Adeno-associated Virus Receptor-binding: Flexible Domains and Alternative Conformations through cryo-Electron Tomography of AAV2 and AAV5 complexes

Recombinant forms of adeno-associated virus (rAAV) are vectors of choice in the development of treatments for a number of genetic dispositions. Greater understanding of AAV’s molecular virology is needed to underpin needed improvements in efficiency and specificity. Recent advances have included identification of a near universal entry receptor, AAVR, and structures by cryo-electron microscopy (EM) single particle analysis (SPA) that revealed, at high resolution, only the domains of AAVR most tightly bound to AAV. Here, cryogenic electron tomography (cryo-ET) is applied to reveal the neighboring domains of the flexible receptor. For AAV5, where the PKD1 domain is bound strongly, PKD2 is seen in three configurations extending away from the virus. AAV2 binds tightly to the PKD2 domain at a distinct site, and cryo-ET now reveals four configurations of PKD1, all different from that seen in AAV5. The AAV2 receptor complex also shows unmodeled features on the inner surface that appear to be an equilibrium alternate configuration. Other AAV structures start near the 5-fold axis, but now β-strand A is the minor conformer and, for the major conformer, partially ordered N-termini near the 2-fold axis join the canonical capsid jellyroll fold at the βA-βB turn. The addition of cryo-ET is revealing unappreciated complexity that is likely relevant to viral entry and to the development of improved gene therapy vectors. IMPORTANCE With 150 clinical trials for 30 diseases underway, AAV is a leading gene therapy vector. Immunotoxicity at high doses used to overcome inefficient transduction, has occasionally proven fatal and highlighted gaps in fundamental virology. AAV enters cells, interacting through distinct sites with different domains of the AAVR receptor, according to AAV clade. Single domains are resolved in structures by cryogenic electron microscopy. Here, the adjoining domains are revealed by cryo-electron tomography of AAV2 and AAV5 complexes. They are in flexible configurations interacting minimally with AAV, despite measurable dependence of AAV2 transduction on both domains.

equilibrium alternate configuration. Other AAV structures start near the 5-fold axis, but 23 now β-strand A is the minor conformer and, for the major conformer, partially ordered N-24 termini near the 2-fold axis join the canonical capsid jellyroll fold at the βA-βB turn. The 25 addition of cryo-ET is revealing unappreciated complexity that is likely relevant to viral 26 entry and to the development of improved gene therapy vectors. 27 IMPORTANCE: With 150 clinical trials for 30 diseases underway, AAV is a leading gene 28 therapy vector. Immunotoxicity at high doses used to overcome inefficient transduction, 29 has occasionally proven fatal and highlighted gaps in fundamental virology. AAV enters 30 cells, interacting through distinct sites with different domains of the AAVR receptor, 31 according to AAV clade. Single domains are resolved in structures by cryogenic 32 9 Cryo-ET tilt series were acquired on an ThermoFisher Titan Krios (Hillsboro, OR) and 170 recorded with Leginon software (Suloway et al., 2009) on a Gatan K3 direct detector. A 171 magnification of 33,000 X was used with a pixel size of 2.74 Å and a total dose of 100 172 e¯/Å² per tilt series. The tilt angle ranged from -60° to 60° with 2° steps. Exposure time 173 at each tilt step was automatically adapted by the Leginon software according to the tilt 174 angle. The number of frames at each tilt step was automatically set by Leginon 175 according to the exposure time at each tilt step. Dose was fractionated across the 176 frames at each step. Defocus values were set to 5 μm underfocus. 177 Single particle data was also collected on the Titan Krios with the K3 camera using 178 Leginon software. Magnification was 81,000 X and pixelsize was 1.1 Å. The defocus 179 range was set to -1.0 to -3.0 μm. Total dose was ~60 e¯/Å² per image with 50 frames 180 for each micrograph. All frames of each micrograph were aligned using MotionCor2 181 (62). 182 For AAV5 bound with PKD1-2, both single particle and tomography data were collected 183 from the same cryo-EM grid. 184 Single particle image processing 185 CTFFIND4 and GCTF were used to estimate contrast transfer function (CTF) 186 parameters on all motion-corrected micrographs and the best estimate was chosen 187 using resolution evaluation in Appion (63-65). Around 1000 particles were picked using 188 DoG (Difference of Gaussian) Picker and the rotational average of those particles was 189 used as the template for picking using FindEM in Appion (65,66). A total of 26,091 190 particles were picked from 378 micrographs and extracted with a box size of 432 × 432 191 pixels in Appion. 