Junctin is a transmembrane protein of striated muscles, located at the junctional sarcoplasmic reticulum (SR). It is characterized by a luminal C-terminal tail, through which it functionally interacts with calsequestrin and the ryanodine receptor (RyR). Interaction with calsequestrin was ascribed to the presence of stretches of charged amino acids (aa). However, the regions able to bind calsequestrin have not been defined in detail. We report here that, in non-muscle cells, junctin and calsequestrin assemble in long linear regions within the endoplasmic reticulum, mirroring the formation of calsequestrin polymers. In differentiating myotubes, the two proteins colocalize at triads, where they assemble with other proteins of the junctional SR. By performing GST pull-down assays with distinct regions of the junctin tail, we identified two KEKE motifs that can bind calsequestrin. In addition, stretches of charged aa downstream these motifs were found to also bind calsequestrin and the RyR. Deletion of even one of these regions impaired the ability of junctin to localize at the junctional SR, suggesting that interaction with other proteins at this site represents a key element in junctin targeting.
Junctin is an integral membrane protein localized in the sarcoplasmic reticulum (SR) of skeletal and cardiac muscle cells. It is encoded by the gene aspartate beta-hydroxylase (ASPH, also known as AβH-J-J) that can give rise to three functionally distinct proteins through alternative splicing, i.e. the enzyme aspartyl beta-hydroxylase, junctin and junctate. In humans, two promoters – hereafter referred to as P1 and P2 – drive the transcriptional activation of the AβH-J-J locus, with P2 preferentially activating gene transcription in excitable tissues (Feriotto et al., 2006). Two mRNAs are transcribed starting from the P2 promoter, coding for junctin and for one of the two isoforms of junctate. The two proteins share the first 76 aa coding for the cytoplasmic region, the transmembrane domain, and a short luminal charged acidic region, but differ in the largest part of the luminal domain (Hong et al., 2007).
Junctate is a transmembrane protein expressed in the non-junctional SR within striated muscles, which interacts with the inositol-1,4,5-trisphospate receptor (InsP3R) and transient receptor potential cation (TRPC) channels in HEK293 cells (Treves et al., 2004). The protein displays a high binding capacity for Ca2+ and, when overexpressed, the intracellular Ca2+ content of ER/SR stores was increased (Treves et al., 2004; Divet et al., 2007). More recently, junctate has been identified as a Ca2+-sensing structural component of calcium release-activated calcium channel protein 1 (ORAI1) and stromal interaction protein 1 (STIM1) (Srikanth et al., 2012).
Junctin has originally been purified from a cardiac junctional SR membrane subfraction as a calsequestrin-binding protein, suggesting that it plays an important role in Ca2+ homeostasis (Jones et al., 1995). The role of junctin in striated muscle has been further assessed in overexpression or knockout studies. Junctin overexpression in rat cardiomyocytes results in a decrease in Ca2+ release amplitude and contractility (Gergs et al., 2007). Accordingly, junctin overexpression in mouse atria results in decreased expression levels of triadin and ryanodine receptor type 2 (RyR2), which was paralleled by a decrease of the SR Ca2+ load and depressed SR Ca2+ release in aged mice compared to that in young animals (Kirchhefer et al., 2004, 2006). Electron microscopy analysis of mouse myocardium overexpressing canine junctin showed that calsequestrin polymers are tighter and junctional associations between SR and transverse (T)-tubules are increased in size, suggesting that junctin expression affects the packing of calsequestrin and facilitates the association of SR and T-tubules (Zhang et al., 2001). Junctin knockout mice, by contrast, exhibit increased contractility and Ca2+-cycling parameters in cardiac muscle (Yuan et al., 2007), although no significant changes in Ca2+ signalling were observed in skeletal muscle (Boncompagni et al., 2012). Similarly, acute downregulation of junctin in cardiomyocytes significantly increases contraction and Ca2+ transients (Fan et al., 2008), whereas junctin knockdown in skeletal myotubes results in a decrease in Ca2+ release (Wang et al., 2009). Finally, junctin levels are severely reduced in human failing hearts (Gergs et al., 2007; Altshafl et al., 2011).
The functional activity of junctin in striated muscles is mainly linked to its ability to establish dynamic interactions with calsequestrin and ryanodine receptor (RyR) Ca2+-release channels (Lee et al., 2012; Altshafl et al., 2011; Dulhunty et al., 2009; Beard et al., 2008; Jones et al., 1995; Li et al., 2015; Rossi et al., 2020). Analysis of the primary structure of junctin revealed that it is composed by a short N-terminal cytosolic domain, a transmembrane domain and a C-terminal tail that protrudes into the lumen of the SR. Junctin is structurally homologous to triadin, another protein of the SR. Indeed, the two proteins share a luminal tail enriched in stretches of charged aa that, in triadin, form a KEKE motif able to bind calsequestrin and RyR (Kobayashi et al., 2000; Shin et al., 2000). More recently, multiple additional regions in the C-terminal tail of triadin have been shown to interact with calsequestrin-1 (Rossi et al., 2014a). The structural similarity between junctin and triadin suggests that sites that interact with calsequestrin and/or the RyRs are present in junctin. Indeed, interaction between junctin and calsequestrin has been functionally demonstrated (Beard et al., 2009; Dulhunty et al., 2009; Jones et al., 1995), although the position of the binding sites has not yet been defined.
In this study, we performed GST pull-down experiments on skeletal muscle microsomal proteins to identify sequences in the luminal tail of junctin that interact with calsequestrin. We found that the C-terminal tail of junctin contains multiple sites that interact with calsequestrin-1, calsequestrin-2, RyR and triadin. Deletion of these binding sites in recombinant junctin-GFP proteins affects their localization to the junctional SR in developing myotubes, suggesting that protein interactions at this site contribute to junctin localization.
