Subcellular localization of the alpha and beta subunits of the acute myeloid leukemia-linked transcription factor PEBP2/CBF.

Each of the two human genes encoding the alpha and beta subunits of a heterodimeric transcription factor, PEBP2, has been found at the breakpoints of two characteristic chromosome translocations associated with acute myeloid leukemia, suggesting that they are candidate proto-oncogenes. Polyclonal antibodies against the alpha and beta subunits of PEBP2 were raised in rabbits and hamsters. Immunofluorescence labeling of NIH 3T3 cells transfected with PEBP2 alpha and -beta cDNAs revealed that the full-size alpha A1 and alpha B1 proteins, the products of two related but distinct genes, are located in the nucleus, while the beta subunit is localized to the cytoplasm. Deletion analysis demonstrated that there are two regions in alpha A1 responsible for nuclear accumulation of the protein: one mapped in the region between amino acids 221 and 513, and the other mapped in the Runt domain (amino acids 94 to 221) harboring the DNA-binding and the heterodimerizing activities. When the full-size alpha A1 and beta proteins are coexpressed in a single cell, the former is present in the nucleus and the latter still remains in the cytoplasm. However, the N- or C-terminally truncated alpha A1 proteins devoid of the region upstream or downstream of the Runt domain colocalized with the beta protein in the nucleus. In these cases, the beta protein appeared to be translocated into the nucleus passively by binding to alpha A1. The chimeric protein containing the beta protein at the N-terminal region generated as a result of the inversion of chromosome 16 colocalized with alpha A1 to the nucleus more readily than the normal beta protein. The implications of these results in relation to leukemogenesis are discussed.

Construction of a cDNA to the hamster CAD gene and its application toward defining the domain for aspartate transcarbamylase.

cDNA complementary to hamster mRNA encoding the CAD protein, a multifunctional protein which carries the first three enzymes of pyrimidine biosynthesis, was constructed. The longest of these recombinants (pCAD142) covers 82% of the 7.9-kilobase mRNA. Portions of the cDNA were excised and replaced by a lac promoter-operator-initiation codon segment. The resultant plasmids were transfected into an Escherichia coli mutant defective in aspartate transcarbamylase, the second enzyme of the pathway. Complementation of the bacterial defect was observed with as little as 2.2 kilobases of cDNA sequence, corresponding to the 3' region of the mRNA. DNA sequencing in this region of the hamster cDNA reveals stretches which are highly homologous to the E. coli gene for the catalytic subunit of aspartate transcarbamylase; other stretches show no homology. The highly conserved regions probably reflect areas of protein structure critical to catalysis, while the nonconserved regions may reflect differences between the quaternary structures of E. coli and mammalian aspartate transcarbamylases, one such difference being that the bacterial enzyme in its native form is allosterically regulated and the mammalian enzyme is not.

PEBP2 alpha B/mouse AML1 consists of multiple isoforms that possess differential transactivation potentials.

A murine transcription factor, PEBP2, is composed of two subunits, alpha and beta. There are two genes in the mouse genome, PEBP2 alpha A and PEBP2 alpha B, which encode the alpha subunit. Two types of the alpha B cDNA clones, alpha B1 and alpha B2, were isolated from mouse fibroblasts and characterized. They were found to represent 3.8- and 7.9-kb transcripts, respectively. The 3.8-kb RNA encodes the previously described alpha B protein referred to as alpha B1, while the 7.9-kb RNA encodes a 387-amino-acid protein, termed alpha B2, which is identical to alpha B1 except that it has an internal deletion of 64 amino acid residues. Both alpha B1 and alpha B2 associate with PEBP2 beta and form a heterodimer. The alpha B2/beta complex binds to the PEBP2 binding site two- to threefold more strongly than the alpha B1/beta complex does. alpha B1 stimulates transcription through the PEBP2 site about 40-fold, while alpha B2 is only about 25 to 45% as active as alpha B1. Transactivation domain is located downstream of the 128-amino-acid runt homology region, referred to as the Runt domain. Mouse chromosome mapping studies revealed that alpha A, alpha B, and beta genes are mapped to chromosomes 17, 16, and 8, respectively. The last two genes are syntenic with the human AML1 on chromosome 21q22 and PEBP2 beta/CBF beta on 16q22 detected at the breakpoints of characteristic chromosome translocations of the two different subtypes of acute myeloid leukemia. These results suggest that previously described chimeric gene products, AML1/MTG8(ETO) and AML1-EAP generated by t(8;21) and t(3;21), respectively, lack the transactivation domain of AML1.

