Structural Requirements of the Dihydropyridine Receptor (cid:1) 1S II-III Loop for Skeletal-type Excitation-Contraction Coupling*

Residues Leu 720 -Leu 764 within the II-III loop of the skeletal muscle dihydropyridine receptor (DHPR) (cid:1) 1S subunit represent a critical domain for the orthograde excitation-contraction coupling as well as for retrograde DHPR L-current-enhancing coupling with the ryanodine receptor (RyR1). To better understand the molecular mechanism underlying this bidirectional DHPR-RyR1 signaling interaction, we analyzed the critical domain to the single amino acid level. To this end, constructs based on the highly dissimilar housefly DHPR II-III loop in an otherwise skeletal DHPR as an interac-tion-inert sequence background were expressed in dysgenic ( (cid:1) 1S -null) myotubes for simultaneous recordings of depolarization-induced intracellular Ca 2 (cid:2) transients (orthograde coupling) and whole-cell Ca 2 (cid:2) currents (ret-rograde coupling). In the minimal skeletal II-III loop sequence (Asp 734 -Asp 748 ) required for full bidirectional coupling, eight amino acids heterologous between skeletal and cardiac DHPR were exchanged for the corresponding cardiac residues. Four of these skeletal-spe-cific residues (Ala 739 , Phe 741 , Pro 742 , and Asp 744 ) turned out to be essential for orthograde and two of them (Ala 739 and Phe 741 )

Residues Leu 720 -Leu 764 within the II-III loop of the skeletal muscle dihydropyridine receptor (DHPR) ␣ 1S subunit represent a critical domain for the orthograde excitation-contraction coupling as well as for retrograde DHPR L-current-enhancing coupling with the ryanodine receptor (RyR1). To better understand the molecular mechanism underlying this bidirectional DHPR-RyR1 signaling interaction, we analyzed the critical domain to the single amino acid level. To this end, constructs based on the highly dissimilar housefly DHPR II-III loop in an otherwise skeletal DHPR as an interaction-inert sequence background were expressed in dysgenic (␣ 1S -null) myotubes for simultaneous recordings of depolarization-induced intracellular Ca 2؉ transients (orthograde coupling) and whole-cell Ca 2؉ currents (retrograde coupling). In the minimal skeletal II-III loop sequence (Asp 734 -Asp 748 ) required for full bidirectional coupling, eight amino acids heterologous between skeletal and cardiac DHPR were exchanged for the corresponding cardiac residues. Four of these skeletal-specific residues (Ala 739 , Phe 741 , Pro 742 , and Asp 744 ) turned out to be essential for orthograde and two of them (Ala 739 and Phe 741 ) for retrograde coupling, indicating that orthograde coupling does not necessarily correlate with retrograde signaling. Secondary structure predictions of the critical domain show that an ␣-helical (cardiac sequence-type) conformation of a cluster of negatively charged residues (Asp 744 -Glu 751 of ␣ 1S ) corresponds with significantly reduced Ca 2؉ transients. Conversely, a predicted random coil structure (skeletal sequence-type) seems to be prerequisite for the restoration of skeletal-type excitation-contraction coupling. Thus, not only the primary but also the secondary structure of the critical domain is an essential determinant of the tissue-specific mode of EC coupling.
Excitation-contraction (EC) 1 coupling in skeletal muscle is understood as a protein-protein or "mechanical" interaction of two distinct Ca 2ϩ channels, the voltage-gated L-type Ca 2ϩ channel or dihydropyridine receptor (DHPR) and the Ca 2ϩ release channel or ryanodine receptor (RyR1) in the sarcoplasmic reticulum (1, 2; reviewed in Refs. 3 and 4). Therefore, skeletal-type EC coupling is independent from the influx of extracellular Ca 2ϩ (5)(6)(7), in contrast to cardiac EC coupling where Ca 2ϩ influx is required to trigger the release of intracellular Ca 2ϩ from the sarcoplasmic reticulum stores (8), which in turn activates contraction. In skeletal muscle EC coupling the voltage-sensing DHPR undergoes voltage-driven conformational changes that are allosterically communicated to RyR1 via the cytoplasmic loop connecting the homologous repeats II and III of the pore-forming DHPR ␣ 1S subunit (2,9). The II-III loop is not only important for transmitting this orthograde EC coupling signal to the RyR1, it also receives a retrograde, current-enhancing signal from the RyR1 to the DHPR (10,11). Beside unequivocally strong indications for an essential role of the II-III loop for this bidirectional coupling mechanism, accumulating evidence suggests an additional influence of other regions of the DHPR ␣ 1S subunit and/or of the accessory ␤ subunit (12,13).
