Intermodule cooperativity in the structure and dynamics of consecutive complement control modules in human C1r Structural biology Andra ´ sLa ´ ng 1, *, Katalin Szila ´ gyi 2, *, Bala ´ zs Major 2 ,Pe ´ ter Ga ´ l 2 ,Pe ´ ter Za ´ vodszky 2 and Andra ´ s Perczel 1,3 1 Laboratory of Structural Chemistry and Biology, Institute of Chemistry, Eo ¨ tvo ¨ s Lora ´ nd University, Pa ´ zma ´ ny Pe ´ ter se ´ ta ´ ny 1 ⁄ A, Budapest, Hungary 2 Institute of Enzymology, Biological Research Center, Hungarian Academy of Sciences, Budapest, Hungary 3 Protein Modeling Group HAS-ELTE, Institute of Chemistry, Eo ¨ tvo ¨ s Lora ´ nd University, Budapest, Hungary Introduction Complement is an effective serine protease cascade sys- tem in the blood, whose purpose is to opsonize and clear infectious particles and antigens [1]. This is achieved by three parallel pathways, which converge in common final steps to eliminate pathogens. The first pathway is termed the ‘classical’ pathway, although it is not the oldest one in evolutionary terms [2,3]. The classical pathway recognizes immunocomplexes con- Keywords cooperativity; dynamics; flexibility; modularity; NMR-spectroscopy Correspondence A. Perczel, Institute of Chemistry, Eo ¨ tvo ¨ s Lora ´ nd University, H-1518, 112, PO Box 32, Budapest, Hungary Fax: (36 1) 372 2620 Tel: (36 1) 372 2600 E-mail: perczel@chem.elte.hu Website: http://www.szerves.chem.elte.hu *These authors contributed equally to this work (Received 26 May 2010, revised 22 June 2010, accepted 27 July 2010) doi:10.1111/j.1742-4658.2010.07790.x The modular C1r protein is the first protease activated in the classical complement pathway, a key component of innate immunity. Activation of the heteropentameric C1 complex, possibly accompanied by major inter- subunit re-arrangements besides proteolytic cleavage, requires targeted regulation of flexibility within the context of the intramolecular and inter- molecular interaction networks of the complex. In this study, we prepared the two complement control protein (CCP) modules, CCP1 and CCP2, of C1r in their free form, as well as their tandem-linked construct, CCP1CCP2, to characterize their solution structure, conformational dynamics and cooperativity. The structures derived from NMR signal dispersion and secondary chemical shifts were in good agreement with those obtained by X-ray crystallography. However, successful heterologus expression of both the single CCP1 module and the CCP1CCP2 constructs required the attachment of the preceding N-terminal module, CUB2, which could then be removed to obtain the properly folded proteins. Internal mobility of the modules, especially that of CCP1, exhibited considerable changes accompanied by interfacial chemical shift alterations upon the attachment of the C-terminal CCP2 domain. Our NMR data suggest that in terms of folding, stability and dynamics, CCP1 is heavily dependent on the presence of its neighboring modules in intact C1r. Therefore, CCP1 could be a focal interaction point, capable of transmitting information towards its neighboring modules. Abbreviations CCP, complement control protein; CCP1, first complement control protein module of C1r; CCP2, second complement control protein module of C1r; CCP1CCP2, tandem of the two CCPs from C1r; CCP1 single, single CCP1 module; CCP2 single, single CCP2 module; CUB2CCP1CCP2, a trimodular fragment of C1r; CCP1 CCP2, CCP1 module in tandem CCP1CCP2; CCP1 CCP2, CCP2 module in tandem CCP1CCP2; CUB2 CCP1, CUB2 module in tandem CUB2CCP1; DLS, dynamic light scattering; MASP, mannose-binding lectin-associated serine protease; R 1, longitudinal relaxation; R 1, transverse relaxation; RSDM, reduced spectral density mapping; s e, effective correlation time. 3986 FEBS Journal 277 (2010) 3986–3998 ª 2010 The Authors Journal compilation ª 2010 FEBS taining antibodies of IgG or IgM isotypes. The recog- nition is achieved by C1q, a disulfide-linked subunit of the C1 heteropentameric protein, and leads to the autoactivation of the C1r subunit, a modular serine protease. In an extraordinary example of ‘action at a distance’, an activation signal is then mechanically transmitted to two molecules of C1r that are located within the six collagen-like stems of C1q. The exact details of this process are still largely unresolved. The active C1r then activates the third subunit of C1, a homologus serine protease called C1s. Both C1r and C1s are present in duplicate copies within the C1 com- plex, forming a tetramer. Both the autoactivation of C1r and the activation of C1s require large-scale move- ments of the different domains of the proteases. Inter- modular and intramodular flexibilities are thought to be essential in these movements. Expression of recom- binant intact C1r [4] and its domain combinations in insect cells and bacteria, has provided an opportunity to study the mechanism of activation at an intramolec- ular level. The intact C1r is an 80-kDa trypsin-like modular serine protease of 705 residues [5–7]. C1r contains six modules of four different types. A post-translationally hydroxylated Ca-ion-binding epidermal growth factor domain [8–10], sandwiched between two CUB (C1s ⁄ C1r, urchin epidermal growth factor, bone mor- phogenetic protein) modules [11], contributes to the formation of the complex with C1s and C1q in a Ca 2+ -dependent manner [12]. The C-terminal proteo- lytic fragment of C1r is termed c B or catalytic frag- ment, and consists of tandem complement control protein (CCP) and serine protease domains. This frag- ment is responsible for all reported catalytic activity and forms a homodimer in neutral solutions [13]. The CCP2 module increases the thermal stability of the ser- ine protease and enhances its C1s-cleaving activity. CCP modules are common in the complement system [14,15] and can also be found in proteins such as c-ami- nobutyric