Isolation of PEBP2 alpha B cDNA representing the mouse homolog of human acute myeloid leukemia gene, AML1.

Breakpoints of the t(8;21) chromosome translocation in acute myeloid leukemia are clustered within the human gene, AML1, located on chromosome 21 [Miyoshi, H., Shimizu, K., Maseki, N., Kaneko, Y. & Ohki, M. (1991). Proc. Natl. Acad. Sci. USA, 88, 10431-10434]. The product of AML1 has a region about 130 amino acids long that is highly homologous to the Drosophila segmentation gene runt (runt homology region). The cDNA isolated from mouse fibroblasts encoding the alpha-subunit of polyomavirus enhancer binding protein 2 (PEBP2/PEA2) revealed that it also has a runt homology region (E. Ogawa et al., submitted). In this study, a different cDNA clone presumed to represent the mouse homolog of human AML1 (PEBP2 alpha B) was isolated from a cDNA library derived from B cells. The deduced amino acid sequence of PEBP2 alpha B is 99% identical to that of AML1 for the first 241 residues, including the runt homology region, though their sequences diverge thereafter. On the other hand, PEBP2 alpha B and PEBP2 alpha share only 92% and 82% homologies at the amino acid and nucleotide levels respectively, even for the runt homology region, indicating that these proteins are encoded by distinct genes. While PEBP2 alpha is highly expressed in T-cell lines but not in most of the B-cell lines and functions as an activator of T-cell-specific genes, PEBP2 alpha B is expressed in both types of cells. A possible functional relationship between PEBP2 alpha and PEBP2 alpha B is discussed in relation to leukemogenic potential of AML1.

Mutant rho factors with increased transcription termination activities. I. Functional correlations of the primary and secondary polynucleotide binding sites with the efficiency and site-selectivity of rho-dependent termination.

We have characterized rho proteins from mutants of Escherichia coli, rho s-81 and rho s-82, which are hyperactive in termination. The two mutant rho proteins are differentially altered both in termination activities and in RNA interactions. rho s-81 generally elicits enhanced termination on various templates such as phage T7 DNA and a DNA restriction fragment containing the trpE intracistronic rho-dependent terminators, either measured as a whole or examined for individual sites. On the other hand, rho s-82 has strikingly different preferences toward individual termination sites, exhibiting overall termination activities higher or lower than normal, depending on templates. From measurements of the rho ATPase activity with T7 RNA and various homoribopolymers as cofactors, both mutant rho proteins are shown to have broadened RNA base specificities in contrast to the stringent requirement for cytosine observed with the wild-type rho. Functional tests on the two kinds of polynucleotide binding sites known for rho have indicated that rho s-81 is mainly altered in the primary site, whereas rho s-82 is simultaneously affected in the secondary binding site as well as the primary site. Thus, we conclude that the primary and secondary sites contribute distinctly in determining the overall efficiency and site-specificity of termination, respectively. Further analysis of detailed termination points at the trpE and lambda tR1 terminators has revealed that major RNA transcripts generated by the wild-type rho and rho s-81 are notably rich in adenine and poor in cytosine for the 3'-terminal five to ten nucleotides, whereas those preferentially terminated by rho s-82 are conversely richer in cytosine than adenine. This finding suggests that rho may recognize the RNA-DNA hybrid region at the 3' end of a nascent transcript in its secondary binding reaction.

Mutant rho factors with increased transcription termination activities. II. Identification and functional dissection of amino acid changes.

