Macromolecular Assembly of Polycystin-2 Intracytosolic C-Terminal Domain

Home , Polycystin 1, Polycystin 2, TRPV5, TRPV6

Macromolecular assembly of polycystin-2 intracytosolic C-terminal domain

Frederico M. Ferreiraa,b,1, Leandro C. Oliveirac, Gregory G. Germinod, José N. Onuchicc,1, and Luiz F. Onuchica,1

aDivision of Nephrology, University of São Paulo School of Medicine, 01246-903, São Paulo, Brazil; bLaboratory of Immunology, Heart Institute, University of São Paulo School of Medicine, 05403-900, São Paulo, Brazil; cCenter for Theoretical Biological Physics, University of California at San Diego, La Jolla, CA 92093; dNational Institute of Diabetes, Digestive, and Kidney Diseases, Bethesda, MD 20892-2560

Contributed by José N. Onuchic, April 28, 2011 (sent for review March 20, 2011)

Mutations in PKD2 are responsible for approximately 15% of the In spite of the aforementioned information and insights, the autosomal dominant polycystic kidney disease cases. This gene macromolecular assembly of PC2t homooligomer continued to encodes polycystin-2, a calcium-permeable cation channel whose be an open question. In the current work, we present the most C-terminal intracytosolic tail (PC2t) plays an important role in its comprehensive set of analyses yet performed and that show interaction with a number of different proteins. In the present PC2t forms a homotetrameric oligomer. We have proposed a study, we have comprehensively evaluated the macromolecular PC2 C-terminal domain delimitation and submitted it to a range assembly of PC2t homooligomer using a series of biophysical of biochemical and biophysical evaluations, including chemical and biochemical analyses. Our studies, based on a new delimitation cross-linking, dynamic light scattering (DLS), circular dichroism of PC2t, have revealed that it is capable of assembling as a homo- (CD) and small angle X-ray scattering (SAXS) analyses. We have tetramer independently of any other portion of the molecule. also used molecular dynamics simulations employing an all-atom Our data support this tetrameric arrangement in the presence and structure-based model (SBM) to evaluate the macromolecular as- absence of calcium. Molecular dynamics simulations performed with sembly of PC2t, by comparing the structures obtained from the a modified all-atoms structure-based model supported the PC2t tet- simulation including the raw SAXS data and by reconstructing rameric assembly, as well as how different populations are disposed the distribution of states in solution using a system forced by en- in solution. The simulations demonstrated, indeed, that the best- tropy. The simulations determined, in fact, that the best-scored BIOPHYSICS AND scored structures are the ones compatible with a fourfold oligomeric structures are also the ones compatible with tetramerization. state. These findings clarify the structural properties of PC2t domain COMPUTATIONAL BIOLOGY and strongly support a homotetramer assembly of PC2. Results In Silico Analysis of PC2t Molecular Properties. The domain topology utosomal dominant polycystic kidney disease (ADPKD) is of PC2 was initially examined upon the analysis of its primary Athe most common monogenic life-threatening human dis- sequence (Fig. 1). In silico analysis of PC2t polypeptide sequence ease, with a prevalence of 1∶400 to 1∶1;000 (1). Whereas muta- revealed an inverse correlation between hydrophobicity and tions in the PKD1 gene are responsible for approximately 85% tendency of disorder, with significantly low hydrophobicity and of the disease cases, nearly 15% are caused by pathogenic muta- high tendency of disorder in the transition between the trans- tions in PKD2 (1). The PKD2 gene encodes polycystin-2 (PC2), a membrane and intracellular domains, aa 649–681 (Fig. S1). A 968-amino acid (aa) membrane protein with six transmembrane similar pattern was observed for flexibility, particularly for the helices (TM) and both C- (PC2t) and N-termini intracytosolic interface residues between these two domains, where the residues (1, 2). PC2 and the membrane protein polycystin-1 (PC1), the are highly hydrophobic and structurally organized (Fig. S2). The PKD1 gene product, are thought to physically interact by their PC2t middle region (aa 800–830) shows remarkably high average C-termini (3). In recent years, a series of studies have implicated flexibility and tendency of disorder, where a linker is positioned the primary apical cilia in the pathogenesis of polycystic kidney between the EF-hand and the coiled-coil subdomains. The same disease. PC2 is a member of the transient receptor potential behavior was detected in the PC2t final region (aa 940–968). (TRP) channel superfamily (TRPP2), acting as a nonselective ca- These results, therefore, support the assignment of the intracyto- tion channel involved in calcium transport (1, 4, 5). The PC1–PC2 plasmic PC2t domain to the 289 C-terminal residues 680–968. complex plays a key role in this ciliary mechanism, by apparently It must be noted that PC2t is highly charged, with a number of translating physical or chemical stimuli in calcium influx through negatively charged residues yielding theoretical pI of 5.1 and PC2, a process that is followed by significant calcium release from molecular mass of 32.6 kDa. intracellular stores (6). The PC2 topology predicts the pore structure to reside in TMs Dynamic Light Scattering. To access the monodispersity of PC2t 5 and 6, whereas a sensor portion is expected in TMs 1–4 (2, 7). samples and for a rough estimation of its maximum dimension, TRP channels have been shown to form homo and/or heteromul- we performed DLS experiments. Such experiments revealed the timers, a property apparently correlated with quantitative and PC2t samples in the monodispersed state. The estimation of the qualitative aspects of their function (8). Full-length PC2 appears oligomer maximum dimension was 17 4 nm. to follow this pattern, because a recent study supports PC2 assembling as a homotetramer and as a PC2-TRPC1 heterotetramer PC2t Expression, Purification, and Molecular Mass Estimation. The (7). Previous work has suggested that a coiled-coil subdomain coding sequence corresponding to PC2t was amplified by identified in the PC2t (PC2t-CC) plays critical roles in PC2 oligomerization and in its physical interaction with PC1 to form the PC1–PC2 complex, a process that also involves a coiled-coil Author contributions: F.M.F., L.C.O., G.G.G., J.N.O., and L.F.O. designed research; F.M.F. and L.C.O. performed research; F.M.F., L.C.O., J.N.O., and L.F.O. analyzed data; and subdomain in PC1t (1). Another recent study, however, suggests F.M.F., L.C.O., G.G.G., J.N.O., and L.F.O. wrote the paper. þþ that the PC2t-CC assembles as a homotrimer (9). A Ca -bind- The authors declare no conflict of interest.

