Two Cys residues essential for von Willebrand factor multimer assembly in the Golgi

Abstract

Von Willebrand factor (VWF) dimerizes through C-terminal CK domains, and VWF dimers assemble into multimers in the Golgi by forming intersubunit disulfide bonds between D3 domains. This unusual oxidoreductase reaction requires the VWF propeptide (domains D1D2), which acts as an endogenous pH-dependent chaperone. The cysteines involved in multimer assembly were characterized by using a VWF construct that encodes the N-terminal D1D2D′D3 domains. Modification with thiol-specific reagents demonstrated that secreted D′D3 monomer contained reduced Cys, whereas D′D3 dimer and propeptide did not. Reduced Cys in the D′D3 monomer were alkylated with N-ethylmaleimide and analyzed by mass spectrometry. All 52 Cys within the D′D3 region were observed, and only Cys¹⁰⁹⁹ and Cys¹¹⁴² were modified by N-ethylmaleimide. When introduced into the D1D2D′D3 construct, the mutation C1099A or C1142A markedly impaired the formation of D′D3 dimers, and the double mutation prevented dimerization. In full-length VWF, the mutations C1099A and C1099A/C1142A prevented multimer assembly; the mutation C1142A allowed the formation of almost exclusively dimers, with few tetramers and no multimers larger than hexamers. Therefore, Cys¹⁰⁹⁹ and Cys¹¹⁴² are essential for the oxidoreductase mechanism of VWF multimerization. Cys¹¹⁴² is reported to form a Cys¹¹⁴²–Cys¹¹⁴² intersubunit bond, suggesting that Cys¹⁰⁹⁹ also participates in a Cys¹⁰⁹⁹–Cys¹⁰⁹⁹ disulfide bond between D3 domains. This arrangement of intersubunit disulfide bonds implies that the dimeric N-terminal D′D3 domains of VWF subunits align in a parallel orientation within VWF multimers.

Von Willebrand factor (VWF) is an essential hemostatic protein that initiates the formation of a platelet plug by binding to connective tissue exposed at sites of vascular injury and to receptors on the surface of platelets. These functions depend on binding sites for many ligands within the repeated multidomain structure of the VWF subunit: D1-D2-D′-D3-A1-A2-A3-D4-B1-B2-B3-C1-C2-CK. In addition, normal function requires the assembly of VWF into large multimers, and defects in VWF multimerization cause the inherited bleeding disorder, von Willebrand disease (VWD) type 2A (1).

VWF assembly begins in the endoplasmic reticulum (ER) of endothelial cells and megakaryocytes where VWF is synthesized as a precursor that consists of a 22-aa signal peptide, a 741-aa propeptide (domains D1D2), and a 2,050-aa mature subunit. Also in the ER, proVWF monomers form homodimers “tail-to-tail” through interchain disulfide bonds between their CK domains (2–5). The proVWF dimers are transported to the Golgi where they assemble into large multimers “head-to-head” through interchain disulfide bonds between two D3 domains, and the propeptide is cleaved by furin (2, 6).

VWF multimerization in the Golgi involves a unique oxidoreductase mechanism that requires the VWF propeptide, which functions as an endogenous chaperone to promote disulfide-bond formation specifically under acidic conditions (2, 7–9). The reaction appears to proceed through a transient, disulfide-linked intermediate between the propeptide and the D3 region of VWF that forms in the ER and resolves in the Golgi to yield disulfide-linked VWF multimers (10). Partially purified proVWF dimers will form proVWF multimers in vitro at acidic pH, and the reaction is prevented by N-ethylmaleimide (NEM) (11), which indicates that reduced cysteines are required.

VWF multimers purified from blood contain little or no reduced cysteine, suggesting that successful multimer assembly engages all VWF cysteine residues in intrasubunit or intersubunit disulfide bonds (12, 13). However, the subunits at the ends of VWF multimer have not formed intersubunit disulfide bonds and are likely to have reduced Cys residues within their D3 domains. These Cys may have transiently paired with the VWF propeptide (10) and may be key participants in the oxidoreductase mechanism of VWF multimer assembly.

Using a model system that simplifies the analysis of VWF multimer assembly (10, 14), we now have shown that Cys¹⁰⁹⁹ and Cys¹¹⁴² are oxidized when VWF multimerization succeeds and reduced when it does not. In addition, mutation of these Cys residues prevents the assembly of VWF multimers. Thus, the integrity of VWF multimers may depend on just two disulfide bonds between the D3 domains of VWF subunits.

Results

Secreted D′D3 Monomer Contains Reduced Cys.

