Functional interaction of Jun and homeodomain proteins.

We have used the yeast two-hybrid system to identify proteins that interact with the N-terminal region of c-Jun, which is known to be involved in regulatory interactions. One of the proteins identified is the homeodomain-containing protein Hex. The Hex homeodomain is sufficient for interaction; moreover, the homeodomains of several other transcription factors also interact. Mutations within helix III of the Hex homeodomain greatly reduce the interaction. In vitro, c-Jun/c-Fos, JunB/c-Fos, and JunD/c-Fos all interact with the Hex homeodomain more strongly than the respective Jun proteins (or c-Fos) alone, suggesting that heterodimerization exposes reactive regions in the N termini of the Jun proteins. In transfected cells, Hex expression inhibits Jun- or Jun/c-Fos-dependent transcription of a reporter gene; the presence of Hex-binding sites in the promoter enhances the inhibitory effect. Jun-dependent activation of transcription from the basic fibroblast growth factor gene, previously shown to be regulated by both Jun and homeodomain proteins, was also dramatically reduced by Hex expression. Furthermore, in contrast to the reduction of Jun-mediated transcription by Hex, we found that expression of the Drosophila ultrabithorax gene enhanced c-Jun-dependent transcription. We conclude that the functional interaction between members of the Jun and homeodomain families of transcription factors could play a critical role in regulating developmental and differentiation programs.

Conditional gene expression in human intracranial xenograft tumors.

In an initial effort to determine the effect of expressing potentially therapeutic gene products on the growth properties of glioma tumor xenografts, we describe the development of cell lines that can conditionally express beta-galactosidase (beta-gal). To achieve this, we generated stable cell lines that express the modified tetracycline repressor molecule (rtTA) and the beta-gal gene under control of tetracycline-responsive cis-elements. The resulting cell lines express functional beta-gal following treatment with the tetracycline analog doxycyclin (Dox). These cells were then used to form intracranial tumors after injection into the brain using an implantable guide-screw system. The xenografts were found to express beta-gal when the animals were fed drinking water containing Dox. From these studies, we conclude that the expression of a target gene in a human xenograft growing in the brain of a living mouse can be conditionally regulated.

Oncostatin M activates stat DNA binding and transcriptional activity in primary human fetal astrocytes: low- and high-passage cells have distinct patterns of stat activation.

In this study we explored the activation of the JAK/Stat pathway by gp 130 family cytokines in primary human astrocytes. We report that of four gp 130 cytokines tested, only oncostatin M (OnM) resulted in the activation of Stat molecules. To test that the induced molecules were transcriptionally active, transcription from a Stat-responsive reporter plasmid (from the acute-phase gene alpha-2 macroglobulin) transiently transfected into astrocytes was assessed after activation by OnM and was blocked by cotransfection with dominant-negative Stat3 encoding plasmids strongly suggesting that the activation was Stat-mediated. While DNA binding complexes comprised of both Stat1 and Stat3 were induced in low-passage cells, only those containing Stat3 were formed by extracts from high-passage cells. Stat1 protein was detected in the cytoplasm of high-passage cells indicating that the inability to form SIF-B and -C complexes was due to a lack of activation of Stat1 rather than a lack of expression. These results indicate a fundamental difference between low- and high-passage astrocytes in response to cytokine treatment that might result in distinct patterns of gene expression through altered ratios of activated Stat3 and Stat1.

Expression of protein mediators of type I interferon signaling in human squamous cell carcinoma of the skin.

IFN-based therapy has been shown to be active in the treatment of squamous cell carcinoma (SCC) of the skin and has promise for chemoprevention and treatment of several other cancers. In an effort to better understand the molecular mechanism of this activity, we have determined the expression pattern of several of the protein mediators of type I IFN signaling by immunohistochemistry in cutaneous SCC, SCC metastases, and adjacent nonmalignant epithelium from patient biopsies. All of the proteins, signal transducer and activator of transcription (STAT) 1alpha/beta, STAT2, p48, STAT3a, and STAT3beta, are expressed at varying levels in the adjacent epidermis, as well as in other epidermal and dermal cell types. For the majority of samples tested, the expression of one or more of these proteins was reduced in SCC primary tumors compared with the adjacent nonmalignant epithelial cells, as determined by manual scoring. Quantitative densitometry of several samples revealed differences that are statistically significant. Our study provides the first direct evidence for the expression of the IFN-stimulated gene factor 3 (STAT1alpha/beta, STAT2, and p48) and STAT3alpha and STAT3beta mediators of IFN-alpha/beta signaling in human skin and skin-derived SCCs. These data have led to the hypothesis that the loss of IFN sensitivity may contribute to the development and progression of skin SCC.