acid type B receptor subunit 1a [16] and interleukin-2 receptor-a [17]. The CCP of the interleu- kin-2 receptor-a carries another, but atypical, CCP as an insertion at its ‘hypervariable loop’ leading to two strand-swapped CCP-like domains. Typical CCP mod- ules [18] have four highly conserved cysteine residues with an abab disulfide-pairing pattern, as well as a con- served tryptophan and multiple invariable proline, gly- cine and aromatic residues. The fold comprises two antiparallel b-pleated sheets, which are formed by strands B, D, F, G, E and H forming a b-sandwich. The two disulfides are located close to the termini of the module, which define the two apices of the some- what ellipsoidal protein. CCP structures are highly variable, with the invariant parts comprising mainly the b-strands. The loops between strands B and C, known as the ‘hypervariable loop‘, D and E, E and F, F and G, and G and H are particularly tolerant of longer insertions [19]. In multimodular proteins, however, insertions in the F–G and D–E loops might affect inter- modular interactions between consecutive CCP mod- ules because these loops project towards the previous (F–G) or the subsequent (D–E) CCP module. In gen- eral, CCP modules mediate protein–protein [20–22] and ⁄ or protein–carbohydrate [23–25] interactions. Currently available backbone dynamics data, deter- mined by NMR spectroscopy, show a diverse picture of CCP mobility [26–34]. In general, segments with increased mobility are candidates for interaction sites [29]. In the present study, NMR spectroscopy was applied to determine the backbone dynamics of the two CCP modules from human C1r to locate the source of flexibility needed for structural re-arrangement upon autoactivation and to identify possible interaction sites. Results Expression, folding and stability of single and tandem CCP modules To distinguish between the different constructs investi- gated in this study, we used the following notations: first complement control protein module of C1r (CCP1 single ) and second complement control protein module of C1r (CCP2 single ) for the single CCP mod- ules, and CCP1 CCP2 and CCP1 CCP2 for the corre- sponding modules within the covalently linked tandem CCP1CCP2 ( # 1) module pair. The products were expressed as insoluble inclusion bodies. Only a small amount of CCP1 single could be obtained, whereas CCP2 single was successfully produced in the required quantity. However, fusion constructs containing the N-terminal preceding CUB2 module, CUB2CCP1, and a trimodular fragment of C1r (CUB2CCP1CCP2) yielded sufficient amounts of prop- erly folded proteins. The wild-type human protein does not contain any post-translational modifications (e.g. glycosylation) and thus it was expected that the pro- karyotic host fulfills the basic requirements for produc- tion of these constructs. In terms of folding, as judged by differential scan- ning calorimetry and CD spectroscopy, CCP2 single was properly folded when expressed as a single module, whereas folded CCP1 single and CCP1CCP2 ( # 2) could only be obtained by thermolysin digestion of the CUB2CCP1 and CUB2CCP1CCP2 constructs, respec- tively. Furthermore, CD-spectroscopic and differential A. La ´ ng et al. Flexibility and cooperativity of CCP modules FEBS Journal 277 (2010) 3986–3998 ª 2010 The Authors Journal compilation ª 2010 FEBS 3987 scanning calorimetric measurements also revealed that the thermal denaturation of both CCP2 single at 54.3 °C and CCP1CCP2 ( # 2) was reversible (data not shown), in contrast to that of CCP1 single , which is irreversible with a melting temperature of 63.4 °C [35]. Both folded modules are stable under ambient conditions. To investigate possible multimerization ⁄ aggregation of the samples caused by their high concentration (about two to three times higher than that found in serum from human blood), gel filtration and dynamic light scattering (DLS) studies were performed on the pair of modules. DLS showed the approximate molec- ular mass of CCP1CCP2 (17259.6 Da) in the tempera- ture range from 300 to 315K (Fig. 1). In conjunction with gel filtration, this clearly excludes any significant aggregation at the temperature range of our NMR measurements. Analysis of single modules Backbone assignment and secondary chemical shifts: the secondary structure of CCPs 1 H- 15 N correlation spectra of both CCP single modules exhibited excellent signal dispersion in both dimensions with minimal overlaps, indicative of well-folded globu- lar structures at 300 K, pH 4.0–4.5 and pH 7.0. Spin-system identification and 1 H, a H, 15 N backbone assignment of each single module was obtained by 3D N H- 15 N-TOCSY-HSQC and 3D N H- 15 N-NOESY- HSQC. Assignment of CCP1 single N H- 15 N cross-peaks was hampered by its large proportion of glutamine residues (nine out of 73,  12.3%) but almost com- plete assignment was achieved (except for T286–I289). The large 1 H downfield shift (10.46 ppm) of the weak L334 cross-resonance suggests strong H-bonding. All backbone resonances of CCP2 single were assigned at both pH values, except for the four N-terminal resi- dues (A353, S354, M355, I356) and G377. Addition- ally, R399–G401 and E404 resonances were missing only at pH 7.0. Neutralizing the solutions resulted in a few back- bone shifts, as monitored by 2D HSQC spectra. Imid- azole groups of histidine residues (pK a  6.0) are expected to be deprotonated at neutral pH. Significant changes in backbone amide shifts were indeed observed for H335, H348 (CCP1 single ) and H390 (CCP2 single ), and for some nearby residues (Fig. S1). H335 displays a significant downfield shift in both dimensions ( N H and 15 N), whereas the cross-peaks of H348 and H390 migrate mainly along the 15 N dimen- sion upon the elevation of pH. The preceding L334 was not clearly visible at acidic pH (however, it gave an NOE