Identification of molecular determinants required for interaction of ubiquitin-conjugating enzymes and RING finger proteins.

Recent results from several laboratories suggest that the interaction of E2 ubiquitin-conjugating enzymes with the RING finger domain has a central role in mediating the transfer of ubiquitin to proteins. Here we present a mutational analysis of the interaction between the E2 enzyme UbcM4/UbcH7 and three different RING finger proteins, termed UIPs, which, like Parkin, contain a RING1-IBR-RING2 motif. The results show that the E2 enzyme binds to the RING1 domain but not to the other cysteine/histidine-rich domains of the RING1-IBR-RING2 motif. Three regions within the UbcM4 molecule are involved in this interaction: the H1 alpha helix, loop L1, connecting the third and fourth strand of the beta sheet, and loop L2, located between the fourth beta strand and the second alpha helix. Loop L2 plays an important role in determining the specificity of interaction. The effects of L2 mutations on UbcM4/UIP interaction are different for each UIP, indicating that RING finger domains can vary considerably in their structural requirements for binding to E2 enzymes. The result that single amino-acid changes can regulate binding of E2 enzymes to different RING finger proteins suggests a novel approach to experimentally manipulate proteolytic pathways mediated by RING finger proteins.

Interaction of the ring finger-related U-box motif of a nuclear dot protein with ubiquitin-conjugating enzymes.

The U-box domain has been suggested to be a modified RING finger motif where the metal-coordinating cysteines and histidines have been replaced with other amino acids. Known U-box-containing proteins have been implicated in the ubiquitin/proteasome system. In a search for proteins interacting with the ubiquitin-conjugating enzyme UbcM4/UbcH7, we have identified a novel U-box containing protein, termed UIP5, that is exclusively found in the nucleus as part of a nuclear dot-like structure. Interaction between UbcM4 and UIP5 was observed in vivo and in vitro with bacterially expressed proteins. In addition to UbcM4, several other ubiquitin-conjugating enzymes (E2s) that share the same sequence within the L1 loop bind to UIP5. Mutational analysis showed that the U-box, like the RING finger in other proteins, forms the physical basis for the interaction with E2 enzymes. Further support for the structural similarity between U-box and RING finger comes from the observation that, in both cases, the same regions within the UbcM4 molecule are required for interaction. Our results establish at the molecular level a link between the U-box and the ubiquitin conjugating system and strongly suggest that proteins containing U-box domains are functionally closely related to RING finger proteins.

Disruption of the gene encoding the ubiquitin-conjugating enzyme UbcM4 has no effect on proliferation and in vitro differentiation of mouse embryonic stem cells.

The ubiquitin-conjugating enzyme UbcM4, which is identical to the human enzyme UbcH7, was previously shown to be essential for normal mouse development. In order to study the possible role of UbcM4 for cell proliferation and in vitro differentiation, we here describe the establishment and characterization of fibroblast and embryonic stem cell lines with partial or complete inactivation of the UbcM4 gene. ES cell lines in which both alleles of the gene were inactivated by targeted mutagenesis showed no differences in growth rates, cell cycle progression and in vitro differentiation when compared to wild-type ES cells. Fibroblast cell lines with a partially inactivated UbcM4 gene were derived from embryos of the previously described A6 mouse mutant, where retrovirus integration has resulted in a recessive lethal mutation. As in the mutant embryos, steady levels of RNA and protein in the cell lines were reduced by about 70%. The mutant cell lines showed no differences in immortalization kinetics, growth rates and cell cycle progression when compared to wild-type fibroblasts. Taken together, our results strongly suggest that UbcM4-mediated ubiquitination and degradation are not necessary for proteins involved in the maintenance and growth of cells.

A family of structurally related RING finger proteins interacts specifically with the ubiquitin-conjugating enzyme UbcM4.

