Effective Inhibition of Bone Morphogenetic Protein Function By

Published OnlineFirst September 8, 2015; DOI: 10.1158/1535-7163.MCT-14-0956

Small Molecule Therapeutics Molecular Cancer Therapeutics Effective Inhibition of Bone Morphogenetic Protein Function by Highly Speciﬁc Llama-Derived Antibodies Silvia Calpe1, Koen Wagner2, Mohamed El Khattabi3, Lucy Rutten3, Cheryl Zimberlin4, Edward Dolk3, C. Theo Verrips3, Jan Paul Medema4, Hergen Spits2, and Kausilia K. Krishnadath1,5

Abstract

Bone morphogenetic proteins (BMP) have important but signaling. Epitope binning and docking modeling have shed distinct roles in tissue homeostasis and disease, including light into the basis for their BMP specificity. As opposed to the carcinogenesis and tumor progression. A large number of BMP wide structural reach of natural inhibitors, these small mole- inhibitors are available to study BMP function; however, as culestargetthegroovesandpocketsofBMPsinvolvedin most of these antagonists are promiscuous, evaluating specific receptor binding. In organoid experiments, specific inhibition effectsofindividualBMPsisnotfeasible.Becausetheonco- of BMP4 does not affect the activation of normal stem cells. genic role of the different BMPs varies for each neoplasm, Furthermore, in vitro inhibition of cancer-derived BMP4 non- highly selective BMP inhibitors are required. Here, we describe canonical signals results in an increase of chemosensitivity the generation of three types of llama-derived heavy chain in a colorectal cancer cell line. Therefore, because of their variable domains (VHH) that selectively bind to either BMP4, high specificity and low off-target effects, these VHHs could þ to BMP2 and 4, or to BMP2, 4, 5, and 6. These generated VHHs represent a therapeutic alternative for BMP4 malignancies. have high affinity to their targets and are able to inhibit BMP Mol Cancer Ther; 14(11); 1–14. 2015 AACR.

Introduction ical intracellular effectors (3, 4). Translocation to the nucleus of these transcription factors initiates transcription of BMP target Bone morphogenetic proteins (BMP) are growth factors that genes (5, 6). belong to the TGFb superfamily. They consist of around 20 BMPs are emerging as attractive therapeutic targets due to members, classified into distinct subfamilies, depending on their multiple roles in a variety of diseases, in particular cancer. their sequence homology and functionality. BMP2, BMP4 and For many years, the BMP field has been dogged by the assump- BMP5, BMP6, BMP7 form two subgroups that have been tion that many BMPs are functionally interchangeable. How- extensively studied (1). They mediate their function by binding ever, there is increasing evidence indicating the disparate to two molecules of type 1 (BMPR1 ,BMPR1 , or ACVR1) and a b actions of BMPs (7). For instance, whereas BMP4 and BMP2 two molecules of type 2 (BMPR2, ACVRL1, ActRII, or ActRII ) B seem to facilitate metastasis and invasion in several cancers, serine/threonine kinase receptors (2). Depending on the pres- BMP6 and BMP7 present a suppressive role in metastatic breast ence or absence of preformed BMPR complexes, the signaling is cancer and melanomas (8). Most intriguing is the role of BMP4, mediated via the SMAD family of transcription factors follow- which presents opposing roles depending on the neoplasm ingadefined canonical pathway or by less defined noncanon- (7, 9). In certain malignancies, such as glioblastoma (10) and myeloma (11), high levels of BMP4 are associated with less malignant features while in breast (12), ovarian (13), lung 1Center for Experimental and Molecular Medicine, Academic Medical (14), and prostate (15), BMP4 is involved in epithelial–mes- Center, Amsterdam, the Netherlands. 2AIMM Therapeutics, Academic enchymal transition (EMT) and metastatic behavior. In gastro- 3 Medical Center, Amsterdam, the Netherlands. QVQ BV, Utrecht, the intestinal cancers such as pancreas (16), colon (17), and gastric Netherlands. 4Laboratory for Experimental Oncology and Radiobiol- ogy, Center for Experimental Molecular Medicine, Academic Medical cancer (18), BMP4 has been associated with tumor progression Center, University of Amsterdam, Amsterdam, the Netherlands. and chemotherapy resistance. Therefore, to resolve and inhibit 5 Department of Gastroenterology and Hepatology, Academic Medical the individual action of BMP4 in tumors, the generation of Center, Amsterdam, the Netherlands. potent and specific BMP4 inhibitors is warranted. Note: Supplementary data for this article are available at Molecular Cancer Targeting and regulating BMP signaling at the extracellular Therapeutics Online (http://mct.aacrjournals.org/). receptor level is clinically highly attractive (19). Neutralization Corresponding Author: Kausilia K. Krishnadath, Academic Medical Center, of BMP by several natural antagonists and chemical inhibitors Meibergdreef 9, Amsterdam 1105AZ, the Netherlands. Phone: 31-0- has already been proven successful for the suppression of 205666703; Fax: 31-0-206917033; E-mail: [email protected] metastasis in prostate (15), lung (14), as well as breast cancer doi: 10.1158/1535-7163.MCT-14-0956 (20). Albeit significant progress has been made in generating 2015 American Association for Cancer Research. highly specificandlesstoxicsmall-moleculeinhibitorsthat

www.aacrjournals.org OF1

Calpe et al.

