July 1992 — June 1993 ANNUAL REPORT UNIVERSITY OF CALIFORNIA, SAN FRANCISCO SCHOOL OF MEDICINE DEPARTMENT OF BIOCHEMISTRY AND BIOPHYSICS ANNUAL REPORT: JULY 1992-JUNE 1993 HAROLD E. VARMUS, M.D., PROFESSOR MOLECULAR AND GENETIC APPROACHES TO RETROVIRUSES AND ONCOGENES Krissy Bibbins, Graduate Student _ Violetta Bigornia, Staff Research Associate Delicia Caballero, Laboratory Assistant Mario Chamorro, Staff Research Associate Michael Chastain, Postdoctoral Fellow Lucy Godley, Graduate Student Ken Kaplan, Graduate Student (shared with D. Morgan) Mary Jo Kelley, Administrative Assistant Susanna Lee, Postdoctoral Fellow Michael Lochrie, Postdoctoral Fellow Clifford Lowell, Postdoctoral Fellow Hans-Peter Miiller, Postdoctoral Fellow John Murphy, Postdoctoral Fellow Tuan-tu Nguyen, Staff Research Associate Maho Niwa-Rosen, Postdoctoral Fellow Suzanne Ortiz, Specialist Raymond Scott, Laboratory Assistant Supriya Shivakumar, Graduate Student (shared with C. Kenyon) Peter Sorger, Postdoctoral Fellow Richard Sutton, Postdoctoral Fellow lan Taylor, Postdoctoral Fellow Hao Wang, Postdoctoral Fellow Karl Willert, Graduate Student Linda Yushenkoff, Staff Research Associate In this laboratory, retroviruses have traditionally served as points of departure for studying various aspects of eukaryotic cells at the molecular level. Our attempts to understand how retroviruses multiply within infected cells and how retroviruses cause cancers have led to work on a variety of topics, including cell surface receptors, DNA recombination, translational control, inter- and intracellular signaling, hematopoiesis, embryonic development, and mammary carcinogenesis. 1. Early events in virus infection require entry into cells through host-encoded receptors; we have recently cloned avian genes encoding receptors for a subgroup of Rous sar- coma virus (RSV) and shown that the receptors resemble the receptor for low-density lipoproteins (LDL). After viral DNA is synthesized by reverse transcriptase, it is efficiently inserted into chromosomes by a small viral protein, integrase, that performs a recombination reac- tion required for transposition of many types of elements. Recent studies of retroviral integrases identify some of the residues essential for their biochemical activities and the structural determinants that influence choice of integration sites in DNA and chromatin. Synthesis of reverse transcriptase and integrase usually requires efficient shifting of reading frame by ribosomes translating viral RNA, in response to short sequences and RNA structural elements, commonly pseudoknots; the determinants of frameshifting 261 HAROLD E. VARMUS efficiency can now be studied with genetic, structural, and biochemical methods in yeast and mammalian systems. Most of the highly tumorigenic retroviruses carry oncogenes, derived from host proto- oncogenes; several of these belong to a gene family (src) that encodes cytoplasmic pro- tein-tyrosine kinases implicated in differentiated functions; targeted mutations of some of these genes (hck and fgr) have been made in the mouse germ line to determine their physiological roles in hematopoiesis. In efforts to understand the actions of sre protein, the protein has been localized to endosomes in fibroblasts, and extracatalytic regions that mediate interactions with other proteins have been defined with genetic and bio- chemical means. Most weakly tumorigenic retroviruses activate cellular proto-oncogenes by insertion mutations; the Wnt-1 gene, discovered as a target for insertional activation by the mouse mammary tumor virus, belongs to a large family of genes encoding secretory proteins involved in important developmental events in many organisms. In attempts to identify cell-surface receptors for Wnt proteins, we have developed several bioassays for Wnt genes and learned to make cell-free Wnt protein in a complex with the surface antigen of hepatitis B virus. Site-directed mutants of Wnt-1 include temperature-sensi- tive alleles and alleles encoding active, transmembrane proteins. Two Wnt genes have been characterized in C. elegans. Finally, a Wnt-I transgenic model for mammary carcinogenesis has been used to identify additional genetic components of a multistep neoplastic process. THE RETROVIRAL LIFE CYCLE The study of retrovirus replication has been in- structive about eukaryotic cells in many ways during the past twenty years: Retroviruses provided the first source of reverse transcriptase, their proviruses were the first examples of precise recombination products in eukaryotic DNA, and their strategies for gene ex- pression include transcriptional enhancers, glucocor- ticoid responsiveness, alternative splicing, ribosomal