Discoidin Discoveries

Structure

Previews culmination of multiple structural studies and provides a fresh perspective on a critical complex in the NER process. The proposed model also highlights the critical requirement to study full-length proteins and multi-protein complexes to understand how the evolving architecture of the NER machine, and the resulting structural changes induced in the DNA substrate, drive the progression of biochemical steps required for repair.

ACKNOWLEDGMENTS Research on NER in our laboratory is supported by the National Institutes of Health (Grants R01 ES1065561, P01 CA92584, P30 CA68485, and P30 ES00267) and a postdoctoral fellowship from the American Cancer Society (Grant PF-11-27101-DMC). We are grateful to Rachel C. Wright for preparation of the figure.

REFERENCES Das, D., Folkers, G.E., van Dijk, M., Jaspers, N.G.J., Hoeijmakers, J.H.J., Kaptein, R., and Boelens, R. (2012). Structure 20, this issue, 667–675. Hohl, M., Thorel, F., Clarkson, S.G., and Scha¨rer, O.D. (2003). J. Biol. Chem. 278, 19500–19508. Krasikova, Y.S., Rechkunova, N.I., Maltseva, E.A., Petruseva, I.O., and Lavrik, O.I. (2010). Nucleic Acids Res. 38, 8083–8094. Moggs, J.G., Yarema, K.J., Essigmann, J.M., and Wood, R.D. (1996). J. Biol. Chem. 271, 7177–7186. Newman, M., Murray-Rust, J., Lally, J., Rudolf, J., Fadden, A., Knowles, P.P., White, M.F., and McDonald, N.Q. (2005). EMBO J. 24, 895–905. Nouspikel, T. (2009). Cell. Mol. Life Sci. 66, 994–1009. Riedl, T., Hanaoka, F., and Egly, J.M. (2003). EMBO J. 22, 5293–5303. Shao, X., and Grishin, N.V. (2000). Nucleic Acids Res. 28, 2643–2650.

Singh, S., Folkers, G.E., Bonvin, A.M., Boelens, R., Wechselberger, R., Niztayev, A., and Kaptein, R. (2002). EMBO J. 21, 6257–6266. Staresincic, L., Fagbemi, A.F., Enzlin, J.H., Gourdin, A.M., Wijgers, N., Dunand-Sauthier, I., GigliaMari, G., Clarkson, S.G., Vermeulen, W., and Scha¨rer, O.D. (2009). EMBO J. 28, 1111–1120. Tripsianes, K., Folkers, G., Ab, E., Das, D., Odijk, H., Jaspers, N.G., Hoeijmakers, J.H., Kaptein, R., and Boelens, R. (2005). Structure 13, 1849–1858. Tripsianes, K., Folkers, G.E., Zheng, C., Das, D., Grinstead, J.S., Kaptein, R., and Boelens, R. (2007). Nucleic Acids Res. 35, 5789–5798. Tsodikov, O.V., Enzlin, J.H., Scha¨rer, O.D., and Ellenberger, T. (2005). Proc. Natl. Acad. Sci. USA 102, 11236–11241. Tsodikov, O.V., Ivanov, D., Orelli, B., Staresincic, L., Shoshani, I., Oberman, R., Scha¨rer, O.D., Wagner, G., and Ellenberger, T. (2007). EMBO J. 26, 4768–4776.

Discoidin Discoveries Kathryn M. Ferguson1,* 1Department of Physiology, Perelman School of Medicine, University of Pennsylvania, Philadelphia, PA 19104, USA *Correspondence: [email protected] DOI 10.1016/j.str.2012.03.003

