Formation of regulator/target gene relationships during evolution

The Foxn1-like forkhead/winged-helix transcription factor genes have been maintained in single copy throughout chordate evolution. Among other functions, Foxn1 (formerly known as Whn) regulates the expression of hair keratin genes in the hair follicle, which represents an evolutionarily novel organ characteristic of mammals. We show here that fish and mouse Foxn1-like genes are functionally equivalent in hair keratin gene activation, suggesting the absence of functionally relevant changes over the course of several hundred million years of vertebrate evolution. In contrast, the Foxn1-like gene from the cephalochordate Branchiostoma lanceolatum is inactive in this assay because of changes in the region located N-terminal to DNA binding and transcriptional activation domains of the protein. Our results indicate that functionally relevant changes in cis-regulatory regions are not necessarily accompanied by corresponding changes in transcription factor proteins in the formation of evolutionarily novel regulator/target gene relationships.

Gene structure, chromosomal localization, and expression pattern of Capn12, a new member of the calpain large subunit gene family

We report the identification of mouse Capn12, a new member of the calpain large subunit gene family. It possesses potential protease and calcium-binding domains, features typical of the classical calpains. In situ hybridization and Northern blot analysis demonstrate that during the anagen phase of the hair cycle the cortex of the hair follicle is the major expression site of Capn12. The gene was sequenced in its entirety and consists of 21 exons spanning 13 kb with an exon-intron structure typical of the calpain gene family. The last exon of the mouse Actn4 gene overlaps the 3' end of Capn12 but in the opposite orientation. This overlap between the two genes is conserved in the human genome. Three versions of the Capn12 mRNA transcript were identified. They occur as a result of alternative splicing, and two of these encode a protein lacking the C-terminal calmodulin-like domain. Radiation hybrid mapping localized Capn12 to mouse chromosome 7, closely linked to a marker positioned at 10.4 cM. Refined mapping of Capn5, also previously localized to chromosome 7, indicated that it was not closely linked to Capn12, mapping tightly linked to a marker positioned at 48.5 cM.

Evolutionary conservation of gene structures of the Pax1/9 gene family

Based on amino acid sequence comparisons, Pax1 and Pax9 genes are considered to form a subgroup of vertebrate Pax genes. We show here that the gene structures of mouse Pax1, human PAX9 genes are similar to that of a single Pax1/9 related gene in Branchiostoma lanceolatum, AmphiPax1. This supports the hypothesis that Pax1 and Pax9 genes were derived from a single ancestral gene. A refined protein alignment of AmphiPax1, mouse Pax1 and human PAX9 proteins based on the determined exon boundaries indicates that sequence divergence at the C-termini may be related to the unique functions of the Pax1 and Pax9 genes in vertebrates. AmphiPax1 is expressed in adult amphioxus in the pharyngeal endoderm.

Genetically separable determinants of hair keratin gene expression

The nude locus encodes Whn, a transcription factor of the forkhead/winged-helix class. Mutations in Whn cause failure of differentiation of thymic epithelium with a corresponding lack of intrathymic T-cell development; in the skin, differentiation of follicular keratinocytes is disturbed resulting, in the formation of fragile hair shafts. Here, we describe the identification and characterization of a novel nude allele, nu(StL). nu(StL) encodes a truncated Whn transcription factor protein, designated Whn(StL), lacking the activation domain but retaining the characteristic DNA binding domain. In contrast, the previously described Whn(nu) mutant protein lacks both domains. nu(StL)/nu(StL) mice show an alymphoid thymic rudiment and lack of peripheral T cells, similar to nu/nu mice. In the skin, impaired expression of hair keratin genes mHa1, mHa2, mHa3 and mHa4, mHb3, mHb4, mHb5, and mHb6 is observed in a pattern that parallels that of nu/nu mice: both mutant alleles behave as hypomorphs with respect to the expression of these hair keratin genes. However, a significant difference between these two alleles exists for mHa5 expression, which is reduced in nu(StL)/nu(StL) but not in nu/nu mice. We show that the mutant Whn protein in nu/nu mice cannot enter the nucleus, whereas the mutant Whn protein in nu(StL)/nu(StL) mice is present in the nucleus. The antimorphic characteristic of the activation-deficient Whn(StL) protein with respect to mHa5 expression is therefore most likely caused by its non-productive interaction with other proteins at cis-regulatory regions of the mHa5 gene. Our results indicate that the molecular consequences of mutations of the Whn gene can be different and demonstrate an unexpected complexity of transcriptional control mechanisms of hair keratin genes.

