Báo cáo khoa học: Solution structure of human proinsulin C-peptide ppt

Solution structure of human proinsulin C-peptide Claudia Elisabeth Munte 1 , Luciano Vilela 2 , Hans Robert Kalbitzer 3 and Richard Charles Garratt 1 1 Instituto de Fı ´ sica de Sa˜o Carlos, Universidade de Sa˜ o Paulo, Sa˜ o Carlos, Brazil 2 Biomm S.A., Montes Claros, Brazil 3 Institut fu ¨ r Biophysik und Physikalische Biochemie, Universita ¨ t Regensburg, Germany C-peptide is an enzymatic cleavage product that arises from proinsulin during maturation in the b cells of the islets of Langerhans [1,2]. Two endopeptidases cleave the proinsulin at two sites marked by pairs of dibasic amino acids [2,3]. The type-I endopeptidase cleaves at the junction of the B⁄ C chains of proinsulin and the type-II endopeptidase cleaves at the proinsulin C ⁄ A junction. The basic amino acids at both sides are then removed through the action of carboxypeptidase H. After cleavage is complete, C-peptide and insulin are produced and stored in mature secretory granules until they are released in equimolar amounts from b cells [4,5]. The 31-residue C-peptide has long been considered to be merely auxiliary for the correct folding of insulin, lacking any biological activity [6–8]. However, several studies in diabetic patients and animal models during the last 10 years have changed this view and it is now considered to present biological activity by binding to target cells, activating a G-protein-coupled signalling response [9–12]. C-peptide elicits a number of cellular responses, including Ca 2+ influx [9,13] and the activa- tion of a series of enzymes including Na + ⁄ K + -ATPase [9,14], endothelial nitric oxide synthase [10,15], and mitogen-activated protein kinases [16]. Administration of C-peptide to insulin-dependent diabetic patients is accompanied by a concentration-dependent rise in blood flow to the kidneys, muscle, skin, and nerves in the diabetic state [12,17,18]. Although the amino acid sequences of the C-peptide from different species are quite variable, they do present several relatively well conserved sequences, such as the N-terminal acidic region, the glycine-rich central seg- ment, and the highly conserved C-terminal pentapeptide (Fig. 1). It has been recently demonstrated that muta- tions in the N-terminal region have significant effects on the in vitro refolding of proinsulin, probably due to interactions with the A and B chains. It is therefore believed to present an intramolecular chaperone-like function important for proinsulin folding [19]. On the Keywords CA knuckle; NMR; proinsulin C-peptide; protein secondary structure Correspondence H. R. Kalbitzer, Institut fu ¨ r Biophysik und Physikalische Biochemie, Universita ¨ t Regensburg, 93040 Regensburg, Germany Tel: + 49 941 943 2595 E-mail: hans-robert.kalbitzer@biologie. uni-regensburg.de R. C. Garratt, Instituto de Fı ´ sica de Sa˜o Carlos, Universidade de Sa˜o Paulo, Caixa Postal 369, 13560–970 Sa˜o Carlos SP, Brazil Tel: + 55 16 33739881 Fax: + 55 16 33739881 E-mail: richard@if.sc.usp.br (Received 9 March 2005, revised 25 May 2005, accepted 4 July 2005) doi:10.1111/j.1742-4658.2005.04843.x The C-peptide of proinsulin is important for the biosynthesis of insulin, but has been considered for a long time to be biologically inert. Recent studies in diabetic patients have stimulated a new debate about its possible regulatory role, suggesting that it is a hormonally active peptide. We des- cribe structural studies of the C-peptide using 2D NMR spectroscopy. In aqueous solution, the NOE patterns and chemical shifts indicate that the ensemble is a nonrandom structure and contains substructures with defined local conformations. These are more clearly visible in 50% H 2 O ⁄ 50% 2,2,2-trifluoroethanol. The N-terminal region (residues 2–5) forms a type I b-turn, whereas the C-terminal region (residues 27–31) presents the most well-defined structure of the whole molecule including a type III¢ b-turn. The C-terminal pentapeptide (EGSLQ) has been suggested to be respon- sible for chiral interactions with an as yet uncharacterized, probably a G-protein-coupled, receptor. The three central regions of the molecule (resi- dues 9–12, 15–18 and 22–25) show tendencies to form b-bends. We propose that the structure described here for the C-terminal pentapeptide is consis- tent with the previously postulated CA knuckle, believed to represent the active site of the C-peptide of human proinsulin. 4284 FEBS Journal 272 (2005) 4284–4293 ª 2005 FEBS other hand, it is the C-terminal pentapeptide of C-peptide that has been observed to elicit the full activ- ity of intact C-peptide in stimulating Na + ⁄ K + -ATPase [20]. Furthermore it elicits an increase in intracellular calcium [13], and causes phosphorylation of mitogen- activated protein kinases in human renal tubular cells [12]. In addition, the pentapeptide is capable of fully dis- placing C-peptide bound to renal tubular cell mem- branes [21,22], supporting the view that the C-terminal segment may constitute an active site. The glycine-rich central portion also exhibits some stimulatory effects on Na + ⁄ K + -ATPase activity [20], and it is reported to be important for normalization of glucose-induced vascu- lar dysfunction in a rat model [23]. Activities associated with this central region appear not to be particularly residue-specific, as fragments