Tumor Cells Exhibit Deregulation of the Cell Cycle Histone Gene Promoter Factor HiNF-D JOOST HoLTHUIs, THOMAS A. OWEN, ANDRE J. VAN WIJNEN, KENNETH L. WRIGHT, ANNA RAMSEY-EWING, MARY BETH KENNEDY, RUTH CARTER, STEPHEN C. COSENZA, KENNETH J. SOPRANO, JANE B. LIAN, JANET L. STEIN, GARY S. STEIN Cell cycle-regulated gene expression is essential for normal cell growth and develop- ment and loss of stringent growth control is associated with the acquisition of the transformed phenotype. The selective synthesis of histone proteins during the S phase of the cell cycle is required to render cells competent for the ordered packaging of replicating DNA into chromatin. Regulation of H4 histone gene transcription requires the proliferation-specific promoter binding factor HiNF-D. In normal diploid cells, HiNF-D binding activity is regulated during the cell cycle; nuclear protein extracts prepared from normal cells in S phase contain distinct and measurable HiNF-D binding activity, while this activity is barely detectable in G1 phase cells. In contrast, in tumor-derived or transformed cell lines, HiNF-D binding activity is constitutively elevated throughout the cell cycle and declines only with the onset of differentiation. The change from cell cycle-mediated to constitutive interaction of HiNF-D with the promoter of a cell growth-controlled gene is consistent with, and may be functionally related to, the loss of stringent cell growth regulation associated with neoplastic transformation. T HE HUMAN H4 HISTONE GENE (FO108) proximal promoter con- tains two in vivo protein binding domains, sites I and II. These sites have been defined at single-nucleotide resolution, with the protein-DNA contacts confirmed by in vivo deoxyribonuclease (DNase) I protec- tion analysis (1) and native genomic blotting (2). In vitro, the site I and II sequences form specific protein-DNA complexes with at least four distinct nuclear factors (HiNF-A, HiNF-C, HiNF-D, and HiNF-E), as shown by deletion analysis, DNase I footprinting, and dimethylsulfate fingerprinting (3-5). HiNF-A, HiNF-C, and HiNF-E bind inde- pendently to the distally located site I and are present in both actively proliferating and differentiating cells (3, 5) (Fig. 1A). Dele- tion of site I results in a four- to sixfold reduction in H4 histone gene transcription in vitro (3), suggesting that the interactions of these nuclear factors with site I have an auxiliary role in augmenting the H4 histone gene transcription rate. J. Holthuis, T. A. Owen, A. J. van Wijnen, K. L. Wright, A. Ramsey-Ewing, M. B. Kennedy, J. B. Lian, J. L. Stein, G. S. Stein, Department of Cell Biology, Universi- ty of Massachusetts Medical School, 55 Lake Avenue North, Worcester, MA 01655. R. Carter, S. C. Cosenza, K. J. Soprano, Department of Microbiology and Immunology, Temple University School of Medicine, 3400 North Broad Street, Philadel- phia, PA 19140. HiNF-D interacts with a highly con- served histone-specific element (Fig. 2) lo- cated in the distal part of site II (3). This protein-DNA interaction has been implicat- ed as having an essential function in H4 histone gene transcription because deletion of the HiNF-D binding site abolishes expression in vivo (6). Furthermore, the HiNF-D-site II interaction is proliferation- specific. When HL60 promyelocytic leuke- mia cells are induced to differentiate into monocytes, the downregulation of H4 his- tone gene transcription at the completion of the proliferative phase is accompanied by a selective loss of interaction of HiNF-D with site II, as demonstrated both in vivo and in vitro (7) (Fig. 1A). A similar observation has been made in primary cultures of diploid rat calvarial osteoblasts at a transition point in their developmental sequence when pro- liferation and expression of cell growth- related genes are downregulated, and initia- tion of tissue-specific gene expression char- acteristic of the bone cell phenotype occurs (8) (Fig. LA). Hence, the loss of the HiNF- D-site II interaction in the proximal pro- moter of the H4 histone gene may represent an important event in the process whereby proliferation ceases and the genes encoding phenotypic markers of differentiated cells are progressively expressed. Because the occupancy of site II by HiNF-D appears to be essential for H4 histone gene expression, we reasoned that this protein-DNA interaction could be a rate-limiting step for the cell cycle-regulated expression of this gene. We have reported previously that in human HeLa S3 cells, the in vivo protein-DNA interactions at sites I and II of H3 and H4 histone genes persist throughout the cell cycle (1, 9) and that the binding activities of HiNF-A, HiNF-C, and HiNF-D are present during all phases of the cell cycle (3) (Fig. 1B). HeLa cells are continuously proliferating heteroploid cells derived from tumors and have lost the po- tential to differentiate. Therefore, we exam- ined the HiNF-D-site II interaction during the cell cycle of diploid cell types in which normal cell growth control mechanisms are operative. Initially, the site II binding activity of HiNF-D was monitored during the cell cycle of primary rat calvarial osteoblasts. Actively growing osteoblasts were synchro- nized by two cycles of 2 mM thymidine block, resulting in the accumulation of cells at the boundary between Gl and S phases of the cell cycle. After release from the second thymidine block, the cells progressed syn- chronously through the cell cycle, as reflect- ed by DNA synthesis [monitored by pulse labeling with [3H]thymidine and assayed by both in situ autoradiography and deter- mination of trichloroacetic acid (TCA) pre- cipitable radioactivity] and mitotic activity (Fig. 3). At intervals during the cell cycle, cytoplasmic RNA was analyzed for the pres- ence of H4 histone mRNA and nuclear protein extracts were assayed for HiNF-D- site II binding activity by gel retardation assay. Rat osteoblast cells actively engaged in DNA synthesis (S phase cells) showed ten times as much H4 histone mRNA as pre- release cells (those not released from the second thymidine block) or cells in the GI phase, reflecting at a molecular level the high degree of synchrony obtained (Fig. 4). When nuclear extracts prepared from these cells were analyzed in gel retardation assays, a striking cell cycle-dependent alteration in HiNF-D binding activity became evident. While HiNF-D binding activity in S phase nuclear extracts is abundant, it is barely detectable in extracts prepared from pre- release cells and cells in the Gl phase of the cell cycle (Fig. 5A). To determine whether the cell cycle-de- pendent alteration in HiNF-D binding ac- tivity is directly coupled to the process of DNA replication, we inhibited DNA syn- thesis in S phase rat osteoblast cells by treatment with hydroxyurea (1 mM) for 1 hour or 8 hours, after which nuclear protein extracts were prepared and analyzed for SCIENCE, VOL. 247 I454