target BMP type I receptors (21–23), these inhibitors are with 100 ng/mL of BMP2, BMP4, BMP5, BMP6, or BMP7 (R&D still nonselective. Therefore, the clinical use of current BMP Systems) for 4 hours with or without VHHs or Noggin at the inhibitors is questionable (24). For instance, in vivo treatment indicated concentrations. C2C12 mouse myoblasts (stably trans- with the natural antagonist Coco or the chemical inhibitor fected with a reporter plasmid for BMP activity) were a kind LDN-193189 results in reactivation of breast cancer (25) or gift of Dr. L. Zilberberg and Dr. D. Rifkin (ref. 30; New York colon tumor burden (26) in mice with a particular mutatio- University, School of Medicine, New York, NY). CL-11 and nal phenotype, respectively. Therefore, these off-target effects Colo-678 were obtained from the COSMIC project at the Sanger could lead to undesirable side effects if used in the clinic. Institute, UK. The rest of the cell lines were obtained from ATCC. Llama-derived antibodies are small proteins of about 15 kDa All cell lines were obtained more than 6 months ago, and no that, as opposed to conventional antibodies, their antigen-bind- authentication has been performed by the authors. ing fragment is formed by just the variable domain of the heavy chain, hence their denomination as VHHs (27). Because of their Western blot unique structure, VHHs can bind specifically and with high Protein isolation and SDS-PAGE separation were performed as affinity to their antigens, often due to their ability to bind to described previously (31). Briefly, EPC2-hTERT and organoids hidden epitopes within grooves or cavities. This peculiar charac- were lysed with 100 mL of M-PER buffer (Sigma-Aldrich), whereas teristic was leveraged to design specific and highly effective the mouse organoids were lysed at a ratio of 1:20 with T-PER inhibitors of BMP4. In the present study, we describe the char- buffer (Sigma-Aldrich). Cell debris were pelleted and supernatant acterization of five anti-BMP4 llama–derived VHHs. Specificity to was collected. For protein detection, the following antibodies different BMPs allowed their classification into 3 groups: VHHs were used: antiphospho-SMAD1/5/8 at 1:500; p-p38 (T180/ specific for BMP4; BMP2, 4; and BMP2, 4, 5, 6. We demonstrate Y182) at 1:500, pAkt (ser473) at 1:1,000, total Akt at 1:500, total that their binding affinities are greater than those published for p38 at 1:500 (Cell Signaling Technology); anti-b-actin at 1:1,000, Noggin, a natural BMP antagonist, and for conventional anti- anti-SMAD4 at 1:500, anti-pERK (Y204) at 1:1,000 (Santa Cruz); BMP4 antibodies. We show that the VHHs effectively inhibit BMP anti-GAPDH at 1:500 (Millipore); anti-BMP7 MAB3541 and anti- downstream signaling and transcriptional activation of BMP BMP6 AF507 (R&D Systems) at 1:1,000; and anti-BMP5 ab10858 target genes. Our epitope and docking experiments, as well as (Abcam), anti-BMP2 500-P195 (PeproTech), and anti-BMP4 mutagenesis analyses, provide structural evidence for their MAB757 (R&D Systems) all at 1:500. affinity and specificity. These studies show that each VHH binds to a different molecular interface within the BMP molecule and, BMP activity luciferase reporter assay subsequently, unveil the binding regions that shape BMP spec- C2C12 cells were plated in 96-well plates at 5 103 cells per ificity. Finally, we show that our VHHs inhibit endogenous BMP well, and cells were allowed to attach overnight. DMEM (100 mL) function as they can replace Noggin in ex vivo organoid cultures with 0.1% BSA was added in each well. Cells were treated in þ as well as increase in vitro chemosensitivity of BMP4 colorectal triplicate with BMPs, VHHs, or controls at the indicated concen- cancer cells. trations for 16 hours. Wells with unstimulated cells or no cells were added as controls. Luciferase activity was measured by adding 100 mL of luciferase substrate solution from the Bright- Materials and Methods Glo Luciferase Assay System (Promega Benelux). After 3 minutes Generation of llama-derived anti-BMP4 VHHs of incubation, luciferase activity was measured with Synergy To generate VHHs specific for BMP4, we immunized two llamas HT Multi-Mode Microplate Reader (Biotek). Luciferase values with 100 mg of recombinant human BMP4 (R&D Systems) on were normalized to background activity (activity of wells with days 0, 14, day 28 and day 35. At day 44, peripheral blood no cells) and represented as ratio to the activity of BMP-stimulated lymphocytes were purified for RNA extraction and library con- cells. struction. Fragments of about 350 bp in size were cloned in a phage display plasmid and transferred to the Escherichia coli strain ELISA TG1. Two rounds of phage selection were performed to select for Concentrations of BMP present in the conditioned mouse VHH binders to BMP4, which were confirmed by ELISA. Eight organoid media were evaluated via ELISA, according to the high binders to BMP4 were then purified from the soluble manufacturers' protocol: BMP2 (RAB0028), BMP4 (RAB0029), periplasmic fractions using Talon beads (Clontech). Fractions BMP5 (RAB0031), and BMP6 (RAB0032), from Sigma-Aldrich. containing purified VHHs were pooled and their purity analyzed by SDS-PAGE and Coomassie staining. Binding to BMP4 was Surface plasmon resonance again confirmed by ELISA. PCR and site-directed mutagenesis Serial dilutions (2.0–10 mg/mL) of purified VHHs were spot- were used to generate the biheads and the mutant monoheads, tedonanamine-specific SensEye gold-film gel-type SPRchip respectively. Specific details on the generation, isolation, purifi- (Ssens), using a Continuous Flow Microspotter (Wasatch cation, and modification of VHHs can be found in Supplementary Microfluidics) as described (32). BMP binding was analyzed Materials and Methods. on an IBIS MX96 (IBIS Technologies) instrument by perform- ing injections with dilution series (0.05–2.0 mg/mL) of recom- Cell culture binant BMPs on the VHH-coated chip. In each injection, BMPs The human hTERT-immortalized esophageal cell line, EPC2- were injected and incubated for 8 minutes, followed by 15 hTERT, was a kind gift of Prof. A. Rustgi (University of Pennsyl- minutes thorough washing with binding buffer (PBS þ 0.05% vania, Philadelphia, PA; ref. 28) and was cultured as previously Tween-20 þ 0.05% sodium azide) to measure dissociation. described (29). To detect phosphorylation of SMAD1/5/8 in vitro, Epitope binning was done by injecting VHH dilutions EPC2-hTERTs were cultured in 24-well plates and activated (2.0 mg/mL) over the chip, immediately after the dissociation

OF2 Mol Cancer Ther; 14(11) November 2015 Molecular Cancer Therapeutics

Generation of Selective Anti-BMP VHHs

step. Injections with blank binding buffer were used as refer- would also inhibit BMP-mediated transcriptional activity, we ence. After each concatenated injection, the chip was regener- used the BMP-responsive C2C12 cells which are stably trans- ated with 10 mmol/L glycine HCl, pH 2.0. Experimental data fected with the BRE luciferase vector containing the mouse were processed with SPRintX software (IBIS Technologies), and DNA-binding protein inhibitor 1 (ID1) promoter, an imme- kinetic parameters were determined using Scrubber2 software diate early gene that is activated by binding of the phosphor- (BioLogic Software). Binding constants were obtained by glob- ylatedSMADtoitspromoterregion(30).AsshowninFig.1B, al fitting to a one-site binding model. all VHHs were able to inhibit BMP4-mediated transcriptional activity to the same level as Noggin. Both C8 and E7 exerted an HADDOCK modeling inhibitory effect in BMP2 transcriptional activation. E7 also HADDOCK software (high-ambiguity-Driven protein–protein inhibited BMP5 and BMP6, albeit to a lesser extent than the DOCKing; ref. 33) was used to model BMP4 binding to the other BMPs. different VHHs. A 3-dimensional (3D) homology model of BMP4 was generated with BMP2 (PBD 1ES7) as a template using Bivalent VHHs lead to increased functional activity Modeller9.9. Structures of the different VHHs were modeled on Hetero- and homodimeric VHHs have superior antigen-neu- the basis of crystal structures of other VHHs, such as 4B5E (for C4) tralizing activities compared with monomers (35). We con- and SJX (for E7 and C8) based on their similarities at the sequence structed three bivalent VHHs with two consecutive peptide level (74%). A detailed description of the residues chosen for spacers (GGGGS) as a linker. In EPC2-hTERT cells, the dimers docking BMP4 to VHHs can be found in Supplementary Materials inhibited BMP-mediated pSMAD1/5/8 activation at much low- and Methods. er concentration compared with the monomers (Supplemen- tary Fig. S1). To evaluate the extent of this increase in potency, Organoid cultures the IC50 values of the different VHHs were determined in a Crypts were isolated from small intestines of wt mice as dose-dependent inhibition assay using C2C12 cells (Fig. 2). previously described (34). After isolation, 500 crypts were C4C4 displayed a 9-fold increase at inhibiting BMP4-mediated resuspended in 50 mL of polymerized Matrigel per well, in a transcriptional activation as compared with monomer C4 and m 24-well plate. Advanced DMEM/F12 (500 L; Invitrogen) medi- the lowest IC50 value amongst all tested VHHs (Fig. 2A). um containing 50 ng/mL EGF (Tebu-BIO), Fc-Noggin (10%), Similarly, bivalent C8 also resulted in a more potent dimer and Fc-Rspondin (20%), both generated in house, was added (Fig.2BandC),aneffectmorestrikingforBMP2ratherthan per well. Where indicated, VHHs were used in place of Fc- BMP4 inhibition. Albeit better than E7, bihead C8E7 proved to Noggin (C4C4 at 100 ng/mL, C8C8 at 500 ng/mL, C8C8 at 5 be similar at inhibiting both BMP2 and BMP4 as compared mg/mL, E7 at 5 mg/mL). As controls, some crypts were cultured in with C8 (Fig. 2D and E), suggesting that C8 might be the the absence of Noggin or VHHs. At day 3, the supernatant was dominant molecule in this bihead. The specificity of the collected and medium was replaced. Lysates for protein extrac- biheads is retained, except for C8C8, which surprisingly, weakly tion were collected at day 5. inhibits BMP5-mediated transcriptional activity at higher doses (Supplementary Fig. S2). The remaining members of the BMP Chemosensitivity subfamily have low sequence similarities to BMP4 (30% HT29 cells were plated in 96-well plates at a density of 10,000 identities). Nevertheless, they were tested to confirm the spec- cells per well. After 24 hours, they were stimulated with the ificity of the monoheads and biheads. As expected, no inhibi- indicated concentrations of 50-fluorouracil (50-FU), oxaliplatin, tion of BMP9, BMP10, or BMP12 was observed by any of the and cisplatin (Sigma-Aldrich) for 48 hours. The VHHs or Noggin VHHs tested (Supplementary Fig. S3A–S3C). Other members of were added at the same time at a concentration of 500 ng/mL. the TGFb superfamilywerealsotestedtofurtherconfirm the Cell viability was measured by adding the Presto Blue Reagent BMP specificity of the VHHs. No inhibition of SMAD2 phos- (Invitrogen) for 2 hours at 37C, after which absorbance was phorylation was observed after TGFb or Activin A activation measured at 520 nm. (Supplementary Fig. S3D–SDF).

