Caspase-dependent deubiquitination of monoubiquitinated nucleosomal histone H2A induced by diverse apoptogenic stimuli.

Enzymatic deubiquitination of mono-ubiquitinated nucleosomal histone H2A (uH2A) and H2B (uH2B) is closely associated with mitotic chromatin condensation, although the function of this histone modification in cell division remains ambiguous. Here we show that rapid and extensive deubiquitination of nucleosomal uH2A occurs in Jurkat cells undergoing apoptosis initiated by anti-Fas activating antibody, staurosporine, etoposide, doxorubicin and the proteasome inhibitor, N-acetyl-leucyl-leucyl-norlucinal. These diverse apoptosis inducers also promoted the accumulation of slowly migrating, high molecular weight ubiquitinated proteins and depleted the cellular pool of unconjugated ubiquitin. In apoptotic cells, ubiquitin was cleaved from uH2A subsequent to the appearance of plasma membrane blebbing, and deubiquitination of uH2A closely coincided with the onset of nuclear pyknosis and chromatin condensation. Nucleosomal uH2A deubiquitination, poly (ADP-ribose)polymerase (PARP) cleavage and chromatin condensation were prevented in cells challenged with apoptosis inducers by pretreatment with the pan-caspase inhibitor, zVAD-fmk, or by over-expressing anti-apoptotic Bcl-xL protein. These results implicate a connection between caspase cascade activation and nucleosomal uH2A deubiquitination. Transient transfection of 293 cells with the gene encoding Ubp-M, a human deubiquitinating enzyme, promoted uH2A deubiquitination, while an inactive mutated Ubp-M enzyme did not. However, Ubp-M-promoted deubiquitination of uH2A was insufficient to initiate apoptosis in these cells. We conclude that uH2A deubiquitination is a down-stream consequence of procaspase activation and that unscheduled cleavage of ubiquitin from uH2A is a consistent feature of the execution phase of apoptosis rather than a determining or initiating apoptogenic event. Nucleosomal uH2A deubiquitination may function as a cellular sensor of stress in situations like apoptosis through which cells attempt to preserve genomic integrity.

A nonerythroid isoform of protein 4.1R interacts with the nuclear mitotic apparatus (NuMA) protein.

Red blood cell protein 4.1 (4.1R) is an 80- kD erythrocyte phosphoprotein that stabilizes the spectrin/actin cytoskeleton. In nonerythroid cells, multiple 4.1R isoforms arise from a single gene by alternative splicing and predominantly code for a 135-kD isoform. This isoform contains a 209 amino acid extension at its NH2 terminus (head piece; HP). Immunoreactive epitopes specific for HP have been detected within the cell nucleus, nuclear matrix, centrosomes, and parts of the mitotic apparatus in dividing cells. Using a yeast two-hybrid system, in vitro binding assays, coimmunolocalization, and coimmunoprecipitation studies, we show that a 135-kD 4.1R isoform specifically interacts with the nuclear mitotic apparatus (NuMA) protein. NuMA and 4.1R partially colocalize in the interphase nucleus of MDCK cells and redistribute to the spindle poles early in mitosis. Protein 4.1R associates with NuMA in the interphase nucleus and forms a complex with spindle pole organizing proteins, NuMA, dynein, and dynactin during cell division. Overexpression of a 135-kD isoform of 4.1R alters the normal distribution of NuMA in the interphase nucleus. The minimal sequence sufficient for this interaction has been mapped to the amino acids encoded by exons 20 and 21 of 4.1R and residues 1788-1810 of NuMA. Our results not only suggest that 4.1R could, possibly, play an important role in organizing the nuclear architecture, mitotic spindle, and spindle poles, but also could define a novel role for its 22-24-kD domain.

A mutant deubiquitinating enzyme (Ubp-M) associates with mitotic chromosomes and blocks cell division.