The C-terminal tail of junctin contains multiple binding sites for calsequestrin-1 and calsequestrin-2
To date, no high-resolution structure is available for junctin on its own or in complex. To understand the structural context of the various regions seemingly important for the function of junctin, and that might mediate binding to calsequestrin, we used the protein structure prediction server Phyre2, which uses remote homology detection methods (Kelley et al., 2015). However, no reliable model could be predicted, suggesting that the structure of junctin does not resemble any currently known protein structure. One possibility is that the bulk of junctin is intrinsically disordered. To confirm this, we used DISOPRED3 (Jones and Cozzetto, 2015), which indicated that the bulk of the sequence, except for the transmembrane helix and a few flanking residues, are likely to be intrinsically disordered (Fig. S1A). In addition, FoldIndex (https://fold.proteopedia.org/cgi-bin/findex; Prilusky et al., 2005), which uses hydrophobicity as an indicator for the ability to fold, predicted that the bulk of the luminal region does not fold on its own.
However, as predicted by the sequence comparison database program COILS (Lupas et al., 1991), the C-terminal tail of junctin most probably contains two predicted coiled-coil domains (Fig. S1A) – despite the predicted disorder. In addition, based on sequence homology with the KEKE motif that, in triadin, binds calsequestrin (Kobayashi et al., 2000; Shin et al., 2000), a KEKE motif was identified within the first predicted coiled-coil domain, i.e. in the region encompassing aa 86–107 of junctin (Fig. S1B). The KEKE motif is defined as a 15 aa long sequence with a content of >60% lysine (K) and aspartic acid (D) or glutamic acid (E), and lacking five charged aa residues in a row (Realini et al., 1994). According to this definition, we predicted a second KEKE motif in junctin between aa 118 and 132 (Fig. S1B).
Based on these results, we generated three plasmids that encode GST fusion proteins, containing aa 49–210 (GST-junctin49-210), 49–140 (GST-junctin49-140) and 118–210 (GST-junctin118-210) of junctin to perform GST pull-down experiments with detergent-solubilized microsomes of mouse hindlimb skeletal muscles (Fig. S1C). Striated muscles express different calsequestrin isoforms; calsequestrin-1 is expressed in all skeletal muscles, where it represents the predominant isoform, whereas calsequestrin-2 is expressed at low levels in fast-twitch muscle and moderately higher levels in slow-twitch muscles. Cardiac muscle expresses only calsequestrin-2 (Fig. S2).
Western blot analysis performed on microsomes prepared from total hindlimb muscles showed that calsequestrin-1 interacted with GST-junctin49-210, GST-junctin49-140 and GST-junctin118-210 (Fig. 1A). To evaluate the ability of the first KEKE motif in the aa region 49–140 of junctin to bind calsequestrin, pull-down assays were performed with a GST-fusion protein lacking aa 86–94 (GST-junctin49-140Δ86-94). As shown in Fig. 1A, deletion of these nine aa abolished binding to calsequestrin-1. A second region homologous to triadin was found in the aa sequence 60–78 of junctin, where six D residues are conserved in both proteins. As shown in Fig. 1A, a GST-junctin fusion protein, whose six conserved D residues had been changed to Q (junctin-GST49-1406Q), was still able to bind calsequestrin-1.
To better define the calsequestrin-binding region(s) located within aa 118–210 of junctin, two GST-fusion proteins covering aa 118–154 (GST-junctin118-154) and aa 155–210 (GST-junctin155-210) were generated and tested in GST pull-down assays. As shown in Fig. 1A, both GST-junctin118-154 and GST-junctin155-210 were equally able to bind calsequestrin-1. We also extended our binding experiments to calsequestrin-2, the only isoform expressed in cardiac muscle (Fig. S2). Pull-down experiments performed with detergent-solubilized microsomes from mouse cardiac muscle revealed that calsequestrin-2, similarly to calsequestrin-1, was also able to bind all GST-junctin proteins, except the deletion mutant GST-junctin49-140Δ86-94 (Fig. 1A).
To verify whether pull down of calsequestrin-1 from skeletal muscle-solubilized microsomes occurs via a direct interaction with junctin, i.e. without the requirement of additional muscle-specific proteins, a plasmid expressing GFP-tagged calsequestrin-1 was transfected into HEK293T cells and probed with GST-junctin fusion proteins (Fig. 1A). Like proteins from skeletal muscle solubilized microsomes, recombinant calsequestrin-1-GFP expressed in HEK293T cells was pulled down by GST-junctin49-140, GST-junctin118-210, GST-junctin118-154 and GST-junctin155-210. Similar results were observed when recombinant calsequestrin-2-GFP was expressed in HEK293T and tested through binding to GST-junctin proteins.
Several findings indicate that junctin forms a quaternary complex with RyR, triadin and calsequestrin (Zhang et al., 1997). We, therefore, probed GST-junctin fusion proteins for their ability to bind triadin and RyR1. Pull-down experiments were performed with solubilized microsomes from skeletal muscle or HEK293T cells expressing either recombinant RyR1 or triadin. In experiments with solubilized microsomes from skeletal muscle, we observed interaction between RyR1 and junctin-GST118-210, even if a weak interaction was also observed with junctin-GST49-210 and junctin-GST155-210 proteins. The same fusion proteins were able to interact with RyR1 expressed in HEK293T cells. All GST-junctin fusion proteins were able to pull down triadin, although weaker interaction was observed with junctin-GST118-154, in HEK293T cells (Fig. 1B).