ing EF-hand subdomain has also been predicted in PC2t and 1 þþ To whom correspondence may be addressed. E-mail: [email protected], [email protected], been proposed to function as a Ca -sensing switch (2, 10), or [email protected]. whereas the PC2 channel activity has been shown to be regulated This article contains supporting information online at www.pnas.org/lookup/suppl/ by calcium (4, 5, 11). doi:10.1073/pnas.1106766108/-/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.1106766108 PNAS ∣ June 14, 2011 ∣ vol. 108 ∣ no. 24 ∣ 9833–9838 Downloaded by guest on September 25, 2021 to chemical cross-linking assays, shown in Fig. 2D. A blurred band corresponding to the PC2t oligomer was detected, with a middle- position height slightly above the 116-kDa marker. The profile of the log(MM) vs. Rf plot is sigmoidal, so the assumption of linearity is valid only for a finite but useful range (Fig. S5). Because the nonlinearity is even more pronounced for the higher MM markers, a polynomial regression was applied. The apparent mass Fig. 1. (A) Topology of the PC2 predicted domains: in green, the cytosolic calculated for the cross-linked sample was 123 kDa, a mass com- domains; in cyan, the putative extracellular domain; and in magenta, the patible with a tetramer, whereas a value of 36 kDa was obtained transmembrane domain. (B) Topology of PC2t subdomains and motifs as for the negative control sample in accordance to that expected for predicted by PRODOM: in red, EF-hand subdomain and its motif (arrow); a monomer. and in yellow, the coiled-coil subdomain and its motif and the MS sequenced residues (arrow). Structural Features of PC2t Monomer Revealed by Mass Spectroscopy. C standard PCR and cloned in an expression vector. Successful Fig. 2 shows the PC2t controlled proteolysis with trypsin, reveal- PC2t overexpression followed by western blot analysis of the so- ing a structural core of approximately 14.4 kDa insensitive to such luble fraction of cell lysate are shown in Fig. 2A, confirming a a proteolytic effect. This finding is, in fact, inconsistent with the in band immediately above the 30-kDa marker. silico trypsinization pattern obtained for PC2t primary sequence, The PC2t molecular mass (MM) estimation from size-exclu- which predicts 40 restriction sites leading to peptides with masses sion chromatography (SEC) revealed an apparent mass of 354 ranging from 147.1 Da (a lysine) to 2845.4 Da (27 residues). The kDa, a mass consistent with an oligomer of order 11 (Fig. S3). observed data require residues organized in a tridimensional The PC2t MM estimated by native gel electrophoresis (N-PAGE; molecular folding, making the remaining arginine and lysine re- Fig. 2B), on the other hand, was 137 kDa, a value consistent with striction sites not accessible to trypsin activity. Interestingly, the a tetrameric arrangement (Fig. S4). The presence of calcium presence of this molecular core was not originally expected, given the intrinsically disorder probability of the primary sequence. neither significantly altered PC2t retention time obtained in SEC Mass spectroscopy analysis determined the sequence of two experiments nor in the migration pattern observed in N-PAGE peptides resulting from the PC2t macromolecular core. Table S1 analyses. displays the matching between the experimental ionic masses and those predicted by in silico trypsinization. As the peptide se- PC2t Chemical Cross-Linking. To further examine the PC2t homoo- quences were univocally determined and the chain is unique, it ligomerization assembly, we extended the experimental approach was possible to conclude that the primary sequence of the molecular core has at least 71 residues, comprising aa 827–897. This region includes the beginning of the PC2t coiled-coil subdomain and corresponds to 8.1 kDa of the molecular core mass. Because the complete sequence of the molecular core could not be estab- lished, its remaining portion(s) might theoretically flank either or both sides of the sequenced fragment.

