Homer regulates gain of ryanodine receptor type 1 channel complex.

Homer proteins form an adapter system that regulates coupling of group 1 metabotropic glutamate receptors with intracellular inositol trisphosphate receptors and is modified by neuronal activity. Here, we demonstrate that Homer proteins also physically associate with ryanodine receptors type 1 (RyR1) and regulate gating responses to Ca(2+), depolarization, and caffeine. In contrast to the prevailing notion of Homer function, Homer1c (long form) and Homer1-EVH1 (short form) evoke similar changes in RyR activity. The EVH1 domain mediates these actions of Homer and is selectively blocked by a peptide that mimics the Homer ligand. 1B5 dyspedic myotubes expressing RyR1 with a point mutation of a putative Homer-binding domain exhibit significantly reduced (approximately 33%) amplitude in their responses to K(+) depolarization compared with cells expressing wild type protein. These results reveal that in addition to its known role as an adapter protein, Homer is a direct modulator of Ca(2+) release gain. Homer is the first example of an "adapter" that also modifies signaling properties of its target protein. The present work reveals a novel mechanism by which Homer directly modulates the function of its target protein RyR1 and excitation-contraction coupling in skeletal myotubes. This form of regulation may be important in other cell types that express Homer and RyR1.

Homer proteins are adapters that physically bind and functionally couple target proteins (1,2). Homer1, Homer2, and Homer3 are encoded by three mammalian genes whose expression is dynamically regulated by cellular activity (2) and can attain high levels of protein in the nervous system where their functional regulation of excitatory signaling has been studied (3)(4)(5). A common element of structure of all Homer proteins is an N-terminal Enabled/Vasp homology domain (EVH1) essential for binding Homer ligands (6 -7). Crystallographic analysis of the EVH1 domain binding surface has revealed its specific association with polyproline ligands (7) that have previously been defined within plasmalemmal glutamate receptors including mGluR1a 1 and mGluR5a/b (1), inositol 1,4,5-trisphosphate receptors (IP 3 R) localized within endoplasmic and sarcoplasmic reticulum (SR) membranes (1), and cytoplasmic Shank proteins that are part of the N-methyl-D-aspartate receptor-associated PSD-95 complex (8,9). A C-terminal coiled-coil domain is responsible for Homer self-multimerization (10 -13). Although full-length Homer proteins are constitutively expressed in a number of tissues, immediate-early gene products of the Hom-er1 gene including Homer1a and Ania 3 lack the CC domain (3,10). "Short form" Homer1a is rapidly and transiently induced by physiological synaptic stimuli that evoke long term potentiation in the hippocampus (3,10) or in striatum by the addition of dopaminergic drugs (3). Thus, a use-dependent exchange of multimeric and protomeric short forms of Homer appears to be responsible for dynamic regulation of context-dependent signaling in neurons.
The functional influences of Homer adaptors on targeted protein have not been identified. Multi-PDZ domain proteins have been reported to cluster membrane ion channels with the result that the channel become active, but this effect is mimicked by agents that otherwise cross-link the channel (14). Homer binding to mGluR1 was recently reported to define agonist-independent activity of the receptor (15); however, this report did not define a molecular mechanism. The direct consequence of forming Homer-IP 3 R complexes on the dynamics of endoplasmic reticulum/SR Ca 2ϩ release and its possible contribution to temporal and spatial aspects of Ca 2ϩ signals remain unclear.
Recently all three Homer mRNAs have also been detected in striated muscle, and Homer protein has been identified within skeletal and cardiac muscle (4,5). Moreover, putative Homer ligand sequences are found within both the type 1 ryanodine receptor (RyR1) and the ␣ 1S -subunit of the dihydropyridine receptor (DHPR; CACNA1S) of skeletal muscle (7). RyR1 assembles as tetrameric structures within junctional regions of SR where it forms large organized arrays (16). RyR1 forms physical associations with ␣ 1S -DHPR that are essential for engaging reciprocal signaling units that are essential for excitation-contraction (E-C) coupling, a process whereby depolarization in the T-tubules triggers Ca 2ϩ release from SR resulting in muscle contraction (17). Although Homer ligand consensus sequences reside in RyR1, experimental evidence confirming the physical association of Homer protein with RyR1, its functional influence on channel gating, and its pos-sible physiological involvement in regulating E-C coupling are lacking. The presence of putative Homer ligand sequences on both RyR1 and ␣ 1S -DHPR within skeletal myotubes provides an excellent system not only to understand the determinants underlying the interaction of Homer proteins with RyR but also to define their functional and physiological significance. The present paper shows that the direct physical interaction of Homer proteins with RyR1 results in a potent (nanomolar) amplification of channel responses to physiological and pharmacological signals. Homer-RyR1 complexes described here regulate the gain associated with E-C coupling and may likely have broad significance in physical and functional coupling of Ca 2ϩ release units in excitable cells.