Junctin colocalizes with calsequestrin-1 in non-muscle cells
To further analyze the interaction between junctin and calsequestrin-1, recombinant proteins were expressed in non-muscle cells to exclude the involvement of other muscle endogenous proteins. An expression pattern compatible with localization to the endoplasmic reticulum (ER) was observed following transfection of HeLa cells with a plasmid encoding a GFP-tagged junctin (junctin-GFP) (Fig. 2A). A comparable pattern was observed when HeLa cells were transfected with plasmids encoding junctate-GFP or triadin-GFP (Fig. 2B or C, respectively). Following transfection of HeLa cells with a plasmid encoding calsequestrin-1, most cells presented a fluorescence signal consistent with either localization at the ER or a combined pattern characterized by the presence of puncta that are likely to reflect assembly of small regions containing polymers of calsequestrin-1, on top of the ER distribution (Fig. 2D,E). Only in a small percentage of cells (6.64±6.12%; mean±s.d.), localization of calsequestrin-1 was restricted to regions of the ER, with a characteristic elongated shape (Fig. 2F) that might reflect the assembly of large domains containing extended calsequestrin-1 polymers (Table 1). Interestingly, coexpression of calsequestrin-1 with junctin-GFP significantly increased the number of cells in which calsequestrin-1 was distributed within these elongated linear regions of the ER (38.14%±11.93, P=0.0270). In addition, the junctin-GFP fluorescence signal was redistributed completely, i.e. from a diffused ER pattern to elongated regions containing calsequestrin-1 (Fig. 2G). By contrast, of cells coexpressing calsequestrin-1 and either triadin-GFP or junctate-GFP, only a modest, not statistically significant, number presented elongated regions of the ER that contained calsequestrin-1 (15.65%±6.70, P=0.2615 or 14.83%±3.69, P= 0.1330, respectively; Fig. 2J-O).
Evidence of interaction between calsequestrin-1 and junctin was also obtained when studying the dynamic properties of GFP-tagged proteins by Fluorescence Recovery After Photobleaching (FRAP) experiments (Fig. 2P). Cells showing low levels of fluorescence were selected to avoid potential problems due to GFP overexpression. Reticular calsequestrin-1-GFP showed a very high mobile fraction (98.05±5.5%) that was significantly reduced when the protein localized to elongated regions of the ER (47.9±31.5%). No differences were observed in the mobile fraction of polymeric calsequestrin-1-GFP measured in the presence or absence of junctin (47.91±31.46% and 26.67±9.67%, respectively). By contrast, a significant decrease of junctin-GFP mobility was observed in the presence of calsequestrin-1, compared to that measured when junctin-GFP was expressed alone (90.6±13.1% compared to 70.2±15.9%, respectively).
Previous studies by Lee and co-workers have shown that junctin favors calsequestrin-2 decondensation during Ca2+ depletion of the SR (Lee et al., 2012). To evaluate dynamics of calsequestrin-1-GFP polymers, we expressed calsequestrin-1 alone or together with junctin in HeLa cells treated with ionomycin and followed the distribution of calsequestrin-1 in the ER (Fig. 2Q-R; Fig. S4). Upon depletion of Ca2+in the ER, calsequestrin-1-GFP polymers underwent a rapid disassembly; in most cells GFP fluorescence, initially observed in linear structures, redistributed in puncta and, finally, reached a diffuse distribution in the ER. Interestingly, quantitative analysis showed that, in cells expressing only calsequestrin-1-GFP, elongated structures and puncta persisted for longer in the ER, compared to cells coexpressing calsequestrin-1-GFP and junctin (Fig. 2Q-R; Fig. S4).
Calsequestrin-1 and junctin negatively regulate caffeine-induced Ca2+ release in HEK293 cells expressing recombinant RyR1
Calsequestrin and junctin have been described to functionally regulate the RyR1 Ca2+ channel in single-channel lipid bilayer experiments (Wei et al., 2009; Beard and Dulhunty, 2015). To evaluate the direct effect of calsequestrin-1 and junctin, and the complex between calsequestrin-1 and junctin on RyR1 channel activity, HEK293 cells stably expressing recombinant RyR1 were co-transfected with calsequestrin-1-GFP and junctin-GFP fusion proteins (Fig. S3). Cells were loaded with the ratiometric Ca2+ indicator Fura2-AM and stimulated with 10 mM caffeine. Compared with control cells, the peak magnitude of Ca2+ release observed upon caffeine stimulation was significantly reduced in cells expressing calsequestrin-1-GFP or junctin-GFP, or coexpressing calsequestrin-1-GFP and junctin-GFP, suggesting that calsequestrin-1 and junctin exert an inhibitory effect on RyR1 activation.
Junctin interacts with calsequestrin-1 in differentiating rat myotubes
Assembly of mature triads, i.e. the structures formed by a T-tubule and two terminal cisternae of the SR, is a multistep process beginning in the embryonic life and proceeding until the first weeks after birth. These steps can be recapitulated following in vitro differentiation of primary rat myotubes (Rossi et al., 2019). To study junctin expression and interaction with calsequestrin-1 in muscle cells, we coexpressed YFP-tagged junctin (junctin-YFP) and CFP-tagged calsequestrin-1 (calsequestrin-1-CFP) in primary rat myoblasts, and measured protein interaction by Fluorescence Resonance Energy Transfer (FRET) after 4, 8 and 12 days of differentiation. Fig. 3A,B shows the typical distribution of junctin-GFP during myotube differentiation. At day 4 of differentiation junctin-GFP presented a diffuse distribution in the SR of myotubes, indicating that, at this stage, the SR and the T-tubule systems are still immature (Fig. 3A). The typical triadic distribution of junctin-GFP is observed from day 12 onwards, when it arranges in double rows resembling its localization observed in adult skeletal muscle fibers (Fig. 3B). During differentiation, the percentage of FRET between junctin-YFP and calsequestrin-1-CFP progressively increased from day 4 to day 12 of differentiation (Fig. 3C). The same trend, although not statistically significant, was observed for the percentage of FRET measured between triadin-YFP and calsequestrin-CFP (Fig. 3D). Finally, FRET analysis of junctin-CFP and triadin-YFP showed that interaction between these proteins can be detected already at day 4 of differentiation without further increase over the following days (Fig. 3E).