Reduced thiols, presumably in Cys residues, are essential for VWF multimerization (11), but whether the propeptide, D′D3 region, or another factor contains mechanistically significant reduced Cys is unknown. The N-terminal D1D2D′D3 region of the proVWF subunit reproduces essential features of multimer assembly, forming intersubunit disulfide bonds between D3 domains in the Golgi apparatus when expressed in mammalian cells (10), and this model system was used to identify reduced Cys that participate in multimer assembly.

VWF FLAG-D1D2D′D3-c-Myc was expressed in BHK cells, and secreted proteins were reacted with biotin-containing maleimide reagents [supporting information (SI) Fig. 5] to alkylate free thiols (15, 16). Species with alkylated Cys were adsorbed on neutravidin-agarose and identified by Western blotting (Fig. 1). FLAG-tagged bacterial alkaline phosphatase (FLAG-BAP) has two intrachain disulfide bonds and no reduced Cys (17), and FLAG-BAP treated with biotin-maleimide reagents did not bind to neutravidin agarose (Fig. 1B), confirming the specificity of the maleimide reagents for reduced Cys.

Fig. 1.

Secreted D′D3-c-Myc monomer contains reduced Cys, but D′D3-c-Myc dimer and FLAG-D1D2 do not. (A) Media from BHK-fur4-D1D2D′D3-FLAGNT-c-MycCT cells were reacted with 850 μM biotin-BMCC (Pierce, Rockford, IL), biotin-maleimide (Sigma, St. Louis, MO), or maleimide PEO₂-biotin (Pierce) for 1 h at room temperature, followed by addition of 40 mM NEM. Biotinylated and nonbiotinylated D′D3-c-Myc and FLAG-D1D2 were isolated by immunoprecipitation, separated on neutravidin-agarose, and detected by SDS/PAGE and Western blotting with anti-VWF (Upper) or anti-FLAG (Lower) as described in Materials and Methods. The lanes are: total immunoprecipitate (I), fraction bound to neutravidin-agarose (B), and unbound supernatant (S). (B) Bacterial alkaline phosphatase (FLAG-BAP) (2 μg/ml) in media from BHK-fur4-D1D2D′D3-FLAGNT-c-MycCT cells was reacted with biotin-BMCC or biotin-maleimide and analyzed by using anti-FLAG as described for A. Similar results were obtained with maleimide PEO₂-biotin (data not shown).

These BHK cells secrete VWF propeptide (FLAG-D1D2) and a mixture of D′D3-c-Myc monomer and dimer (10). Most of the D′D3-c-Myc dimers and FLAG-D1D2 did not bind to neutravidin-agarose (Fig. 1A), indicating that the majority of these proteins contain only oxidized (disulfide-bonded) Cys. In contrast, all of the D′D3 monomer bound to neutravidin-agarose (Fig. 1A), suggesting that all secreted D′D3 monomer contained reduced Cys that was biotinylated. None of these proteins bound to neutravidin-agarose when alkylated with NEM (data not shown). Similar results were obtained for BHK cells transiently expressing FLAG-propeptide on one plasmid and D′D3-c-Myc on a separate plasmid (data not shown). Thus, D′D3 monomer contains reduced Cys that are not present in D′D3 dimer, and these residues may normally form intersubunit disulfide bonds.

Only Two Cys Are Reduced in D′D3 Monomer.

The oxidation state of Cys residues in secreted D′D3 monomer was established by differential alkylation, proteolytic digestion, and nano-LC-linear quadrupole ion trap Fourier transform cyclotron resonance mass spectrometry (nano-LC-FTMS) (18). The D′D3 monomer was treated with NEM in conditioned medium, to alkylate any reduced Cys and prevent disulfide rearrangement. The modified product was purified and then reduced completely and alkylated with 4-vinylpyridine (4VP) or iodoacetamide (IAA). N-linked oligosaccharides at Asn⁸⁵⁷, Asn¹¹⁴⁷, and Asn¹²³¹ (19) were removed with protein N-glycosidase F to simplify MS analysis. Samples were digested with various proteases including trypsin, thermolysin, Asp-N, pepsin, Arg-C, chymotrypsin, and Glu-C. Chymotrypsin sometimes cleaved after 4VP-modified Cys, which was useful for identifying Cys¹²²⁵ and Cys¹²²⁷. Double digestion with Glu-C/Asp-N was performed to isolate a peptide containing Cys¹¹⁵⁷ and Cys¹¹⁶⁵ (Table 1).

Table 1.