cross-peak to the H335 amide), whereas it could be clearly identified at neutral pH. Furthermore, N H of M351, R399, A400, G401 and E404 are only observable at pH 4.5. Comparison of secondary a H chemical shifts at acidic and neutral conditions revealed no major differ- ences, and shifts under both conditions were in good agreement with the secondary structure observed in the crystal structures of C1r CCPs [36,37] and other CCP modules (Figs S2–S9). Therefore, we conclude that a change in pH within our examined range did not sig- nificantly disturb modular integrity, although local effects were shown. Both single module structures are therefore essentially similar to those observed in the crystal structure of multimodular constructs. We note that, in general, conserved residues do not necessarily exhibit similar chemical shifts caused by differences in their microenvironment [38]. Relaxation data of single modules To characterize the inherent flexibility of the single modules, longitudinal (R 1 ) and transverse ( R 2 ) relaxa- tion, as well as { 1 H}- 15 N NOE measurements, were made at 11.7 T. These relaxation data were acquired and analyzed for both CCP single modules at pH 4.0–4.5 and pH 7.0 at 300K (Table S1). The R 2 ⁄ R 1 value, which satisfactorily correlates with the overall re-orientation time (s c ) of well-folded pro- teins, was calculated for CCP single modules at both pH values. Because of more complete assignment, we Fig. 1. Good correspondence between the calculated (dotted horizontal line) and measured ( # 2, unfilled triangle) molecular mass of thermolysin-treated CCP1CCP2 indicates monomer forms in the temperature range of 300–315 K. In the same temperature range, the first construct ( # 1, filled circle) is not a monomer, as shown by DLS data. Flexibility and cooperativity of CCP modules A. La ´ ng et al. 3988 FEBS Journal 277 (2010) 3986–3998 ª 2010 The Authors Journal compilation ª 2010 FEBS focused on relaxation data obtained at the lower pH (data acquired at pH7.0 are listed in Tables S1 and S2). Although larger (9358.6 Da), the lower R 2 ⁄ R 1 value of CCP2 single (2.681 ± 0.356) suggests a faster overall re-orientation than that for CCP1 single (8564.7 Da and 2.957 ± 0.359) (Table S1). The E–F loop of CCP2 single could not be fully traced in the available X-ray structures [Protein Data Bank (PDB) entries 1gpz, 2qy0], and the corresponding residues (R399–Q407) exhibited an R 2 ⁄ R 1 ratio (2.276 ± 0.352) below the average, with a low { 1 H}- 15 N NOE (0.301) also. Corresponding cross-resonances were absent at the higher pH value. All these data indicate that the E–F loop is mobile on the ps to ns timescale. High R 2 ⁄ R 1 values, indicating slow timescale motion (lsto ms), were observed for residues K419–K423 and E425, giving a value of 3.134 ± 0.435 for the G–H loop average (K419–I427). For CCP1, a high R 2 ⁄ R 1 value was observed for C341, D344, R349 and A350 (R 2 ⁄ R 1 > 3.4) and indi- cated ls to ms timescale NH vector re-orientation. Except for the N-terminus, low R 2 ⁄ R 1 values (< 2.5) for residues F301, T302, H335 and S336 showed effec- tive relaxation mechanisms on the ps to ns timescale. Model-free analysis of single modules Model-free analysis of CCP1 single and CCP2 single at both pH values was applied in order to obtain a detailed pic- ture of residual motions of all the amide NHs within the molecules. Local NH flexibility is typically given in terms of the square of the generalized order parameters, S 2 , as the rate of physical restriction of NH motion, and of effective correlation time, s e , on the ps to ns range. Extremely fast local motion is indicated by dissecting S 2 to provide its proportion, S f 2 . We applied isotropic motional description because of its simplicity and robustness. Here, we focus on the backbone dynamics at pH 4.0–4.5 (Tables S3 and S4) whereas the results at pH 7.0 are given in (Tables S5 and S6). Both CCP single modules at both pH values exhibited high S 2 values (> 0.8) in general, which are character- istic of well-folded globular proteins. In CCP1 single , small S 2 values of NH (S 2 < 0.8), representing larger amplitude motions, were found at the N-terminus (up to L307) extending to the ‘HVL’ (except D299 and I303) at pH 4.0. Increased local backbone flexibility was also detected at loops E–F (N331–S336) and G–H (R349–A350) and at the C-terminus (Fig. 2 and Table S3), as well as in the ‘hypervariable’ B–C loop and the D–E turn. Dissection of ps to ns timescale motions by introducing the S f 2 parameter was necessary for resi- dues F301, T302 and H335, which showed very rapid internal motion. By contrast, significant contributions from the slower ls to ms scale backbone motions (R ex ) were observed for residues A350, R349, C341, E300 and N331. For CCP2 single , residues with low S 2 values (< 0.8) were located at the N-terminus (up to L365 except for C359) and C-terminus, in the D–E turn around P392 and in the highly mobile E–F loop (T398–Q407) at pH 4.5 (Fig. 2 and Table S4). Crystallographic B-factors determined for this loop [36,37] were in accordance with our NMR results. Furthermore, the observed backbone proton ( a H and N H) chemical shifts of most of these residues were close to random-coil values con- sistent with increased mobility (Fig. S2). Residues exhibiting more restricted motion are located not only in b -strands (from B to H) but also in the ‘hypervari- able’ B–C loop. Conformational exchange (R ex ) was detected for residues E425, E421, G408, K423, T398 and K419. Distinct rapid local motion was indicated by the significant s e of G424 and G401, for which resi- dues the inclusion of an S f 2 term was also found to be necessary. In general, similar mobility patterns were observed at neutral pH for both modules (Tables S5 and S6). However, the A–B turn in CCP1 single loses much of