Nature Structural Biology
9, 288 - 292 (2002)
Published online: 25 February 2002; | doi:10.1038/nsb769
Structural basis of BLyS receptor recognitionDeena A. Oren1, Yuling Li2, Yulia Volovik1, Tina S. Morris2, Chhaya Dharia1, Kalyan Das1, Olga Galperina2, Reiner Gentz2
& Eddy Arnold11 Center for Advanced Biotechnology and Medicine and Department of Chemistry and Chemical Biology, Rutgers University, 679 Hoes Lane, Piscataway, New Jersey 08816, USA. 2 Department of Protein Development, Human Genome Sciences, 9410 Key West Avenue, Rockville, Maryland 20850, USA.
Correspondence should be addressed to Eddy Arnold arnold@cabm.rutgers.eduB lymphocyte stimulator (BLyS), a member of the tumor necrosis factor (TNF) superfamily, is a cytokine that induces B-cell proliferation and immunoglobulin secretion. We have determined the three-dimensional structure of BLyS to 2.0 Å resolution and identified receptor recognition segments using limited proteolysis coupled with mass spectrometry. Similar to other structurally determined TNF-like ligands, the BLyS monomer is a -sandwich and oligomerizes to form a homotrimer. The receptor-binding region in BLyS is a deeper, more pronounced groove than in other cytokines. The conserved elements on the 'floor' of this groove allow for cytokine recognition of several structurally related receptors, whereas variations on the 'walls' and outer rims of the groove confer receptor specificity.B lymphocyte stimulator (BLyS, also known as TALL-1 (ref. 1), THANK2, BAFF3 and zTNF4 (ref. 4)), a member of the tumor necrosis factor (TNF) superfamily of cytokines, induces B-cell proliferation and immunoglobulin secretion and is a key regulator of peripheral B-cell populations in vivo4,
5. As with other members of the TNF ligand family, BLyS is a type-II membrane protein that is cleaved at the cell surface, forming a soluble protein6. Although members of the TNF ligand family show significant sequence diversity, they are structurally related7,
8,
9,
10,
11. BLyS and other TNF ligand family members are biologically active as trimers12,
13. Similar to other members of the TNF family, BLyS is a ligand that interacts with several receptors. BLyS was initially shown to interact with TACI (transmembrane activator and CAML interactor) and BCMA (B-cell maturation antigen)14. Both of these receptors were found to bind APRIL (a TNF-like ligand most similar to BlyS) as well15,
16. A third receptor, termed BAFF-R17 or BR-3 (ref. 18), has been identified. This receptor apparently does not interact with APRIL17 or any other TNF family members other than BLyS18. Experiments using transgenic animals have shown that the interaction of BLyS with TACI and BCMA plays a role in the development of autoimmune disease14. At the same time, BLyS is a crucial factor for the normal development of B-cells, and this function seems to be mediated through a BCMA-independent pathway19. In all the known TNF receptor−ligand complexes10,
12,
20,
21, the receptor-binding site is formed by two monomers of the ligand trimer (as initially noted by Eck and Sprang8 and summarized by Idriss et al.22). We determined the crystal structure of BLyS in order to shed light on the receptor specificity of BLyS. Understanding the structural basis of this specificity and of the unique ability of BLyS to regulate B-cell development may create opportunities for the design of new drugs for treating immune disorders.
