Generation of selective anti-BMP VHHs
SUPPLEMENTARY MATERIAL
Effective inhibition of Bone Morphogenetic Protein function
by highly specific llama-derived antibodies
Including:
-Supplementary materials and methods
-Supplementary figure legends (Figures S1 to S5)
-Supplementary references
SUPPLEMENTARY MATERIALS AND METHODS
Immunization and Library construction
The immunizations of the llamas were approved by the Utrecht University ethical committee (DEC#: 2007.III.01.013). Two llamas were immunized with 100µg of recombinant human BMP4 (R&D Systems). The schedule of the immunizations consisted of a priming immunization on day 0, followed by boost immunizations on day 14, day 28 and day 35. The immune response was measured in serum taken at day 44 and day 28, using solid phase ELISA as previously described (1). At day 44 peripheral blood lymphocytes (PBLs) were purified from 150 ml blood using a ficoll gradient (GE Healthcare Bio-Sciences AB) for RNA extraction and library construction. RNA isolated from these PBLs was converted into cDNA using SuperscriptIII kit (Life Technologies). DNA corresponding to antigen binding domains of IgG was amplified by PCR to generate ~700 bp fragments corresponding to the antigen binding domain of the heavy chain antibodies (VHH). To facilitate cloning, a SfiI restriction site was introduced at the 5’ in a nested PCR- step. The purified 700 bp fragment was digested with BstEII (a restriction site found in the hinge region of heavy chain antibodies) and SfiI, and the resulting ~350 bp VHH was cloned in a phage-display plasmid. The resulting library was transferred to Escherichia coli strain TG1 [supE hsd_5 thi (lac-proAB) F(traD36 proAB_ lacIq lacZ_M15)] by electroporation.
Selection and Screening of BMP4 binders
Different concentrations of BMP4 (2 µg, 0.4 µg, 0.08 µg and 0 µg) were coated onto wells of a 96-wells Maxisorp plate. After blocking with 4% marvel in PBS, phages were added at ~1011 phages per well. Phages were allowed to bind to the attached antigens for 2 hours at room temperature. Plates were washed extensively with PBS containing 0.05% Tween-20 and with PBS before elution of the bound phages with 100mM Triethylamine pH ~12. Phages were rescued by infection of E. coli strain TG1 and subsequent selection on agar plates containing 2% glucose and 100 µg/ml ampicillin. For the production of phages for the second round selection, E. coli TG1 rescued from successful selections were infected with helper phage and grown overnight at 37oC in the presence of ampicillin (100 g/ml) an kanamycin (25 g/ml) for the production of new phages, which were subsequently purified form culture supernatant and panned on immobilized BMP4 for a second round of selection. The second round of selection was similar to the first round, except for the low coating of BMP4 (0.5µg, 0.1µg, 0.02µg and 0µg), and 100-fold less phages (~109 phages per well) were used for selection. Single colonies were selected from the different outputs of the selections and grown in sterile 96-wells plates. Individual VHH binding to BMP4 were screened using the periplam fraction of the 96 different clones as follows: periplasmic fractions were prepared by inducing log-phase cultures in 96-deep wells plate with 1mM isopropyl-β-D-thiogalactopyranoside (IPTG) and growing for 16 hours at 37oC. Subsequently, cells were separated from medium by centrifugation. Bacteria pellets were resuspended into 100 µl PBS and frozen overnight. After thawing the bacteria suspensions, cell debris was separated from the periplasmic fractions, which contain functional VHH. VHH bound to 200 ng BMP4 immobilized on the Maxisorp plate wells, were detected using mAb 9E10 directed against the myc-tag and a secondary anti-mouse coupled to horseradish peroxidase, followed by the colorimetric measurementof converted Ortho-Phenylenediamine (OPD) in the presence of H2O2 at 490 nm.