Affinity analysis of the VHHs Results BMP specificity and binding affinities of the VHHs was deter- Specificity and functional activity of the anti-BMP4 VHHs mined by surface plasmon resonance (SPR) analysis. Noggin and The immune phage display libraries constructed from BMPR1a were used as controls and bound all BMPs, concurring peripheral blood lymphocytes (PBL) of llamas immunized with published observations (Table 1; refs. 36, 37). Among all with BMP4 resulted in the isolation of five VHHs with high measured monomers, C8 and E7 were the only ones binding to affinity for BMP4 binding. In BMP-responsive EPC2-hTERT BMP2 (Fig. 3A), with slightly higher affinity for BMP4 (Table 1), cells (29), a normal human squamous esophageal cell line explaining the lower IC50 values of C8 and E7 for BMP4 than for with no detectable basal levels of phosphorylated SMAD1/5/8, BMP2 (Fig. 2). The affinities of E7 and C8 to BMP2 are similar all VHHs were able to inhibit BMP4-mediated SMAD1/5/8 but their binding kinetics differ. Whereas E7 binds to BMP2 phosphorylation (Fig. 1A). much faster, C8 dissociates from BMP2 at a much slower pace Because of the high similarities (60%–80%) between the than E7. This could explain why C8 is better than E7 at inhibiting mature regions of BMP4 and other members of the BMP family, BMP2-mediated signaling and transcriptional activation (Fig. 1). we tested the specificity of the VHHs with regard to BMP2, Because of poor coupling to the plate, binding affinities for BMP5, BMP6, and BMP7. We found that whereas A1, C3, and monomer C4 could not be obtained. C4 were strictly BMP4 specific, C8 also inhibited BMP2-medi- The C8 dimers had increased affinity for BMP2 and BMP4 ated signaling (Fig. 1A). To determine whether our VHHs (Table 1). The lower KD values observed for C8C8 and C8E7 result

www.aacrjournals.org Mol Cancer Ther; 14(11) November 2015 OF3

Calpe et al.

A BMP4 BMP2 BMP5 1.5 1.5 1.5

1.0 1.0 1.0 ns

0.5 0.5 0.5 relative to β -actin relative relative to β -actin relative to β -actin relative pSMAD1/5/8 intensity pSMAD1/5/8 intensity pSMAD1/5/8 intensity 0.0 0.0 0.0 A1 C3 C4 C8 E7 A1 C3 C4 C8 E7 A1 C3 C4 C8 E7 Uns Uns Uns BMP5 BMP4 BMP2

pSMAD 1/5/8 pSMAD 1/5/8 pSMAD 1/5/8

β-Actin β-Actin β-Actin

BMP6 BMP7 1.5 1.5

1.0 1.0

0.5 0.5 relative to β -actin relative relative to β -actin relative pSMAD1/5/8 intensity

pSMAD1/5/8 intensity 0.0 0.0

A1 C3 C4 C8 E7 A1 C3 C4 C8 E7 Uns Uns BMP6 BMP7 pSMAD 1/5/8 pSMAD 1/5/8

β -Actin β-Actin B 1.5 BMP4 1.5 BMP2 1.5 BMP5

1.0 1.0 1.0

0.5 0.5 0.5

Relative ID1-promoter activity Relative 0.0 ID1-promoter activity Relative 0.0 ID1-promoter activity Relative 0.0

A1 C3 C4 C8 E7 A1 C3 C4 C8 E7 A1 C3 C4 C8 E7 Uns Uns Uns BMP4 BMP2 BMP5 Noggin Noggin Noggin BMP4 BMP2 BMP5

1.5 BMP6 1.5 BMP7

1.0 1.0

0.5 0.5

Relative ID1-promoter activity Relative 0.0 ID1-promoter activity Relative 0.0

A1 C3 C4 C8 E7 A1 C3 C4 C8 E7 Uns Uns BMP7 BMP6Noggin Noggin BMP6 BMP7

OF4 Mol Cancer Ther; 14(11) November 2015 Molecular Cancer Therapeutics

Generation of Selective Anti-BMP VHHs

mainly from lower dissociation rates as compared with the target a different area within the wrist epitope, a not surprising monomers (Table 1). Interestingly, the association rate for finding due to the small nature of the VHHs and the large C8E7 to both BMP2 and BMP4 is similar to the one observed contact area of the this epitope. Cross-binding with a commer- for C8, confirming the assumption that C8 is the prevalent cial anti-BMP4 (R&D) was detected on all VHHs and controls, molecule in this bihead. The BMP-binding affinities of biheads demonstrating that the epitope of this anti-BMP4 is remarkably C4C4 and C8C8 were higher than those found for Noggin or for a differentthanthatoftheVHHsanditresidesinanarea commercially available conventional anti-BMP4 antibody from nonoverlapping Noggin or BMPR1a (Fig. 3B). R&D Systems (MAB757; Table 1). In keeping with the functional experiments, E7, C8C8, and Docking models of C4, C8, and E7 C8E7 bound to BMP5. However, the affinities for these inter- To identify putative binding sites of the VHHs to BMP4, actions are low, which might explain why functional inhibition modeling using HADDOCK software was performed (33). of BMP5 by these VHHs is not as striking as the one for BMP4. Modeling of C4 to BMP4 showed that C4 interacts with the Notably, and in contrast to their monomers, unspecificbinding "hydrophobic groove" of the wrist epitope of BMP4 (Fig. 4). In of the bivalents to diverse BMPs was observed. C4C4 was this model, loop 2 of BMP4 packs mainly hydrophobically against shown to bind BMP2, C8C8 to BMP5, and C8E7 to BMP7 residues from both CDR2 and CDR3 of C4. P50 of BMP4 sustains (Fig.3A).Becauseofthepresenceoftwobindingsitesper hydrophobic interactions with I57 of VHH C4 and D46 of BMP4 bihead, the change for rebinding BMP after initial dissociation forms a hydrogen bond with S102 of C4 (Fig. 4A). In contrast, the is higher, explaining the increase in potency but the loss in preb1 and b1 regions mainly interact with CDR3 residues. Of note specificity. These equilibrium dissociation constants, however, is D99 that forms salt bridges with both K12 and R15 of BMP4, a befall within the ranges of 2 to 123 nmol/L, 300 times higher similar interaction was observed between K15 of BMP2 and D46 than the ones observed for BMP4 binding (10–100 pmol/L), of BMPR1a (37). which might explain the lack of functional implications for Modeling of C8 to BMP4 revealed that C8 interacts with the these interactions (Table 1). other contact point of the wrist epitope. In this model, C8– BMP4 binding appears to be driven by interactions between Epitope binning residues in the CDR3 region of C8 and the "hydrophobic The solved crystal structures of BMP2 and BMP7 with pocket" of the wrist epitope of BMP4 (Fig. 4B). In the hydro- their receptors have provided the basis of receptor binding phobic surface of the ahelix of BMP4A, L66 sticks out to interact to BMP ligands (37–39). Two epitopes are involved in the with T105 of C8. Residues in loop 1 of BMP4B provide a binding of BMPs to their receptors (Supplementary Fig. S4): hydrophobic area allowing for interactions like P109 with the "wrist epitope," a large concave area involved in the W31 of BMP4 and the double salt bridge between D30 of binding to type 1 receptors (40) and the "knuckle epitope," BMP4 and R106 of C8 (Fig. 4B). Also, F110 and F102 of C8 a convex molecular interface involved in type 2 receptor make multiple contacts with the area of BMP4 containing D30. binding (41). Y107 of C8 forms a hydrogen bond with K101 located at the In an effort to determine the binding epitopes and explain inner side of the b8 strand of BMP4. the different BMP specificities between C4, C8, and E7, we HADDOCK modeled both CDR1 and CDR2 residues of E7 used SPR sandwich cross-binding or "epitope binning" assays to be involved in binding to the knuckle epitope of BMP4 (Supplementary Table S1). Notably, besides BMPR1a, all mole- (Fig. 4C). The nitrogen atom from the main chain of T56 in E7 cules showed weak self-binding in the cross-binding assay forms a hydrogen bond with S88 of BMP4. L57 from E7 is also (Fig. 3B). C8 and E7 do not share the same epitope, as they could in close proximity to S88. V32 from the CDR1 of E7 is in close cross-bind with each other in both directions. Furthermore, proximity to a hydrophobic area formed by the residues whereas both E7 and C8 compete with Noggin for BMP4 binding, located in the outer sides of the b3 (A34) and b8strands E7 but not C8 could still bind to BMPR1a-bound BMP4. This (L90) of BMP4. Finally, W47 from E7 interacts with Q39 of suggests that the epitope of C8 maps to the wrist epitope, whereas BMP4 forming a hydrogen bond. This interaction is reminis- the E7 epitope maps to a region different to that of BMPR1a but cent of the one in type 2 receptor ActR-IIB and BMP2, in which targeted by Noggin. Because Noggin binds to both BMPR1a and W60 (a highly conserved residue among type 2 receptors) BMPR2 epitopes (42), it is most likely that the E7 epitope extends to the hydrophobic core of the knuckle epitope of therefore maps to the BMPR2-binding site. BMP2 (43). Because of technical difficulties of coating C4 into the SPR Mutation analysis was performed to confirm the HADDOCK chip, we decided to use C4C4 for these experiments. C8 and E7, models. Some of the residues shown to be involved in VHH– but not Noggin or BMPR1a,wereabletoweaklybindBMP4 BMP4 binding were substituted singly by in vitro mutagenesis. bound to C4C4 (Fig. 3B). Therefore, C4C4 binds an epitope Except for I57 of C4, all mutations affected the potency of the that is different from C8 and E7 but overlaps with that of different VHHs (Fig. 4D and E), suggesting that the mutated Noggin and BMPR1a. These results suggest that C4 and C8 residues are important binding determinants.

