Structure and potential C-terminal dimerization of a recombinant mutant of surfactant-associated protein C in chloroform/methanol Burkhard Luy 1 , Alexander Diener 2 , Rolf-Peter Hummel 3 , Ernst Sturm 3 , Wolf-Ru¨ diger Ulrich 4 and Christian Griesinger 5 1 Institut fu ¨ r Organische Chemie und Biochemie, Technische Universita ¨ tMu ¨ nchen, Garching, Germany; 2 Institut fu ¨ r Organische Chemie, Johann Wolfgang Goethe-Universita ¨ t Frankfurt, Germany; 3 Department of Physical Organic Chemistry and 4 Department of Chemical Research, Altana Pharma AG, Konstanz, Germany; 5 Max Planck Institut fu ¨ r Biophysikalische Chemie, Go ¨ ttingen, Germany The solution structure of a recombinant mutant [rSP- C (FFI)] of the human surfactant-associated protein C (hSP-C) in a mixture of chloroform and methanol was determined by high-resolution NMR spectroscopy. rSP- C (FFI) contains a helix from Phe5 to the C-terminal Leu34 and is thus longer by two residues than the helix of porcine SP-C (pSP-C), which is reported to start at Val7 in the same solvent. Two sets of resonances at the C-terminus of the peptide were observed, which are explained by low-order oligomerization, probably dimerization of rSP-C (FFI) in its a-helical form. The dimerization may be induced by hydrogen bonding of the C-terminal carboxylic groups or by the strictly conserved C-terminal heptapeptide segment with a motif similar to the GxxxG dimerization motif of glycophorin A. Dimerization at the heptapeptide segment would be consistent with findings based on electrospray ionization MS data, chemical cross-linking studies, and CNBr cleavage data. Keywords: dimerization; NMR spectroscopy; surfactant; surfactant protein C (SP-C). Surfactant-associated protein C (SP-C) is a 34–35-amino- acid peptide which is highly conserved among species (Table 1). It is part of the protein–phospholipid complex that is secreted into the alveolar space [1] and is responsible for lowering of the alveolar surface tension. Recombinant (r)SP-C (FFI) surfactant (Venticute) has proved to be highly effective in animal experiments [2,3] as well as in pilot clinical trials [4,5]. The structure of porcine SP-C (pSP-C) has been solved in CDCl 3 /CD 3 OH/ 0.1 M HCl (32 : 64 : 5, v/v/v), and it has been found that the peptide forms an a-helix from residue 7 to the C-terminal residue 34 [6]. The N-terminal structure as well as the hydrophobic a-helix seems to be conserved in the micellar environment as shown for the N-terminal 17 residues of pSP-C in fully deuterated dodecylphospho- choline micelles [7]. A second set of resonances was found for the full-length pSP-C peptide in chloroform/methanol at the C-terminus, which was explained by partial oxidation of the methionine residue M32. In general, samples of the lipophilic pSP-C are not completely stable in chloroform/methanol mixtures and form a gel-like b-sheet aggregate after several days at 10 °C[8].Amutant of the human SP-C (hSP-C) has been produced recom- binantly by omitting the residue [Phe() 1)] that is only partially present and performing the following substitu- tions: C4F, C5F and M32I. The rationale behind the substitutions is that the two cysteine residues are naturally palmitoylated, which would have been difficult to achieve for a bacterially expressed protein. The mutation of residue 32 was to prevent the undesired putative oxidation of methionine. In this article, we present the structure of the rSP-C (FFI) mutant in CDCl 3 /CD 3 OH (1 : 1, v/v) with a comparison with the structure of pSP-C. A second set of C-terminal signals is explained by the coexistence of monomeric and oligomeric (probably dimeric) rSP-C (FFI). Materials and methods Preparation of the sample For the studies on rSP-C (FFI) (Altana Pharma AG, Konstanz, Germany; WO patent no. 95/32992), we used the solid substrate consisting of the peptide (90%), HCl (4%), propan-2-ol (3%), water (2%), and methyl ester (1%). Samples of rSP-C (FFI) were prepared by dissolving 3–12 mg of the powder in 600 lLCDCl 3 /CD 3 OH (1 : 1, v/v) or CDCl 3 /CD 3 OD (1 : 1, v/v). The resulting rSP- C (FFI) concentration was 1.1–4.4 m M , respectively. The solid peptide was stored at )20 °C, and the prepared samples were stored in liquid nitrogen between NMR measurements. Dissolved samples had a lifetime of  72 h at 10 °C. Over time, the dissolved peptide maintained identical NMR chemical shifts, but strongly reduced intensity, indicating similar aggregation to b-sheet-like Correspondence to C. Griesinger, Max Planck Institut fu ¨ r Biophysi- kalische Chemie, Abt. NMR based Structural Biology, Am Fassberg 11, 37077 Go ¨ ttingen, Germany. Fax: + 49 551201 2202, Tel.: + 49 551201 2201, E-mail: cigr@nmr.mpibpc.mpg.de Abbreviations: SP-C, surfactant-associated protein C; hSP-C, human SP-C; pSP-C, porcine SP-C; rSP-C, recombinant human SP-C; rSP-C (FFI), FFI variant of recombinant human SP-C; TACSY, taylored correlation spectroscopy. (Received 17 December 2003, revised 1 March 2004, accepted 23 March 2004) Eur. J. Biochem. 