A new ubiquitin-processing protease (Ubp-M) has been identified in mammalian cells that is phosphorylated at the onset of mitosis and dephosphorylated during the metaphase/anaphase transition. The carboxyl-terminal domain of this 823-aa protein can be phosphorylated in vitro with either extracts of mitotic cells or purified cdc-2/cyclin B complexes. Recombinant Ubp-M is able to deubiquitinate histone H2A in vitro, and the phosphorylated form is also enzymatically active. Wild-type Ubp-M, transiently expressed as green fluorescent protein-fusion proteins, localizes in the cytoplasm of cultured cells, but mutant forms, lacking an active-site cysteine, associate closely with mitotic chromosomes during all stages of cell division and remain within the nucleus during the postmitotic period. Cells transfected with plasmids containing mutant Ubp-M genes stop dividing and eventually undergo apoptosis. Ubp-M may deubiquitinate one or more critical proteins that are involved in the condensation of mitotic chromosomes, possibly acting selectively on histones H2A and H2B, the major ubiquitinated proteins of chromatin.

Asynchronous regulation of splicing events within protein 4.1 pre-mRNA during erythroid differentiation.

Protein 4.1 is an 80-kD structural component of the red blood cell (RBC) cytoskeleton. It is critical for the formation of the spectrin/actin/protein 4.1 junctional complex, the integrity of which is important for the horizontal strength and elasticity of RBCs. We and others have previously shown that multiple protein 4.1 mRNA isoforms are generated from a single genomic locus by several alternative mRNA splicing events, leading to the insertion or skipping of discrete internal sequence motifs. The physiologic significance of these motifs: (1) an upstream 17-nucleotide sequence located at the 5' end of exon 2 that contains an in-frame ATG initiation codon, the inclusion of which by use of an alternative splice acceptor site in exon 2 allows the production of a 135-kD high-molecular-weight isoform present in nonerythroid cells; (2) exon 16, which encodes a 21-amino acid (21aa) segment located in the 10-kD "spectrin/actin binding domain" (SAB), the presence of which is required for junctional complex stability in RBCs. Previous studies by our group and others suggested that, among blood cells, this exon was retained only in mature mRNA in the erythroid lineage. Exon 16 is one of a series of three closely linked alternatively spliced exons, generating eight possible mRNA products with unique configurations of the SAB. In this communication, we report studies of the expression of both the translation initiation region and the SAB region during induced erythroid maturation in mouse erythroleukemia (MEL) cells. We have found that only two of eight possible combinatorial patterns of exon splicing at the SAB region are encountered: the isoform lacking all three exons, present in predifferentiated cells, and the isoform containing only exon 16, which increases in amount during erythroid differentiation. The protein isoform containing the 21aa segment encoded by exon 16 efficiently and exclusively incorporates into the membrane, whereas the isoform lacking this 21aa segment remains in the cytoplasm, as well as the membrane. In contrast with exon 16, the erythroid pattern of exon 2 splicing, i.e., skipping of the 17-base sequence at the 5' end, was found to be already established in the uninduced MEL cells, suggesting strongly that this regulated splicing event occurs at an earlier stage of differentiation. Our results demonstrate asynchronous regulation of two key mRNA splicing events during erythroid cell maturation. These findings also show that the splicing of exon 16 alters the intracellular localization of protein 4.1 in MEL cells, and appears to be essential for its targeting to the plasmalemma.

In vitro assembly of multiprotein complexes containing alpha, beta, and gamma tubulin, heat shock protein HSP70, and elongation factor 1 alpha.

We have isolated two sets of multiprotein complexes from supernatants from high-speed centrifugation of nocodazole-arrested CHO cells. One set, assembled in vitro after a 37 degrees C incubation in the presence of ATP or GTP, is composed of equivalent amounts of alpha- and beta-tubulin and a 50-kDa protein, provisionally identified as elongation factor 1 alpha. These complexes, which are heterogeneous in size when analyzed by sucrose gradient ultracentrifugation, also contain the cognate form of heat shock protein HSP70 and gamma-tubulin, a tubulin isoform of low abundance, along with other proteins known to be involved in the regulation of mitosis. Similar but distinct complexes assemble in vitro if the same extracts are incubated at 37 degrees C without added nucleotides; multiprotein complexes generated under these conditions lack HSP70 but contain instead a 43-kDa protein identified as an actin isoform. Both sets of assembled complexes exhibit a globular substructure when analyzed by electron microscopy, and their size distribution suggests that they assemble by the step-wise addition of smaller precursors. The properties of these multiprotein complexes and their presence in cells arrested in a stage between prophase and metaphase suggest that they may be precursors to mitotic centrosomes and are possibly involved in mitotic spindle nucleation.

