Calcium-Activated Neutral Proteinase of Human Brain: Subunit Structure and Enzymatic Properties of Multiple Molecular Forms

Abstract Calcium-activated neutral proteinase (CANP) was purified 2,625-fold from postmortem human cerebral cortex by a procedure involving chromatography on diethylaminoethyl (DEAE)-cellulose, phenyl-Sepharose, Ultrogel AcA-44, and DEAE-Biogel A. The major active form of CANP exhibited a molecular weight of 94–100 kilodaltons (Kd) by gel filtration on Sephacryl 300 and consisted of 78-Kd and 27-Kd subunits. Two-dimensional gel electrophoresis resolved the small subunit into two molecular species with different isoelectric points. CANP degraded most human cytoskeletal proteins but was particularly active toward fodrin and the neurofilament protein subunits (145 Kd > 200 Kd > 70 Kd). The enzyme required 175 μMCa²⁺ for half-maximal activation and 2 mM Ca²⁺ for optimal activity toward [methl-¹⁴C]azocasein. Other divalent metal ions were poor activators of the enzyme, and some, including copper, lead, and zinc, strongly inhibited the enzyme. Aluminum, a neurotoxic ion that induces neurofilament accumulations in mammalian brain, inhibited the enzyme 47% at 1 mM and 100% at 5 mM A second CANP form lacking the 27-Kd subunit was partially resolved from the 100-Kd heterodimer during DEAE-Biogel A chromatography. The 78-Kd monomer exhibited the same specific activity, calcium ion requirement, pH optimum, and specificity for cytoskeletal proteins as the 100-Kd heterodimer, suggesting that the 27-Kd subunit is not essential for the major catalytic properties of the enzyme. The rapid autolysis of the 27-Kd subunit to a 18-Kd intermediate when CANP is exposed to calcium may explain differences between our results and previous reports, which describe brain mCANP in other species as a 76-80-Kd monomer or a heterodimer containing 76-80-Kd and 17-20-Kd subunits. The similarity of the 100-Kd human brain CANP to CANPs in nonneural tissues indicates that the heterodimeric form is relatively conserved among various tissues and species.     

A structural model of human erythrocyte protein 4.1

Limited proteolysis and specific chemical cleavage methods have enabled a detailed structural characterization of human erythrocyte protein 4.1. This protein is composed of two chemically very similar polypeptide chains (a and b) with apparent molecular masses of 80,000 and 78,000 daltons. Cleavage of protein 4.1 at cysteine residues by 2-nitro-5-thiocyanobenzoic acid produces a series of doublets which differ by approximately 2,000 daltons and have identical peptide maps. Alignment of these peptides by mapping analysis has localized 4 cysteine residues within a 17,000-dalton segment on both a and b polypeptides. Mild chymotryptic treatment at 0 degrees C cleaves protein 4.1 primarily in three central locations and generates two families of unrelated peptides. Analysis of these fragments in two-dimensional gels and by peptide mapping reveals an unusual polarity in protein 4.1 structure in that each polypeptide chain contains two segments, one relatively acidic the other basic, that are segregated at opposite ends of the molecule. The basic region is digested into a cysteine-rich 30,000-dalton domain which resists further breakdown while the acidic region is readily degraded into smaller fragments. The peptides derived from the acidic region all appear as doublets suggesting that protein 4.1 a and b polypeptides differ close to the terminus of the acidic end. Similar phosphorylation sites occur on both polypeptides within a segment some 24,000-34,000 daltons from the acidic terminus.     

Characterization of two distinct disulfide-linked B-G molecules in the chicken

Alloantisera specific for B-G antigens recognized a complex of molecules of apparent molecular weights of 90 and 98 Kd under nonreducing conditions and molecules of 40, 44, and 48 Kd under reducing conditions on both embryo- and adult-derived peripheral red blood cells (RBC). The chicken B-G molecules produced a unique two-dimensional "diagonal" pattern. Two antisera permitted the characterization of the complex B-G molecular profile as a homodimer composed of 48-Kd subunits and as a heterodimer composed of 40- and 44-Kd subunits. A rabbit antiserum produced against B-G molecules preferentially recognized the 48-Kd reduced molecules, suggesting that the 90-Kd molecule was a homodimer composed of two 48-Kd molecules. One B-G reagent was capable of recognizing only the 98-Kd nonreduced B-G molecule that gave rise to 40- and 44-Kd molecules under reducing conditions, suggesting that the 98-Kd molecule was a heterodimer composed of 44- and 40-Kd subunits. Adult chicken B-G2 molecules produced a variety of two-dimensional isoelectric focusing/sodium dodecyl sulfate-polyacrylamide gel electrophoresis (IEF/SDS-PAGE) patterns depending on the characteristics of the reagent employed in the immunoprecipitation. B-G molecules were immunoprecipitated from primitive and definitive chicken RBCs but not from any nonerythroid cells tested. B-G molecules were not expressed by avian erythroblastosis virus (AEV)-transformed erythroleukemia cells, nor were they induced to appear with butyric acid-induced erythroid differentiation.     

