PROTEIN DNA INTERACTIONS

JAN FENG, Ph.D., Associate Member

The aim of the research conducted in this laboratory is to study the structure/function relationships of proteins and their interactions with DNA at the atomic level. X-ray crystallography, molecular biology, and computer modeling are the experimental tools employed by our group. Our current research focuses on the mechanisms of site-specific DNA recombination, protein-DNA interactions, and proteins associated with signal transduction pathways.

DNA recombination is one of the most basic biological processes in living organisms playing an essential role in diversifying, maintaining, and transferring genes. Any miscue in the process can cause serious and sometimes fatal consequences for the cell or organism. Our research is concerned with the molecular mechanism of a model site-specific DNA recombination system, the Hin reaction, involving an inversion of a 1kb segment of DNA in Salmonella. This reaction controls the alternate expression of two flagella genes, which enables the bacteria to evade host cell attack. The inversion requires three proteins, Hin, factor for inversion stimulation (Fis), and HU. Biochemical data have suggested that the initiation of the reaction requires two recombination sites bound with Hin dimers paired together in association with an enhancer bound with Fis proteins, forming a nucleoprotein complex called the invertasome.

DNA binding proteins are generally classified into sequence-specific DNA binding proteins and nonsequence-specific DNA binding proteins. Biochemical and structural studies on these proteins, particularly on sequence-specific DNA binding proteins, have generated a wealth of information on protein-DNA interactions. This knowledge has greatly enhanced our understanding of how proteins recognize their target DNA sites.

Our efforts have focused on determining the crystal structure of a sequence-specific DNA binding protein, Fis, in complex with its target sites. First discovered for its activity in stimulating inversion reactions in the Hin (recombinase)- and Gin (invertase)-mediated site-specific DNA recombination in Salmonella and phage Mu, respectively, the Fis protein has been shown to function in a variety of biological processes including transcriptional activation, transcriptional repression and DNA replication. Fis is a protein of 11 kilodaltons with 98 amino acids, and binds DNA as a homodimer. One of the most interesting biological functions of Fis is its DNA binding property. Of more than 60 biologically functional sites that are specifically targeted by Fis, no consensus sequence is evident. Unlike the target sites of most dimeric DNA-binding proteins, the Fis-binding sites lack an apparent dyad symmetry. It is not clear whether such sequential asymmetry is related to the specificity of Fis binding. Biochemical data suggest that Fis induces severe DNA bending upon binding. It is possible that the specificity of Fis binding arises from the unique sequence composition of its binding sites, which are capable of undergoing conformational changes in order to accommodate the binding surface of the Fis protein.

The crystal structure of Fis (Yuan, et al., Proc. Natl. Acad. Sci. USA 88:9558, 1991) showed that it adopts an a-helical structure with four helices in each subunit. The two monomers are interlocked with extensive interactions between them. The helix-turn-helix DNA binding motif constitutes the main component of the DNA binding surface. Circular permutation assays showed that Fis induces large bends in DNA and that the overall bending angle is dependent on the sequences flanking both sides of a 15-base pair core binding site. Based on a computer model of the Fis/DNA complex, Cu²⁺-mediated site-specific cleavage experiments identified potential protein-DNA contacts that are responsible for the observed DNA bending. It is apparent that an induced fit mechanism is involved in the Fis/DNA recognition process. We are currently in the process of working towards determining the crystal structure of Fis complexed with various DNA sequences.

The significant role of nonsequence-specific DNA binding proteins in biological processes has only recently been fully appreciated. High mobility group 1 (HMG-1)-like proteins are among the most abundant proteins in eukaryotic cells. They have been implicated in a number of basic cellular reactions including DNA replication, transcription, and recombination. They are commonly described as architectural proteins, in light of their apparent ability to facilitate biological reactions by creating a looped DNA structure that is optimal for the assembly of nuclear protein complexes. Recent studies have also shown that HMG-1 like proteins can stimulate p53 activation and V(D)J recombination in vitro.

The functional roles of HMG-1-like proteins are closely related to their DNA binding properties. These proteins are known to bind preferentially to unusual DNA structures such as four-way Holliday junctions, cruciform DNA, and prebent DNA structures. Perhaps one of the more intriguing DNA binding properties of the HMG-1 like proteins is the induction of DNA bending; this property is apparently directly responsible for their architectural functions. To understand how these architectural proteins regulate cellular reactions, we must have detailed information on their interactions with the DNA, particularly on protein-induced DNA bending. However, the nature of non-specific DNA binding, which inevitably generates non-homogeneous complexes in solution, has so far hindered our ability to study HMG-1 protein-DNA interactions.

We have developed a novel method to study the DNA bending induced by non-specific HMG-1 proteins. This method is based on a fusion protein that has been engineered to link a sequence specific DNA-binding domain (DBD) with the HMG-1 protein. The sequence-specific DBD of the fusion protein is expected to target its binding site, thus bringing the non-specific HMG-1 domain onto the DNA at a localized position, producing homogeneous complex molecules in solution (Figure 1). A fusion gene linking the Hin-DBD with NHP6A, a yeast HMG-1-like protein, has been constructed. The fusion protein was expressed in BL21 Escherichia coli cells, and was purified to homogeneity by ion-exchange column chromatography. Electrophoretic mobility shift assays have demonstrated that the fusion protein indeed binds specifically to a DNA. We also carried out DNA binding competition assays, which measures the DNA binding affinity of proteins. These experiments showed that the fusion protein has higher affinity for the target DNA site than either domains (Hin-DBD and NHP6A) alone. Knowing that the fusion protein binds DNA with high affinity and that it is localized onto a specific target site, we performed circular permutation assays, an electrophoretic method that measures the DNA bending angles. The result of this experiment suggests that the fusion protein bends DNA by 71º.