Cys-containing peptides in protease digestions of D′D3 by using nano-LC-FTMS

Sequence	Start–End	Modification	Theoretical m/z	Observed m/z	Enzyme
SCRPPMVKL	766–774	1 4VP	568.3100	568.3116	Pepsin
LVCPADNLR	774–782	1 IAA	529.2772	529.2772	Trypsin
DNLRAEGLECTKTCQNY	779–795	2 4VP	1,084.4992	1,084.5011	Asp-n
CMSMGCVSGCLCPPGMVR	799–816	4 4VP	1,125.9860	1,125.9957	Arg-c
CVSGCLCPPGM	804–814	3 4VP	691.3004	691.3028	Pepsin
CVALER	821–826	1 4VP	398.2131	398.2129	Trypsin
CPCFHQGK	827–834	2 IAA	517.2214	517.2219	Trypsin
IGCNTCVCR	844–852	3 IAA	570.2420	570.2420	Trypsin
WNCTDHVCDATCS	856–868	3 IAA	813.2931	813.2961	Trypsin
VCDATCSTI	862–870	2 4VP	561.7522	561.7537	Chymotrypsin
DATCSTIGMAHYLTF	864–878	1 4VP	868.3952	868.3973	Asp-n
DGLKYLFPGECQYVLVQ	879–895	1 4VP	1,039.0274	1,039.0302	Asp-n
CQYVLVQDYCGSNPGTFR	889–906	2 4VP	1,130.5200	1,130.5277	Arg-c
ILVGNKGCSHPSVKCKKR	907–924	2 4VP, 4 acetyl-K	1,166.6248	1,166.6248	Trypsin
GCSHPSVK	913–920	1 4VP	460.2270	460.3322	Trypsin
KQTYQEKVCG	985–994	1 4VP	644.8220	644.8241	Pepsin
LCGNFDGIQNND	995–1006	1 4VP	707.8070	707.8094	Thermolysin
DFGNSWKVSSQCA	1020–1032	1 4VP	767.3438	767.3448	Asp-n
VSSQCADTR	1027–1035	1 4VP	536.2486	536.2485	Trypsin
VPLDSSPATCHNNIMK	1037–1052	1 4VP	916.4457	916.4449	Trypsin
DSSCRILTS	1057–1065	1 4VP	1,086.5254	1,086.5264	Asp-n
ILTSDVFQDCNK	1062–1073	1 IAA	720.3460	720.3469	Trypsin
DVCIY	1082–1086	1 4VP	717.3282	717.3298	Asp-n
SIGDCACF	1093–1100	2 4VP	513.2149	513.2166	Pepsin
DTCSCESIGDCACF	1087–1100	3 4VP, 1 NEM	947.3515	947.3586	Chymotrypsin
DCACFC	1096–1101	3 4VP, 1 NEM	996.3412	996.3452	Asp-n
YAHVCAQHGKVVT	1107–1119	1 4VP	759.3883	759.3909	Pepsin
TATLCPQSCEER	1122–1133	2 IAA	726.3169	726.3170	Trypsin
ENGYECEWR	1137–1145	1 NEM	655.7596	655.7599	Trypsin
ENGYECEWR	1137–1145	1 4VP	645.7646	645.7658	Arg-c
WRYNSC	1144–1149	1 4VP	467.2060	467.2081	Pepsin
YNSCAPACQVTCQHPEPL	1146–1163	3 IAA	1,066.4540	1,066.4550	Trypsin
VTCQHPEPLAC	1155–1165	2 4VP	704.3316	704.3361	Glu-c/asp-n
ACPVQCVEGCH	1164–1174	3 4VP	730.8178	730.8217	Chymotrypsin
CPPGKILDELL	1177–1187	1 4VP	651.8606	651.8631	Chymotrypsin
DELLQTCV	1184–1191	1 4VP	1,025.4978	1,025.5004	Asp-n
DELLQTCVDPEDCPVC	1184–1199	3 4VP	1,047.4551	1,047.4573	Asp-n
NPSDPEHCQIC	1215–1225	2 4VP	726.8060	726.8082	Chymotrypsin
HCDVVNL	1226–1232	1 4VP	904.4351	904.4401	Chymotrypsin
DVVNLTCEACQ	1228–1238	2 4VP	702.8186	702.8207	Asp-n

All sequences were confirmed from manual interpretation of the MS/MS spectra. Observed m/z were obtained by selected ion extraction by using Xcalibur software to search for m/z values of singly, doubly, or triply charged species.

Cys modified by NEM, 4VP, and IAA have distinct masses, and the specific modifications of all Cys except for Cys⁹²¹ were identified in these digests by nano-LC-FTMS (Table 1). Several Lys residues flank Cys⁹²¹, which prevented its identification in tryptic digests; 4VP-modified Cys⁹²¹ was observed in a sample reacted with sulfosuccinimidyl acetate to acetylate Lys residues and prevent their recognition by trypsin. Thus, all 52 Cys in D′D3 were characterized: 50 were alkylated only by 4VP or IAA and therefore involved in intrachain disulfide bonds; Cys¹⁰⁹⁹ and Cys¹¹⁴² were alkylated by NEM, indicating that they are reduced in D′D3 monomer and potentially form intersubunit disulfide bonds in VWF multimers.