Dimer stability as a determinant of differential DNA binding activity of Stat3 isoforms.

Stat3alpha and Stat3beta are two Stat3 isoforms with marked quantitative differences in their DNA binding activities. To examine the molecular basis of the differential DNA binding activities, we measured DNA binding strength and dimer stability, two possible mechanisms responsible for these differences. Stat3alpha and Stat3beta showed no difference in DNA binding strength, i.e. they had similar association and dissociation rates for DNA binding. However, competition analyses performed with dissociating reagents including an anti-phosphotyrosine antibody, SH2 domain protein, and a phosphopeptide demonstrated that Stat3beta dimers are more stable than Stat3alpha dimers. We report here that dimer stability of activated forms plays a critical role in determining DNA binding activity of Stat3 isoforms. We found that C-terminal deletions of Stat3alpha increased both DNA binding activity and dimer stability of Stat3alpha. Our findings suggest that the acidic C-terminal region of Stat3alpha does not interfere with the DNA binding of activated Stat3alpha dimers, but destabilizes the dimeric forms of Stat3alpha. We propose that dimer stability described in vitro may be the underlying mechanism of in vivo stability of activated Stat3 proteins, regulating dephosphorylation of tyrosine 705.

Activation of stat3 and stat1 DNA binding and transcriptional activity in human brain tumour cell lines by gp130 cytokines.

In this study we examine the activation of the latent Stat family of transcription factors by the gp130 family of cytokines in cell lines derived from human brain tumours. Of the cytokines tested, oncostatin M resulted in the most dramatic induction of Stat1 and Stat3 in all cell lines analysed, as assessed by the formation of protein/DNA complexes. Interleukin-6, leukemia inhibitory factor, and ciliary neurotrophic factor also induced Stat complexes more selectively and to a lesser magnitude than oncostatin M. The kinetics of Stat1 and Stat3 activation was rapid and transient; the nuclear accumulation of DNA binding-proficient Stat protein was detected in the nucleus within minutes of cytokine induction. The transcriptional potential of the oncostatin M-activated Stat molecules was demonstrated in two glioma cell lines (U87-MG, SNB-19) by transient transfection experiments using a Stat-responsive reporter plasmid. Oncostatin M-dependent transcription from this reporter plasmid was reduced to uninduced levels by the inclusion of a dominant-negative Stat3 molecule, demonstrating that Stat molecules were responsible for the induction. These studies demonstrate that oncostatin M is the most potent activator of Stat molecules in a variety of brain tumour-derived cell lines, an observation that could have implications affecting the balance between proliferation/apoptosis of these cells.

c-Src activates the DNA binding and transcriptional activity of Stat3 molecules: serine 727 is not required for transcriptional activation under certain circumstances.

Stat3 proteins are constitutively activated in cells transformed by v-Src and the proteins have been shown to interact directly. Subsequent studies have shown that Stat3 is required for cellular transformation of NIH fibroblasts by v-Src, suggesting a potential role for Stat3 in aberrant cell growth. Stat3 is phosphorylated on a single tyrosine (tyrosine 705) which is required for effective dimer formation. An additional phosphorylation event (serine 727) is believed to be required for full transcriptional activity of Stat1 and Stat3 molecules. Here we report that c-Src activates the DNA binding activity of Stat3alpha, Stat3beta, and three Stat3 mutants, one in which serine 727 was replaced by alanine (Stat3alphaS727A) and C-terminal truncated molecules Delta48 and Delta55. Consistent with this finding is a general increase in the tyrosine 705-phosphorylated Stat3 in cells cotransfected with c-Src. Furthermore, transcription from an alpha-2 macroglobulin reporter gene is activated by Stat3alphaS727A to the same magnitude as compared to Stat3alpha and Stat3beta in the presence of c-Src. These results suggest that serine 727, contained in a consensus MAP kinase recognition site and shown to be the only serine in Stat3 phosphorylated in epidermal growth factor (EGF) treated cells, is not necessary for transcriptional activity comparable to wild-type Stat3alpha or Stat3beta when activated by c-Src in COS-7 cells.