containing amino acid substitutions or non-natural d-amino acid, are also partly active. However, a peptide comprising amino acid residues 11–19 derived from the central portion of the C-peptide is unable to displace cell membrane-bound human C-peptide, suggesting that the mechanisms asso- ciated with this region are different from those of the C-terminal segment [22]. Structural models for proinsulin and for the C-pep- tide have been suggested in recent years, on the basis of empirical analyses [23,24] and spectroscopic experi- ments, such as NMR [25,26], photochemically induced dynamic nuclear polarization [25], Fourier transformed infrared [27] and CD [26]. Many of the resulting models are consistent, at least in part, but there is remaining conflict, principally about the probable C-terminal act- ive site. With a view to determining structural motifs within the human proinsulin C-peptide that are consis- tent with the clinical and physiological results thus far reported for C-peptide fragments, we performed the high-resolution 2D NMR studies presented here. Results Sequential assignments and secondary structure The C-peptide was studied in different solvents, in aqueous solution (95% H 2 O ⁄ 5% D 2 O) and in mix- tures of trifluoroethanol with water (50% H 2 O ⁄ 50% 2,2,2-trifluoroethanol-d2, and 20% H 2 O ⁄ 80% 2,2,2- trifluoroethanol-d2). We succeeded in obtaining com- plete spin-system assignments under these conditions. Data were deposited in the BioMagResBank (accession code 6623). The NOE path for the sample in water is shown in Fig. S1, and the same pattern is preserved in the other two samples. The chemical shifts in aqueous solution deviate clearly from those observed for random-coil peptides. This effect is strengthened when 2,2,2-trifluoroethanol is added, which is known to stabilize secondary-struc- ture formation in most cases [28]. In nonisotope enriched samples especially, the chemical shifts of the a-protons can be used to predict the secondary struc- ture in well-folded proteins. In peptides existing in a fast equilibrium between different partially folded con- formations, it can be used to predict the secondary- structure content of the time or ensemble average. The differences DdH a of the measured chemical shifts from random coil conformation values published by Wishart et al. [29] have been depicted for the three samples in Fig. 2A. In general, the tendencies visible in aqueous solution are enhanced by the addition of 2,2,2-trifluoro- ethanol, that is the content of the corresponding local structures in the ensemble is increased by 2,2,2-trifluoro- ethanol. In well-folded proteins, consecutive positive DdH a values are indicative of b-strands, whereas con- secutive negative values are characteristic of helices. In our partially folded peptide, they indicate a helical tendency (small negative DdH a ) for the Glu1–Glu11 and Gln22–Glu27 sequences, suggesting that these resi- dues belong to either short unstable helices or turns coexisting in solution. The NOE contact maps displayed in Fig. 2B,C show a summary of sequential and intermediate-range NOEs for the samples in water and in 50% H 2 O ⁄ 50% 2,2,2- trifluoroethanol-d2. Together with the 3 J NHa coupling constants and the secondary chemical shifts (Fig. 2A), the NOE pattern characterizes the local structure of the peptide. As already mentioned, the analysis of the chemical shifts in aqueous solution clearly shows that the peptide adopts nonrandom structures in the time average. However, as even after the addition of 2,2,2- trifluoroethanol no long-range NOEs could be Fig. 1. Amino acid sequences of proinsulin C-peptide from nine mammalian species. Conserved residues are marked in bold. C. E. Munte et al. Solution structure of C-peptide FEBS Journal 272 (2005) 4284–4293 ª 2005 FEBS 4285 observed, the peptide is not expected to occur in a unique, compactly folded state but rather in an exten- ded conformation or, more likely, in multiple conform- ational states. As is to be expected, a significant increase in the number of NOEs could be observed in the sample in 50% H 2 O ⁄ 50% 2,2,2-trifluoroethanol- d2. Despite some ambiguities, there are some NOEs that are not observable in the sample in water, but present in the sample in 50% H 2 O ⁄ 50% 2,2,2-trifluoro- ethanol-d2, as can be seen in Fig. 3 for the H N -H N - contact region. Addition of 2,2,2-trifluoroethanol shifts the equilibrium to states with higher structural organ- ization, particularly in the sequentially highly con- served C-terminal region. A B C Fig. 2. Local structure of the C-peptide as obtained from the devia- tions of 1 H a chemical shifts from random-coil values and from the NOE contact map. (A) Values of the conformation-dependent sec- ondary shifts DdH a are plotted with solid bars: in black for the C-peptide in water, in dark grey for the C-peptide in 50% H 2 O ⁄ 50% 2,2,2-trifluoroethanol-d2, and in pale grey for the C-pep- tide in 20% H 2 O ⁄ 80% 2,2,2-trifluoroethanol-d2. (B,C) The intensi- ties of the sequential proton–proton NOE connectivities d NN (i,i +1), d aN (i,i +1),d bN (i,i +1)(d instead of N for proline residues) for the peptide in water (B) and in 50% H 2 O ⁄ 50% 2,2,2-trifluoroethanol-d2 (C) are represented as strong, medium and weak by the height of the bars; existing but ambiguous NOE cross-peaks are marked in grey. The observed