Purification of anti-BMP VHHs
Sequences of the VHH binders were cut from the phagemid using the restriction enzymes SfiI and BstEII and cloned into the expression plasmid pMEK222. In this way, VHH expression was controlled by a lac-promoter, its secretion to the periplasm was directed with PelB signal sequence and the VHH sequence was fused to a FLAG and a His tags at the C terminus. DNA sequences of the VHHs were confirmed after subcloning. VHHs were produced in E.coli in the presence of 1 mM IPTG for 5 hours at 37oC. Subsequently, the VHHs were purified from the soluble periplasmic fractions via the hexa-histidine tag using Talon beads (Clontech) as previously described (1). Fractions containing purified VHHs were pooled and dialyzed overnight against PBS at 4oC. Yield and purity of the VHHs were analysed by SDS-PAGE and Coomassie staining. Purified VHHs binding to BMP4 was confirmed by ELISA as described for the VHH screening. Briefly, serial dilutions of Talon-purified VHHs (in duplicate) from 1000nM up to 100pM were added to wells coated with 250 ng BMP4. Amount of bound VHH was measured using an anti-VHH antibody and a secondary anti-rabbit coupled to peroxidase.The DNA sequences of selected clones were determined at Macrogen (Meibergdreef 31, 1105 AZ, Amsterdam, The Netherlands).
Generation of bivalent VHHs
PCR was used to amplify the VHH sequences of C4, C8 and E7. Different primer sets were designed to amplify the VHHs, which will be located at the N terminus and which will be located at the C terminus of the bivalent molecules. The primers at the 3’ of the N-terminal VHH and at the 5’ of the C-terminal VHH encode a flexible sequence (GS-linker) represented by a repeat of the pentapeptide “G-G-G-G-S”. These same primers contain a unique restriction site (BamHI). After PCR amplification, the generated fragments were digested with a unique N-terminal restriction site (SfiI) and BamHI for the VHH that will be located at the N terminus, and with BamHI and a unique C-terminal restriction site (BstEII) for VHH that will be located at the C terminus. The two fragments were ligated into pMEK222, which was digested with SfiI and BstEII. This resulted in the homodimers C4C4, C8C8 and the heterodimer C8E7, all with the linker length of 10 amino acids. Purified biheads were produced as described above for the monoheads.
Generation of VHH variants by site-directed mutagenesis
Site directed mutagenesis by overlap extension PCR (2) was used to substitute a single amino acid in the designated VHH. Two divergently oriented and partly complementary primers were designed with the substituted sequence. These primers were subsequently used together with the VHH outward primers to amplify two fragments with the VHH sequence. The generated two DNA fragments with an overlapping region were then used as template for a second PCR to glue the 2 fragments together and restore the complete VHH sequence. Final PCR fragment was digested with the restriction enzymes with unique recognition sites at the N and C terminus of the VHH and cloned in the expression plasmid pMEK222.Substitutions were confirmed by sequencing and VHH mutants were produced and purified as indicated before.
Specificity VHHs
C2C12 cells were plated in 96-well plates when they reached confluency, cells were treated in triplicate with 25ng/ml of BMP9, 50ng/ml of BMP10 or 50ng/ml of BMP12 (all from R&D Systems), for 16 hours. VHHs or Noggin were added at a concentration of 5ug/ml.C2C12 cells or HEK293 were plated on 12-well plates and starved for 5 hours, before stimulation with 2ng/ml of TGF- (R&D Systems) or 100ng/ml Activin-A (R&D Systems), respectively. VHHs at the indicated concentrations were added at the same time. After 1 hour cells were lysed, and protein isolation and SDS-PAGE separation were performed as described previously (3). For protein detection the following antiboides were used: antiphospho-SMAD2 at 1:1000 (Cell Signalling).
HADDOCK Modelling
Residues located in the complementary-determining regions (CDR1 and CDR2) that differ from the germline as a result of one or more mutations in the codon, as well as residues in CDR3, were chosen as the VHH active sites for the docking. Therefore, for C4, those residues were:T31,K53,I57,D99, A110, P112; for C8: S44, L100, R101, F102, R106 Y107, R108 ,F110; and for E7: R29, G30, V32, G55, T56, L57, G68 and L96.
Epitope binning experiments suggested that C4 interacts with the molecular interface involved in BMP4-BMPR1a binding. Because C4 can bind and inhibit BMP4- but not BMP2-mediated activation, C4 most likely targets a BMPR1a binding region in which differences in residues between BMP2 and BMP4 are observed. Examination of the sequences of BMP2 and BMP4 reveals that the hydrophobic groove of the wrist epitope is the only BMPR1a-binding area where differences between BMP2 and BMP4 exist (K12, N13 and R15). Therefore this region was used for docking C4 to BMP4. Residues located in this region and used for the docking were: S12, S13 and K15 (different between BMP2 and BMP4) as well as F49, P50, L51, A52, D53, H54 and S69 (identical between BMP2 and BMP4) (Supplementary Fig.S4B). The docking results were grouped into 10 clusters. Cluster 1 contained 78 structures and presented a HADDOCK score of -86.7 +/-8.7.