Nevertheless, one sequence portion of the skeletal DHPR ␣ 1S II-III loop (Leu 720 -Leu 764 ) was previously shown to be essential for bidirectional coupling (11,14,15). These 45 residues inserted into the corresponding regions of ␣ 1 subunit chimeras that contain II-III loops incapable of direct skeletal-type (Ca 2ϩindependent) coupling, like the cardiac II-III loop (11,14) or the highly heterologous II-III loop of the housefly (Musca domestica) DHPR (15), fully restored orthograde and retrograde signaling when expressed in dysgenic (␣ 1S -null) myotubes. Based on the observation that bidirectional coupling was unaffected by drastic alterations of the sequence surrounding skeletal residues Leu 720 -Leu 764 in the chimeric Musca loop, it was concluded (15) that the critical domain Leu 720 -Leu 764 of the skeletal DHPR II-III loop represents a potential site of proteinprotein interaction necessary for the functional coupling of DHPR and RyR1, whereas adjacent II-III loop sequences play little or no role. However, the mechanism by which this domain activates the RyR1 upon depolarization or receives the retrograde signal is still poorly understood. In an effort to elucidate the structural bases of this bidirectional DHPR-RyR1 signaling interaction, we fine-mapped the critical DHPR-RyR1 interaction domain down to the single amino acid level in chimeras with the highly dissimilar Musca II-III loop (15) as a protein-protein interaction-inert neutral sequence background. Our results indicate a striking structure-function correlation between EC coupling properties of the individual chimeras or point mutants and the predicted secondary structure of the interaction domain. Whenever amino acid exchanges resulted in conversion of a predicted random coil structure (skeletal sequence-type) to an ␣-helical structure (cardiac sequence-type) of a negatively charged amino acid cluster in the center of this DHPR-RyR1 interaction domain, skeletal-type EC coupling was significantly reduced. On the other hand, a predicted random-coiled structure of the negatively charged cluster seems to be required for direct, skeletal-type coupling. Thus, the secondary structure of the DHPR-RyR1 interaction domain in the DHPR ␣ 1 subunit II-III loop may be a decisive factor in the tissue-specific mode of interaction between the DHPR and the RyR1 in skeletal muscle EC coupling.

EXPERIMENTAL PROCEDURES
Construction of DHPR Chimeras-The design of the DHPR II-III loop chimeras used in the present study was based on the sequence of chimeras GFP-SkLM and GFP-SkLMS 45 (15). Briefly, GFP-SkLM is composed of rabbit skeletal muscle DHPR ␣ 1S sequence (16) except for a II-III loop derived from the housefly body wall muscle DHPR ␣ 1 subunit (17) (residues Asp 665 -Ser 790 ). For GFP-SkLMS 45 , 45 rabbit ␣ 1S residues (Leu 720 -Leu 764 ) were introduced into the Musca II-III loop of GFP-SkLM, thus substituting for Musca ␣ 1 residues Glu 724 -Thr 755 . Sequence portions of rabbit skeletal muscle (Sk) (16), rabbit cardiac muscle (C) (18), or carp white skeletal muscle (K) (19) DHPR ␣ 1 subunits were introduced into the Musca (M) II-III loop ("L") of GFP-SkLM. Chimeras were N-terminal fused to a modified GFP as described previously (20). Chimeras had amino acid compositions in their II-III loops as follows: GFP-SkLMS 30 , Sk sequence (Leu 720 -Glu 749 ) replaced a portion of the Musca ␣ 1 sequence of GFP-SkLM (residues Glu 724 -Gly 740 ). For the succeeding chimeras, a different portion of Musca ␣ 1 sequence of GFP-SkLM (residues Leu 738 -Thr 755 ) was exchanged by the following sequences: GFP-SkLMS 31  The cDNAs of the DHPR II-III loop chimeras were cloned into a proprietary mammalian expression vector (20) and were constructed as follows (nucleotide numbers (nt) are given in parenthesis, and asterisks indicate restriction enzyme (RE) sites introduced by PCR technique using proofreading Pfu Turbo DNA polymerase (Stratagene)).