Genomic structure of the locus encoding protein 4.1. Structural basis for complex combinational patterns of tissue-specific alternative RNA splicing.

Protein 4.1 (P4.1) is a multifunctional protein with heterogeneity in molecular weight, intracellular localization, tissue- and development-specific expression patterns. We have analyzed the genomic structure of the locus encoding mouse P4.1 and have systematically analyzed diverse P4.1 mRNA isoforms expressed in erythroid and nonerythroid tissues. Our results indicate that the mouse protein 4.1 gene, over 90 kilobases long, comprises at least 23 exons (13 constitutive exons, 10 alternative exons) interrupted by 22 introns. The donor and acceptor splice site sequences match the consensus sequences for the exon-intron boundaries of most eukaryotic genes. No significant sequence difference was observed between splice junctions of alternative and constitutive exons. Apparently, most alternative exon-encoded peptides are located within particular functional domains of the P4.1 protein: two peptides encoded by alternative exons 4 and 5 are located near or within the glycophorin/calmodulin binding domain, whereas three other alternative exon-encoded peptides (19-amino acid encoded by exon 14, 14-amino acid encoded by exon 15, and 21-amino acid encoded by exon 16) are located near or within the spectrin-actin binding domain. Selective use of exon 2', which carries an upstream translation initiation codon (AUG), may produce an elongated P4.1 isoform (135 kDa) that is predominantly expressed in nonerythroid tissues. Combinatorial splicing of these exons may generate different isoforms that exhibit complicated tissue-specific expression patterns.

Molecular analysis of insertion/deletion mutations in protein 4.1 in elliptocytosis. I. Biochemical identification of rearrangements in the spectrin/actin binding domain and functional characterizations.

Protein 4.1 (80 kD) interacts with spectrin and short actin filaments to form the erythrocyte membrane skeleton. Mutations of spectrin and protein 4.1 are associated with elliptocytosis or spherocytosis and anemia of varying severity. We analyzed two mutant protein 4.1 molecules associated with elliptocytosis: a high molecular weight 4.1 (95 kD) associated with mild elliptocytosis without anemia, and a low molecular weight 4.1 (two species at 68 and 65 kD) associated with moderate elliptocytosis and anemia. 4.1(95) was found to contain a approximately 15-kD insertion adjacent to the spectrin/actin binding domain comprised, at least in part, of repeated sequence. 4.1(68/65) was found to lack the entire spectrin-actin binding domain. The mechanical stability of erythrocyte membranes containing 4.1(95) was identical to that of normal membranes, consistent with the presence of an intact spectrin-actin binding domain in protein 4.1. In contrast, membranes containing 4.1(68/65) have markedly reduced mechanical stability as a result of deleting the spectrin-actin binding domain. The mechanical stability of these membranes was improved following reconstitution with normal 4.1. These studies have thus enabled us to establish the importance of the spectrin-actin binding domain in regulating the mechanical stability of the erythrocyte membrane.

Full-length sequence of the cDNA for human erythroid beta-spectrin.

Spectrin is the major molecular consituent of the red cell membrane skeleton. We have isolated overlapping human erythroid beta-spectrin cDNA clones and determined 6773 base pairs of contiguous nucleotide sequence. This includes the entire coding sequence of beta-spectrin. The sequence translates into a 2137 amino acid, 246-kDa peptide. beta-Spectrin is found to consist of three distinct domains. Domain I, at the N terminus, is a 272-amino acid region lacking resemblance to the spectrin repetitive motif. Sequences in this region exhibit striking sequence homology, at both nucleotide and amino acid levels, to the N-terminal "actin-binding" domains of alpha-actinin and dystrophin. Between residues 51 and 270 there is 55% amino acid identity to human dystrophin, with only four single amino acid gaps in alignment. Domain II consists of 17 spectrin repeats. Several sequence variations are observed in typical repeat structure. Homology to alpha-actinin extends beyond domain I into the N-terminal portion of domain II. Domain III, 52 amino acid residues at the C terminus, does not adhere to the spectrin repeat motif. Combining knowledge of spectrin primary structure with previously reported functional studies, it is possible to make several inferences regarding structure/function relationships within the beta-spectrin molecule.