In this issue of Structure, Carafoli et al. investigate the mode of antibody-mediated inhibition of the discoidin domain receptor 1 (DDR1). These studies also provide new insight into activation of the DDRs, which are unique among receptor tyrosine kinases in the composition of their extracellular regions. Like many receptor tyrosine kinases (RTKs) the two members of the discoidin domain receptor family (DDR1 and DDR2) regulate fundamental cellular process such as proliferation, differentiation and adhesion (Leitinger, 2011; Lemmon and Schlessinger, 2010). Dysregulation of these receptors is linked to a number of human diseases, including fibrotic disorders, atherosclerosis, and cancer. Mounting evidence suggests that DDRs are relevant therapeutic targets (Valiathan et al., 2012). Inhibitory antibodies, such as those described by Carafoli et al. (2012) in this issue of Structure, will be critical in evaluating the potential of DDR inhibition in disease settings. DDRs are unusual RTKs in several respects. DDR activators—extracellular

matrix collagens in their native triple-helical conformation—are unique among RTK ligands. Short triple-helical collagenderived peptides are sufficient to activate DDRs, suggesting that receptor clustering induced by collagen may not be an essential aspect of DDR activation. It is believed that DDRs exist as preformed dimers in cells and that collagen binding activates signaling by altering the conformation of these dimers, rather than by inducing receptor oligomerization as seen for many RTKs (Lemmon and Schlessinger, 2010). The intracellular events following collageninduced DDR activation are broadly similar to those with other RTKs. The receptors undergo autophosphorylation, recruit signaling adaptors (including Shc and Nck), and activate downstream signaling

568 Structure 20, April 4, 2012 ª2012 Elsevier Ltd All rights reserved

cascades including the mitogen-activated protein kinase pathway. However, the timing of these events is unusually slow and sustained. Exactly how this relates to signaling outcome is unclear. A further unusual feature of DDRs is the unique domain composition of the extracellular region, of which Carafoli et al. (2012) present the most complete picture to date. They describe the X-ray crystal structure of the DDR1 extracellular region (ECR) lacking only the z50 amino acid, presumed unstructured, juxtamembrane (JM) region. The ECR of DDR1 contains two discoidin (DS) domains, members of the coagulation factor V/VIII type C superfamily. No other RTK contains extracellular DS domains, but these domains are found in a number of other proteins

Structure

Previews A

B

Figure 1. Insights from the Structure of the Extracellular Region of DDR1 Bound to an Inhibitory Fab (A) View of the Fab/DDR1 complex highlighting key structural features. (B) Two possible models for collagen (red) mediated DDR1 activation. In the first, the conserved patch (pink) is involved in inter-receptor contacts in the dimer, whereas in the second, this region is a secondary, presumed low affinity, binding site that promotes ligand crosslinking of two receptor molecules. Linking these events to intracelluar kinase domain activation is complicated by the presence of a long, unstructured extracellular juxtamembrane region (dashed line). Binding of mAb to the DS-like domain sterically blocks formation of the signaling active dimer.

involved in cell adhesion. Carafoli et al. (2009) previously determined the crystal structure of the N-terminal DS domain of DDR2 bound to a triple-helical collagen peptide. The DS domain has a barrel-like fold, and collagen binds to three loops or spikes that protrude from one of its ends—a common site for ligand binding to this structural family. The structure of the DDR1 ECR now confirms that the second globular domain also shares structural similarity with DS domains. The second DS-like domain has a long insert between the first two b strands of the DS core that augments the protrusions on the ‘‘ligand-binding’’ end of the barrel and contributes to an extensive interface between the two domains. DS domains are often found as tandem repeats, and the relative orientation of the two domains in DDR1 is reminiscent of that observed in neuropilin (Vander Kooi et al., 2007). However, the packing of the two domains is much looser in DDR1 (surface complementarity [Sc] value of 0.56) than in neuropilin (Sc value of 0.77). The looser coupling between the two domains in DDR1 suggests that alterations in the relative orientation of the DS and DS-like domains upon receptor activation might be a source of conformational rearrangement in the DDR1 dimer. This almost complete view of the DDR1 ECR is just one of several important advances that were enabled by the panel of inhibitory antibodies developed and described by Carafoli et al. (2012). Crystals of the DDR1 ECR could only be ob-