A hydrophobic pocket constitutes the other contact point of the wrist epitope (Supplementary Fig. S4A). Residues within the αhelix of BMP2A and residues in loop1, β7 and β8 of BMP2B fit like a “knob-into-pocket” with a F85 of a molecule of BMPR1a (orange residues in Supplementary Fig. S4B). Because C8 can only bind to BMP2 and BMP4 and not the rest of the BMPs, the epitope for C8 presumably resides in a BMPR1a binding region displaying differences between BMP2,4 and BMP5,6,7 but not overlapping with C4. A large number of non-conserved residues between these two subgroups of BMPs are found in this region (blue residues in Supplementary Fig.S4B). Since they might represent a determinant for C8 specificity, residues within the hydrophobic pocket and different between BMP2,4 and BMP5,6,7 were used for docking: V26, W30, N31, N68, S69, M89. HADDOCK grouped 191 structures into 6 clusters. Cluster 1 presented a score of -142.6 +/- 2.9 that included 138 structures.
Our epitope mapping experiments suggested that E7 might bind to the knuckle epitope. Moreover, all the residues in this region are highly conserved between BMPs, a notion that corresponds to the promiscous nature of E7. The ‘knuckle” epitope, is formed by residues of only BMP2B in which four regions are involved in binding to BMPR2 (4). In this epitope, a small shallow pocket is formed by hydrophobic residues surrounded by a ring of polar and non-polar amino acids. The central hydrophobic residues have been shown to be binding determinants for this molecular interface and were therefore used to model E7 with BMP4: S88, L90, A34, H39 and L100 (Supplementary Fig.S4B). HADDOCK clustered 173 structures into 4 clusters. Cluster 1 with 150 structures gave the highest score: -83.5 +/-5.
SUPPLEMENTARY FIGURE LEGENDS
Figure S1:VHH biheads showincreased functional activity as compared to monoheads.
EPC2-hERT cells were activated with 100ng/ml of human BMP4 (A,C,E) or BMP2 (B,D,F) for 4 hours. VHH biheads at the indicated concentrations were added at the same time as the BMPs. Phosphorylation of SMAD1/5/8 was detected by western blot. Equivalent protein loading was confirmed by detection of GAPDH. Protein quantification was measured with ImageJ. Error bars represent standard deviations of the mean, calculated from 3 independent experiments. ***= <0.0001, **=<0.01, *=<0.1. Stadistical analysis was done using a two-tailed P-test.
Figure S2: VHH biheads show differential specificity to BMPs.
C2C12 were activated with human BMP2 (50ng/ml), BMP4 (5ng/ml), BMP5 (200ng/ml), BMP6 (50ng/ml) or BMP7 (200ng/ml) for 16 hours. VHH biheads were added at the same time at the indicated concentrations. Human Fc-Noggin was added at a concentration of 5g/ml. Luciferase activity was assayed with the Bright-Glo™ Luciferase Assay System. Luciferase values were normalized to background activity and represented as ratio to the activity of BMP-stimulated cells. N= Noggin. Error bars represent standard deviations of the mean, calculated from at least three independent experiments, with experimental triplicates each. ***= <0.0001, **=<0.01, *=<0.1. Stadistical analysis was done using a two-tailed P-test. (A) C4C4, (B) C8C8, (C) C8-E7.
Figure S3: VHH do not inhibit other members of the BMP subdfamily or the TGF- superfamily.
C2C12 were activated with human A) BMP9 (25ng/ml), B) BMP10 (50ng/ml) or C) BMP12 (50ng/ml), for 16 hours. VHH or Noggin were added at the same time at a concentration of 5g/ml. Luciferase activity was assayed with Bright-Glo™ Luciferase Assay System (Promega). Luciferase values were normalized to background activity and represented as ratio to the activity of BMP-stimulated cells. Error bars represent standard deviations of the mean, calculated least three independent experiments, with experimental triplicates each. *=<0.1. Stadistical analysis was done using a two-tailed P-test. D) C2C12 were activated with 2ng/ml of TGF- (R&D Systems) for one hour. VHHs at the indicated concentrations were added at the same time. Western blots of phospho SMAD2 are representative of two experiments. E) HEK293 cells were activated with 100ng/ml of Activin-A (R&D Systems) for one hour. VHHs at the indicated concentrations were added at the same time. Western blots of phospho SMAD2 are representative of two experiments.