its local mobility (S 2 > 0.8), which might be the result of D299 side-chain deprotonation and carboxylate inter- action with the amide NH of C354, as observed in both crystal structures [36,37]. This atypical b-turn is therefore not only sequentially, but also dynamically, unique in the known CCP folds. Analysis of the covalently linked CCP1CCP2 module pair Spectral properties of CCP1CCP2: chemical shift perturbation analysis as a result of CCP1 and CCP2 Two different CCP1CCP2 constructs were investigated. The first ( # 1) contained a non-native stretch of three residues at the N-terminus originating from the con- struct. The second ( # 2) was the thermolysin-digested CUB2CCP1CCP2. At 280 K, 300 K and pH 4.0, or at 300 K and pH 7.0, HSQC spectra of both CCP1CCP2 constructs (Figs S10 and S11) showed highly overlap- ping peaks in the middle of the amide region (close to random coil values) and also gave resonances previ- ously unobserved in single CCPs. Additionally, many peaks were weak. These observations suggest that the proteins were not properly folded under the conditions applied. However, at a slightly elevated temperature (320 K, pH 4.0 and 315 K, pH 7.0), a better-quality HSQC spectrum with more resolved peaks could be obtained for the second construct ( # 2) initially A. La ´ ng et al. Flexibility and cooperativity of CCP modules FEBS Journal 277 (2010) 3986–3998 ª 2010 The Authors Journal compilation ª 2010 FEBS 3989 containing the CUB2 module and obtained by therm- olysin digestion (Figs S10 and S11). All data reported below refer to this construct. Backbone NH cross-resonances of residues in the single and tandem CCP modules were compared at 300 K and pH 7.0. At neutral pH, fewer resonance overlaps were observed than under acidic conditions (Fig. S12). Chemical shift changes in the tandem CCP1CCP2 construct relative to the free modules are shown in Fig. 3. In general, the chemical shift changes were small with the exception of the vicinity of two aromatic residues, namely Y325 in CCP1 (D–E turn) and Y381 of CCP2 (C–D turn), which were located at the interface between the two modules. Remarkable changes were also observed in the intermodular linker and at the F–G turn in CCP2. This suggests that both modules maintain their modular integrity; nevertheless, there is a well-defined interface region between the two modules that is formed primarily by the two aromatic residues. CCP1CCP2 flexibility (relaxation data and reduced spectral density values) Relaxation parameters for CCP1CCP2 were obtained at 315 K and pH 7.0 (Fig. 4). In general, CCP1 CCP2 A B Fig. 2. General order parameters for CCP single modules (A, CCP1; and B, CCP2) at acidic (filled circle) and neutral (unfilled triangle) pH and 300 K. The overall rotation correlation times are as indicated (acidic ⁄ neutral, respectively) completed with the value calculated by HydroPro. The positions of b-strands are indicated with black boxes at the bottom of each panel. Flexibility and cooperativity of CCP modules A. La ´ ng et al. 3990 FEBS Journal 277 (2010) 3986–3998 ª 2010 The Authors Journal compilation ª 2010 FEBS has smaller R 1 values than CCP1 CCP2 , whereas the opposite is found for R 2 values. Consequently, the average R 2 ⁄ R 1 ratio is higher in CCP1 CCP2 than in CCP1 CCP2 , which is exactly the reverse of the situation observed for the CCP1 single and CCP2 single modules and probably reflects the complex interdependence of the modules in terms of internal dynamics and may also be a consequence of the large anisotropy (devia- tion from the ideal spherical shape) of the tandem module pair relative to the single modules. This Fig. 3. Interaction of tandem CCP modules is restricted to its interface (pH 7.0 and 300 K). Changes of residues at the interface region in both modules are indicated with color-coded arrows. Interaction of CCP modules is mapped on both faces of the surface representation of the crystal structure (2qy0). b-strands are indicated with black boxes. Color-coding: yellow > 0.15; orange > 0.30; red > 0.50. Combined chemical shifts were obtained as described previously [39]. The linker colored light brown is slightly ambiguous. AB CD CCP single CCP single CCP single CCP single Fig. 4. Relaxation data of the CCP modules. (A) R 1 ( 543.0 ± 56.2 ms), (B) R 2 ( 75.6 ± 12.3 ms), (C) { 1 H}- 15 N NOE (0.608 ± 0.163) and (D) R 2 ⁄ R 1 of CCP1CCP2 at pH7.0, 315 K (filled circle) and of CCP single at pH7.0, 300 K (hollow triangle). The positions of the b-strands are indicated with black boxes at the bottom of each panel. Horizontal solid and dotted lines indicate mean and 1 SD values for single and tandem constructs. A. La ´ ng et al. Flexibility and cooperativity of CCP modules FEBS Journal 277 (2010) 3986–3998 ª 2010 The Authors Journal compilation ª 2010 FEBS 3991 discrepancy is also apparent in the absolute values of the calculated rotational diffusion correlation times, as the s c of CCP1 single (s c ‡ 5.0 ns) at both pH values is larger than that of CCP2 single (s c  4.7 ns), which is otherwise a larger molecule, and hydrodynamic calcu- lations yield values of 5.2 and 5.6 ns, respectively. R 2 values (13.32 ± 2.14) are consistent with the increase in molecular size relative to the single modules (CCP1 single : 7.47 ± 1.35 and CCP2 single : 7.16 ± 1.02). { 1 H}- 15 N NOE and R 1 values, primarily indicative of motions on the ps-ns timescale, show major changes mainly in the CCP1 module. Low { 1 H}- 15 N NOE val- ues indicate remarkably rapid (ps to ns) local mobility in residues of the B–C loop and, to a lesser extent, in the E–F and G–H loops; the R 1 values indicate mobil- ity changes in the E–F loop. In general, the { 1 H}- 15 N NOE values show a much more diverse distribution in CCP1 CCP2 than in CCP1 single , possibly corresponding to an overall gain of fast timescale flexibility. By