Figure 1. Anti-BMP4 VHH speciﬁcity to the different BMPs. A, phosphorylation of SMAD1/5/8 in EPC2-hERT cells after activation with 100 ng/mL of human BMP2, 4, 5, 6, or 7 in the presence of 5 mg/mL VHHs. Error bars, SDs of the mean, calculated from 3 independent experiments. B, C2C12 cells were activated with human BMPs in the presence or absence of 5 mg/mL of VHHs or human Fc-Noggin. After 16-hour incubation, luciferase activity was assayed. Error bars, SDs of the mean, calculated from 3 independent experiments, with experimental triplicates each. , P 0.0001. Statistical analysis was done using a two-tailed P test.

www.aacrjournals.org Mol Cancer Ther; 14(11) November 2015 OF5

Calpe et al.

A BMP4 activation

1.5 C4 IC50 = 7.5 ng/mL (500 pmol/L) C4C4 IC50 = 1.3 ng/mL (45 pmol/L)

1.0

0.5

0.0 0.1 1 10 100 1,000 Concentration (ng/mL) Relative ID1-promoter activity Relative –0.5

BCBMP4 activation BMP2 activation IC = 1.6 μg/mL (106 nmol/L) C8 IC50 = 7.9 ng/mL (527 pmol/L) C8 50 C8C8 IC = 3.8 ng/mL (126 pmol/L) C8C8 IC50 = 62 ng/mL (2 nmol/L) 1.5 50 1.5 Figure 2. IC50 curves of VHH biheads and 1.0 1.0 monoheads. C2C12 cells were activated with human BMP4 at 5 ng/mL (A, B, D) or BMP2 at 50 ng/mL (C and E) with 0.5 0.5 increasing concentrations of VHHs. Data are representative from 2 experiments and with experimental 0.0 0.0 triplicates each. 0.1 1 10 100 1,000 10 100 1,000 10,000 Concentration (ng/mL) Concentration (ng/mL) Relative ID1-promoter activity Relative –0.5 ID1-promoter activity Relative –0.5

DEBMP4 activation BMP2 activation μ E7 IC50 = 176 ng/mL (12 nmol/L) E7 IC50 = 1.6 g/mL (106 nmol/L) 1.5 C8E7 IC50 = 31.3 ng/mL (1 nmol/L) 1.5 C8E7 IC50 = 356 ng/mL (11 nmol/L)

1.0 1.0

0.5 0.5

0.0 0.0 0.1 1 10 100 1,000 10 100 1,000 10,000 Concentration (ng/mL) Concentration (ng/mL) Relative ID1-promoter activity Relative –0.5 ID1-promoter activity Relative –0.5

VHHs inhibit endogenous BMP activity in ex vivo organoid natants of these cultures confirmed the functionality of these cultures VHHs (Fig. 5C). Whereas specific BMP4 inhibition provided by The ability of our VHHs to provide functional inhibition of C4C4 resulted in a partial decrease in the BRE luciferase endogenous BMP was tested in mouse intestinal cultures of activity, concomitant inhibition of BMP2 and BMP4 was suf- stem cells. Inhibition of BMP activity is one of the require- ficient to result in an almost complete blockage of BRE lucif- ments to maintain intestinal stem cells in ex vivo cultures of erase activity. Further inhibition of BMP5 and BMP6 in these "organoids," structures that mirror the intestinal crypt villus cultures, by the remaining VHHs (or by Noggin), did not result units. In these cultures, BMP inhibition is usually provided by in a further decrease in BRE luciferase activity, revealing a the addition of the natural BMP inhibitor Noggin to the media minor role for other BMPs in these cultures (Fig. 5C). Although (34). ELISA (Fig. 5A) and Western blot (Fig. 5B) analysis these results manifest the capacity of C4C4 to inhibit endog- revealed that BMP2, BMP4, BMP5, and BMP6 are expressed enous BMP4 function, they suggest that inhibition of BMP4 is and secreted in organoids cultured for 3 days in the absence of not sufficient to maintain intestinal crypt cultures. Indeed, the Noggin. To check the ability of the VHHs to provide inhibition number of organoids counted from the crypts cultured for 5 of these endogenous BMPs and to sustain organoid cultures ex days with C4C4 was as low as the crypts cultured without vivo, freshly isolated crypts were cultured with the different Noggin (Fig. 5D). Furthermore, the average size of these VHHs. Analysis of the BRE-luciferase activity in the super- organoids was small and presented features comparable to the

OF6 Mol Cancer Ther; 14(11) November 2015 Molecular Cancer Therapeutics

Generation of Selective Anti-BMP VHHs

Table 1. Kinetic constants for BMP binding BMP2 BMP4 BMP5 BMP6 BMP7

Molecules ka kd KD ka kd KD Molecules KD KD KD A1 —— —51.4 29.8 574 C8C8 123,000 —— (6.0) (9.6) (184) (6,000) C3 —— —28.3 43.7 1,980 E7 197,000 — 80,000 (9.1) (4.3) (970) (44,000) (10,000) C4 ND ND ND ND ND ND C8E7 95,000 50,000 31,000 C4C4 37 70.6 2,150 19.2 0.2 9.7 (7,000) (20,000) (2,000) (11.5) (3.6) (760) (2.4) (0.03) (1.6) hNoggin 68,000 50,000 80,000 C8 21.9 19.1 880 21.6 8.1 392 (5,000) (10,000) (10,000)

(2.2) (0.3) (100) (1.6) (3.2) (180) BMPR1a 60,000 2,400 131,000 C8C8 20.2 1.3 62.8 24.2 0.7 31 (4,000) (3,000) (9,000) (5.0) (0.3) (3.1) (6.5) (0.2) (9) E7 44.6 47.2 1,070 60.1 30.8 512 (3.2) (3.1) (130) (3.1) (5.1) (73) C8E7 19.4 10.7 554 34.6 3.4 99 (3.1) (1.0) (38) (6.6) (0.8) (12) hNoggin 17.1 16.2 961 47.2 3.9 91 (2.5) (0.9) (91) (11.9) (0.8) (38)

BMPR1a 28.2 14 498 16.3 9.2 635 (2.0) (2.0) (33) (4.5) (2.8) (268) aBMP4 —— —22.8 36.2 1,590 (0.7) (0.4) (30) 4 1 1 5 1 NOTE: VHHs and controls are immobilized on a chip. ka in 10 s (mol/L) , kd in 10 s , KD in pmol/L. Abbreviations: —, no binding detected; ND, not determined (VHH could not be immobilized).