Three-dimensional structure of BLyS The BLyS structure was solved by molecular replacement using the homolog TRAIL as a starting model, and data were collected using high-energy synchrotron X-radiation (Table 1). The BLyS monomer adopts the TNF-like jellyroll fold as in other representatives of this family. A structure-based sequence alignment among members of this cytokine family (Fig. 1) reveals that the Greek-key motif of the five-stranded -sheets is conserved throughout the family despite the low sequence identity. Using these structural alignments, the calculated identities between BLyS and the homologs are TNF- , 15%; CD40L, 16%; RANKL, 19%; TRAIL, 18%; and TNF- , 20%. The identities occur primarily in -strands C, D, F, G and H, which constitute the core of the jellyroll fold (Fig. 1), whereas the differences occur in the loop regions AA'', CD, DE, EF and GH. In contrast to related TNF-family ligands, BLyS does not have the short GH -helix, is truncated in loops CD and EF, and contains large inserts between strands A and A'' and between strands D and E. The AA" loop has an insertion of two short -strands, a and a', that form a hairpin motif (Fig. 2b), which does not participate in -sheet formation but widens the molecule. Similarly, the DE loop contains a four-residue insert, protrudes from the surface and forms inter-trimer contacts reminiscent of a handshake. As a result of these changes, the BLyS homotrimer measures 52 Å high (along the three-fold axis) and 60 Å wide, compared with 58 Å and 57 Å, respectively, for TNF- . Three BLyS monomers make extensive contacts (buried surface area = 5,700 Å2), forming the trimer, with the -sheets inclined 30° relative to the three-fold axis (Fig. 2a,b).
 | | Figure 1. Structural alignment of TNF-like cytokines. |  |  |  | Alignments are as calculated with the Swiss PDB Viewer38 using the PDB entries 1TNF (TNF- )8, 1TNR (TNF- )12, 1D4V (TRAIL)21, 1ALY (CD40L)32 and 1JTZ (RANKL)11. APRIL was added to the alignment4. Boxes denote -strands assigned by PDB Viewer, and the bar denotes the -helix as assigned in the PDB entries. Residues that contact the respective receptors are colored pink, and locations of mutations that affect receptor binding are shown in yellow. Residues that contact (within 4 Å) both TNF-R and DR-5 when docked with the cytokine are in cyan. Green letters indicate disulfide-bonded Cys residues, and residues that bind metals are red. The asterisk stands for a nine-residue insert in TRAIL with the sequence SNTLSSPNS. Also displayed is a ribbon diagram of the BLyS monomer, drawn using RIBBONS39. Strands are labeled following the TNF- notation7 and color-coded as follows: A, a, a' and A" are red; B and B', green; C, yellow; D, blue; E, magenta; F, cyan; G, lavender; and H, brown. The two antiparallel sheets forming the jellyroll comprise strands A'', A, H, C and F in one sheet, and B', B, G, D and E in the other sheet.
Full Figure and legend (88K) |
|  |
 | | Figure 2. BLyS structure. |  |  |  | Ribbon diagrams (using RIBBONS39) of a, BLyS (left) and TNF- (right) are shown down the three-fold axis of the trimer, and b, rotated 90° about the horizontal viewing axis. Strands of two of the three monomers are color coded as in Fig. 1. Gln 234 is green; Asn 243, yellow; Mg2+, brown; and waters, red. The metals have a coordination sphere of six, and metal−ligand distances range between 2.4 Å and 2.8 Å. The BLyS trimer has a wider and shorter shape than that of TNF- . c, SIGMAA35-weighted electron density map using 2mFo − Fc coefficients and contoured at 1 . Shown is the region of the three-fold axis (vertical in the viewing plane), highlighting the high quality of the refined electron density, the geometry of the metal coordination and the presence of a dioxane molecule. The electron density associated with the hydrated metals is colored in magenta for clarity.
Full Figure and legend (130K) |
|  |
 | |  | Near the narrow end of the BLyS trimer (the side probably presented to the B-cell receptor), a complex containing two hydrated magnesium ions binds along the three-fold axis (Fig. 2). This metal-binding site probably plays an important role in stabilizing the trimer because it links adjacent monomers at Gln 234 and bridges the Asn 243 side chains and Asn 235 main chain oxygens through the metal hydration sphere. In addition, treatment of BLyS with EDTA reduced biological activity and induced aggregate formation, as observed on size-exclusion HPLC. An analogous metal chelation exists in the homolog TRAIL, which contains a zinc ion bound in a related position (6.4 Å from Mg1 when superimposing the core -strands) along the three-fold axis that interacts with Cys 230 sulfhydryls from each monomer9.