con- trast, the large flexibility of the E–F loop in CCP2 seems to be retained in the tandem construct. R 2 val- ues, bearing information on ls to ms motions, are higher in both modules of the tandem construct than in the free modules, with increases apparent at the interfaces A–B (E300), D–E (Y325) and C–D (Y381), near the F–G turns of CCP2 (T411, C412, I417 and W418) and near the linker region (G360). For the tandem module, the model-free approach did not yield satisfactory results [e.g. the rotational diffusion correlation time obtained from NMR data (9.318 ns for CCP1 and 9.872 ns for CCP2) deviated remarkably from that estimated by hydrodynamic calculations (12.640 ns) based on the crystal structure] [37]. Therefore, we turned to reduced spectral density analysis (RSDM) [40,41]. RSDM evaluates three such values sensitive for slow or overall (at x=0), interme- diate (x N ) and rapid (0.87x H ) local motions. Such analysis of CCP1CCP2 clearly shows the shift of J(0) to larger values compared with CCP single , in accor- dance with the slower re-orientation (i.e. higher s c )of the molecule (Fig. 5). The smaller shift observed for the values of CCP1 is in agreement with the smaller difference in the R 2 ⁄ R 1 ratio of the corresponding resi- dues after attachment of CCP2. The larger dispersion along both dimensions in CCP1CCP2 probably indi- cates the increased anisotropy of the molecule. In CCP1, the outliers along J(0) are A350 [the largest J(0); filled circles in Fig. 5 ] in CCP1 single and S336 [the smallest J(0); open triangles in Fig. 5 ] in CCP1 CCP2 , indicating the slow and rapid N H re-orientations of these residues, respectively. Discussion We have successfully expressed and purified the two CCP modules of the human complement protein C1r, both individually and as a fused construct. Whereas folded CCP2 was easily produced, properly folded CCP1 could only be obtained using a thermolysin- cleaved, folded CUB2CCP1 construct. Similarly, enzy- matic cleavage of CUB2CCP1CCP2 resulted in CCP1CCP2, which was folded and stable, as judged by NMR and dynamic light-scattering measurements. NMR signal dispersion and secondary chemical shifts showed that the obtained proteins were properly folded and their structures were consistent with the general CCP fold. According to DLS measurements, undesired aggregation or oligomerization did not occur AB Fig. 5. Intermediate versus slow timescale-sensitive reduced spectral-density values correlated for residues from CCP1 (A) and from CCP2 (B) both in single at 300 K (hollow triangle) and in tandem at 315 K (filled circle) constructs at pH7.0. The solid line represents reduced spectral-density values reduced to single motion in a fully isotropic case. For simplicity, residues from the linker (355KIKD) are not shown. Flexibility and cooperativity of CCP modules A. La ´ ng et al. 3992 FEBS Journal 277 (2010) 3986–3998 ª 2010 The Authors Journal compilation ª 2010 FEBS under the conditions applied and therefore the constructs could be reliably used to assess the interde- pendence of the modules. This was corroborated by the calculated s c values of the tandem construct based on R 2 ⁄ R 1 data, which do not indicate any increase in molecular size above that expected based on hydrody- namic calculations. Internal mobility data are also in good agreement with previous structural studies; this was most prominently exemplified by the S 2 parame- ters of the E–F loops of the modules where the increased mobility observed for CCP2 was also reflected by the corresponding poor electron density in the crystal structure [36,37]. These observations were valid at both at pH 4.0–4.5 and pH 7.0 for the single modules, indicating no major conformational transi- tion upon pH change. Chemical shift changes detected upon the transition from acidic to neutral conditions were primarily located in the sequential or structural vicinity of histi- dines and thus may reflect minor conformational changes induced by the protonation ⁄ deprotonation of the imidazole side-chains [42]. This might affect ionic contacts of histidines with aspartate and ⁄ or glutamate residues. In the light of atomic structures of CCP1 and CCP2 (2qy0 [36] and 1gpz [37]), this perturbation might indicate that in CCP1, the side chain of H335 is fully exposed, whereas that of H348 is close to the backbone N H of V340. The protonation change of the only His in CCP2, namely H390, affects residues at strand B (368GDF sequence) and therefore might indi- cate an ionic interaction with D369. Inspection of four complement molecules from the ‘classical’ and ‘lectin’ pathways, each containing two CCPs, indicates that the homologus interaction seems important to dock the b-strand B to strand D (Table S7). Such ionic con- tacts, like in C1r CCP2, may contribute to fold stabil- ity in mannose-binding lectin-associated serine protease (MASP)1 ⁄ 3 CCP1, MASP2 CCP1 and CCP2, and C1s CCP2, whereas in C1r CCP1 and C1s CCP1, polar and hydrophobic interactions are likely to be dominant, respectively. The presence of the CCP1 module, preceding the catalytic fragment, is required to form the c B dimer (made up of two CCP1CCP2SP molecules) at neutral pH, but dimerization does not occur at pH values lower than 5.5 [13]. Because the largest changes upon pH alterations were detected at and near histidines (H335 and H348), these residues are strong candidates for inducing minor conformational changes and ⁄ or forming pH-dependent interaction sites on CCP1. Surface turns and loops with increased mobility can provide easily variable protein–protein interaction sites in proteins. In CCP1, the A–B turn has a unique sequence among CCPs, as F301 occupies a position of a Gly that is conserved in other modules. Side-chain atoms of F301 form aromatic stacking interactions with those of Y325 and act in synergy with the