Since it is known that Hin-DBD does not bend DNA, it is reasonable to assume that the observed DNA bending can be attributed to the binding of NHP6A.

Efforts to determine DNA bending by other HMG-1 like proteins are underway. We believe the knowledge learned from this study will help us to understand the architectural roles of HMG-1 proteins in cells.

STRUCTURAL AND FUNCTIONAL STUDIES OF THE AKT PROTEIN. TANG, CRASTO, FENG, in collaboration with TSICHLIS^c

Akt, also known as PKB, is a cytoplasmic serine-threonine protein kinase encoded by the akt proto-oncogene. Since its discovery by Drs. Testa and Tsichlis and colleagues at Fox Chase (Bellacosa et al, Science 254:274, 1991), the Akt protein has been the subject of intense research effort to explore its function in cellular signaling. Recent studies have attributed a number of important biological responses to Akt, including cell survival and gluconeogenesis.

There appears to be two Akt regulatory pathways in the cell: the growth factor-dependent pathway, which is mainly associated with the phosphoinositide 3-kinase (PI3K), and the pathways induced by cellular stress such as heat shock and hyperosmolarity, which activate the p38/HOG1 kinase cascade. Based on studies of growth factor mediated Akt activation, it is now known that PI3K generated phosphoinositol is a direct substrate of Akt. Upon binding to its specific phospholipid, INS(3,4)P, the Akt protein is activated in vivo. Biochemical data suggest that the activation of Akt is closely related to the conformational changes of its structure.

Akt is a protein of 480 amino acids and can be divided into two main domains: the N-terminal Pleckstrin homology (PH) domain (residues 1-109), and the kinase domain (residues 148-411). The two domains are linked by a highly acidic segment of 37 amino acids (residues 110-147), which is unique to Akt. The remaining residues (412-480) constitute the carboxy-terminal tail of the protein.

We have built a computer model of the Akt protein based on available structural information on PH and kinase domains. Our model revealed two positive pockets as potential phospholipid binding sites on the surface of the Akt PH domain. Site-directed mutagenesis has produced evidence that supports the potential functions for these pockets. One of the more challenging questions regarding the structure of Akt is the relative orientation of the PH and the kinase domains. The acidic linker may play a critical role in determining the conformations of active and inactive Akt. Mutational studies also suggested that the carboxy-terminal tail of Akt regulates activation of this protein.

We are currently working towards a structure determination of the intact Akt protein by X-ray crystallographic techniques to provide a structural basis for the mechanism of Akt activation.

Based on the crystal structure of DNA polymerase b (pol b)/DNA complex determined by Kraut and colleagues in 1996, we have identified a catalytic groove on the Nterminal 8-kDa domain of pol b. The helix-hairpin-helix (HhH) motif, which forms one side of the catalytic groove, plays a major role in mediating specific recognition of deformed local DNA structure including pre-incised apurinic/apyrimidinic (AP) sites. When the HhH motif in a protein is bound to DNA, it positions its hairpin loop parallel to the phosphate backbone. Since this provides only limited surface contact, it is unlikely such binding could achieve high affinity with regular DNA. On the other hand, in binding to its natural substrate, which is a pre-incised AP site, the HhH motif of the 8-kDa domain may locate the 5' deoxyribose phosphate group at the opening of the catalytic groove by electrostatic interactions. A re-positioning of the 5' deoxyribose phosphate group may allow for the establishment of additional contacts, thereby securing the binding. Indeed, in vitro DNA binding assays show that the 8-kDa domain has a significantly higher binding affinity for nicked AP sites than for AP nucleotides that have not been nicked. We are interested in determining the structure of the 8-kDa domain bound with DNA containing a nicked AP site. This structure should provide the basis for understanding the unique DNA recognition by the HhH motif.

BELLACOSA, A., CHAN, T.O., AHMED, N.N., DATTA, K., MALSTROM, S., STOKOE, D., McCORMICK, F., FENG, J., TSICHLIS, P. Akt activation by growth factors is a multiple-step process: the role of the pH domain. Oncogene 17(3):313-325, 1998.

MATSUMOTO, Y., KIM, K., KATZ, D.S., FENG, J. Catalytic center of DNA polymerase b; for excision of deoxyribose phosphate groups. Biochemistry 37(18):6456-6464, 1998.

FENG, J., CRASTO, C.J., MATSUMOTO, Y. Deoxyribose phosphate excision by the N-terminal domain of the polymerase b: the mechanism revisited. Biochemistry 37:(27):9605-9611, 1998.

^§   Fox Chase researcher

^a   D. Katz: Present address-Agnes Irwin School, Bryn Mawr, PA 19010

^b   R. Johnson: UCLA School of Medicine, Los Angeles, CA 90005

^c   P. Tsichlis: Present address-Kimmel Cancer Institute, Thomas Jefferson University, Philadelphia, PA 19107

Illustrations or unpublished data in these reports should not be used without permission of the author.

Fox Chase Cancer Center

Scientific Report 1998