Circular Dichroism Analysis of PC2t. The far-UV PC2t spectra evaluation supports a calcium-sensitive structural organization at the secondary level (Fig. S6). The PC2t circular dichroism spectrum shows double ellipticity minima at 208 and 222 nm and a maximum ellipticity at 198 nm, findings consistent with an α-helical configuration. The 208-nm minimum is deeper than that of the 222-nm and slightly shifted to a smaller wavelength due to the presence of disordered residues. Complete calcium depletion in the solution, in turn, decreased the ellipticity at 222-nm and 208-nm minima, and decreased its maximum at 198 nm. The 208-nm minimum, moreover, was further displaced to 206 nm, supporting an increase of disordered residues. Deconvolution of the PC2t spectrum in presence of calcium showed secondary structure contends of 68% of helices, 10% of strands, 10% of turns, and 12% of disordered residues. The complete removal of calcium, on the other hand, led to secondary structure contends of 56% of helices, 17% of strands, 8% of turns, and 18% of disordered residues.

Fig. 2. (A) PC2t expression is confirmed by immunoblotting with antihistag SAXS Solution Analyses of PC2t. The PC2t particle envelope was (Sigma–Aldrich H1029): lane 1, MM standard markers; lanes 2 and 3, respec- assessed by SAXS data solution analyses. Fig. 3A displays the tively, soluble and insoluble fractions of overexpressed PC2t; and lane 4, SAXS curves of PC2t in presence of 2.0 mM CaCl2 (PC2t-Ca) positive western blotting showing PC2t immediately above the 30-kDa and 2.0 mM EGTA (PC2t-EGTA). These results revealed MM marker in crude extract. (B) PC2t N-PAGE: lane 1, MM standard markers; PC2t-Ca and PC2t-EGTA apparent masses of 129 kDa and and lane 2, PC2t oligomer immediately bellow the 140-kDa MM marker. 139 kDa, respectively. Guinier analyses showed that the PC2t (C) Controlled proteolysis of PC2t with trypsin: lane 1, MM markers; lanes 1∶250 samples were diluted and monodispersed in both conditions, 2 to 5, PC2t samples treated with protease:protein mass ratio at 4 °C as demonstrated by the linearity of the curves at very low angles. for 5, 15, 30, and 60 min, respectively. The PC2t structural core is observed at the same height as the 14.4-kDa MM marker. (D) Chemical cross-linking of The observation that both linear extrapolations are virtually par- PC2t samples: lanes 1 and 2, MM markers; lanes 3 to 5, treated with EGS allel, in turn, indicates that the PC2t Guinier radii are very close: 1 mM for 60, 40, and 20 min of incubation, respectively; lanes 6 to 8, EGS 52 2 Å in the presence and 54 1 Å in the absence of calcium. 0.5 mM for 60, 40, and 20 min of incubation. In the reciprocal space, both curves superimpose very well in the

9834 ∣ www.pnas.org/cgi/doi/10.1073/pnas.1106766108 Ferreira et al. Downloaded by guest on September 25, 2021 Fig. 3. (A) PC2t small angle X-ray scattering intensities measured at D11A-LNLS in the presence of 2.0 mM CaCl2 or 2.0 mM EGTA and their corresponding Guinier extrapolations (inside); (B) the distance distribution functions between electron density pairs for PC2t in both conditions; (C) the respective Kratky plots; and (D) superimposition between the theoretical and experimental scattering curves from different reconstruction models. The curves were arbitrarily displaced on the vertical axis for clarity.

low angle regimen. Such superimposition, however, is not overall folding, followed by selection of the closest model to the observed for wider angles in the second and third sectors of the centroid of the largest clusters, were applied to identify a more scattering curves (Fig. 3A). Those differences are better appre- representative ensemble of conformers. One hundred indepen- BIOPHYSICS AND