MATERIALS AND METHODS
Preparation of SR Membranes-Junctional SR membrane vesicles enriched in RyR1 were prepared from skeletal muscle of New Zealand White rabbits according to the method of Saito et al. (18) with some modifications (19). The preparations were stored in 10% sucrose, 10 mM HEPES, pH 7.4, at Ϫ80°C until needed. Functional experiments reported were performed on four different SR preparations.
Expression and Purification of Homer Construct-GST fusion constructs were made by PCR amplifying the H1c open reading frame and the N-terminal 360-bp fragment with in frame primers with SalI and NotI sites and inserting the PCR products into pGEX4T-2 (Pharmacia Corp.). Homer1-EVH1 W24A mutant was made with the Quik-Change TM site-directed mutagenesis kit (Stratagene). Mena EVH1 GST was a gift from Dr. Leahy (Johns Hopkins and Howard Hughes Medical Institute). GST fusion plasmids of H1c, Homer1-EVH1 W24A, and Mena were transformed into BL21 cells, and positive clones were expanded. The cells were lysed by sonication, and the lysate was added to 1 ml of glutathione-agarose (Sigma) and sequentially washed as described previously (8,9). A slurry of glutathione-agarose beads loaded with fusion protein was incubated with 5 units of biotinylated thrombin (Novagen). Purified Homer proteins were dialyzed against phosphatebuffered saline (PBS) at 4°C overnight.
Construction and Expression of Putative Homer-binding Mutations in RyR1 cDNA-A 0.5-kb fragment was isolated between nucleotide 5064 (Pml site) and nucleotide 5582 (BglII site), and one 0.9-kb fragment (between nucleotide 14312, ClaI site and 15230, XbaI site) was isolated from the full-length RyR1 cDNA in pHSV/RyR1 to make the F1777R and F1782R mutations, respectively. The fragments were cloned in pBluescript-SK(ϩ), and a QuikChange TM site-directed mutagenesis kit (Stratagene) was used to introduce F1777R/F1782R into the fragments. After confirming their sequence, the mutated fragments were cloned into the full-length RyR1. Wild type and mutated cDNAs were packaged in HSV-1 amplicon virions and expressed in 1B5 (RyR null) myotubes as previously described (21).
Co-immunoprecipitation and GST Pull-down of Homer-RyR1 Complex-Rat skeletal muscle was homogenized and sonicated in 20 volumes of ice-cold 10 mM MOPS buffer containing 1% CHAPS with protease inhibitors and centrifuged for 10 min at 100,000 ϫ g 4°C. Three microliters of anti-Homer immune serum (8) or preimmune serum was added to 60 l of tissue extract and incubated overnight at 4°C. 50 l of protein A-Sepharose slurry was washed, added to the antibody-extract mixture, and incubated for an additional 2 h. The beads were washed three times with ice-cold MOPS/CHAPS. Bound protein was eluted with 4% SDS loading buffer and analyzed by SDS-PAGE and Western blot. For pull-down assays skeletal muscle detergent extracts were prepared as described for co-immunoprecipitation and mixed with glutathione beads loaded with GST or GST-Homer proteins, incubated overnight, and washed three times with ice-cold MOPS/CHAPS.
FIG. 1. Physical association of homer with RyR. The physical interactions between RyR1 and Homer proteins were analyzed by coimmunoprecipitation (a), GST-pull-down assays (b), and Western analysis of purified junctional SR membranes (c). In a, extracts of skeletal muscle SR in 1% CHAPS buffer were incubated with Homer antibodies (Homer1c, Homer2, and Homer3). The immunoprecipitates were probed with anti-RyR antibody 34C (22). Control lanes (lanes denoted as H1 PI, H2 PI, and H3 PI) utilized the preimmune serum of each antibody. The last lane (6 Lysate) was the offered sample. In b, pulldown assays with GST-Homer fusion proteins and detergent extracts of skeletal muscle enriched in RyR1 were performed. RyR1 bound Homer1c and Homer3 GST proteins, whereas they did not bind to GST alone. In c, junctional SR membranes from four rabbits were separately prepared (preparations (preps) I-IV) and probed for RyR1 and Homer proteins by Western blot analysis. Junctional SR (30 g of protein from each preparation) was applied onto SDS-PAGE, transferred to polyvinylidene difluoride membrane, and probed by anti-RyR1, Homer1c, and Homer3 antibodies.