We have determined the nucleotide sequences of three mutant rho genes encoding hyperfunctional rho proteins (rho S) together with their parent allele, rho-ts702. These mutant rho factors contain the following amino acid changes as deduced from their sequences: (1) the thermo-labile mutant, rho-ts702, has Thr304 substituting for Ala; (2) rho S-77 and rho S-81, which are selectively altered in the primary polynucleotide binding site, share an identical mutation, Leu3----Phe; (3) rho S-82, which is altered in both the primary and secondary polynucleotide binding sites, carries three amino acid substitutions together, Leu3----Phe, Asp156----Asn and Thr323----Ile. Dissection and functional characterization of each mutation in rho S-82 have revealed that Ile323 alone is responsible for alterations in both the secondary RNA interaction and the terminator selectivity observed with the original mutant, rho S-82. Taken together, these results not only confirm our proposal in the accompanying paper that the primary and secondary RNA binding sites differently contribute in determining the overall efficiency and site-specificity of termination, respectively, but also support the possibility that these binding sites exist as structurally distinct domains in rho protein. In contrast, Asn156 was shown to cause decreased termination efficiency, though it had no influence on RNA interactions. Thus, this amino acid residue appears to be associated with still another rate-determining step of termination, for instance, interactions between rho and RNA polymerase. On the basis of Chou-Fasman secondary structure predictions as well as amino acid sequence comparison with F1-ATPase, we discuss how the proposed domains are structurally and functionally related to the putative ATPase reactive center of rho protein.

Structural and functional dissections of transcription termination factor rho by random mutagenesis.

Transcription termination factor rho from Escherichia coli is a homohexamer of 419 amino acid subunits and catalyzes an ATP-dependent release of nascent RNA transcripts. A rho monomer has three distinct domains functioning independently at the first approximation: the amino-terminal one quarter containing a primary RNA-binding site, the central 270-amino acids region constituting an ATP-binding domain with homologies to F1-ATPase, and the carboxy-terminal remainder with unknown function(s). To further delineate the structural and functional organizations of rho protein, we undertook its random mutagenesis using error-prone polymerase chain reactions with the carboxy-terminal 100-amino acid region chosen as the initial target. From 14 mutants identified, rho protein was purified and characterized in vitro. Of these, 11 mutants are defective in termination in vivo and show decreased activities in various partial functions examined: ATP binding; RNA binding; and ATPase activities dependent on three cofactors with decreasing efficacies, poly(C), lambda cro RNA and poly(U). A few of them are also affected in the putative secondary RNA-binding site that is functionally coupled to ATP hydrolysis. By contrast, the three other mutants are hyperactive in termination, poly(U)-dependent ATPase activity, and RNA interaction at the primary site. In these properties, the hyper-terminating mutants strikingly resemble the "super rho" mutant formerly found in the amino-terminal domain. Taken together, these findings indicate that the carboxy-terminal region plays a pivotal role in functionally coupling the RNA and ATP-binding domains, plausibly by acting as an interface for their interaction within or across individual subunits. In light of the reported X-ray crystallographic structure of F1-ATPase, we propose a model for the tertiary and quaternary structure of rho that is consistent with the observed mutational effects as well as a number of structural and functional properties characteristic of rho.

The quaternary geometry of transcription termination factor rho: assignment by chemical cross-linking.

