The molecular architecture of dihydropyrindine receptor/L-type Ca2+ channel complex

Dihydropyridine receptor (DHPR), an L-type Ca2+ channel complex, plays an essential role in muscle contraction, secretion, integration of synaptic input in neurons and synaptic transmission. The molecular architecture of DHPR complex remains elusive. Here we present a 15-Å resolution cryo-electron microscopy structure of the skeletal DHPR/L-type Ca2+ channel complex. The DHPR has an asymmetrical main body joined by a hook-like extension. The main body is composed of a “trapezoid” and a “tetrahedroid”. Homologous crystal structure docking and site-specific antibody labelling revealed that the α1 and α2 subunits are located in the “trapezoid” and the β subunit is located in the “tetrahedroid”. This structure revealed the molecular architecture of a eukaryotic Ca2+ channel complex. Furthermore, this structure provides structural insights into the key elements of DHPR involved in physical coupling with the RyR/Ca2+ release channel and shed light onto the mechanism of excitation-contraction coupling.

The slices, from slice 1 to slice 18, represent the putative ion-conduction channel direction of DHPR proposed by Wolf et al 11 and others 13,14 , viewed from putative extracellular side to the cytoplasmic side. This direction is orthogonal to the ion-conduction channel direction we revealed (ref. Fig. 2b). The distance between slices is 2 pixels (1.43 Å/pixel); the thickness of each slice is 2 pixels. A large niche leading the channel to run side-way into the hydrophobic core of lipid bilayer is illustrated (indicated by arrows).  When displayed at high threshold, the "hook" shrinks and splits into parts while the main body maintains its shape even at high threshold; when displayed at low threshold, however, the "hook" expands and joins to the main body at two ends and transform into a "handle" (indicated by arrows).

Supplementary Discussion
The ion-conduction channel The ion-conduction channel proposed in literature 11,13,14 is orthogonal to the ion-conduction channel direction we revealed (ref. Fig. 2b). It appears a "channel" existing in this direction in our EM-map (Suppl. Fig. S2). This "channel", however, cannot act as the ion-conduction channel for following reasons: first, the densities surrounding the hole do not display psudo-4-fold symmetry, which is contradict to the Ca 2+ channel model 2 ; second, the central hole is not surrounded by densities throughout the channel-it runs sideway in the middle of the map (indicated by arrows in Suppl. Fig.   S2), which would run into the hydrophobic core of the lipid bilayer; and third, the density-enclosed channel (from the 1st slice to the 4th slice) is only 1.14 nm, far too short to go across a lipid bilayer.

Comparison of our EM-map with EM-maps in literature
Consistent with previous studies 9,11-14 , our EM map has similar profile with the previously published EM-maps: the DHPR is composed of a main body and a characteristic hook/handle-like density conjoined to the main body. Compared with other maps, the shape of the main body is more close to Serysheva et al. 9 ("diamond" vs. "heart"). Using the hook as a register structure, we overlaid our EM map with that of  Fig. 1a, lane 2 & lane 4), we believe that the molecular mass in the maps of Wolf et al. and others must be larger than 450 kDa therefore contain non-protein mass, most likely detergent and lipids. The difference in detergent and lipid amount is likely caused by different purification methods. In our new method, we washed the column extensively with buffer containing low concentration of detergent before the final elution step and diluted the purified DHPR prior to EM analysis (see Methods), thereby minimised the amount of detergents and lipids attached to the DHPR complex. As the α2 subunit is extra-membrane and connected to the δ subunit just via a disulfide bond, it is mobile. During image processing, the individual particle images were aligned, added together, then averaged to enhance signal-to-noise ratio. Due to the mobile nature, the positions of the α2 subunit in individual particle images would be different from each other. As a result, part of the α2 subunit density would be "averaged out" by image processing. Therefore, it appears smaller than it should be and the majority of the "trapezoid" density in the EM map is contributed mainly by the α1 subunit (MW~176kDa).

Assignments of subunits
Previous studies have assigned the hook region as the α2 subunit of DHPR 11-14 . Based on our structure and membrane topology, however, we believe that the hook region represents part of the α2/δ subunit. As shown in Fig. 3a, the position recognized by the anti-α2 antibody is located on top of α1, which is conjunct to the hook region, suggesting 6 that the top region above α1 corresponds to part of α2. As the hook region is connected to the region identified as α2, it is reasonable to assume that the hook region is an extension of the α2/δ subunit. The α2 subunit is linked to the δ subunit via a disulphide bond and is proposed to be located in the extracellular side 17 . Currently, there are two models concerning how is the α2/δ subunit associated with the DHPR complex-besides a direct interaction of the α2 subunit with the α1 subunit in the extracellular side, there exists another element that "anchors" the α2 subunit to the plasma membrane: the first model proposed that the δ subunit is a transmembrane protein with a single transmembrane helix, which interacts with the transmembrane domain of the α1 subunit and anchors the α2 subunit to the membrane 33 ; the second model proposed that the δ subunit is not transmembrane, but instead extracellular and anchored via a GPI link to the membrane 34 .
As our EM structure cannot resolve secondary structures (i.e., alpha helices), we cannot judge which model is correct. However, our EM structure reveals that the δ subunit is not associated with the α1 subunit, this provides a possibility that the interaction of the α2/δ subunit with the DHPR complex could be dynamic, it could associate to/dissociate from the DHPR complex under different conditions, providing a dynamic way of modulation of the DHPR function.
As the α2 subunit is linked to the δ1 subunit via a disulfide bond and is proposed to be located in the extracellular side, this region should be mobile. To test this hypothesis, we examined the density profiles of our EM map by changing display threshold. When displayed at high threshold, the "hook" shrinks and splits into parts while the main body maintains its shape even at high threshold; when displayed at low threshold, however, the "hook" expands and joins to the main body at two ends and transform into a "handle" 7 (Suppl. Fig. S4). This demonstrates that the hook region is indeed mobile. Serysheva et al.
showed that the "handle" (equivalent of the "hook" in our map) is located at the side of the top part of the "heart"-shaped DHPR, whereas in our case the "hook" is attached to the side of the lower part of the "diamond"-shaped DHPR. The position difference of the "hook/handle" in the EM maps from different research groups again demonstrates that this region is mobile. The mobile nature of the α2/δ subunit implies that the interaction of the α2/δ subunit with the DHPR complex could be dynamic; it could associate to/dissociate from the DHPR complex under different conditions, providing a dynamic way of modulation of the DHPR function. The positions of the hook/handle region in the EM maps are probably not its original position as in the membrane-bound state due to this mobility. The original position of the hook/handle region should be addressed by reconstitution of the DHPR into lipid bilayer and investigated by electron tomography or electron crystallography.