Single Channel Kinetics in Bilayer Lipid Membranes (BLM)-RyR1
channels reconstitution was performed as described previously (19). The reconstituted channels exhibiting "low P o type" gating behavior were the focus of the study. RyR1 is extremely sensitive to Cys oxidation during preparation of SR (19,27) and results in 60 -70% of reconstitutions in BLM yielding channels that exhibit "high P o type" gating behavior, whereas a small fraction of RyR1 channel reconstituted from these preparations exhibit low P o type gating behavior (19,(27)(28)(29). Low P o type channels maintain low open probability over a large range of cis (cytoplasmic) Ca 2ϩ . Because intact muscle possesses a highly reducing environment, the functional consequences of adding Homer proteins to the cis chamber focused on reconstituted channels exhibiting low P o type gating behavior to better reflect the native environment within junctions of intact skeletal muscle (19,27). Single channel activity was measured at Ϫ40 mV (applied cis relative to the trans) using a patch clamp amplifier (Dagan 3900). The data were filtered at 1 kHz before acquired at 10 kHz by a DigiData 1200A (Axon Inst., Foster City, CA). The data were analyzed using pClamp 6 (Axon) without additional filtering. Average P o was calculated from Ն1 min of recording.
Macroscopic Ca 2ϩ Flux Measurement-Ca 2ϩ transport across SR membranes was performed using the absorbance dye antipyrylazo III (30). SR membranes (50 g/ml) were equilibrated at 37°C in a transport buffer consisting of 92 mM KCl, 20 mM K-MOPS (pH 7.0), 7.5 mM pyrophosphate, and 250 M antipyrylazo III. MgATP (1 mM), 10 g/ml creatine phosphokinase, and 5 mM phosphocreatine were present to regenerate ATP. Ca 2ϩ flux was monitored by measuring antipyrylazo III 710 -790 nm absorbance at 2-4-s intervals using a diode array spectrophotometer (model 8452A, Hewlett Packard). Pyrophosphate was previously shown to accumulate into SR and support active Ca 2ϩ loading without forming precipitates (30); however, the presence of pyrophosphate buffers extravesicular Ca 2ϩ 10-fold, so that in its presence the increase in free Ca 2ϩ is ϳ1/10 th of the total Ca 2ϩ added.
Calcium Imaging of 1B5 Myotubes-The cultured and differentiated 1B5 (RyR null) myotubes were transduced with either wild type or F1777R/F1782R RyR1 amplicons as previously described (21). Within 24 -36 h after transduction, the cells were loaded with the calcium indicator dye Fluo-4 at 37°C, for 30 min in imaging buffer (125 mM NaCl, 5 mM KCl, 2 mM CaCl 2 , 1.2 mM MgSO 4 , 6 mM glucose, and 25 mM HEPES, pH 7.4) supplemented with 0.05% bovine serum albumin and 5 M Fluo-4 (Molecular Probes Inc., Eugene, OR). The cells were washed with imaging buffer before being transferred to a Nikon Diaphot microscope. Fluo-4 was excited at 494 nm using the output of a DeltaRam fluorescence imaging system (Photon Technology International, Princeton, NJ). Fluorescence emission was measured at 516 nm using a 40ϫ quartz objective. Data presented as the 516 nm emissions of Fluo-4 were collected with an intensified CCD camera (Solamere Technology Group) from individual cells. Caffeine-and K ϩ -evoked Ca 2ϩ transients were generated by exposing the cells for 10 s to increasing concentrations of caffeine or KCl. Each test was followed by a 30-s wash allowing the intracellular Ca 2ϩ to return to base line. The data were presented as 1) the percentages of myotubes responding to a given stimulus and 2) the normalized magnitude of the transient, where F o is the mean base-line fluorescence at rest and F is the peak fluorescence in response to the stimulus. The data were recorded from at least 30 myotubes/treatment. showing that H1a-EVH1 is 3-fold less potent than H1c for enhancing [ 3 H]ryanodine binding. d and e show that Homer enhances responsiveness of RyR1 to Ca 2ϩ and caffeine, respectively. SR vesicles (50 g/ml) were incubated as described under "Materials and Methods" for measuring macroscopic Ca 2ϩ fluxes. After 160 M Ca 2ϩ was actively loaded into SR vesicles, 8 nM H1c (II) or the same volume of PBS (I) was introduced. After about 5 min, 80 M Ca 2ϩ (d; Ca 2ϩ -induced Ca 2ϩ release) or 10 mM caffeine (e; caffeine-induced Ca 2ϩ release) was added to initiate release of Ca 2ϩ accumulated during the loading phase. The initial Ca 2ϩ release rates in the presence (II) or absence (I) of H1c were analyzed and plotted. The initial rate of Ca 2ϩ -induced Ca 2ϩ release increased from 0.002 Ϯ 0.03 to 0.337 Ϯ 0.029 mol mg Ϫ1 min Ϫ1 in the presence of H1c (n ϭ 16). Caffeine-induced Ca 2ϩ release increased from 0.009 Ϯ 0.045 to 0.249 Ϯ 0.076 mol mg Ϫ1 min Ϫ1 in the presence of H1c (n ϭ 11).

