A new cytoplasmic interaction between junctin and ryanodine receptor Ca2+ release channels

ABSTRACT Junctin, a non-catalytic splice variant encoded by the aspartate-β-hydroxylase (Asph) gene, is inserted into the membrane of the sarcoplasmic reticulum (SR) Ca2+ store where it modifies Ca2+ signalling in the heart and skeletal muscle through its regulation of ryanodine receptor (RyR) Ca2+ release channels. Junctin is required for normal muscle function as its knockout leads to abnormal Ca2+ signalling, muscle dysfunction and cardiac arrhythmia. However, the nature of the molecular interaction between junctin and RyRs is largely unknown and was assumed to occur only in the SR lumen. We find that there is substantial binding of RyRs to full junctin, and the junctin luminal and, unexpectedly, cytoplasmic domains. Binding of these different junctin domains had distinct effects on RyR1 and RyR2 activity: full junctin in the luminal solution increased RyR channel activity by ∼threefold, the C-terminal luminal interaction inhibited RyR channel activity by ∼50%, and the N-terminal cytoplasmic binding produced an ∼fivefold increase in RyR activity. The cytoplasmic interaction between junctin and RyR is required for luminal binding to replicate the influence of full junctin on RyR1 and RyR2 activity. The C-terminal domain of junctin binds to residues including the S1–S2 linker of RyR1 and N-terminal domain of junctin binds between RyR1 residues 1078 and 2156.