medium-range NOEs d NN (i,i +2), d aN (i,i +2), d bN (i,i +2),d aN (i,i +3),d ab (i,i + 3) are indicated by lines connecting the two residues that are related by the NOE. The absence of some medium-range connectivities may be due to ambiguous or nonexisting NOEs. J-coupling constants 3 J NH-Ha are displayed by open circles for J > 8 Hz, filled circles for J < 6 Hz and crosses for values between 6 and 8 Hz. A B Fig. 3. Amide region of the 2D NOESY spectrum for the C-peptide. The NOEs for the peptide in (A) water and (B) 50% H 2 O ⁄ 50% 2,2,2-trifluoroethanol-d2 are labelled on the spectrum. Some sequential NOEs are unresolved because of resonance overlap. Solution structure of C-peptide C. E. Munte et al. 4286 FEBS Journal 272 (2005) 4284–4293 ª 2005 FEBS Three-dimensional structure The chemical shifts of the peptide indicate that only a small proportion of the peptide is locally folded in aqueous solution. This is in line with the relatively small number of intermediate range NOEs observable under these conditions. The chemical shifts indicate that with increasing 2,2,2-trifluoroethanol concentra- tion the content of local secondary structures increases. The best-resolved NOESY spectra were obtained for 50% H 2 O ⁄ 50% 2,2,2-trifluoroethanol-d2, therefore this sample was used for the subsequent structure calcula- tions. The NOESY spectrum for the sample in 20% H 2 O ⁄ 80% 2,2,2-trifluoroethanol-d2 showed a lower spectral quality with broad resonance lines (probably owing to the increased viscosity of the solution and exchange of amide protons with 2,2,2-trifluoroethanol hydroxy protons). For the 50% H 2 O ⁄ 50% 2,2,2-trifluoroethanol sam- ple, a total of 268 NOE distance restraints were obtained in the NOESY spectra and used in the final structure calculation. In addition, 10 3 J NHa coupling constant restraints were obtained from an analysis of the COSY spectra. On the basis of both the total and NOE energies, the 30 structures that presented the lowest energies were selected for further analysis. The structural statistics can be seen in Table 1. The relat- ively large NOE energies probably reflect the overall flexibility of the structure, which may cause conflicting NOEs. However, the number of NOEs with violations larger than 0.5 A ˚ (five) is rather small and concerns mainly the ill-defined central region of the peptide. The backbone superimposition of the best structures did not reveal a defined tertiary structure, so a search for structured regions was undertaken by superimpo- sing the peptide main chain in sections. RMSDs were calculated within a sliding window of four, five and six amino acids, as shown in Fig. 4A. The regions Ala2– Leu5, Gln9–Leu12, Gly15–Ala18 and Gln22–Ala25 all show a reasonable superimposition marked as local minima in the red curve, indicating the presence of more locally structured fragments in these portions of the peptide. The best superimposition, however, occurs in the C-terminal region, comprising the last five resi- dues (Glu27–Gln31), which presents an RMSD for all backbone atoms of 0.10 A ˚ . These five regions also all exhibit a higher density of experimental restraints, as can be seen in Fig. 4B. The C-peptide therefore appears to be subdivided into a series of regions with better-defined structures connected by regions with a limited number of NOEs. The lack of observable con- tacts between these regions may be due to a real spa- tial separation between protons (greater than 4.5 A ˚ ) and ⁄ or ambiguous NOE cross-peaks that could not be assigned. These five regions have been individually analysed, and a superimposition of the main-chain atoms of the 30 selected structures can be seen in Fig. 4C. Ala2–Leu5 In all structures the distances O(2)–H N (5) (between the carbonyl oxygen of residue 2 and the amide hydrogen of residue 5), as well as the angle defined by O(2)–N H (5)–H N (5) are compatible with the presence of a hydrogen bond (distance ¼ 1.9–2.3 A ˚ ; angles ¼ 30–38°). The /, w angles of residues 3 and 4 indicate a type I b-turn. This appears to be the second most highly structured part of the molecule after the C-ter- minal pentapeptide (see below). Gln9–Leu12, Gly15–Ala18 and Gln22–Ala25 In these three regions, the main-chain superimposition does not indicate any well-defined structural element, as can be seen for example in Fig. 4C for the region Gly15–Ala18. The Ramachandran plot shows a large Table 1. Structural statistics. C-peptide at pH 7.0 and 283 K, in 50% H 2 O ⁄ 50% 2,2,2-trifluoroethanol -d2. Number of experimental restraints NOEs 268 Intraresidual (i,i)134 Sequential (i,i+1) 90 Medium-range (i,i+j;1<j £ 4) 44 Long-range (i,i+j;4<j )0 3 J NHa 10 Structural statistics for the 30 lowest energy structures (from 800 calculated) Energy (kcalÆmol -1 ) Total 549 ± 19 Bond 47.1 ± 3.1 Angle 115.7 ± 5.8 Improper 8.1 ± 1.0 van der Waals 85.9 ± 8.9 NOE 280 ± 12 Coupling 12.4 ± 3.3 RMSDs (A ˚ ) a Whole peptide 4.731 (5.830) Amino acid 2–5 0.130 (0.944) Amino acid 9–12 0.490 (1.394) Amino acid 15–18 0.309 (0.591) Amino acid 22–25 0.350 (1.129) Amino acid 27–31 0.103 (0.844) a All backbone atoms; values in parentheses all non-hydrogen atoms. C. E. Munte et al. Solution structure of C-peptide FEBS Journal 272 (2005) 4284–4293 ª 2005 FEBS 4287 dispersion for the four residues in all three regions. For these regions the number of distance restraints is