Cytohesin-1 regulates beta-2 integrin-mediated adhesion through both ARF-GEF function and interaction with LFA-1

Intracellular signaling pathways, which regulate the interactions of integrins with their ligands, affect a wide variety of biological functions. Here we provide evidence of how cytohesin-1, an integrin-binding protein and guanine-nucleotide exchange factor (GEF) for ARF GTPases, regulates cell adhesion. Mutational analyses of the beta-2 cytoplasmic domain revealed that the adhesive function of LFA-1 depends on its interaction with cytohesin-1, unless the integrin is activated by exogenous divalent cations. Secondly, cytohesin-1 induces expression of an extracellular activation epitope of LFA-1, and the exchange factor function is not essential for this activity. In contrast, LFA-1-mediated cell adhesion and spreading on intercellular cell adhesion molecule 1 is strongly inhibited by a cytohesin-1 mutant, which fails to catalyze ARF GDP-GTP exchange in vitro. Thus, cytohesin-1 is involved in the activation of LFA-1, most probably through direct interaction with the integrin, and induces cell spreading by its ARF-GEF activity. We therefore propose that both direct regulation of the integrin and concomitant changes in the membrane topology of adherent T cells are modulated by dissectable functions of cytohesin-1.

Forkhead/winged-helix transcription factor Whn regulates hair keratin gene expression: molecular analysis of the nude skin phenotype

The molecular characteristics of the nude phenotype (alopecia and thymic aplasia) in humans and rodents are unknown. The nude locus encodes Whn, a transcription factor of the forkhead/winged-helix class. Expression of Whn in HeLa cells induced expression of human hair keratin genes Ha3-II and Hb5. Correspondingly, in nude mice, which are homozygous for a loss-of-function mutation of Whn, expression of mouse mHa3 and mHb5 hair keratin genes is severely reduced. Characterization of a previously identified nude allele, nu(Y), revealed a mis-sense mutation (R320C) in the DNA binding domain of Whn. This mutant protein is unable to activate hair keratin gene expression in HeLa cells. When the Whn transcription factor was expressed in two parts, one containing the N-terminal DNA binding domain and the other the C-terminal activation domain, no activation of hair keratin genes in HeLa cells was observed. However, when these two proteins were noncovalently linked by means of synthetic dimerizers, hair keratin gene expression was induced. This finding suggests that target gene activation by Whn depends on the structural integrity and physical proximity of DNA binding and activation domains, providing a molecular framework to explain the loss-of-function phenotypes of all previously characterized nude mutations. Our results implicate Whn as a transcriptional regulator of hair keratin genes and reveal the nude phenotype as the first example of an inherited skin disorder that is caused by loss of expression rather than mutation of keratin genes.

Diverse mRNA expression patterns of the mouse calpain genes Capn5, Capn6 and Capn11 during development

Calpains are a family of related proteins, originally classified on the basis of their calcium dependence and protease activity. Here we report the mRNA expression patterns during mouse development of the recently identified Capn5, Capn6 and Capn11 genes. The major expression sites of Capn5 during embryogenesis are the developing thymus, sympathetic and dorsal root ganglia. Capn6 mRNA is exclusively expressed during embryogenesis predominantly in developing skeletal and heart muscle overlapping closely with Capn3 expression domains. Expression was also observed in specific cells of the lung, kidney and placenta and in various epithelial cell types where the Capn6 mRNA appeared to be localized within the cell to the basal and apical ends. Capn11 mRNA is restricted exclusively to spermatocytes and only during the later stages of meiosis.

Whn and mHa3 are components of the genetic hierarchy controlling hair follicle differentiation

The molecular basis of the characteristic hair growth disorder in nude mice that carry a defective Whn transcription factor gene is unknown. A comparison of mRNA populations from wild-type and nude mice back skin by representational difference analysis revealed the absence of acidic hair keratin gene 3 (mHa3) mRNA in mutant mice. Whn and acidic hair keratin genes are co-expressed in hair follicles, nail forming regions and filiform papillae of the tongue: expression of the mHa3 gene is generally detectable about 1 day after Whn mRNA and rapidly ceases in its absence. Whn is strongly expressed during the anagen (growth) phase of the hair cycle in matrix, cortex and outer root sheath; its expression rapidly declines during catagen and is undetectable in telogen phases. In nude mice, low levels of mHa3 expression are maintained in nails and whisker follicles, whereas expression is completely absent in pelage hair follicles and filiform papillae. Thus, the nude phenotype represents the first example of an inherited skin disorder that is associated with the loss of expression rather than structural mutation of keratin genes. The distinct molecular difference between pelage and whisker follicles correlates with the improved mechanical stability of vibrissae in nude mice, implicating mHa3 as an important structural component of the hair shaft.