INTRODUCTION
Contraction in the heart and skeletal muscle depends on the release of Ca 2+ from the intracellular sarcoplasmic reticulum (SR) Ca 2+ store. Inherited or acquired changes in Ca 2+ release lead to skeletal and cardio-myopathies and cardiac death (Györke and Carnes, 2008). The efficacy of Ca 2+ release depends an influence of the 'Ca 2+ load' in the SR on the activity of ryanodine receptor (RyR) Ca 2+ release channels, which is facilitated by a Ca 2+dependent interaction between CSQ proteins (also known as CASQ) and the RyR through the 'anchoring' protein junctin, which is a non-catalytic splice variant encoded by the aspartate-bhydroxylase (Asph) gene (Dulhunty et al., 2009;Dulhunty et al., 2012;Wei et al., 2009a). A second 'anchoring' protein, triadin also binds to CSQ and the RyR, but does not transmit signals from CSQ to the RyR in skeletal muscle in vitro (Wei et al., 2009a), although it does influence Ca 2+ release during excitationcontraction coupling in intact myotubes (Goonasekera et al., 2007) and anchoring of CSQ to the junctional SR membrane (Boncompagni et al., 2012). Different isoforms of the RyR, CSQ and triadin are expressed in cardiac and skeletal muscle, whereas the same isoform of junctin is expressed in both muscle types.
The functions of junctin, triadin and CSQ in SR Ca 2+ signalling have been extensively explored in transgenic animals. Triadin-, junctin-or CSQ-null animals survive, but their longevity and ability to tolerate stress is compromised (Altschafl et al., 2011;Dainese et al., 2009;Oddoux et al., 2009). Thus, normal expression of each of the proteins is required for normal function. In contrast to the transgenic studies there are few reports addressing the molecular interactions between either triadin or junctin and the RyR in vitro. The regions of junctin that support functional interactions with RyRs are unknown. Junctin contains a short N-terminal cytoplasmic domain, a single SR membrane-spanning domain and a longer luminal C-terminal domain (Fig. 1A) that contains binding sites for CSQ and the RyR (Altschafl et al., 2011;Kobayashi et al., 2000). The binding site for junctin on RyR1 has not been determined, but two junctin-binding regions on luminal domains of RyR2 have been identified (Altschafl et al., 2011). Our aim here has been to examine the functional interactions between junctin and skeletal RyR1 and cardiac RyR2 channels and to explore the role of the cytoplasmic and luminal domains of junctin in modulating RyR activity. The results show the expected interaction between the luminal domain of junctin (as well as one of its KEKE motifs) with the luminal, but not the cytoplasmic domains, of RyR1 and RyR2 channels. In addition we find an unexpected and substantial interaction between the cytoplasmic domain of junctin and the cytoplasmic, but not the luminal, domains of RyR1 and RyR2.
The cytoplasmic interaction appears to be essential to reproduce the effect of full-length junctin on the RyR channels.
binding to RyRs. FLjun bound to both RyR2 and RyR1 as assessed by anti-junctin co-immunoprecipitation (Fig. 1C, lanes 1 and 2), or when RyRs were conjugated to anti-RyR-antibody-conjugated protein-A/G-agarose (n52-3, included in averages in Fig. 1D). There was similar association between Cjun and RyR2 or RyR1, as assessed by both anti-RyR (Fig. 1C, lanes 3 and 4) and anti-junctin antibody co-immunoprecipitation (n52-3, included in averages in Fig. 1D). Comparable amounts of Cjun and FLjun bound to the RyRs (Fig. 1D). The density of RyR2 bands was lower than RyR1 owing to a reduced antibody affinity for RyR2. No FLjun, Cjun, RyR1 or RyR2 associated with protein-A/G-agarose in the absence of antibody, or in the presence of antibody but absence of epitope (e.g. Fig. 1C, lanes 5 and 6).
We next assessed the binding between RyRs and the cytoplasmic N-terminal domain of junctin (Njun), using coimmunopreciptation of RyR2 by biotinylated Njun (Fig. 1E). As it was assumed that the RyR interactions with junctin occurred only between the luminal domains of the proteins, it was surprising that RyR1 and RyR2 bound to the biotinylated Njun peptide (Fig. 1E, lanes 2 and 5). Given that RyRs isolated from heart or skeletal muscle bound to NeutrAvidin agarose beads, but recombinant RyR1 and RyR2 did not (Fig. 1E, lanes 1 and 4), recombinant channels were used in this experiment. However a similar interaction between muscle-isolated RyRs and the cytoplasmic Njun domain is supported by the functional and binding data described below. We did find that some RyR bound to a scrambled Njun peptide (Fig. 1E, lanes 3 and 6), but this was ,20% of that bound by Njun (Fig. 1F). Neither adding 1% Triton X-100 to the wash buffer, nor increasing the volume of wash buffer or number of washes, altered the binding. This limited binding was considered to be either non-specific or non-functional as the scrambled peptide did not alter channel activity (see Fig. 3). It is perhaps not surprising that scrambled Njun (with an isoelectric point of 9.70, containing two glutamic acids, three lysines and three histadines) associated weakly and non-specifically with RyRs, without functional consequences, as the large cytoplasmic domain of the RyR contains hydrophilic surfaces. Attempts at complimentary coimmunoprecipitation of Njun by RyRs was unsuccessful as the Njun peptide bound to protein-A/G-agarose (n54).
Full-length junctin activates RyR2 and RyR1 channels from heart and skeletal muscle RyR channels (both native RyR channels in SR vesicles or RyR channels purified from SR vesicles) added to the solution on the cis side of the lipid bilayer incorporate into bilayers with their cytoplasmic side facing that solution and luminal side facing the trans solution (Beard et al., 2002;Laver, 2005). Therefore these solutions are referred to as cytoplasmic and luminal solutions, respectively. The average conductances for the RyR channels were 284.869.1 pS for RyR1 or 266.6610.5 for RyR2, which is in the range reported previously with physiological [Ca 2+ ]s of 1 mM cis and 1 mM trans (Laver et al., 1995;Tinker and Williams, 1993;Wei et al., 2009a). Channel activity increased when FLjun was added to the luminal solution at a maximal activating concentration of The lower panel shows a western blot (anti-junctin antibody). (C) Co-immunoprecipitation of RyR1 by FLjun and 6-his Cjun, or RyR2 by FLjun and by 6-his Cjun (lanes 1 to 4). No RyR1 or RyR2 bound to the beads or the anti-junctin antibody in the absence of FLjun or Cjun (lanes 5 and 6). (D) Mean6s.e.m. densitometry values of RyR:FLjun or RyR:Cjun (RyR2, n55; RyR1, n54). *P,0.05 for RyR2 values significantly less than RyR1; P-values below bins are for differences between (a) FLjun and Cjun bound to RyR2 or RyR1, and (b) FLjun bound to RyR1 and FLjun bound to RyR2, or Cjun bound to RyR1 and Cjun bound to RyR2. (E) Co-immunoprecipitation of RyR2 and RyR1 by biotinylated Njun or biotinylated scambled Njun peptide (Njun sc ). (F) Mean6s.e.m. densitometry values for RyR:Njun or RyR:Njun sc (RyR2, n57; RyR1, n58). *P,0.05 for RyR:Njun sc values significantly less than RyR:Njun; P-values below bins (a) are for differences between Njun and Njun sc binding to RyR2 or RyR1.
213 nM [the molar equivalent of 5 mg/ml, as used previously (Wei et al., 2009a)] ( Fig. 2A-C), associated with an increase in mean open time and a decrease in mean closed time (Fig. 2D,E). The open and closed time distributions (t o and t c, respectively) were described by three time constants (t 1, from 1-10 ms; t 2 , from 10-50 ms or t 3 , from 50-500 ms) that were not altered by FLjun, although there was a significant increase in long openings in t o3 (and fewer in t o1 ), and a decrease in long closures in t c3 (with more in t c1 ) (Fig. 2F,G).
In control experiments, FLjun was added to the luminal side of native RyR channels that remained associated with endogenous junctin (see Fig. 4 below and Wei et al., 2009a). The rationale was that if endogenous junctin was bound to native RyRs, then exogenous FLjun added to the luminal solution could not bind to or activate the channels. Consistent with this hypothesis, there was no significant effect of luminal FLjun on either native RyR1 or native RyR2 channels (Fig. 2H).