Cys¹⁰⁹⁹.

A singly charged ion with m/z = 996.3462 was observed in an Asp-N digest, corresponding to the predicted m/z = 996.3418 for the peptide ¹⁰⁹⁶DCACFC¹¹⁰¹ with two 4VP modifications and one NEM modification (Fig. 2A). The MS/MS spectrum for this ion showed that Cys¹⁰⁹⁷ was modified by 4VP because of the presence of the b₂ ion at the expected m/z = 324.09 and the b₃ ion at m/z = 395.18 (Fig. 2B). Similarly, the y₂ ion at m/z = 374.15 indicated that Cys¹¹⁰¹ was 4VP-modified. However, Cys¹⁰⁹⁹ was modified by NEM, as shown by the y₃ ion at m/z = 602.18, the y₄ ion at m/z = 673.18, and the y₅ ion at m/z = 881.27. NEM modification on Cys¹⁰⁹⁹ was confirmed by the b₄ and b₅ ions at m/z = 623.09 and 770.18, respectively (Fig. 2B). This peptide also was observed as a doubly charged ion with m/z = 498.6753 (data not shown). Further evidence that Cys¹⁰⁹⁹ was NEM-modified was provided from a doubly charged (m/z = 947.3586) chymotryptic peptide (¹⁰⁸⁷DTCSCESIGDCACF¹¹⁰⁰), whose MS/MS spectrum identified Cys¹⁰⁸⁹, Cys¹⁰⁹¹, and Cys¹⁰⁹⁷ as 4VP-modified and Cys¹⁰⁹⁹ as NEM-modified (data not shown). Therefore, Cys¹⁰⁹⁹ is reduced in the D′D3 monomer.

Fig. 2.

Nano-LC-FTMS of D′D3 monomer peptides containing NEM-modified Cys¹⁰⁹⁹ and Cys¹¹⁴². (A) MS spectrum of the DC[4VP]AC^[NEM]FC[4VP] singly charged ion at m/z = 996.3452. The peptide corresponds to Asp¹⁰⁹⁶–Cys¹¹⁰¹ of VWF. (B) MS/MS spectrum acquired from the DC[4VP]AC^[NEM]FC[4VP] ion. The asterisk indicates a DCACFC ion with neutral loss of one 4VP modification. (C) MS spectrum of the ENGYEC^[NEM]EWR doubly charged ion at m/z = 655.7599, corresponding to Glu¹¹³⁷-Arg¹¹⁴⁵ of VWF. (D) MS/MS spectrum acquired from the ENGYEC[NEM]EWR ion. Complete lists of assigned ions are given in SI Tables 2 and 3.

Cys¹¹⁴².

A doubly charged ion with m/z = 655.7599 was detected in a trypsin digest corresponding to the predicted m/z = 655.7596 for the peptide ¹¹³⁷ENGYEC[NEM]EWR¹¹⁴⁵ (Fig. 2C). The identity of the peptide was confirmed by its MS/MS spectrum (Fig. 2D), which contains the expected y₄ (¹¹⁴²C[NEM]EWR¹¹⁴⁵), y₅ (¹¹⁴¹EC[NEM]EWR¹¹⁴⁵), y6 (¹¹⁴⁰YEC[NEM]EWR¹¹⁴⁵), and y7 (¹¹³⁹GYEC[NEM]EWR¹¹⁴⁵) ions at m/z = 718.36, 847.36, 1010.45, and 1067.45, respectively. The b₇ ion also was present at the predicted m/z = 950.73 for NEM modification of Cys¹¹⁴². These data show that Cys¹¹⁴² is reduced in the D′D3 monomer, which is consistent with the identification of a Cys¹¹⁴²-Cys¹¹⁴² intersubunit disulfide bond in plasma VWF multimers by using direct methods (20).

The corresponding 4VP-modified doubly charged ion has a predicted m/z = 645.7646, which corresponds to the minor tryptic peptide, ¹¹³⁷ENGYEC[4VP]EWR¹¹⁴⁵ (SI Fig. 6), that was observed at m/z = 645.7658 (Table 1). Detection of both NEM- and 4VP-modified peptides confirms that the nano-LC-FTMS method identifies both species, when present. The occurrence of some 4VP-modified Cys¹¹⁴² may reflect incomplete modification of reduced Cys when NEM is added directly to conditioned medium.

Cys¹⁰⁹⁹ and Cys¹¹⁴² Are Required for VWF Multimer Assembly.