Transcription termination factor rho from Escherichia coli is a ring-shaped homohexamer of 419 amino acid subunits and catalyzes an ATP-dependent release of nascent RNA transcripts. Previous chemical cross-linking studies suggested that the rho hexamer might have D3 symmetry with three isologous dimers as protomers. However, our recent mutational analysis of rho alongside its putative structural homology to F1-ATPase rather argued for C6 symmetry. To resolve this discrepancy, we have re-investigated the pattern of cross-linking of rho using various cross-linkers with different functional groups and spacer lengths. Upon reaction with dimethyl suberimidate followed by SDS-polyacrylamide gel electrophoresis, rho protein generated a series of cross-linked oligomers up to hexamers, of which dimers migrated as distinct doublet bands of approximately equal intensities. However, the lower band became much stronger than the upper one with dimethyl adipimidate and difluorodinitrobenzene, and vice versa with disuccinimidyl glutarate, disuccinimidyl suberate and disulfosuccinimidyl tartarate. Furthermore, the trimeric products also produced doublet bands, whose relative intensities were again variable with cross-linkers, but in an inverse correlation with those of the dimer bands. These results combined with theoretical considerations support a C6 symmetry model in which cross-linking is assumed to occur stochastically at one of two alternative sites within each subunit interface with variable relative frequencies depending on cross-linkers. The D3 symmetry is excluded, for the putative trimeric subspecies should always retain mutually equal intensities in that case. Detailed inspections of the cross-linking kinetics further revealed a moderate characteristic of C3 symmetry for the rho hexamer such that the collective as well as relative rates of cross-linking at the two available sites could fluctuate between alternating interfaces. The final model designated as C3/6 is also compatible with other functional and structural properties known for rho.

Three-dimensional reconstruction of transcription termination factor rho: orientation of the N-terminal domain and visualization of an RNA-binding site.

The Escherichia coli rho transcription termination protein is a hexameric helicase, and is believed to function by separating an RNA-DNA hybrid. Unlike hexameric DNA helicases, where a single strand of DNA passes through the central channel, it has been proposed that the RNA wraps around the outside of the ring. We have generated a three-dimensional reconstruction of rho, and localized a tRNA molecule bound to the primary RNA-binding site to the outside of the ring. An atomic structure of the N-terminal domain of rho fits into our reconstruction uniquely, with the residues involved in RNA-binding on the outside of the ring. Although rho shares a common structural core with the F1-ATPase and other hexameric helicases, there has been a divergence in function due to rho's N-terminal domain, which has no homology to other helicases.

Dual mutations in the AML1 and FLT3 genes are associated with leukemogenesis in acute myeloblastic leukemia of the M0 subtype.

Point mutations of the transcription factor AML1 are associated with leukemogenesis in acute myeloblastic leukemia (AML). Internal tandem duplications (ITDs) in the juxtamembrane domain and mutations in the second tyrosine kinase domain of the Fms-like tyrosine kinase 3 (FLT3) gene represent the most frequent genetic alterations in AML. However, such mutations per se appear to be insufficient for leukemic transformation. To evaluate whether both AML1 and FLT3 mutations contribute to leukemogenesis, we analyzed mutations of these genes in AML M0 subtype in whom AML1 mutations were predominantly observed. Of 51 patients, eight showed a mutation in the Runt domain of the AML1 gene: one heterozygous missense mutation with normal function, five heterozygous frameshift mutations and two biallelic nonsense or frameshift mutations, resulting in haploinsufficiency or complete loss of the AML1 activities. On the other hand, a total of 10 of 49 patients examined had the FLT3 mutation. We detected the FLT3 mutation in five of eight (63%) patients with AML1 mutation, whereas five of 41 (12%) without AML1 mutation showed the FLT3 mutation (P=0.0055). These observations suggest that reduced AML1 activities predispose cells to the acquisition of the activating FLT3 mutation as a secondary event leading to full transformation in AML M0.

Construction and characterization of plasmid and lambda phage vector systems for study of transcriptional control in Escherichia coli.

We constructed a family of lambda phage and plasmid vectors which facilitate cloning and quantitative analysis of transcriptional regulator in both single and multiple copies. Their expression system was modified from the ara-trp-lac fusion operon of plasmid pMC81 [Casadaban and Cohen, J. Mol. Biol. 138 (1980) 179-207], which is designed to assay both promoters and terminators with a single vehicle. To eliminate transcriptional and translational polar effects liable to occur in the original fusion operon upon insertion of a foreign nucleotide sequence, intracistronic Rho-dependent terminators, that are present within the trpB gene and distal to the cloning site were deleted, and DNA spacers containing stop codons were introduced immediately before and after the cloning site. In analysis of the cloned trp regulatory region, the lambda phage system faithfully reproduced the tight regulation by tryptophan characteristic to the natural trp operon on the E. coli chromosome, whereas the plasmid counterpart exhibited a substantially relaxed response. Comparative studies on the relative strengths of various promoters and terminators have further demonstrated that the lambda phage vector system permits accurate assays of exceptionally strong promoters like Ptrp and lambda pL without disturbing the bacterial growth, while being sensitive enough for detecting low-level transcription under the control of weak promoters or potent terminators. Cloning with the lambda phage vector can be greatly facilitated by transferring the target regulatory site precloned with the plasmid onto the phage genome through in vivo recombination.