The N-terminal domain of junctin activates RyR channels
Njun added to the solution on the cytoplasmic side of the lipid bilayer dramatically increased RyR activity (Fig. 3A,B). RyR2  (Sigworth and Sine, 1987). The probability of events falling into each time constant is plotted against the time constant (in ms), for control data (open circles) and with FLjun (filled circles) for RyR2 (F) or RyR1 (G). Horizontal and vertical bars indicate the s.e.m. for the time constant and probability, respectively, and are contained within the symbols if not visible; Pvalues below bins test are for differences between the probability of events (P, upper row) or time constant (t, lower row) with FLjun and control. *P,0.05 between control and FLjun data. (H) Mean6s.e.m. open probability of native RyR2 (n58) and RyR1 channels (n512) exposed to FLjun, relative to open probability prior to exposure; P-values below the bins in C-E and in H are for differences between FLjun and control parameters. The rate of Ca 2+ release from Ca 2+ -loaded SR vesicles after blocking SERCA with thapsigargin. Left, immediate effect of Njun on cardiac SR (n54), and skeletal SR (n54). Right, effect of preincubation with 5 mM Njun on caffeine induced Ca 2+ release through RyR2 (5 mM caffeine, n57) and RyR1 (0.5 mM caffeine, n511). *P,0.05 between control and junctin construct data. In C-F, I-L and M, P-values for differences between Njun or scrambled Njun and control are given below the bins.  Fig. 2, for luminal addition of Cjun to purified RyR2 (n518) and RyR1 (n516) (black bins); cytoplasmic Cjun on purified RyR2 (n532) and RyR1 (n512) (blue bins); cytoplasmic Cjun on native RyR2 (n58) and RyR1 (n56) (orange bin); and cytoplasmic recombinant FLjun on purified RyR2 (n512) and (I) Effect of 5 mM Cjun on the relative rate of Ca 2+ release from SR vesicles immediately after application (left, cardiac n55; skeletal n55) and on caffeine-induced Ca 2+ release after 20 min pre-incubation and 10 min exposure during Ca 2+ loading (right, cardiac n55; skeletal n55). P-values for differences between rates with Cjun and buffer are given below the bins. (J) Western blots probed for RyR, skeletal triadin (Trisk95), CSQ, triadin 2 and junctin. From left to right: cardiac SR (SR C ), purified RyR2, skeletal SR (SR sk ), and purified RyR1. (K,L) Mean6s.e.m. densitometry data from western blots of SR and purified RyR from cardiac (n54) and skeletal (n54) muscle. Mean6s.e.m. RyR:junctin; the P-value for the difference between cardiac and skeletal SR is give below the bin (K). Junctin:RyR ratios show no significant junctin density in the purified preparations; P-values for differences between SR and purified RyR are given below the bins; # P,0.05, reflecting less junctin (L). open probability increased 6.962.4-fold and RyR1 increased 5.061.0-fold (Fig. 3C). This was significantly greater than the 2.860.44 (RyR2, P51.2610 23 ) and 1.960.14 (RyR1, P51.9610 24 ) increases produced by luminal FLjun, but was similarly due to longer open times (Fig. 3D) and briefer closures (Fig. 3E). As with FLjun, there was no change in the open or closed time constant values, but a decrease in the number of longest closed events and increase in the number of longest opening events (Fig. 3G,H). Channel activity was not altered by addition of the scrambled cytoplasmic Njun peptide (Fig. 3F) or by luminal Njun (Fig. 3I), indicating that there is a specific action of Njun on the cytoplasmic side of the RyRs.
To assess whether Njun was bound to RyR1 in vivo, we again utilised native RyR channels in SR vesicles containing endogenous junctin. As with FLjun, the rationale was that if the N-terminal domain of endogenous junctin was bound to native RyRs, then exogenous Njun added to the cytoplasmic solution could not bind to the channels. Indeed there was no significant change in relative gating parameters after adding 213 nM Njun to the cytoplasmic side of native RyRs resident in the membrane of SR vesicles ( Fig. 3J-L). We also examined the effects of Njun on Ca 2+ release from SR vesicles, through native RyRs, which are oriented with their cytoplasmic surface facing the extravesicular solution (Saito et al., 1984). Njun was added at 5 mM because high peptide concentrations are required to instantly alter Ca 2+ release from the SR (Dulhunty et al., 1999). Consistent with the single channel results, Njun did not immediately alter resting Ca 2+ release from cardiac or skeletal SR (Fig. 3M, left). Similarly, 20 min pre-incubation in 5 mM Njun (plus 10 min exposure during Ca 2+ loading steps) did not alter caffeineinduced Ca 2+ release (Fig. 3M, right), which reflects the physiological Ca 2+ -induced Ca 2+ release process (Pagala and Taylor, 1998). Therefore, we suggest that most Njun-binding sites are occupied by the N-terminal tail of endogenous junctin and thus that Njun binding to RyRs might occur in vivo.