ones observed for the organoids grown in the absence of HT29, was used to test the effect of BMP4 inhibition in terms of Noggin (Fig. 5E). In contrast, inhibition of BMP2 and BMP4 proliferation. No effect was observed when cultured with the provided by the addition of C8C8 (at 500 ng/mL) was enough VHHs or Noggin alone (data not shown). However, both VHHs to sustain self-renewal of intestinal crypts. These organoids C4C4andC8C8provedtoincreasethechemosensitivitywhen presented the same morphologic characteristics as the ones combined with different types of chemotherapeutics. This is cultured with Noggin, manifested by the presence of several demonstrated by the fact that combining the VHHs with crypt-like budding structures, where the intestinal stem cells carboplatin, 50-FU, or oxaliplatin had a significant stronger together with the Paneth cells reside (Fig. 5E). Additional effect on cell viability as compared with chemotherapy alone inhibition of BMP5 provided by C8C8 at higher doses (5 (Fig. 6C–E). The effect on cell viability by Noggin was not as mg/mL) or BMP5, 6 by E7 did not provide superior growth effective as observed for C4C4 or C8C8. As SMAD-independent advantage to the cultures when compared with C8C8 at lower signaling has been shown to mediate BMP-induced chemore- dose. Similar findings were observed when cultured with sistance (44), we next studied whether the VHHs would inhibit Noggin (Fig. 5D and E). In sum, these experiments show that noncanonical BMP4 signals in HT29 cells. Figure 6F shows that inhibition of both endogenous BMP2 and BMP4 is required to C4C4 and C8C8 inhibit the phosphorylation of the kinases sustain self-renewal of small intestine stem cells in vitro and p38,Akt,aswellasERK.AsHT29cellsharbormutationsinthe this can effectively and selectively be provided by C8C8, the SMAD4 gene, the SMAD pathway is inactive in these cells and BMP2, 4–specificVHH. therefore the inhibition of pSMAD by the VHHs has no functional consequences (Fig. 6F). Together, these results show that Inhibition of BMP4 by the VHHs increase the chemosensitivity the VHHs increase chemosensitivity of HT29 cells through the þ of BMP4 colorectal cancer cells inhibition of noncanonical BMP4 signals. BMPsignalinghasbeenshowntobeimplicatedinchemore- sistance in colorectal cancer cell lines (44). We wanted to determine whether this effect could be attributable to BMP4 Discussion being secreted by the cancer cells and, if therefore, could be VHHs are emerging as promising clinical tools with a grow- inhibited by our BMP4-specific VHHs. A panel of colorectal ing number of benefits compared with conventional antibodies cancer cells was tested for the presence of active BMPs in their (45) such as low immunogenicity, cost-effective production, conditioned media (Fig. 6A). To that end, the C2C12 cells were high stability, consistent activity, and ease of manipulation. cultured with supernatants from different cancer cell lines. Their advantageous characteristics have prompted their usage When normalized for the BMPs present in normal fetal calf for research, diagnostic, and therapeutic applications (46, 47). serum, we found a high variability in BRE luciferase activity Their small size and structural properties make VHHs better across cell lines, indicating differences in the amounts of BMPs adapted to bind to epitopes in hidden grooves, such as the ones being secreted by the cell lines. Incubation of the conditioned present in the homodimeric BMP structures. In our study, we media of the cancer cell lines with the different VHHs showed have generated and characterized three types of BMP-specific that certain lines were secreting exclusively BMP4 and not inhibitors and compared their activity with that of the most BMP2 or other BMPs, as C4C4 was already sufficient to potent and generally expressed natural BMP inhibitor Noggin completely inhibit this signal (Fig. 6B). One of those cell lines, (Fig. 6G).

www.aacrjournals.org Mol Cancer Ther; 14(11) November 2015 OF7

Calpe et al.

A BMP2 BMP4 120 E7 100 600 E7 600 C4C4 90 80 C8C8 C8 400 60 400 60 C8 C8E7 C4C4 A1 C8E7 40

Response (RU) 30 Response (RU) 200 Response (RU) Response (RU) 200 20 C8C8 A1/non-binding 0 non-binding non-binding 0 0 0 non-binding 0 300 600 900 1,200 0 300 600 900 1,200 0 300 600 900 1,200 0 300 600 900 1,200 Time (s) Time (s) Time (s) Time (s) monoheads biheads monoheads biheads BMP5 BMP6 BMP7 40 200 150 C8E7 C8E7 C8E7 30 150 100 20 100

E7 10 50 E7 Response (RU) 50 Response (RU) Response (RU) non-binding C8C8 0 non-binding 0 non-binding 0 0 300 600 900 1,200 0 300 600 900 1,200 0 300 600 900 1,200 Time (s) Time (s) Time (s) mono- + biheads mono- + biheads mono- + biheads

B C8 E7 C4C4 750 2,000 αBMP4 αBMP4 600 αBMP4 1,600 500 400 1,200

250 C8 800

Response (RU) E7 Response (RU) 200 Response (RU)

C4C4 400 C4C4 + C8 non-binding E7 C4C4 + C8 0 non-binding 0 0 non-binding 0 500 1,000 1,500 2,000 0 500 1,000 1,500 2,000 0 500 1,000 1,500 2,000 Time (s) Time (s) Time (s)

mNoggin BMPR1a aBMP4

300 120 120 E7 αBMP4 C8 90 αBMP4 90 200

60 60

E7 100 Response (RU) 30 Response (RU) 30 Response (RU) mNoggin αBMP4 C4C4 non-binding non-binding mNoggin 00 0 non-binding 0 400 800 1,200 0 400 800 1,200 0 400 800 1,200 Time (s) Time (s) Time (s)

Figure 3. Kinetic analysis for BMP binding. A, binding of VHHs to the different BMPs was tested in a direct SPR assay. The graphs show two replicate ligand injections at just one concentration of ligand (0.4 mg/mL for BMP2 and BMP4 and 2.0 mg/mL for BMP5, BMP6, and BMP7). For clarity, just one replication is shown. B, "epitope binning" of VHH-coated chip plates. BMP4 (1.0 mg/mL) was injected on the chip followed by the injection of a second VHH (2.0 mg/mL) or the corresponding controls.

OF8 Mol Cancer Ther; 14(11) November 2015 Molecular Cancer Therapeutics

Generation of Selective Anti-BMP VHHs

A B C4

BMP4A BMP4A C8

BMP4B BMP4B

Wrist epitope Wrist epitope Hydrophobic groove Hydrophobic pocket

Figure 4. HADDOCK docking of the VHHs to BMP4 binding. Ribbon diagrams of the docking of C4 (A), C8 (B), and E7 (C) binding to BMP4. Hydrogen bonds and salt bridges are represented as dashed lines. The enlarged ribbon diagrams show the residues constituting the molecular interfaces. Residues corresponding to the VHH paratope are represented as magenta sticks. Residues forming the BMP4 epitope are represented as yellow sticks and CD2.0 labeled in green (A), orange (B), or red (C). Function of the generated mutant variants in inhibiting ID1 promoter 1.5 activity in C2C12 cells after stimulation with BMP2 (50 ng/mL; D) or BMP4 1.0 (5 ng/mL; E) for 16 hours. Uns, unstimulated C2C12 cells. Values Relative 0.5 calculated from 3 independent E7

experiments, with experimental ID1-promoter activity triplicates each. BMP4A 0.0 C4 BMP4B Uns C8 E7 BMP2 C4 I57C4 D99 C8 P109 C8 Y107A E7 V32YE7 L57Y BMP2 E 1.5

Knuckle epitope 1.0 Relative 0.5 ID1-promoter activity

0.0

C4 C8 E7 Uns BMP4 C4 I57 C4 D99 C8 P109 E7 V32YE7 L57Y C8 Y107A BMP4

One highly interesting and unique group of VHHs that we knowledge, these are the first VHHs that can effectively inhibit classified as the C4-like antibodies inhibits BMP4-mediated func- BMP4 function, while leaving important homeostatic functions of tion only, in contrast to Noggin, that also inhibits BMP2, 5, 6, and the other BMPs intact. Unselective inhibition of BMP signaling in 7. The homo-bihead C4C4 proved to be the VHH with the highest mice has been shown to increase tumor burden (26) and activa- affinity to BMP4, 10 times higher than that of Noggin. To our tion of metastatic dormancy of breast cancer cells (25). Thus,

www.aacrjournals.org Mol Cancer Ther; 14(11) November 2015 OF9

Calpe et al.