RESULTS AND DISCUSSION
Homer Proteins Physically Interact with RyR1-An N-terminal EVH1 domain mediates Homer binding to target proteins by contacting specific proline-rich (Homer ligand) sequences found within mGluR1, mGluR5, and the IP 3 R (7). Putative Homer ligand sequences are also present in RyR1 (7). Because Homer proteins are expressed in skeletal muscle (2), we examined the possible in vivo association of Homer with RyR in immunoprecipitation assays from detergent (1% CHAPS) extracts of rat skeletal muscle. Antibodies for Homer1c (H1c), Homer2, and Homer3 co-precipitated RyR1, whereas the preimmune sera failed to do so (Fig. 1a). The interaction of RyR1 with Homer proteins was independently confirmed using GST pull-down assays with GST-Homer1c and GST-Homer3 fusion protein and detergent extracts of skeletal muscle (RyR1) (Fig.  1b; lanes GST-H1c and GST-H3). To examine the presence of endogenous Homer proteins in commonly used preparations of junctional SR enriched in RyR1, we isolated several membrane preparations from rabbit skeletal muscle by sucrose density gradient centrifugation and probed them by Western blot analysis with RyR1-and Homer-specific antibodies (Fig. 1c). In every junctional SR preparation tested, H1c co-purified with RyR1, although the relative amounts of the H1c varied from preparation to preparation. Homer3 was also co-purified with RyR1 but was not detected in every junctional SR preparation examined (Fig. 1c). In all four preparations that were analyzed, H1c and Homer3 proteins were less enriched than RyR1. Because Homer is a cytosolic protein, these results are consistent with the physical association of Homer with RyR1 and indicate that the co-enrichment of Homer relative to RyR1 varies from preparation to preparation. Presumably, Homer proteins variably dissociated from the RyR1 complex during purification of junctional SR membranes.
H1C and H1-EVH1 Enhance RyR1 Gating-To examine the functional consequence of the protein-protein interaction, we determined whether Homer could modify RyR1 channel gating behavior using the BLM method. Single RyR1 channels were reconstituted with an artificial BLM by inducing fusion of purified junctional SR vesicles. Those channels exhibiting the low P o gating mode (19,(27)(28)(29) were selected for examining the influence of exogenous Homer protein added to the cytoplasmic (cis) side of the channel measured under voltage clamp conditions. Both the long form H1c (Fig. 2a) and short form H1-EVH1 lacking the coiled-coil domain (Fig. 2b) were highly potent and efficacious toward enhancing RyR1 channel open probability (P o ). With 5 nM H1c added to the cis chamber, P o increased ϳ10-fold. H1-EVH1 was equally effective but 7-fold less potent (34 nM H1-EVH1 needed to enhance P o ϳ10-fold). The murine homolog of Ena (Mena) possesses an EVH1 domain that is structurally similar to Homer but binds a distinct proline-rich sequence (7). Mena lacked detectable activity on channel P o at the highest concentration tested (50 nM). Both H1-EVH1 and H1c shortened channel closed dwell time by over 100-fold without significantly altering open dwell time (data not shown). We also examined the effects of the immediateearly gene product Homer1a-EVH1 on high P o gating mode channels reconstituted in BLM. Homer1a-EVH1 (40 nM) signif-  (24 -26). When GST was included in the binding medium, occupancy of [ 3 H]ryanodine was indistinguishable from control (Fig. 2c). Similarly, [ 3 H]ryanodine binding was unaltered by inclusion of Mena (Յ100 nM). In marked contrast, 50 nM H1c or its