Overproduction of transcription termination factor Rho in Escherichia coli.

A plasmid system has been constructed which allows high-level expression of the rho gene of Escherichia coli under the control of the pL promoter and the N-antitermination regulatory system of bacteriophage lambda. The pL-directed synthesis of Rho crucially depends on the lambda N gene product and is promoted most effectively when this product is supplied from the N gene cloned on a separate compatible plasmid with a moderate copy number. The requirement for N can be circumvented partly, but not completely, by deletion of the region preceding the rho structural gene. Attempts were also made to optimize the construction of rho-expression plasmids by adjusting the orientation and location of pL and rho inserts on the pBR322 vector. With optimal conditions, Rho protein is overexpressed 100-fold and can become as much as 10% of the total cellular protein. Using this plasmid system, Rho can be purified with a yield of more than 20 mg from 10 g of induced cells.

Nucleotide sequence and molecular evolution of mouse retrovirus-like IAP elements.

We determined the nucleotide (nt) sequences of cDNA and genomic clones for murine intracisternal type A particle (IAP) elements, which are retrovirus-like repetitive sequences in rodent genomes. The nucleotide sequence of the cDNA resembled that of retrovirus RNA genomes in its lack of the U5 sequence within the 3' long terminal repeat. By sequence comparison of our clones with reported rodent IAP elements, we located the probable gag, pol and env gene regions. The sequences for the pol, env and the 3' two-thirds of the gag region were conserved among the IAP elements. In the regions, synonymous substitutions occurred more frequently than non-synonymous ones, which suggested that the regions in question were functionally constrained until fairly recently. The rate of nucleotide substitutions in the regions was estimated to be 6-10 X 10(-9) nt per site per year, and significantly higher than that of the cellular genes. These rates may exemplify a characteristic of the nucleotide substitutions for an endogenous retrovirus. The sequence homology between the IAP element and IgE-binding factor gene is discussed.

A simple screening for mutant DNA binding proteins: application to murine transcription factor PEBP2alpha subunit, a founding member of the Runt domain protein family.

Mouse transcription factor PEBP2 (polyomavirus enhancer-binding protein (2) is composed of two distinct subunits alpha and beta. The alpha subunit has an ability to bind the specific DNA sequences, which is enhanced by formation of a heterodimer with the beta subunit. The DNA binding and heterodimerization activities of the alpha subunit are both localized within a 128-amino-acid (aa) region termed as the Runt domain for its homology to the Drosophila segmentation gene runt. To characterize the molecular determinants for these activities, the Runt domain was randomly mutagenized and produced in E. coli as a secreted form. Using E. coli culture supernatant, the DNA binding and heterodimerization of mutant Runt domains were analyzed by gel retardation assay. Nine randomly picked single-aa substitution mutants showed various functional alterations in DNA binding and heterodimerization either separately or simultaneously. This observation suggests that the structure of Runt domain is highly ordered and is quite sensitive to modulations in its primary structure. The method presented here provides a simple and quick method to characterize a large number of mutant DNA binding proteins.

Expression of the gag gene of human T-cell leukemia virus type I in Escherichia coli and its diagnostic use.