Other molecules were also observed bound to the protein. Molecules of dioxane (the crystallizing agent) were identified along the three-fold axis, interacting with the phenyl rings of Phe 165 and Phe 194, and in other hydrophobic pockets. Also, two citrate molecules were located at the interface between the two trimers in the asymmetric unit where the DE loops 'shake hands'. The citrates interact with each other through the hydroxyl groups, and their carboxylates bind to Arg 214, Lys 216, His 218, Glu 223, Lys 252 and Glu 254. Citrate-containing buffer also stabilizes BLyS better than formulations containing phosphate, Tris or acetate. The presence of citrate in the structure may explain this stability, as well as the observation that using citrate as a buffer has a pronounced effect on the ionic strength required to elute BLyS from an ion exchange column (10−50 mM sodium citrate versus 500−1000 mM sodium chloride).
Deep groove and receptor-binding specificity of BLyS A comparison of the molecular surfaces among the trimeric form of BLyS (Fig. 3) and other TNF-like ligands has revealed a unique shape with three pronounced grooves on the surface. The TNF-R and DR5 (death receptor-5, TRAIL receptor) receptors bind to analogous grooves of their respective ligands. A cleft noted11 in the structure of RANKL is also visible in other related cytokines but is most extensive and deep in BLyS (Fig. 3). The groove is created by loops from two adjacent monomers (Fig. 3). One 'wall' of the groove contains loop DE with some residues of loops aa' and GH, and the other wall contains loops EF, Aa and a'A". The deepest portion (or 'floor') of this groove consists primarily of -strands D, E and F. Residues with surface accessible side chains are Ala 207, Leu 211, Gln 213 and Arg 214 from strand D; Thr 228, Leu 229, Phe 230, Arg 231 and Ile 233 from strand E; and Ala 251, Lys 252, Leu 253, Glu 254 and Asp 257 from strand F. The groove winds around the surface of the trimer and has a shape appropriate for binding elongated receptors (Fig. 4a). Loops DE and AA" form the most extensive contacts with cytokine receptors (PDB entries 1TNR12 and 1D4V21 or 1D0G23). Modeling potential interactions of BLyS with TNF-R (Fig. 4a) suggests that the outer rim of the groove (loops DE and the -hairpin of loop AA") would lead to steric conflict. These residues would permit receptors to discriminate between TNF (or other cytokines) and BLyS. The residues involved in creating the surface of this groove and putative receptor-binding site are from adjacent monomers (green, Fig 4a). Of those residues, the homolog APRIL shares residues Leu 200, Arg 214, Thr 228, Leu 229, Phe 230, Arg 231, Ile 233, Leu 253, Asp 257 and Phe 278 with BLyS (Fig. 4a, red). The majority of these shared residues are located on the floor of the groove, suggesting that the floor is used as a common binding motif for TACI, BCMA and BAFF-R to BLyS and APRIL. Variations in residues on the groove walls would permit BAFF-R to discriminate against APRIL.