frameshifting, nonsense suppression, and intricate protein-nucleic acid interactions during virus assem- bly. The need to understand these and other aspects of retroviruses has become more urgent with the dis- covery that AIDS is caused by a retrovirus, the hu- man immunodeficiency virus (HIV). Our current work on the retroviral life cycle is focused upon three events: 1) the entry of virus into cells via host-encoded receptors; 2) the integration of viral DNA into host chromosomes; and 3) the syn- thesis of polymerase gene products by ribosomal frameshifting. Although we continue to perform some of these studies with avian and murine retro- viruses, we give increasing attention to HIV. Retroviral Receptors Entry of retroviruses into cells depends upon host-encoded transmembrane proteins that serve as receptors for viral envelope glycoproteins. The re- 262 markable specificity of virus-host interactions has been known for over twenty years from studies of the polymorphic envelope proteins of avian retroviruses, yet little biochemical information is available about the receptors or about the nature of their interactions with viral envelope glycoproteins. Over the past few years, receptors for several animal viruses have been identified as members of the superfamily of immunoglobin genes; perhaps the most important example is the lymphocyte cell-surface antigen CD4, a simple transmembrane glycoprotein that is required for attachment of HIV to target cells. Recently, the receptor for ecotropic murine leukemia virus (MLV) was shown to be a very different type of protein, an amino acid permease with fourteen transmembrane domains. Paul Bates and John Young have recently cloned the gene encoding the receptor for one of the five subgroups of RSV. This was accomplished by a gene transfer method in which chicken DNA was in- troduced into mammalian cells, which lack functional receptors; target cells that acquired a receptor gene were identified by infection with subgroup A avian virus vectors carrying a selectable marker. Although cDNA clones have been difficult to obtain, we have deduced the coding sequence of the gene by exon trapping in a retroviral vector. The results were surprising: The putative receptor is a very small protein, just over 100 amino acids in length, with a single transmembrane domain and an ectodomain that closely resembles a portion of the binding region of the low-density lipoprotein (LDL) receptor. In addi- tion, the gene appears to be differentially spliced, generating at least two membrane-bound proteins (one of which has been shown by Hao Wang to be GPI linked) and a secreted protein. Antisera against the ectodomain have been used by Paul and David Chu (an MSTP student on rotation) to show that the gene produces a cell surface glycoprotein in transfected cells and that virus infection of normal cells can be blocked immunologically. Hao Wang is now trying to assess the normal composition and function of the receptor, to ask (with Judy White’s laboratory) whether it can bind directly to viral envelope protein, and to examine its structure (in collaboration with Dave Agard’s laboratory). (Some of these studies continue to be pursued with John Young, now on the UCSF faculty at the Gladstone Institute for Virology and Immunology.) Mike Lochrie is using methods similar to those employed with the avian virus receptor to clone re- ceptors for feline leukemia viruses and to analyze a known receptor for subgroup B feline viruses. In addition, Richard Sutton, in collaboration with Dan Littman, has begun to try to clone a receptor for the human T cell leukemia virus (HTLV). Proviral Integration Like many transposable elements from plants, bacteria, yeast, and insects, retroviral proviruses can be found at many different sites in host genomes, but are always joined to host DNA at the same sites in vi- ral DNA. The provirus contains viral genes arranged as they are in viral RNA (most commonly: 5’-gag- pol-env-3'), flanked by long terminal repeats (LTRs) that are generated during reverse transcription and used for regulation of transcription. The LTRs termi- nate with short inverted repeats that form part of the att site required for integration, and the entire provirus is flanked by short direct repeats of cellular origin generated during the integration step. The only viral protein known to be required specifically for in- tegration, the IN protein, is encoded by the 3' end of the pol gene. Current studies of retroviral integration depend largely upon reconstruction of the reaction under in vitro conditions. This was first accomplished with vi- ral nucleoprotein complexes from cells recently in- fected with MLV. It is now possible to study