The ubiquitin-conjugating enzyme UbcM4 was previously shown to be necessary for normal mouse development. As a first step in identifying target proteins or proteins involved in the specificity of UbcM4-mediated ubiquitylation, we have isolated seven cDNAs encoding proteins that specifically interact with UbcM4 but with none of the other Ubcs tested. This interaction was observed in yeast as well as in mammalian cells. With one exception, all UbcM4-interacting proteins (UIPs) belong to a family of proteins that contain a RING finger motif. As they are structurally related to RING finger proteins that have recently been shown to play an essential role in protein ubiquitylation and degradation, the possibility is discussed that UIPs are involved in the specific recognition of substrate proteins of UbcM4.

Developmental expression of the CaM kinase II isoforms: ubiquitous gamma- and delta-CaM kinase II are the early isoforms and most abundant in the developing nervous system.

CaM kinase II constitutes a family of multifunctional protein kinases that play a major role in Ca2+-mediated signal transduction. As a first step in understanding their possible function in mouse development we characterized the expression patterns of all CaM kinase II isoforms (alpha, beta, gamma and delta) starting in prenatal development. Remarkably, only the ubiquitous gamma- and delta-CaM kinase II are expressed during early development. Their distribution suggests a special role in the developing nervous system and in mature excitable tissues. Additionally, we describe the murine betaM-CaM kinase II, a variant of the 'brain-specific' beta-CaM kinase II, which is highly expressed in skeletal muscle.

alphaKAP is an anchoring protein for a novel CaM kinase II isoform in skeletal muscle.

Ca2+/calmodulin-dependent protein kinase II (CaM kinase II) is present in a membrane-bound form that phosphorylates synapsin I on neuronal synaptic vesicles and the ryanodine receptor at skeletal muscle sarcoplasmic reticulum (SR), but it is unclear how this soluble enzyme is targeted to membranes. We demonstrate that alphaKAP, a non-kinase protein encoded by a gene within the gene of alpha-CaM kinase II, can target the CaM kinase II holoenzyme to the SR membrane. Our results indicate that alphaKAP (i) is anchored to the membrane via its N-terminal hydrophobic domain, (ii) can co-assemble with catalytically competent CaM kinase II isoforms and target them to the membrane regardless of their state of activation, and (iii) is co-localized and associated with rat skeletal muscle CaM kinase II in vivo. alphaKAP is therefore the first demonstrated anchoring protein for CaM kinase II. CaM kinase II assembled with alphaKAP retains normal enzymatic activity and the ability to become Ca2+-independent following autophosphorylation. A new variant of beta-CaM kinase II, termed betaM-CaM kinase II, is one of the predominant CaM kinase II isoforms associated with alphaKAP in skeletal muscle SR.

Structure of the gene encoding the ubiquitin-conjugating enzyme Ubcm4, characterization of its promoter, and chromosomal location.

Ubiquitin-conjugating enzymes (E2 or Ubc) play a key role in the post-translational modification of proteins by ubiquitylation. They are encoded by a large family of genes that are closely related to each other. In this paper we present the first complete structural analysis, including the promoter and the chromosomal location, of a member of this family, the mouse Ubcm4 gene. At the genomic level the Ubcm4 gene spans approx. 50kb and is composed of four exons. Only about 1% of the total gene codes for amino acids. The four different Ubcm4 specific RNAs encode the same protein and differ only in the length of the 3' untranslated region. The polyadenylation signals used by the four different RNAs are all within the 3' terminal exon. At the 5' end of the gene, multiple transcriptional start sites were mapped within a region of 25bp. The region proximal to the initiation sites does not contain a TATA box and is not GC-rich. Transient chloramphenicol acetyltransferase assays, however, showed that this region can promote the expression of a reporter gene and that 15bp upstream of the first initiation site were sufficient for basal expression. The Ubcm4 gene was mapped by interspecific backcross analysis to the proximal region of mouse chromosome 16.

Molecular cloning and chromatin structure analysis of the murine alpha1(I) collagen gene domain.