A B

OrganoidsDuodenumJejunumIleum 300 BMP2

200 BMP4

100 BMP5

N.D BMP6 0 BMP concentration (pg/mL) BMP concentration

BMP4 BMP2 BMP5 BMP6

C D

80 20

60 15

10 40

5 20 Number of organoids Relative ID1-promoter activity Relative 0 0 Backg. No Noggin BMP4 BMP BMP BMP No Noggin BMP4 BMP BMP BMP Noggin inh.2/4 2/4/5 2/4/5/6Noggin inh. 2/4 2/4/5 2/4/5/6 C4C4 inh.inh. inh. C4C4 inh. inh. inh. C8C8 C8C8 E7 C8C8 C8C8 E7 500 ng/mL 5 μg/mL 500 ng/mL 5 μg/mL

E No Noggin Fc-Noggin BMP4 inhibition C4C4

BMP2/4 inhibition BMP2/4/5 inhibition BMP2/4/5/6 inhibition C8C8 500 ng/mL C8C8 5 mg/mL E7

Figure 5. VHH biheads inhibit endogenous BMP function. A, ELISA determinations of BMP secretion in mouse intestinal organoids cultured in the absence of Nogginfor3days.BMP5 levels were not detected (N.D). B, Western blot analysis of lysates of mouse intestinal organoids grown under No Noggin conditions for 3 days. Lysates from mouse duodenum, jejunum, and ileum were used as positive controls. C, BMP activity of conditioned media of organoids cultured at the indicated conditions for 3 days. Error bars, SDs of the mean as compared with No Noggin (black bar). Data collected from at least 5 experiments with experimental triplicates each. , P 0.0001; , P 0.01; , P 0.1. Statistical analysis was done using a two-tailed P test. D, number of organoids formed in cultures of mouse intestinal organoids grown for 5 days at the indicated conditions. E, representative bright-ﬁeld images of mouse intestinal organoids grown under the indicated conditions for 5 days.

OF10 Mol Cancer Ther; 14(11) November 2015 Molecular Cancer Therapeutics

Generation of Selective Anti-BMP VHHs

A B 15 Colon cancer cell lines 1.5 HT-29 HCT116

10 1.0

0.5 5

in conditioned media 0.0 in conditioned media Noggin

Relative ID1-promoter activity Relative 0 Noggin C4C4 C8C8 Relative ID1-promoter activity Relative

C4C4 CL11 Media HT29 − SW837 0.5 Colo320 Colo678 HCT116 C8C8 HUTU-80 NCI-H716

C HT29 + carboplatin DEHT29 + 5'FU HT29 + oxaliplatin Chemotherapy alone 140 200 100 + C4C4 + C8C8 120 80 150 + Noggin 100 60 100 80 40

Viability (%) Viability (%) 50 Viability (%) 60 20

40 0 0 0 10 100 1,000 0 1 10 100 1,000 0 1 10 100 1,000 Carboplatin concentration (μmol/L) 5'FU concentration (μmol/L) Oxaliplatin concentration (μmol/L)

FG HT29 BMPR1a Wrist Uns. +C4C4 +C8C8 Hydrophobic SMAD1/5/8 HT29 HCT116 Wrist SMAD4 groove Hydrophobic pocket Total GAPDH SMAD1/5/8 BMP4A HT29 HT29 Knuckle epitope

Uns. Uns. +C4C4 +C8C8 +C4C4 +C8C8 pP38 pAkt BMPR2

Total P38 Total Akt BMP4B C4 C8 epitope epitope E7 epitope Uns. +C4C4 +C8C8 pERK

GAPDH

Figure 6. Anti-BMP4 VHHs inhibit cancer-derived BMP signals. A, C2C12 cells were incubated with conditioned media from a panel of different colon cancer cell lines. B, C2C12 cells were incubated with conditioned media of the indicated cancer cell lines. C4C4, C8C8, or human Fc-Noggin were added at the same time at a concentration of 500 ng/mL for 16 hours. HT29 cells were treated with carboplatin (C), 50-FU (D), or oxaliplatin (E) at the indicated concentrations with or without VHHs or Noggin. After 48 hours, the viability was measured. Experiments were repeated at least 3 times with experimental triplicates each. , P 0.01; , P 0.1. Statistical analysis was done using a two-tailed P test. F, HT29 cells were starved overnight and treated with the VHHs or Noggin at a concentration of 500 ng/mL for 1 hour. SMAD4 expression was tested on lysates of resting HT29 and HCT116. Blots are representative of at least 3 experiments. G, space-ﬁlled view of BMP4 binding to BMPR1a and BMPR2 where the proposed model of BMP4-VHH binding is shown.

www.aacrjournals.org Mol Cancer Ther; 14(11) November 2015 OF11

Calpe et al.

when using BMP inhibition as therapy, avoiding the deleterious of the knuckle epitope are invariant between the BMPs, in side effects of unchecked cancer stem cell proliferation is of major particular, S88 at the center is highly conserved (43). This importance. In the intestinal organoid cultures, inhibition of explains why targeting this region renders E7 nonselective to BMP4 alone by C4C4 does not affect stem cell proliferation. We BMPs. Moreover, the knuckle epitopeexhibitsaconcavehydro- also found that in SMAD4 colorectal cancer cells, inhibition of phobic area with no deep pockets that perfectly complements the BMP4 noncanonical pathway by C4C4 increases the chemo- theconvexnatureofE7.Becauseofthestructuralconformation sensitivity to several chemotherapeutic agents. As this mutational of the knuckle epitope, binding of type 2 receptors to BMPs is profile has also been linked to other BMP-mediated oncogenic characterized as low affinity binding (43), a notion that might functions in colorectal cancer such as migration and invasion explain the lower affinity of E7 in comparison to the other (48), it is tempting to speculate that those features would also be VHHs. Albeit present, the binding of C4 on C8-or E7-bound inhibited by C4C4 treatment. Therefore, C4-like antibodies have BMP4 coated chips was not as striking as the one observed an important potential for clinical, diagnostic, and therapeutic between C8 and E7 (Fig. 3B). This could be explained by the uses in a subset of colorectal cancers. As BMP4 has also been effect of a small allosteric competition between C4 and the shown to contribute to carcinogenesis in breast (12), lung (14), other two VHHs, as the hydrophobic groove of the wrist prostate (15), and gastric (18) cancers, the oncologic applicability epitope is in close proximity to the other two binding interfaces of C4-like antibodies is considerable. targeted by C8 and E7 (Fig. 6G). C8C8 belongs to a second group of inhibitors (C8-like), One of the limitations of our HADDOCK modeling is the which inhibit both BMP2 and BMP4 action, with affinities lack of crystal structures for BMP4 and the VHHs. Nevertheless, higher to those found for Noggin. At lower doses, C8C8 can mutational analysis of the paratopes of the VHHs supports our be used as a potent BMP2, 4–specific inhibitor. This effect is modeling data (Fig. 4D and E). Furthermore, experiments with clearly exemplified in our organoid experiments, in which heterodimer BMP4 and BMP2 confirm that whereas C4 and E7 C8C8 was as effective in maintaining intestinal organoid cul- target an epitope formed by one BMP monomer only, C8 binds turesasNoggin.Becauserecombinant Noggin is expensive, to an epitope formed by two monomers as it is unable to inhibit bihead C8C8 offers an easy accessible and cheaper alternative heterodimers of BMP4 or BMP2 (Supplementary Fig. S5A and for ex vivo stem cell cultures. S5B). E7 is the most promiscuous of the studied VHHs, as its A notable result of our studies is the fact that all VHHs compete specificity is extended to BMP2, 4, 5, and 6, sharing a similar for binding with Noggin. This is explained by the wide reach of pattern of inhibition with Noggin (49), albeit presenting a much contact points of Noggin, which simultaneously masks both lower potency. Because homobivalency of C8 and C4 has resulted BMPR1a and BMPR2 epitopes (42). This remarkable structural in higher efficacy for binding and potency, it is tempting to reach is conserved among BMP antagonists and might explain speculate that the generation of E7E7 might result in a homo- their lack of BMP specificity (50). While the binding interface of bihead that will closely resemble and share the same BMP- Noggin to BMP is about 1,400 Å2 (42), the size of a VHH paratope inhibitory capabilities of Noggin. is usually 700 Å2 (51). The generation of highly effective VHHs Indiscriminate binding for both receptors and antagonists is that target small areas within the epitopes for BMPR binding one of the most puzzling features of the BMP family. While therefore demonstrates that neutralization of small areas is previous studies hypothesized that only a small subset of enough to antagonize ligand activity. An interesting result is the residues might shape binding specificity and affinity (40), our fact that the commercial anti-BMP4 antibody (R&D Systems) results have provided important evidence of the location of presents lower affinity than our antibodies. This antibody targets such residues. Our epitope binning experiments and subse- a BMP4-specific area not recognized by Noggin or BMPR1a. quent HADDOCK modeling have shown that each VHH binds Therefore, the N-terminal heparin-binding domain (refs. 52, to a different BMP4–BMPR binding region (Fig. 6G). Because 53; highly different between BMP2 and BMP4) represents a likely the other BMPR receptors might share similar binding mechan- region to be targeted by this antibody (Supplementary Fig. S4B). isms as BMPR1a andBMPR2,itistemptingtospeculatethatthe Indeed, functional experiments with a BMP4 lacking the heparin- VHHs would also inhibit their binding to BMP4. binding domain, confirms this assumption (Supplementary Fig. C4-like VHHs target the hydrophobic groove of the wrist epitope S5C). These data therefore seem to indicate the targeting the of BMP4, a BMPR-binding interface composed of one BMP4 small BMPR-binding area would provide higher affinity than monomer and presenting differences in residues between BMP2 targeting the heparin-binding domain. and BMP4 (Supplementary Fig. S4B). The docking model There is now overwhelming evidence supporting the notion showed that these nonconserved residues are involved in bind- that large functional differences between highly similar BMPs ing with residues from the CDR2 and CDR3 of C4, therefore exist. Besides cancer, BMPs are also involved in a variety of validating its BMP4 specificity. Interestingly, residues located in diseases, including bone abnormalities, cardiovascular, and met- loop 2 of the groove epitope represent a hotspot of binding to abolic disorders. Because BMPs present opposing functions, the BMPR1a (36), which could explain the high affinity binding importance of specific inhibition for each individual BMP extends observed for C4C4. C8-like VHHs target the hydrophobic pocket of beyond the oncology field. We have generated small, potent, and the wrist epitope of BMP2 and BMP4, an epitope formed by specific inhibitors of BMP molecules that target the contact points residues from both monomers. This is an area in which non- involved in binding to the receptors. In particular, our studies identical and nonisofunctional residues between BMP2, 4 and have underscored a previously unappreciated molecular interface BMP5, 6, 7 reside (Supplementary Fig. S4B), which explains that ultimately dictates BMP specificity. By targeting a small region why C8 binds only to one BMP subgroup. Finally, E7 targets the within the large wrist epitope, namely the hydrophobic groove, knuckle epitope, composed of residues from one monomer. we have developed a highly potent BMP4-specific antibody that Interestingly, all the binding determinants located at the core could represent a promising therapeutic strategy for sensitizing