The C-terminal domain of junctin inhibits RyR channels
We expected that luminal Cjun would activate RyRs in the same way as FLjun, due to an assumed dominance of the luminal interaction between the proteins. Contrary to expectation Cjun inhibited the channels ( Fig. 4A-C, solid black bins). The open probability (P o ) fell to 0.4860.06 of control in RyR2 and to 0.6860.17 in RyR1, due to a small reduction in open times (Fig. 4D) and a larger increase in closed times (Fig. 4E). The open and closed time constants were not affected by Cjun, but there were fewer long open events and more long closed events (Fig. 4F,G). This was likely to be a specific effect of the native Cjun because denaturing the protein by boiling for 10 min removed its effects on channel open probability (Fig. 4H).
Cjun added to the cytoplasmic (cis) solution caused a small significant decline in channel activity ( Fig. 4C-E, blue bins), which was significantly less than the fall in activity with Cjun in the luminal solution (RyR2, P51.8610 23 ; RyR1, P50.047). This was not due to Cjun crossing the bilayer to access luminal RyR domains, as subsequent addition of luminal Cjun produced normal inhibition and, conversely, luminal addition of Cjun did not prevent the effect of subsequent cytoplasmic addition (n56, RyR1 and n512, RyR2). To determine whether the C-terminal domain of endogenous junctin bound to the cytoplasmic side of native RyRs, cytoplasmic Cjun was added to native channels where it inhibited RyR2 but not RyR1 (Fig. 4C-E, orange bins). In contrast, however, the rate of Ca 2+ release from SR vesicles was unaffected by 5 mM Cjun either immediately after addition or during caffeine-induced Ca 2+ release after a 20 min preincubation plus 10 min exposure during Ca 2+ loading (Fig. 4I). Finally, recombinant FLjun with a His-tag-protected N-terminus added to the cytoplasmic side of purified channels, did not consistently change gating ( Fig. 4C-E, grey bins). These inconsistent results suggest that any cytoplasmic effect of Cjun is weak and non-specific. As with scrambled Njun, this is not surprising given the overall highly charged nature of both Cjun and the cytoplasmic RyR domains.
The strong action of Cjun on the luminal side of purified RyRs was not due to the peptide binding to residual CSQ associated with the solubilised RyR as there is no detectable CSQ, triadin or junctin in these preparations ( Fig. 4J-L).

Combined actions of Cjun and Njun on RyR channels
Given that junctin appears to regulate RyRs through its N-and C-terminal domains, we examined the combined effect of cytoplasmic Njun plus luminal Cjun. Njun added first to the cytoplasmic solution increased RyR activity (1st black bin, Fig. 5A-F) to levels significantly greater than luminal FLjun (cyan bins). Subsequent luminal addition of Cjun reduced the open probability (2nd black bin), to levels that remained greater than control, but were not significantly different from those with luminal FLjun.
In the complementary experiment, P o fell when luminal Cjun was added first, but then increased with cytoplasmic Njun addition, although the activity remained less than that for the control (grey bins, Fig. 5A-F). This was in contrast to the greater than control activity with cytoplasmic Njun alone ( Similar changes in open probability with FLjun (luminal), or Njun (cytoplasmic) followed by Cjun (luminal), were mainly due to changes in closed times (Fig. 5C,F, red ovals). These results suggest that increased RyR activity with luminal FLjun might depend on the cytoplasmic N-tail binding first, followed by interactions with the luminal C-domain.
The data in Fig. 5 support our assumptions (1) that FLjun inserts into the bilayer, as do RyRs and variety of other membrane proteins, and (2) that after insertion, the short Nterminal domain is exposed to the cytoplasmic solution, while the highly charged C-terminus remains on the luminal side. The consistent increase in RyR activity after FLjun addition to the luminal solution suggests that FLjun routinely inserts into the bilayer in the same orientation.