We have isolated molecular clones of genomic mouse DNA spanning 55 kb, including the entire coding region of the murine alpha1(I) collagen (Col1a1) gene and 24 kb of 5' and 13 kb of 3'-flanking sequences, and have performed a detailed chromatin structure analysis of these sequences. Several new DNase-I-hypersensitive sites were identified. The distal 5'-flanking region contains two clusters of DNase-I-hypersensitive sites located between 7 and 8 kb and between 15 and 20 kb upstream of the start site of transcription, respectively. Several of these sites were shown to be present in collagen-producing, but not in non-producing cells, indicating that they are associated with transcription of the gene and may function in its regulation. One strong constitutive DNase-I-hypersensitive site at -18.5 kb was also cleaved by endogenous nucleases. The 3'-flanking region of the gene contains a DNase-I-hypersensitive site located 6 kb downstream of the end of the gene, as well as sequences that can induce a non-B DNA structure. Because these latter sequences coincide with DNase-I-hypersensitive sites in the homologous human gene, our results suggest that some regulatory elements may play a role in gene regulation, not by specific protein-DNA interactions but by virtue of their ability to induce a non-B DNA structure and/or an alternate chromatin conformation. A comparison of the murine and human Col1a1 domains shows a similar, although not identical, distribution of DNase-I-hypersensitive sites, indicating a conserved arrangement of regulatory elements. Our results strongly suggest that these new sites constitute regulatory elements which are involved in the transcriptional regulation and/or chromatin loop organization of the Col1a1 gene, and they are now amenable for functional analyses.

Provirus integration into a gene encoding a ubiquitin-conjugating enzyme results in a placental defect and embryonic lethality.

Ubiquitin-conjugating enzymes (E2 or Ubc) constitute a family of conserved proteins that play a key role in ubiquitin-dependent degradation of proteins in eukaryotes. We describe here a transgenic mouse strain where retrovirus integration into an Ubc gene, designated UbcM4, results in a recessive-lethal mutation. UbcM4 is the mouse homologue of the previously described human UbcH7 that is involved in the in vitro ubiquitination of several proteins including the tumor suppressor protein p53. The provirus is located in the first intron of the gene. When both alleles are mutated the level of steady-state mRNA is reduced by about 70%. About a third of homozygous mutant embryos die around day 11.5 of gestation. Embryos that survive that stage are growth retarded and die perinatally. The lethal phenotype is most likely caused by impairment of placenta development as this is the only organ that consistently showed pathological defects. The placental labyrinth is drastically reduced in size and vascularization is disturbed. The UbcM4 mouse mutant represents the first example in mammals of a mutation in a gene involved in ubiquitin conjugation. Its recessive-lethal phenotype demonstrates that the ubiquitin system plays an essential role during mouse development.

An alternative, nonkinase product of the brain-specifically expressed Ca2+/calmodulin-dependent kinase II alpha isoform gene in skeletal muscle.

The gene for the alpha isoform of Ca2+/calmodulin-dependent kinase II (alpha CaMKII) codes for a multifunctional protein kinase that is found exclusively in the brain. Here we show that in skeletal muscle, an alternative nonkinase product, hereafter referred to as alpha KAP (alpha CaMKII association protein), is expressed from the same gene. alpha KAP consists of a C-terminal region that is identical to the association domain of alpha CaMKII, with the exception of 11 amino acids inserted in the variable region. The N-terminal sequence of alpha KAP is highly hydrophobic and not present in any known CaMKII protein. The catalytic and regulatory domains of alpha CaMKII are missing in alpha KAP. Analysis of the exon-intron structure revealed that the alpha KAP transcript is derived from the alpha CaMKII gene by alternative promoter usage and RNA splicing. The transcriptional start site of alpha KAP mRNA is located within an intron of the alpha CaMKII gene. Therefore, the relationship between alpha KAP and alpha CaMKII is that of a gene within a gene. Immunostaining using anti-alpha KAP antibodies suggests that alpha KAP is associated with sarcomeres of skeletal muscle fibers. On the basis of its primary structure and specific location, the possible function of alpha KAP as an anchoring protein for CaMKII is discussed.

Tissue- and stage-specific activation of an endogenous provirus after transcription through its integration site in the opposite orientation.