fusion protein, GST-H1c, enhanced the binding of 5 nM [ 3 H]ryanodine by nearly 5-fold (Fig. 2c). The actions of GST-H1c measured in the presence [ 3 H]ryanodine at a concentration well below its K D (0.5 nM) resulted in ϳ8-fold enhancement of occupancy (Fig. 2c, right panel) consistent with the effects of Homer protein on enhancing channel P o in BLM studies (Fig. 2a). At a concentration of 5 nM [ 3 H]ryanodine, the addition of 50 nM H1-EVH1 mimicked the activity of full-length H1c but exhibited smaller (Ͻ2-fold) but statistically significant (p Ͻ 0.05) enhancement in binding compared with control. When tested at subsaturating [ 3 H]ryanodine (0.5 nM), the addition of 150 nM H1-EVH1 increased binding ϳ5-fold, whereas 50 nM H1c increased binding ϳ8-fold (Fig. 2c). This demonstrates that both forms have a similar ability to enhance RyR1 activity consistent with their ability to enhance the P o of channels reconstituted in BLM (Fig. 2b). The 3-fold difference in the potency observed with the long and short forms of Homer on RyR1 could be attributed to the multivalency of H1c that may optimize conformational interaction with RyR1 oligomeric units. As a further control, we found that the W24A point mutant of H1-EVH1, which does not bind Homer ligands (7), had negligible activity on altering the binding of [ 3 H]ryanodine to RyR1 under the stated assay condition (Fig. 2c). These results support the hypothesis that specific protein-protein interactions between Homer proteins and RyR1 can variably enhance the gain of signaling events mediated through Ca 2ϩ -induced Ca 2ϩ release (CICR), and hence Ca 2ϩ released from SR, by influencing the fraction of RyR channels within junctional SR responsive to physiological activators.
To further test the hypothesis, [ 3 H]ryanodine-binding curves were performed in the presence of nearly optimal Ca 2ϩ for channel activation (1 M) and in the absence or presence of Homer protein (Fig. 3a). Scatchard analysis revealed that H1c enhanced the binding of [ 3 H]ryanodine over the complete range of concentrations, enhancing maximum occupancy 3-fold (from B max ϭ 4.2 and 12.7 pmol mg Ϫ1 in the absence and presence of 50 nM H1c, respectively) without significantly altering binding affinity (K D ϭ 7.95 and 9.06 nM, respectively; Fig. 3b). These results indicate that formation of H1c-RyR1 complexes in vitro significantly increased the density of Ca 2ϩ -responsive channels within the SR membrane. Because Homer protein appears to dissociate from SR preparations during biochemical enrichment of RyR1, the ability of exogenous Homer to increase the density of functional RyR1 able to bind ryanodine in a Ca 2ϩdependent manner likely results from formation of new protein-protein complexes that may have been lost during preparation of junctional SR membranes.
Although in skeletal muscle CICR is not necessary to initiate E-C coupling, nor is it necessary to engage retrograde signaling (32), it is generally agreed that Ca 2ϩ is an important physiological modulator of RyR1 in skeletal muscle. Several ligands have been shown to influence the sensitivity of RyR to activation by Ca 2ϩ (24,26). Using [ 3 H]ryanodine binding analysis near K D for the radioligand, H1c (50 nM) did not significantly shift the sensitivity of RyR1 to activation by Ca 2ϩ relative to control (EC 50 ϭ 210 versus 280 nM Ca 2ϩ , respectively; Fig. 3c). However, H1c did enhance maximal occupancy 2.5-fold, indicating an enhanced "gain" of RyR1 toward activation by Ca 2ϩ consistent with an increase in the density of Ca 2ϩ -responsive RyR1 complexes.