the re- action using only purified IN protein (usually pre- pared here from yeast expression systems) and la- beled oligonucleotides representing att sites and tar- gets. It is now known that two independent events are mediated by IN protein: 1) removal of two nu- cleotides from the 3’ end of each strand in the linear form of viral DNA, and 2) staggered cutting of the ANNUAL REPORT: JULY 1992-JUNE 1993 target DNA, with a concerted, ATP-independent join- ing of newly created 5' ends to the viral 3' ends. Andy Leavitt has isolated the HIV-1! IN protein from a yeast expression system in order to study the att site, target site, and IN sequence determinants of the steps in the integration reaction. He has shown that both steps in the reaction require a highly con- served CA dinucleotide at the site of 3' processing; that additional internal sequences confer specificity for the HIV IN protein; and that the selection of in- tegration sites is dependent upon the sequence of the target oligonucleotide. In conjunction with Lily Shiue, who has made site-directed mutants of HIV-1 IN, several amino acid residues crucial for biochemi- cal function have been identified. Some of these mutant genes have also been tested in an HIV-based virus vector system. Mike Chastain is now constructing HIV-MLV chimeric IN proteins in efforts to map functional domains. In addition, HIV- 1 IN protein has been purified on a large scale to at- tempt X-ray crystallography in collaboration with Bob Rose in the Stroud laboratory. To study the integration reaction in a fashion that more closely resembles what occurs in infected cells, Peter Pryciak has shown that MLV nucleopro- tein complexes mediate integration in vitro into target DNA assembled into nucleosomes. Through the use of minichromosomes in which nucleosomes are accu- rately positioned, he has been able to map integration sites in relation to nucleosomes. This was done ini- tially by cloning and sequencing reaction products, but the analysis has been remarkably simplified and enhanced by Peter’s development of a polymerase chain reaction (PCR)-based assay that scores the position and frequency of integration events within several hundred nucleotides of a PCR primer site. The findings reveal that integration occurs more efficiently into DNA packaged nucleosomes than into nucleosome-free regions of DNA; that the sites most available for integration are in the major groove of DNA facing away from the nucleosomal core (thus occurring with a periodicity of about 10 bp, one turn of the DNA helix); and that nucleotide sequence also influences the selection of integration site. DNA- binding proteins, such as the yeast alpha-2 protein, block integration, creating a “footprint” in the PCR- based assay. These results imply that the retroviral integra- tion machine is sensitive to the status of eukaryotic chromosomes and may be able to survey their func- tional and structural properties in vivo, as well as in vitro. Peter and Hans-Peter Miiller have shown that co-infection of cells with MLV and simian virus 40 .(SV40, a virus with a small circular DNA genome) allows integration of MLV DNA into the many copies of SV40 minichromosomes in a strikingly nonrandom pattern. Hans-Peter is using various 263 HAROLD E. VARMUS strategies to bend DNA targets and thereby determine how site preference occurs. For example, when DNA is bent by contact with bacterial CAP protein or by bringing together ends of short linear duplexes, the distribution of integration sites is altered and resem- bles patterns observed with chromatin. Ribosomal Frameshifting All retroviruses synthesize reverse transcriptase and IN protein as gag-pol polyproteins that must be subsequently cleaved by a viral protease to form mature functional proteins. Yet most retroviruses en- code gag and pol in different frames. Retroviruses solve this problem by directing ribosomes to slip back one nucleotide while decoding regions of gag- pol mRNAs in which the reading frames overlap. The slippage occurs at defined frequencies (generally 5-25%), in response to specific sequences at the frameshift sites (e.g. A AAU UUA or A AAA AAC), and requires secondary structural elements just down- stream from the site. Similar frameshifting occurs during synthesis of a coronavirus RNA polymerase, the transposase of the prokaryotic element, IS1, and a subunit of E. coli DNA polymerase III. The nature and function of RNA secondary structures downstream of frameshift sites continue to be important and incompletely resolved issues. Mario Chamorro, in conjunction with Neil Parkin, has precisely defined an RNA pseudoknot required for high-frequency frameshifting on mouse mammary tumor virus (MMTV) RNA, at the overlap of the gag and protease genes. Structural features of this and other pseudoknots are now being studied in collaboration with Ignacio Tinoco’s laboratory at the University of California, Berkeley, using a combina- tion of genetic, biochemical, and physical methods (including NMR). Preliminary results confirm the existence of the predicted pseudoknot. Moreover, substitution of a known pseudoknot for the MMTV pseudoknot indicates that not all pseudoknots are able to promote frameshifting. Neil and Mario have re- cently confirmed the importance of a hypothetical stem-loop downstream of the HIV-1 frameshift site; although the integrity of the stem-loop appears to have only minor effects upon frameshifting in vitro, it has a more dramatic effect when the gag-pol region of HIV-1 is expressed in cultured cells. Mario’s study of the MMTV gag-pro frameshift site has also suggested that the efficiency of frameshifting is strongly influenced by the nature of the tRNAs decoding the frameshift site. Susanna Lee has begun to study retroviral frameshifting in yeast, where genetic means can be used to identify tRNA’s, other trans-active factors, and unsuspected cis-active elements that affect frameshift rates. By screening colonies for produc- 264 tion of B-galactosidase or selecting for high levels of CUP! protein, Susanna has assembled a set of strains with variable levels of frameshifting in response to MMTV- and MHIV-derived signals. Two complementation groups have been identified for cloning mutant genes have begun. RETROVIRAL ONCOGENESIS AND PROTO- ONCOGENES Retroviruses competent to induce tumors are conveniently grouped in two categories: 1) those highly oncogenic agents that carry oncogenes transduced from normal cells, and 2) those less efficient agents that lack their own oncogenes. Among the first group are viruses employing over twenty distinctive oncogenes (v-onc’s). Each v-onc is derived from a cellular proto-oncogene, and these in turn are often members of gene families. The second group of oncogenic retroviruses, those lacking their own oncogenes, includes several viruses producing a wide spectrum of diseases. Viruses of this second type usually act as insertional mutagens, enhancing the expression of adjacent cellular proto- oncogenes as an initial step in tumorigenesis. Some of these genes have also been transduced to form retroviral oncogenes, but many were discovered as novel targets for insertion mutation. A surprising number of proto-oncogenes have proved to encode growth factors, growth factor recep- tors, components of cytoplasmic signaling networks, or transcription factors. Still, the biochemical steps in normal growth control remain incompletely de- fined, as are the mechanisms responsible for conver- sion of a normal cell to a cancer cell by activated oncogenes. It is increasingly evident that cancer re- sults from a series of genetic events, in which inacti- vation of tumor suppressor genes is likely to be at least as important as activation of oncogenes. Furthermore, the role of proto-oncogenes in devel- opmental events has become better appreciated, but it is complicated by the complementary functions of several members within gene families. Our current studies of proto-oncogenes are largely confined to two gene families: 1) the src family, the first member of which was discovered as the cellular progenitor of the v-src oncogene of RSV, and 2) the Wnt family, the first member of which was found as the target for insertional mutation by MMTV. The src Gene Family v-src encodes a membrane-associated phospho- protein of 60,000 daltons (pp60%~5"5) that displays protein kinase activity in vitro and induces phospho- rylation of tyrosine residues in several proteins in vivo. The functionally significant target molecules have not been identified, however, and the physio- logical functions of the cellular progenitor, c-src, are not known, Mutations that augment the tyrosine ki- nase activity of pp60 appear to be required to convert c-sre into an oncogene. Studies of c-src have been strongly influenced in the past few years by the recognition that it is one of about eight genes encod- ing very similar but differentially regulated proteins. c-sre is expressed in virtually all cells, often at high levels; its mRNA is differentially spliced in the nervous system; and c-sre protein is phosphorylated by the cdc2-encoded kinase during mitosis. Nevertheless, as shown by Phil Soriano (Baylor), in- activation of the mouse c-src gene by targeted homologous recombination produces an unexpectedly limited phenotype: osteopetrosis (excessive bone de- position) resulting from defective osteoclast function. These findings