For each of the subsequent chimeras and point mutants SOE-PCR was used to generate MfeI-ClaI* PCR fragments, which were coligated with the SalI*-MfeI fragment of GFP-SkLMS 45 (nt Ϫ5-M2024) into subclone pMux according to the ligation protocol as described for GFP-SkLMS 31 .
For simplicity, we will refer to the GFP-tagged chimeras in the following text without the GFP attachment whenever appropriate. The integrity of all cDNA sequences generated and modified by PCR was confirmed by sequence analysis (MWG Biotech, Ebersberg, Germany).
Cell Culture and Transfection-Myotubes of the homozygous dysgenic (mdg/mdg) mouse cell line GLT were cultured as previously described (22). At the onset of myoblast fusion (2 days after addition of differentiation medium) GLT cultures were transfected using FuGENE (Roche Diagnostics). 3-5 days after transfection, expressing myotubes were used for biophysical experiments after identification by their GFP fluorescence.
Electrophysiological Characterization-The whole-cell patch clamp technique (23) was used to simultaneously record voltage-gated L-type Ca 2ϩ currents and depolarization-induced intracellular Ca 2ϩ transients (24). Patch pipettes (borosilicate glass, Harvard apparatus LTD) had resistances of 1.5-3 M⍀ when filled with (mM): 145 Cs-aspartate, 2 MgCl 2 , 10 HEPES, 0.1 Cs-EGTA, 2 Mg-ATP (pH 7.4 with CsOH). Fluorescent signals of intracellular Ca 2ϩ transients were collected microphotometrically (Photon Technology International, South Brunswick, NJ) during whole-cell recordings by including 0.2 mM pentapotassium Fluo-4 (Molecular Probes, Eugene, OR) in the pipette solution. The extracellular bath solution contained (mM): 10 CaCl 2 , 145 tetraethylammonium chloride, 10 HEPES (pH 7.4 with tetraethylammonium hydroxide). Currents were recorded with an Axopatch 200B amplifier (Axon Instruments Inc., Foster City, CA). Data acquisition and command potentials were controlled by pClamp software (version 7.0; Axon Instruments). Test pulses were preceded by a 1-s prepulse to Ϫ30 mV to inactivate endogenous T-type Ca 2ϩ channels (25). Currents were determined with 200-ms depolarizing steps from a holding potential of Ϫ80 mV to test potentials between Ϫ50 and 80 mV in 10-mV increments. Leak currents were digitally subtracted by a P/3 prepulse protocol. Recordings were low-pass Bessel-filtered at 2 kHz and sampled at 5 kHz. As shown in Equation 1, peak currents were normalized to total cell capacitance (pA/pF), plotted as a function of test potential, and fitted according to where G max is the maximal L-channel conductance of the cell, V rev is the extrapolated reversal potential of the Ca 2ϩ current, V 1/2 is the potential for half-maximal activation of G max , and k is a slope factor. Only currents with a maximal voltage error Ͻ10 mV attributable to series resistance were analyzed. Transient changes in fluorescence (⌬F) were normalized by the resting fluorescence (F). For the recordings of immobilization-resistant intramembrane charge movements (gating currents), ionic Ca 2ϩ currents were blocked by the addition of 0.5 mM Cd 2ϩ and 0.1 mM La 3ϩ to the extracellular recording solution, and depolarizing steps of the prepulse protocol were reduced to 50 ms. As recently reported (26), the OFF component of the immobilization-resistant charge is less likely to be contaminated by a nonlinear outward current than the ON component. Thus, charge movements were analyzed from the integrated OFF components (26). The voltage dependence of Ca 2ϩ conductance (G), of intracellular Ca 2ϩ release (⌬F/F), and of gating currents (Q) were fitted according to a Boltzmann distribution as shown in Equation 2 where A is G, ⌬F/F, or Q and A max is G max , (⌬F/F) max , or Q max . V 1/2 is the potential at which A ϭ A max /2, and k is a slope factor. Recordings were performed at room temperature (ϳ23°C), and data are presented as mean Ϯ S.E. Statistical significance was determined by unpaired Student's t test. Data were analyzed using Clampfit® 8.0 (Axon Instruments, Foster City, CA) and SigmaPlot® 6.0 (SPSS Science, Chicago, IL) software.