Homer Enhances CICR and Caffeine-A functional corre- FIG. 4. Enhancement of RyR1 channel function by Homer is specifically blocked by competing peptide. H1c was preincubated with 1000ϫ synthetic wild type Homer-binding peptide (Pept wt ), synthetic peptide missing the critical phenylalanine (Pept mut ), or PBS at 37°C for 10 min. After completion of preloading of SR with 160 M Ca 2ϩ , Homer in the presence or absence of Pept wt or Pept mut was introduced. The final concentration of the H1c in the transport buffer was 20 nM (a) or 8 nM (b and c). 10 M ryanodine (Ry) or Ruthenium Red (RR) was added to activate or inhibit RyR1, whereas 0.5 M thapsigargin (TG) was introduced to inhibit SR Ca 2ϩ pumps. In b and c, the final concentration of Ca 2ϩ used to induce Ca 2ϩ release was 60 M. Summary data represent the mean of 21 (a), 6 (b), or 6 (c) repetitions of these experiments. late to the results obtained from radioligand receptor binding analysis was undertaken to understand how H1c influences macroscopic Ca 2ϩ efflux elicited by either Ca 2ϩ (CICR) or caffeine from actively loaded skeletal muscle SR vesicles (Fig.  4, b and c). This assay utilizes pyrophosphate to support active loading of Ca 2ϩ into the vesicles (30). After calcium loading, the addition of H1c to a subthreshold concentration (8 nM) that did not directly initiate net Ca 2ϩ efflux from the SR vesicles enhanced the initial rate of RyR1-mediated Ca 2ϩ elicited by subsequent addition of either optimal 80 M total Ca 2ϩ (8.0 M free Ca 2ϩ ) or saturating caffeine (10 mM) 7.9-and 8.2-fold, respectively. Thus, H1c enhanced the maximum efficacy of both Ca 2ϩ and caffeine for activation of RyR1.
Homer Modulation of RyR1 Is Blocked by Specific Peptide-The specificity of H1c for enhancing Ca 2ϩ release from SR was examined by preincubating H1c with a synthetic "blocking" peptide that mimics the Homer ligand within the C terminus of mGluR-1 (1). Fig. 4a (trace I) demonstrates that addition of saturating (20 nM) H1c induced net Ca 2ϩ release from actively loaded SR vesicles. In the absence of added Homer, Ca 2ϩ was not released until the addition of either ryanodine (10 M; trace III) or the SR Ca 2ϩ pump inhibitor thapsigargin (0.5 M; trace IV). The effect of H1c could be blocked by the addition of 50 M wild type peptide (Pept wt ) but not by a point mutant of the blocking peptide (Pept mut ) that lacks Homer binding activity (1) (Fig. 4a, trace II). Pept wt does not block release by subsequent addition of ryanodine. Pept wt , but not Pept mut , also prevented the ability of H1c to amplify CICR after active loading of 160 M Ca 2ϩ into SR (Fig. 4, b and c). CICR induced by the addition of 60 M Ca 2ϩ in the presence of 8 nM H1c exhibited indistinguishable rates of Ca 2ϩ release in the presence or absence of Pept mut (Fig. 4c, traces I and II). Subsequent addition of Ruthenium Red (10 M), an RyR channel blocker, resulted in active re-uptake of Ca 2ϩ into SR vesicles. However, in the presence of Pept wt (Fig. 4b, trace III), the addition of 60 M Ca 2ϩ displayed a net Ca 2ϩ uptake into SR because SR Ca 2ϩ pump-transport into the vesicles surpassed the amount of release through RyR1. Active accumulation could be interrupted abruptly and was quickly followed by Ca 2ϩ release after subsequent addition of 10 M ryanodine (Fig. 4, b and c, trace IV). As further evidence of its specificity, we confirmed that Pept wt did not affect modulation of RyR1 by the scorpion toxin maurocalcine, a potent activator of RyR1 (31) (not shown).
Taken together, these results revealed that physical association of H1c with RyR1 dramatically enhanced Ca 2ϩ respon-siveness of RyR1 at the level of the single channel and underlies the increased "gain" of RyR1-mediated Ca 2ϩ efflux units within junctional SR (defined by two independent measures, macroscopic Ca 2ϩ efflux (CICR) and [ 3 H]ryanodine binding density). These actions of Homer protein exhibited a high degree of molecular specificity and were mediated by their unique EVH1 domain.
RyR1 Homer-binding Domain Regulates Gain-Conformational changes in IP 3 R caused by IP 3 binding and store depletion was shown to provide a retrograde signal to opening of store-operated Ca 2ϩ channels in the plasma membrane (33,34), similar to the bi-directional signaling between RyR1 and ␣ 1S -DHPR in skeletal muscle that is essential for engaging E-C coupling (17). Moreover, evidence for conformational coupling between RyR and store-operated Ca 2ϩ channels has been recently presented (35,36). Thus, conformational coupling appears to be a fundamental mechanism involved in generating localized Ca 2ϩ signals. Both RyR and ␣ 1S -DHPR possess discrete polyproline sequences that are putative Homer ligands. RyR1 contains a putative Homer-ligand domain between amino acids 1773 and 1783. A second putative Homer binding domain can be found in RyR1 (PPPGYA; amino acids 804 -809), where the essential hydrophobic phenylalanine is substituted with tyrosine.