 | |  |
 | | Figure 4. Putative BLyS−receptor interactions. |  |  |  | a, Superimposed TNF-R (red ribbon) docked on BLyS surface representation, color-coded by monomer as in Fig. 3, with groove residues in green. The orientation on the left is as in Fig. 3. The middle image is the same but rotated 90° about the horizontal viewing axis. On the right, groove residues in common between BLyS and APRIL are colored red, implicating contacts with TACI and BCMA receptors. The residues forming the groove from adjacent monomers are Gln 148, Ile 150, Ala 151, Asp 152, Ser 153, Glu 154, Leu 169, Leu 170, Phe 172, Leu 200, Thr 202, Ile 270, Ser 271, Leu 272, Asp 273, Glu 274, Asp 275 and Phe 278 from one monomer, and Thr 190, Tyr 192, Ala 207, Gly 209, His 210, Leu 211, Gln 213, Arg 214, Lys 216, His 218, Phe 220, Asp 222, Glu 223, Leu 224, Leu 226, Val 227, Thr 228, Leu 229, Phe 230, Arg 231, Ile 233, Ala 251, Lys 252, Leu 253, Glu 254 and Asp 257 from another monomer. Those in common with APRIL are underlined. b, PAWS coverage analysis, mapping fragments found in SELDI binding assays of TACI and BMCA to areas in the BLyS sequence. Red boxes highlight areas of strongest coverage. Binding-site mapping was done by in situ trypsin digestion of the captured ligand, followed by mass spectrometric identification of retained fragments. Arrows mark BLyS -strands.
Full Figure and legend (48K) |
|  | The three receptors known to bind and be activated by BLyS share little sequence identity, but they all contain at least one Cys-rich domain. As seen in the complex between TNF and TNF-R, the Cys-rich region of the receptor forms contacts with loops AA" and DE of TNF- . BAFF-R, the receptor with the highest affinity for BLyS, has the shortest sequence, containing only one Cys-rich domain. A ProDom24 database search (aided by PredictProtein25) probed using the BAFF-R sequence revealed BCMA as the most similar, specifically in the Cys-rich region, the transmembrane domain and an intracellular portion consisting of residues GEDPGTTPGHSVPVPA. In a receptor-binding study using SELDI affinity mass spectrometry26, we show that the a'A" loop, the B' and B strands, and strands C and D of the molecule are centrally involved (Fig. 4b) in the interaction of BLyS with both recombinant BCMA and TACI receptors, as indicated by the relatively large number of retained fragments of BLyS that map to these areas. The data support the assumption that BLyS interacts similarly with its receptors as other TNF ligands interact with their respective receptors. TACI and BCMA are unable to mediate the survival activity of BLyS, and the interaction of BAFF-R with BLyS was recently determined to be important to peripheral B-cell survival19,
27. This highlights the ability of the unique surface of BLyS to interact differently with several receptors.
Conclusions In summary, the structure of BLyS has revealed a distinctive binding groove formed by adjacent monomers within the trimer that permits the cytokine to discriminate among closely related receptors. The floor of the groove seems to harbor shared receptor-binding elements that permit recognition of the three receptors TACI, BCMA and BAFF-R, whereas variations on the outer rims of the groove confer specificity to the interaction. This model, supported by evidence obtained using SELDI affinity mass spectrometry, provides a basis for understanding cytokine receptor-binding specificity and the unique regulation of immune function by BLyS. We now have a model that explains both cross-reactivity and specificity. By targeting areas that are implicated in receptor discrimination, developing drugs that can selectively modulate the immunoregulatory functions of BLyS should be possible.
Note added in proof: While this paper was in press, an independent study of sTALL-1 (BLyS) was reported28. The authors solved an icosahedrally symmetric assembly of BLyS to 3 Å resolution, in which neighboring BLyS trimers interact through pairs of DE loops ('flaps'). They propose that this virus-like particle is the biologically active form of BLyS. In our structure, pairwise DE loop contacts form the majority of intertrimer contacts, both within the asymmetric unit ('handshake' mediated by citrate) and in a network linking trimers by crystallographic symmetry.
Methods Protein expression and purification. BLyS was expressed in insect Sf9 cells using a recombinant baculovirus system4 and was secreted as the mature protein (residues 134−285). Sf9 cell supernatant was treated with 10 mM CaCl2 in slightly alkaline conditions. BLyS was purified through Poros PI-50 (Applied BioSystem), Sephacryl S200 size exclusion (Amersham Pharmacia Biotech), Toyopearl Hexyl 650C (TosoHaas) and DEAE sepharose (Amersham Pharmacia Biotech) columns. The final purified BLyS protein was diafiltered into a buffer containing 25 mM sodium citrate and 125 mM NaCl, pH 6.