Regions in RyR1 that interact with junctin
Cjun-and Njun-binding sites on RyR1 were examined using deletion mutations of rabbit RyR1 ( Fig. 6A; Fig. 7A, see Materials and Methods). Although RyR preparations were obtained from different species for reasons of availability and compatibility with affinity chromatography materials, the results appear to be largely species independent. It is notable that FLjun, Cjun and Njun bound to full-length RyR1 from rabbit skeletal muscle, full length recombinant mouse RyR2 (Fig. 1C-F), full-length recombinant rabbit RyR1 (Fig. 6B,C; Fig. 7B,D) and interacted consistently with full-length RyR2 from sheep heart (Figs 2-5).

RyR1-binding sites for Njun
Njun bound to all RyR1 constructs containing residues 182-2156 (I, IX, X and XI; Fig. 7B,C), with little binding to the scrambled Njun peptide. Construct XII bound similarly to Njun and the scrambled Njun peptide (Fig. 7B), suggesting non-specific binding. None of the I, IX, X, XI or XII constructs bound to neutravidin-agarose in the absence of peptide. However the 1-1078 construct (XIII) bound to neutravidin agarose and to the scrambled Njun peptide in greater amounts than Njun (Fig. 7B).
Construct 1079-2156 (XIV) bound to Njun only, not to the scrambled Njun peptide or neutravidin agarose (Fig. 7B), indicating highly specific binding. To asses specific Njun binding, the scrambled Njun peptide densities for each construct were subtracted from Njun densities (Fig. 7C). The results indicate that an N-terminal RyR1-binding site for Njun lies between residues 1097 and 2156 (XIV). Although Njun appears not to bind specifically to residues RyR1 1-1097 (XIII), a conclusion for this construct is complicated by its binding to streptavidin agarose.

RyR-binding residues in Cjun
Cjun binds to the pore loop of RyR2 (Altschafl et al., 2011), a region of high charge density that, in RyR1, also binds to triadin (Goonasekera et al., 2007). A KEKE motif in triadin has been implicated in binding to RyR1 ; hence, we explored RyR interactions with a similar KEKE motif that is present in the C-terminal junctin residues 86-107 (Fig. 8A,B). The KEKE peptides (Fig. 8B), like Cjun, added to the luminal solution, inhibited RyR channels with significant increases in

(B) Co-immunoprecipitation of recombinant RyR1 and RyR1 fragments by
Cjun bound to the anti-junctin antibody conjugated to protein-A/G-agarose. RyR fragments were identified by using anti-GFP antibody, Cjun was identified by using antijunctin antibody. (C) Mean6s.e.m. density of RyR1 constructs, relative to the density of Cjun in the same lane as shown in B (n53-5). *P,0.05 for differences between WT RyR1 and indicated fragment, P-values are given beneath the bins.  , which has a similar KEKE content (Fig. 8B). These opposite actions were not explored further, but might be due to the slightly different KEKE sequence in triadin and junctin, or to the extended hydrophobic C-terminal tail of the triadin peptide allowing additional RyR interactions. In contrast to Cjun, and similar to the triadin KEKE 200-232 peptide , the KEKE peptide in the cytoplasmic solution had no effect on RyR channels (Fig. 8C-E). In conclusion, the KEKE motif in the junctin 86-107 region can bind to RyR1 and RyR2 and reproduce the effects of Cjun on the luminal side of the channels. Affinity chromatography reveals a strong association between the KEKE peptide and purified RyR1 or RyR2 (Fig. 8F,G).