Different regions in the luminal sequence of junctin contribute its to localization to the junctional SR
To investigate the role of the intraluminal region of junctin regarding its localization to the junctional compartment of the SR, GFP-tagged full-length junctin or junctin deletion mutants lacking selected regions of the luminal tail were expressed in primary myoblasts (Fig. 4). Expression of full-length junctin-GFP resulted in a fluorescence signal superimposable to that of the endogenous RyRs, used as a marker of the junctional SR compartment at triads (Fig. 4A-C). A GFP-fusion protein deleted of the region comprising aa 103–210 (junctin Δ103-210-GFP), did not localize to the junctional SR but formed large transversal bands in correspondence of the isotropic (I) band of the sarcomere (Fig. 4D-F), suggesting that targeting of junctin at the junctional SR requires the presence of the entire luminal region. To confirm this hypothesis, we expressed chimeric proteins composed of sarcolipin, an integral protein of the longitudinal SR, and either the region of junctin comprising aa 103–210 or aa 53–210. Fig. 4G-I shows the localization of GFP-sarcolipin to the longitudinal SR, with a typical localization, resembling the organization of the longitudinal SR around the I and M bands. A chimeric protein composed of the entire sarcolipin sequence followed by the region comprising aa 53–102 of junctin, localized to the longitudinal SR, without significant difference to sarcolipin (Fig. 4J-L). By contrast, a chimeric protein in which the region comprising aa 53–210 of junctin was fused in-frame with the C-terminal region of sarcolipin, localized to the junctional SR, indicating that the luminal sequence of junctin can redirect a protein of the longitudinal SR to the junctional domain of the SR (Fig. 4M-O).
We next wanted to verify which region in the intraluminal tail contributes to junctin targeting to the junctional SR. For this, GFP-junctin proteins deleted of selected region of the luminal tail were expressed in differentiating myotubes. A junctin-GFP fusion protein deleted of the aa region 53–102, containing the KEKE motif homologous to the calsequestrin-binding domain in triadin, resulted in protein localization to irregular large puncta distributed throughout the entire cell (Fig. 4P-R). Similarly, a junctin-GFP fusion protein deleted of the aa region from 86–94 (junctin Δ86-94-GFP), which was shown to be required for calsequestrin binding in GST pull-down experiments, was unable to localize to the junctional SR (Fig. 4S-U). Deletion of aa region 53–76, containing the six D residues conserved in the aa sequences of junctin and triadin, resulted in partial localization of the protein to puncta and the junctional SR (Fig. 4V-X). Finally, a junctin-GFP protein, in which the six D residues were mutated to Q (junctin-6Q GFP) was also unable to localize to the junctional SR (Fig. 4Y-AA).
To verify the role of the region between aa 96–210 in targeting junctin to the junctional SR, we expressed three plasmids encoding junctin proteins that lack aa 96–168 (junctin Δ96-168-GFP), 169–210 (junctin Δ169-210-GFP) or 190–220 (junctin Δ190-210-GFP). The first two proteins were unable to localize to the junctional region of the SR but covered most of the I band – as expected for a protein of the longitudinal SR (Fig. 4AB-AD and AE-AG). However, deletion of the last 20 aa of junctin did not affect junctin localization to the junctional SR (Fig. 4AH-AJ).
At the junctional SR of striated muscles, junctin forms a multiprotein complex that includes RyR, triadin and calsequestrin. Published data clearly indicate that junctin plays a role in the structural organization of the junctional SR where, together with triadin, it mediates retention of calsequestrin in close proximity to RyRs (Tijskens et al., 2003; Zhang et al., 2001; Boncompagni et al., 2012; Lee et al., 2012).
Since most of interactions that link junctin with proteins of the junctional SR are expected to occur around the luminal tail, we focused our attention to this region, which represents three quarters of the length of the entire protein sequence. Here, we report our results of experiments aimed at identifying domains in the luminal region of junctin, involved in establishing interactions with calsequestrin, triadin and RyR1, and in its localization at the junctional SR.
Pull-down experiments performed on solubilized mouse skeletal muscle microsomes showed that GST-junctin49-140 interacts with calsequestrin-1. Computer-aided analysis of the junctin aa sequence revealed the presence of two coiled-coil domains. Within the first one, a KEKE motif (K1 in Fig. S1B) within the region comprising aa 86–107 is significantly similar to the KEKE motif identified in triadin, which represents the first identified site of interaction with calsequestrin. According to Realini and co-workers, KEKE motifs are sequences of ≥12 aa that are devoid of W, Y, F or P residues, contain >60% of K and E/D, and lack five positively or negatively charged residues in a row (Realini et al., 1994). On this basis, we identified in the first coiled-coil domain, a second minimal region corresponding to a KEKE motif (K2 in Fig. S1B) in the region from aa 118–132. Although both KEKE motifs are present in GST-junctin49-140, deletion of half of the K1 sequence in GST-junctin49-140Δ86-94 resulted in loss of interaction with calsequestrin-1 – despite the presence of the K2 motif. This indicates that efficient binding of GST-junctin49-140 to calsequestrin-1 either requires the entire coiled-coil domain – including both K1 and K2 – or that the K2 motif alone is not functional. Our observation that the luminal tail of junctin does not fold on its own, let us believe that our approach to delete specific regions in junctin was unlikely to have interfered with any 3D fold of junctin itself. Most likely, parts of junctin become ordered when bound to its binding partner(s).