2D classification and 3D classification were conducted to choose good 192 particles in Relion3-beta. The previous 2.5 Å resolution single particle cryo-EM 193 reconstruction of the AAV5-PKD12 complex (EMD22988) was low pass filtered to 60 Å 194 resolution and used as the initial reference for 3D refinement by 67). 195 A total of 15,052 particles were selected for gold standard auto-refinement. Icosahedral 196 symmetry was applied for auto-refinement. After auto-refinement, CTF refinement and 197 beam tilt refinement, a final map of 2.88 Å resolution was achieved. 198

Tomography image processing 199
Tilt series were aligned using Protomo software within Appion (Winkler and Taylor, 200 2006;Lander et al., 2009;Noble and Stagg, 2015). Following that, the image stack for 201 tilt series were imported into EMAN2/e2tomo. Alignment parameters from Protomo were 202 imported into EMAN2/e2tomo with home-made scripts, and tomograms were directly 203 calculated with imported parameters. Contrast transfer function (CTF) parameters were 204 estimated on all micrographs inside E2tomo. For AAV2 bound with PKD1-2, 127 virions 205 (7620 asymmetric subunits) were manually picked and extracted with E2tomo and a box 206 size 288 x 288 pixels. Similarly, 85 virions (5100 asymmetric subunits) were picked and 207 extracted for AAV5 bound with PKD1-2. The same 2.4 Å AAV2-PKD12 reconstruction 208 (EMD-0553) was low pass filtered, now to 50 Å and used as the initial reference for 209 alignment. For the complex of PKD1-2 with AAV5, extracted 3D subtomograms were 210 aligned using the new 2.88 Å SPA reconstruction of the PKD12-AAV5 complex, low 211 pass filtered to 50 Å. Subtilt refinement was then used to align the individual 2D particle 212 images in each tilt, and apply a per-particle-per-tilt CTF correction. The sub-tilt refined 213 3D particles were exported from EMAN2/e2tomo and then imported into the program I3 214 (68, 69). Icosahedral symmetry was applied to generate 60 copies of each particle such 215 11 that each possible asymmetric unit was overlaid onto the same frame of reference. The 216 particles were then translated and rotated to center on the 3-fold spike in a "spike up" 217 standard orientation. At this point, particles were re-extracted with box size of 90 x 90 218 pixels surrounding a single 3-fold spike, facilitating classification of asymmetric units. In 219 order to improve the signal/noise ratio, the re-extracted subtomograms, containing one 220 trimer, were binned by 2. Then classification was conducted on one asymmetric unit to 221 reveal the PKD1 domain (AAV2 bound with PKD1-2) or PKD2 domain (AAV5 bound 222 with PKD1-2). Even though the resulting classes are based on a single asymmetric 223 unit, the classes were re-expanded by three-fold symmetry to better illustrate the 224 context of the extra domains. 225

Model fitting 226
All the tomography maps for AAV2 and AAV5 bound with PKD1-2 were aligned to the 227 same frame of reference. High resolution SPA reconstructions EMD0553 and EMD9672 228 were aligned with the global (overall) average of the subtomograms of AAV2 bound with 229 57). Similarly, the map for the newly obtained 2.88 Å single particle 230 reconstruction for the AAV5/PKD1-2 complex was aligned to the global average 231 subtomogram of the same complex. PDBid 6ihb, with its well-ordered βA, was used for 232 the atomic model of the AAV2 viral protein, while PKD2 was taken from the higher 233 resolution PDBid 6nz0 (43, 57). Likewise, the atomic model of VP protein and PKD1 234 from PDBid 7kpn was docked as a single rigid body trimer into the newly obtained 2.88 235 Å reconstruction for the AAV5/PKD1-2 complex and used to interpret the tomography 236 maps. In summary, the tomographic reconstructions for AAV2 and AAV5 complexes 12 were calculated in the same frame of reference, then high resolution SPA 238 reconstructions and atomic models were overlaid, facilitating comparisons. 239 The previously unseen domains were modeled using the following. Atomic models for 240 PKD1 and PKD2 were taken from PDB entries 7kpn and 6nz0 respectively, and were 241 fitted, as rigid domains, into the newly revealed domain densities separately for the 4 242 AAV2 classes and the 3 AAV5 classes using Chimera (70). 243

Contact analysis 244