Heterogeneity of mRNA and protein products arising from the protein 4.1 gene in erythroid and nonerythroid tissues.

Immunologically cross-reactive isoforms of the cytoskeletal element protein 4.1 have been identified in many tissues in which they exhibit heterogeneity of molecular weight, abundance, and intracellular localization. To examine the basis for isoform production in erythroid and nonerythroid tissues, we have compared the structure and expression of cDNAs isolated from human erythroid and nonerythroid sources. We have encountered cDNAs representing many distinct mRNA sequences. These exhibit complete nucleotide sequence homology along most of their lengths. Differences were confined to five sequence blocks designated Motifs I-V, which were present or absent in each mRNA moiety. Motif I was expressed only in erythroid cells; it encodes 21 amino acids in a well-characterized spectrin/actin binding domain. Motif II, located near the COOH terminus of the 80-kD "erythroid" protein 4.1 molecule is present in the vast majority of transcripts from both erythroid and nonerythroid cells. Motifs IV and V alter the 5' untranslated region: simultaneous insertion of Motif IV and deletion of Motif V in the untranslated region inserts a new initiator methionine and establishes a contiguous open reading frame encoding a novel 135-kD protein 4.1 molecule. By immunochemical analysis we have identified the longer isoform in cells. Our results are most consistent with tissue-specific alternative mRNA splicing of transcripts of the protein 4.1 gene to yield numerous isoforms. These isoforms exhibit tissue specificity and alter strategic portions of the molecule. Moreover, we describe a novel high molecular weight form of protein 4.1 that arises by splicing events which allow translation at an upstream site.

The complete cDNA and polypeptide sequences of human erythroid alpha-spectrin.

Overlapping human erythroid alpha-spectrin cDNA clones were isolated from lambda gt11 libraries constructed from cDNAs of human fetal liver and erythroid bone marrow. The composite 8001-base pair (bp) cDNA nucleotide sequence contains 187-bp 5'- and 528-bp 3'-untranslated regions and has a single long open reading frame of 7287 bp that encodes a polypeptide of 2429 residues. As previously described (Speicher, D. W., and Marchesi, V. T. (1984) Nature 311, 177-180), spectrin is composed largely of homologous 106-amino acid repeat units. From the amino acid sequence deduced from the cDNA, alpha-spectrin can be divided into 22 segments. Segments 1-9 and 12-19 are homologous and can therefore be considered repeats; the average number of identical residues in pairwise comparisons of these repeats is 22 out of 106, or 21%. Of these 17 repeats, 11 are exactly 106 amino acids in length, whereas five others differ from this length by a single residue. Segments 11, 20, and 21, although less homologous, appear to be related to the more highly conserved repeat units. The very N-terminal 22 residues, segment 10, which is atypical both in length and sequence, and the C-terminal 150 residues in segment 22 appear to be unrelated to the conserved repeat units. The sequence of the erythroid alpha-spectrin polypeptide chain is compared to that of human alpha-fodrin and chicken alpha-actinin to which it is related. alpha-Spectrin is more distantly related to dystrophin.

Sequence and exon-intron organization of the DNA encoding the alpha I domain of human spectrin. Application to the study of mutations causing hereditary elliptocytosis.

We have determined the exon-intron organization and the nucleotide sequence of the exons and their flanking intronic DNA in cloned genomic DNA that encodes the first 526 amino acids of the alpha I domain of the human red cell spectrin polypeptide chain. From the gene sequence we designed oligonucleotide primers to use in the polymerase chain reaction technique to amplify the appropriate exons in DNA from individuals with three variants of hereditary elliptocytosis characterized by the presence of abnormal alpha I spectrin peptides, 46-50 and 65-68 kD in size, in partial tryptic digests of spectrin. The alpha I/68-kD abnormality resulted from a duplication of leucine codon 148 in exon 4: TTG-CTG to TTG-TTG-CTG. The alpha I/50a defect was associated in different individuals with two separate single base changes in exon 6: CTG to CCG (leucine to proline) encoding residue 254, and TCC to CCC (serine to proline) encoding residue 255. In another individual with the alpha I/50a polypeptide defect, the nucleotide sequence encoding amino acid residues 221 through 264 was normal. The alpha I/50b abnormality resulted from a single base change of CAG (glutamine) to CCG (proline) encoding residue 465 in exon 11 in two unrelated individuals. In a third individual with alpha I/50b-kD hereditary elliptocytosis, the entire exon encoding residues 445 through 490 was normal. The relationship of the alpha I domain polypeptide structure to these mutations and the organization of the gene is discussed.

