Robert Shapiro, Ph. D.

Associate Professor of Pathology, Harvard Medical School

Center for Biochemical & Biophysical Sciences & Medicine

One Kendall Square Building 600, Third Floor

Cambridge, MA 02139

Phone: 617-621-6132; FAX: 617-621-6111

E-mail:  Robert_Shapiro@hms.harvard.edu

 

 Research Overview

 Structure and function of angiogenin

         Human angiogenin (Ang) is an unusual member of the pancreatic RNase superfamily that induces blood vessel formation in vivo.  Although Ang contains counterparts for the key catalytic residues of bovine pancreatic RNase A, it cleaves standard RNase substrates 10⁵ - 10⁶ times less efficiently than does RNase A.  Despite this apparent weakness, the enzymatic activity of Ang appears to be essential for biological activity:  replacements of important active site residues invariably diminish ribonucleolytic and angiogenic activities in parallel, and a substitution that increases enzymatic activity also enhances angiogenic potency.  Crystal structures of two of the inactive variants show that there are no significant changes beyond the replacement site, indicating that loss of biological activity is directly attributable to disruption of the catalytic apparatus.  The identities of the in vivo substrates or ligands of Ang are now being investigated by the research group of Dr. Guo-fu Hu at this Center.

         A major focus of our studies has been to understand the structural basis for the unique enzymatic properties of Ang.  Our approach combines site-directed mutagenesis, kinetic mapping, and X-ray crystallography (in collaboration with Prof. K. R. Acharya of the University of Bath, U.K.). Thus far, the roles of many active site components in catalysis, binding, or modulation of activity have been determined.  In addition, several surprising structural features of Ang that contribute to the attenuation of enzymatic activity have been revealed.  The most striking of these is the obstruction of the putative pyrimidine-binding site by Gln117, which lies on the C-terminal 3₁₀ helix.  Modeling demonstrates that a conformational change to open this site is required in order for Ang to bind and cleave RNA substrates: i.e., that the native Ang structure observed by crystallography and NMR is inactive.  This raises the possibility that Ang undergoes activation at the appropriate time and location in vivo by binding to other cellular components or through post-translational modification.  We are now working with crystallographer K. R. Acharya and NMR spectroscopist F. Ni (National Research Council of Canada Biotechnology Research Institute) to define the active conformation of Ang by determining structures of superactive Ang variants and complexes of Ang with pyrimidine-containing inhibitors.

Development of small-molecule inhibitors of angiogenin

             Angiogenin was first isolated from medium conditioned by human colon adenocarcinoma cells, and has been shown to play a critical role in the establishment and/or metastatic spread of a wide range of human tumor xenografts in athymic mice, most likely by contributing to tumor angiogenesis.  Moreover, Ang expression has been found to be elevated in numerous types of human cancers, and in many instances a specific association of Ang with cancer aggressiveness and/or progression has been demonstrated.  These findings identify Ang as an attractive target for anticancer therapy.

             One strategy for the development of Ang antagonists is to design molecules that bind tightly to the ribonucleolytic active center of the protein, which has been demonstrated to be a key actor in the angiogenic mechanism.  Part of our effort in this area has focused on nucleotide-based compounds, and has identified several Ang inhibitors that may be useful as starting points for structure-based design.  Parallel work has been performed with RNase A as a model system that is more amenable to crystallographic study at this stage.  Recently, we have been able to expand our approach to include non-nucleotide compounds through a new high-throughput assay, which is being used to screen large compound libraries in collaboration with the Harvard Institute of Chemistry and Cell Biology.  Several new leads have been identified that bind much more tightly than any of the nucleotides.  These will now be used for rational design of improvements based on models of ligand complexes and subsequently actual 3D structures to be determined by our collaborators K. R. Acharya and F. Ni.  Predictions of binding modes of inhibitors and the energetic effects of proposed modifications are performed with the programs AutoDock and LUDI, respectively; AutoDock has been validated using several difficult test cases from the literature where minor alterations of inhibitor structures caused major unexpected changes in binding modes.

 Human placental RNase inhibitor

           Human placental RNase inhibitor (RI) is a 50-kDa cytosolic protein that binds mammalian pancreatic RNase superfamily enzymes with dissociation constants (K_d values) ranging from 10^-13 M to below 10^-15 M.  Its tightest-binding natural ligand is angiogenin (Ang), with a K_d value of 0.5 fM reflecting extremely slow dissociation of the complex (t_½ = 73 days) and rapid association that approaches the diffusion limit.  hRI is constructed almost entirely of tandem alternating 29- and 28-residue leucine-rich repeat (LRR) units with average 40% sequence identity.  LRRs are common motifs utilized for protein-protein interactions and have been found in more than 70 proteins with diverse functions, including signal transduction, cell cycle control, cell adhesion, and enzyme regulation.  The interactions of RI with its various ligands provides a particularly intriguing and powerful system for studying the structural basis for molecular recognition in general and the function of LRRs in particular because of the extraordinarily high affinities achieved and the broad specificity of RI for proteins that typically share only 25-35% sequence identity.  Moreover, suitably engineered derivatives of RI (with greater selectivity for Ang and increased extracellular stability) might be useful as therapeutic agents for treatment of cancer and other angiogenin-dependent diseases.

             We are investigating the molecular basis for tight-binding of RI to Ang, RNase A, and other RNases by single-site and multi-site mutagenesis, together with X-ray crystallography (in collaboration with Prof. K. R. Acharya, University of Bath, U.K.).  The crystal structures of the complexes of porcine RI with RNase A (determined by Kobe and Deisenhofer) and human RI with Ang (determined by the Acharya laboratory) show that both interfaces are large, and that the ligands dock similarly, although few of the specific interactions formed are analogous.  Mutational analysis of the Ang complex, however, reveals that the binding energy is focused largely in a single relatively small "hot spot" containing residues from the C-terminal region of RI and the active site of the enzyme; only one other part of the interface, a region rich in tryptophans, makes an important energetic contribution.  Within both the hot spot and Trp-rich regions, residues function cooperatively such that mutational effects are superadditive, even in many instances where there are no obvious structural linkages.  The RI-RNase A complex contains a similarly-positioned hot spot, but only one major contact (involving the catalytic lysine of the enzyme) appears to be conserved.  The binding energy is much more widely distributed over the interface in this complex, and the combined effects of replacing multiple residues are markedly subadditive, rather than superadditive.  Thus, RI recognizes Ang and RNase A in largely distinctive ways.

             Efforts are now underway to determine whether the general "themes" suggested by the RI-Ang and RI-RNase A complexes (e.g., anchoring of the ligand through contacts with the active site; the use of cooperativity) apply to the interactions of the inhibitor with other ligands, such as human RNase 2 (also known as eosinophil-derived neurotoxin).  Additional areas being actively pursued are the generation of "minimized" RI derivatives based on the hot spot region and the investigation of the role of the 32 cysteine residues of RI in conferring sensitivity to oxidation.

 Current research group

Jeremy L. Jenkins, Ph. D.  -- Structure-based design of angiogenin inhibitors

Kapil Kumar, Ph. D. -- Engineering and minimization of protein RNase inhibitor

Michael Brady, B.S. -- Stabilization of protein RNase inhibitor

 Previous group members (since 1996)

Matthew Crawford, B. A.

Richard Kao, Ph. D.

Daniel P. Teufel (visiting student from University of Bath)

Melisa Ruiz-Gutierrez, B. S.

Anwar Jardine, Ph. D.

Marsha Crochierre, B. S.

Chang-Zheng Chen, Ph. D.

Cecilia Roh, B. S.

 Recent publications