DISCUSSION
Here, we report a new in vitro interaction between the N-terminal cytoplasmic residues of junctin and the cytoplasmic domains of RyR1 and RyR2, in addition to the expected luminal interactions. We show that a binding site for the N-terminal domain of junctin resides in residues 1078-2156 of RyR1 and suggest that this cytoplasmic binding must be established before the luminal interaction in order to replicate the normal regulation of RyR channels by junctin. In addition to the cytoplasmic binding site, the results indicate that there is a luminal RyR1 binding site for junctin between residues 4535-4830, which is the region containing the S1-S2 linker, which was already known to bind to junctin in human RyR2 (Altschafl et al., 2011). The N-and Cterminal parts of this linker region are homologous in rabbit RyR1 and RyR2, although the overall similarity in residues 4535-4830 is only 75% (BLASTP 2.2.30+; supplementary material Table S4). The very similar effects of junctin on RyR1 and RyR2 activity indicate that the functional binding is to regions conserved in the two RyR isoforms. Finally, this work reveals differences between the functional outcome of junctin and triadin interactions with RyRs, supporting the idea that two 'anchoring' proteins have independent roles in Ca 2+ homeostasis. These differences include (1) several binding sites for junctin, in contrast to a single triadin-binding site in the RyR pore loop (Goonasekera et al., 2007), (2) RyR activation by the N-terminal domain of junctin, in contrast to inhibition by N-terminal triadin residues (Groh et al., 1999), and (3) the luminal KEKE-rich region of junctin (amino acids 86-107) inhibits RyRs, whereas the equivalent triadin region (amino acids 200-232) activates RyR1 . These differences are consistent with reports  Fig. 2)  that junctin communicates signals from CSQ1 (CASQ1) to RyR1 (Wei et al., 2009a), whereas triadin influences excitationcontraction coupling (Goonasekera et al., 2007).

The significance of N-terminal junctin binding to RyRs
The results suggest that the cytoplasmic site is physiological as it appears to be occupied by the endogenous junctin N-tail in the native channels. As noted above, we suggest that when FLjun is added to the luminal bilayer solution, its hydrophobic transmembrane domain (residues 23-43) inserts into the bilayer such that the short N-terminus is located in the cytoplasmic solution and the C-terminal domain remains in the luminal solution (Wei et al., 2009a). Given that the actions of luminal FLjun can only be replicated when Njun is added first to the cytoplasmic solution and then Cjun to the luminal solution, we suggests that FLjun insertion into the bilayer might allow the cytoplasmic interaction to anchor FLjun to the RyRs and facilitate the luminal interaction.

Similar actions of junctin on RyR1 and RyR2
It is notable that FLjun, Njun and Cjun bind similarly to cardiac and skeletal RyRs, with little RyR isoform dependence in the functional effect of either cytoplasmic or luminal interactions. This is not surprising as the same junctin splice variant of the aspartate-b-hydroxylase gene is expressed in the heart and skeletal muscle. Although there is considerable overall sequence diversity between RyR1 and RyR2, some regions of high similarity exist. One explanation for the similar results is that the junctin-binding sites are in regions of sequence similarity in RyR1 and RyR2. Curiously, CSQ1 inhibits RyR1 by binding to junctin (Wei et al., 2009a), whereas CSQ2 (CASQ2) activates RyR1 and RyR2 (Wei et al., 2009b). If RyR1 or RyR2 activation by CSQ2 is also mediated through junctin, CSQ1 and CSQ2 must cause different conformational changes in junctin.

Multiple junctin-binding sites on RyRs
The C-terminal domain of junctin binds to the S1-S2 (M5-M6) linker and the pore loop of RyR2 (Altschafl et al., 2011) and is likely to bind to the same regions in RyR1. We found that the residues including the largely uncharged S1-S2 linker region of RyR1 bound to Cjun. The KEKE motif could well interact with charged residues in the pore loop as does the similar KEKE motif in triadin (Goonasekera et al., 2007;Wium et al., 2012). Both the cytoplasmic RyR binding site and S1-S2 linker binding would help sustain full-length junctin binding to mutant RyRs with acidic residues in the pore loop replaced by alanine (Goonasekera et al., 2007). KEKE motifs in junctin and triadin are implicated in binding not only to RyRs but also to CSQ. The same KEKE motif in triadin reportedly binds to both CSQ and the RyR, raising the possibility that both cannot bind to triadin at the same time (Dulhunty et al., 2009). That triadin cannot convey signals from CSQ1 to RyR1 could be explained if CSQ1 binding to triadin disrupted the association of triadin with RyR1 (Wei et al., 2009a). In contrast to triadin, junctin could convey signals from CSQ1 to RyR1 (Wei et al., 2009a), due to multiple RyR-junctin (this manuscript) and junctin-CSQ (Kobayashi et al., 2000) binding sites. Thus, junctin could remain anchored to RyR1, while also binding CSQ1 and allowing CSQ1 to influence RyR1 activity (Wei et al., 2009a).
There is a complex interaction between triadin, junctin, CSQ and RyRs in living cells. Our previous studies in myotubes (Goonasekera et al., 2007) and other studies in C2C12 cells (Wang et al., 2009) suggest that triadin is more involved in excitation-contraction coupling, whereas junctin supports resting RyR activity and CSQ regulation of RyRs (Wei et al., 2009a). However junctin knockout, triadin knockout or triadin and junctin double-knockout suggest that the triadin-CSQ complex has a greater impact on junctional structure than the junctin-CSQ complex (Boncompagni et al., 2012). It is likely that dynamic functional effects and long-term structural effects reflect different aspects of Ca 2+ homeostasis and that the transgenic results are complicated by changes in structure and in expression of CSQ and other junctional proteins.