271, 2076–2085 (2004) Ó FEBS 2004 doi:10.1111/j.1432-1033.2004.04106.x structures as observed for natural pSP-C in the solvent used. Because of the limited lifetime, samples were prepared immediately before NMR measurements. NMR measurements 2D 1 H-NMR spectra were recorded on Bruker DRX 800, DMX 600, AMX 600 and AMX 400 spectrometers in the pure-phase absorption mode using the States-TPPI method [9]. All spectra were recorded at 10 °C, and processing and baseline corrections were performed using the standard Bruker software XWINNMR . The complete set of experiments recorded is given in Table 2. The 1 H-NMR chemical shifts were calibrated relative to trimethylsilane. The residual water signal and the signal of the hydroxy proton of CD 3 OH are degenerate at 4.8 p.p.m. and were reduced using presaturation [10]. Before Fourier transformation, the time domain data were multiplied with shifted squared sinebell window functions. The vicinal scalar coupling constants 3 J NHa were deter- mined using the SIAM-TACSY and Keeler–Titman approaches [11,12] using macros written by T. Prasch for the program FELIX (Felix 95; MSI, San Diego, CA, USA). Signal overlap in the 800-MHz NOESY made peak integration unreliable. So, instead, signal height of the cross-peaks was used for a conservative estimation of the maximum distances and classification of cross-peaks as weak, medium and strong. For the calibration of the intensities of the NOE peaks, a statistical analysis of the d aN (i,i+3) signals of residues 11–30 was performed using typical values for an ideal a-helix [13]. The a-helical structure of this part of the peptide is clearly evident from H a chemical shifts [14,15]. Results NMR assignment Sequence-specific 1 H-NMR assignment was achieved by standard procedures for small proteins [13] using the computer program NDEE (Spin Up, Lu ¨ nen, Germany). Owing to the high abundance of the amino acids valine, leucine and isoleucine in the sequence of rSP-C (FFI), there was extensive overlap in the homonuclear 1 H-NMR spectra. Nevertheless, almost all spin systems (vide infra) could be assigned from the TOCSY spectra (Fig. 1A) and the DQF-COSY spectra (not shown) collected under identical conditions (Table 3). The unique spin systems His8, Lys10, Arg11 and Ala29, and the pairs of Phe and Pro residues and Gly28 and Gly33 were unambiguously identified, as well as 10 of the 11 valines. The N-terminal Gly1 shows a single very broad H N /H a cross-peak. Although all 34 amino acids were found, the spin systems of seven leucines, five isoleucines and the residual valine could only be unambiguously identified using sequential NOE information. The high dispersion of the 800-MHz NOESY spectrum made it possible to obtain the complete assignment of rSP- C (FFI) (Fig. 1B,C). Starting from the unambiguously identified residues, we were able to carry out the sequential assignment for residues 1–17 and 24–34 by d aN and d NN cross-peaks. As an a-helical secondary structure was assumed from chemical-shift arguments, d aN (i,i+3) and d aN (i,i+4) NOE cross-peaks were used, leading to the assignment of the residual amino acids 18–23. We encountered special difficulties in identifying the following connectivities: the chemical shifts of the amide Table 2. NMR experiments. Sample Experiment Spectrometer frequency (MHz) Data matrix Processed matrix Mixing time (ms) Total time (h) 1.1 m M rSP-C (FFI) in CDCl 3 /CD 3 OH (1 : 1) TOCSY 600 4096 · 768 4096 · 1024 70 11 NOESY 600 4096 · 768 4096 · 1024 50 8 NOESY 600 4096 · 768 4096 · 1024 100 8 NOESY 600 4096 · 768 4096 · 1024 200 8 DQF-COSY 600 4096 · 1024 4096 · 1024 – 12 4.4 m M rSP-C (FFI) in CDCl 3 /CD 3 OH (1 : 1) NOESY 800 8192 · 1024 8192 · 1024 50 24 1.1 m M rSP-C (FFI) in CDCl 3 /CD 3 OH (1 : 1) SIAM-TACSY 600 4096 · 400 4096 · 1024 70 12 1.1 m M rSP-C (FFI) in CDCl 3 /CD 3 OH (1 : 1) NOESY 400 4096 · 1024 4096 · 1024 50 12 Table 1. Amino-acid sequences of several SP-C polypeptides, including human, porcine and recombinant human SPC with FFI substitution [rSP- C (FFI)]. Species Amino-acid sequence Numbering 1 11 21 31 hSP-C (F) GIPCCPVHLK RLLIVVVVVV LIVVVIVGAL LMGL rSP-C (FFI) GIPFFPVHLK RLLIVVVVVV LIVVVIVGAL LIGL pSP-C L RIPCCPVNLK RLLVVVVVVV LVVVVIVGAL LMGL Cow SP-C LIPCCPVNIK RLLIVVVVVV VLVVVIVGAL LMGL Rat SP-C RIPCCPVHLK RLLIVVVVVV LVVVVIVGAL LMGLH Canine SP-C GIPCFPSSLK RLLIIVVVIV LVVVVIVGAL LMGLH Ó FEBS 2004 Recombinant mutant of surfactant protein C (Eur. J. Biochem. 271) 2077 protons of Leu21, Val23 and Leu31 are degenerate so there was a large overlap in the d NN cross-peaks. As the amide protons of Ile22 and Leu30 also overlapped, the assignment was even more difficult. The identical H a chemical shifts of Leu12, Leu13 and Leu21 caused further problems in the sequential assignment. The same occurred for the d NN connectivities to Val27 because Ile26 and Gly28 have almost identical amide proton chemical shifts. Except for some side chain protons of Ile14, Ile22 and Ile26, all 1 H resonances of rSP-C (FFI) were assigned. Stereochemical assignments for Fig. 1. NMR assignment. (A) Assignment of the spin systems of 32 nonproline residues out of the 34 amino acids of rSP-C (FFI) illustrated in the TOCSY experiment with a mixing time of 70 ms. Shown is the so-called fingerprint region where the well-dispersed H N protons are correlated to the H a and side chain protons. The spin systems of Lys10 and Arg11 are