ciated in the real space by comparing their respective distance dent reconstructions were also performed for each calcium COMPUTATIONAL BIOLOGY distribution functions between electron density pairs [P(r)], gen- and EGTA dataset using the entire resolution range with the pro- erated by the indirect Fourier transform of the scattering curves gram GASBOR (16). The calcium restorations were clustered in with the GNOM program (12, 13). The PC2t P(r) functions in the four groups, whereas the EGTA set was clustered in nine groups presence and absence of calcium are shown in Fig. 3B, revealing (the isolated clusters were discarded). The superimposition of the distance distributions slightly differ. In presence of calcium, PC2t theoretical and experimental scattering curves is shown in Fig. 3D. presents a narrower P(r), centered in a smaller gyration radius The comparison between the PC2t models from both restoration (Rg) as a consequence of a smaller maximum dimension (Dmax). methods are represented in Table 1, whereas the SAXS models In contrast, the absence of calcium suppresses the typical P(r) from both reconstruction methods are presented in Fig. 4. multidomain shoulders and oscillations corresponding to intra- and intersubunit distances, and Rg and Dmax swell toward a wider Tridimensional Structure Simulation and Analysis of PC2t Homooligo- distribution of distances. The expansions of Rg and Dmax deter- mer. The I-TASSER C-Score function was employed to predict 55 9 0 5 mined by calcium removal were from . . Å and 175 Å to models based on alignments and convergence of structural as- 56 8 0 6 . . Å and 188 Å, respectively. Calcium-induced differ- semblies (17). Despite their limitations, these models are suffi- ences in the PC2t scattering intensities were also detected by C cient to investigate the main features of the protein assembly. the Kratky plot (Fig. 3 ). The PC2t-Ca Kratky plot presents a Three of the best-scored models (Fig. S7) were chosen to simu- narrow maximum at low-resolution range and a more dispersive late the trimer, tetramer, and pentamer arrangements (Fig. 5). profile for higher angles. Calcium removal, in turn, slightly shifted Only the lowest score configuration is in satisfactory agree- the low-resolution maximum to lower s, indicating that the R g ment with the SAXS data. The computational approaches are value is larger than the one obtained in the presence of calcium. described in SI Appendix. Twelve simulations of each possible In addition, some organization was gained for wider angles. These findings suggest a PC2t multidomain arrangement whose Table 1. Average structural parameters of the dummy residue flexibility is modulated or altered upon calcium binding. models (DRM) from both restoration methods The low-resolution particle shapes of PC2t-Ca and PC2t- EGTA homomultimers were restored from the scattering inten- DAMMIN GASBOR sities using two ab initio procedures. The particle envelopes were ¼ 0 15 −1 PC2t-Ca PC2t-EGTA PC2t-Ca PC2t-EGTA primarily calculated for low-resolution data (smax . Å ) with the simulated annealing protocol implemented in the DAM- Mass (kDa) 129.0 (1.1%) 139.2 (6.7%) MIN program (14). No symmetry was applied because higher or- Dmax (Å) 175 188 175 188 2 Rg (Å) 55.2 56.2 55.8 56.5 der restraints neither improve the χ fit between the theoretical 3 V (nm ) 3,899 4,201 44.9 44.9 and experimental scattering profiles, nor stabilized the normal- Number of 5,594 5,642 1,156 1,156 ized standard discrepancies (NSD) among several independent dummy residues reconstructions. Thirty independent simulations for the calcium Free parameters* 8,306 8,590 17.13 17.52 and EGTA sets were filtered and averaged with the DAMAVER Discrepancy χ 0.70 0.45 1.75 1.38 program (15). All of the ab initio models obtained from SAXS Resolution† 42.1 42.1 20.1 20.1 data yielded similar physical parameters. In the second ab initio These data include DAMMIN and GASBOR output parameters for PC2t in method, the entire sets of the scattering data were employed in an the presence and the absence of calcium. The protein molecular mass was attempt to identify the PC2t structural subdomains. The unique- obtained using SAXSMOW. ¼ ð − Þπ−1 ness of the shape restorations was also verified by comparing the *Number of Shannon channels is given by Ns Dmax smax smin . † ¼ 2π∕ NSDs of the superimposing solutions. Cluster analyses based on R smax.

Ferreira et al. PNAS ∣ June 14, 2011 ∣ vol. 108 ∣ no. 24 ∣ 9835 Downloaded by guest on September 25, 2021 SAXS data using CRYSOL (20). The best fit model for all simulations was achieved for a PC2t tetrameric arrangement in both calcium and EGTA conditions. Fig. 5B shows the superimposition between the best fit model and the PC2t-Ca experimen- χ ¼ 11 4 ¼ 51 8 C–E tal data ( . and Rg . Å), whereas Fig. 5 display the superimposition between the corresponding model and χ ¼ 7 5 ¼ 53 8 the PC2t-EGTA experimental data ( . and Rg . Å). Simulations allowing increased degrees of deformation have not improved the χ values. An optimized combination of the best simulated models (SM) enhanced the agreement between the theoretical and experimental data for the EGTA condition. The combination of 70% of the χ ¼ 7 5 ¼ 53 8 χ ¼ 8 4 SM1 model ( . and Rg . Å), 20% of SM2 ( . and ¼ 53 5 χ ¼ 8 6 ¼ 54 0 Rg . Å), and 10% of SM3 ( . , Rg . Å) yielded a combined χ of 7.1. Combinations of the calcium-simulated models, on the other hand, have not improved the agreement between theoretical and experimental data in comparison to results obtained with individual models. Discussion In addition to a coiled-coil subdomain, the PC2 C-terminal intracytosolic domain includes an EF-hand subdomain and a flexible linker (10), structural features consistent with our results. This C-tail is highly relevant to the protein biology, interacting with a number of binding partners (17). In the current study, in silico Fig. 4. Final averaged PC2t SAXS envelopes. GASBOR reconstructions for analyses based on flexibility, hydrophobicity and tendency of (A) PC2t-EGTA and (B) PC2t-Ca complete datasets; and DAMMIM reconstruc- disorder have allowed a suitable delimitation of the polycystin- tions for (C) PC2t-EGTA and (D) PC2t-Ca low-resolution data. (Models in blue 2 C-terminal cytosolic domain. After a series of procedural ad- are Ca and in red are EGTA; all models are presented in the same orientation). justments, an appropriate level of soluble expression of the entire trimer, tetramer, and pentamer arrangements were performed PC2t was achieved. Construct nucleotide sequencing, positive immunochemical assays and mass spectroscopy-based peptide using a modified version of the all-atoms SBM (18, 19), with dif- sequencing of proteolysis-derived products ensured the accuracy ferent constraints from minimal to maximal monomer separation. of each step. The assemblies obtained were compared to the experimental The PC2t apparent mass obtained from size-exclusion chromatography was in disagreement with the results yielded by the other methods. It must be noted, however, that these SEC results were interpreted as a large deviation of the PC2t oligomer from globularity. This conclusion was corroborated by the maximum dimension derived from DLS and SAXS analyses. N-PAGE results, moreover, revealed an apparent mass highly agreeable with the value derived from its aminoacid sequence for four iden- tical subunits (relative error of 5%). Charge effects were not observed, because the PC2t pI and the running pH were sufficiently close. The tetrameric hypothesis was also supported by the mass estimation of PC2t oligomer resulting from the chemical cross- link analysis (relative error of 6%). A second band of higher MM was also observed when using higher EGS concentrations. The experiment shows, indeed, that this band is an EGS concentration and time artifact, because it is absent in the set of experiments run under lower EGS concentration. This explanation is particularly strengthened considering the EGS concentration range (50-fold excess). A third independent set of results supports a tetrameric organization for PC2t. The SAXS MM estimation, performed in solution, revealed a high level of agreement between the PC2t determined mass and the mass expected for a tetrameric arrangement, both in the presence and in the absence of calcium (relative Fig. 5. (A) Monomeric molecular model of PC2t predicted by I-TASSER. Five errors of 1% and 7%, respectively). The PC2t MM difference different predictions were made but only this configuration is in satisfactory observed in our analysis was in fact expected because different agreement with the SAXS data. (B) Superimposition between the PC2t-Ca ex- volumes were found in the two conditions but the MM estimation perimental data (blue) and the theoretical scattering curve from the simu- was calculated using the same density, 1.37 gcm−3 (18). The lated model SM1-Ca (black). (C), (D), and (E) Superimposition between the three outlined sets of data, therefore, lead to a proposed model PC2t-EGTA experimental data (red) and the theoretical scattering curve, re- of tetrameric assembly for PC2t. This model is supported by data spectively, for the tetrameric simulated models SM1-EGTA, SM2-EGTA, and from Zhang et al., based on work with the full-length product and SM3-EGTA (black). SM1-EGTA shows a good fit in the very low angle region, SM2-EGTA shows a better fit in the middle angle region and SM3-EGTA shows multimer assembly evaluation with atomic force microscopy, a better fit in the higher angle region (all of them indicated by arrows). (F) followed by consistent functional results (7). Notably, the homo- Superimposition between the PC2t-EGTA scattering data and the theoretical tetrameric arrangement is also observed in other members of the þþ scattering curve as a combination of the models presented in (C), (D), and (E). TRP cation channel superfamily, including the Ca -permeable