suggest that src gene family members complement each other in many cell types, so that phenotypes may be best studied in specialized cells or revealed more generally after inactivation of combi- nations of such genes. To these ends, in collaboration with Soriano’s laboratory, Cliff Lowell has made targeted null muta- tions of the hck and c-fgr genes, two src gene family members expressed mainly in the hematopoietic lin- eage. Homozygous mice have been derived for these mutations, but, thus far, only subtle phenotypic changes have been found. Macrophages from hck- deficient mice phagocytize latex beads inefficiently, but even doubly homozygous Ack and fer animals appear overtly healthy and have apparently normal bone marrow and circulating blood cells. However, the doubly homozygous mice are unable to protect themselves against infection with Listeria monocytogenes. In addition, mice lacking both src and Ack are anemic and leukopenic and lack splenic germinal centers. Maho Niwa is studying the basis of this hematopoietic phenotype. Ken Kaplan has used immunofluorescent and cell fractionation methods to localize c-src protein in cultured fibroblasts. Although the product of v-src is mostly present in the plasma membrane, especially in adhesion plaques, p60°-s’c is associated mainly with endosomes, particularly in the region of the micro- tubule organizing center in interphase cells and near the spindle pole bodies during mitosis. Ken and Krissy Bibbins are introducing c-src mutants into src- deficient cells (supplied by Soriano) to study the de- terminants of the localization pattern and search for functional consequences of the association with en- dosomes. Preliminary findings suggest that src may have a role in maintenance of cell adhesion and that the kinase domain is neither necessary nor sufficient to direct sre protein to focal adhesions. ANNUAL REPORT: JULY 1992-FUNE 1993 Several years ago, we isolated a host-dependent mutant of v-src, a mutant that transformed chicken cells efficiently and mammalian cells poorly, and found a single amino acid change in the most highly conserved region of a non-catalytic domain, called SH2, that mediates interactions with phosphotyrosine and is found in many kinds of signaling proteins. As evidence accumulated for the importance of the SH2 domain, Hisamaro Hirai introduced many site- directed mutations into these regions of a c-src gene made active for transformation. Several of these mutants are also host-dependent, with greater trans- forming activity in either avian or mammalian cells. Multiple approaches are being taken to exploit the properties of these mutants to probe the interac- tions between src proteins and other cellular proteins. John Murphy is seeking chicken genes and mouse cell mutants that restore the ability of host-dependent SH2 mutants to transform mouse cells. In addition, he is attempting to generate intragenic, second site mutants that restore competence to transform mouse cells and testing the idea that differences in protein- tyrosine phosphatases may account for the host range phenomena. Helene Boeuf and Krissy Bibbins have produced wild-type and mutant SH2 domains as fu- sion proteins in E. coli. They have used these proteins as affinity reagents with extracts of mammalian and avian ceils and with phosphotyrosine peptides on beads. The results are largely consistent with the recently published structure of the src SH2 domain bound to phosphotyrosine, although at least one residue predicted to contact phosphotyrosine can be altered without apparent loss of binding activity. SH2 mutants, even those with strong host-depen- dence for transformation, bind phosphotyrosine-con- taining proteins in avian or mammalian cell extracts in accord with their ability to bind phosphotyrosine- containing peptides on beads. Thus, host-dependence remains unexplained, but an important phenomenon. The Wnt Gene Family The Wnt-/ gene was discovered in 1982, using the technique known as “transposon tagging” to iden- tify genes that serve as targets for insertion mutation during mammary tumor induction by MMTV. About 75% of tumors in C3H mice have MMTV insertion mutations that activate expression of Wnt-/. (In mammary tumors in some other mouse strains, addi- tional genes have been isolated as targets for MMTV insertion mutations. The best-studied of these, the int-2 gene, is a member of the fibroblast growth fac- tor gene family. Another, int-3, encodes a transmem- brane receptor of unknown function.) The Drosophila homolog of Wnr-/ is the seg- ment polarity gene, wingless. Furthermore, mammals 265 HAROLD E. VARMUS contain several genes closely related to Wnt-J and homologs have been cloned from a