We therefore tested the hypothesis that Homer is involved in modulating the strength of depolarized-induced Ca 2ϩ release in intact cells. 1B5 dyspedic skeletal myotubes express all the proteins known to be associated with the mature triadic junction but lack expression of the three RyR isoforms (23,37). Transduction of 1B5 myotubes with wild type RyR1 cDNA reconstituted bi-directional signaling between RyR1 and ␣ 1S -DHPR and directed the organization of ␣ 1S -DHPR particles into organized arrays of tetrads (38,39). Homer1 protein was detected in 1B5 myotubes by immunocytochemical and Western blot analyses (Fig. 5a).
The possible functional role of one of the putative Homerbinding sites of RyR1 was tested by incorporating site-directed mutations (F1777R/F1782R) within the full-length protomer. The substitutions were chosen based on the recent data showing that hydrophobic phenylalanine C-terminal to polyprolines are essential for Homer binding (7). RyR protomers from 1B5 myotubes expressing full-length wild type and F1777R/F1782R RyR1 proteins exhibited the same mobility by SDS-PAGE (not shown) and show similar levels of expression by immunocytochemistry. At higher magnification both wt and F1777R/ F1782R RyR1 exhibited the characteristic punctate pattern of shows the responses of 1B5 myotubes expressing wild type and the putative HLM (F1777R/F1782R RyR1) to threshold (1 mM) and saturating (30 mM) caffeine. 40% fewer myotubes responded to the threshold concentration of caffeine, whereas all myotubes responded to saturating caffeine. The normalized amplitude of responses to threshold caffeine were 56% smaller (p Ͻ 0.001) in HLM versus wild type but were not different at saturating caffeine. The data were analyzed from 68 and 92 wild type and HLM myotubes, respectively. c and d show the K ϩ -evoked Ca 2ϩ transients following expression within the myotubes, which has previously been shown to be consistent with targeting of RyR protein to peripheral junctions (Fig. 5b) (38,39).
Both wild type and the putative Homer ligand mutation (HLM; F1777R/F1782R) RyR1 responded to caffeine challenges. However, the percentage of HLM myotubes responding to a threshold concentration of caffeine (1 mM) was 39.6% lower than wild type myotubes (43.5% versus 72.0% of n ϭ 92 and 68 myotubes, respectively; Fig. 6a, left bars). When challenged with saturating caffeine (30 mM), the frequency of responses with HLM and wild type myotubes were both 100%. The normalized magnitude of the Fluo-4 fluorescence signal (F/F o ) is a quantitative measure of the amount of Ca 2ϩ mobilized in response to direct activation of release channels by caffeine. The magnitude of the responses to 1 mM caffeine was 56.6% smaller (F/F o ϭ 6.8% versus 15.7%; p ϭ 0.003) in myotubes expressing HLM. However, 30 mM caffeine elicited Ca 2ϩ responses whose magnitude were indistinguishable between wild type and HLM myotubes, indicating that the SR stores of both cell types were filled with Ca 2ϩ to similar levels (Fig. 6a, right bars).
Expression of either wild type or HLM cDNAs in 1B5 myotubes also restored Ca 2ϩ responses to K ϩ depolarization consistent with restoration of skeletal type excitation-contraction coupling, and the amplitude of this response increased in a saturable manner with increasing K ϩ concentration between 10 and 100 mM (Fig. 6, c and d). The expressed proteins produced Ca 2ϩ transients that were kinetically identical, and the apparent potency for producing K ϩ -evoked Ca 2ϩ transients was unaltered by the mutation (EC 50 ϭ 26.6 ϩ 1.2 and 27.0 ϩ 1.2 mM for wild type and mutant protein, respectively). Importantly, however, the maximum amplitude of the K ϩ -evoked Ca 2ϩ transients produced by cells expressing HLM were 33% smaller than those produced by wild type RyR1 (mean F/F o ϭ 76 ϩ 0.85 and 58 ϩ 2.0; n ϭ 34 and 48 cells for wild type and HLM, respectively; p Ͻ 0.001), demonstrating Homer's importance in modulating depolarization induced Ca 2ϩ release. The possible existence of a second Homer ligand domain within RyR1 (amino acids 804 -809) and its functional role remains to be investigated.