Skeletal Muscle DHPR II-III Loop Residues
Asp 734 -Leu 764 Are Required for the Bidirectional Signaling Interaction with RyR1-In previous studies, residues Leu 720 -Leu 764 of the rabbit skeletal muscle DHPR ␣ 1S II-III loop were demonstrated to be essential for the full restoration of skeletal-type EC coupling (11,12,14,15) as well as for retrograde coupling, which is the RyR1-mediated enhancement of Ca 2ϩ influx through the DHPR (11,12,15). To better understand the molecular mechanism underlying the direct skeletal DHPR-RyR1 interaction, we identified the minimum sequence within ␣ 1S residues Leu 720 -Leu 764 required for the bidirectional coupling. Because chimera GFP-SkLMS 45 (Fig. 1A) had coupling properties comparable with wild-type GFP-␣ 1S (15), it was used as the reference for skeletal-type EC coupling and enhancement of Ca 2ϩ currents. On the other hand, chimera GFP-SkLM (Fig. 1B), which contains the entire Musca DHPR II-III loop, was used as the reference for constructs not supporting orthograde or retrograde signaling (15). First, we further restricted the bidirectional signaling domain by generating two daughter chimeras of chimera SkLMS 45 : SkLMS 30 containing the upstream sequence Leu 720 -Glu 749 and SkLMS 31 containing the downstream Asp 734 -Leu 764 portion of the S 45 insert with a 15-amino acid overlap (Fig. 1B). Simultaneous recordings of depolarization-induced intracellular Ca 2ϩ transients and whole-cell Ca 2ϩ currents from GFP-fluorescing cells revealed that the Ca 2ϩ current densities as well as the amplitude and voltage dependence of Ca 2ϩ transients of SkLMS 31 were similar to those of SkLMS 45 , whereas the values of SkLMS 30 were similar to SkLM (p Ͼ0.05) (Fig. 1, C and D). As previously described (15), the small Ca 2ϩ transients of SkLM and hence those of SkLMS 30 appeared to be a direct consequence of Ca 2ϩ influx through the DHPR. In contrast, SkLMS 31 fully restored both orthograde and retrograde coupling with RyR1.
Fine-mapping of the Skeletal DHPR II-III Loop Sequence Asp 734 -Leu 764 Important for Bidirectional Coupling-To identify individual amino acids of the skeletal muscle DHPR II-III loop essential for bidirectional signaling, sequence portions within the S 31 insert of chimera SkLMS 31 were exchanged for the corresponding rabbit cardiac (18) or for the evolutionary distant carp skeletal (19) DHPR sequences (Fig. 2A). The latter diverges from the S 31 insert in six amino acids; nonetheless, bidirectional coupling was fully restored by the carp sequence portion Asp 750 -Leu 780 in chimera SkLMK 31 (Fig. 2B). Next, we analyzed the similarities and differences between the S 31 insert and the corresponding cardiac sequence Ser 865 -Leu 895 . As depicted in Fig. 2A, the two ␣ 1 subunits differ in 13 of the 31 residues, 5 of which are conservative Asp/Glu and Glu/Asp exchanges. In both sequences Asp and Glu residues form a central cluster of 7 negative charges. Among the total of 13 amino acid exchanges, 10 are positioned in the N-terminal half, and only 3 are located in the C-terminal half. Neither skeletaltype EC coupling nor Ca 2ϩ current enhancement was restored when the entire 31-residue sequence was cardiac. Chimera SkLMC 31 yielded small Ca 2ϩ currents and marginal Ca 2ϩ transients similar to SkLM (Fig. 2B). Interestingly, when only the C-terminal half of the 31-amino acid insert Glu 880 -Leu 895 was cardiac sequence (chimera SkLMS 15 C 16 ) full bidirectional coupling was restored. Thus, the N-terminal half (Asp 734 -Asp 748 ) of the S 31 insert was sufficient to restore both orthograde and retrograde signaling with RyR1, regardless of whether cardiac or skeletal sequence was positioned downstream. This was not surprising because this downstream sequence is highly conserved between the two isoforms. By contrast, if the highly heterologous Musca sequence were located downstream of the S 15 portion, bidirectional coupling was abolished (chimera SkLMS 30 ) (Fig. 2B).