indicated by rectangles as both contain a second H N in the side chain. The N-terminal Gly1 appears as a weak and very broad peak. All H a chemical shifts of residues 5–31 show an upfield shift compared with random-coil data indicating an a-helical structure in an empirical pattern-recognition approach [13,16]. (B) H N -H N region of the 800-MHz NOESY experiment. Sequential d NN (i,i+1) connectivities can be found for all nonproline amino acids. For the C-terminal residues 31–34, a second set of resonances can be sequentially assigned indicated by the prime in the annotation of the corresponding NOE connectivity. (C) H N -H a region of the 800-MHz NOESY experiment. All resolved interresidual NOE connectivities are annotated. In particular, the d Na (i,i+3) and d Na (i,i+4) connectivities are indicators of an a-helical secondary structure. Intraresidual signals are not annotated. (D) Summation of the experimental NMR data. Shown are all resolved NOE connectivities, where thin bars indicate distances > 4.0 A ˚ , medium bars distances of 3.0–4.0 A ˚ , and thick bars distances < 3.0 A ˚ . The d Na (i,i+3), d Na (i,i+4) as well as the d NN (i,i+2) and the strong d NN (i,i+1) connectivities clearly show the a-helical structure of rSP-C (FFI). In addition, 3 J NHa coupling constants are summarized, with small circles indicating couplings < 5.0 Hz and large circles for constants > 6.0 Hz. Pentagons classify the exchange properties of amide protons in weak exchange (filled pentagons), medium exchange (open pentagons) and strong exchange (no pentagon) as described in the text. 2078 B. Luy et al.(Eur. J. Biochem. 271) Ó FEBS 2004 the diastereotopic groups were inferred from NOEs through floating chirality calculations. Second set of resonances Closer inspection of the spectra revealed two sets of resonances for Gly28, Leu30, Ile32, Gly33 and Leu34, which differ mainly in the chemical shifts of the amide protons and the c protons of Ile32 and Leu34. A compar- ison of the spectra showed different relative intensities of the two sets of resonances with respect to the concentration of rSP-C (FFI) in CDCl 3 /CD 3 OH (1 : 1, v/v) and the age of the sample. For a systematic analysis, freshly prepared samples with concentrations of 0.7–3.5 m M were used in NOESY experiments with a mixing time of 50 ms. At low concentration, the two sets of signals were almost equally strong, whereas at higher concentrations of rSP-C, one of the signal sets was more predominant. Attempts to fit the relative intensities of the two sets of resonances to a quantitative monomer–dimer equilibrium model failed (data not shown). However, the concentration dependence shown in Fig. 5 can be considered an indication of intermolecular interaction. The comparable linewidths of the signals of the two sets of resonances still suggest that monomeric and dimeric units are involved. Amide proton exchange The exchange properties of the amide protons were obtained from a 400-MHz NOESY spectrum of rSP-C (FFI) in CDCl 3 /CD 3 OD (1 : 1, v/v) with the sample freshly prepared about 1 h before the experiment. All measurable H a -H N cross-peaks were integrated and com- pared with the integrals of the 800-MHz NOESY spectrum. The most intense signals were taken as 100% relative intensity, making the assumption that no significant exchange occurred in the given time frame within the center of the well-ordered a-helix. The relative intensities of the H a -H N cross-peaks of residues His8, Ala29 and Leu31 were about 50% of those recorded in the 800-MHz NOESY spectrum in CDCl 3 /CD 3 OH (1 : 1, v/v), and the intensities of residues 9–28 and 30 were 80% or higher. From these estimates of the relative intensities, hydrogen bonds for the structure calculations were assumed for His8 to Leu31. The amide protons of residues 1–7 and 32–34 could not be detected in the fully deuterated solvent. Structure of rSP-C (FFI) Using the empirical pattern-recognition approach [16], the combination of strong sequential d NN connectivities, obser- vation of a significant number of d aN (i,i+3), d ab (i,i+3), and d aN (i,i+4) connectivities, 3 J NHa coupling constants of less than 5 Hz for all non-Gly residues in the polypeptide segment Phe5, Val7–Leu30, and retarded amide proton exchange for residues 8–31 indicate that rSP-C (FFI) forms alonga-helix comprising approximately residues 5–34. For a more precise definition of the structure of rSP- C (FFI), a set of 203 intraresidual, 201 interresidual and seven ambiguous NOE-derived upper distances were used together with 23 / angles derived from 3 J(H N ,H a ) coupling constants as input data for a structure calculation using the program XPLOR [17]. In addition, we introduced 24 hydro- gen bonds derived from the slow exchange rate of the amide protons. No stereospecific assignments were used in the floating chirality simulated annealing protocol. For residues 28–34, we used only the set of resonances with the stronger intensities because identical relative NOEs were observed for the two species. For the structure calculations, we used a standard simulated annealing protocol designed for proteins [18]. After an initial energy minimization involving 50 optimiza- tion steps with