When BHK-fur4 cells were transfected with FLAG-D1D2D′D3-c-Myc containing the mutations C1099A or C1142A, D′D3 monomer was secreted efficiently, and little D′D3 dimer was detected (Fig. 3A). Cells expressing the double mutant (C1009A/C1142A) secreted only D′D3 monomer. Thus, Cys¹⁰⁹⁹ and Cys¹¹⁴² are essential for intersubunit disulfide-bond formation between D′D3 dimers. Reaction with biotin-maleimide reagents as described in Fig. 1 showed that all of these mutant D′D3 monomers contained reduced Cys residues (data not shown).

Fig. 3.

Mutation of Cys¹⁰⁹⁹ or Cys¹¹⁴² impairs intersubunit disulfide bond formation. (A) BHK cells were transiently transfected with FLAG-D1D2D′D3-c-Myc (WT, lane 1), FLAG-D1D2D′D3-c-Myc (C1099A) (lane 2), FLAG-D1D2D′D3-c-Myc (C1142A) (lane 3), or FLAG-D1D2D′D3-c-Myc (C1099A/C1142A) (lane 4). Conditioned media were analyzed by SDS/PAGE under nonreducing conditions and Western blotting with antI–VWF. (B) BHK cells were transiently transfected with full-length VWF (WT, lane 1), VWF (C1099A) (lane 2), VWF (C1142A) (lane 3), or VWF (C1099A/C1142A) (lane 4). VWF multimers in conditioned media were analyzed by SDS-agarose gel electrophoresis and Western blotting with anti-VWF (32).

The same mutations were introduced into full-length VWF for expression in BHK cells (Fig. 3B). VWF C1099A and VWF C1099A/C1142A did not form multimers and were secreted only as dimers. VWF C1142A was secreted mainly as dimers but did form some tetramers and a trace of hexamers, in agreement with our previous studies (20). Thus, both of the Cys residues that are reduced in monomeric D′D3 appear to be necessary for the efficient assembly of full-length VWF multimers.

Discussion

When VWF subunits assemble into multimers, intersubunit disulfide bonds form between D3 domains except possibly at the extreme ends of the multimer, where Cys residues that normally make intersubunit bonds may remain unpaired. These cysteines were identified in a truncated VWF construct that permits the preparative isolation of D′D3 monomers (Fig. 1), which may correspond to the terminal subunits of VWF multimers.

The D′D3 region of VWF contains 52 Cys residues, but a relatively small number of reduced Cys was expected. Proteolytic digestion of plasma VWF indicated that intersubunit disulfide bonds must be confined to the 28 cysteines in the D3 domain from Cys¹⁰⁴⁶ through Cys¹²³⁷; 8 of these are reported to form intrachain disulfide bonds (12), and we had shown previously that Cys¹¹⁴² makes a Cys¹¹⁴²–Cys¹¹⁴² intersubunit bond (20), leaving 17 Cys that might be reduced in monomeric D′D3. In fact, only two reduced Cys were found, Cys¹⁰⁹⁹ and Cys¹¹⁴² (Fig. 2). Interestingly, mutation of either Cys¹⁰⁹⁹ or Cys¹¹⁴² prevented multimer assembly (Fig. 3), and a simple explanation for this finding would be that Cys¹⁰⁹⁹ pairs directly with Cys¹¹⁴² on another subunit. However, our previous direct isolation of a Cys¹¹⁴²–Cys¹¹⁴² intersubunit bond suggests instead that VWF multimers probably contain just two disulfide bonds, Cys¹⁰⁹⁹–Cys¹⁰⁹⁹ and Cys¹¹⁴²–Cys¹¹⁴², between the N termini of the subunits (Fig. 4). This arrangement implies that the N termini of VWF subunits align in a parallel orientation during multimer assembly and within finished multimers.

Fig. 4.

Disulfide bond structure of the D′D3 region of VWF. The 52 Cys residues in the D′D3 region of VWF (Ser⁷⁴¹-Gly¹²⁴¹) are shown as numbered ovals. Locations of N-linked oligosaccharides (CHO) are shown. Intrasubunit disulfide bonds that have been characterized (12) are shaded black [based on protein sequencing of plasma VWF (12)] or hatched (based on ref.12 and Table 1). Intersubunit disulfide bonds are indicated in orange. At least one of Cys¹²²², Cys¹²²⁵, and Cys¹²²⁷ (yellow) makes at least one intrachain disulfide bond with a subset of Cys¹⁰⁹⁷, Cys¹¹⁰¹, Cys¹¹⁵⁷, Cys¹¹⁷³, Cys¹¹⁷⁷, and Cys¹¹⁹⁰ (blue).

A mutation affecting one of these Cys residues has been observed in an unusual form of VWD type 2A, initially named “type IIC Miami”, that is characterized by dominant inheritance and a markedly increased plasma concentration of VWF with very small multimers (21). In two families, the causative mutation was C1099Y (22). Thus, mutation of Cys¹⁰⁹⁹ prevents multimer assembly (Fig. 3) (22) but is compatible with efficient VWF secretion and normal intravascular survival in vivo (22).