To investigate whether the S 15 insert (Asp 734 -Asp 748 ) of chimera SkLMS 15 C 16 could be further reduced without impairing bidirectional coupling, two additional chimeras were examined. In chimera SkLMCSC, the cluster of negatively charged residues was skeletal, embedded in an otherwise cardiac 31-residue cassette. This chimera was unable to restore bidirectional coupling (Fig. 2B) indistinguishable from SkLMC 31 (p Ͼ 0.05). In the second chimera, SkLCS 18 (Fig. 2B), the cluster of negatively charged residues was cardiac instead of skeletal compared with the 31-residue cassette of SkLMS 15 C 16 . Despite the fact that only conservative Asp/Glu exchanges occurred within this negatively charged cluster, SkLCS 18 lost its ability to restore retrograde or full orthograde coupling as had been previously shown (11).
To exclude the possibility that the reduction of EC coupling and of Ca 2ϩ currents observed for chimeras SkLMC 31 and SkLMCSC resulted from reduced membrane expression, we recorded immobilization-resistant charge movements (gating currents) as an accurate measure of expression density of voltage-gated ion channels in the surface membrane. The maximum charge movements (Q max ) obtained from SkLMC 31 and SkLMCSC were compared with that of SkLMS 15 C 16 , which fully restored bidirectional coupling. Regardless of the ability to restore bidirectional coupling or not, Q max values of chimeras SkLMS 15 C 16 (5.8 Ϯ 0.6 nC/F, n ϭ 15), SkLMC 31 (6.1 Ϯ 0.8 nC/F, n ϭ 10), and SkLMCSC (5.6 Ϯ 0.6 nC/F, n ϭ 15) were statistically indistinguishable (p Ͼ 0.7). Q max values of chimeras SkLM, SkLMS 45 , and SkLCS 18 had been determined in previous studies (11,15) and are highly comparable with our recent data. Thus, the reduction of Ca 2ϩ transients and Ca 2ϩ currents of our II-III loop chimeras cannot be simply explained by reduced membrane expression levels but rather by an impairment of bidirectional coupling. Residues of the central negatively charged amino acid cluster are indicated by an encircled minus sign. Chou-Fasman secondary structure predictions (␤-turn probability, 0.75 ϫ 10 Ϫ4 ) are represented underneath the sequence by loops for ␣-helical, a dashed line for random coil, and a zig-zag line for ␤-sheet conformation. Bar graphs indicate the mean Ϯ S.E. of the maximal depolarization-induced intracellular Ca 2ϩ transients (⌬F/F) max and were normalized to the reference chimera SkLMS 45 . High values like those of SkLMS 45 and SkLMS 31 represent skeletal-type EC coupling. The mean Ϯ S.E. of the maximal whole-cell Ca 2ϩ conductance G max is indicated at right; number of recordings in parentheses. ** indicates high significant difference (p Ͻ 0.001) compared with chimera SkLMS 45 . a indicates that chimera GFP-SkLCS 18 had been analyzed for bidirectional coupling in a previous study (11) and is listed here for comparison.
Up to this point the results suggested that the 15 skeletal residues Asp 734 -Asp 748 cannot be further reduced without significantly impairing bidirectional signaling. Downstream of this skeletal sequence the adjacent 16-residue sequence is also essential for bidirectional coupling. However, this sequence is highly homologous between skeletal and cardiac DHPR and could be of either origin as long as the upstream domain was skeletal.
The Predicted Secondary Structure of the Negatively Charged Cluster Correlates with Bidirectional Coupling Properties-Beyond the search for a putative interaction motif determined by the primary structure, we also considered the primary structuredependent differences in the secondary structure of the S 31 II-III loop region. The secondary structure of this domain predicted by Chou and Fasman (27) revealed a striking structurefunction correlation in our chimeras. According to this algorithm (and others; not shown) the skeletal sequence (rabbit and carp) forms a random coil in all but the very last residues, which are ␣-helical. In the cardiac channel the centrally located negatively charged cluster is predicted to form an ␣-helix. This sequence contains a conserved proline (in position Pro 750 , corresponding to Pro 881 of the cardiac sequence), a residue that is commonly considered to be a helix breaker. However, because a large number of ␣-helical domains in membrane and globular proteins contain prolines (28), it is not necessary to assume that the predicted ␣-helix terminates at Pro 881 . In the Musca channel, which lacks the negatively charged cluster, a short ␤-sheet exists in the corresponding region. Most interestingly, in all chimeras in which the central negative amino acid cluster was lacking, as in the Musca II-III loop sequence of chimera SkLM, or was predicted as an ␣-helix, as in chimeras SkLMC 31 , SkLMS 30 , SkLCS 18 , and SkLMCSC (Fig. 2B), bidirectional signaling was significantly reduced (p Ͻ 0.001) compared with SkLMS 31 or SkLMS 15 C 16 . In contrast, when this negatively charged amino acid cluster was predicted to form a random coil, as in chimeras SkLMS 31 , SkLMK 31 , and SkLMS 15 C 16 , depolarization-induced intracellular Ca 2ϩ transients and Ca 2ϩ currents were statistically indistinguishable (p Ͼ 0.05) from those of SkLMS 45 and hence from wild-type GFP-␣ 1S (15) in which the entire critical domain or the entire II-III loop was skeletal, respectively. Thus, secondary structure appears to be an important determinant of the tissue-specific mode of DHPR-RyR interaction.