9836 ∣ www.pnas.org/cgi/doi/10.1073/pnas.1106766108 Ferreira et al. Downloaded by guest on September 25, 2021 channels TRPV6 (19) and TRPC1 (20). A recent study proposes were applied. The combination of three simulated models at dif- a tetrameric assembly of PC2 including the C- and N-terminal ferent ratios yielded the same χ value whereas providing informa- dimerization domains (21). Our data, however, have demon- tion on the flexibility of the system. This approach, however, strated that the homotetrameric assembly can occur indepen- can be employed to obtain details on the geometrical nature of dently of PC2 N-terminus. This is a key structural point, the oligomerization interface. Structure-based simulations have showing that the C-tail itself is capable of directing assembly energy functions based on the predicted structures. In both cases, of PC2. however, our simulation results are in accordance with the experi- In the initial stages of refining the purification protocol, we mental findings, supporting our proposed PC2t homotetrameric observed a coexisting lower band in some of our N-PAGE experi- arrangement. The comparison of theoretical profiles and experiments that appeared to be related to the assembly of noncano- mental SAXS data demonstrated that the trimeric or pentameric nical thrombin cleavage products. Whereas we cannot exclude arrangements are inconsistent (see details in SI Appendix). the possibility that this band is a homotrimeric form of PC2t, Mixing multiple structures does not substantially improve the it would be formed in very minor quantities under specific PC2t analysis, indicating further assembly rigidity and therefore physicochemical conditions, coexisting with far predominating higher organization. This supports the notion that calcium in- homotetramers. creases the assembly stability. This region also provides informa- The PC2t oligomerization state may have become a controver- tion about the monomer extension and compactness of the sial issue due to interpretation of data based on incomplete PC2 assembly. The intermediate s region shows the presence of multiple C-terminal delimitation. Although a functional PC2 trimeric conformations. By simply mixing three representative structures, a channel has not been reported to present, recent studies are better theoretical adjustment to the experimental data is obtained. compatible with this hypothetical arrangement (9, 22). These The large s region shows the existence of dynamical fluctuations analyses, however, were performed with incomplete PC2ts. Our varying from local ones all the way up to assembly compactness. proposed PC2t delimitation, in turn, shows that the full-length Our dynamic oligomerization hypothesis was also supported by PC2 cytoplasmic domain encompasses regions that are necessary the SAXS data solution, using a recently developed approach and sufficient for the homotetramer formation that are absent in (23–25). The applied molecular dynamics strategy was proposed PC2t segments previously evaluated. for protein quaternary structure elucidation, particularly to such Our MS results are consistent with a PC2t multiflexible- a flexible and complex system, and should be useful for other sys- domain organization. Controlled proteolysis showed that at least tems. Our results, in fact, are in accordance with all experimental