The physiological role of junctin
It remains our hypothesis that junctin is involved in maintaining RyR activity at rest and in transmitting signals from CSQ to RyRs. This is a significant role given the deleterious effects of changes in resting RyR activity or 'leak' in causing myopathies such as central core disease and malignant hyperthermia in skeletal muscle (Dainese et al., 2009;Paolini et al., 2007) and in precipitating cardiac arrhythmia (Jiang et al., 2004;Terentyev et al., 2006). Indeed junctin-knockout and knockout or mutation of CSQ both increase channel leak and can lead to arrhythmia (Altschafl et al., 2011;Faggioni and Knollmann, 2012). In bilayer studies, experimental removal of CSQ2 from RyR2 increases the sensitivity of RyR2 to changes in luminal Ca 2+ , in a manner that could lead to arrhythmia . Similar changes in sensitivity to luminal [Ca 2+ ] are seen with junctin knockout (Altschafl et al., 2011) and are likely due to removal of the link to CSQ2.

Conclusions
The results reveal a strong interaction between the N-terminal tail of junctin and the cytoplasmic side of RyR1 and RyR2 channels and indicate that this interaction dictates the functional consequences of the full-length protein binding to skeletal and cardiac RyR channels. A modulatory effect of the luminal Cterminal domain on the cytoplasmic N-terminal activation is essential to achieve the normal regulatory action of the full protein on channel activity. The multiple junctin-binding sites on RyR1 and RyR2 raise the unexpected possibility that the influence of this anchoring protein on RyR activity is modified by factors in the cytoplasm as well as in the SR lumen.

Materials
DNA restriction and modifying enzymes were from New England Biolabs and Roche Applied Science; T4 DNA ligase was from Promega; anti-GFP from Roche; anti-RyR (34C, which recognizes both RyR1 and RyR2, DSHB) from the Developmental Studies Hybridoma Bank). The antibody against a junctin peptide (SKHTHSAKGNNQKRKN, with a cysteine added to the N-terminus and conjugated to KLH, GL Biochem, Shanghai) was from IMVS Veterinary Services, Australia. PCR primers were from GeneWorks, Australia.

Species
RyR1, RyR2 and junctin was sourced from several different species for various reasons including amounts required, compatibility with affinity chromatography materials and availability. Identity and homology between isoforms and species where known are given in supplementary material Tables S1-S4.
The Cjun-pET15b.ep plasmid was transformed into E. coli BL21/DE3. Expression of the N-terminal His-tagged protein was induced by 0.6 mM IPTG. Cjun was purified from the soluble lysate fraction using Buffer A (50 mM phosphate buffer PH 8 and 300 mM NaCl) with Ni-agarose affinity chromatography. The solution was exchanged for 20 mM MOPS, pH 7.4 and 150 mM NaCl, then Cjun concentrated (centrifugal filter, Millipore) and stored at 280˚C. Cjun-pHUE was also transformed and poly His-ubiquitin-tagged Cjun expressed in E. coli BL21DE3 (as Cjun-pET15b.ep). The poly His-ubiquitin tag was removed as described previously (Baker et al., 2005;Catanzariti et al., 2004) and Cjun purified, concentrated and stored at 280˚C.
All FLjun, Cjun and RyR constructs were sequenced by the JCSMR Biomolecular Resource Facility.

Peptide synthesis
The following peptides were synthesised by the JCSMR BRF, with purification using HPLC and mass spectroscopy: Njun, Peptides for affinity chromatography were constructed with an Nterminal biotin tag. Rabbit KEKE motif peptides were tested because RyR1 and some FLjun were isolated from rabbit muscle. The canine KEKE motif was also tested because recombinant canine FLjun was also used (see isolation of FLjun below) and Cjun was derived from canine cardiac junctin. Note that the KEKE sequence is identical in canine junctin (Fig. 8) and human junctin (as used by Altschafl et al., 2011).

SR vesicle isolation, RyR purification
Rabbit skeletal SR and sheep ventricular SR vesicles were isolated and RyRs purified as described previously (Wei et al., 2009b). Sucrose gradient fractions containing RyRs (Western blot detection) were concentrated, 15 ml aliquots frozen and stored at 280˚C.