Figure S4: BMP binding to BMPRs.
A) Space filled view of BMP2 binding to BMPR1a and BMPR2 (PBD entry 2H62). The insets show an enlarged cartoon diagram of each of the molecular binding interfaces. The wrist is composed by two contact points: a hydrophobic groove (formed by residues located at the preβ, β1 and loop2) and a hydrophobic pocket (composed by residues of the αhelix from BMP2Aand residues from the loop1, β7 and β8 from BMP2B). The knuckle epitope is formed by residues of monomer BMP2B located in the-strandsβ3, β4, β7 and β8. B) Sequence alignment of mature BMPs. NCBI accession numbers of full immature BMPs: BMP2, NP_001191.1; BMP4, AAH20546; BMP5, NP_066551.1; BMP6, NP_001709.1, BMP7, NP_001710.1. Location of secondary structures such as β strands, loops and αhelix are adapted from (5). Dashes represent identical residues. In total 41 residues have been involved in BMP2-receptor interactions (6,7). Residues shaded in green are residues involved in the hydrophobic groove of the wrist epitope; in orange in the hydrophobic pocket of the wrist epitope; and in red in the knuckle epitope (4). Residues in green are groove residues non-conserved between BMP2 and BMP4. Pocket residues non-conserved between BMP2,4 and BMP5,6,7 are represented in dark orange. Heparin binding domains are represented as bolded purple residues(8).
Figure S5: Epitope Mapping with heterodimer BMPs.
C2C12 were activated with 10ng/ml of human BMP4 and BMP4/6 (A) or 50ng/ml of BMP2, BMP2/6 or BMP2/7 (B), from RD systems. Error bars represent standard deviations of the mean, calculated from at least three independent experiments, with experimental triplicates each. C) Heparin domain mutant (HDBMP4) was purchased from Prepotech (AF-120-05ET) and used at 500ng/ml. Noggin (100ng/ml), VHHs (5g/ml) and anti-BMP4 (5g/ml) were added at the same timewith the BMPs and incubated for 16 hours. Error bars represent standard deviations of the mean, calculated from a representative experiment, with experimental triplicates. N= Noggin.
SUPPLEMENTARY REFERENCES
1. El Khattabi M, Adams H, Heezius E, Hermans P, Detmers F, Maassen B, et al. Llama single-chain antibody that blocks lipopolysaccharide binding and signaling: prospects for therapeutic applications. Clin Vaccine Immunol CVI. 2006;13:1079–86.
2. Ho SN, Hunt HD, Horton RM, Pullen JK, Pease LR. Site-directed mutagenesis by overlap extension using the polymerase chain reaction. Gene. 1989;77:51–9.
3. Milano F, van Baal JWPM, Buttar NS, Rygiel AM, de Kort F, DeMars CJ, et al. Bone Morphogenetic Protein 4 Expressed in Esophagitis Induces a Columnar Phenotype in Esophageal Squamous Cells. Gastroenterology. 2007;132:2412–21.
4. Nickel J, Dreyer MK, Kirsch T, Sebald W. The crystal structure of the BMP-2:BMPR-IA complex and the generation of BMP-2 antagonists. J Bone Joint Surg Am. 2001;83-A Suppl 1:S7–14.
5. Kirsch T, Nickel J, Sebald W. BMP-2 antagonists emerge from alterations in the low-affinity binding epitope for receptor BMPR-II. EMBO J. 2000;19:3314–24.
6. Kirsch T, Sebald W, Dreyer MK. Crystal structure of the BMP-2-BRIA ectodomain complex. Nat Struct Biol. 2000;7:492–6.
7. Allendorph GP, Vale WW, Choe S. Structure of the ternary signaling complex of a TGF-beta superfamily member. Proc Natl Acad Sci U S A. 2006;103:7643–8.
8. Ruppert R, Hoffmann E, Sebald W. Human bone morphogenetic protein 2 contains a heparin-binding site which modifies its biological activity. Eur J Biochem FEBS. 1996;237:295–302.
1