– BIOPHYSICS AND 71 aa, residues 827 897, are folded in a tridimensional arrange- analyses. The homotetrameric molecular model for PC2t should

ment. Most of the 40 restriction sites were found exposed to serve as a starting point to focus on questions directed to the COMPUTATIONAL BIOLOGY trypsin activity, suggesting sequence flexibility and/or partial un- PC2 channel architecture, gating mechanism and roles in ADPKD. folding. This structural concept is supported by the CD analyses, which showed an increasing number of disordered residues in the Methods absence of calcium. This effect was also observed for wider angles Computational Analysis. Analysis of PC2t primary sequence allowed the iden- in the PC2t Kratky analysis. tification of the membrane interface residues and its intracytosolic domain. The PC2t N-PAGE, DLS, and the Guinier analyses for calcium We were able to exclude the last TM residues without removing aminoacids and EGTA conditions indicate that the SAXS experiments were potentially important for folding by quantifying hydrophobicity (26), mean conducted with homogeneous and monodispersed samples, flexibility (27) and tendency of disorder (28). The MARCOIL program (29) validating the subsequent data treatment. The experimental P(r) was used for coiled-coil prediction and PRODOM (30) for scanning known structural motifs and subdomains. profiles, Dmax and Rg from samples in the presence and absence of calcium suggest that the PC2 intracellular domain adopts a multilobular prolate arrangement. In addition, the Kratky plot Cloning, Expression, Purification, and Dynamic Light Scattering. Specific primers were designed for unidirectional cloning of the PKD2 nucleotide segment analysis revealed an interesting mechanism of calcium-induced corresponding to PC2t. Several plasmids and cell lines were tested to define conformational change. Both independent ab initio reconstruc- the appropriate insert-vector system (31). PC2t was purified from cell lysate tions for PC2t-Ca and for PC2t-EGTA molecular envelopes using immobilized affinity chromatography, following a size-exclusion chro- yielded very consistent results (Table 1). The data maximum re- matography step. solution did not allow determination of the spatial positions of The monodispersity of the PC2t samples and the maximum dimension of their secondary structure elements, but allowed the obtainment the oligomer was estimated using the right angle dynamic light scattering of the overall PC2t shapes in the presence and the absence of analyzer Zetasizer μV (Malvern Instruments). PC2t samples of 60 μM were calcium. These data provide important insights into the relative analyzed at 10 °C in the same conditions adopted in the SAXS experiments. position of their subdomains. It should be emphasized that the PC2t P(r) observations in both conditions were translated into Estimation of PC2t Molecular Mass and Chemical Cross-linking. The PC2t homo- their respective molecular envelopes. As made clear in Fig. 4, multimer MM was estimated by native gel electrophoresis performed in a polyacrylamide gradient of 10% to 15% (M/V), 112 mM acetate and PC2t presents a multilobular prolate shape, a feature even more 112 mM Tris at pH 6.4, using the PhastSystem (GE Healthcare). PC2t MM pronounced for the wider-angle reconstructions. The calcium- was derived from the linear regression of the log(MM) vs. Rf (retardation induced conformational modifications can be also observed when factor) plot. An additional estimation of PC2t MM was obtained using comparing the PC2t molecular envelopes associated with the pre- size-exclusion chromatography on a Superdex200-10/300 column (GE Health-

sence and the absence of this cation. The analysis of wider-angle care). PC2t MM was estimated from linear regression of the log(MM) vs. Kav reconstructions revealed a lower number of clusters for the PC2t- plot, where Kav is the gel-phase distribution coefficient. Ca condition, suggesting a more rigid packing as outlined by For chemical cross-linking analyses, 20-μM PC2t samples in PBS were incu- the Kratky analysis. Our data suggest that the slender portion bated in the presence of EGS to the final concentrations of 0.5 mM and of the PC2t-EGTA wider-angle molecular envelopes includes 1.0 mM. The MM of the PC2t cross-linked samples were calculated from the coiled-coil subdomain and the bulk part of the model encom- the polynomial regression of the log(MM) vs. Rf plot. passes the EF-hand subdomain, whereas the final flexible region Mass Spectroscopy. Controlled trypsinization was carried out at several pro- of PC2t remains to be determined. Higher resolution techniques, tease:protein molar ratios, temperatures and incubation times. Products of however, have to be applied to confirm this arrangement. interest yielded by proteolysis were purified by electrophoresis on 16% tri- A single rigid structure is expected to bring limited information cine gels, excised and submitted to an additional proteolytic process. The MS from a SAXS analysis. Better adjustments were obtained for the evaluation of the resulting peptides was carried out with a quadrupole time- PC2t-EGTA data when appropriate mixtures of simulated models of-flight mass spectrometer micrOTOF-QII (Bruker), in the direct injection

Ferreira et al. PNAS ∣ June 14, 2011 ∣ vol. 108 ∣ no. 24 ∣ 9837 Downloaded by guest on September 25, 2021 mode. The quantified ionic masses were compared to primary sequence Automated molecular modeling was also applied for PC2t monomer folding databases using the search engine MASCOT (32). using the online I-TASSER server 3D protein predictor (38).