Predetermined chromosomal deletion encompassing the Nf-1 gene

Complex chromosomal rearrangements (deletions, inversions, translocations) are a hallmark of human tumour cells. Yet, the generation of animal models for gross chromosomal abnormalities still presents a formidable challenge. Here, we describe a versatile procedure for chromosomal engineering that was used to generate an ES cell line with a megabase deletion encompassing the tumour suppressor gene neurofibromatosis-1 (Nf-1) on mouse chromosome 11, which is often deleted in tumours of neural crest origin. Homologous recombination into sites flanking Nf-1 was used to introduce artificial sequences (triple-helix, loxP, vector backbone) that can be employed for in vitro recovery of intervening sequences or the generation of in vivo deletions. This strategy may be developed into a scheme by which large chromosomal regions with precisely defined end points may be excised from mammalian cells and reintroduced after suitable in vitro modification.

CAPN11: A calpain with high mRNA levels in testis and located on chromosome 6

Calpains are a superfamily of related proteins, some of which have been shown to function as calcium-dependent cysteine proteases. In mammals, eight different calpains have been identified. We report the identification of a new mammalian calpain gene, CAPN11. The predicted protein possesses the features typical of calpains including potential protease and calcium-binding domains. The CAPN11 mRNA exhibits a highly restricted tissue distribution with highest levels present in testis. Radiation hybrid mapping localized the gene to human chromosome 6, within a region mapped to p12. Phylogenetic analysis suggests that, in mammals, the predicted CAPN11 protein is most closely related to CAPN1 and CAPN2. However, of the calpain sequences available, the predicted CAPN11 sequence exhibits greatest homology to the chicken micro/m calpain. Thus CAPN11 may be the human orthologue of micro/m calpain. The discovery of this new calpain emphasizes the complexity of the calpain family, with members being distinguished on the basis of protease activity, calcium dependence, and tissue expression.

The role of matrix metalloproteinase activity in the maturation of human capillary endothelial cells in vitro

Vessel maturation during angiogenesis (the formation of new blood vessels) is characterized by the deposition of new basement membrane and the downregulation of endothelial cell proliferation in the new vessels. Matrix remodeling plays a crucial, but still poorly understood role, in angiogenesis regulation. We present here a novel assay system with which to study the maturation of human capillary endothelial cells in vitro. When human dermal microvascular endothelial cells (HDMEC) were cultured in the presence of dibutyryl cAMP (Bt2) and hydrocortisone (HC), the deposition of a fibrous lattice of matrix molecules consisting of collagens type IV, type XVIII, laminin and thrombospondin was induced. In basal medium (without Bt2 and HC), HDMEC released active matrix metalloproteinases (MMPs) into the culture medium. However, MMP protein levels were significantly reduced by treatment with Bt2 and HC, while protein levels and activity of endogenous tissue inhibitor of MMPs (TIMP) increased. This shift in the proteolytic balance and matrix deposition was inhibited by the specific protein kinase A inhibitors RpcAMP and KT5720 or by substituting analogues without reported glucocorticoid activity for HC. The addition of MMP inhibitors human recombinant TIMP-1 or 1,10-phenanthroline to cultures under basal conditions induced matrix deposition in a dose-dependent manner, which was not observed with the serine protease inhibitor epsilon-amino-n-caproic acid (ACA). The deposited basement membrane-type of matrix reproducibly suppressed HDMEC proliferation and increased HDMEC adhesion to the substratum. These processes of matrix deposition and downregulation of endothelial cell proliferation, hallmarks of differentiating new capillaries in the end of angiogenesis, were recapitulated in our cell culture system by decreasing the matrix-degrading activity. These data suggest that our cell culture assay provides a simple and feasible model system for the study of capillary endothelial cell differentiation and vessel maturation in vitro.

A novel serine protease overexpressed in the hair follicles of nude mice

We have studied changes in gene expression between normal and nude mouse skin. Using the method of representational difference analysis, a cDNA encoding a novel serine protease designated BSSP was cloned. Our results show that in skin this gene is predominantly expressed in the sebaceous gland of the hair follicle and in the distal part of the outer root sheath. Thus, BSSP is the first serine protease known to be expressed in the sebaceous gland. In nude mouse skin, this gene is overexpressed.

Disruption of the KEX1 gene in Pichia pastoris allows expression of full-length murine and human endostatin

Endostatin is a potent angiogenesis inhibitor. In order to isolate sufficient quantities of soluble protein for in vivo studies in mice, we expressed murine endostatin in Pichia pastoris. Analysis of the expressed protein by mass spectrometry indicated that the protein was truncated. N-terminal sequence analysis determined that the N-terminus was intact, suggesting that the C-terminal lysine was missing. In Saccharomyces cerevisiae, Kex1p can cleave lysine and arginine residues from the C-terminus of peptides and proteins. We hypothesized that the KEX1 homologue in P. pastoris is responsible for the loss of the C-terminal lysine of endostatin. To test this hypothesis, we cloned and disrupted the P. pastoris KEX1 gene. Although the overall amino acid identity between the P. pastoris and the S. cerevisae Kex1p is only 36%, the amino acid residues involved in the catalytic activity or close to the active residues are highly conserved. Disruption of the KEX1 reading frame allowed expression of murine and human endostatin with the C-terminal lysine. The KEX1 disruption strain may be a useful tool for the expression of other proteins with a C-terminal basic amino acid. Addition of a lysine to the C-terminus of recombinant proteins may protect the C-terminus from degradation by other carboxypeptidases.