OF12 Mol Cancer Ther; 14(11) November 2015 Molecular Cancer Therapeutics

Generation of Selective Anti-BMP VHHs

þ BMP4 colorectal cancer cells to chemotherapy. Its novel struc- Writing, review, and/or revision of the manuscript: S. Calpe, K. Wagner, tural format makes is remarkably suited to overcome the limita- M. El Khattabi, L. Rutten, C. Zimberlin, E. Dolk, H. Spits, K.K. Krishnadath tions that affect the clinical and research applications of current Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): M. El Khattabi, E. Dolk BMP inhibitors. Study supervision: K.K. Krishnadath

Disclosure of Potential Conﬂicts of Interest Acknowledgments No potential conﬂicts of interest were disclosed. The authors thank laboratory members of the K. Krishnadath's laboratory for reading the manuscript. They also thank colleagues at the LEXOR lab for Authors' Contributions providing the supernatants of the cancer cell lines. Conception and design: S. Calpe, C.T. Verrips, J.P. Medema, K.K. Krishnadath Development of methodology: S. Calpe, K. Wagner, M. El Khattabi, E. Dolk, Grant Support K.K. Krishnadath This work was supported by the European Research Council starting grant Acquisition of data (provided animals, acquired and managed patients, Targets4Barrett (S. Calpe and K.K. Krishnadath). provided facilities, etc.): S. Calpe, K. Wagner, M. El Khattabi, C. Zimberlin Analysis and interpretation of data (e.g., statistical analysis, biostatistics, Received November 10, 2014; revised August 19, 2015; accepted August 24, computational analysis): S. Calpe, K. Wagner, L. Rutten, K.K. Krishnadath 2015; published OnlineFirst September 8, 2015.

References 1.BragdonB,MoseychukO,SaldanhaS,KingD,JulianJ,NoheA. 18. Ivanova T, Zouridis H, Wu Y, Cheng LL, Tan IB, Gopalakrishnan V, et al. Bone morphogenetic proteins: a critical review. Cell Signal 2011;23: Integrated epigenomics identifies BMP4 as a modulator of cisplatin sen- 609–20. sitivity in gastric cancer. Gut 2013;62:22–33. 2. Miyazono K, Kamiya Y, Morikawa M. Bone morphogenetic protein 19. Hong CC, Yu PB. Applications of small molecule BMP inhibitors receptors and signal transduction. J Biochem (Tokyo) 2010;147: in physiology and disease. Cytokine Growth Factor Rev 2009;20: 35–51. 409–18. 3. Mueller TD, Nickel J. Promiscuity and specificity in BMP receptor activa- 20. Owens P, Pickup MW, Novitskiy SV, Giltnane JM, Gorska AE, Hopkins CR, tion. FEBS Lett 2012;586:1846–59. et al. Inhibition of BMP signaling suppresses metastasis in mammary 4. Ehrlich M, Horbelt D, Marom B, Knaus P, Henis YI. Homomeric and cancer. Oncogene 2015;34:2437–49. heteromeric complexes among TGF-b and BMP receptors and their roles in 21. Hao J, Ho JN, Lewis JA, Karim KA, Daniels RN, Gentry PR, et al. In vivo signaling. Cell Signal 2011;23:1424–32. structure-activity relationship study of dorsomorphin analogues identifies 5. Ten Dijke P, Korchynskyi O, Valdimarsdottir G, Goumans M-J. Controlling selective VEGF and BMP inhibitors. ACS Chem Biol 2010;5:245–53. cell fate by bone morphogenetic protein receptors. Mol Cell Endocrinol 22. Engers DW, Frist AY, Lindsley CW, Hong CC, Hopkins CR. Synthesis and 2003;211:105–13. structure–activity relationships of a novel and selective bone morphoge- 6. Miyazono K, Maeda S, Imamura T. BMP receptor signaling: transcriptional netic protein receptor (BMP) inhibitor derived from the pyrazo- targets, regulation of signals, and signaling cross-talk. Cytokine Growth lo[1.5-a]pyrimidine scaffold of Dorsomorphin: the discovery of ML347 Factor Rev 2005;16:251–63. as an ALK2 versus ALK3 selective MLPCN probe. Bioorg Med Chem Lett 7. Ehata S, Yokoyama Y, Takahashi K, Miyazono K. Bi-directional roles of 2013;23:3248–52. bone morphogenetic proteins in cancer: another molecular Jekyll and 23. Sanvitale CE, Kerr G, Chaikuad A, Ramel M-C, Mohedas AH, Reichert S, Hyde? Pathol Int 2013;63:287–96. et al. A new class of small molecule inhibitor of BMP signaling. PLoS ONE 8. Singh A, Morris RJ. The Yin and Yang of bone morphogenetic proteins in 2013;8:e62721. cancer. Cytokine Growth Factor Rev 2010;21:299–313. 24. Vogt J, Traynor R, Sapkota GP. The specificities of small molecule inhibitors 9. Kallioniemi A. Bone morphogenetic protein 4—a fascinating regulator of of the TGFb and BMP pathways. Cell Signal 2011;23:1831–42. cancer cell behavior. Cancer Genet 2012;205:267–77. 25. Gao H, Chakraborty G, Lee-Lim AP, Mo Q, Decker M, Vonica A, et al. The 10. Piccirillo SGM, Reynolds BA, Zanetti N, Lamorte G, Binda E, Broggi G, et al. BMP inhibitor Coco reactivates breast cancer cells at lung metastatic sites. Bone morphogenetic proteins inhibit the tumorigenic potential of human Cell 2012;150:764–79. brain tumour-initiating cells. Nature 2006;444:761–5. 26. Whissell G, Montagni E, Martinelli P, Hernando-Momblona X, Sevillano 11. Holien T, Sundan A. The role of bone morphogenetic proteins in myeloma M, Jung P, et al. The transcription factor GATA6 enables self-renewal of cell survival. Cytokine Growth Factor Rev 2014;25:343–50. colon adenoma stem cells by repressing BMP gene expression. Nat Cell Biol 12. Ampuja M, Jokim€aki R, Juuti-Uusitalo K, Rodriguez-Martinez A, Alarmo 2014;16:695–707. E-L, Kallioniemi A. BMP4 inhibits the proliferation of breast cancer cells 27. Deffar K, Shi H, Li L, Wang X, Zhu X. Nanobodies - the new concept and induces an MMP-dependent migratory phenotype in MDA-MB-231 in antibody engineering. Afr J Biotechnol [Internet]. 2009 [cited 2014 Apr cells in 3D environment. BMC Cancer 2013;13:429. 15]; 8. Available from: http://www.ajol.info/index.php/ajb/article/view/ 13. Theriault BL, Shepherd TG, Mujoomdar ML, Nachtigal MW. BMP4 induces 60807. EMT and Rho GTPase activation in human ovarian cancer cells. Carcino- 28. Harada H, Nakagawa H, Oyama K, Takaoka M, Andl CD, Jacobmeier B, genesis 2007;28:1153–62. et al. Telomerase induces immortalization of human esophageal kerati- 14. Hao J, Lee R, Chang A, Fan J, Labib C, Parsa C, et al. DMH1, a small nocytes without p16INK4a inactivation. Mol Cancer Res MCR 2003;1: molecule inhibitor of BMP type i receptors, suppresses growth and inva- 729–38. sion of lung cancer. PLoS One 2014;9:e90748. 29. Mari L, Milano F, Parikh K, Straub D, Everts V, Hoeben KK, et al. A pSMAD/ 15. Lee Y-C, Cheng C-J, Bilen MA, Lu J-F, Satcher RL, Yu-Lee L-Y, et al. BMP4 CDX2 complex is essential for the intestinalization of epithelial metapla- promotes prostate tumor growth in bone through osteogenesis. Cancer Res sia. Cell Rep 2014;7:1197–210. 2011;71:5194–203. 30. Zilberberg L, Dijke P ten, Sakai LY, Rifkin DB. A rapid and sensitive bioassay 16. Virtanen S, Alarmo E-L, Sandstrom€ S, Ampuja M, Kallioniemi A. Bone to measure bone morphogenetic protein activity. BMC Cell Biol 2007;8:41. morphogenetic protein -4 and -5 in pancreatic cancer—Novel bidirectional 31. Milano F, van Baal JWPM, Buttar NS, Rygiel AM, de Kort F, DeMars CJ, et al. players. Exp Cell Res 2011;317:2136–46. Bone morphogenetic protein 4 expressed in esophagitis induces a colum- 17. Deng H, Makizumi R, Ravikumar TS, Dong H, Yang W, Yang W-L. Bone nar phenotype in esophageal squamous cells. Gastroenterology morphogenetic protein-4 is overexpressed in colonic adenocarcinomas 2007;132:2412–21. and promotes migration and invasion of HCT116 cells. Exp Cell Res 32. Lokate AMC, Beusink JB, Besselink GAJ, Pruijn GJM, Schasfoort RBM. 2007;313:1033–44. Biomolecular interaction monitoring of autoantibodies by scanning