Endogenous proviruses of the Moloney murine leukemia retrovirus (Mo-MuLV) are transcriptionally blocked in early embryos and in general remain silent even when the tissues have become permissive to the expression of newly integrated copies. Eventually, activation in presumably very few cells initiates rapid superinfection leading to viremia and leukemia, but the processes leading to provirus activation are unknown. Differences in the onset and development of viremia between several mouse strains carrying an endogenous Mo-MuLV (Mov lines) are attributed to a chromosomal position effect, but neither cell type nor stage of provirus activation is known for any strain. We have now monitored the appearance of viral transcripts and particles in the Mov13 strain, which carries the Mo-MuLV provirus in inverted orientation in the first intron of a collagen gene (Col1a1) with well-characterized transcriptional activity. We report obligatory tissue- and stage-specific virus activation in osteoblasts and odontoblasts. The significance of this activation pattern is indicated by the fact that of the great variety of cells expressing the wild-type collagen gene, only these two cell types can also transcribe the mutant allele including its viral insert. We propose that this transcription of the proviral genome, albeit in the opposite direction, leads to the activation of the viral promoter.

An optimized electroporation protocol applicable to a wide range of cell lines.

A number of transfection methods for mammalian cells are available; however, many cell lines may appear resistant to efficient transfection, or at best, necessitate lengthy optimization procedures in recommended protocols. We describe here an electroporation protocol that yields highly efficient gene transfer (20%-100% of surviving cells) in all 19 cell lines tested so far, including lymphoid, myeloid, glial, fibroblast and COS cells. Unlike earlier electroporation protocols, adaptation of this protocol to a cell line of interest only requires the optimization of a single parameter, i.e., the voltage, and can be performed within a day. Furthermore, it is also superior to transfection protocols for other methods (calcium phosphate precipitation, lipofection, DEAE-dextran transfection and transferrinfection) in terms of reproducibility, economy and average transfection efficiency for a wide variety of cell lines.

Interaction of several related GC-box- and GT-box-binding proteins with the intronic enhancer is required for differential expression of the gb110 gene in embryonal carcinoma cells.

The molecular mechanisms by which expression of a gene is down-regulated after differentiation of F9 embryonal carcinoma cells into parietal endoderm-like cells was studied by characterizing the cis- and trans-regulatory elements of the gb110 gene. This gene encodes a putative RNA helicase, and its expression is down-regulated when F9 cells are differentiated with retinoic acid and cyclic AMP. The 5'-flanking region of the gene has all of the features of a GC-rich island promoter and seems to play only a minor role, if any, in the regulated expression. A 133-bp enhancer in the first intron was identified by transient chloramphenicol acetyltransferase assays that activated expression in undifferentiated F9 cells about 50- to 100-fold. As this enhancer was not active in differentiated F9 cells, it seems to be the prime mediator of the differentiation-specific down-regulation of the gb110 gene. Four different protein-binding sites, three of which contain GC- and GT-box motifs, were identified in the enhancer element. The fourth site, interacting with previously described transcription factor FTZ-F1/ELP, seems to be of minor importance for the activity of the enhancer. Mutational analysis showed that the cooperative interaction of several most likely related proteins with the three GC- and GT-box motifs was required for full enhancer activity. On the basis of their binding properties, at least two of these proteins seem to be identical or closely related to ubiquitous transcription factor Sp1. One of the GT-box-binding proteins was present in undifferentiated F9 cells but not, however, in its differentiated derivatives. The cell specificity of this transcription factor explains why the gb110 gene is not expressed or expressed only at low levels in parietal endoderm-like cells.

Restricted expression of Mov13 mutant alpha 1(I) collagen gene in osteoblasts and its consequences for bone development.