Zinc-binding of endostatin is essential for its antiangiogenic activity

Endostatin is a potent angiogenesis inhibitor in vitro and in vivo. We used the yeast Pichia pastoris to express and purify soluble endostatin. It was discovered that metal chelating agents can induce N-terminal degradation of endostatin. We theorized that a metal was removed from endostatin which changed the conformation and allowed a contaminating protease to degrade the N-terminus. Atomic absorption and amino acid analysis of endostatin purified from Pichia pastoris and mammalian cells showed a 1:1 molar ratio of Zn2+ to protein. Ding et al. have shown that histidines 1, 3, 11, and aspartic acid 76 coordinate the Zn2+ atom (1). An H1/3A double, an H11A, and a D76A single mutant of endostatin were not able to regress Lewis lung carcinoma. We conclude that the ability of endostatin to bind Zn2+ is essential for its antiangiogenic activity.

Zinc-dependent dimers observed in crystals of human endostatin

The crystal structure of human endostatin reveals a zinc-binding site. Atomic absorption spectroscopy indicates that zinc is a constituent of both human and murine endostatin in solution. The human endostatin zinc site is formed by three histidines at the N terminus, residues 1, 3, and, 11, and an aspartic acid at residue 76. The N-terminal loop ordered around the zinc makes a dimeric contact in human endostatin crystals. The location of the zinc site at the amino terminus, immediately adjacent to the precursor cleavage site, suggests the possibility that the zinc may be involved in activation of the antiangiogenic activity following cleavage from the inactive collagen XVIII precursor or in the cleavage process itself.

Expressed sequences as candidates for a novel tumor suppressor gene at band 13q14 in B-cell chronic lymphocytic leukemia and mantle cell lymphoma

Deletions affecting the interval between the RB1 gene and marker D13S25 at band 13q14 are the most frequent genetic abnormalities of B-cell chronic lymphocytic leukemia (B-CLL) and indicate the presence of a novel tumor suppressor gene in this region. In the current study, a high resolution physical map of fragments spanning one megabasepair (Mb) of genomic DNA at the critical 13q14 segment was constructed. To define the minimal region of loss within the RB1 and D13S25 interval, we screened 322 B-CLLs for deletions at either of the two loci. Thirty mantle cell lymphomas (MCLs) were included in the analysis because we observed a 13q14 deletion pattern similar to B-CLL in this disease. The incidence of 13q14 deletions was 51% in B-CLL and 70% in MCL, respectively. No frequent loss of the BRCA2 gene at band 13q12 was found. Detailed deletion mapping at band 13q14 with probes from the RB1-D13S25 interval lead to the identification of a critical deletion region 400 kb in size. Within this region two segments were most frequently affected, one at D13S272 120 kb in size and another 240 kb distal of D13S272 80 kb in size. From these two segments expressed sequences were identified as candidates for the putative 13q14 tumor suppressor gene involved in the pathogenesis of B-CLL and MCL.

Genomic organization of mouse Capn5 and Capn6 genes confirms that they are a distinct calpain subfamily

CAPN5 and CAPN6 are recently identified human and mouse genes lacking a calmodulin-like domain with homology to the calpain family of proteases. To clarify their relationship to the known calpains, we have compared their genomic organization and chromosome location with other human calpain gene family members. In the mouse, both genes have 11 introns of identical location, with 6 of these being similar in location to those of the known vertebrate members. Surprisingly, there were no splice junctions in common with the nematode gene tra-3, the calpain with highest homology to CAPN5 and CAPN6. CAPN5 is localized on human chromosome 11, closely linked to the mu-calpain gene CAPN1. CAPN6, which is expressed only in the placenta, is localized on the X chromosome, to which no other calpain has yet been mapped.

Positional cloning and gene identification

After genetic mapping and physical representation of a particular genomic region containing the gene underlying a particular Mendelian trait, a successful positional cloning strategy depends on the efficient detection and analysis of genes in the critical interval. Several gene detection strategies are presently available to compile an inventory of genes from large genomic regions. Here, the principle of these methods is briefly reviewed and their relative value for positional cloning projects compared.