Four Individual Skeletal DHPR Residues Are Essential for Skeletal-type EC Coupling-To test whether this structurefunction correlation could also be observed on the level of point mutants, we performed skeletal to cardiac amino acid exchanges in the S 15 insert of chimera SkLMS 15 C 16 and compared the effect of these substitutions on secondary structure and function. First we established that SkLMS 15 C 16 had full skeletal coupling properties by comparing its intracellular Ca 2ϩ release properties and characteristics of Ca 2ϩ current enhancement with those of SkLMS 45 , which has previously been demonstrated to resemble wild-type ␣ 1S in EC and retrograde coupling (15). SkLMS 45 and SkLMS 15 C 16 not only had identical (⌬F/F) max and G max values (Fig. 2B) but were also statistically indistinguishable (p Ͼ 0.05) in terms of voltagedependence of intracellular Ca 2ϩ transients and of the wholecell inward currents (Fig. 3). Consequently, chimera SkLMS 15 C 16 fully restored bidirectional coupling.
Among the 15 residues of the S 15 insert (Asp 734 -Asp 748 ) required for full restoration of bidirectional coupling in chimera SkLMS 15 C 16 , 10 residues were different between the skeletal and cardiac DHPR sequences (Fig. 4A). Two of these 10 amino acids (Ser 738 and Asp 748 ) could be omitted from this analysis because they are already mutated in the corresponding carp sequence of chimera SkLMK 31 to Pro 754 and Val 764 , respectively, without any significant reduction of bidirectional coupling (Fig.  2). The presence of Val 764 within the negatively charged amino acid cluster of the carp sequence is of particular interest because this resulted in the loss of one negative charge without any significant impact on skeletal-type EC coupling (Fig. 2).
The comparison of intracellular Ca 2ϩ release of each one of the remaining eight single amino acid substitutions with the reference chimera SkLMS 15 C 16 showed that the exchange of the four skeletal residues, Ala 739 , Phe 741 , Pro 742 , and Asp 744 , to their cardiac counterparts significantly (p Ͻ 0.01) reduced skeletal-type EC coupling (Fig. 4). Therefore, these four skeletal amino acids (Fig. 4A, boxed) are essential for skeletal-type EC coupling. Three of the eight point mutants were conservative Asp/Glu exchanges. Even though two of them are adjacent to each other (D744E and D745E), the impact of these substitutions on orthograde signaling differed considerably. Point mutant D744E with a predicted ␣-helix in the central negatively charged amino acid cluster showed significantly reduced Ca 2ϩ transients. In contrast, mutations D745E as well as D740E, which had no impact on the random coil structure, restored skeletal-type EC coupling. Step depolarizations from Ϫ50 to ϩ80 mV were applied in 10-mV increments (200-ms test pulses) from a holding potential of Ϫ80 mV. B, comparison of the voltage-dependence of depolarization-induced Ca 2ϩ transients (⌬F/F, upper) and of peak Ca 2ϩ current densities (pA/pF, lower) of SkLMS 45 and SkLMS 15 C 16 . Values represent the mean Ϯ S.E. of 10 -35 recordings and were statistically identical (p Ͼ 0.05) for both chimeras tested.