Characterization of the calmodulin-binding site of nonerythroid alpha-spectrin. Recombinant protein and model peptide studies.

An important function of the mammalian nonerythroid alpha-spectrin chain (alpha-fodrin) that distinguishes it from the closely related erythroid isoform is its ability to bind calmodulin. By analysis of a series of deleted recombinant spectrin fusion proteins, we have identified a region in the nonerythroid alpha chain involved in calcium-dependent binding of calmodulin. The region is distinctive in that the sequence is absent from the homologous domain of the erythroid alpha chain and diverges from the normal internal repeat structure observed throughout other spectrins. In order to determine limits of this functional site, a synthetic peptide as small as 24 residues was shown to compete with either recombinant or brain alpha-spectrin in binding to calmodulin. The active peptide, which was derived from a segment between repeats 11 and 12, was composed of the following sequence: Lys-Thr-Ala-Ser-Pro-Trp-Lys-Ser-Ala-Arg-Leu-Met-Val-His-Thr-Val-Ala-Thr-Phe-Asn - Ser-Ile-Lys-Glu. Comparison of this sequence with functional sites in other diverse calcium-dependent calmodulin-binding proteins has revealed a structural motif common to all of these proteins, namely clusters of hydrophobic residues interspersed with basic residues. When folded into alpha-helical conformations, these binding sites are predicted to form amphipathic structures.

Synthetic peptides mimic the assembly of transmembrane glycoproteins.

The composition of the intramembranous domains of many receptors are remarkably uniform, yet there is evidence that many transmembrane proteins associate together to form specific noncovalent homo- or heterocomplexes within the membrane. We have synthesized peptides corresponding to transmembrane domains of glycophorin A, glycophorin C, and the interleukin 2-receptor Tac antigen to study the interactions between transmembrane domains in vitro. Synthetic transmembrane glycophorin A peptide formed a complex with native glycophorin and glycoproteins of erythrocyte and K562 cell membranes that was reversible, specific, and could be demonstrated in a natural bilayer system in the absence of detergents. Synthetic glycophorin C and interleukin 2-receptor Tac antigen transmembrane peptides, although similar in amino acid composition, did not interact with glycophorin and did not inhibit the binding of the synthetic glycophorin A transmembrane peptide to native glycophorin. It is proposed that the transmembrane segments of receptor proteins contain not only the structural information necessary for insertion and anchoring but specific binding sites that mediate interactions between transmembrane glycoproteins.

Erythrocyte membrane skeleton phosphoproteins: identification of two unrelated phosphoproteins in band 4.9.

Human erythrocyte membrane band 4.9 is phosphorylated by several erythrocyte protein kinases. Chromatography of erythrocyte membrane skeleton proteins on DEAE-Sephacel produces two proteins with relative mobilities, on gel electrophoresis, similar to that of band 4.9. The first, with a molecular mass of 49 kDa, is quite basic (pI greater than 8) while the second, 50.5 kDa, is slightly acidic (pI = 6.2). Comparative two-dimensional peptide mapping reveals that both proteins are present in band 4.9 on one-dimensional gels of total erythrocyte membrane proteins and membrane skeleton proteins. The 49 kDa protein, but not the 50.5 kDa protein, binds to actin filaments in a sedimentation assay. In intact erythrocytes metabolically labeled with [32P]orthophosphate, the 49 kDa protein is phosphorylated by protein kinase C, cAMP-dependent protein kinase, and protein kinases which are active in the absence of exogenous kinase activators. In contrast, the 50.5 kDa protein is phosphorylated by protein kinase C but not by the other protein kinases examined. Finally, two-dimensional peptide mapping was employed to compare the 49 kDa protein and a 57 kDa protein which copurifies with, and has many characteristics of, the 49 kDa protein. Significant similarities were found in both 125I-labeled chymotryptic peptide maps and 32P-labeled tryptic peptide maps, suggesting that the 49 kDa and 57 kDa proteins are closely related.