We also observed binding between calsequestrin and the GST-junctin155-210 protein, which was contained within a second coiled-coil domain of junctin. The C-terminal region of junctin has also a substantial number of K and E/D residues, even if their distribution does not conform to the definition of a KEKE motif according to Realini and co-workers. This suggests, in addition to KEKE motifs, the existence of two coiled-coil domains that support the formation of homomeric interactions (resulting in junctin oligomers) or of heteromeric interactions with other proteins that contain predicted coiled-coil domains. Although structure analyses of calsequestrin have not revealed any coiled coil regions, it is interesting to note that triadin contains several predicted coiled-coil domains in its luminal tail. Indeed, binding to calsequestrin has been reported within several regions of the luminal tail of triadin (Rossi et al., 2014a).
Experiments using recombinant calsequestrin-1 or -2 expressed in HEK293T cells provided results almost identical to those observed with muscle extracts, indicating that the observed interactions represent direct protein–protein contacts between calsequestrin-1 or -2 and junctin.
Additional pull-down assays showed that the intraluminal tail of junctin also interacts with RyR1 and triadin. Interaction with RyR1 was narrowed down to the region between aa 155–210. Previous work had identified binding sites for RyR1 within the cytoplasmic (aa 1–22) and in the luminal region (aa 86–107) of junctin (Li et al., 2015), whereas binding sites for RyR2 had been identified in junctin aa regions 47–77 and 78–210 (Altshafl et al., 2011). Since we focused in our studies on the interactions between the luminal junctin sequence and other SR proteins, we did not test the cytoplasmic region of junctin. In our studies, no interaction between the GST-junctin49-140 protein and RyR1 was observed. However, interaction with RyR1 was observed with GST-junctin155-210, which partially overlaps with the aa sequence 78–210 described by Altshafl et al. (2011). Experiments with solubilized skeletal muscle microsomes indicated that triadin was also efficiently pulled down by all GST-junctin fusion proteins tested. Interestingly, interaction with triadin was also observed with GST-junctin49-140Δ86-94 fusion protein that did not bind calsequestrin. This might be due to the three-dimensional conformation of junctin-GST49-140Δ86-94, showing differences in binding capacity and/or binding affinities for calsequestrin and triadin. Interactions between triadin and junctin were also tested by expressing recombinant triadin proteins in HEK293T cells. The obtained results confirmed our data gained with skeletal muscle microsomes, even though the interaction between triadin expressed in HEK293T cells and GST-junctin-118-154 was weaker than that observed with muscle microsomes.
The establish interactions with junctin has been proposed to influence calsequestrin polymerization (Zhang et al., 2001; Lee et al., 2012; Perni et al., 2013). However, whereas overexpression of junctin has been shown by Zhang et al. (2001) to result in increased calsequestrin polymerization, Lee and co-workers showed that the presence of junctin favors depolymerization of calsequestrin-2 after Ca2+ depletion (Lee et al., 2012). In our experiments, interaction between junctin and calsequestrin-1 was also observed following transfection into HeLa cells. When expressed alone, both junctin and calsequestrin-1 were mostly found to be homogeneously distributed in the ER. However, when junctin and calsequestrin-1 were coexpressed, they colocalized to regions of the ER in a speckle-like structure likely to reflect areas of polymerization, suggesting the existence of a stable interaction between the two proteins within the ER lumen. Indeed, the mobile fraction of junctin was found to significantly decrease when measured in cells coexpressing calsequestrin-1 organized in a polymeric form. In line with results obtained by Lee and co-workers, we also found that the presence of junctin favors calsequestrin-1 decondensation in HeLa cells following ER Ca2+ depletion, further supporting the hypothesis that interaction between these two proteins is important for the intracellular control of Ca2+ homeostasis (Lee et al., 2012). The effect of calsequestrin-1, junctin and of the junctin-calsequestrin-1 complex on RyR1 channel activity was evaluated by caffeine-induced Ca2+ release experiments performed in HEK293 cells stably expressing RyR1 channels. Single channel lipid bilayer experiments showed that calsequestrin-1 acts as a negative regulator of RyR1 opening (Beard and Dulhunty, 2015; Wei et al., 2009). However, Qin and co-workers reported that calsequestrin-1 had no effect on RyR1 channels (Qin et al., 2009). Data regarding the effect of junctin on RyR1 are more controversial, since either inhibition or activation of channel opening were reported and, in junctin knockout mice, no effect was observed in Ca2+-release experiments (Beard and Dulhunty, 2015; Wei et al., 2009; Li et al., 2015; Boncompagni et al., 2012). RyR1 inhibition was also confirmed by our experiments in HEK293 cells, where coexpression of RyR1 with calsequestrin-1 or junctin resulted in a decreased peak amplitude of Ca2+ release following 10 mM stimulation with caffeine, compared to cells expressing only RyR1 channels. Coexpression of calsequestrin-1 and junctin resulted in a caffeine-induced Ca2+ release comparable to that observed in cells expressing either one of the two proteins, suggesting that – at least in a non-muscle cell environment – the effects junctin and calsequestrin-1 have on regulation of RyR1 are not additive.