Next we examined whether this structure-function correlation is also valid for the point mutants A739P, F741T, and P742T, which also failed to restore full skeletal-type EC coupling like point mutant D744E. In fact, F741T, which showed the strongest reduction of Ca 2ϩ transients, and P742T, which showed a more moderate reduction of orthograde signaling, have a central ␣-helix and thus clearly support this hypothesis. Only in A739P was the impairment of orthograde signaling not correlated to an ␣-helix, thus appearing to be an exception to this general rule. However, in this case not the presence of an ␣-helical structure but an additional bend introduced by the proline might weaken the signaling interaction of the DHPR with RyR1. On the other hand, in SkLMK 31 a proline occurs in the adjacent position (corresponding to the rabbit skeletal Ser 738 ) but had no adverse effect on EC coupling. Thus, the position of the proline in the interaction domain might be decisive in inducing, or not, an effect on skeletal EC coupling. Interestingly, the impact on orthograde signaling by the SkLMS 15 C 16 point mutations did not in all cases correlate with retrograde signaling. In previous studies and up to this point of our present study, a significant reduction in orthograde signaling correlated with a significant reduction in retrograde signaling (10,11,15, and see Fig. 2). This strict correlation was not observed with all SkLMS 15 C 16 point mutants. P742T and D744E showed significantly impaired skeletal-type EC coupling but revealed Ca 2ϩ current enhancement comparable with SkLMS 15 C 16 (Fig. 4B). Taken together, our data obtained by the analysis of single amino acid substitutions reveal that among the four skeletal residues (Ala 739 , Phe 741 , Pro 742 , and Asp 744 ) essential for skeletal-type EC coupling, two (Ala 739 and Phe 741 ) are also essential for the reception of retrograde signaling. Thus our results indicate that some individual amino acids are involved in an unidirectional interaction with RyR1. Moreover, the correlation between the secondary structure and the mode of DHPR-RyR1 interaction was also valid for single amino acid substitutions. DISCUSSION The results reported here characterize the minimal 31-amino acid sequence in the II-III loop of the ␣ 1S subunit that supports skeletal muscle EC coupling and the RyR1-dependent Ca 2ϩ current amplification. This sequence contains three structural characteristics important for the bidirectional interaction of the ␣ 1 subunit and the RyR1: 1) a motif of four essential skeletal residues in the N-terminal half of the 31-amino acid sequence, 2) a cluster of negatively charged residues in the center of this sequence, and 3) the secondary structure of this negative amino acid cluster. Because this critical EC coupling sequence was analyzed in the heterologous Musca II-III loop inserted in the ␣ 1S background, this study does not address the possible contributions of additional sites involved in the reciprocal interaction with the RyR1. In fact, a recent study shows that, whereas deletion of amino acids 720 -765 abolished EC coupling, a construct that restored 17% of control skeletal-type Ca 2ϩ transients (12) after the II-III loop sequence 671-690 was deleted in addition to the 720 -765 sequence. Moreover, resto- ration of retrograde coupling originally assessed in dyspedic (RyR1-null) myotubes by expression of recombinant RyR1 seems to be always larger (ϳ5-fold; Refs. 10 and 29) than that observed with II-III loop chimeras in dysgenic myotubes (ϳ2fold; Refs. 11, 15, and this work). Finally, transferring the ␣ 1S II-III loop sequence 720 -764 into the heterologous low voltageactivated Ca 2ϩ channel ␣ 1H was not sufficient to restore skeletal EC coupling (30). Together, these and other studies (9,13,31) suggest that parts of ␣ 1S outside the II-III loop and/or auxiliary DHPR subunits also participate directly or indirectly in the specific interaction with the RyR1. However, within the II-III loop the 31-amino acids (734 -764) in which the upstream 15 amino acids need to be skeletal but the downstream 16 residues can be of either skeletal or cardiac origin is the only critical sequence, and each of the three structural characteristics mentioned above are decisive determinants for skeletal muscle EC coupling.
Single amino acid substitution showed that in this sequence only four of the residues heterologous in ␣ 1S and ␣ 1C were essential for skeletal muscle EC coupling. Other residues may also contribute to the signaling but not in the tissue-specific manner, because their replacement for the cardiac counterparts did not significantly affect EC coupling. The importance of one of these essential residues, Asp 744 , has already been suggested by previous work. An insert of 18 skeletal residues, Phe 725 -Pro 742 , including three of the four essential skeletal residues but not Asp 744 , inserted in either ␣ 1C (CSk58; Ref. 14) or in the cardiac II-III loop within ␣ 1S (SkLCS 18 ; Ref. 11 and Fig. 2B) resulted in only marginal restoration of EC coupling. Interestingly, of the four residues essential for skeletal EC coupling only two are also essential for the retrograde interaction of the RyR1 with the ␣ 1 subunit. Thus, the two signaling mechanisms utilize overlapping but not identical II-III loop sequences. A similar unidirectional coupling has been observed in studies of RyR1/RyR2 chimeras expressed in dyspedic myotubes. Nakai et al. (32) described a region in the RyR1 (R9) that was solely responsible for retrograde coupling, whereas an upstream region was required for signaling in both directions. If multiple domains of the RyR1 and the DHPR cooperate in the signaling (32-34), a possible explanation for the divergence of orthograde and retrograde coupling in the point mutants P742T and D744E could be that conformational changes induced by these substitutions impaired their orthograde signaling, whereas the physical interaction for the reception of retrograde signaling was still intact.