Functional Significance of Interactions between RyR1 and Homer Proteins-The present study demonstrates that Homer binding directly enhances responses of RyR1 to physiological and pharmacological stimuli. This activity is conferred by the EVH1 domain and does not require the cross-linking adapter function of Homer. However, multimeric H1c does appear to significantly enhance the potency toward enhancing RyR1 responses. Homer is the first example of an "adapter" that also modifies signaling properties of its target protein. Multi-PDZ domain proteins are reported to activate ion channels, but this activity is mediated by simple physical cross-linking (14). The nanomolar affinities required for channel modulation with fulllength RyR and the complete long and short forms of Homer indicate a structurally specific interaction. Previous studies that used peptides to block the binding of EVH domains with their target proteins suggested affinities in the micromolar range (1,8). Because our studies used full-length proteins, it is inferred that additional structure is important for optimizing these binding interactions. Thus, RyR1 that is bound by constitutively expressed full-length Homer could reinforce its known adapter role, resulting in signaling microdomains that are uniquely primed for Ca 2ϩ signals.
The present findings reveal that Homer proteins through their direct interaction with RyR1 can modulate the gain of E-C coupling. This can occur in three ways. First, the interaction of constitutively expressed H1c could bind to one to four channel protomers to regulate cooperativity associated with channel activation. Second, H1c multimers could cross-link multiple RyR1 oligomers to facilitate coupled gating among channels elicited by FKBP12 (40). Because localized Ca 2ϩ sparks occur upon release of Ca 2ϩ through a cluster of RyR (41), it is interesting to note that H1c has been recently shown to induce sparks in frog skeletal muscle (42). Homer-RyR associations may therefore be a significant physiological regulator of spark activity that contributes to Ca 2ϩ waves that sweep through cells and subsequently activate myocyte contraction. Third, multimeric H1c could directly cross-link RyR1 with ␣ 1S -DHPR to enhance the efficiency of E-C coupling in a manner already described for mGluR1 and IP 3 R (1, 7). Whether Homer interactions with a putative binding site on ␣ 1S -DHPR contribute to modulation of E-C coupling remains to be determined.
Homer short form (i.e. H1a-EVH1) is encoded by a prematurely terminated transcript that lacks the coiled-coil domain that is required for the formation of Homer multimers. The currently favored mechanism for how Homer proteins regulate signaling functions suggests that the H1-EVH1, an immediateearly gene product, is rapidly expressed in response to cellular activity, whereas long form H1c is constitutively expressed. In this scheme, H1c plays an adapter function, linking target proteins such as mGluR1 and IP 3 R in neurons. Immediateearly expression of H1-EVH1 acts in a dominate-negative fashion by competing with existing H1c complexes, thereby impairing the formation and maintenance of Homer-mediated cluster of protein complexes (2). However, experimental evidence for a more dynamic role for Homer proteins in cellular regulation has recently been defined (13,(43)(44)(45). For example, experiments using single particle tracking demonstrated that the composition of Homer-mediated clustering of protein complexes is highly dynamic. The exchange of the proteins between dispersed and clustered states is fast and is regulated during physiological processes (44,45). The present results suggest that the catalytic activity of Homer proteins toward RyR may provide one mechanism linking Ca 2ϩ signals with dynamic regulation of Homer-mediated protein clustering.
One anticipated consequence of Homer catalytic activity is to increase the spatial fidelity of signaling beyond what can be achieved by simple physical coupling. Homer and RyR proteins are widely expressed, and it may be anticipated that their physical and functional interactions identified here contribute to temporal and spatial patterning of intracellular Ca 2ϩ signaling. In skeletal muscle, release of Ca 2ϩ through RyR1 is the key event linking membrane depolarization and mechanical activity during E-C coupling. In neurons, Homer proteins play a special role in adaptation to cellular activity. CICR has been demonstrated in a number of neuronal cell types (46 -48). Evidence for the role of RyR in the regulation of specific aspects of neuronal plasticity, dendritic growth, and spatial learning has recently been demonstrated (49 -52). The present results predict that RyR1-Homer interactions may also contribute elements of structural and functional regulation important for synaptic plasticity that are likely to direct signal processing and shape neuronal circuits. 10 s of varied K ϩ challenge on dyspedic 1B5 myotubes transduced with either wild type or HLM cDNA. Ca 2ϩ transients were independent of extracellular Ca 2ϩ entry and represent skeletal type excitation-contraction coupling. Although the K ϩ concentration producing half-maximal amplitude was not different (26.1 versus 26.7 mM), the maximal amplitude of the transients was 33% smaller (p Ͻ 0.001) for RyR1 lacking the critical phenylalanines within a putative Homer ligand domain. The measurements are the means from 34 and 47 individual cells for wild type and HLM.