Circular Dichroism. The PC2t circular dichroism spectra were assessed by far- Conformational Searching Employing Structure-Based Models. SBM simulations UV CD spectroscopy using a J-815 Jasco spectropolarimeter equipped with a have been very successful in understanding functional and folding aspects of Peltier temperature control unit. The CD spectra represented the average of protein complexes (23, 39–43), including interpretation of SAXS data for pro- 20 accumulations, using a scanning speed of 50 nm∕ min, a spectral band- tein complexes (24). In this study, we have added additional terms to the SBM width of 0.5 nm, and a response time of 1 s. PC2t samples of 1 to 2.5 μM were energy function (described in detail in SI Appendix) to provide us the free- measured in 10 mM NaCl, 10% glycerol in 5 mM Tris, pH 7.0, and 2.0 mM dom to adjust the theoretical model to the appropriate protein monomer CaCl2 or EGTA. The CD spectra deconvolution was performed with the elongation and the degree of compactness of the full complex. program CDSSTR (33) using the reference database SP175 (34). ACKNOWLEDGMENTS. We thank Glaucius Oliva for financial support SAXS Data Collection and Processing. Small angle X-ray scattering data were [Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq) collected at the D11A-SAXS beamline (35) in the Brazilian Synchrotron Light INCT-INBEQMeDI] and access to his laboratory. We thank Hannes Fischer Laboratory electron storage ring. Scattering data covering momentum for valuable discussion and manuscript revision; Andressa P. A. Pinto for −1 −1 transfers between 0.01 Å < s < 0.3 Å * were recorded with a MAR145 im- excellent technical support; and Paul C. Whitford for helpful suggestions age plate detector (Marresearch), using the 1.1488 Å wavelength, sample- on modeling. This work was mainly supported by Brazilian agency Fundação detector distance of 1290.98 mm and samples from 3 to 8 mg mL−1. The sam- de Amparo à Pesquisa do Estado de São Paulo (FAPESP) Grant 2006/ 03098-8 (F.M.F. and L.F.O.) and LIMs of University of São Paulo School of ple and buffer scattering image intensities were normalized to the intensity Medicine. Additional support was provided by The Johns Hopkins Polycystic of the incident beam for fluctuation correction and subtracted, following the Kidney Disease (PKD) Research and Clinical Core Center [National Institutes of integration with the FIT2D program (36). The scattering data were processed Health (NIH) P30 DK090868]. L.C.O. was partially supported by CNPq. Work in and normalized against the forward scattering intensity. The PC2t MM, in San Diego was supported in part by the Center for Theoretical Biological turn, was determined by the SAXSMOW program (37). Physics sponsored by the National Science Foundation (NSF) Grant PHY-0822283 and NSF Grant MCB-1051438. Tridimensional Structure Prediction. A simulation approach was used to char- acterize the complex PC2t tridimensional arrangement. In the current work, *The scattering vector s ¼ 4 sinðθÞλ−1, where 2θ is the scattering angle. All envelops and molecular dynamics and SAXS were combined as sources of information. molecules were drawn using PyMOL (www.pymol.sourceforge.net).