We finally investigated whether the regions that mediate the interaction with calsequestrin-1 also affect its localization to the junctional SR in skeletal muscle cells. Tijskens et al. have shown that, in developing cardiomyocytes, junctin is not necessary for localization of calsequestrin to nascent junctional SR (Tijskens et al., 2003). We expressed GFP-tagged full-length junctin and selected deletion mutants in primary rat skeletal myoblasts. As expected, full-length junctin localized to the junctional SR in myotubes differentiated for 12 days. However, a GFP-tagged mutant lacking aa 53–102, thereby only containing the K1 motif, localized to puncta randomly distributed along the entire SR. Deletion of aa 53–76, resulted in partial junctin localization to puncta and junctional SR. The region comprising aa 53–76 is located upstream K1 and, according to Altshafl et al. (2011), contains one RyR1-binding site. However, we were unable to detect this binding site under our experimental conditions. Mutagenesis of six D residues within aa 60–78 also resulted in protein mislocalization. Similarly, deletion of aa 86–94 that, in GST pull-down experiments, were shown to be required for binding to calsequestrin-1, resulted in only a partial localization of the protein at the junctional SR.
Deletion of regions 118–154 and 155–210, that can bind calsequestrin-1 in GST pull-down experiments, resulted in GFP-junctin proteins unable to localize at the junctional SR, whereas a junctin-GFP protein lacking aa 190–210 maintained the ability to localize at the junctional SR.
Altogether these data suggest that deletion or mutagenesis of the aa within the RyR1-binding site described by Altshafl et al. (2011) resulted in perturbed protein targeting to the junctional SR – despite the presence of intact calsequestrin-binding domains. In the same way, deletion of only one of the calsequestrin-1-binding regions (aa 118–154 and 155–210) identified in GST pull-down experiments prevents the correct targeting of junctin at the junctional SR.
In conclusion, we report binding of calsequestrin to three regions within junctin (aa 49–140, 118–154 and 155–210). Triadin also interacted with these three regions, whereas RyR1 binding was restricted to aa 155–210. Any deletion within the luminal tail of junctin, except for the most-C-terminal 20 aa, prevented localization to the junctional SR domain. In agreement with previous published results, localization of junctional proteins to this domain within the SR is likely to be caused by protein–protein interactions, probably started by contacts with junctophilins that can be seen as the structural organizers of triads (Rossi et al., 2019). Accordingly, triadic localization of junctin, which – even in differentiated triads– displays a relatively high mobility, might depend on the combination of multiple interactions with other junctional SR proteins, such as calsequestrin, triadin and RyR1.
MATERIALS AND METHODS
All reagents used were from Sigma-Aldrich (St Louis, MO), unless otherwise specified.
Skeletal muscle tissue lysate and microsome preparation
Animals (Mus musculus) were housed in a standard environment maintained at constant temperature and humidity, and with free access to food and water. Animals used for experiments were 4-month-old male C57BL/6J mice. Experimental procedures and sacrifice were performed in agreement with D.lgs. 26/2014 following University of Siena ethical committee (OBPA) approval and MIUR authorization (n° 643/2016-PR). Heart, tibialis anterior, extensor digitorum longus (EDL), gastrocnemius and soleus muscles were dissected and homogenized with the TissueRupture device (Qiagen GmbH, Hilden, Germany) in RIPA lysis buffer (Cell Signaling Technology, Danvers, MA) supplemented with 1 μM final concentration of phenylmethylsulfonyl fluoride (PMSF). Homogenized samples were centrifuged for 10 min at 10,000 rpm (Centrifuge model 5424, Eppendorf, Hamburg, Germany); supernatants were recovered and used for experiments or preserved at −80°C. Protein concentration was determined using the Bradford assay (Bio-Rad, Hercules, CA).
Microsomes were prepared from skeletal and cardiac muscle of 4-month-old male C57BL/6J mice or from HEK293T cells transfected with the plasmids of interest (Rossi et al., 2014a). Tissue samples or cultured cells were homogenized in ice-cold buffer A (0.32 M sucrose, 5 mM HEPES pH 7.4, and 0.1 mM PMSF) using a potter for cells or a homogenizer for tissues. Homogenates were centrifuged at 7000 g for 5 min at 4°C. Supernatants were centrifuged at 100,000 g for 1 h at 4°C (Sorvall Discovery 90 Ultracentrifuge, Hanau, Germany). Microsomes were resuspended in buffer A and stored at −80°C. Protein concentration was quantified using the Bradford protein assay kit (Bio-Rad).
Microsome solubilization and GST pull-down assay
Microsomes were solubilized according to the method previously described (Rossi et al., 2019) with minor modifications. Briefly, microsomes were solubilized at a protein concentration of 1 mg/ml in buffer containing 2% Triton X-100, 1 M NaCl, 1 mM dithiothreitol, 20 mM Tris-HCl pH 7.4 and protease inhibitor mixture for 1 h, at 4°C. Microsomes prepared from HEK293T cells were solubilized at a protein concentration of 1 mg/ml for 3 h at 4°C in buffer containing 10 mM Tris-HCl pH 7.4, 150 mM NaCl, 5 mM EDTA, 1% Triton X-100, 10% glycerol, 1 mM Na3VO4, 1 mM PMSF and protein inhibitor mixture. Solubilized proteins were obtained by centrifugation (Beckman Coulter Ti90 rotor, Brea, CA) at 48,300 rpm for 45 min at 4°C. 250 μg of solubilized microsomal proteins were incubated with 25 μg of GST fusion proteins in a buffer containing 10 mM Tris-HCl pH 7.4, 150 mM NaCl, 5 mM EDTA, 1% Triton X-100, 10% glycerol, 1 mM Na3VO4, 1 mM PMSF and protein inhibitor mixture for 2 h at 4°C. After incubation, the GST fusion protein complexes were washed three times with 20 mM Tris-HCl pH 7.4, 150 mM NaCl, and 0.2% Triton X-100. Bound proteins were eluted by boiling in SDS-PAGE sample buffer and subjected to SDS-PAGE.