D299 (carbox- ylate Od2)–C354 (amide H) hydrogen bond to promote an unconventional geometry of the A–B turn with increased flexibility when the D299 side-chain carboxyl- ate is protonated (at pH 4.0). In CCP1 CCP2 at pH 7.0, the ‘hypervariable’ B–C loop can be characterized by rapid local motion, a feature that may be linked to its interaction capacity with other molecules, as it is clearly seen in the crystal structures [36,37]. The E–F loop has different dynamics in CCP1 and CCP2, dominated by a slow timescale (ls to ms) conformational exchange in the former, while showing increased mobility (S 2 values as low as 0.5) on the ps to ns timescale in the latter. The G–H loop can be characterized by significant motions on the ls to ms timescale in both of the free modules. Thus, these loops are strong candidates for binding sites of other complement and ⁄ or regulatory proteins. The large insertion between E and F strands in C1r CCP2 is atypical in CCP modules; in particular, it is absent from the three closest human CCP2 homologs (MASP-1 ⁄ 3, MASP-2 and C1s). Modules are often defined as domains (i.e. autono- mous folding and functional units) that occur in diverse proteins. Thus, constructs containing domains and their combinations that were different from those of the modular C1r protein were expected to be easily obtainable in folded and functional forms. However, both the single CCP1 and the tandem CCP1CCP2 construct required an alternative strategy for efficient production: only variants expressed in fusion with the CUB2 module at their N-terminus were expressed and folded properly and could then be cleaved to yield the desired modules. This suggests that the CUB2 domain might have a chaperoning role in the folding of the CCP1. By contrast, interdependence of the two CCP modules in the tandem construct is apparent, such as the observed spectral properties of CCP1CCP2 at 300 and 320 K, as well as changes both in chemical shifts and mobility parameters compared with those of the free domains. Residues in the B–C, E–F and G–H loops of CCP1 display deviations between CCP1 single and CCP1 CCP2 in both of their { 1 H}- 15 N NOE and R 2 values (these R 2 changes are far greater than the gen- eral increase of R 2 values according to the molecular mass change). This is also consistent with the other- wise elusive observation that the rotational correlation times of the free modules are in reverse order com- pared with those in the CCP1CCP2 tandem construct. Comparison of C1r CCP data with previous CCP module pair structures and mobility data shows that A. La ´ ng et al. Flexibility and cooperativity of CCP modules FEBS Journal 277 (2010) 3986–3998 ª 2010 The Authors Journal compilation ª 2010 FEBS 3993 the C1r inter-CCP flexibility is most likely similar to that described for the CCP3CCP4 of viral complement proteins [26] as a few hydrophobic interface residues are well conserved. Although there are also important differences, namely a shorter ‘HVL‘ loop, and the presence of a tyrosine in strand C of CCP4 in place of an asparagine in C1r CCP2 (N379), and based on the results of the present perturbation study, we suggest that the inter-CCP flexibility in C1r is restricted by two nearby aromatic residues (Y325 and Y381). In summary, our results suggest that folding, flexibility and the implicated partner-binding ability of C1r CCP1 are all affected by neighboring modules in the intact C1r molecule. These results imply that CCP1 might act as a central effector site towards partner molecules as it is capable of sensing alterations in neighboring modules caused by various effects. In particular, the B–C, E–F and G–H loops in CCP1, which are affected by the pres- ence of CCP2, are candidates for such a function. Internal mobility is a key factor in a multitude of biomolecular interactions [43], and complement prote- ases are surely no exception. Our observation that the internal dynamics of a module can be modulated by neighboring domains offers a new way to understand the regulation of multidomain proteins and challenges the generally accepted view that domains are indepen- dent functional and folding units. Therefore, the reduc- tionist approach for modular proteins (i.e. dissection to smaller building blocks), their analysis and subse- quent extrapolation to the full-length protein is not always feasible for understanding the biological func- tion (e.g. [44,45]). Nevertheless, it is clear that consecu- tive CCP modules do exemplify a wide range of cases, from independence to tight intermodular contacts [46–48], manifesting an excellent opportunity for evolu- tion to fine-tune macromolecular behavior and interac- tions. Our results are in agreement with the previous observations of Major et al. [35], showing that the CUB2–CCP1 fragment plays an important regulatory role in the autoactivation of C1r. The structure of the CUB2 domain changes considerably upon Ca 2+ bind- ing, and this effect is likely to be transmitted by the CCP1 module towards the catalytic region of C1r. Materials and methods Construction of recombinant plasmids for expression of the CCP modules of C1r The cDNA fragments corresponding to the CCP2 (I356– V433), CCP1CCP2 (I289–V433), CUB2CCP1 (Q173–D358) and CUB2CCP1CCP2 (Q173–V433) modules of C1r were amplified by PCR using the full-length cDNA template. For the amplification procedure the following forward and reverse primers were used, respectively: CCP2: CGCGCTAGCATGATCAAGGACTGTGGG CAGCCC and CGC GAATTCTCACACTGGCAAGC ACCGAGGAATCT; CCP1CCP2: CGC GCTAGCATGATCATCAAGTGCC CCCAGCCC and CGC GAATTCTCACACTGGCAA GCACCGAGGAATCT; CUB2CCP1: CGC GCTAGCATGACTCAGGCTGAG TGCAGCAGC and CGC GAATTCTCAGTCCTTGA TCTTGCATCTGGG; CUB2CCP1CCP2: CGC GCTAGCATGACTCAGGCT GAGTGCAGCAGC and CGCGAATTCTCACACT GGCAAGCACCGAGGAATCT. The PCR products were digested with NheI and EcoRI (cleavage