insufficient to define a unique conformation, and most structures do not exhibit distances consistent with the presence of hydrogen bonds. A slight tendency to induce a turn in the peptide main chain could be clas- sified as a bend, with distances between residues i and i +3 below 7 A ˚ . Glu27–Gln31 As shown in Fig. 4C, this region is in sharp contrast with the remainder of the structure characterized by the excellent main-chain superimposition of the 30 structures. The RMSD for this pentapeptide is signifi- cantly smaller than that for the tetrapeptides described above, showing that this region is by far the most highly structured part of the C-peptide. The side chains for this short segment also seem to be well defined, especially that of Leu30. The O(27)–H N (30) distance and the O(27)–N H (30)–H N (30) angle are consistent with the presence of a hydrogen bond (1.9–2.7 A ˚ and 21–29°, respectively). This bond seems to be bifurcated in which the backbone carbonyl of residue 27 is also hydrogen-bonded to the H N (31) atom, with O(27)–H N (31) distance of 1.7–2.8 A ˚ and O(27)–N H (31)–H N (31) angle of 12–21°. The two predicted hydrogen bonds are indicated in Fig. 5A. The /,w angles of Gly28 and Ser29 characterize a type III¢ b-turn, as can be confirmed in the Ramachandran plot (Fig. 5B). Leu30 exhibits a poorly favoured, but not forbidden main-chain conformation for a leucine. The type III¢ b-turn is extremely well defined, showing /,w angles for Gly28–Leu30, which consistently reside in the same regions of the Ramachandran plot for all 800 structures initially generated by the simulated A B C Fig. 4. Structured regions of the C-peptide in 50% H 2 O ⁄ 50% 2,2,2-trifluoroethanol-d2. (A) RMSD calculated from the peptide main- chain superimposition within a sliding win- dow of four (red), five (pale blue) and six (dark blue) amino acids. (B) Density of experimental distance restraints (blue lines). (C) Superimposition of the main-chain atoms of the 30 selected structures for the C-pep- tide in 50% H 2 O ⁄ 50% 2,2,2-trifluoroethanol- d2, for the N-terminal region Ala2–Leu5, the central region Gly15–Ala18, and the C-ter- minal region Glu27–Gln31 (the main chains are indicated in black and the side chains in grey). Solution structure of C-peptide C. E. Munte et al. 4288 FEBS Journal 272 (2005) 4284–4293 ª 2005 FEBS annealing protocol. To eliminate the possibility that the type III¢ b-turn may result from the force field only and not the experimental NOEs, the structures obtained were submitted to a new refinement at low temperature in the absence of experimental restraints. In none of the models so produced did the type III¢ b-turn persist, showing it to be a consequence of the experimental NOEs. In larger protein structures, type III¢ b-turns are relatively rare, and its appearance in these simulations suggested the possibility that the number of experimen- tal restraints may not be sufficient to distinguish it from its more common mirror image, the type III b-turn. To eliminate this possibility, the /,w angles of Ser29–Leu30 for one of the structures were changed in such a way as to convert the type III¢ b-turn into a type III b-turn. However, the resulting interatom distances found in the new model were inconsistent with the NOESY spectra measured for the C-peptide. Discussion The analysis of the chemical shifts and NOEs of the C-peptide dissolved in solution shows that it is neither well folded nor has a random structure. The data are typical for a structural ensemble in fast equilibrium on the NMR time scale, favouring some local structures. The addition of 2,2,2-trifluoroethanol shifts the equi- librium in the accessible conformational space towards specific local structures. However, as is often found for peptides in aqueous solution, some typical NOE con- tacts that are present in 2,2,2-trifluoroethanol are still observed with reduced intensity in water, and the chemical shift deviations from the random-coil values are qualitatively still in agreement with the structure found in the presence of 2,2,2-trifluoroethanol. This is usually interpreted as the existence of a small popula- tion of the local and global structural states stabilized by 2,2,2-trifluoroethanol, which are mixed with other ‘random’ structures. An indication of such an avera- ging would be a concentration dependence on the cosolvent, the extrapolation of which to zero would lead to nonrandom, qualitatively still correct values for the chemical shift changes and interatomic distances. Such behaviour is found for the C-peptide in our stud- ies. The local structures determined in the presence of 2,2,2-trifluoroethanol, especially the extremely well- defined structure found for the C-terminal region, would thus also exist in low populations in aqueous solution and would be stabilized in a less polar envi- ronment, as is to be expected during the interaction with its receptor or with cell membranes. Water is in general excluded from these interactions favouring the formation of this structure. Therefore the human pro- insulin C-peptide structure presented in this work is expected to be physiologically relevant, despite the nonphysiological conditions used for the structure determination itself (presence of 2,2,2-trifluoroethanol). Human proinsulin C-peptide dissolved in 50% H 2 O ⁄ 50% 2,2,2-trifluoroethanol-d2 does not present a