Production and purification of GST-fusion proteins
GST fusion proteins were produced in BL21 bacterial strain following induction with 1 mM isopropyl β–D-thiogalactopyranoside (IPTG) for 3 h at 30°C. Cells were harvested, centrifuged at 4000 rpm in a Megafuge 1.0R centrifuge (Haereus, Hanau, Germany) for 10 min at 4°C. The pellet was resuspended in cold buffer containing phosphate buffered saline (PBS), 1% Triton X-100, 20 mM EDTA and lysed by sonication on ice. The soluble fraction was obtained by centrifugation at 13,200 rpm for 15 min at 4°C in a Mini centrifuge (Centrifuge model 5424, Eppendorf, Hamburg, Germany). The fusion proteins were immobilized by incubating 1 ml of the soluble fraction with 100 μl of beads of glutathione-Sepharose 4B (GE Healthcare, UK) for 10 min and washed three times with 1 ml of a buffer containing PBS and 1% Triton X-100. Beads were finally resuspended with an equal volume of PBS.
Western blot analysis
Protein samples were separated by SDS-PAGE (5% or 10%), as described (Rossi et al., 2019). Filters were incubated with the following primary antibodies: rabbit polyclonal anti-triadin (kindly provided by Prof. Isabelle Marty, Grenoble Institut des Neurosciences; Fourest-Lieuvin et al., 2012), used at 1:10,000; mouse monoclonal anti-calsequestrin-1, clone VIIID12, catalogue number MA3-913, batch VD301442, Thermo Fisher Scientific (Waltham, MA), used at a 1:1000; rabbit anti-ryanodine receptor type 1, produced in our laboratory (Giannini et al., 1995), used at a 1:3000; rabbit polyclonal anti-calsequestrin-2, catalogue number C3868, batch 107K4824, Sigma-Aldrich, used at 1:000; mouse monoclonal anti-GFP, clone GF28R, catalogue number MA5-15256, batch TH264296, Thermo Fisher Scientific, used at a 1:2000.
Primary antibodies were diluted in blocking buffer overnight at 4°C with agitation. Filters were washed three times with washing buffer (0.5% non-fat dry milk, 50 mM Tris-HCl pH 7.4, 150 mM NaCl, 0.2% Tween-20) for 10 min each and incubated with horseradish peroxidase-conjugated secondary antibody (ECL anti mouse IgG, catalogue number NA931, batch 9557566 and ECL anti rabbit IgG, catalogue number NA9340, batch 9565032 from GE Healthcare, Chicago, IL) and detected using the ECL system (ECL Western Blotting Substrate, Promega, Madison, WI.
Generation of expression vectors
Human junctin (NM_032467.4), junctate (NM_032468.5), triadin (NM_006073.4), sarcolipin (NM_003063.3) and calsequestrin-1 (NM_001231.5) and 2 (NM_001232.4) cDNAs were cloned into pEGFP-N1, pEGFP-N2 or pEGFP-N3 vectors, according to the translation reading frame (Takara, Kusatsu, Shiga, Japan) using available restriction sites (Rossi et al., 2014a,b). Junctin mutants and sarcolipin chimeric proteins were generated by PCR using specific primers to amplify the regions of interest. The amplified sequences were cloned in pEGFP-N1, pEGFP-N2 or pEGFP-N3 vectors, according to the translation reading frame and sequenced using an ABI Prism 7900 apparatus (Applied Biosystems). Full-length RyR1 was cloned into the pcDNA3.1 vector (Invitrogen, Carlsbad, CA) (Rossi et al., 2002). For FRET analysis, triadin, junctin and calsequestrin-1 cDNAs were cloned into the pEYFP or the pECFP vectors (Takara) (Rossi et al., 2014a). For GST pull-down experiments, junctin deletion mutants were generated by PCR using specific primers to amplify the regions of interest. Primer use were: for GST-Junctin49-140: Forward: 5′-GTTGACTATGAGGAAGTTCTA-3′; Reverse: 5′-CTTCCTGGATAGGTCTGCATC-3′. Primers for GST-Junctin49-210: Forward: 5′-GTTGACTATGAGGAAGTTCTA-3′; Reverse: 5′-GCCGTTTCTTTTCTGGGTATT-3′. Primers for GST-Junctin49-210Δ86-94: Forward: 5′-GTTGACTATGAGGAAGTTCTA-3′; Reverse: 5′-TTCTTTAGTGAGTTCGGCTACCCCACTGGG-3′ and Forward: 5′-CCCAGTGGGGTAGCCGAACTCACTAAAGAA-3′; Reverse: 5′-CCCAACAACAACAATTGCATT-3′. Primers for GST-Junctin118-210: Forward: 5′-GAGAGAAAAAAGGGAAG-3′; Reverse 5′-GCCGTTTCTTTTCTGGGTATT-3′. Primers for GST-Junctin118-154: Forward: 5′-GAGAGAAAAAAGGGAAG-3′; Reverse: 5′-TTTCTCTTTTTCTCTGTC-3′. Primers for GST-Junctin155-210: Forward 5′-GTGGACCTAGAAAAAAGT-3′; Reverse: 5′-GCCGTTTCTTTTCTGGGTATT-3′). The amplified sequences were cloned into the pGEX4T1 vector (GE Healthcare, Chicago, IL).
Primary cultures of rat skeletal muscle cells, culture of cell lines and DNA transfection
Myoblasts were obtained from hind limb muscles of 2-day-old rats (Sprague-Dawley; Envigo, Indianapolis, IN). Cell suspensions were plated on 0.025% laminin coated LabTek chambers (Sarstedt, Nümbrecht, Germany) or 0.025% laminin-coated glass coverslips. Cells were grown at 37°C under 5% CO2. After 2 days, cells were transfected with plasmid DNA, using the Lipofectamine-Plus method (Thermo Fisher Scientific) following manufacturer's instruction. Myoblasts were induced to differentiate with DMEM containing 2 mM L-glutamine, 100 μg/ml streptomycin, 100 U/ml penicillin, 1 mM sodium pyruvate, 1 mM dexamethasone, 0.05 mM hydrocortisone, supplemented with 10% heat-inactivated fetal bovine serum (FBS) and 5% heat-inactivated horse serum.