Cell type-specific differences in the transcriptional control of the mouse gene coding for the alpha 1 chain of collagen type I (Col1a1) have been revealed previously with the help of the Mov13 mouse strain which carries a retroviral insert in the first intron of the gene. Transcription of this mutant Col1a1 allele is completely blocked in all mesodermal cell types tested so far, with the exception of the odontoblast where it is expressed at an apparently normal rate (Kratochwil et al. [1989] Cell 57:807-816). To define the tissue specificity of the mutant allele more precisely, we have now studied its expression in osteoblasts, another skeletogenic cell type which, like odontoblasts, produces high amounts of collagen I. Evidence for transcription of the Mov13 allele was obtained by in situ hybridization in homozygous (M/M) and heterozygous (M/+) bone tissue, in grafts as well as in vivo. The presence of mouse collagen I and the development of bone tissue were demonstrated in M/M skeletal elements grown on the chick chorioallantoic membrane (CAM). Further support for expression of the mutant gene was obtained from two 16 day M/M fetuses in vivo. Bone tissue of diverse embryological origin (vertebrae and ribs of somitic origin, long bones derived from lateral plate, calvariae from head paraxial mesoderm, and mandibulae from head neural crest) expresses the mutant allele. However, in situ hybridization experiments indicate that only a subpopulation of osteoblasts is capable of transcribing it at a high rate, resulting in severe impairment of bone development in grafts and in vivo. Therefore, osteoblasts, in comparison to odontoblasts and fibroblast-like cells, assume an intermediate position with respect to transcription of the Mov13 allele. We suggest that this diversity in the utilization of the mutant collagen gene reflects cell type-specific differences in the transcriptional regulation of the wild type (wt) Col1a1 gene.

Consecutive inactivation of both alleles of the gb110 gene has no effect on the proliferation and differentiation of mouse embryonic stem cells.

The gene gb110 was originally identified by provirus integration in the Mov10 mouse strain and encodes a 110-kDa protein with potential GTP-binding activity. The gene is evolutionarily conserved, and its expression is controlled in a developmentally and cell-cycle-specific manner, suggesting that it has an important function in differentiation and development. As a first step in studying the functional role of gb110, embryonal stem (ES) cell lines were derived in which both gb110 alleles were inactivated by consecutive gene targeting via homologous recombination. The first allele was interrupted by integration of a neomycin resistance-encoding gene (neo) and, subsequently, the second allele by integration of a hygromycin B resistance-encoding gene (hyg). Selection for homologous recombination was achieved by using promoter and AUG codon-deficient hyg or neo whose expression was dependent on integration into the host genome next to the transcriptional and translational start signals. The efficiency of gb110 gene targeting was very high, with 85-100% of all drug-resistant colonies having undergone homologous recombination. ES cells lacking a functional gb110 were indistinguishable from the wild-type ES cells, indicating that this gene is not required for normal ES cell proliferation and differentiation in vitro.

Structure, expression, and chromosome location of the gene for the beta subunit of brain-specific Ca2+/calmodulin-dependent protein kinase II identified by transgene integration in an embryonic lethal mouse mutant.

The transgenic mouse strain CAT40 carries in its germ line one copy of a DNA construct consisting of the chloramphenicol acetyltransferase gene and the immunoglobulin heavy-chain enhancer. We show that transgene integration has resulted in a recessive lethal mutation that leads to death of homozygous CAT40 embryos shortly after implantation. The transgene has integrated adjacent to the 3' end of the gene coding for the beta subunit of the brain-specific Ca2+/calmodulin-dependent protein kinase II (Camk-2). The complete cDNA sequence of the Camk-2 gene and most of its exon/intron structure was determined. The deduced amino acid sequence is highly homologous to the previously described rat protein. The chromosomal location of the Camk-2 locus was mapped by interspecific backcross analysis to the proximal region of mouse chromosome 11. This region lacks previously identified recessive embryonic lethal mutations. During embryonic development, Camk-2-specific transcripts are first seen in the head section of 12.5-day-old embryos, and in adult mice the gene is expressed almost exclusively in the brain. Although transcription of the Camk-2 gene in heterozygous CAT40 mice is affected by transgene integration, it is unlikely that this gene is responsible for the mutant phenotype, since it is not expressed in blastocysts and the first transcripts during normal development are detected after the death of homozygous CAT40 embryos. Transgene integration is accompanied by a large deletion of cellular DNA; death is therefore most likely caused by the loss of a gene or genes that are important for early postimplantation development.