The N-terminal region of VWF (domains D1D2D′D3) profoundly affects the pattern of intersubunit disulfide bonds in recombinant VWF. For example, VWF residues Arg¹²⁰⁴–Asn¹⁴⁷⁹ contains seven Cys residues: Cys¹²²², Cys¹²²⁵, Cys¹²²⁷, Cys¹²³⁴, and Cys¹²³⁷ from the C terminus of the D3 domain, and Cys¹²⁷² and Cys¹⁴⁵⁸ from the adjacent domain A1. Plasma VWF contains intrachain disulfide bonds linking Cys¹²³⁴–Cys¹²³⁷ and Cys¹²⁷²–Cys¹⁴⁵⁸, and the pairing of the remaining Cys residues is unknown (12). When the Arg¹²⁰⁴–Asn¹⁴⁷⁹ fragment is expressed in mammalian cells with a signal peptide, it is secreted efficiently as a disulfide-linked homodimer (23) with interchain disulfide bonds involving one or more of Cys¹²²², Cys¹²²⁵, and Cys¹²²⁷; mutation of any one of these three Cys prevents dimerization (24). However, larger constructs containing the rest of domains D′D3 (e.g., D′D3A1) do not dimerize unless the propeptide (D1D2) also is included (7, 8, 10). Furthermore, Cys¹²²², Cys¹²²⁵, and Cys¹²²⁷ make only intrachain disulfide bonds in monomeric D′D3 (Table 1 and SI Fig. 7), and mutation of all three of them does not impair multimer assembly by full-length VWF (20). Therefore, Cys¹²²², Cys¹²²⁵, and Cys¹²²⁷ probably do not form interchain disulfide bonds in native VWF but can do so when the D1D2D′D3 regions, which are required for normal pH-dependent multimerization in the Golgi, are absent.

This conclusion is supported by structural studies of plasma VWF. Trypsin digestion yields a major ≈116-kDa product that contains disulfide-linked ≈50-kDa and ≈13-kDa polypeptides (25). Some reports describe this ≈116-kDa species as a homodimer of ≈50-kDa polypeptides, but the data included in these reports confirm the presence of a disulfide-linked ≈10- to 13-kDa component (26, 27). The ≈50-kDa fragment consists of Val¹²¹²–Lys¹⁴⁹¹ (28). The ≈13-kDa fragment begins with Gln¹⁰⁵³ (29) and could extend to Arg¹¹³⁶ (9.4 kDa) or Arg¹¹⁴⁵ (10.5 kDa) but probably not to the next trypsin site at Lys¹¹⁸¹ (≈18 kDa). Therefore, Cys¹⁰⁹⁹–Cys¹⁰⁹⁹ and/or Cys¹¹⁴²–Cys¹¹⁴² disulfide bonds between two ≈13-kDa fragments would produce a ≈116-kDa tetramer with the composition (≈13 kDa + ≈50 kDa)₂. This structure implies that at least one of Cys¹²²², Cys¹²²⁵, and Cys¹²²⁷ forms an intrachain disulfide bond with Cys residues in the D3 domain whose disulfide pairing is not yet known (Fig. 4). Candidates include Cys¹⁰⁹⁷ or Cys¹¹⁰¹, which are within the ≈13-kDa component of the 116-kDa trypsin-digestion product.

How Cys¹⁰⁹⁹ and Cys¹¹⁴² actually form intersubunit disulfide bonds remains to be determined. A transient disulfide-linked complex between the VWF propeptide (D1D2) and D′D3 has been identified in the ER during VWF biosynthesis, and this complex appears to rearrange in the Golgi to yield VWF multimers (10). Cys¹⁰⁹⁹ or Cys¹¹⁴² might reasonably participate in this oxidoreductase reaction intermediate, in which case, rearrangement would yield intersubunit disulfide bonds between D3 domains with transfer of electrons to yield reduced Cys residues on D1D2. However, secreted D1D2 does not contain reduced Cys residues (Fig. 1), which suggests that unidentified electron acceptors in the Golgi apparatus can rapidly oxidize them. Alternatively, D1D2 may not contain Cys residues that participate directly in VWF multimer assembly. Instead, mechanistically important thiols may be confined to the D3 domains, and D1D2 may promote multimer assembly mainly by approximating them.

Materials and Methods

Plasmid Constructs and Protein Expression.