The VMD "atomselect" command was used to identify additional potential residue 245 contacts contributed by the AAVR domains that had not previously been resolved in 246 single particle analysis (Humphrey et al., 1996). Distances between the newly revealed 247 PKD1 (AAV2) / PKD2 (AAV5) and respective viral proteins were calculated. The 248 "atomselect" command lists the residue numbers of all residues that have any atom 249 approaching within 4.5 Å. 250

Results 251
The structure of AAV2 bound with PKD1-2 252 Cryogenic electron tomographic (cryo-ET) tilt series were acquired for AAV2 bound by 253 the PKD1-2 domains of AAVR and tomograms were reconstructed. AAV2 subvolumes 254 were aligned and then subdivided into individual trimers and all aligned trimers were 255 averaged ( Figure 1B1). This revealed densities at about 20 Å resolution, corresponding 256 to the VP capsid protein and PKD2, consistent with the published single particle 257 reconstructions of the AAV2/PKD1-2 complexes (43, 57). The atomic model of the 258 AAV2 spike fits the global average well, defining the density for the viral protein and the 259 PKD2 domain of AAVR ( Figure 1B2). As with the single particle analysis, the AAV2 viral 260 13 protein and the PKD2 domain are readily apparent, but there was no sign of the PKD1 261 domain. In order to reveal PKD1, tomographic subclassification was performed using a 262 trimer subvolume that extended mostly outwards beyond PKD2 for the AAV2/PKD1-2 263 complex. This revealed additional features corresponding to four distinct conformations 264 for PKD1 ( Figure 1C). The features are of the correct shape and size for a PKD domain, 265 and an atomic model of PKD1 fits well into the map of each class ( Figure 1D indicates that the PKD2 domain is more constrained when bound to AAV5 than PKD1 is 280 when it is bound to AAV2. The PKD2 domain becomes better defined upon 281 classification, focusing on the area outside PKD1, which yielded three distinct classes 282 ( Figure 2B). As with AAV2, the extra densities are the correct shape and size for a PKD 283 domain, and an atomic model of PKD2 fits well into the class maps ( Figure 2C). In each 284 case, the first residue of the PKD2 model is within 19 Å of the C-terminal residue of 285 PKD1, close enough to be bridged by the 5-residue domain linker. Variation in PKD2 286 orientation among classes of the AAV5 complex is modest ( Figure 2C) which is 287 consistent with conformations that are more constrained than the AAVR conformations 288 that we observed with AAV2. 289

Hybrid analysis: Integration with Cross-linking Mass Spectrometry (XL-MS) 290
Meyer et al. reported mass spectroscopic identification of amino acids in AAV2 and 291 AAV-DJ that were cross-linked with cyanurbiotindimercaptopropionylsuccinimide 292 (CBDPS) that has a spacer length of 14 Å (43). Atomic models can be compared with 293 these distance constraints, with the caveats that the tomographic classes are at ~20 Å 294 resolution and without definition of side chains, and that the distances are measured 295 from cryo-ET samples that were not cross-linked, so do not reflect any remodeling of 296 local structure on cross-linking (Table 1). 297 Given that at best nanometer-level consistency should be expected, class 4 provides 298 plausible explanation for 4 of 5 observed cross-links, one involving the tightly bound 299 PKD2 and three involving the C-terminus of PKD1 which is closest to PKD2 and the 300 virus surface. A rationalization of K338 cross-linking (minimal 36 Å) is more tenuous, 301 requiring remodeling of the lysine side chains. It seems more likely that tomography is 302 sampling four of many possible PKD1 orientations, and that any of the larger population 303 could be captured in cross-linking. In other words, the cross-linking reflects a highly 304 flexible receptor with many domain orientations, of which a subset, perhaps the most 305 stable, are sampled in the tomographic classes. 306

Comparison of the AAV2 and AAV5 complexes with PKD1-2 307
When AAV2 is bound by PKD1-2, the PKD2 domain has the highest affinity but PKD1 308 has measurable impact, while, for AAV5, PKD1 appears to be the only domain involved 309 (54). These results, coming from binding and transduction analysis of domain-swap and 310 deletion mutants are supported and rationalized by the current tomography study. The 311 tomography reveals other differences in the PKD1/PKD2 domain modes of binding to 312 the two serotypes. For PKD1-2 bound to AAV2, no density is revealed for the PKD1 313 domain in the global average of aligned subtomograms, indicating a high level of 314 heterogeneity of the PKD1 domain. For the AAV5 complex, even though weak, density 315 of the previously-missing PKD2 