Structure and expression of a gene encoding a putative GTP-binding protein identified by provirus integration in a transgenic mouse strain.

The Mov-10 mouse strain was derived by infection of preimplantation embryos with the Moloney murine leukemia virus and carries one copy of the provirus in its germ line. Here we show that the provirus has integrated into an evolutionarily conserved gene that can code for a protein of 110 kDa containing the three consensus elements characteristic for GTP-binding proteins. The Mov-10 locus was expressed in a variety of cell types, including embryonal carcinoma and embryonic stem cells. Transcription of the gene was down-regulated about 10-fold when F9 embryonal carcinoma cells are differentiated into parietal endodermlike cells and about 2-fold when they are differentiated into visceral endodermlike cells. High levels of Mov-10 transcripts were also found at different stages of embryonal development and in the testes and thymus of adult animals. Expression was cell cycle controlled, with steady-state RNA levels significantly higher in growth-arrested than in growth-stimulated cells. The results suggest that the Mov-10 locus has an important function in development and/or control of cell proliferation. The provirus was shown to have integrated into intron 1 of the gene without disrupting expression, indicating that integration into intronic sequences of a transcription unit does not necessarily affect transcription. This result together with previous results from the Mov-13 mouse strain suggested that proviruses exert their mutagenic effect only by integration in specific sites, such as cis-regulatory DNA elements.

Retroviral integration sites in transgenic Mov mice frequently map in the vicinity of transcribed DNA regions.

Transcription of cellular sequences flanking proviral insertion sites was studied in several Mov mouse strains, each of which carried one copy of the Moloney murine leukemia virus in its germ line. In three out of five randomly chosen Mov strains, the provirus had integrated in the vicinity of DNA regions transcribed in the embryonal stem cell line CCE and the embryonal carcinoma cell line F9. Assuming that CCE and F9 cells are developmentally equivalent to the early embryonic cells that were infected to establish the Mov strains, our results suggest that retroviruses integrate preferentially into actively transcribed DNA regions.

Transcription of a mutant collagen I gene is a cell type and stage-specific marker for odontoblast and osteoblast differentiation.

The Mov13 allele of the mouse alpha 1(I) collagen gene carries a retroviral insert in its first intron and had been reported to be transcriptionally silent. We have recently shown, however, that this mutant gene is expressed in odontoblasts of transplanted teeth derived from homozygous and heterozygous carrier embryos. The expression of the Mov13 allele has now been followed throughout in vivo development of mandibular teeth and bone in heterozygous animals, by in situ hybridization with a probe that specifically recognizes transcripts of the mutant gene. We show that the onset of its transcription precisely coincides with the final differentiation of odontoblasts and the onset of dentinogenesis, i.e. on day E16 for the incisor and at birth for the first molar. The mutant allele is also transcribed in osteoblasts of mandibular bone, again starting precisely with the onset of osteogenesis (day E13/14). No other cells were seen to transcribe the mutant gene. By these criteria, transcription of the Mov13 allele constitutes a true differentiation marker for odontoblasts and osteoblasts. Expression of the mutant allele in these two specialized cell types, in contrast to its transcriptional block in all other mesodermal cells ('fibroblasts'), suggests tissue-specific differences in the regulation of the alpha 1(I) collagen gene.

Structure and chromosomal mapping of a highly polymorphic repetitive DNA sequence from the pseudoautosomal region of the mouse sex chromosomes.

The pseudoautosomal region of the Mov15 mouse strain is marked by a Moloney murine leukemia provirus. The sequences flanking the Mov15 provirus were molecularly cloned and shown to consist of a tandemly repeated sequence of 31 nucleotides. Copy number variation of this repeat most likely accounts for the polymorphism in the mouse pseudoautosomal region detected with a probe from the flanking sequences. In situ hybridization to metaphase chromosomes showed heavy labeling of the pairing region of the X and Y chromosomes. The repetitive sequence was also found at the subtelomeric region of three autosomes. A similar level of amplification as the one seen on the sex chromosomes seems to be present on chromosomes 9 and 13. Lower copy number appear to be present on chromosome 4.