The mUBC9 murine ubiquitin conjugating enzyme interacts with the E2A transcription factors.

The ubiquitin-mediated degradation of cellular proteins requires the sequential activity of E1, E2 and, in some cases, E3 enzymes. Using the yeast two-hybrid system, we have cloned 1.0- and 2.5-kb cDNAs encoding the identical murine E2, or ubiquitin conjugating enzyme by virtue of its interaction with the E2A transcription factor. This cDNA encodes the 158-amino-acid protein, mUBC9, which has considerable sequence homology to UBC9 from Saccharomyces cerevisiae and HUS5 from Schizosaccharomyces pombe and is identical to the human UBC9 protein. HUS5 is essential for DNA damage repair, whereas UBC9 is necessary for G2/M progression. The human protein has been shown to correct the UBC9 defect in yeast. Antisera raised against bacterially expressed mUBC9 fusion protein recognize a murine cellular protein of approximately 18 kDa, corresponding to the predicted mobility. Unlike E2A, the mUBC9 protein level is not regulated by serum growth factors. The activity of the apparent homologues UBC9 and HUS5 suggests that mUBC9 may be involved in the degradation of key nuclear proteins that regulate cell cycle progression.

Functional differences between Stat3alpha and Stat3beta.

Stat3beta is a short form of Stat3 that differs from the longer form (Stat3alpha) by the replacement of the C-terminal 55 amino acid residues of Stat3alpha by 7 residues specific to Stat3beta. In COS cells transfected with Stat3 expression plasmids, both Stat3alpha and Stat3beta were activated for DNA binding and transcription by the same set of growth factors and cytokines and both, when activated, formed homodimers and heterodimers with Stat1. Only Stat3beta was active in the absence of added cytokine or growth factor. Activation of each form, including constitutive activation of Stat3beta, was correlated with the phosphorylation of tyrosine 705. Activated Stat3beta in transfected COS cells was more stable and had greater DNA-binding activity than activated Stat3alpha. However, relative to DNA-binding activity, Stat3alpha showed greater transcriptional activity than Stat3beta. A mutant of Stat3alpha lacking its highly acidic C-terminal 48 amino acids had properties indistinguishable from Stat3beta. We conclude that Stat3alpha and Stat3beta have significantly different properties due to the presence or absence of the acidic C-terminal tail of Stat3alpha rather than the C-terminal sequence peculiar to Stat3beta. In addition to its effect on transcription, we speculate that the acidic tail may destabilize the active dimeric form of Stat3alpha, resulting in lower DNA-binding activity of the Y705-phosphorylated form compared to Stat3beta and in more rapid dephosphorylation.

In vitro activation of Stat3 by epidermal growth factor receptor kinase.

Stat proteins are SH2 domain-containing transcription factors that are activated in cells by various cytokines and growth factors. In the case of cytokines whose receptors lack protein kinase activity, phosphorylation-activation is mediated by members of the JAK family of tyrosine protein kinases. In the case of growth factors whose receptors have intrinsic tyrosine protein kinase activity, it is thought that Stat proteins can be activated either directly by the receptor or indirectly through JAK proteins. To test the possibility of direct activation, we have used purified Stat3 alpha, Stat3 beta, and epidermal growth factor receptor kinase produced in recombinant baculovirus-infected Sf9 insect cells. The Stat proteins formed a stable complex with the receptor kinase, and they were phosphorylated on tyrosine by the receptor kinase and activated for binding to DNA, properties shared with Stat proteins purified from Sf9 cells coexpressing JAK1 or JAK2. Both JAK-phosphorylated Stat3 beta and Stat3 beta phosphorylated in vitro by the receptor kinase were 20-50 times more active on a molar basis for DNA binding than phosphorylated Stat3 alpha. We conclude that Stat3 isoforms can be directly phosphorylated and thereby activated in vitro by the epidermal growth factor receptor kinase.

Granulocyte colony-stimulating factor activation of Stat3 alpha and Stat3 beta in immature normal and leukemic human myeloid cells.