Spectrin, fodrin and protein 4.1-like proteins in differentiating rat germ cells

The presence and the distribution of proteins of the membrane skeleton in differentiating germ cells of the rat has been investigated. Immunofluorescence and immunoblotting analysis, performed using monoclonal and polyclonal antibodies to human erythroid alpha-spectrin and protein 4.1 and to brain spectrin (fodrin), demonstrated the presence of analogues of spectrin and fodrin in spermatocytes and round spermatids and of protein 4.1-like molecules in spermatocytes, spermatids and spermatozoa. Spectrin and fodrin showed molecular weights comparable to those of their analogues in erythrocytes but a distinct intracellular distribution. Fodrin was localized along the plasma membrane while spectrin appeared associated with the regions of the Golgi apparatus and of the developing acrosome. Antibodies to protein 4.1 recognized molecules with a molecular weight not comparable with that in erythrocytes, and their presence in spermatozoa was confined to specific regions of the head and of the tail.     

O-N-acetyl-D-glucosamine moiety on discrete peptide of multiple protein 4.1 isoforms regulated by alternative pathways

Erythrocyte protein 4.1 is a cytoplasmic protein that possesses a protein-saccharide modification structure, an O-N-acetyl-D-glucosamine (GlcNAc) moiety. We determined the amino acid sequence of the proteolytic fragment containing the O-GlcNAc moiety after labeling the saccharide with [3H]galactose in the presence of bovine milk galactosyltransferase. Glycosylation appears to occur on one or more serine or threonine residues in the following sequence: Thr-Ala-Gln-Thr-Ile-Thr-Ser-Glu-Thr-Pro-Ser-Ser-Thr-Thr-Thr-Thr-Gln-Ile-Thr-Lys . This sequence corresponds to the carboxyl-terminal half of the 34-amino acid peptide in the 22/24-kDa carboxyl-terminal domain of protein 4.1, which is one of the discrete peptides regulated by alternative RNA splicing. Multiple protein 4.1 isoforms in erythroid and nonerythroid cells including major components of erythrocyte membrane proteins, 4.1a and 4.1b, appear to contain this sequence since most immunochemically reactive proteins were labeled with [3H]galactose, with the exception of several variant polypeptides. These results appear to suggest the functional or biological significance of the O-GlcNAc linkage in protein 4.1.     

Ca2(+)-dependent regulation of the spectrin/actin interaction by calmodulin and protein 4.1

The Ca2(+)-dependent regulation of the erythroid membrane cytoskeleton was investigated. The low-salt extract of erythroid membranes, which is mainly composed of spectrin, protein 4.1, and actin, confers a Ca2+ sensitivity on its interaction with F-actin. This Ca2+ sensitivity is fortified by calmodulin and antagonized by trifluoperazine, a potent calmodulin inhibitor. Additionally, calmodulin is detected in the low-salt extract. These results suggest that calmodulin is the sole Ca2(+)-sensitive factor in the low-salt extract. The main target of calmodulin in the erythroid membrane cytoskeleton was further examined. Under native conditions, calmodulin forms a stable and equivalent complex with protein 4.1 as determined by calmodulin affinity chromatography, cross-linking experiments, and fluorescence binding assays with an apparent Kd of 5.5 x 10(-7) M irrespective of the free Ca2+ concentration. Domain mapping with chymotryptic digestion reveals that the calmodulin-binding site resides within the N-terminal 30-kDa fragment of protein 4.1. In contrast, the interaction of calmodulin with spectrin is unexpectedly weak (Kd = 1.2 x 10(-4) M). Given the content of calmodulin in erythrocytes (2-5 microM), these results imply that the major target for calmodulin in the erythroid membrane cytoskeleton is protein 4.1. Low- and high-shear viscometry and binding assays reveal that an equivalent complex of calmodulin with protein 4.1 regulates the spectrin/actin interaction in a Ca2(+)-dependent manner. At a low Ca2+ concentration, protein 4.1 potentiates the actin cross-linking and the actin binding activities of spectrin. At a high Ca2+ concentration, the protein 4.1-potentiated actin cross-linking activity but not the actin binding activity of spectrin is suppressed by Ca2+/calmodulin. The Ca2(+)-dependent regulation of the spectrin/protein 4.1/calmodulin/actin interaction is discussed.     