sites underlined) and ligated into the pET-17b expression vector (Novagen, Darmstadt, Germany). As a result, the recombinant proteins contain an extra tripeptide (A-S-M) at their N-terminus. The constructs were verified by DNA sequencing. Expression, isotope labeling, renaturation and purification of the recombinant proteins Expression, inclusion-body isolation and solubilization were performed as previously described [13]. For isotope label- ing, cultures were grown on M9 minimal medium supple- mented with thiamine, trace metals [49], ampicillin and chloramphenicol. Starter culture was grown for 4–5 h, and then the cells were collected and used to inoculate 1 L of minimal medium supplemented with 1 g of 15 NH 4 Cl (National Institute of Research and Development for Isoto- pic and Molecular Technologies, Cluj-Napoca, Romania) for 15 N-labeling and 2 g of 13 C-glucose (Cambridge Isotope Laboratories, Inc., Andover, MA, USA) for 13 C-labeling. Cells were grown in a BioStat B (Braun, Sartorius, Go ¨ ttingen, Germany) fermentor for 12 h. The solubilized proteins (20 mgÆmL )1 ) were diluted 400-fold into the refold- ing buffers [50 mm Tris ⁄ HCl (pH 8.3), 5 mm EDTA and 145 mm NaCl in the case of CCP2, or 2 m GuHCl in the case of CCP1CCP2] or 125-fold for CUB2CCP1 and CUB2CCP1CCP2 (pH 8.5, 750 mm Arg, 500 mm GuHCl and 5 mm CaCl 2 ). In each case, the refolding buffers con- tained 3 mm reduced glutathione and 1 mm oxidized gluta- thione. The renaturation process was conducted at 15 °C overnight (2 days at 10 °C for CUB2-containing con- structs). The solutions of the renatured proteins were dia- lyzed against 50 mm Tris ⁄ HCl (pH 8.3) containing 145 mm NaCl, and filtered through a glass filter, or, for CUB2-con- taining constructs, were dialyzed twice against 20 mm Tris ⁄ HCl (pH 8.0) containing 5 mm NaCl and 5 mm CaCl 2 , and filtered through a 0.22-lm membrane filter. Renatured proteins were purified on an SP Sepharose XL column Flexibility and cooperativity of CCP modules A. La ´ ng et al. 3994 FEBS Journal 277 (2010) 3986–3998 ª 2010 The Authors Journal compilation ª 2010 FEBS (Pharmacia Biotech, Uppsala, Sweden) equilibrated with 50 mm Tris ⁄ HCl (pH 8.3) containing 50 mm NaCl, and the elution was conducted with a linear, 50–1000 mm gradient of NaCl. The recombinant proteins were further purified using an SP Sepharose HP column (GE Healthcare, Little Chalfont, Buckinghamshire, UK), and the fractions were identified by SDS ⁄ PAGE. CUB2CCP1 and CUB2CCP1C CP2 renatured proteins were purified by Q Sepharose ion-exchange chromatography and gel filtration on a Sephacryl S100 column. Limited proteolysis of C1r CCP1 and CCP1CCP2 and determination of protein concentration Both CUB2-containing constructs were digested with thermolysin at 37 °C [50] at an enzyme ⁄ substrate molar ratio of 1 : 40. The reaction was stopped by the addition of 10 mm EDTA. The products were dialyzed against 50 mm Na-acetate, 10 mm NaCl, 5 mm EDTA (pH 4.0). C1r CCP1 and CCP1CCP2 were purified by chromatography on an SP Sepharose HP cation-exchange column in the same buffer and eluted by application of an increasing ionic-strength gradient. Both fragments were verified by MS. The concentration of recombinant proteins was deter- mined by measuring the absorbance at 280 nm using the absorption coefficients 1.1, 1.6 and 1.4 (1%, 1 cm) for the CCP1, CCP2 and CCP1CCP2 fragments, respectively. For calculating absorption coefficients, we used the method of Gill et al. [51], taking disulfide bridges into account. The relative molecular mass values calculated from the amino acid sequences were 8565, 9359 and 17 260 for CCP1 single , CCP2 single and CCP1CCP2, respectively. NMR spectroscopy All NMR experiments were acquired on a Bruker DRX500 NMR spectrometer using a protein solution of  1.5 mm at 300, 310, 315 and 320 K. Samples were dissolved, each at a final concentration of 10 mm, in Na-acetate ⁄ NaCl buffer containing 2 mm NaN 3 in an H 2 O:D 2 O ratio of 9 : 1, at pH values of 4.0, 4.5 and 7.0. Spectra typically contained 4K*64 data points in 2D experiments and 2K*256*64 data points in 3D experiments. Data processing and resonance frequency assignment were completed using NMRPipe [52], Sparky [53] and Xeasy [54]. Whereas backbone resonance assignment of the CCP single could be achieved using standard 3D 15 N-TOC- SY-HSQC and 15 N-NOESY-HSQC spectra [55], resonances in the CCP1CCP2 construct were assigned by triple-reso- nance experiments [HNCA and HN(CO)CA [56], as well as CBCACONH [57,58] and HNCACB experiments]. Based on heteronuclear 1 H- 15 N correlation experiments [59], chemical shift differences were calculated as reported previously [39]. For investigations of H N backbone dynamics, R 1 and R 2 relaxation measurements, as well as { 1 H}- 15 N NOE experi- ments, were performed [60]. Peak intensities were deter- mined using Sparky [53], and relaxation parameters were fitted with the Levenberg–Marquardt algorithm. The R 1 and R 2 delay times applied can be found in Table S8. Hydrodynamic calculations for estimating the rotational correlation times were performed using the program Hydro- Pro [61]. Identification of residues with slow and fast timescale motions was performed as described by Clore et al. [62]. Residues with a high R 2 ⁄ R 1 (mean + 1 SD) ratio are likely to exhibit R ex contribution, whereas low R 2 ⁄ R 1 (mean ) 1SD) is indicative of s e contribution. These resi- dues, along with those that produced unresolved (overlap- ping) peaks were generally excluded from initial s c calculations (listed in Tables S1–S6). Dynamics interpretation of relaxation values Backbone dynamics of CCPs were calculated from relaxa- tion parameters using both the model-free approach [63] and RSDM [41]. Model-free analysis was carried out with the program Tensor2 [63] using an isotropic diffusion model, whereas RSDM