well-defined global tertiary structure. No long-range inter-residual NOEs could be assigned in the NOESY A B Fig. 5. Structure of the C-terminal pentapeptide of the C-peptide in 50% H 2 O ⁄ 50% 2,2,2-trifluoroethanol-d2. (A) One selected model showing the two predicted hydrogen-bonds. (B) Ramachandran plot, indicating the main-chain conformation of Glu27 (green dots), Gly28 (red dots), Ser 29 (blue dots) and Leu30 (black dots). C. E. Munte et al. Solution structure of C-peptide FEBS Journal 272 (2005) 4284–4293 ª 2005 FEBS 4289 spectra, which would be essential for the convergence of the models to a well-defined, compact tertiary struc- ture. However, a detailed analysis of the models obtained shows the existence of five regions with rather well-defined local structures. The N-terminal region (Ala2–Leu5) possesses the basic features of a type-I b-turn. The potential helical structure initiated by this turn is consistent with the lower frequency chemical-shift deviations in this region and with previous results, including theoretical predic- tions [23,24] and NMR spectroscopy [26]. However, there is a discrepancy about the size of the structured region. Unlike in previous studies, the nascent helix encountered here is broken by Gln6, which is evi- denced by its random-coil H a chemical shift and by the random /,w-angle distribution in the Ramachan- dran plot of the experimental structures. The unambig- uous absence of an NOE between the H a proton of Asp4 and the H N proton of Val7 (expected always to exist in a helix because of separation within the range 3.3–3.5 A ˚ between these atoms) endorses our conjec- ture that the helix is short. Despite the fact that Gln6 is highly conserved in C-peptides from different spe- cies, Leu5 is found to be replaced in all species but humans by a proline, a known helix breaker. Recent experiments performed with proinsulin reveal that deletions or alanine mutations of the N-terminal acidic amino acids of the C-peptide result in the formation of large aggregates during in vitro refolding [19]. It is sug- gested that these results indicate that the highly con- served acidic N-terminal part of the C-peptide may have some intramolecular chaperone-like function in the folding of the insulin precursor. The presence of a highly conserved N-terminal tetrapeptide also suggests the existence of a functionally active site in the B ⁄ C junction of proinsulin, and has been proposed to con- stitute part of the type I endopeptidase recognition site [19,30]. The b-turn structure that has been found for the Ala2–Leu5 region in this study is coherent with these previous findings. The superimposition of the three central regions (Gln9–Leu12, Gly15–Ala18 and Gln22–Ala25) is not as good as that of the N-terminal region, with little indication of any well-defined, typical secondary struc- ture. The number of experimental restraints in these regions is low, probably resulting from ambiguities in the NOESY spectra, making it impossible to conclude if the three regions are really unordered or not. Previ- ous results with the use of spectroscopic methods, such as Fourier transformed infrared [27], NMR and CD [25,26], confirm that these regions tend to be dis- ordered. Published studies of Na + ⁄ K + -ATPase activ- ity in rat renal tubule segments (the stimulation of which by C-peptide has been previously described [18]) revealed the existence of peptide fragments derived from part of the central portion of the molecule that exhibited some stimulatory activity [20]. In human C-peptide, the sequence from residues 13–19 (GGGP- GAG) is unusual in that it is nearly symmetrical with respect to the central proline, possesses solely nonpolar residues, and has a high content of the nonchiral amino acid glycine. These residues have been proposed to form a turn-like structure, which is relevant to non- chiral interactions with membranes [23]. The slight ten- dency to a bend found in the three central regions is consistent with these results. The C-terminal region (Glu27–Gln31) contrasts sharply with the remainder of the structure in that it presents an excellent superimposition for the 30 mod- els. The Glu27–Leu30 tetrapeptide forms a type-III¢ b-turn stabilized by a hydrogen bond between the Glu27 carbonyl and the Leu30 amide. This hydrogen bond is bifurcated and also involves the amide from Gln31 (the highly conserved C-terminal residue of the C-peptide). These results are consistent with the existence of a well-defined structure for the EGSLQ C-terminal pentapeptide. Although a type-III¢ b-turn is a secondary-structure element not commonly found in polypeptides, it is favoured by the presence of glycine in position (i + 1) and serine in (i + 2), both of which are able to adopt a left-handed helical conformation. Residue Leu30 possesses an excellent side-chain super- imposition among the structures and seems to be sta- bilized by van der Waals interactions. In 2D NMR experiments comparing proinsulin and insulin [25], the authors described perturbations to the 2D NMR resonances assigned to the hydrophobic core of the insulin moiety of proinsulin. These perturba- tions were reversed by site-specific cleavage at the C ⁄ A junction but not the B ⁄ C junction. The authors suggest the existence of a stable local structure at the C ⁄ A junction, which has been designated the ‘CA knuckle’, involving a