Detergent dissociation of bovine liver phosphomannosyl binding protein

We have reported previously the isolation and partial characterization of a 215-kilodalton (Kd) phosphomannosyl binding protein from bovine liver membranes [3,9]. In the present studies evidence is presented that the binding protein is an aggregate. Four N-terminal amino acids were detected, and the complex could be dissociated into subunits. Bovine liver membranes were extracted with the detergent, Zwittergent, in the presence of protease inhibitors. The extract was subjected to affinity chromatography on phosphomannan-Sepharose 4B, and proteins with apparent Mr values of 215 and 57 Kd were eluted with mannose 6-phosphate. As reported previously, extraction with Triton X-100 yielded only the higher molecular weight material. When the binding protein was incubated at 4 degrees C in the presence of Zwittergent TM 3-14 the 215-Kd form slowly dissociated into smaller subunits; after two months, the major species had an apparent Mr of 57 Kd. The subunits derived from the binding protein were recognized by antiserum raised against purified binding protein. Dissociation of the binding protein by Zwittergent was enhanced by incubation at 37 degrees C, the presence of dithiothreitol, and low pH values. The subunit mixture enriched in the 57-Kd subunit had a lowered ability to bind ligands containing the phosphomannosyl recognition marker. Binding was partially restored (greater than 48% of the initial value) when dissociated receptor was back exchanged with Triton X-100.     

Inactivation of the human papillomavirus-16 E6 oncoprotein by organic disulfides

We are investigating compounds that could be useful in the treatment of neoplastic lesions of the cervix by acting on the oncoprotein E6 of human papillomavirus-16. The E6 protein contains two potential zinc-binding domains that are required for most of its functions. We have published tests that measure (i) the release of zinc ions after chemical alteration of the cysteine groups of these zinc-binding domains (TSQ assay), (ii) the interaction of E6 with the cellular proteins E6AP and E6BP (BIACORE assay), and (iii) the viability of tumor cell lines that require the continuous expression of HPV oncoproteins (WST1 assay). Based on these tests, we identified 4.4'-dithiodimorpholine as a potential lead compound. In this study we examined whether the dithiobisamine moiety of 4,4'-dithiodimorpholine may be an important molecular prerequisite for further drug development in this system. We have evaluated 59 new substances including organic disulfides and those containing the dithiobisamine moiety, as well as structural analogues. The compounds with significant reactivity in all three assays were observed only for dithiobisamine derivatives with saturated cyclic amines and aryl substituted piperazines. The identity of these substances suggests that the N-S-S-N moiety is necessary but not sufficient for reactivity in our assays, and that dithiobisamine based substances are useful as lead compounds that target the cysteine groups of HPV-16 E6 zinc fingers.     

Beta spectrin bestows protein 4.1 sensitivity on spectrin-actin interactions

The ability of protein 4.1 to stimulate the binding of spectrin to F-actin has been compared by cosedimentation analysis for three avian (erythrocyte, brain, and brush border) and two mammalian (erythrocyte and brain) spectrin isoforms. Human erythroid protein 4.1 stimulated actin binding of all spectrins except the brush border isoform (TW 260/240). These results suggested that the beta subunit determined the protein 4.1 sensitivity of the heterodimer, since all avian alpha subunits are encoded by a single gene. Tissue-specific posttranslational modification of the alpha subunit was excluded by examining the properties of hybrid spectrins composed of the purified alpha subunit from avian erythrocyte or brush border spectrin and the beta subunit of human erythrocyte spectrin. A hybrid composed of avian brush border alpha and human erythroid beta spectrin ran on nondenaturing gels as a discrete band, migrating near human erythroid spectrin tetramers. The actin-binding activity of this hybrid was stimulated by protein 4.1, while either chain alone was devoid of activity. Therefore, although both subunits were required for actin binding, the sensitivity of the spectrin-actin interaction to protein 4.1 is a property uniquely bestowed on the heterodimer by the beta subunit. The singular insensitivity of brush border spectrin to stimulation by erythroid protein 4.1 was also consistent with the absence of proteins in avian intestinal epithelial cells which were immunoreactive with polyclonal antisera sensitive to all of the known avian and human erythroid 4.1 isoforms.     