conjugated gradients, a high temperature phase with 2000 K was simulated for 32.5 ps in which all upper limits built the active constraints. The following step was the first cooling phase from 2000 K to 1000 K in 25 ps with the dihedral angles as additional constraints. After the Table 3. Chemical shifts of rSP-C (FFI). Residue H N H a H b Others Gly1 8.23 3.73 Ile2 8.61 4.45 1.90 c1.66, 1.00; d1.23,0.95 Pro3 4.38 2.15, 1.99 c2.10; d3.95, 3.72 Phe4 8.08 4.49 3.18, 3.09 d7.17; e7.27; f7.19 Phe5 8.46 4.68 3.29 d7.28; e7.42; f7.36 Pro6 4.25 2.34, 2.00 c2.14; d3.65 Val7 7.62 3.69 2.28 c1.13, 1.01 His8 8.05 4.47 3.35, 3.29 d7.22; e8.74 Leu9 8.13 3.97 1.70, 1.60 c1.65; d1.03, 0.98 Lys10 7.95 3.91 2.03 c1.64, 1.50; d1.79; e2.93; f2.92 Arg11 7.89 3.94 2.02, 1.99 c1.70; d3.30, 3.24; e7.50; 1g7.18; 2g6.68 Leu12 7.82 4.01 1.69 c1.81; d0.94 Leu13 8.04 4.01 1.89 c1.71; d0.95 Ile14 7.77 3.64 2.08 c1.94, 1.20; d0.97, 0.93 Val15 7.67 3.52 2.40 c1.17, 1.01 Val16 8.01 3.51 2.33 c1.15, 1.00 Val17 8.03 3.52 2.33 c1.16, 1.02 Val18 8.15 3.57 2.32 c1.14, 1.03 Val19 8.36 3.57 2.32 c1.15, 1.00 Val20 8.35 3.49 2.31 c1.15, 1.00 Leu21 8.25 4.01 1.99, 1.93 c1.75; d1.02, 0.94 Ile22 8.30 3.60 2.16 c1.17; d0.99 Val23 8.25 3.53 2.40 c1.16, 0.99 Val24 8.59 3.57 2.42 c1.17, 1.02 Val25 8.28 3.71 2.39 c1.17, 1.03 Ile26 8.45 3.70 2.09 c1.95, 0.98; d1.17 Val27 8.93 3.59 2.23 c1.14, 1.02 Gly28 8.45 3.88, 3.77 8.42 3.86, 3.77 Ala29 8.21 4.09 1.62 8.19 4.08 1.62 Leu30 8.25 4.18 2.12, 2.03 c1.60; d0.99 Leu31 8.30 4.14 2.07 c1.60; d0.99, 0.96 8.28 4.13 2.07 g1.60; d0.99,0.96 Ile32 7.70 4.35 2.18 c1.64,1.51; d1.02 7.64 4.38 2.18 g1.61,1.55; d1.02 Gly33 7.95 4.09, 3.87 7.91 4.09, 3,85 Leu34 8.02 4.51 1.73 c1.78; d1.00 8.08 4.55 1.64 g1.74; d1.00 Ó FEBS 2004 Recombinant mutant of surfactant protein C (Eur. J. Biochem. 271) 2079 second cooling phase from 1000 K to 100 K in 10 ps, a second energy minimization was performed with 200 steps of conjugated gradients for each structure. The rmsd values and the distance and dihedral angle violations for the best 10 out of 60 structures are given in Table 4. The final structures shown in Figs 3 and 4 were determined by an additional refinement in vacuo including the experimental restraints, full charges, and a dielectric constant set to e ¼ 4r ij using a heating and cooling protocol. Figure 2 shows MOLMOL stereographic projections [19] of the heavy atoms of rSP-C (FFI). The structure of rSP- C (FFI) is a well-defined a-helix ranging from Phe5 to Leu34. Note that the distribution of the / and w angles indicates an a-helical structure up to Phe5, although residue 6 is a proline. Strong evidence for this comes from the unambiguously identified d aN (i,i+3) and d aN (i,i+4) cross- signals for Phe5 and Pro6 (cf. Fig. 1D). Discussion Comparison of rSP-C (FFI) with pSP-C The 34-residue peptide rSP-C (FFI) contains mainly apolar amino acids, i.e. 11 valines, seven leucines and five isoleucines, and forms a well-defined a-helix along residues 5–34 dissolved in CDCl 3 /CD 3 OH (1 : 1, v/v). The solution structure of pSP-C with 76% sequence identity (Table 1) in CDCl 3 /CD 3 OH/0.1 M HCl (32 : 64 : 5, v/v/v) was investigated by Johansson et al. [6]. To compare the structure of pSP-C with rSP-C (FFI), we show in Fig. 3 the differences in chemical shifts of the H N and H a signals of the corresponding residues. It can be seen that the chemical shifts for residues 10–29 are almost identical, with slightly greater variations at nonidentical amino acids. Only the N-terminal nine residues show significant chemical-shift differences mainly introduced by the sequence deviations at residues 4, 5 and 8. This difference at the N-terminus can also be seen when the two resulting structures shown in Fig. 4 are compared. Whereas the backbone of the central a-helix is very well defined in both structures, the N-terminal variability for the pSP-C is greater than that of rSP-C (FFI). This reflects the NOE- data-based fact that rSP-C (FFI) has a defined a-helix comprising residues 5–34, whereas for pSP-C an a-helical region at residues 7–34 has been reported [6]. However, the slow deuterium exchange for Leu9 and small distances d Na (i,i+3) and d ab (i,i+3) for Pro6 and Val7 suggest that even pSP-C adopts an a-helix starting with capping at residue Cys5 [8]. Substitution of acylated Cys with Phe in the polypeptide seems to influence the N-terminal a-helix formation including Pro6 in rSP-C (FFI). A possible explanation is the occurrence of aromatic interactions between Phe5 and His8 which may lead to stabilization of the extended a-helix. The structures of both pSP-C and rSP-C (FFI) were determined in chloroform/methanol, an environment in which hydropho- bic elements can move freely. Membranous environments such as the surfactant, however, have a directional effect on the hydrophobic palmitoylated Cys and Phe residues and on the charged Lys and Arg residues at positions 10 and 11, which probably results in slightly different N-terminal structures for the SP-C variants in their biologically active form. The central helix of pSP-C has a slightly lower rmsd value than that of rSP-C, probably because of the longer stretch of Val residues, leading to extremely stable stacking. In