Co-immunoprecipitation
Co-immunoprecipitation was used to assess Cjun binding to native or recombinant RyRs and RyR1 constructs (Beard et al., 2008;Wei et al., 2006). 100 mg of purified RyRs or 3 mg of Cjun were diluted to 100 ml in immunoprecipitation buffer (0.025 M Tris-HCl, 0.15 M NaCl, 0.001 M EDTA, 1% NP-40, 5% glycerol, pH 7.4) and pre-cleared by rotation for 2 h at 4˚C with 100 ml of washed control agarose resin. 10 mg of antijunctin was diluted to 100 ml in coupling buffer (0.01 M sodium phosphate, 0.15 M NaCl, pH 7.2), rotated with 25 ml of washed and equilibrated protein-A/G-agarose for 2 h at room temperature with addition of 3 ml sodium cyanoborohydride, then washed, quenched, and re-equilibrated with immunoprecipitation buffer. The antibody-agarose was then incubated overnight at 4˚C with either 3 mg of the pre-cleared Cjun or with immunoprecipitation buffer alone, then incubated with 100 mg of pre-cleared RyR or RyR1 constructs for 2 h at room temperature. Each incubation was followed by three or four washes and centrifugation steps. In the reverse assay, 10 mg of 34C anti-RyR1 and RyR2 or -GFP antibody was diluted to 100 ml in coupling buffer, bound to protein-A/G-agarose, coupled with 10 mg of pre-cleared RyRs or RyR1 constructs, then incubated with 10 mg of pre-cleared Cjun. Proteins were eluted by boiling for 5 min in 15-20 ml of Laemmli sample buffer and 5 to 10 ml was loaded onto SDS-PAGE gels. Importantly, a constant volume was added for each series of experiments. Proteins were separated by SDS-PAGE, transferred onto polyvinyl difluoride (PVDF) membrane and immunoprobed with 34C anti-RyR1 and RyR2 or -GFP antibody to detect RyR, or with anti-junctin antibody to detect Cjun.

NeutrAvidin agarose affinity chromatography
Binding of Njun and scrambled Njun to native or recombinant RyRs or RyR1 fragments was detected by affinity chromatography (Rebbeck et al., 2011;Wium et al., 2012). 100 ml of NeutrAvidin-agarose (Thermoscientific, Rockford, IL) was washed twice in wash buffer (PBS plus 0.1% SDS) and then resuspended to 100 ml in the same buffer. 100 mg of biotinylated peptide was diluted to 1 mg/ml in binding/wash buffer (PBS plus 0.1% SDS), added to 100 ml of agarose slurry, incubated for 2.5 h at room temperature, and then washed five times with centrifugation (1000 g for 3 min) in wash buffer. Purified RyR protein (90 mg) was diluted in wash buffer plus protease inhibitor (AEBSF, Sigma-Aldrich, USA, or Complete EDTA-free protease Inhibitor Cocktail, Roche Diagnostics, Germany), precleared with control agarose resin (200 ml) for 2 h at room temperature, incubated with NeutrAvidin-agarose-bound biotinylated peptide or NeutrAvidinagarose alone overnight at 4˚C, then washed five times as above. Protein was eluted at 65˚C for 10 min in 15-20 ml of Laemmli sample buffer and 5 to 10 ml was loaded onto SDS-PAGE gels. Importantly, a constant volume was added for each series of experiments. Proteins were separated on SDS-PAGE, transferred onto PVDF membrane and immunoprobed with 34C anti-RyR1 and -RyR2 or -GFP antibody to detect RyR protein, or with StrepTactin to detect biotinylated peptides.
Note that FLjun isolated from rabbit skeletal muscle was initially used in binding and bilayer experiments. However, the extraction was lengthy and yields were very low and we obtained better yields with recombinant canine junctin. There was no observable difference between results obtained with the recombinant FLjun and FLjun isolated from skeletal muscle and both was combined in average data.

Statistics
Data are presented as mean6s.e.m. and significance was evaluated using paired or unpaired Student's t-test or ANOVA as appropriate. The number of observations (n) are given. As similar results were obtained at +40 mV and 240 mV in bilayer experiments, data for each potential was normally included in average data (i.e. two observations for each channel under each condition). To reduce effects of normal variability in the control parameters for RyRs, data with junctin constructs (P o J, T o J, T c J) are expressed relative to control (P o C, T o C, T c C) (e.g. logP o J2logP o C).