Crystallization. The sparse matrix29 approach was used for crystal screening. The only condition that gave crystals was 35% (v/v) dioxane in water (Crystal Screen II, condition #4; Hampton Research). Crystals were grown in hanging drops containing 1 l of 20 mg ml-1 BLyS in 25 mM sodium citrate, 125 mM NaCl, pH 6, 1 l of 25% (v/v) dioxane and 25 mM MgCl2 suspended over 25% dioxane and 25 mM MgCl2 reservoir. Crystals appeared overnight. Crystals were flash-cooled for data collection by rapid transfer into 25% (v/v) glycerol, 25% (v/v) dioxane and 25 mM MgCl2, followed by direct placement into the liquid N2 stream. The lattice had hexagonal symmetry, with unit cell dimensions of a = b = 123.58 Å, c = 161.23 Å, = = 90° and = 120°, and the space group was subsequently found to be P65. Crystal density measurements using Ficoll gradients indicated six BLyS monomers per asymmetric unit30. The Matthews coefficient for these crystals was calculated to be 3.58 Å3 Da-1 for a solvent content of 65%, assuming 912 residues.
Data collection and processing. All data were collected from one flash-cooled crystal on the Cornell High Energy Synchrotron Source (CHESS) in Ithaca, NY, on the F1 beamline using the ADSC Quantum4 detector system and = 0.942Å. A total of 250 oscillations of 1° were collected using 40 s exposures and a crystal-to-detector distance of 190 mm. Intensities were integrated, reduced, scaled (I / (I) > -3) and postrefined with the DENZO/SCALEPACK package31. Data collection statistics are shown in Table 1.
Structure determination. Attempts at structure determination via MIR and MAD procedures were unsuccessful but yielded molecular envelopes indicating that the asymmetric unit contained a dimer of trimers. Because of low sequence identity and the large number of molecules (six) in the 96-kDa asymmetric unit, the molecular replacement procedure was challenging. Search models for molecular replacement were chosen from homologs TNF (1TNR)12, TRAIL (1D0G)23 and CD40 ligand (1ALY)32, using monomers and/or trimers, after truncating side chains to Ala. A 'pruned' model created from TRAIL (18% identity with BLyS) that retained side chains for residues identical with BLyS (on the basis of sequence alignments) yielded solutions with the best statistics in AMoRe33 (Table 1) and was used to locate one trimer and then the other to complete the asymmetric unit. The same solution was obtained using the TNF- model. Phases were calculated with CNS34 using SIGMAA35-weighting and refined by density modification using solvent 'flipping' assuming 60% solvent content. The refined phases yielded remarkably clear and minimally biased electron density maps at both 3.5 Å and 2.0 Å resolution without requiring noncrystallographic symmetry averaging. All the protein structure segments that differed between the model and BLyS were apparent in this map, including new loops and disulfide bonds, as well as ordered ligands and solvent molecules.
Model building and refinement. One monomer of 143 amino acids (residues 142−285) was modeled within 48 h, and then replicated and transformed using the 'lsq' module in O36 onto the other five locations in the asymmetric unit. One round of simulated annealing, initially at 2,000 °C with maximum likelihood refinement, reduced the values of R-factor to 26.0% and Rfree to 28.4% using 30.0−2.0 Å resolution data. Subsequent cycles of refinement involved addition of ligands (metals, citrates and dioxanes) and water molecules with final R-factor and Rfree of 18.9% and 20.9%, respectively (|F| 1 (F)). The final model consisted of 7,473 nonhydrogen atoms. N-terminal residues 134−141 of each of the monomers are unobserved, and residues 204−206 in each monomer have weak density. The overall geometry is good, with no non-Gly, non-Pro residues in the disallowed regions of the Ramachandran plot.