wide range of vertebrate and invertebrate species (see below). These genes have been strongly implicated in devel- opmental events by their restricted patterns of expres- sion, particularly in the embryo; by the consequences of null mutations (cerebellar and mid-brain malfor- mations after targeted mutation of mouse Wat-/, segmentation defects with wingless mutations); and by the induction of axis duplication after injection of Wnt mRNAs into the early frog embryo. The nucleotide sequence of Wnt-] cDNA pre- dicts a protein of 370 amino acids, with a signal pep- tide, a cysteine-rich carboxyterminus, and four N- linked glycosylation sites. Over the past few years, we have shown that the primary product is subject to multiple modifications in cultured cells, including proteolytic cleavage and several glycosylations, producing at least five distinguishable forms of Wnt-1 protein in the secretory pathway. At least two of these forms are secreted and associated with the extracellular matrix and the cell surface. Jan Kitajewski also found that Wnt-1 proteins are associated with a 78 Kd protein known as BiP, which binds to certain secretory proteins in the endoplasmic reticulum. Unlike most other secretory proteins, Wnt pro- teins have not been obtained in a soluble, biologically active form. This fact has severely limited studies of the structure and function of Wnt proteins and may account for the failure to identify cell surface recep- tors for Wnt proteins. We have attempted to compensate for this problem by developing assays for Wnt genes in cultured cells. Although most cell lines do not appear to respond to the introduction of Wnt genes, we have found two that do: a mouse mammary epithelial cell line, called C57MG, that responds by dramatic morphological change and enhanced growth potential; and the rat pheochromocytoma line, PC12, that changes shape, becomes more adherent, and loses responsiveness to NGF. In addition, because Wnt proteins are secreted and appear to act locally, we can assay genes in a paracrine manner by expressing them in a non responsive cell and observing changes in adjacent, non expressing responsive cells (C57MG). Site-directed mutants of Wnt-/ generated by John Mason and Jan Kitajewski have shown that the signal peptide is required for entry into the secretory pathway, association with BiP, and biological activ- ity. Mutation of conserved cysteine residues results in loss of autocrine and paracrine activities. None of the glycosylation sites appears to be important for bi- ological activity: Wnt-J is active in both autocrine and paracrine assays even when all four glycosylation sites have been mutated. Two mutants, one altering a 266 cysteine residue and one a glycosylation site, are temperature-sensitive. Jan, Dave Leonardo (on a rotation project), and Neil Parkin have found that Wnt-1 chimeras with heterologous transmembrane domains can induce partial transformation of CS7MG cells and axis duplication in Xenopus embryos. Karl Willert has been attempting to exploit some of these mutants, especially the ts mutants, to implicate biochemical mechanisms, such as tyrosine Phosphorylation, in the Wnt signaling pathway. He has described the step in NGF-induced signaling that is blocked by Wnt-1 protein in PC12 cells; the trk- encoded NGF receptor is autophosphorylated in a normal fashion in Wnt-1-expressing PC12 cells, and other early consequences of NGF occur normally, but the cell fails to respond by production of neurite and neuronal markers. Jan Taylor and Sophie Roy (a postdoctoral fellow in the Ganem laboratory) recently found that hepatoma cells co-transfected with vectors expressing Wnt-1 protein and hepatitis B surface antigen efficiently secrete particles that contained both proteins in immunoprecipitable form. Importantly, the particles exhibit biological activity. They induce morphological changes in C57MG cells and can be used to attach the cells to plastic. The availability of a cell-free, biologically active form of Wnt protein might be used to seek Wnt receptors and genes that encode them. Transgenic mice carrying a Wnt-/ gene acti- vated by an MMTV LTR exhibit marked epithelial hyperplasia in the mammary glands in both male and female mice, and the females have a high incidence of mammary carcinoma indistinguishable from the virus-induced disease. In addition, some males have mammary carcinomas, and a few transgenic animals have developed salivary gland carcinomas. The transgenic mice document the oncogenic potential of Wnt-1, but they also indicate that this single oncogene is not sufficient for tumorigenesis: Only a few cells appear to develop into mammary carcinomas after several months of age, despite