nonstandard secondary structure, accessible to solvent and not involving distant regions of the C-peptide [25]. These results are wholly consistent with the structure presented in this work for the C-terminal region of the human proinsulin C-peptide. In the same C-peptide fragment activity experiments described above for the central region, a second fragment invol- ving residues 27–31 was found to elicit stimulatory effects on the Na + ⁄ K + -ATPase [20]. In addition, the C-terminal pentapeptide was also effective in the dis- placement binding studies in which C-peptide had been previously bound to membranes of several human cell types (renal tubular cells, skin, fibroblasts, and saph- enous vein endothelial cells) [21]. The results, which Solution structure of C-peptide C. E. Munte et al. 4290 FEBS Journal 272 (2005) 4284–4293 ª 2005 FEBS are sequence-specific, point to the existence of a C-pep- tide-specific receptor that recognizes the C-terminal pentapeptide [20,21]. Thus a well-defined secondary structure is to be expected for the EGSLQ pentapep- tide, perhaps the previously foreseen ‘CA knuckle’, which may well represent the elusive active site of the C-peptide itself [25]. Our results showing an extremely well-defined structure for the Glu27–Gln31 region are compatible with these concepts. Proinsulin is processed to produce insulin by the action of two distinct endopeptidases followed by the loss of two basic amino acids via carboxypeptidase H cleavage acting on both the B-peptide and C-peptide. It is well known that the removal of the C-terminal argi- nines from the B-chain of insulin is essential for the interaction of mature insulin with its receptor [2]. How- ever, the processing of the C-peptide is rather curious if it does not have any useful physiological role. The loss of the two C-terminal basic amino acids from the proin- sulin C-peptide has been shown to be fundamental to its biological activities [21], conceivably because of the exposure of the highly conserved Gln31, suggesting an important functional role for this residue. The conclu- sion drawn here that the C-terminal region of the proc- essed C-peptide is its most highly structured portion, consistent with a physiological role, provides an explan- ation for C-peptide C-terminal processing during insulin maturation. Gly28–Ser29, which are expected to be structurally important for the maintenance of the type III¢ b-turn, are not absolutely conserved in different species, being notably absent in both rat C-peptides (Fig. 1). This is probably connected with either the existence of species-specific C-peptide receptors or a nonconserved C-peptide activity across species. It is relevant to note therefore that rat C-peptide failed to bind to human cells [11,22], suggesting that the structure found in the present work for the human C-terminal pentapeptide is not expected to exist in the rat homologues. Indeed in the rat C-terminal pentapeptides, the Gly28–Ser29 sequence is replaced by Val–Ala, which would not be expected to form a type III¢ b-turn principally because of the bifurcation of the valine b-carbon which makes the left-handed a-helical region of the Ramachandran space inaccessible. Finally, and most importantly, our results support the idea of structured N-terminal and C-terminal regions for the peptide. The latter has been previously suggested to form the active site of the human proinsu- lin C-peptide, and has been baptised the CA knuckle. The results described here have gone further than previous studies in terms of the structural characteriza- tion of this region. We suggest that the knuckle, as described here, represents the most structurally ordered segment of an otherwise flexible peptide. However, whether this structure is assumed by the peptide under physiological conditions remains to be demonstrated. It is hoped that our observation that the most highly structured region of the molecule appears also to correspond to the predicted active site will stimulate further investigation. Experimental procedures Sample preparation Human C-peptide, a byproduct of the industrial prepar- ation of human recombinant insulin, was a gift from Biomm S.A. A stock solution of 6 mm unlabelled human C-peptide in distilled water was used to prepare three dif- ferent samples: C-peptide in 95% H 2 O ⁄ 5% D 2 O(v⁄ v), in 50% H 2 O ⁄ 50% 2,2,2-trifluoroethanol-d2 (v ⁄ v), and in 20% H 2 O ⁄ 80% 2,2,2-trifluoroethanol-d2 (v ⁄ v). The pH of each was adjusted to 7.0 by the addition of appropriate quanti- ties of NaOH. As internal reference 2,2-dimethyl-2-silapen- tane-5-sulfonate was added to a final concentration of 0.05 mm. NMR spectroscopy NMR experiments were performed on a Bruker DRX-600 spectrometer (proton frequency of 600 MHz). All spectra were recorded at 283 K. The water signal was suppressed by selective presaturation. 