1. Torres VE, Harris PC (2009) Autosomal dominant polycystic kidney disease: The last 24. Yang S, Park S, Makowski L, Roux B (2009) A rapid coarse residue-based computational 3 years. Kidney Int 76:149–168. method for X-ray solution scattering characterization of protein folds and multiple 2. Mochizuki T, et al. (1996) PKD2, a gene for polycystic kidney disease that encodes an conformational states of large protein complexes. Biophys J 96:4449–4463. integral membrane protein. Science 272:1339–1342. 25. Jamros MA, et al. (2010) Proteins at work: a combined small angle X-RAY scattering 3. Qian F, et al. (1997) PKD1 interacts with PKD2 through a probable coiled-coil domain. and theoretical determination of the multiple structures involved on the protein Nat Genet 16:179–183. kinase functional landscape. J Biol Chem 285:36121–36128. 4. González-Perrett S, et al. (2001) Polycystin-2, the protein mutated in autosomal 26. Kyte J, Doolittle RF (1982) A simple method for displaying the hydropathic character of dominant polycystic kidney disease (ADPKD), is a Ca2þ-permeable nonselective cation a protein. J Mol Biol 157:105–132. channel. Proc Natl Acad Sci USA 98:1182–1187. 27. Bhaskaran R, Ponnuswamy PK (1988) Positional flexibilities of amino acid residues in 5. Vassilev PM, et al. (2001) Polycystin-2 is a novel cation channel implicated in defective globular proteins. Int J Pept Protein Res 32:241–255. intracellular Ca(2+) homeostasis in polycystic kidney disease. Biochem Biophys Res 28. Dosztányi Z, Csizmók V, Tompa P, Simon I (2005) The pairwise energy content esti- Commun 282:341–350. mated from amino acid composition discriminates between folded and intrinsically – 6. Zhou J (2009) Polycystins and primary cilia: primers for cell cycle progression. Annu Rev unstructured proteins. J Mol Biol 347:827 839. Physiol 71:83–113. 29. Delorenzi M, Speed T (2002) An HMM model for coiled-coil domains and a comparison – 7. Zhang P, et al. (2009) The multimeric structure of polycystin-2 (TRPP2): Structural- with PSSM-based predictions. Bioinformatics 18:617 625. functional correlates of homo- and hetero-multimers with TRPC1. Hum Mol Genet 30. Servant F, et al. (2002) ProDom: Automated clustering of homologous domains. Brief – 18:1238–1251. Bioinform 3:246 251. 8. Catterall WA (2000) Structure and regulation of voltage-gated Ca2þ channels. Annu 31. Berrow NS, et al. (2006) Recombinant protein expression and solubility screening in – Rev Cell Dev Biol 16:521–555. Escherichia coli: A comparative study. Acta Crystallogr D Biol Crystallogr 62:1218 1226. 32. Perkins DN, Pappin DJ, Creasy DM, Cottrell JS (1999) Probability-based protein identi- 9. Yu Y, et al. (2009) Structural and molecular basis of the assembly of the TRPP2/PKD1 fication by searching sequence databases using mass spectrometry data. Electrophor- complex. Proc Natl Acad Sci USA 106:11558–11563. esis 20:3551–3567. 10. Ćelić A, Petri ET, Demeler B, Ehrlich BE, Boggon TJ (2008) Domain mapping of the poly- 33. Sreerama N, Woody RW (2000) Estimation of protein secondary structure from circular cystin-2 C-terminal tail using de novo molecular modeling and biophysical analysis. dichroism spectra: Comparison of CONTIN, SELCON, and CDSSTR methods with an J Biol Chem 283:28305–28312. expanded reference set. Anal Biochem 287:252–260. 11. Koulen P, et al. (2002) Polycystin-2 is an intracellular calcium release channel. Nat Cell 34. Whitmore L, Wallace BA (2008) Protein secondary structure analyses from circular Biol 4:191–197. dichroism spectroscopy: Methods and reference databases. Biopolymers 89:392–400. 12. Svergun DI, Semenyuk AV, Feigin LA (1988) Small-angle-scattering-data treatment by 35. Kellermann G, et al. (1997) The small-angle X-ray scattering beamline of the Brazilian the regularization method. Acta Crystallogr A 44:244–250. synchrotron light laboratory. J Appl Crystallogr 30:880–883. 13. Svergun DI (1992) Determination of the regularization parameter in indirect- 36. Hammersley AP, Svensson SO, Hanfland M, Fitch AN, Hausermann D (1996) Two- transform methods using perceptual criteria. J Appl Crystallogr 25:495–503. dimensional detector software: From real detector to idealised image or two- 14. Svergun DI (1999) Restoring low resolution structure of biological macromolecules theta scan. High Pressure Res 14:235–248. – from solution scattering using simulated annealing. Biophys J 76:2879 2886. 37. Fischer H, Oliveira Neto M de, Napolitano HB, Polikarpov I, Craievich AF (2010) 15. Volkov VV, Svergun DI (2003) Uniqueness of ab initio shape determination in small- Determination of the molecular weight of proteins in solution from a single small- – angle scattering. J Appl Crystallogr 36:860 864. angle X-ray scattering measurement on a relative scale. J Appl Crystallogr 43:101–109. 16. Svergun DI, Petoukhov MV, Koch MH (2001) Determination of domain structure of 38. Zhang Y (2008) I-TASSER server for protein 3D structure prediction. BMC Bioinfor- – proteins from X-ray solution scattering. Biophys J 80:2946 2953. matics 9:40–48. – 17. Witzgall R (2007) TRPP2 channel regulation. Handb Exp Pharmacol 179:363 375. 39. Noel JK, Whitford PC, Sanbonmatsu KY, Onuchic JN (2010) SMOG@ctbp: Simplified 18. Fischer H, Polikarpov I, Craievich AF (2004) Average protein density is a molecular- deploymnet of structure-based models in GROMACS. Nucleic Acids Res 38: – weight-dependent function. Protein Sci 13:2825 2828. W657–W661. 19. Hoenderop JGJ, et al. (2003) Homo- and heterotetrameric architecture of the epithe- 40. Oliveira LC, Schug A, Onuchic JN (2008) Geometrical features of the protein folding 2þ lial Ca channels TRPV5 and TRPV6. EMBO J 22:776–785. mechanism are a robust property of the energy landscape: a detailed investigation of 20. Barrera NP, et al. (2007) AFM imaging reveals the tetrameric structure of the TRPC1 several reduced models. J Phys Chem B 112:6131–6136. channel. Biochem Biophys Res Commun 358:1086–1090. 41. Whitford PC, et al. (2009) An all-atom structure-based potential for proteins: Bridging 21. Feng S, et al. (2008) Identification and functional characterization of an N-terminal minimal models with all-atom empirical forcefields. Proteins 75:430–441. oligomerization domain for polycystin-2. J Biol Chem 283:28471–28479. 42. Clementi C, Nymeyer H, Onuchic JN (2000) Topological and energetic factors: what 22. Molland KL, Narayanan A, Burgner JW, Yernool DA (2010) Identification of the struc- determines the structural details of the transition state ensemble and “en-route” tural motif responsible for trimeric assembly of the C-terminal regulatory domains of intermediates for protein folding? An investigation for small globular proteins. polycystin channels PKD2L1 and PKD2. Biochem J 429:171–83. J Mol Biol 298:937–953. 23. Kim SJ, Dumont C, Gruebele M (2008) Simulation-based fitting of protein–protein 43. Lammert H, Schug A, Onuchic JN (2009) Robustness and generalization of structure- interaction potentials to SAXS experiments. Biophys J 94:4924–4931. based models for protein folding and function. Proteins 77:881–891.

9838 ∣ www.pnas.org/cgi/doi/10.1073/pnas.1106766108 Ferreira et al. Downloaded by guest on September 25, 2021