www.aacrjournals.org Mol Cancer Ther; 14(11) November 2015 OF13

Calpe et al.

surface plasmon resonance microarray imaging. J Am Chem Soc 2007;129: action of BMP-2 and activin type IIB receptor. BMC Struct Biol 14013–8. 2007;7:6. 33. Dominguez C, Boelens R, Bonvin AMJJ. HADDOCK: a protein-protein 44. Voorneveld PW, Kodach LL, Jacobs RJ, van Noesel CJM, Peppelenbosch docking approach based on biochemical or biophysical information. J Am MP, Korkmaz KS, et al. The BMP pathway either enhances or inhibits the Chem Soc 2003;125:1731–7. Wnt pathway depending on the SMAD4 and p53 status in CRC. Br J Cancer 34. Sato T, Vries RG, Snippert HJ, van de Wetering M, Barker N, Stange DE, et al. 2015;112:122–30. Single Lgr5 stem cells build crypt-villus structures in vitro without a 45. Muyldermans S. Nanobodies: natural single-domain antibodies. Annu Rev mesenchymal niche. Nature 2009;459:262–5. Biochem 2013;82:775–97. 35. Mukherjee J, Tremblay JM, Leysath CE, Ofori K, Baldwin K, Feng X, et al. A 46. Harmsen MM, De Haard HJ. Properties, production, and applications of novel strategy for development of recombinant antitoxin therapeutics camelid single-domain antibody fragments. Appl Microbiol Biotechnol tested in a mouse botulism model. PLoS One 2012;7:e29941. 2007;77:13–22. 36. Keller S, Nickel J, Zhang J-L, Sebald W, Mueller TD. Molecular recognition 47. De Meyer T, Muyldermans S, Depicker A. Nanobody-based products as of BMP-2 and BMP receptor IA. Nat Struct Mol Biol 2004;11:481–8. research and diagnostic tools. Trends Biotechnol 2014;32:263–70. 37. Kirsch T, Sebald W, Dreyer MK. Crystal structure of the BMP-2-BRIA 48. Voorneveld PW, Kodach LL, Jacobs RJ, Liv N, Zonnevylle AC, Hoogen- ectodomain complex. Nat Struct Biol 2000;7:492–6. boom JP, et al. Loss of SMAD4 alters BMP signaling to promote colorectal 38. Allendorph GP, Vale WW, Choe S. Structure of the ternary signaling cancer cell metastasis via activation of Rho and ROCK. Gastroenterology complex of a TGF-beta superfamily member. Proc Natl Acad Sci U S A 2014;147:196–208. 2006;103:7643–8. 49. Zimmerman LB, De Jesus-Escobar JM, Harland RM. The Spemann orga- 39. Greenwald J, Groppe J, Gray P, Wiater E, Kwiatkowski W, Vale W, et al. The nizer signal noggin binds and inactivates bone morphogenetic protein BMP7/ActRII extracellular domain complex provides new insights into the 4. Cell 1996;86:599–606. cooperative nature of receptor assembly. Mol Cell 2003;11:605–17. 50. Lin SJ, Lerch TF, Cook RW, Jardetzky TS, Woodruff TK. The structural basis 40. Kirsch T, Nickel J, Sebald W. BMP-2 antagonists emerge from alterations in of TGF-beta, bone morphogenetic protein, and activin ligand binding. the low-affinity binding epitope for receptor BMPR-II. EMBO J 2000;19: Reprod Camb Engl 2006;132:179–90. 3314–24. 51. Desmyter A, Decanniere K, Muyldermans S, Wyns L. Antigen specificity and 41. Nickel J, Dreyer MK, Kirsch T, Sebald W. The crystal structure of the BMP-2: high affinity binding provided by one single loop of a camel single-domain BMPR-IA complex and the generation of BMP-2 antagonists. J Bone Joint antibody. J Biol Chem 2001;276:26285–90. Surg Am 2001;83-A Suppl 1:S7–14. 52. Ruppert R, Hoffmann E, Sebald W. Human bone morphogenetic protein 42. Groppe J, Greenwald J, Wiater E, Rodriguez-Leon J, Economides AN, 2 contains a heparin-binding site which modifies its biological activity. Kwiatkowski W, et al. Structural basis of BMP signalling inhibition by the Eur J Biochem 1996;237:295–302. cystine knot protein Noggin. Nature 2002;420:636–42. 53. Choi YJ, Lee JY, Park JH, Park JB, Suh JS, Choi YS, et al. The identification of 43. Weber D, Kotzsch A, Nickel J, Harth S, Seher A, Mueller U, et al. A a heparin binding domain peptide from bone morphogenetic protein-4 silent H-bond can be mutationally activated for high-affinity inter- and its role on osteogenesis. Biomaterials 2010;31:7226–38.

OF14 Mol Cancer Ther; 14(11) November 2015 Molecular Cancer Therapeutics

Effective Inhibition of Bone Morphogenetic Protein Function by Highly Specific Llama-Derived Antibodies

Silvia Calpe, Koen Wagner, Mohamed El Khattabi, et al.

Mol Cancer Ther Published OnlineFirst September 8, 2015.

Updated version Access the most recent version of this article at: doi:10.1158/1535-7163.MCT-14-0956

Supplementary Access the most recent supplemental material at: Material http://mct.aacrjournals.org/content/suppl/2015/09/05/1535-7163.MCT-14-0956.DC1

E-mail alerts Sign up to receive free email-alerts related to this article or journal.

Reprints and To order reprints of this article or to subscribe to the journal, contact the AACR Publications Subscriptions Department at [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://mct.aacrjournals.org/content/early/2015/10/20/1535-7163.MCT-14-0956. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.