SELDI mass spectrometry and data analysis. A surface-enhanced laser desorption-ionization (SELDI) approach was used to identify regions involved in BLyS binding to receptors TACI and BCMA37. Recombinant receptor proteins, tagged with an immunoglobulin Fc domain, were expressed in CHO cells (TACI) or baculovirus-infected insect cells (BCMA) and tested for binding activity by BIACORE and cell-based assays. The receptors were then covalently bound to PS2 ProteinChipTM Arrays (Ciphergen Biosystems) and subsequently incubated with recombinant BLyS. After removal of unbound material, the complexes were digested with a high concentration of trypsin. Unretained digest fragments were removed by a stringency wash. The energy-absorbing molecule, -hydroxy-cinnaminic acid (CHCA) in 10% (v/v) formic acid and 10% (v/v) ethanol, was added, and chips were analyzed on a Ciphergen PBS 2, as well as a PE Sciex Qstar with protein chip interface. PS2 data with four-point external calibration achieve a mass accuracy of 50−100 p.p.m., and QStar data have 5 p.p.m. accuracy. Fragment matches and distributions were analyzed using PAWS (Protein Analysis Worksheet, Proteometrics).
Coordinates. The coordinates have been deposited in the Protein Data Bank (accession code 1KXG).
Received 6 December 2001; Accepted 23 January 2002; Published online: 25 February 2002.
REFERENCES
- Shu, H.B., Hu, W.H. & Johnson, H. J. Leukoc. Biol. 65, 680−683 (1999). | PubMed | ISI | ChemPort |
- Mukhophadhyay, A., Ni, J., Zhai, Y., Yu, G.-L. & Aggarwal, B.B. J. Biol. Chem. 274, 15978−15981 (1999). | Article | PubMed |
- Schneider, P. et al. J. Exp. Med. 189, 1747−1756 (1999). | Article | PubMed | ISI | ChemPort |
- Moore, P.A. et al. Science 285, 260−263 (1999). | Article | PubMed | ISI | ChemPort |
- Mackay, F. et al. J. Exp. Med. 190, 1697−1710 (1999). | Article | PubMed | ISI | ChemPort |
- Mariani, S.M., Matiba, B., Armandola, E.A. & Krammer, P.H. J. Cell Biol. 137, 221−229 (1997). | Article | PubMed | ISI | ChemPort |
- Eck, M.J., Ultsch, M., Rinderknecht, E., de Vos, A.M. & Sprang, S.R. J. Biol. Chem. 267, 2119−2122 (1992). | PubMed | ISI | ChemPort |
- Eck, M.J. & Sprang, S.R. J. Biol. Chem. 264, 17595−17605 (1989). | PubMed | ISI | ChemPort |
- Hymowitz, S.G. et al. Biochemistry 39, 633−640 (2000). | Article | PubMed | ISI | ChemPort |
- Cha, S.S. et al. J. Biol. Chem. 275, 31171−31177 (2000). | Article | PubMed | ISI | ChemPort |
- Lam, J., Nelson, C.A., Ross, F.P., Teitelbaum, S.L. & Fremont, D.H. J. Clin. Invest. 108, 971−979 (2001). | Article | PubMed | ISI | ChemPort |
- Banner, D.W. et al. Cell 73, 431−445 (1993). | Article | PubMed | ISI | ChemPort |
- Kanakaraj, P. et al. Cytokine 13, 25−31 (2001). | Article | PubMed | ISI | ChemPort |
- Gross, J.A. et al. Nature 404, 995−999 (2000). | Article | PubMed | ISI | ChemPort |
- Marsters, S.A. et al. Curr. Biol. 10, 785−788 (2000). | Article | PubMed | ISI | ChemPort |