rSP- C (FFI) this homogeneous stacking is interrupted by Ile14 and Ile22, which may introduce slight mobility into the hydrophobic a-helix. However, this increased mobility still leaves the central helix quite rigid and does not seem to be important, as it was shown in mutation studies that SP-C retains its function even after the replacement of all valines by leucines or other a-helical amino-acid sequences [20,21]. Two sets of resonances Two sets of resonances were found for rSP-C (FFI) at the C-terminal residues Gly28, Leu30, Ile32, Gly33 and Leu34. Similar duplication of resonances has been reported for pSP-C, affecting residues Val27, Ala29, Leu30, Leu31 and Met32 [6]. In the case of pSP-C, the additional signals were explained by partial oxidation of Met32 to methionine sulfoxide. In the case of rSP-C (FFI), a different explan- ation must be found for the second set of resonances because Met32 is substituted by Ile32. The careful studies on pSP-C show a variation of 20–50% of the minor populated Table 4. Analysis of the 10 best calculated structures before and after the refinement. Before refinement After refinement E tot (kcalÆmol )1 ) 165.9 ± 12.7 (142.4.181.6) ) 265.4 ± 7.5 ()266.9 … )246.1) Distance violations Number > 0.5 A ˚ 1.1 ± 0.6 (0 … 2) 0 Sum (A ˚ ) > 0.1 A ˚ 0.68 ± 0.36 (0… 1.14) 0 Maximum (A ˚ ) 0.51 ± 0.18 (0 … 0.57) 0 Torsion-angle violations Number > 0.5 A ˚ 0 1.2 ± 0.9 (0 … 3) Sum (°) 0 13.0 ± 11.9 (0 … 34.9) Maximum (°) 0 9.1 ± 6.0 (0. 16.7) Rmsds (A ˚ ) Backbone (8–33) 0.59 ± 0.19 (0.36.0.99) 0.34 ± 0.06 (0.24.0.41) Heavy atoms (8–33) 1.05 ± 0.18 (0.90.1.44) 0.82 ± 0.13 (0.67.1.00) Backbone (18–28) 0.23 ± 0.08 (0.14.0.41) 0.07 ± 0.02 (0.04.0.10) Heavy atoms (18–28) 0.61 ± 0.08 (0.51.0.77) 0.45 ± 0.12 (0.35.0.68) 2080 B. Luy et al.(Eur. J. Biochem. 271) Ó FEBS 2004 set of resonances among samples prepared from different batches. We observed the same variation even in samples prepared from the same batch. A closer look at the acquired spectra indicates a dependence of the relative population of the signals on the overall SP-C concentration. As a consequence, we acquired a set of 2D NOESY spectra with identical mixing times but different concentrations of rSP- C (FFI) in CDCl 3 /CD 3 OH (1 : 1, v/v). The relative popu- lations of the two sets of resonances in these spectra with respect to the overall SP-C concentration are shown in Fig. 5. The dependence observed is a clear indication of intermolecular interaction. The relatively narrow linewidths of the observed signals led to the conclusion that oligomers of low order are present, probably monomeric and dimeric units, but trimeric or tetrameric units may also be possible; larger oligomers can be excluded because the linewidths would have to be significantly broader than observed. The linewidths of the two sets of resonances do not differ significantly, therefore the two oligomers must be of comparable size, and a monomer/tetramer equilibrium, for example, cannot explain the observed signals. The absence of further resonances implies that we are observing specific oligomers. Finally, chemical shifts of the Ha resonances are a clear indication that both oligomers are mainly a-helical and that their structures differ only slightly. The NMR data therefore point to the coexistence of a monomeric and homodimeric a-helical form of rSP-C (FFI). Fig. 2. Stereographic projection of the best 10 out of 60 structures of rSP-C (FFI). (A) Side view of the heavy atoms of the full-length peptide. (B) View from the bottom along residues 15–27 of the tightly packed a-helix. Ó FEBS 2004 Recombinant mutant of surfactant protein C (Eur. J. Biochem. 271) 2081 The literature on SP-C describes many oligomerization processes, most of which are either aggregates with mainly b-sheet-like or undetermined structure. Specific oligomeri- zation, i.e. dimerization, is only reported in a few cases: MS data provide evidence for dimeric SP-C [22,23], and chemical cross-linking studies also show mainly a specific dimer of mature SP-C (Fig. 8C in [24]). Yet unpublished high-resolution Fourier-transform ion-cyclotron-resonance MS, light-scattering and CD experiments reveal the exist- ence of an a-helical dimer at acidic pH ([25]; A. Seidl, G. Maccarone, N. Youhnovski, K. P. Schaefer and M. Przybylski, unpublished data). CNBr cleavage data even put the dimerization site near Met32 at the C-terminus, i.e. at the site at which the dual resonances are observed [23]. The coexistence of monomeric and homodimeric rSP- C(FFI) as derived from the NMR data therefore corres- ponds well to other reported experimental observations. Fibril formation The data from Fig. 5 could not be fitted to a simple monomer–dimer equilibrium model, but this is not surpri- sing considering that rSP-C (FFI), like pSP-C, shows a complete transition to b-sheet fibrils over time [8,26,27]. Immediately after rSP-C (FFI) is dissolved in chloroform/ methanol, short, fiber-like impurities of up to 1 mm length are observed in solution and on the glass walls of the NMR tube on visual inspection. This indication of already formed fibrils makes it necessary to describe rSP-C (FFI) by at least a three-state model with two a-helical states, probably