the diffuse hyperplasia of mammary tissue. We are therefore looking for cellular genes that might collaborate with Wnt-/ dur- ing tumorigenesis. By crossing our Wat-/ transgenic mice with transgenic mice that carry the int-2 gene under control of the MMTV LTR, we observed a marked acceleration of tumorigenesis in bitransgenic mice compared to the Wat-I transgenics. Another approach also illustrates the collaboration of Wnt-J and int-2: Greg Shackleford and Helen Kwan infected Wnt-J transgenic mice with MMTV; mammary tumors appeared at least one to two months earlier in virgin or breeding females after MMTV infection than in control animals, and most of the tumors had new MMTV proviruses in a pattern that implies clonal growth of an infected cell. Nearly half of these tumors had proviral insertions in int-2 or in hst (an adjacent gene that also belongs to the FGF gene family). Linda Yuschenkoff and Lucy Godley are per- forming additional crosses between our Wnt-/ trans- genic animals and animals with interesting transgenes expressed in the mammary gland.(int-3) and animals with targeted mutations in tumor suppressor genes (e.g., the p53 gene and Rb gene, both implicated in human breast cancer). In collaboration with Larry Donehower, we have found that tumors arise much earlier in the absence of a p53 gene. Lucy is also using differential cDNA cloning to identify genes that may influence conversions from hyperplastic to neoplastic to metastatic phases of carcinogenesis, and making transgenic mice that express wild-type and mutant p53 genes in the mammary gland. ANNUAL REPORT: JULY 1992-JUNE 1993 As an alternative approach to the function of the Wnt gene family, we have been collaborating with Cynthia Kenyon’s laboratory to study the two nematode homologs of mammalian and insect Wnt genes. Both Ce-Wnt genes have been sequenced (by Lily Shiue, Greg Shackleford, and Supriya Shivakumar) and the predicted proteins show about 35% amino acid identity with other Wnt proteins, with nearly complete conservation of cysteine residues. Supriya has identified a psoralen-induced mutant of C. elegans-Wnt-] that appears to produce embryonic lethality. She is now testing genomic clones and several derivatives for their ability to res- cue the mutant or produce additional phenotypes, and she and Greg Jongeward are seeking more mutants of the two Ce-Wnt genes. Antisera have also been raised against Ce-Wnt protein (by Supriya and Neil Parkin) and are being used to attempt to localize sites of expression. PUBLICATIONS Bates, P., J.A.T. Young, and H.E. Varmus (1993). A putative receptor for subgroup A Rous sarcoma virus is related to the LDL receptor. Cell 74: 1043-1051. Bibbins, K.B., H. Boeuf, and H.E. Varmus (1993). Binding of the src SH2 domain to phosphopeptides is determined by residues in both the SH2 domain and the phosphopeptides. Mol. Cell. Biol. 13: 7278-7287. Bishop, J.M., M. Kirschner, and H-E. Varmus (1993). Policy Forum: Science and the new administration. Science 259: 444-445, Leavitt, A.D., L. Shiue, and H.E. Varmus (1993). Site-directed mutagenesis of HIV-1 integrase demonstrates that integrase functions are separable in vitro. J. Biol. Chem. 268: 2113-2119. Parkin, N.T., J. Kitajewski, and H.E. Varmus (1993). Activity of Wnt-1 as a transmembrane protein. Genes & Development 7: 2181-2193. Shackleford, G.M., K. Willert, J. Wang, and H.E. Varmus (1993). Wnt-1 proto-oncogene induces changes in morphology, gene expression, and growth factor responsiveness in PC12 cells. Neuron 11: 865-875. Shackleford, G.M., C.A. MacArthur, H.C. Kwan, and H.E. Varmus (1993). Mouse mammary tumor virus infection accelerates mammary carcinogenesis in Wnt-1 transgenic mice by insertional activation of int-2/Fef-3 and hstilFgf-4. Proc. Natl. Acad. Sci. USA 90: 740-744. Shackleford, G.M., S. Shivakamur, L. Shiue, J. Mason, C. Kenyon, and H.E. Varmus (1993). Two wat genes in Caenorhabditis elegans. Oncogene 8: 1857-1864. Stamatoyannopoulos, G., A.W. Neinhus, P.W. Majorus, and H.E. Varmus, eds. (1993). The Molecular Basis of Blood Diseases. W.B. Saunders, Philadelphia. Young, J.A.T., P. Bates, and H.E. Varmus (1993). Isolation of the chicken gene encoding the receptor for subgroup A- specific avian leukosis and sarcoma virus. J. Virology 67: 1811-1816. IN PRESS Miller, H.-P., P.M. Pryciak, and H.E. Varmus (1993). Retroviral integration machinery as a probe for DNA structure and associated proteins (1993). Cold Spring Harbor Symposia on Quantitative Biology, in press. Verderame, M.F., and H.E. Varmus (1993). Highly conserved amino acids in the SH2 and catalytic domains of v- sre are altered in naturally occurring, transformation-defective alleles. Oncogene, in press. 267