Plasmids encoding human VWF fragments were derived from pSVHVWF1.1 (30): pSVH-D3-FLAGNT/c-MycCT encodes VWF Met¹–Gly¹²⁴¹ with a FLAG tag (DYKDDDDK) after the signal peptide between Cys²² and Ala²³ and a c-Myc tag (EQKLISEEDL) after Gly¹²⁴¹; pSVH-propeptide-FLAGNT is similar but has a stop codon after Arg⁷⁶³; pSVH-D′D3Δpro-c-Myc encodes VWF Met¹–Ala³³, followed by Ala⁷⁶⁴–Gly¹²⁴¹, c-Myc tag, and stop codon; pSVH-D′D3Δpro encodes VWF Met¹–Ala³³, followed by Ala⁷⁶⁴–Gly¹²⁴¹ (10).

Plasmids pSVH-D3-FLAGNT/c-MycCT (C1099A), pSVH-D3-FLAGNT/c-MycCT (C1142A), and pSVH-D3-FLAGNT/c-MycCT (C1099A/C1142A) were made by using a QuikChange XL kit (Stratagene, La Jolla, CA) to mutate plasmid pSVH-D3-FLAGNT/c-MycCT. A similar strategy was used to make mutations in full-length VWF to yield plasmids pSVH-VWF (C1099A), pSVH-VWF (C1142A), and pSVH-VWF (C1099A/C1142A).

BHK cells were grown in Dulbecco's modified Eagle's medium (Invitrogen, Carlsbad, CA) plus 10% heat-inactivated FBS and 2 mM glutamine. BHK cells were transiently transfected by using Lipofectamine PLUS in Opti-MEM (Invitrogen). Stably transfected BHK cell lines were prepared as described for BHK-fur4-D1D2D′D3-FLAGNT-c-MycCT cells (10, 31).

Detection of Reduced Cys.

After reaction with biotin-maleimide reagents (Fig. 1 and SI Fig. 5), media from BHK-fur4-D1D2D′D3-FLAGNT-c-MycCT cells (10) were incubated overnight at 4°C with 1:100 volume of anti-c-Myc antibody or anti-FLAG M2 antibody (Sigma), respectively, and protein G-Sepharose (1:10 volume) (GE Healthcare, Piscataway, NJ). Beads were washed three times with 50 mM Tris·HCl (pH 7.4)/300 mM NaCl/5 mM EDTA/0.02% NaN₃/0.1% Triton X-100/1 mM NEM, and proteins were eluted in 1% SDS/1 mM NEM at 100°C for 5 min. The supernatant (80 μl) was diluted with 400 μl of 50 mM Tris·HCl (pH 7.4)/300 mM NaCl/5 mM EDTA/0.02% NaN₃/0.2% SDS/0.5% Triton X-100/1 mM NEM and incubated with 100 μl of neutravidin-agarose (Pierce) overnight at 4°C. Unbound protein in the supernatant was immunoadsorbed and eluted as described above. After washing the neutravidin-agarose beads, bound proteins were eluted by heating to 100°C in 100 μl of 1% SDS/1 mM NEM for 5 min.

Samples of total immunoprecipitate and neutravidin-bound and unbound material were heated to 100°C in SDS sample buffer with or without 10% 2-mercaptoethanol for 5 min, electrophoresed on 4–15% gradient SDS/PAGE (Bio-Rad, Hercules, CA) or 4% SDS/PAGE (Invitrogen) gels, and transferred to PVDF membranes (Bio-Rad). Blots were incubated at room temperature for 1 h in 50 mM Tris·HCl (pH 7.4)/150 mM NaCl/0.1% Tween 20/0.5% casein and incubated overnight at 4°C in either a 1:2,000 dilution of FLAG M2-HRP antibody (Sigma) or a 1:1,000 dilution of HRP-conjugated rabbit polyclonal anti-VWF (P226; DAKO, Carpenteria, CA). Blots were developed by using the ECL-Plus kit (GE Healthcare).

D′D3 Monomer Production.

BHK cells stably transfected with pSVH-D′D3Δpro were grown in FreeStyle 293 Expression Medium (Invitrogen). Conditioned medium was collected after 72 h, and 144 μM PMSF and 10 mM NEM were added. The medium was applied onto a HiTrap Q column (5 ml; GE Healthcare) in 20 mM Hepes (pH 7.4)/20 mM NaCl, developed with an NaCl gradient (0.02–1 M). Fractions containing D′D3 were concentrated by ultrafiltration (Vivaspin 20; Sartorius, Goettingen, Germany) and chromatographed on a C₄ column (10 × 150 mm; Vydac, Hesperia, CA) developed with an acetonitrile gradient (0–100%) in 0.1% TFA. Fractions containing D′D3 were dialyzed against 20 mM Hepes (pH 7.4)/150 mM NaCl, concentrated, and chromatographed on Superdex 200 by using an AKTA system (GE Healthcare). Purified D′D3 was stored at −80°C.

Enzymatic Digestions.