tained in complex with the Fab fragment from one of these antibodies (Figure 1A). Antibodies were generated using recombinant soluble DDR1 ECR (including the JM region), and seven were found to inhibit collagen-induced phosphorylation of DDR1 in cultured cells. Inhibition was as effective with the Fab fragments from these antibodies, indicating that binding to DDR1 per se, rather than receptor clustering, is responsible for the effect. Interestingly, all of the antibodies bind to the membrane proximal DS-like domain and, surprisingly, they do not block binding of collagen peptides. The inhibitory effect of these antibodies on collagen-induced DDR1 phosphorylation must therefore occur through some indirect or steric effect. A key question is what interactions might be sterically blocked by antibody binding to the DS-like domain? Unfortunately the present study cannot provide a complete answer to this question, but several observations provide fuel for speculation and, importantly, direction for additional investigation. Molecular mechanisms have been proposed for two other inhibitory antibodies that inhibit receptor function through steric effects—the binding of efalizumab to the lymphocyte function-associated antigen 1 (LFA-1) (Li et al., 2009) and the binding of matuzumab to the epidermal growth factor receptor (EGFR) (Schmiedel et al., 2008). The epitopes for these antibodies do not overlap with the ligand-binding sites on their cognate receptors, but they are close by. When

bound to LFA-1, efalizumab sterically impairs access of the ICAM-1 to its binding site on the receptor (clashing with regions of the ligand that are not involved in binding). Matuzumab achieves its inhibitory effect on EGFR by sterically restricting the EGF-induced conformational changes in the receptor that are required for ligand-induced dimerization. The 3E9 antibody described by Carafoli et al. (2012)—and all other seven inhibitory antibodies studied—bind to the DSlike domain, quite distant from the primary collagen-binding site. It seems highly unlikely that their inhibitory effects result from steric restriction of collagen binding. Rather, as suggested by Carafoli et al. (2012), it seems likely that these antibodies prevent the DDR1 dimer from adopting the conformation required for activation of the intracellular kinase domains to exert their inhibitory effects. Unfortunately, there is currently no detailed information about the nature of any DDR1 dimer, either in the absence or presence of collagen. The extracellular regions of DDRs are monomeric. Further, in the intact receptor, the long, presumed unstructured, extracellular JM regions make it hard to predict how contacts between the DS or DS-like domains could be coupled to events inside the cell. Although the soluble DDR1 protein is monomeric in solution, the authors interrogate the relevance of one plausible crystallographic dimer. Disruption of this dimer interface does not abolish DDR1 activation. However, this analysis identifies a conserved surface patch on the DS domains that is essential for DDR1 signaling. Carafoli et al. (2012) offer two possible roles for this conserved patch that are illustrated in Figure 1B. In the first, the conserved surface patch on the DS domain contributes to essential interreceptor contacts in the DDR1 dimer. The alternative, and favored (by the authors), model is that this region represents a second, lower affinity collagen-binding site. In this model, collagen could crosslink two receptor molecules to influence dimer conformation. Further investigation is still needed to resolve exactly how collagen binding leads to activation of DDR1. ACKNOWLEDGMENTS The author’s work is supported by NIH R01 CA112552.

Structure 20, April 4, 2012 ª2012 Elsevier Ltd All rights reserved 569

Structure

Previews Leitinger, B. (2011). Annu. Rev. Cell Dev. Biol. 27, 265–290.

Schmiedel, J., Blaukat, A., Li, S., Kno¨chel, T., and Ferguson, K.M. (2008). Cancer Cell 13, 365–373.

Carafoli, F., Bihan, D., Stathopoulos, S., Konitsiotis, A.D., Kvansakul, M., Farndale, R.W., Leitinger, B., and Hohenester, E. (2009). Structure 17, 1573– 1581.

Lemmon, M.A., and Schlessinger, J. (2010). Cell 141, 1117–1134.

Valiathan, R.R., Marco, M., Leitinger, B., Kleer, C.G., and Fridman, R. (2012). Cancer Metastasis Rev. Published online February 26, 2012. 10.1007/s10555-012-9346-z.

Carafoli, F., Mayer, M.C., Shiraishi, K., Pecheva, M.A., Chan, L.Y., Nan, R., Leitinger, B., and Hohenester, E. (2012). Structure 20, this issue, 688–697.

Li, S., Wang, H., Peng, B., Zhang, M., Zhang, D., Hou, S., Guo, Y., and Ding, J. (2009). Proc. Natl. Acad. Sci. USA 106, 4349–4354.