monomer and dimer, and b-sheet fibrils that cannot be observed by high-resolution NMR because of their high molecular mass. A three-state model with monomeric, nonhelical and b-fibril states has already been presented [8]. Interestingly, the existence of an a-helical transition state (SPC # in [8]) was proposed in that publication, which would Fig. 4. Comparison of the 10 best structures of rSP-C (FFI) (left) and pSP-C (right). The backbone of the a-helix is shown. Clearly visible is the better defined secondary structure of rSP-C (FFI) near the N-terminus. Fig. 5. Concentration dependence of the relative integrals of the two sets of resonances observed at the C-terminus. Ratios are given for well- resolved residues Ile32, Gly33 and Leu34. Fig. 3. Differences in the chemical shifts of rSP-C (FFI) compared with pSP-C [6] for the H N (A) and the H a protons (B). Whereas residues 10–29 show almost identical chemical shifts, residues at the N-terminus and C-terminus differ more strongly. 2082 B. Luy et al.(Eur. J. Biochem. 271) Ó FEBS 2004 match the potential a-helical dimer found here. The interpretation of the potential dimeric state as the transition state to b-fibril formation would also match recent solid- state and liquid-state NMR results, which suggest that the smallest fibril diameter in b-amyloid fibrils is due to a parallel b-sheet dimer [28,29] and also that the minimum unit needed for fibril growth of a-synuclein is a dimer [30]. The disappearance of high-resolution NMR signals after several days at 10 °C in chloroform/methanol shows that the equilibrium state of rSP-C (FFI) is the b-sheet-like multimer. The a-helical states are therefore not equilibrated, and, in addition to the observed concentration dependence, a dependence on the age of the prepared samples can be predicted in the given solvent. It should be noted that neither rSP-C (FFI) nor pSP-C [8] show any transition to b-fibrils in dodecylphosphocholine micelles even after several weeks at room temperature. Sample handling and the situation in vivo SP-C is very difficult to handle. In general, basic conditions should be avoided and properties of the molecule depend strongly on the conditions for synthesis, the kind of purification used, and the aggregation states it was trans- ferred to. In this study, we relied on the elaborate procedure developed by Altana Pharma and only suspended the powder provided directly in chloroform/methanol. The NMR spectra yielded good results and therefore there appeared to be no need to change the method. Whether oligomerization can be avoided by different sample treat- ment remains to be proven. The local environment of the molecule also has a large impact on its behavior. Wild-type SP-C, like rSP-C (FFI), is solely monomeric at micromolar concentrations in pure organic solvents but has a strong tendency to aggregate in more hydrophilic environments. Relatively high concentra- tions can be obtained in dodecylphosphocholine micelles in which SP-C is stable for months in its a-helical form [8]. The surfactant consists of  1% by weight of SP-C [31]. The concentration of rSP-C (FFI) and extracted pSP-C in the NMR studies is therefore similar to the concentration of SP-C in its natural environment, although it shows a slow transition to b-sheet fibrils. However, whether the homo- dimer in chloroform/methanol is representative of the biologically active SP-C in the surfactant cannot be judged from the experiments presented. A hint may be gained from chemical cross-linking data on mature SP-C in cytosolic vesicles of A549 cells (Fig. 8C in [24]), which provide evidence of dimer formation during trafficking. Potential dimerization site The evidence suggests dimerization of SP-C, and it might be allowed to speculate on the potential dimerization site. The C-terminus of rSP-C (FFI) only contains apolar side chains and it can be assumed that it is situated at the hydrophobic palmitoyl chains of the surfactant phospholipids. In this environment, hydrogen-bonding interactions and strong hydrophobic associations are most likely to be the source of intermolecular attraction. A minor dimerization motif can be found in the C-terminal carboxylic group. Similar to the dimer formation of acetic acid, SP-C may form a dimer via hydrogen bonding (Fig. 6A). The acidic conditions of the NMR sample as well as the natural environment of SP-C would allow such a dimer formation. However, in the acidic NMR sample, relatively fast hydrogen exchange rates are expected which do not match the slow exchange regime observed for the two sets of resonances. Therefore, hydro- gen bonding of the carboxylic group is unlikely to be the cause of the observed dimerization, but we cannot exclude it. An alternative dimerization motif can be found in the strictly conserved C-terminal heptapeptide segment-span- ning residues Gly28 to Leu34: the heptapeptide segment of rSP-C (FFI), as well as all other SP-C variants, has an AxxxG pattern that perfectly matches the requirements for helix–helix association as described in [32]. Interestingly, the residues for which double resonances are observed are all within the strictly conserved heptapeptide segment with the AxxxG motif (Fig. 