D′D3 (1 μg/ml) in TBS containing 1% RapiGest SF (Millipore, Billerica, MA) was reduced with 25 mM Tris(2-carboxyethyl)phosphine for 2 h at room temperature and alkylated with 25 mM 4VP or iodoacetamide (IAA) for 2 h at room temperature. Samples were desalted and buffer exchanged by ultrafiltration (YM-3 Microcon; Millipore).

Protein N-glycosidase F digestions were done in 100 mM NH₄HCO₃ (pH 8.0) overnight at 37°C with 50 units of enzyme per microgram of substrate. Trypsin or thermolysin (Sigma) digestions were performed in 50 mM NH₄HCO₃ (pH 8.0) at an enzyme-to-substrate ratio of 1:50 (wt/wt) at 37°C overnight. Aspartic-N (Asp-N; Princeton Separations, Freehold, NJ) digestions were performed in 50 mM NH₄HCO₃ (pH 8.0) at an enzyme-to-substrate ratio of 1:100 (wt/wt) at 37°C overnight. Pepsin (Princeton Separations) digestions were done in 50 mM ammonium acetate (pH 4.0) at an enzyme-to-substrate ratio of 1:50 (wt/wt) at 37°C overnight. Arginine-C (Arg-C; Princeton Separations) digestions were done at 37°C overnight in 50 mM NH₄HCO₃ (pH 8.0)/7.7 mM DTT/1 mM calcium acetate, at an enzyme-to-substrate ratio of 1:50 (wt/wt). Chymotrypsin (Princeton Separations) digestions were carried out at 37°C overnight in 50 mM NH₄HCO₃ (pH 8.0)/1 mM CaCl₂ at an enzyme-to-substrate ratio of 1:50 (wt/wt). Double digests with Glutamic-C (Glu-C; Princeton Separations) and Asp-N (Glu-C/Asp-N) were done sequentially with initial addition of Glu-C at an enzyme-to-substrate ratio of 1:20 (wt/wt) at 37°C overnight in 50 mM NH₄HCO₃ (pH 8.0), followed by digestion with Asp-N as described above.

To limit trypsin digestion to Arg residues, Lys residues of reduced and alkylated D′D3 were modified with a 25-fold excess of sulfosuccinimidyl acetate (Pierce) in 0.1 M NaHCO₃ (pH 8.5) for 2.5 h at room temperature. The reaction was quenched by adding 1 M glycine (pH 8.0) for 1 h at room temperature.

Peptides were desalted by adsorption on NuTip Hypercarbon cartridges (Glygen, Columbia, MD) equilibrated with 0.1% formic acid and elution with 0.1% formic acid, 60% acetonitrile. Peptides were lyophilized and dissolved in 0.1% formic acid for analysis.

Mass Spectrometry.

Mass spectrometry was performed by using a linear quadrupole ion trap Fourier transform cyclotron resonance mass spectrometer (LTQ-FTMS; Thermo Scientific, Waltham, MA) interfaced to a nanoliquid chromatograph (Eksigent nano-LC; Eksigent, Livermore, CA) as described (18). Theoretical lists of peptide masses were generated for each enzymatic digest considering 0–3 missed cleavages and including modification of all Cys by NEM and/or 4VP or by NEM and/or IAA, also including acetylated Lys where appropriate. Modification of Cys (residue mass = 103.0092 Da) by NEM yields S-ethylsuccinimidocysteine (residue mass = 228.0569 Da), by 4VP yields S-pyridylethylcysteine (residue mass = 208.0670 Da), and by IAA yields carboxamidomethylcysteine (residue mass = 160.0306 Da). Acetylation of Lys increases the residue mass from 128.0950 Da to 170.1055 Da). By using selected ion extraction and Xcalibur software (Thermo Scientific), MS spectra were analyzed for m/z signals corresponding to the singly, doubly, and triply protonated species. MS/MS spectra were searched against a database of D′D3 sequence y by using MASCOT (Matrix Science, Oxford, U.K.). For each Cys residue, searches were performed for all permutations of NEM-modified Cys and 4VP-modified Cys or NEM-modified Cys and IAA-modified Cys, as appropriate. Except for Cys¹⁰⁹⁹ and Cys¹¹⁴², when an observed peptide ion indicated modification of any Cys by 4VP or IAA, the corresponding ion (or ions) with Cys modified by NEM was searched for but was not detected.

Abbreviations

4VP

4-vinylpyridine

C[NEM]

Cys modified by N-ethylmaleimide, or S-ethylsuccinimidocysteine

C[4VP]

Cys modified by 4-vinylpyridine, or S-pyridylethylcysteine

ER

endoplasmic reticulum

IAA

iodoacetamide

NEM

N-ethylmaleimide

VWF

von Willebrand factor.