domain is apparent in the global average of the aligned 316 subtomograms. This difference between global averages indicates that PKD1 in AAV2 317 is more heterogeneous than PKD2 in AAV5. This is further confirmed by the classes for 318 the extra PKD1/PKD2. As shown in Figure 3, the extent of variability in AAV2 is much 319 higher than AAV5 with a wider range of orientations. Furthermore, for PKD1-2 bound to 320 AAV2, three out of four of the classes are in extended conformations with obtuse angles 321 between the two PKD domains and the fourth class has the two PKD domains folded 322 back on each other. For PKD1-2 bound to AAV5, the two domains are always at an 323 acute angle, folded back towards one another and contacting near the hinge in an 324 antiparallel hairpin configuration. The extra PKD1/PKD2 also differs on the contact with 325 viral proteins. Consistent with the accessory role of PKD1 in AAV2 cellular entry, one 326 the four classes of PKD1 appears to have some contact with the viral protein, whereas 327 for AAV5 bound there appear to be no contacts with PKD2 (see below). 328

Contact between newly revealed extra PKD1/2 with VP protein 329
The details of interactions between PKD2 with AAV2 and PKD1 with AAV5 have been 330 discussed in the respective high resolution single particle cryo-EM analyses (43,59). 331 Here, contact analyses are added for the newly revealed flexible PKD1 in its AAV2 332 complex and PKD2 in its AAV5 complex, but with a caveat that needs to be 333 emphasized. The resolution of the tomographic classes (and therefore precision of 334 atomic models) is low, ~20 Å. One must therefore be very cautious in interpreting 335 is extra density that projects in toward the center of the virus that is not accounted for by 347 the known AAV atomic models (Figure 4). Interestingly, the extra density was only 348 observed in our AAV2/PKD1-2 structure ( Figure 4A) but not the AAV5/PKD1-2 structure 349 ( Figure 4B). Model fitting shows that this protruding density is located in close proximity 350 to residue 237 of the VP protein ( Figure 4C). Maps for two previous AAV2/PKD1-2 351 17 single particle cryo-EM reconstructions are mostly quite similar, but differ here. The 352 Zhang et al. structure at 2.4 Å (EMD9672) is similar to most prior AAV structures, 353 interpretable from residue 219, with β-strand A, running anti-parallel to βB before a 354 hairpin turn at Gly236-Asp237 that connects the two strands (57). This map for βA is 355 slightly weaker than βB and other strands, but only slightly, with only a slight hint of 356 disorder (Fig. 4D) Relevant to these observations, it has been established that the AAV capsid is 363 assembled from VP1, 2, and 3 in a roughly 1:1:10 ratio, sharing much of their sequence 364 and structure, but differing in their N-terminal extensions (see Introduction) (71). 365 Disordered features in the same general area were first seen as "fuzzy globules" in a 366 2001 nanometer-resolution SPA reconstruction of empty AAV2 virus-like particles (72). 367 Assignment as parts of VP1 and/or VP2 was supported by structures lacking "fuzzy 368 globules" either for mutants in which VP1/2 were deleted, or in capsids following heat 369 treatment that was known to expose VP1u on the exterior of AAV (73). However, 370 doubts emerged with the absence of the "fuzzy globules" in reconstructions of AAV1 371 vectors with varying DNA content, again by the same group (74). For the most part, 372 these disordered features have not been noted in subsequent structures. However, they 373 did resurface in the cryo-EM reconstruction of the AAV2 R432A mutant, in which map 374 was missing at 3.7 Å resolution for βA. When this map was viewed at 5 Å resolution, 375 there was feature extending from the 236-7 hairpin turn that was interpreted as four 376 residues extending toward the general area of the "fuzzy globules" (75). The map at 5 377 Å resolution had not been deposited, so for comparison to our cryo-ET, the 3.7 Å map 378 (EMD8100) was low-pass filtered to 11 Å resolution. At this lower resolution, we see 379 not just 4 amino acids heading from the βA-βB turn towards the 2-fold (75), but 380 additionally we see the larger unmodeled feature that is also present in the AAV2-381 PKD12 tomography. Furthermore, this feature is the same as the "fuzzy globules" seen 382 earlier in AAV2 VLPs at 11 Å resolution (72). For both our PKD12 complex and the 383 R432A packaging mutant, higher resolution cryo-EM SPA indicates that the presence of 384 the unmodeled feature is accompanied by loss of βA, i.e. it is an alternative 385 configuration for the N-terminal residues. In the prior 3.8 Å reconstruction of the wild-386 type AAV2, βA was clear, and we now add that when low-pass filtered to 11 Å, 387 EMD8099 shows is no indication of the partially ordered alternative configuration seen 388 in the R432A mutant (75). 