Antiangiogenic therapy of experimental cancer does not induce acquired drug resistance

Acquired drug resistance is a major problem in the treatment of cancer. Of the more than 500,000 annual deaths from cancer in the United States, many follow the development of resistance to chemotherapy. The emergence of resistance depends in part on the genetic instability, heterogeneity and high mutational rate of tumour cells. In contrast, endothelial cells are genetically stable, homogeneous and have a low mutational rate. Therefore, antiangiogenic therapy directed against a tumour's endothelial cells should, in principle, induce little or no drug resistance. Endostatin, a potent angiogenesis inhibitor, was administered to mice bearing Lewis lung carcinoma, T241 fibrosarcoma or B16F10 melanoma. Treatment was stopped when tumours had regressed. Tumours were then allowed to re-grow and endostatin therapy was resumed. After 6, 4 or 2 treatment cycles, respectively, no tumours recurred after discontinuation of therapy. These experiments show that drug resistance does not develop in three tumour types treated with a potent angiogenesis inhibitor. An unexpected finding is that repeated cycles of antiangiogenic therapy are followed by prolonged tumour dormancy without further therapy.

Antiangiogenic therapy of experimental cancer does not induce acquired drug resistance

Acquired drug resistance is a major problem in the treatment of cancer. Of the more than 500,000 annual deaths from cancer in the United States, many follow the development of resistance to chemotherapy. The emergence of resistance depends in part on the genetic instability, heterogeneity and high mutational rate of tumour cells. In contrast, endothelial cells are genetically stable, homogeneous and have a low mutational rate. Therefore, antiangiogenic therapy directed against a tumour's endothelial cells should, in principle, induce little or no drug resistance. Endostatin, a potent angiogenesis inhibitor, was administered to mice bearing Lewis lung carcinoma, T241 fibrosarcoma or B16F10 melanoma. Treatment was stopped when tumours had regressed. Tumours were then allowed to re-grow and endostatin therapy was resumed. After 6, 4 or 2 treatment cycles, respectively, no tumours recurred after discontinuation of therapy. These experiments show that drug resistance does not develop in three tumour types treated with a potent angiogenesis inhibitor. An unexpected finding is that repeated cycles of antiangiogenic therapy are followed by prolonged tumour dormancy without further therapy.

Characterization of mouse and human nude genes

The differentiation of primitive epithelial precursor cells in the thymic primordium into subcapsular, cortical, and medullary epithelial cells of the mature thymus requires the activity of the nude gene product Whn. Whn is also required for proper keratinization of the hair shaft. We determined the nucleotide sequence of a 58 kilobase region on mouse chromosome 11 that encompasses the mouse nude gene and part of the two neighboring genes, encoding a sodium/dicarboxylate co-transporter and the retinal protein 4. Using cross-hybridization, the human orthologue of the mouse nude gene was isolated. The human WHN protein also consists of 648 amino acids, 85% of which are identical to the mouse protein. Like the mouse gene, the human gene consists of eight coding exons and utilizes two alternative first exons in a tissue-specific fashion. Sequences upstream of the two alternative first exons display promoter activity in heterologous reporter assays. Whereas both promoters appear to be active in skin (albeit at different levels), only the most upstream element is active in the thymus, indicating that transcriptional activity of the whn gene is subject to complex regulation. Nucleotide sequence database comparisons reveal that among other winged-helix genes, the HTLF and HTLFL1 genes are most closely related to whn, although the exon/intron structure of the human HTLF gene in the DNA binding domain differs from that of whn.

A new subfamily of vertebrate calpains lacking a calmodulin-like domain: implications for calpain regulation and evolution

Calpains are calcium-dependent intracellular nonlysosomal proteases that are believed to participate in signal transduction. In vertebrates, five different calpains have so far been identified, of which three, mu-, m-, and mu/m-calpain, are ubiquitously expressed while the other two, nCL-1 (p94) and nCL-2, exhibit a restricted tissue distribution. We have identified two new vertebrate calpain genes, Capn5 and Capn6. The human and mouse amino acid sequences of these new calpains are the most divergent of the vertebrate calpains identified. They possess most of the residues conserved in calpain family members but the C-terminal region lacks any homology to the calmodulin-like domain of other vertebrate calpains. They both exhibit significant homology over the entire coding region to the protein encoded by the gene tra-3, involved in nematode sex determination, and Capn5 may represent its vertebrate orthologue. The predicted Capn6 protein lacks critical active site residues and may not be proteolytically active. Both genes are differentially expressed in human tissues with highest RNA levels for Capn5 occurring in the testis, liver, trachea, colon, and kidney, while Capn6 is highly expressed only in the placenta sample of the 50 tissues examined. Phylogenetic analysis suggests that the vertebrate calpains arose through a series of gene duplication events that began before the initial divergence of the vertebrate and invertebrate lineages. The discovery of these two new calpains highlights a hitherto unknown complexity of the calpain family with subclasses perhaps possessing different modes of regulation.