Vander Kooi, C.W., Jusino, M.A., Perman, B., Neau, D.B., Bellamy, H.D., and Leahy, D.J. (2007). Proc. Natl. Acad. Sci. USA 104, 6152–6157.

REFERENCES

Phosphorylation Meets Proteolysis Martin Renatus1,* and Christopher J. Farady1 1Novartis Institutes for BioMedical Research, Forum 1, Novartis Campus, CH-4002 Basel, Switzerland *Correspondence: [email protected] DOI 10.1016/j.str.2012.03.006

Phosphorylation is a reversible post-translational modification that regulates many proteins and enzymes, including proteases, as shown by two recent publications. Huang and colleagues and Vela´zquez-Delgado and Hardy (this issue of Structure) describe how phosphorylation activates the protease activity of the deubiquitinating enzyme DUBA and how it inhibits caspase-6, respectively.

Post-translational protein modification through phosphorylation is central to the regulation of key cellular processes. The human ‘‘kinome’’ (Manning et al., 2002; http://kinase.com/human/kinome/), consisting of at least 518 different kinases, catalyzes millions of distinct phosphorylation events. Another well studies posttranslational event is proteolysis, which is catalyzed by members of the human ‘‘degradome’’ (http://degradome.uniovi. es/dindex.html). It is well appreciated that so-called limited proteolysis, where a specific protease cleaves a specific substrate at one or several specific sites, leading to its activation, deactivation, or subcellular relocation rather than its complete degradation, is implicated in many cellular events ranging from cell division (Dephoure et al., 2008) to apoptosis (Kurokawa and Kornbluth, 2009). Although the two events are fundamentally different—phosphorylation is a reversible modification, whereas proteolysis is irreversible—there is mounting evidence that kinases and proteases work hand-in-hand. For example, the regulation of both cell proliferation and apoptosis is dependent on the interplay between protease and kinases (Lo´pez-

Otı´n and Hunter, 2010). Caspase-3dependent processing and consequent inactivation of the serine-threonine protein kinase AKT1 turns off survival pathways, whereas some deubiquitinating proteases (DUBs) stabilize kinases by removing ubiquitin tags from proteins otherwise destined for degradation by the ubiquitin-proteasome pathway. Likewise, phosphorylation is known to activate or inactivate proteases. Although it is somehow easier to rationalize how a proteolytic event leads to the activation or inactivation of a functional protein, the structural changes induced by phosphorylation are more subtle. Two recent papers provide exciting insight into how phosphorylation can directly regulate protease activity. While Huang et al. (2012) provide a structural explanation of how phosphorylation activates the deubiquitinating protease DUBA, Vela´zquezDelgado and Hardy (2012; this issue of Structure) show that introducing a mutation that mimics a biological phosphorylation event inactivates the apoptotic protease caspase-6. The deubiquitinating activity of human deubiquitinase DUBA is strictly dependent on the phosphorylation of Ser177

570 Structure 20, April 4, 2012 ª2012 Elsevier Ltd All rights reserved

by the casein kinase II (CK2). Indeed, the ligand-free, unphosphorylated DUBA rests in an inactive state, as the substrate binding site is misaligned and parts of the molecule appear to be highly mobile. By itself, CK2 phosphorylation does not induce any structural changes that would be consistent with protease activation. The active conformation is induced only upon ubiquitin binding, as if the enzyme ‘‘folds around its substrate’’ (Huang et al., 2012). The phosphate group stabilizes the substrate-protease interaction but does not directly interact with the active site residues (Figure 1). It clamps together two helices and appears in turn to stabilize the core of the DUBA structure and is involved in direct interactions with C-terminal part of the ubiquitin substrate. Substrate-induced activation has previously been observed for other, structurally diverse DUBs such as UCH-L3 and USP7 (Hu et al., 2002; Johnston et al., 1999). In their ligand free forms, these enzymes exist in an inactive resting state. Ubiquitin binding induces the maturation of the otherwise obstructed and misaligned substrate binding site and catalytic center. This substrate-dependent