389 Whereas we see the partially ordered alternative configuration in our complex of AAV2 390 with the PKD1-2 fragment of AAVR, the nominally similar complex of AAV2 with a 391 PKD1-5 fragment has density for βA that is only marginally weaker than βB (EMD 9671) 392 (57), so the more usual configuration of βA predominates. Thus, there is not a simple 393 and deterministic receptor-triggered conformational switch. Indeed weak, but 394 recognizable density for Trp234 in the 2.4 Å structure of the PKD1-2 complex (43) 395 indicates an equilibrium (favoring the alternative configuration) that might reflect 396 incomplete PKD1-2 binding or an intrinsic and perhaps dynamic finely-balanced 397 19 equilibrium. The latter is consistent with a history of VLP structures where the 398 alternative configuration is seen occasionally (72) corresponding to 89 typically sized amino acids. Thus, the disordered region, which is 408 centered on a two-fold axis, could contain two copies of the N-terminal 35 residues of 409 VP3 (before βB). Alternatively, each could contain a single copy of either VP2 (65 + 35 410 = 100 residues before the βA-βB turn) or part of VP1 (202 + 35 = 237 residues), noting 411 that there are a total of ~12 copies combined of VP1 and VP2, but 30 x 2-fold axes. For 412 conformers headed towards the 5-fold, up to one in five would have access to the 413 exterior through the pore, while others might be part of the "basket-like" disordered 414 structure, seen surrounding the 5-fold on the inner surfaces of some, but not all AAVs 415 (76). 416

Comparison of single particle reconstructions for AAV5 bound with PKD1-2 417
A single particle analysis of the AAV5/PKD1-2 was also performed, using particles from 418 the same grid that was subject to tomographic analysis. The reconstruction, at 2.88 Å 419 resolution, agrees well with the previous 2.5 Å map (59), similarly resolving PKD1, but 420 20 not PKD2. The new SPA and tomography data were collected from the same sample 421 grid, so detection of PKD2 is a result of the technique not the sample ( Figure 5B-D). 422 While the high-resolution SPA structures of PKD1 are mostly very similar, there are 423 differences in the N terminal residues ( Figure 5A, arrows). This is at the same region 424 where the two prior single particle analyses had been modeled differently (58,59)  configuration. Given a lack of side chains, the identity of the extra features cannot be 441 determined unambiguously. They are best described as a β-hairpin "U" ( Figure 5B). 442 Hypothetically, one could account for the two arms of the "U" separately with two 443 21 additional configurations for βA-βA' of AAVR PKD1. The melting of β-sheet hydrogen 444 bonds to spring βA' loose seems implausible and there is not the diminution of density 445 for βA' expected if βA' had alternate conformers. 446 Using high contour levels for the new map, we see that the predominant configuration of 447 PKD1 is as modeled by Silveria et al. (Fig. 5). It is most likely that the new features 448 belong to a single peptide distinct from that previously modeled. In the discussion, two 449 possibilities will be considered, either that the unaccounted density is part of a second 450 AAVR subunit, or that it is a fragment of hitherto unseen N-terminal regions of an AAV 451 capsid protein. AAV2 presented more of an enigma, because the same mutational analysis found that 466 PKD2 was most important for AAV2 entry, but PKD1 also enhanced transduction, 467 though to lesser extent. PKD2, the more critical for transduction, had previously been 468 resolved by SPA (43, 57), but not the "accessory" PKD1. Prior to the SPA structures of 469 AAV5-AAVR complexes (58, 59), we hypothesized that the unseen PKD1 might be 470 interacting loosely with AAV2 at a site corresponding to the (yet to be determined) 471 AAV5-PKD1 interface. The cryo-ET shows that none of PKD1 locations of any of the 472 four classes in the AAV2 complex bear any resemblance to PKD1 as bound by AAV5. 473 However, one of the four PKD1 classes has some direct contact with AAV2 proteins. 