The nude gene encodes a sequence-specific DNA binding protein with homologs in organisms that lack an anticipatory immune system

In the mouse, the product of the nude locus, Whn, is required for the keratinization of the hair shaft and the differentiation of epithelial progenitor cells in the thymus. A bacterially expressed peptide representing the presumptive DNA binding domain of the mouse whn gene in vitro specifically binds to a 11-bp consensus sequence containing the invariant tetranucleotide 5'-ACGC. In transient transfection assays, such binding sites stimulated reporter gene expression about 30- to 40-fold, when positioned upstream of a minimal promotor. Whn homologs from humans, bony fish (Danio rerio), cartilaginous fish (Scyliorhinus caniculus), agnathans (Lampetra planeri), and cephalochordates (Branchiostoma lanceolatum) share at least 80% of amino acids in the DNA binding domain. In agreement with this remarkable structural conservation, the DNA binding domains from zebrafish, which possesses a thymus but no hair, and amphioxus, which possesses neither thymus nor hair, recognize the same target sequence as the mouse DNA binding domain in vitro and in vivo. The genomes of vertebrates and cephalochordates contain only a single whn-like gene, suggesting that the primordial whn gene was not subject to gene-duplication events. Although the role of whn in cephalochordates and agnathans is unknown, its requirement in the development of the thymus gland and the differentiation of skin appendages in the mouse suggests that changes in the transcriptional control regions of whn genes accompanied their functional reassignments during evolution.

Identification of interaction partners for the basic-helix-loop-helix protein E47

Helix-loop-helix proteins constitute a family of transcription factors with the potential to form homo- and hetero-dimers mediated by the helix-loop-helix domain. Oncogenic mutations in such genes can disrupt the equilibrium of protein-protein interactions in the affected cell. In order to assess the biological consequences of such mutations, the full complement of interacting proteins must be known. To identify proteins interacting with the basic-helix-loop-helix domain of the ubiquitously expressed E47 protein, a 'sandwich'-screening procedure was developed which distinguishes between homo- and hetero-oligomers, and specifically excludes the detection of complexes which cannot bind DNA. Nine distinct cDNAs were identified which encode proteins with apparent basic-helix-loop-helix domains, including a novel clone termed eip1 which is distantly related in the basic-helix-loop-helix domain to the Drosophila enhancer-of-split m7 protein. Using epitope-tagging, interaction of E47 basic-helix-loop-helix protein with the eip1 protein encoded by this novel cDNA was confirmed by immunoprecipitation experiments in COS7 cells. Interaction was also observed in the yeast two-hybrid system. Three cDNAs encoding proteins without basic-helix-loop-helix domains were also found to interact in the sandwich-expression screen. Interactions with human PARP and mouse replication factor 1a were confirmed using glutathione transferase-tagged cDNAs. A cDNA encoding part of the nucleolin protein sequence interacted with the E47 basic-helix-loop-helix only when fused to a beta-galactosidase tag.

Endostatin: an endogenous inhibitor of angiogenesis and tumor growth

We previously identified the angiogenesis inhibitor angiostatin. Using a similar strategy, we have identified endostatin, an angiogenesis inhibitor produced by hemangioendothelioma. Endostatin is a 20 kDa C-terminal fragment of collagen XVIII. Endostatin specifically inhibits endothelial proliferation and potently inhibits angiogenesis and tumor growth. By a novel method of sustained release, E. coli-derived endostatin was administered as a nonrefolded suspension. Primary tumors were regressed to dormant microscopic lesions. Immunohistochemistry revealed blocked angiogenesis accompanied by high proliferation balanced by apoptosis in tumor cells. There was no toxicity. Together with angiostatin data, these findings validate a strategy for identifying endogenous angiogenesis inhibitors, suggest a theme of fragments of proteins as angiogenesis inhibitors, and demonstrate dormancy therapy.

A yeast artificial chromosome contig spanning the mouse immunoglobulin kappa light chain locus

A single contig spanning the entire mouse immunoglobulin kappa light chain (Igk) locus on chromosome 6 has been established using yeast and bacterial artificial chromosome clones. Detailed mapping of the Igk locus indicates that a member of the Igk-V2 gene family, located about 3.5 megabases upstream of the Igk-J-C complex, is the most distal functional Igk-V gene. Sequence analyses of Igk-V genes and anonymous DNA segments provide indications for internal duplications at the 5' end of the Igk-V locus and identify the likely origin of Igk-V orphon gene clusters located elsewhere in the mouse genome.

Expression of peptides encoded by exons in cloned mammalian DNA

New synthetic approaches, such as combinatorial chemistry, provide a rich source of potential drug candidates. At the same time, the human genome initiative and other large-scale sequencing projects provide a large number of novel drug targets. However, the functional analysis of thousands of new genes remains a major challenge for the future. A systematic strategy for genome-wide functional analysis of genes could employ the fact that at least some modules in multi-domain proteins are encoded in individual exons. Exon amplification provides information about coding regions of most genes that is independent of their transcriptional status; exon amplification from entire mammalian genomes has been demonstrated. Here, we describe the development of an exon-trap system, lambdaGEE (for genomic exon expression), that couples exon amplification with the expression of exon-encoded peptides.