was performed with an in-house program using the equations described in Krizova et al. [41]. Gel filtration and DLS measurements Gel filtration and DLS experiments were performed in 10 mm Tris ⁄ 1mm EDTA (pH 7.0). For gel filtration, a Superose 12 column (Pharmacia, Stockholm, Sweden) was used at room temperature ( 300 K). The flow rate was 1mLÆmin )1 and UV absorbance at 280 nm was monitored for detection of the proteins. For DLS, a DynaPro Titan (Wyatt Technology Co., Santa Barbara, CA, USA) instru- ment was used with DP-TH-03 (laser power 60 mW; range used 10–15%) and a DT-TC-04 temperature-controlled microsampler. Acknowledgements The authors are grateful to Zolta ´ nGa ´ spa ´ ri and Luka ´ s ˇ Z ˇ ı ´ dek for their help in the interpretation of relaxation data and for their guidance when performing RSDM. We also thank A. K. Fu ¨ ze ´ ry for critical reading of the manuscript. This work was supported by grants from ICGEB (CRP ⁄ HUN08-03), the Hungarian Scientific Research Fund (OTKA K72973, NK-77978 and NI- 68466) and A ´ nyos Jedlik grant NKFP 07 1-MA- SPOK07 from the Hungarian National Office for Research and Technology. References 1 Zipfel PF & Skerka C (2009) Complement regulators and inhibitory proteins. Nat Rev Immunol 9, 729–740. A. La ´ ng et al. Flexibility and cooperativity of CCP modules FEBS Journal 277 (2010) 3986–3998 ª 2010 The Authors Journal compilation ª 2010 FEBS 3995 [...]... Identification of the C1q-binding Sites of Human C1r and C1s: a refined three-dimensional model of the C1 complex of complement J Biol Chem 284, 19340–19348 ´ ´ ´ 13 Kardos J, Gal P, Szilagyi L, Thielens NM, Szilagyi K, ´ ´ ´ Lorincz Z, Kulcsar P, Graf L, Arlaud GJ & Zavodszky ˜ P (2001) The role of the individual domains in the structure and function of the catalytic region of a modular serine protease, C1r J... shifts of residues from different CCP modules Fig S10 Set of HSQC spectra of CCP1CCP2 (#2) collected at 280, 300 and 320 K at pH 4.0 Fig S11 Set of HSQC spectra of CCP1CCP2 (#2) collected at 300 and 315 K at pH7.0 Fig S12 Overlay of HSQC spectra collected at 300 K of single CCP1 and CCP2 and of pair of modules at neutral and acidic pH Doc S1 Intermodule cooperativity in the structure and dynamics of consecutive. .. Zavodszky P & Gal P (2008) Revisiting the mechanism of the autoactivation of the complement protease C1r in the C1 complex: structure of the active catalytic region of C1r Mol Immunol 45, 1752–1760 37 Budayova-Spano M, Lacroix M, Thielens NM, Arlaud GJ, Fontecilla-Camps JC & Gaboriaud C (2002) The crystal structure of the zymogen catalytic domain of complement protease C1r reveals that a disruptive mechanical... Barlow PN (2004) Structural analysis of the complement control protein (CCP) modules of GABA(B) receptor 1a: only one of the two CCP modules is compactly folded J Biol Chem 279, 48292–48306 ´ ´ 35 Major B, Kardos J, Kekesi KA, L}rincz Z, Zavodszky o ´ P & Gal P (2010) Calcium-dependent conformational flexibility of a CUB domain controls activation of the complement serine protease C1r J Biol Chem 285,... Mildvan AS (1992) NMR docking of a substrate into the X-ray structure of staphylococcal nuclease Proteins 13, 275–287 50 Arlaud GJ, Gagnon J, Villiers CL & Colomb MG (1986) Molecular characterization of the catalytic domains of human complement serine protease C1r Biochemistry 25, 5177–5182 51 Gill SC & von Hippel PH (1989) Calculation of protein extinction coefficients from amino acid sequence data Anal... regulators of complement activation J Mol Biol 272, 253–265 Uhrinova S, Lin F, Ball G, Bromek K, Uhrin D, Medof ME & Barlow PN (2003) Solution structure of a functionally active fragment of decay-accelerating factor Proc Natl Acad Sci USA 100, 4718–4723 Smith BO, Mallin RL, Krych-Goldberg M, Wang X, Hauhart RE, Bromek K, Uhrin D, Atkinson JP & Barlow PN (2002) Structure of the C3b binding site of CR1 (CD35),... Mark L, Ball G, Persson J, Lindahl G, Uhrin D, Blom AM & Barlow PN (2006) Human C4b-binding protein, structural basis for interaction with streptococcal M protein, a major bacterial virulence factor J Biol Chem 281, 3690–3697 31 Hoshino M, Hagihara Y, Nishii I, Yamazaki T, Kato H & Goto Y (2000) Identification of the phospholipid-binding site of human beta(2)-glycoprotein I domain V by heteronuclear magnetic... module pair of human complement protease C1r binds Ca2+ with high affinity and mediates Ca2+-dependent interaction with C1s J Biol Chem 274, 9149–9159 46 Kirkitadze MD, Henderson C, Price NC, Kelly SM, Mullin NP, Parkinson J, Dryden DT & Barlow PN (1999) Central modules of the vaccinia virus complement control protein are not in extensive contact Biochem J 344, 167–175 47 Kirkitadze MD, Krych M, Uhrin D,... congregate in a polyanion recognition patch on human factor H revealed in three-dimensional structure J Biol Chem 281, 16512–16520 33 Henderson CE, Bromek K, Mullin NP, Smith BO, Uhrı´ n D & Barlow PN (2001) Solution structure and dynamics of the central CCP module pair of a poxvirus complement control protein J Mol Biol 307, 323–339 34 Blein S, Ginham R, Uhrin D, Smith BO, Soares DC, Veltel S, McIlhinney... Complete amino acid sequence of the A chain of human complement- classical-pathway enzyme C1r Biochem J 241, 711–720 6 Arlaud GJ & Gagnon J (1983) Complete amino acid sequence of the catalytic chain of human complement subcomponent C1-r Biochemistry 22, 1758–1764 ´ ´ 7 Lacroix M, Ebel C, Kardos J, Dobo J, Gal P, ´ Zavodszky P, Arlaud GJ & Thielens NM (2001) Assembly and enzymatic properties of the catalytic . Intermodule cooperativity in the structure and dynamics of consecutive complement control modules in human C1r Structural biology Andra ´ sLa ´ ng 1, *,. reverse of the situation observed for the CCP1 single and CCP2 single modules and probably reflects the complex interdependence of the modules in terms of internal