- Wu, Y. et al. J. Biol. Chem. 275, 35478−35485 (2000). | Article | PubMed | ISI | ChemPort |
- Thompson, J.S. et al. Science 293, 2108−2111 (2001). | Article | PubMed | ISI | ChemPort |
- Yan, M. et al. Curr. Biol. 11, 1547−1552 (2001). | Article | PubMed | ISI | ChemPort |
- Schiemann, B. et al. Science 293, 2111−2114 (2001). | Article | PubMed | ISI | ChemPort |
- Singh, J. et al. Protein Sci. 7, 1124−1135 (1998). | PubMed | ISI | ChemPort |
- Mongkolsapaya, J. et al. Nature Struct. Biol. 6, 1048−1053 (1999). | Article | PubMed | ISI | ChemPort |
- Idriss, H.T. & Naismith, J.H. Microsc. Res. Tech. 50, 184−195 (2000). | Article | PubMed | ISI | ChemPort |
- Hymowitz, S.G. et al. Mol. Cell 4, 563−571 (1999). | Article | PubMed | ISI | ChemPort |
- Corpet, F., Gouzy, J. & Kahn, D. Nucleic Acids Res. 26, 323−326 (1998). | Article | PubMed | ISI | ChemPort |
- Rost, B. Methods Enzymol. 266, 525−539 (1996). | Article | PubMed | ISI | ChemPort |
- Weinberger, S.R., Morris, T.S. & Pawlak, M. Pharmacogenomics 1, 395−416 (2000). | Article | PubMed | ChemPort |
- Gross, J.A. et al. Immunity 15, 289−302 (2001). | Article | PubMed | ISI | ChemPort |
- Liu, Y. et al. Cell 108, 383−394 (2002). | Article | PubMed | ISI | ChemPort |
- Jancarik, J. & Kim, S.H. J. Appl. Crystallogr. 24, 409−411 (1991). | Article | ISI | ChemPort |
- Westbrook, E.M. Methods Enzymol. 114, 187−196 (1985). | Article | PubMed | ISI | ChemPort |
- Otwinowski, Z. & Minor, W. In International tables for crystallography, Vol. F (eds Rossmann, M.G. & Arnold, E.) 226−235 (Kluwer Academic Publishers, Dordrecht; 2001).
- Karpusas, M. et al. Structure 3, 1031−1039 (1995). | Article | PubMed | ISI | ChemPort |
- Navaza, J. Acta Crystallogr. A 50, 157−163 (1994). | Article | ISI |
- Brünger, A.T. et al. Acta Crystallogr. D 54, 905−921 (1998). | Article | PubMed | ISI |
- Read, R.J. Acta Crystallogr. A 42, 140−149 (1986). | Article | ISI |
- Jones, T.A., Zou, J.Y., Cowan, S.W. & Kjeldgaard, M. Acta Crystallogr A 47, 110−119 (1991). | Article | PubMed | ISI |
- Kuwata, H., Yip, T.T., Yip, C.L., Tomita, M. & Hutchens, T.W. Adv. Exp. Med. Biol. 443, 23−32 (1998). | PubMed | ISI | ChemPort |
- Guex, N. & Peitsch, M.C. Electrophoresis 18, 2714−2723 (1997). | PubMed | ISI | ChemPort |
- Carson, M. J. Appl. Crystallogr. 24, 958−961 (1991). | Article | ISI |
- Nicholls, A., Sharp, K. & Honig, B. Proteins Struct. Funct. Genet. 11, 281−296 (1991). | PubMed | ChemPort |
Acknowledgments The authors thank M. Zhang for her contribution in purifying BLyS protein, T. Kwong for excellent work preparing mass spectrometry experiments, other Arnold lab members and the staff at the Cornell High Energy Synchrotron Source and BioCARS at the Advanced Photon Source for assistance, C. Rosen for valuable discussions and enthusiastic support of the collaboration and the Arnold lab gratefully acknowledges HGS for financial support.
Competing interests statement:
The authors declare competing financial interests. |