The second characteristic of the critical 31-amino acid sequence is the central cluster of negatively charged amino acids. This cluster is found in all L-type Ca 2ϩ channels of vertebrates but not in that of Musca; its presence is necessary but not sufficient for skeletal muscle EC coupling. Exchanges of an aspartic acid (Asp 748 ) in this cluster for either a glutamic acid in zebrafish 2 or an uncharged valine in carp (19) had no adverse effect on skeletal EC coupling. On the other hand, the mere presence of this negatively charged amino acid cluster is not sufficient for EC coupling if its secondary structure is predicted to be ␣-helical, as in the case of the cardiac sequence. Thus, the secondary structure of this cluster of negatively charged residues appears to be a third determinant of skeletal muscle EC coupling. This hypothesis was supported by a striking correlation of the secondary structure of all analyzed chimeras and point mutants predicted with the Chou-Fasman algorithm (27) and their EC coupling properties. Whenever exchanges of sequence portions or single amino acids within the DHPR II-III loop interaction domain induced an ␣-helical conformation of the central negatively charged amino acid cluster, signaling interaction with RyR1 was impeded or at least significantly reduced. Conversely, a secondary structure predicted as random coil seemed to be a prerequisite for skeletal-type EC coupling. Whereas empirical statistical methods of secondary structure predictions are of limited predictive value, the Chou-Fasman secondary structure prediction is conceptually the simplest model and the most widely used (35). However, their predictions need to be tested with more experimental structural methods. The random coil structure of this region in the skeletal DHPR II-III loop was recently confirmed by NMR and CD (36); however, NMR analysis of the corresponding cardiac region is still elusive. If our structure predictions are correct, it is likely that not only the presence or absence of individual residues but also the consequences of such substitutions on the secondary structure of the adjacent negatively charged cluster determine the tissue-specific mode of interaction between the DHPR and the RyR1.
Taken together, the analyses of primary and secondary structures of the minimal essential EC coupling domain in the II-III loop of ␣ 1S allows drafting of the following alternative models of II-III loop-RyR1 interactions (Fig. 5). First, the motif of critical residues may specifically interact with a corresponding sequence of RyR1 and the adjacent amino acids, including the negatively charged cluster, are important in that they enable this interaction. Any changes of the four critical amino acids into their cardiac homologues abolish the specific interaction with the RyR1 because those changes lead to an ␣-helical structure of the negative cluster that now masks the specific interaction site (Fig. 5A). Alternatively, not the motif of 2 GenBank TM accession number AY495698.

FIG. 5. Two alternative models of the DHPR II-III loop-RyR1 interaction.
A, schematic representation of the motif of four critical residues (dark spheres, residues indicated in one-letter code and italic above) within the DHPR ␣ 1S II-III loop required for skeletal-type EC coupling. The four skeletal amino acids interact specifically with the RyR1 (Binding). Changes in the amino acid composition that lead to an ␣-helical structure of the adjacent negatively charged cluster (residues indicated by a minus) mask the specific interaction site (Masking). ECC, excitation-contraction coupling, is indicated by an arrow. B, same symbols as in panel A. In contrast to panel A, not the motif of specific residues but the cluster of negatively charged residues interacts with the RyR1 via electrostatic interactions (Attraction/Repulsion). Changes in the amino acid composition that lead to an ␣-helical structure of the adjacent negatively charged cluster impair its mode of interaction with the RyR1 (Distortion). specific residues but the cluster of negatively charged amino acids may be the site of interaction with the RyR1 (Fig. 5B). In this case the interaction is likely to be electrostatic attraction or repulsion, and the adjacent critical residues determine the secondary structure and consequently the function of the interaction site. In either case, our present findings demonstrate that both the primary and the secondary structures of the participating sequences of the ␣ 1 subunit and the RyR1 need to be considered to understand the protein-protein interactions involved in skeletal muscle EC coupling.