2D data sets were recorded with 4096 complex t 2 points and 1024 t 1 increments; phase- sensitive detection in the t 1 direction was obtained with time- proportional phase incrementation [31]. NOESY [32] spectra were recorded with a mixing time of 80, 100, 150 and 200 ms to check for possible spin diffusion effects and to allow nor- mally weak NOEs to become more apparent. TOCSY [33] spectra were recorded with spin-lock times of 80 ms using a MLEV-17 [34] sequence. DQF-COSY spectra were obtained as described by Rance et al. [35]. The time-domain data were processed using the xwinnmr package (Bruker) and evalu- ated with the program aurelia [36]. Experimental restraints Assignment of resonance lines was performed according to the standard strategy for homonuclear spectroscopy [37] using DQF-COSY and TOCSY spectra for the identification of the spin systems and NOESY spectra for the sequence- specific and NOE assignment. Amide–Ha coupling constants 3 J NHa were determined from a slightly exponentially filtered DQF-COSY spectrum by fitting the antiphase signals to a pair of Lorentzians using the corresponding routine from aurelia. Upper and lower boundaries for coupling constant C. E. Munte et al. Solution structure of C-peptide FEBS Journal 272 (2005) 4284–4293 ª 2005 FEBS 4291 restraints were set to 0.5 Hz. The NOESY cross-peaks were integrated by the automated segmentation procedure of the program aurelia, and distances were calculated applying the initial slope approximation. A set of well-resolved methy- lene resonances (assumed interproton distance of 1.76 A ˚ ) were taken as reference distances. The upper and lower dis- tance bounds were taken as described in [38]. Structure calculations and analysis Structures of the C-peptide in 50% H 2 O ⁄ 50% 2,2,2-trifluoro- ethanol-d2 were calculated by simulated annealing using the program CNS 1.0 [39], starting from extended structures. High-temperature torsion-angle dynamics were run for 15 ps at an initial temperature of 3000 K. The system was then slowly cooled to a temperature of 0 K in 25 K steps over a period of 5 ps. At 0 K, a final stage of 150 steps of Powell minimization was performed to yield the final structures. The final values of the force constants used in the simulated annealing calculations are as follows: 1 kcalÆmol )1 ÆA ˚ )2 for bond lengths, 1 kcal Æ mol )1 Ærad )2 for angles and improper torsions, 1 kcalÆmol )1 ÆA ˚ )4 for the quadratic van der Waals repulsion term, 300 kcalÆmol )1 ÆA ˚ )2 for NOE-derived distance restraints and 1 kcalÆmol )1 Æ Hz )2 for the 3 J NHa coupling con- stant restraints. Analysis of secondary-structure elements and calculation of RMSD values were performed using the program molmol 2.6 [40]. Co-ordinates for the 30 lowest energy structures have been deposited in the Protein Data Bank (accession code 1T0C). Acknowledgements This work was supported by Fundac¸ a ˜ o de Amparo a ` Pesquisa do Estado de Sa ˜ o Paulo (FAPESP), Brazil, grant 96 ⁄ 12386-3, the DFG, and the Bayerische Fors- chungsstiftung. References 1 Steiner DF, Cunningham D, Spiegelman L & Aten B (1967) Insulin biosynthesis: evidence for a precursor. Science 157, 697–700. 2 Steiner DF, Bell G & Tager HS (1995) Chemistry and biosynthesis of pancreatic protein hormones. In Endo- crinology (DeGroot L, ed), pp. 1296–1328. Saunders, Philadelphia. 3 Smeekens SP, Montag AG, Thomas G, Albiges-Rizo C, Carroll R, Benig M, Phillips LA, Martin S, Ohagi S, Gardner P, et al. (1992) Proinsulin processing by the subtilisin-related proprotein convertases furin, PC2, and PC3. Proc Natl Acad Sci USA 89, 8822–8826. 4 Rubenstein AH, Clark J, Melani F & Steiner DF (1969) Secretion of proinsulin C-peptide by pancreatic beta cells and its circulation in blood. Nature 224, 697–699. 5 Clark PM (1999) Assays for insulin, proinsulin (s) and C-peptide. Ann Clin Biochem 36, 541–564. 6 Kitabchi AE (1977) Proinsulin and C-peptide: a review. Metabolism 26 , 47–587. 7 Steiner DF (1978) On the role of the proinsulin C-pep- tide. Diabetes 27, 145–148. 8 Hoogwerf BJ, Bantle JP, Gaenslen HE, Greenberg BZ, Senske BJ, Francis R & Goetz FC (1986) Infusion of synthetic human C-peptide does not affect plasma glu- cose, serum insulin, or plasma glucagon in healthy sub- jects. Metabolism 35, 122–125. 9 Ohtomo Y, Aperia A, Sahlgren B, Johansson B-L & Wahren J (1996) C-peptide stimulates rat renal tubular Na + ,K + -ATPase activity in synergism with neuropep- tide Y. Diabetologia 39, 199–205. 10 Kunt T, Forst T, Pfuetzner A, Beyer J & Wahren J (1999) The physiological impact of proinsulin C-peptide. Pathophysiology 5 , 257–262. 11 Wahren J, Ekberg K, Johansson J, Henriksson M, Pramanik A, Johansson B-L, Rigler R & Jo ¨ rnvall H (2000) Role of C-peptide in human physiology. Am J Physiol Endocrinol Metab 278, E759–E768. 12 Johansson J, Ekberg K, Shafqat J, Henriksson M, Chibalin A, Wahren J & Jo ¨ rnvall H (2002) Molecular effects of proinsulin C-peptide. Biochem Biophys Res Commun 295 , 1035–1040. 13 Shafqat J, Juntti-Berggren L, Zhong Z, Ekberg K, Ko ¨ hler M, Berggren P-O, Johansson J, Wahren J & Jo ¨ rnvall H (2002) Proinsulin C-peptide and its analogues induce intracellular Ca 2+ increases in human renal tubular cells. Cell Mol Life Sci 59, 1185–1189. 14 De La Tour DD, Raccah D, Jannot MF, Coste T, Rougerie C & Vague P (1998) Erythrocyte Na ⁄ K ATPase activity and diabetes: relationship with C-peptide level. Diabetologia 41, 1080–1084. 15 Scalia R, Coyle KM, Levine BJ, Booth G & Lefer AM (2000) C-peptide inhibits leukocyte–endothelium interac- tion in the microcirculation during acute endothelial dysfunction. FASEB J 14, 2357–2364. 16 Kitamura T, Kimura K, Jung BD, Makondo K, Okamoto S, Canas X, Sakane N, Yoshida T & Saito M (2001) Proinsulin C-peptide rapidly stimulates mitogen- activated protein kinases in Swiss 3T3 fibroblasts: requirement of protein kinase C, phosphoinositide 3-kinase and pertussis toxin-sensitive G-protein. Biochem J 355, 123–129. 