ice-cold MQ water to remove salts (52), interacting proteins were eluted in 0.1% TFA in 50% acetonitrile for 30 min while rotating at room temperature. The elution step was repeated once with fresh TFA before precipitation and centrifugation, and finally, the pellets were air-dried.

Viral transduction
H9c2 cells were cultured to a confluency of about 80% before being transduced with adenovirus serotype 5 encoding mouse syndecan-4 (catalog no. ADV-271493, Vector Laboratories) or empty vector (null virus) (catalog no. 1300, Vector Laboratories). Virus titer used was 2 ϫ 10 7 plaque-forming units/ml. After 24 h of incubation, the cells were washed in PBS and freshly added media. 48 h after induction, the cells were fractionated as described above.

RNA isolation and quantitative real-time PCR (qRT-PCR)
RNA was isolated from LV tissue using RNeasy mini (catalog no. 74104, Qiagen Norge, Oslo, Norway). RNA concentration was determined using the Nanodrop ND-1000 spectrophotometer (Thermo Fisher Scientific, Waltham, MA). For cDNA synthesis, iScript cDNA synthesis kit (catalog no. 170-8891, Bio-Rad), was used according to the manufacturer's protocol. Gene expression levels were determined using pre-designed TaqMan assays detected on a QuantStudio3 (Applied Biosystems by Thermo Fisher Scientific) and analyzed via the QuantStudio Design and Analysis Software. TaqMan assays used were Rn00561900_m1 (Sdc4) and Rn00820748_g1 (Rpl32) (LifeTech by Thermo Fisher Scientific). SDC4 expression was normalized to RPL32.

Mass spectrometry
The proteins were precipitated using 2D Clean Up kit (80-6484-51, GE Healthcare) according to the manufacturer's instructions. The precipitated proteins were dissolved in 40 l of 0.2% ProteaseMAX TM surfactant, trypsin enhancer (Promega), in 50 mM NH 4 HCO 3 followed by protein reduction, alkylation, and in-solution digestion with trypsin (Promega) ON at 37°C. After digestion, the samples were centrifuged at 14,000 ϫ g for 10 min, and trypsin was inactivated by adding 100 l of 1% TFA, and the samples were again centrifuged at 14,000 ϫ g for 10 min. Peptides were desalted and concentrated before MS by the STAGE-TIP method using a C18 resin disk (3M Empore). The peptides were eluted with 80 l of 80% acetonitrile, 0.1% FA, dried, and solubilized in 7 l of 0.1% FA for MS analysis.

Cardiac syndecan-4 interactome reveals MLP connection
Electron) capillary with 25-cm bed length. The flow rate used was 0.3 l/min, and the solvent gradient was 5% B to 30% B in 120 min and then 90% B wash in 20 min. Solvent A was aqueous 0.1% formic acid, and solvent B was 100% acetonitrile in 0.1% formic acid. Column temperature was kept at 60°C.
The mass spectrometer was operated in the data-dependent mode to automatically switch between MS and MS/MS acquisition. Survey full-scan MS spectra (from m/z 400 to 1200) were acquired in the Orbitrap with resolution r ϭ 70,000 at m/z 200 (after accumulation to a target of 3,000,000 ions in the quadruple). The method used allowed sequential isolation of the most intense multiply-charged ions, up to 10, depending on signal intensity, for fragmentation on the HCD cell using high-energy collision dissociation at a target value of 100,000 charges or maximum acquisition time of 100 ms. MS/MS scans were collected at 17,500 resolution at the Orbitrap cell. Target ions already selected for MS/MS were dynamically excluded for 30 s. General MS conditions were as follows: electrospray voltage, 2.1 kV; no sheath and auxiliary gas flow, heated capillary temperature of 250°C, normalized HCD collision energy 25%. Ion selection threshold was set to 5e4 counts. Isolation width of 3.0 Da was used.
The resulting MS raw files were submitted to the MaxQuant software (version 1.5.2.8) for protein identification using the Andromeda search engine. Carbamidomethyl (C) was set as a fixed modification, and protein N-acetylation and methionine oxidation were set as variable modifications. First search peptide tolerance of 20 ppm and main search error of 6 ppm were used. Trypsin without proline restriction enzyme option was selected, with two allowed miscleavages. The minimal unique ϩ razor peptide number was set to 1, and the allowed false discovery rate was 0.01 (1%) for peptide and protein identification. Label-free quantitation (LFQ) was employed with default settings. The UniProt database with "Rattus" entries was used for the database searches. Perseus software was used for the statistical analysis of the label-free quantitation results. Known contaminants as provided by MaxQuant and identified in the samples were excluded from further analysis. Proteins had to be identified with at least two peptides to be considered for further analysis.

Bioinformatics
For enrichment analysis, the Database for Annotation, Visualization and Integrated Discovery (DAVID), version 6.8, was used. Official gene names were imported into the search tool, and Homo sapiens was chosen for both species and background with the following setting: count (2) and EASE score (0.1) (Modified Fisher Exact p value) (54). For categorizing the interactome into Biological Process, the Human Protein Reference Database (HPRD) was used (55).

Statistics
For quantitative comparisons and determination of significance between the MS samples, LFQ intensities were loaded in Perseus (version 1.4.0.20). Those values were log2-transformed, and a t test was done. For the three large-scale biotin-pS179/syndecan-4 pulldown analyses (each run in triplicates), p values of 0.05 or below were accepted as statistically significant.