6B). Attempts to model two distinct structures for the two sets of resonances failed because of massive overlap of the side chain resonances in the region of interest in particular. However, as mentioned above, we can conclude from chemical-shift arguments that the two structures should be very similar and are a-helical in character. For the same reasons, it was impossible to obtain a structure of the potential dimer based on intermonomeric NOEs. A theoretical model based on the monomeric structure presented in this paper and computational dock- ing studies is derived in the following paper [33]. Conclusion We have derived by NMR spectroscopy the high-resolution 3D structure of rSP-C (FFI) dissolved in CDCl 3 /CD 3 OH (1 : 1, v/v). The lipophilic peptide forms a tight a-helix for residues 5–34 which is two residues longer than the a-helix Fig. 6. Potential dimerization motifs for rSP-C (FFI). (A) Hydrogen bonding at the C-terminal carboxy group may lead to dimerization. (B) Comparison of the amino-acid sequences of glycophorin A and rSP-C (FFI) shows a potential AxxxG dimerization motif similar to the van der Waals dimer of glycophorin A [33–35] at the strictly conserved heptapeptide segment where two sets of resonances are observed. Ó FEBS 2004 Recombinant mutant of surfactant protein C (Eur. J. Biochem. 271) 2083 observed in pSP-C, with 76% sequence identity in the same solvent. Both peptides show two sets of resonances for a number of C-terminal residues. Because of the lack of Met we can exclude oxidation to methionine sulfoxide as the cause of the second set of resonances for rSP-C (FFI), which was previously assumed in the case of pSP-C [6]. Studies on the concentration dependence of the dual resonances together with the narrow linewidth of the NMR signals suggest the coexistence of a monomeric and dimeric a-helical structure in the given solvent. There are two potential dimerization sites in SP-C: the C-terminal carboxylic group may form a dimer via hydrogen bonding; the C-terminal heptapeptide segment, which is conserved in all known SP-C species, contains an AxxxG motif that closely resembles the GxxxG helix–helix dimer motif of glycophorin A. Even though the latter dimerization motif is consistent with other experimental results and therefore highly likely, additional studies such as point mutations at the potential dimerization site are necessary to unambigu- ously determine the origin of the intermolecular interaction that leads to the second set of resonances. Acknowledgements C.G. gratefully acknowledges support from the DFG, the MPG, and the Fonds der Chemischen Industrie. B.L and A.D. were supported by the Fonds der Chemischen Industrie. B.L. is also supported by the DFG (Emmy Noether LU 835/1–1). We thank Bettina Elshorst for help with NDEE , Michael Nilges for help with the XPLOR protocols, and Michael Przybylski (University of Konstanz) for providing his results before publication. Special thanks go to Michael K. Gilson (CARB, Rockville, MD, USA) for many detailed scientific discussions. References 1. Goerke, J. (1998) Pulmonary surfactant: functions and molecular composition. Biochim. Biophys. Acta 1408, 79–89. 2. Ha ¨ fner, D., Germann, P.G. & Hauschke, D. (1998) Effects of rSP- C surfactant on oxygenation and histology in a rat-lung-lavage model of acute lung injury. Am.J.Respir.Crit.CareMed.158, 270–278. 3. Audrey, J.D., Alan, H.J., Ha ¨ fner, D. & Ikegami, M. (1998) Lung function in premature lambs and rabbits treated with a recom- binant SP-C surfactant. Am.J.Respir.Crit.CareMed.157, 553–559. 4. Spragg, R.G., Lewis, J., Wurst, W. & Rathgeb, F. (2000) Treat- ment of ARDS with rSP-C surfactant. Am.J.Respir.Crit.Care Med. 161, A47. 5. Walmrath, D., De Vaal, J.B., Bruining, H.A., Kilian, J.G., Papazian, L., Hohlfeld, J., Vogelmaier, C., Wurst, W., Schaffer, P., Rathgeb, F., Grimminger, F. & Seeger, W. (2000) Treatment of ARDS with recombinant SP-C (rSP-C) based synthetic surfactant. Am. J. Respir. Crit. Care Med. 161, A379. 6. Johansson, J., Szyperski, T., Curstedt, T. & Wu ¨ thrich, K. (1994) The NMR structure of the pulmonary surfactant-associated polypeptide Sp-C in an apolar solvent contains a valyl-rich alpha- helix. Biochemistry 33, 6015–6023. 7. Johansson,J.,Szyperski,T.&Wu ¨ thrich, K. (1995) Pulmonary surfactant-associated polypeptide SP-C in lipid micelles: CD stu- dies of intact SP-C and NMR secondary structure determination of depalmitoyl-SP-C (1–17). FEBS Lett. 362, 261–265. 8. Szyperski, T., Vandenbussche, G., Curstedt, T., Ruysschaert, J.M., Wu ¨ thrich, K. & Johansson, J. (1998) Pulmonary surfactant- associated polypeptide C in a mixed organic solvent transforms from a monomeric alpha-helical state into insoluble beta-sheet aggregates. Protein Sci. 7, 2533–2540. 9. Marion, D., Ikura, M., Tschudin, R. & Bax, A. (1989) Rapid recording of 2D NMR-spectra without phase cycling: application to the study of hydrogen exchange in proteins. J. Magn. Reson. 85, 393–399. 10. Wider, G., Hosur, R.V. & Wu ¨ thrich, K. (1983) Suppression of the solvent resonance in 2D NMR-spectra of proteins in H 2 Osolu- tion. J. Magn. Reson. 52, 130–135. 