17 Johansson B-L, Kernell A, Sjo ¨ rberg S & Wahren J (1993) Influence of combined C-peptide and insulin administration on renal function and metabolic control in diabetes type 1. J Clin Endocrinol Metab 77, 976–981. 18 Forst T, Kunt T, Pohlmann T, Goitom K, Engelbach M, Beyer J & Pfu ¨ tzner A (1998) Biological activity of C-peptide on the skin microcirculation in patients with Solution structure of C-peptide C. E. Munte et al. 4292 FEBS Journal 272 (2005) 4284–4293 ª 2005 FEBS insulin-dependent diabetes mellitus. J Clin Invest 101, 2036–2041. 19 Chen L-M, Yang X-W & Tang J-G (2002) Acidic resi- dues on the N-terminus of proinsulin C-Peptide are important for the folding of insulin precursor. J Bio- chem 131, 855–859. 20 Ohtomo Y, Bergman T, Johansson B-L, Jo ¨ rnvall H & Wahren J (1998) Differential effects of proinsulin C-peptide fragments on Na + ,K + -ATPase activity of renal tubule segments. Diabetologia 41, 287–291. 21 Rigler R, Pramanik A, Jonasson P, Kratz G, Jansson OT, Nygren P-A ˚ , Sta ˚ hl S, Ekberg K, Johansson B-L, Uhle ´ nS,et al. (1999) Specific binding of proinsulin C- peptide to human cell membranes. Proc Natl Acad Sci USA 96, 13318–13323. 22 Pramanik A, Ekberg K, Zhong Z, Shafqat J, Henriks- son M, Jansson O, Tibell A, Tally M, Wahren J & Jo ¨ rnvall H (2001) C-peptide binding to human cell membranes: importance of Glu27. Biochem Biophys Res Commun 284, 94–98. 23 Ido Y, Vindigni A, Chang K, Stramm L, Chance R, Heath WF, DiMarchi RD, Di Cera E & Williamson JR (1997) Prevention of vascular and neural dysfunction in diabetic rats by C-peptide. Science 277, 563–566. 24 Snell CR & Smyth DG (1975) Proinsulin: a proposed three-dimensional structure. J Biol Chem 250, 6291–6295. 25 Weiss MA, Frank BH, Khait I, Pekar A, Heiney R, Shoelson S & Neuriger LJ (1990) NMR and photo- CIDNP studies of human proinsulin and prohormone processing intermediates with application to endopepti- dase recognition. Biochemistry 29, 8389–8401. 26 Henriksson M, Shafqat J, Liepinsh E, Tally M, Wahren J, Jo ¨ rnvall H & Johansson J (2000) Unordered structure of proinsulin C-peptide in aqueous solution and in the presence of lipid vesicles. Cell Mol Life Sci 57, 337–342. 27 Xie L & Tsou C-L (1993) Comparison of secondary structures of insulin and proinsulin by FTIR. J Prot Chem 12, 483–487. 28 Rajan R & Balaram P (1996) A model for the interac- tion of trifluoroethanol with peptides and proteins. Int J Peptide Protein Res 48, 328–336. 29 Wishart DS, Bigam CG, Holm A, Hodges RS & Sykes BD (1995) 1 H, 13 C and 15 N random coil NMR chemical shifts of the common amino acids. I. Investigations of nearest-neighbor effects. J Biomol NMR 5, 67–81. 30 Gross DJ, Villa-Komaroff L, Kahn CR, Weir GC & Halban PA (1989) Deletion of a highly conserved tetra- peptide sequence of the proinsulin connecting peptide (C-peptide) inhibits proinsulin to insulin conversion by transfected pituitary corticotroph (AtT20) cells. J Biol Chem 264, 21486–21490. 31 Marion D & Wu ¨ thrich K (1983) Application of phase sensitive two-dimensional correlated spectroscopy (COSY) for measurements of 1 H- 1 H spin-spin coupling constants in proteins. Biochem Biophys Res Commun 113, 967–974. 32 Jeener J, Meier BH, Bachmann P & Ernst RR (1979) Investigation of exchange processes by two-dimensional NMR spectroscopy. J Chem Phys 71, 4546–4553. 33 Braunschweiler L & Ernst RR (1983) Coherence trans- fer by isotropic mixing: application to proton correla- tion spectroscopy. J Magn Reson 53, 521–528. 34 Bax A & Davis DG (1985) MLEV-17-based two-dimen- sional homonuclear magnetization transfer spectro- scopy. J Magn Reson 65, 355–360. 35 Rance M, Sørensen OW, Bodenhausen G, Wagner G, Ernst RR & Wu ¨ thrich K (1983) Improved spectral reso- lution in COSY 1 H NMR spectra of proteins via double quantum filtering. Biochem Biophys Res Commun 117, 479–481. 36 Neidig K-P, Geyer M, Go ¨ rler A, Antz C, Saffrich R, Beneicke W & Kalbitzer HR (1995) AURELIA: a pro- gram for computer-aided analysis of multidimensional NMR-spectra. J Biomol NMR 6, 255–270. 37 Wu ¨ thrich K (1986) NMR of Proteins and Nucleic Acids. Wiley, New York. 38 Kalbitzer HR & Hengstenberg W (1993) The solution structure of the histidine-containing protein (HPr) from Staphylococcus aureus as determined by two-dimensional 1 H-NMR spectroscopy. Eur J Biochem 216, 205–214. 39 Brunger AT, Adams PD, Clore GM, Delano WL, Gros P, Grosse-Kunstleve RW, Jiang J-S, Kuszewski J, Nilges M, Pannu NS, et al. (1998) Crystallography & NMR system: a new software suite for macromolecular structure determination. Acta Crystallogr D Biol Crstal- logr 54, 905–921. 40 Koradi R, Billeter M, Engeli M, Guntert P & Wu ¨ thrich K (1996) molmol: a program for display and analysis of macromolecular structures. J Mol Graph 14, 51–55. Supplementary material The following material is available for this article online: Fig. S1. NOE path of the C-peptide in water. C. E. Munte et al. Solution structure of C-peptide FEBS Journal 272 (2005) 4284–4293 ª 2005 FEBS 4293 . the active site of the C-peptide of human proinsulin. 4284 FEBS Journal 272 (2005) 4284–4293 ª 2005 FEBS other hand, it is the C-terminal pentapeptide of C-peptide. sequences of proinsulin C-peptide from nine mammalian species. Conserved residues are marked in bold. C. E. Munte et al. Solution structure of C-peptide FEBS