An expression plasmid, pHY202, was constructed which directs the synthesis of a fusion protein encoded by the gag sequence of human T-cell leukemia virus type I (HTLV-I) inserted into the lacZ' gene. Escherichia coli cells harboring pHY202 produced the 43-kDal LacZ'-Gag fusion protein with a yield of approx. 0.3% of total soluble proteins. The fusion protein is specifically recognized by monoclonal antibodies against the Gag proteins p19 and p24, and could be applicable for the diagnosis of HTLV-I infection, because almost all sera from HTLV-I carriers gave a positive response in the enzyme-linked immunosorbent assay (ELISA) employing the LacZ'-Gag hybrid protein purified by immunoaffinity column chromatography.

Cloning, mapping and expression of PEBP2 alpha C, a third gene encoding the mammalian Runt domain.

PEBP2/CBF is a heterodimeric transcription factor composed of alpha and beta subunits. Previously, we reported two distinct mouse genes, PEBP2 alpha A and PEBP2 alpha B, which encode the alpha subunit. PEBP2 alpha B is the homologue of human AML1, encoding the acute myeloid leukemia 1 protein. AML1 and human PEBP2/CBF beta were detected independently at the breakpoints of two characteristic chromosome translocations observed frequently in two subtypes of acute myeloid leukemia. The PEBP2 alpha proteins contain a 128-amino-acid (aa) region highly homologous to the Drosophila melanogaster segmentation gene runt. The evolutionarily conserved region, named the Runt domain, harbors DNA-binding and heterodimerizing activities. In this study, we identified the third Runt-domain-encoding gene, PEBP2 alpha C, which maps to 1p36.11-p36.13 in the human chromosome and encodes a 415-aa protein. PEBP2 alpha C forms a heterodimer with PEBP2 beta, binds to the PEBP2 site and transactivates transcription, similar to PEBP2 alpha A and PEBP2 alpha B.

Construction of a new plasmid vector that can express cloned cDNA in all translational reading frames.

Construction of a bacterial expression vector, pSI4001, is described. The vector contains the lac promoter-operator and three sets of ribosome-binding sites (RBSs) tandemly arranged in all possible reading frames. cDNA can be directly cloned downstream from these translational start points in the fixed and proper orientation by using the method of Okayama and Berg [Mol. Cell. Biol. 3 (1982) 280-289]. The open reading frame of any cDNA inserted may be automatically aligned in phase with either of the three ATG start codons, thus enabling its expression with a maximum theoretical probability of unity. Fusion with the lacZ gene (coding for beta-galactosidase) has shown that at least two of the three translation initiation sites exhibit high expression capacities and the remaining one can also function at a lower but significant rate. We used the vector to construct a bovine pituitary cDNA library, from which clones coding for prolactin were detected by immunological screening with an efficiency as high as two in three clones. The construction with triple RBSs should also provide a unique experimental model to study the regulation of overlapping translations.

Further analysis of cDNA clones for maize phosphoenolpyruvate carboxylase involved in C4 photosynthesis. Nucleotide sequence of entire open reading frame and evidence for polyadenylation of mRNA at multiple sites in vivo.

Four clones of cDNA for phosphoenolpyruvate carboxylase [EC 4.1.1.31] were obtained from a maize green leaf cDNA library by colony hybridization. The largest cDNA was of full-length (3335 nucleotides), being 243 nucleotides longer than the cDNA cloned previously [(1986) Nucleic Acids Res. 14, 1615-1628]. Alignment of the sequence for the N-terminal coding region found in two of the four clones with the sequence reported previously, established the sequence of the entire coding region for the enzyme. The sequencing of 3'-untranslated region of the clones revealed that the poly(A) tract is attached at multiple sites in vivo.

Polyclonal and monoclonal antibodies monospecific to MMTV LTR orf protein produced in E. coli.

Monoclonal and polyclonal antibodies specific to an open reading frame of the mouse mammary tumor virus long terminal repeat were generated using an open reading frame-beta-galactosidase fusion protein produced in E. coli. Both antibodies reacted with the open reading frame-beta-galactosidase fusion protein but not with beta-galactosidase alone using an immunoblotting technique. It is concluded that these antibodies were specific for the protein encoded by the open reading frame of the mouse mammary tumor virus long terminal repeat.