474 This is consistent with PKD1 playing an accessory role not strictly required for, but 475 enhancing cellular transduction. Note, however, that only one of the four AAV2 classes 476 appears to make contact and the contact is not extensive. Thus, it is not surprising that 477 there can be the observed wide-ranging heterogeneity in domain orientation, the four 478 classes spanning a 120° rotation about the interdomain hinge. While we would expect 479 the more populated orientations to rise to the top of classification, there might well be 480 diversity beyond the four discretely classed orientations (as indicated by the XL-MS), 481 and it is not surprising that PKD1 was not detectable by SPA. Clearly the level of 482 interactions between PKD1 and either AAV2 or PKD2 are insufficient to restrict 483 conformational heterogeneity, so one wonders whether the interactions with AAV2 can 484 be strong enough to have a measurable direct impact upon transduction through avidity. 485 It seems more likely that either PKD1 increases the availability or stability of AAVR in a 486 state compatible with the binding of AAV2 to PKD1, or that there is a different step in 487 AAV entry in which PKD1 has a role. 488 23 Completely unanticipated was the unmodeled density on the inside surface of the 489 AAV2/PKD1-2 complex (but not the AAV5/PKD12 complex). It correlates inversely with 490 the strength of βA density, density for βA being much weaker when the unmodeled 491 features are seen. Thus, it appears that we are observing an equilibrium between two 492 states, one with an ordered βA extending from the 5-fold region, and the other with a 493 partially ordered N-terminal region coming from the inner surface protrusion, skipping 494 βA, and joining the jellyroll fold capsid protein at the βA-βB hairpin turn. The volume of 495 the inner protrusion is commensurate with that expected of the N-terminal 35 residues 496 of two VP3 meeting at a 2-fold axis, although one cannot rule out partial occupancy by 497 VP2 or VP1. Whether and how this equilibrium in N-terminus location is influenced by 498 receptor-binding far away on the outside surface are unknown. 499 Another surprise was the previously unseen fragments of β-strand structure adjacent to 500 PKD1 in its complex with AAV5. They lacked distinctive features to identify by 501 sequence. Nevertheless, there are a limited number of plausible possibilities. The N-502 terminal regions of the capsid proteins have never been seen at high resolution. While 503 in this study partially ordered structures were seen on the interior surface of AAV2, 504 crystal structures of some AAVs and autonomous parvoviruses have indicated that a 505 fraction of N-termini (of at least VP3) might be external: partially ordered density running 506 down the 5-fold pore from the outside is interpreted as the connection to the start of the 507 β-barrel on the inside surface (78-81). The absence of density on the 5-fold axis in the 508 AAV5 single particle analyses lessens the likelihood that the unaccounted features are 509 previously unresolved N-terminal parts of the viral protein outside the capsid. 510 24 Alternatively, the extra peptides could come from unmodeled regions of AAVR. Dimers 511 and higher oligomers are seen in preparations of PKD1-2 constructs (and MBP-PKD1-5 512 fusions) (43, 61). To date, AAVR dimers have not been observed bound to AAV5, but 513 one cannot exclude the possibility that a small fraction of receptors in the complex are 514 dimerized, with disorder that precludes EM observation of most of the second subunit. 515 This work is a testament to the value of combining multi-technique, multi-scale 516 approaches for flexible complexes, and in recognizing gaps in our understanding 517 through exclusive reliance on high resolution structure. A plan for multiple 518 contingencies involved not only integration of different EM techniques, but also 519 upstream redundancy in expression constructs, both of which were needed for a more 520 robust and holistic understanding. It is noted that the first application of cryo-ET, to a 521 complex of AAV2 with a PKD1-5 MBP fusion construct, led to a very low resolution 522 visualization that lacked domain definition or perception of conformational heterogeneity 523 (43). It was only with a smaller construct, His6-PKD1-2, that higher binding occupancy 524 was achieved and conformational heterogeneity from domains 3-5 was eliminated, 525 making it possible to classify the remaining heterogeneity and resolve distinct 526 configurations for the two proximal domains. On the technical side, it is noted that fully 527 automated classification of subvolume tomograms within a symmetrical particle was not 528 yet possible. It is hoped that examples like this will inspire on-going algorithm 529 development, so that future applications will not be limited by the laboriousness of 530 interactive classification.