HeLa and HEK293T cells (ATCC Manassas, VA) were cultured in DMEM supplemented with 10% heat-inactivated FBS, 2 mM L-glutamine, 100 μg/ml streptomycin, 100 U/ml penicillin, 1 mM sodium pyruvate. Cells were transfected using the Lipofectamine-Plus method following manufacturer's instruction.
For localization studies, immunofluorescence staining was performed as described (Rossi et al., 2019). A mouse monoclonal antibody recognizing RyRs was used at a dilution of 1:1000 to detect the junctional SR regions in differentiated myotubes (mouse anti RyR1, clone 34C, catalogue number MA3-925, batch QJ219752, Thermo Fisher Scientific). A mouse monoclonal anti-calsequestrin-1 antibody (clone VIIID12, catalogue number MA3-913, batch VD301442, Thermo Fisher Scientific) at a 1:1000 dilution was used to detect recombinant calsequestrin-1 proteins expressed in HeLa cells. Alexa Fluor Plus 555-conjugated goat anti-mouse secondary antibodies (catalogue number A32727, batch UL287768, Thermo Fisher Scientific, Jackson Laboratories) were used for immunofluorescence detection following manufactorer's instruction.
Fluorescence recovery after photobleaching
FRAP experiments were performed in HeLa cells expressing the GFP fusion proteins of interest, using a confocal laser scanning microscope (ZEISS LSM 510, Carl Zeiss, Jena, Germany). Cells were imaged in 140 mM NaCl, 5 mM KCl, 10 mM glucose, 1 mM MgCl2, 0.1 mM CaCl2, 20 mM HEPES and 0.4 mM EGTA at 37°C. A 63×1.4 NA Plan-Apochromat oil immersion objective was used and cells were imaged with a pinhole aperture of 4.96 Airy units, corresponding to a confocal section thickness of 3.5 μm. GFP fluorescence before photobleaching and its recovery after photobleaching was measured with the 488 nm line of argon laser with low laser power (0.5%). After the acquisition of ten pre-bleach images, a 50% photobleaching was performed using the argon laser lines 458, 477 and 488 nm. The photobleached area consisted in a circle of 1.08 μm in diameter. Recovery was measured by time-lapse imaging at 50–300 ms intervals over a period of 1–10 min, until fluorescence level reached a plateau. Throughout the experiment, fluorescence intensities were acquired for the bleached region (Ifrap), for the whole cell (Iwhole), and for the background (Ibase). Data analyses were performed using the IgorPro software (WaveMetrix Inc. Portland, OR). Statistical analysis was performed using two-tailed Student’s t-test on data from three independent experiments. Variance was not significantly different for values of reticular junctin-GFP and junctin-GFP+polymeric calsequestrin-1 and for reticular calsequestrin-1-GFP and polymeric calsequestrin-1+junctin. Welch's correction was applied for values of reticular calsequestrin-1-GFP and polymeric calsequestrin-1-GFP.
Fluorescence resonance energy transfer
Intracellular Ca2+ measurements and Ca2+ depletion by ionomycin
HEK293 cells stably expressing RyR1 channels (Rossi et al., 2002) were loaded with the ratiometric Ca2+ indicator 5 µM Fura 2-AM in Krebs-Ringer-HEPES medium (125 mM NaCl, 5 mM KCl, 1.2 mM KH2PO4, 1.2 mM MgSO4, 2 mM CaCl2, 6 mM glucose and 25 mM HEPES, adjusted to pH 7.4 with NaOH) for 30 min at room temperature in the dark followed by a 30-min deesterification period. Fura 2 fluorescence was recorded using an inverted-stage microscope (Axiovert 200, Carl Zeiss) with a 40×objective. Fura 2 was excited alternately at 340 and 380 nm, using the pE-340fura CoolLED dual monochromator (Carl Zeiss). Emitted light was filtered at 510 nm and collected with an intensified CCD camera (Axiocam 702 mono, Carl Zeiss). Ratio images were acquired and Ca2+ signaling was analyzed using computer software (Physiology module of ZEN Pro software, Carl Zeiss). Data are represented as the mean±s.e.m. Statistical analysis was performed using two-tailed Student’s t-test with Welch correction.
For Ca2+-depletion experiments, cells were treated with ionomycin at a final concentration of 5 µM in Ca2+-free Krebs-Ringer-HEPES medium. Time-lapse high-resolution images were acquired every 10 s, using a confocal laser scanning microscopy (Zeiss LSM 510 Meta, Carl Zeiss). Distribution of the calsequestrin-1-GFP fluorescence signal was evaluated by counting the number of cells showing linear structures, puncta and a homogeneous distribution in the ER at t=0 and t=190 s after addition of ionomycin. The relative percentage of each class of distribution was calculated and expressed as the mean±s.e.m from three independent experiments. Image analysis was performed with ImageJ software (National Institutes of Health).
Methodology: S.L., F.V.P., D.O.A.; Formal analysis: D.R., F.V.P.; Writing - original draft: D.R., V.S.; Visualization: E.P.; Supervision: V.S.; Project administration: V.S.; Funding acquisition: V.S.
This work was supported by Fondazione Telethon (grant number: GGP19291) and the Ministry of Education, University and Research (MIUR) PRIN 2015 (grant number: 2015ZZR4W3) to V.S.
Peer review history
The peer review history is available online at https://journals.biologists.com/jcs/article-lookup/doi/10.1242/jcs.259185.
The authors declare no competing or financial interests.