The whn transcription factor encoded by the nude locus contains an evolutionarily conserved and functionally indispensable activation domain

Mutations in the whn gene are associated with the phenotype of congenital athymia and hairlessness in mouse and rat. The whn gene encodes a presumptive transcription factor with a DNA binding domain of the forkhead/ winged-helix class. Two previously described null alleles encode truncated whn proteins lacking the characteristic DNA binding domain. In the rat rnu allele described here, a nonsense mutation in exon 8 of the whn gene was identified. The truncated whnrnu protein contains the DNA binding domain but lacks the 175 C-terminal amino acids of the wild-type protein. To facilitate the identification of functionally important regions in this region, a whn homolog from the pufferfish Fugu rubripes was isolated. Comparison of derived protein sequences with the mouse whn gene revealed the presence of a conserved acidic protein domain in the C terminus, in addition to the highly conserved DNA binding domain. Using fusions with a heterologous DNA binding domain, a strong transcriptional activation domain was localized to the C-terminal cluster of acidic amino acids. As the whnrnu mutant protein lacks this domain, our results indicate that a transactivation function is essential for the activity of the whn transcription factor.

Two genetically separable steps in the differentiation of thymic epithelium

The development of the thymus depends initially on epithelial-mesenchymal and subsequently on reciprocal lympho-stromal interactions. The genetic steps governing development and differentiation of the thymic microenvironment are unknown. With the use of a targeted disruption of the whn gene, which recapitulates the phenotype of the athymic nude mouse, the WHN transcription factor was shown to be the product of the nude locus. Formation of the thymic epithelial primordium before the entry of lymphocyte progenitors did not require the activity of WHN. However, subsequent differentiation of primitive precursor cells into subcapsular, cortical, and medullary epithelial cells of the postnatal thymus did depend on activity of the whn gene. These results define the first genetically separable steps during thymic epithelial differentiation.

Genomic Organization of the ATM gene

The ATM gene was recently identified and found to be responsible for the genetic disorder ataxiatelgiectasia. The major ATM transcript is 13 kb. Using long-distance PCR, we determined the genomic structure of this gene and identified all of its exon-intron boundaries. The ATM gene spans approximately 150 kb of genomic DNA and consists of 66 exons. The initiation codon falls within exon 4. The last exon is 3.8 kb and contains the stop codon and a 3'-untranslated region of about 3600 nucleotides.

Dynamic changes in gene expression during in vitro differentiation of mouse embryonic stem cells

The expression pattern of protein tyrosine kinases (PTK) and phosphatases (PTP) was determined during the first eight days of in vitro differentiation of mouse embryonic stem (ES) cells. DNA fingerprinting of catalytic domains amplified from cDNA revealed dynamic changes in expression of previously described genes. A novel PTP is expressed in undifferentiated ES cells, and is down-regulated during in vitro differentiation.

An intragenic deletion in the human PTPN6 gene affects transcriptional activity

An intragenic deletion in the human PTPN6 gene is described. The PTPN6 gene maps to chromosome 12p12-13 and is shown to possess two alternative first exons. A 1.7-kb deletion occurring in the intron between the two alternatively used first exons is the result of an illegitimate recombination between two Alu-type repeats. The deletion increases the transcriptional activity of the distal promotor.

YAC/P1 contigs defining the location of 56 microsatellite markers and several genes across a 3.4-cM interval on mouse chromosome 11

The characterization of three YAC/P1 contigs from adjacent segments of the central region of mouse Chromosome (Chr) 11 is described. These contigs are based upon 63 YACs and 40 P1 recombinants. From these clones, 185 end sequences were obtained, of which 147 sequences could be converted into sequence-tagged sites and mapped within the three contigs. Deletions were detected in 16 out of 63 YACs; 19 of 63 YACs were found to be chimeric. No such aberrations were found in P1 recombinants. A total of 22 public and 34 newly developed microsatellite markers were unambiguously localized to and ordered in the contigs. In the cryb1/Nf1 interval of the central contig, several new genes have been identified by exon trapping and precisely localized with respect to known STS markers.

New member of the winged-helix protein family disrupted in mouse and rat nude mutations

Mutations at the nude locus of mice and rats disrupt normal hair growth and thymus development, causing nude mice and rats to be immune-deficient. The mouse nude locus has been localized on chromosome 11 (refs 3, 4) within a region of < 1 megabase. Here we show that one of the genes from this critical region, designated whn, encodes a new member of the winged-helix domain family of transcription factors, and that it is disrupted on mouse nu and rat rnuN alleles. Mutant transcripts do not encode the characteristic DNA-binding domain, strongly suggesting that the whn gene is the nude gene. Mutations in winged-helix domain genes cause homeotic transformations in Drosophila and distort cell-fate decisions during vulval development in Caenorhabditis elegans. The whn gene is thus the first member of this class of genes to be implicated in a specific developmental defect in vertebrates.