Granulocyte colony-stimulating factor (G-CSF) is the cytokine critical for directing neutrophilic granulocyte differentiation. Acute myelogenous leukemia (AML) cells, which frequently arise from this lineage, respond aberrantly to G-CSF by proliferating without differentiating. The basis for this abnormal responses is unknown. In the present study, we investigated whether G-CSF signaling in immature normal and leukemic human myeloid cells diverges at the level of activation of signal transducers and activators of transcription (STAT) proteins. We compared the profile of STAT proteins activated in G-CSF-stimulated immature normal and leukemic human myeloid cells. G-CSF activated Stat3 alpha in all AML cell lines examined except HL60 and in three of six uncultured AML patient samples. In normal human CD34+ bone marrow cells and HL60 cells, both reported to differentiate in response to G-CSF, G-CSF did not activate Stat3 alpha; rather, it activated only an 83 kD form of Stat3 that proved to be the human homologue of a short form of Stat3, Stat3 beta. Because the transcriptional activity of Stat3 beta is distinct from Stat3 alpha, these results suggest that the balance of the two Stat3 isoforms in myeloid cells may influence the cellular pattern of gene activation and consequently the ability of these cells to differentiate in response to G-CSF.

Cooperative transcriptional activity of Jun and Stat3 beta, a short form of Stat3.

To identify proteins that regulate the transcriptional activity of c-Jun, we have used the yeast two-hybrid screen to detect mammalian polypeptides that might interact functionally with the N-terminal segment of c-Jun, a known regulatory region. Among the proteins identified is a short form of Stat3 (designated Stat3 beta). Stat3 beta is missing the 55 C-terminal amino acid residues of the long form (Stat3 alpha) and has 7 additional amino acid residues at its C terminus. In the absence of added cytokines, expression of Stat3 beta (but not Stat3 alpha) in transfected cells activated a promoter containing the interleukin 6 responsive element of the rat alpha 2-macroglobulin gene; coexpression of Stat3 beta and c-Jun led to enhanced cooperative activation of the promoter. Nuclear extracts of cells transfected with a Stat3 beta expression plasmid formed a complex with an oligonucleotide containing a Stat3 binding site, whereas extracts of cells transfected with a Stat3 alpha plasmid did not. We conclude that there is a short form of Stat3 (Stat3 beta), that Stat3 beta is transcriptionally active under conditions where Stat3 alpha is not, and that Stat3 beta and c-Jun are capable of cooperative activation of certain promoters.

Selection of Arabidopsis cDNAs that partially correct phenotypes of Escherichia coli DNA-damage-sensitive mutants and analysis of two plant cDNAs that appear to express UV-specific dark repair activities.

To resist terrestrial UV radiation, plants employ DNA-damage-repair/toleration (DRT) activities, as well as shielding mechanisms. Little is known about the structure and regulation of plant DRT genes. We isolated DRT cDNAs from Arabidopsis thaliana, by selecting for complementation of Escherichia coli mutants lacking all bacterial defenses against UV-light damage to DNA. These mutants are phenotypically deficient in recombinational and mutagenic toleration (RecA-), excision repair (Uvr-) and photoreactivation (Phr-). Among 840 survivors of heavily UV-irradiated (10(-7) survival) mutants harboring plasmids derived from an Arabidopsis cDNA library in the vector lambda YES, we identified four unique plant cDNAs, designated DRT100, DRT101, DRT102, and DRT103. Drt101 and Drt102 activity were specific for UV-light damage, and complemented both UvrB- and UvrC- phenotypes in the dark. Apparent Uvr- correction efficiencies were 1 to 40% for Drt101, and 0.2 to 15% for Drt102, depending on the UV fluence. Drt101 and Drt102 showed no extensive amino-acid homology with any known DNA-repair proteins. Drt100 appeared to correct RecA-, rather than Uvr-, phenotypes. Although the light dependence of Drt103 activity was consistent with its identification as a photoreactivating enzyme, its predicted amino-acid sequence did not resemble known photolyase sequences. The N-terminal coding sequence of Drt101 suggests that it is targeted to chloroplasts, as reported for Drt100. These cDNAs afforded only modest increases in survival during the original selection procedure. The fact that they were readily isolated nevertheless suggests that selections may be made powerful enough to overcome barriers to expression and function in bacteria, at least for cDNAs of reasonable abundance.

Bacteriophage P1 Bof protein is an indirect positive effector of transcription of the phage bac-1 ban gene in some circumstances and a direct negative effector in other circumstances.