11. Prasch, T., Gro ¨ schke, P. & Glaser, S.J. (1998) SIAM, a novel NMR experiment for the determination of homonuclear coupling constants. Angew. Chem. Int. Ed. 37, 802–806. 12. Titman, J.J. & Keeler, J. (1990) Measurement of homonuclear coupling-constants from NMR correlation spectra. J. Magn. Reson. 89, 640–646. 13. Wu ¨ thrich, K. (1986) NMR of Proteins and Nucleic Acids. Wiley, New York. 14. Wishart, D.S., Sykes, B.D. & Richards, F.M. (1991) Relationship between nuclear-magnetic-resonance chemical-shift and protein secondary structure. J. Mol. Biol. 222, 311–333. 15. Wishart, D.S., Sykes, B.D. & Richards, F.M. (1992) The chemical- shift index: a fast and simple method for the assignment of protein secondary structure through NMR- spectroscopy. Biochemistry 31, 1647–1651. 16. Wu ¨ thrich, K., Billeter, M. & Braun, W. (1984) Polypeptide sec- ondary structure determination by nuclear magnetic-resonance observation of short proton–proton distances. J. Mol. Biol. 180, 715–740. 17. Bru ¨ nger, A.T. (1992) X-PLOR: a System for X-Ray Crystallo- graphy and NMR. Yale University Press, New Haven, CT. 18. Nilges, M. & O’Donoghue, I.S. (1998) Ambiguous NOEs and automated NOE assignment. Prog. NMR Spectrosc. 32, 107–139. 19. Koradi, R., Billeter, M. & Wu ¨ thrich, K. (1996) MOLMOL: a program for display and analysis of macromolecular structures. J. Mol. Graph. 14, 51–55. 20. Nilsson, G., Gustafsson, M., Vandenbussche, G., Veldhuizen, E., Griffiths, W.J., Sjovall, J., Haagsman, H.P., Ruysschaert, J.M.,Robertson,B.,Curstedt,T.&Johansson,J.(1998) Synthetic peptide-containing surfactants: evaluation of trans- membrane versus amphipathic helices and surfactant protein C poly-valyl to poly-leucyl substitution. Eur. J. Biochem. 255, 116–124. 21. Clercx, A., Vandenbussche, G., Curstedt, T., Johansson, J., Jornvall, H. & Ruysschaert, J.F. (1995) Structural and functional importance of the C-terminal part of the pulmonary surfactant polypeptide Sp-C. Eur. J. Biochem. 229, 465–472. 22. Mayer-Fligge, P., Volz, J., Kru ¨ ger,U.,Sturm,E.,Gernandt,W., Scha ¨ fer, K.P. & Przybylski, M. (1998) Synthesis and structural characterization of human-identical lung surfactant SP-C protein. J. Pept. Sci. 4, 355–363. 23. Przybylski, M., Maier, C., Ha ¨ gele, K., Bauer, E., Hannappel, E., Nave, R., Melchers, K., Kru ¨ ger, U. & Scha ¨ fer, K.P. (1994) Pri- mary structure elucidation, surfactant function and specific for- mation of supramolecular dimer structures of lung surfactant associated SP-C proteins. In Hodges, R.S. & Smith, J.A., eds. Peptides, Chemistry, Structure and Biology, pp. 338–340. Escom Science Publishers, Leiden. 24. Wang, W.J., Russo, S.J., Mulugeta, S. & Beers, M.F. (2002) Biosynthesis of surfactant protein C (SP-C). J. Biol. Chem. 277, 19929–19937. 25. Seidl, A. (2003) Massenspektrometrische Analyse: Chemische Modifizierung und Synthese Von Lipoproteinen.PhDThesis, Gorre-Verlag, Konstanz, Germany. 26. Johansson, J. (2001) Membrane properties and amyloid fibril formation of lung surfactant protein. Biochem. Soc. Trans. 29, 601–606. 2084 B. Luy et al.(Eur. J. Biochem. 271) Ó FEBS 2004 27. Johansson, J. (2003) Molecular determinants for amyloid fibril formation: lessons from lung surfactant protein C. Swiss Medical Weekly 133, 275–282. 28. Petkova, A.T., Ishii, Y., Balbach, J.J., Antzutkin, O.N., Leapman, R.D., Delaglio, F. & Tycko, R. (2002) A structural model for Alzheimer’s beta-amyloid fibrils based on experimental con- straints from solid state NMR. Proc. Natl Acad. Sci. USA 99, 16742–16747. 29. Tycko, R. (2003) Applications of solid state NMR to the struc- tural characterization of amyloid fibrils: methods and results. Prog. Nucl. Magn. Reson. Spectrosc. 42, 53–68. 30. Fernandez, C.O., Hoyer, W., Zweckstetter, M., Jares-Erijman, E.A., Subramaniam, V., Griesinger, C., Jovin, T.M. (2004) NMR of alpha-synuclein complexes with polyamines elucidates the mechanism and kinetics of induced aggregation. EMBO J. in press. 31. Kru ¨ ger,P.,Schalke,M.,Wang,Z.,Notter,R.H.,Dluhy,R.A.& Lo ¨ sche, M. (1999) Effect of hydrophobic surfactant peptides SP-B and SP-C on binary phospholipid monolayers. I. fluorescence and dark-field microscopy. Biophys. J. 77, 903–914. 32. Eilers, M., Patel, A.B., Liu, W. & Smith, S.O. (2002) Comparison of helix interactions in membrane and soluble alpha-bundle pro- teins. Biophys. J. 82, 2720–2736. 33. Kairys, V., Gilson, M.K., Luy, B. (2004) Structural model for an AxxxG-mediated dimer of surfactant-associated protein C. Eur. J. Biochem. 271, 2086–2092. 34. MacKenzie, K.R., Prestegard, J.H. & Engelman, D.M. (1997) A transmembrane helix dimer: structure and implications. Science 276, 131–133. 35. Smith, S.O., Song, D., Shekar, S., Groesbeek, M., Ziliox, M. & Aimoto, S. (2001) Structure of the transmembrane dimer inter- face of glycophorin A in membrane bilayers. Biochemistry 40, 6553–6558. Ó FEBS 2004 Recombinant mutant of surfactant protein C (Eur. J. Biochem. 271) 2085 . Structure and potential C- terminal dimerization of a recombinant mutant of surfactant-associated protein C in chloroform/methanol Burkhard Luy 1 , Alexander. electrospray ionization MS data, chemical cross-linking studies, and CNBr cleavage data. Keywords: dimerization; NMR spectroscopy; surfactant; surfactant protein C (SP -C) . Surfactant-associated