The sequence complexity of exons trapped from the mouse genome

BACKGROUND

A central issue in genome analysis is the identification and characterization of coding regions. Estimating the coding complexity of vertebrate genomes by measuring the kinetic complexity of mRNA populations and by sequence analysis of cDNAs is limited by the fact that any given source of mRNA represents a very biased sample of all genes. Exon trapping is a method that enables the identification of genes irrespective of their transcriptional status.

RESULTS

Exons were trapped from the entire mouse genome, and the resulting fragments cloned. About 7% of a random sample of exons taken from this library have significant structural homology or sequence similarity to previously sequenced genes. Using cDNAs derived from several stages of mouse development, evidence for expression of about 62% of this sample of exons was found. These data suggest that the great majority of 'exons' in the library are derived from genes. We estimate that the fraction of the genome contained in trapped exons is 2.4%; this corresponds to a sequence complexity of about 72 megabases.

CONCLUSIONS

The library of exons trapped from the entire mouse genome probably represents one of the least biased and most comprehensive libraries of mouse coding regions, and should therefore prove very useful for finding genes during genome mapping and sequencing.

Two large insert vectors, lambda PS and lambda KO, facilitate rapid mapping and targeted disruption of mammalian genes

The construction and the testing of two lambda phage vectors are described that greatly simplify the tasks of mapping genomic DNA and making replacement-type gene-targeting vectors for mammalian cells from a library of isogenic genomic DNA. The first vector, lambda PS, accommodates up to 20 kb and allows inserts to be automatically subcloned in plasmid form because of the presence of loxP sites flanking the insert. The second vector, lambda KO, accommodates up to 16.7 kb and allows inserts to be automatically subcloned as plasmids containing HSVtk genes that are positioned flanking the inserted genomic DNA. We have prepared highly redundant libraries from genomic DNA of 129/Sv-strain mice for the construction of targeting vectors. In our scheme, the locus of interest is characterized using a library made in lambda PS. For instance, suitable flanking probes can be derived to determine targeting events. The final targeting construct with flanking HSVtk genes is obtained using the lambda KO cloning vector. The entire procedure is exemplified by successful targeting of the X-linked mouse hprt locus.

Exon amplification from complete libraries of genomic DNA using a novel phage vector with automatic plasmid excision facility: application to the mouse neurofibromatosis-1 locus

The identification of transcription units in the vicinity of chromosomal lesions found in tumours is an essential step in the identification of new oncogenes. Here, we describe a lambda phage vector system for genomic exon-trapping (lambda GET), which dramatically simplifies the task of exon amplification from genomic DNA. The vector accommodates about 6.5 to 19 kb of DNA and allows inserts to be automatically subcloned as multi-copy plasmids containing splice donor and acceptor sites positioned flanking the inserted genomic DNA. RNA transcripts derived from such plasmids are processed in vivo and exons contained within the inserted genomic fragments become flanked by known sequences in the resulting mRNAs. RNA-based PCR can then be used for subsequent cloning and sequence analysis of trapped exons. We have exploited the large cloning capacity of lambda GET to construct highly redundant complete genomic libraries from Sau3AI partially digested vertebrae DNAs. Using this system, we have analysed a region of about 1 MB around the mouse neurofibromatosis-1 locus and have identified novel transcription units flanking the Nf-1 gene.

A yeast artificial chromosome contig on mouse chromosome 11 encompassing the nu locus

Mutations at the nude locus disrupt the homing process of T cell progenitor cells to the thymic rudiment, a key aspect of T cell differentiation. Here, we map the nude locus to a set of overlapping yeast artificial chromosomes (YAC) clones covering a genetic interval of about 0.5 centi Morgan on mouse chromosome 11. These results provide a suitable starting point to molecularly clone the nude gene.

A versatile expression vector for the in vitro study of protein-protein interactions: characterization of E47 mutant proteins

Several mutants of the E47 protein, a member of the family of basic/helix-loop-helix (b-HLH) transcriptional regulators, were examined for their ability to homo- and heterodimerize with the protein product of the T-cell oncogene tal-1/SCL. For this purpose, a novel bacterial expression system was developed in which proteins are expressed as fusions appended to glutathione-S-transferase via a thrombin cleavage site and either one or four protein kinase recognition sites embedded in a glycine-rich domain. Since the interaction domain can be purified away from the glutathione-S-transferase moiety and the radioactive label is located in a flexible N-terminal tag, protein folding should occur normally. Our studies with E47 proteins prepared in this system indicate that the ratio between E47 homodimers and E47/tal-1 heterodimers can dramatically shift upon subtle mutations in the loop region and the second helix of the E47 protein. This unexpected results suggests a novel mechanism to alter the equilibrium between different transactivating protein complexes of the b-HLH class.