Previous genetic studies have suggested that the Bof protein of bacteriophage P1 can act as both a negative and a positive regulator of phage gene expression: in bof-1 prophages, the ref gene and a putative phage ssb gene are derepressed, but expression of an operator-semiconstitutive variant of the phage ban gene (bac-1) is markedly reduced. An explanation of this apparent duality is suggested by recent reports that Bof is a corepressor of genes that are regulated by the phage C1 repressor, including the autoregulated c1 gene itself. Here we show, by means of operon fusions to lacZ, that the balance points between Bof-mediated decreases in c1 expression and Bof-mediated increases in C1 efficacy are different among various C1-regulated genes. Thus, expression of Bof by P1 prophages affects some genes (e.g., bac-1 ban) positively, and others (e.g., ref) negatively. Even at bac-1 ban, where the positive indirect effect of Bof is physiologically dominant, Bof can be seen to act as a corepressor if C1 is supplied from a nonautoregulated (ptac-c1) source, eliminating the effect of Bof on C1 synthesis.

The bof gene of bacteriophage P1: DNA sequence and evidence for roles in regulation of phage c1 and ref genes.

The C1 repressor of bacteriophage P1 acts via 14 or more distinct operators. This repressor represses its own synthesis as well as the synthesis of other gene products. Previously, mutation of an auxiliary regulatory gene, bof, has been shown to increase expression of some C1-regulated P1 genes (e.g., ref) but to decrease expression of others (e.g., ban). In this study the bof gene was isolated on the basis of its ability to depress stimulation of Escherichia coli chromosomal recombination by the P1 ref gene, if and only if a source of C1 was present. C1 alone, but not Bof alone, was partially effective. The bofDNA sequence encodes an 82-codon reading frame that begins with a TTG codon and includes the sites of the bof-1(Am) mutation and a bof::Tn5 null mutation. Expression of ref::lacZ and cl::lacZ fusion genes was partially repressed in trans by a P1 bof-1 prophage or by plasmid-encoded C1 alone, which was in agreement with effects on Ref-stimulated recombination and with previous indirect evidence for c1 autoregulation. Repression of both fusion genes by plasmid-encoded C1 plus Bof or by a P1 bof+ prophage was more complete. When the C1 source also included a 0.7-kilobase region upstream from C1 which encodes the coi gene, repression of both c1::lacZ and ref::lacZ by C1 alone or by C1 plus Bof was much less effective, as if Coi interfered with C1 repressor function.

Enhancement of Escherichia coli plasmid and chromosomal recombination by the Ref function of bacteriophage P1.

The Ref activity of phage P1 enhances recombination between two defective lacZ genes in the Escherichia coli chromosome (lac- x lac- recombination). Plasmid recombination, both lac- x lac- and tet- x tet-, was measured by transformation of recA strains, and was also assayed by measurement of beta-galactosidase. The intracellular presence of recombinant plasmids was verified directly by Southern blotting. Ref stimulated recombination of plasmids in rec+ and rec(BCD) cells by 3-6-fold, and also the low level plasmid recombination in recF cells. RecA-independent plasmid recombination, either very low level (recA cells) or high level (recB recC sbcA recA cells), was not stimulated. Ref stimulated both intramolecular and intermolecular plasmid recombination. Both normal and Ref-stimulated lac- x lac- chromosomal recombination, expected to be mostly RecBC-dependent in wild-type bacteria, were affected very little by a recF mutation. We have previously reported Ref stimulation of lac- x lac- recombination in recBC sbcB bacteria, a process known to be RecF-dependent. Chromosomal recombination processes thought to involve activated recombination substrates, e.g., Hfr conjugation, P1 transduction, were not elevated by Ref activity. We hypothesize that Ref acts by unknown mechanisms to activate plasmid and chromosomal DNA for RecA-mediated recombination, and that the structures formed are substrates for both RecF-dependent (plasmid, chromosomal) and Rec(BCD)-dependent (chromosomal) recombination pathways.

Suppression of UAA and UGA termination codons in mutant murine leukemia viruses.

Genomes of mammalian type C retroviruses contain a UAG termination codon between the gag and pol coding regions. The pol region is expressed in the form of a gag-pol fusion protein following readthrough suppression of the UAG codon. We have used oligonucleotide-directed mutagenesis to change the UAG in Moloney murine leukemia virus to UAA or UGA. These alternate termination codons were also suppressed, both in infected cells and in reticulocyte lysates. Thus, the signal or context inducing suppression of UAG in wild-type Moloney murine leukemia virus is also effective with UAA and UGA. Further, mammalian cells and cell extracts contain tRNAs capable of translating UAA and UGA as amino acids. To our knowledge, this is the first example of natural suppression of UAA in higher eucaryotes.