Binding of Bodian''s Silver and Monoclonal Antibodies to Defined Regions of Human Neurofilament Subunits: Bodian''s Silver Reacts with a Highly Charged Unique Domain of Neurofilaments

Cleavage at cysteine and chymotrypsin digestion were applied to two human neurofilament (NF) subunits, low- and high-molecular-weight NF (NF-L and NF-H), to locate the regions reacting with Bodian's silver stain and with several monoclonal antibodies, including NF-specific antibodies and one that recognizes all intermediate filaments (anti-IFA). Our findings indicate that whereas anti-IFA recognizes the highly conserved rod domain, all the NF-specific antibodies, as well as Bodian's silver, react with the carboxy-terminal tailpiece of NF subunits. The silver binding sites in NF-L are located in a carboxy-terminal 12-Kd chymotrypsin fragment, a highly charged, unique domain of NF.     

The role of disulfide bonds and alpha-helical coiled-coils in the biosynthesis of type XIII collagen and other collagenous transmembrane proteins

Type XIII collagen is a type II transmembrane protein with three collagenous (COL1-3) and four noncollagenous domains (NC1-4). The human alpha1(XIII) chain contains altogether eight cysteine residues. We introduced point mutations to six of the most N-terminal cysteine residues, and we show here that the two cysteines 117 and 119 at the end of the N-terminal noncollagenous domain (NC1) are responsible for linking the three alpha1(XIII) chains together by means of interchain disulfide bonds. In addition, the intracellular and transmembrane domains have an impact on trimer formation, whereas the cysteines in the transmembrane domain and the COL1, the NC2, and the C-terminal NC4 domains do not affect trimer formation. We also suggest that the first three noncollagenous domains (NC1-3) harbor repeating heptad sequences typical of alpha-helical coiled-coils, whereas the conserved NC4 lacks a coiled-coil probability. Prevention of the coiled-coil conformation in the NC3 domain is shown here to result in labile type XIII collagen molecules. Furthermore, a new subgroup of collagenous transmembrane proteins, the Rattus norvegicus, Drosophila melanogaster, and Caenorhabditis elegans colmedins, is enlarged to contain also Homo sapiens collomin, and Pan troglodytes, Mus musculus, Tetraodon nigroviridis, and Dano rerio proteins. We suggest that there is a structurally varied group of collagenous transmembrane proteins whose biosynthesis is characterized by a coiled-coil motif following the transmembrane domain, and that these trimerization domains appear to be associated with each of the collagenous domains. In the case of type XIII collagen, the trimeric molecule has disulfide bonds at the junction of the NC1 and COL1 domains, and the type XIII collagen-like molecules (collagen types XXIII and XXV) and the colmedins are similar in that they all have a pair of cysteines in the same location. Moreover, furin cleavage at the NC1 domain can be expected in most of the proteins.     

Structural consequences of cysteine substitutions C1977Y and C1977R in calcium-binding epidermal growth factor-like domain 30 of human fibrillin-1

The largest group of disease-causing mutations affecting calcium-binding epidermal growth factor-like (cbEGF) domain function in a wide variety of extracellular and transmembrane proteins is that which results in cysteine substitutions. Although known to introduce proteolytic susceptibility, the detailed structural consequences of cysteine substitutions in cbEGF domains are unknown. Here, we studied pathogenic mutations C1977Y and C1977R, which affect cbEGF30 of human fibrillin-1, in a recombinant three cbEGF domain fragment (cbEGF29-31). Limited proteolysis, 1H NMR, and calcium chelation studies have been used to probe the effect of each substitution on cbEGF30 and its flanking domains. Analysis of the wild-type fragment identified two high affinity and one low affinity calcium-binding sites. Each substitution caused the loss of high affinity calcium binding to cbEGF30, consistent with intradomain misfolding, but the calcium binding properties of cbEGF29 and cbEGF31 were surprisingly unaffected. Further analysis of mutant fragments showed that domain packing of cbEGF29-30, but not cbEGF30-31, was disrupted. These data demonstrate that C1977Y and C1977R have localized structural effects, confined to the N-terminal end of the mutant domain, which disrupt domain packing. Cysteine substitutions affecting other cbEGF disulfide bonds are likely to have different effects. This proposed structural heterogeneity may underlie the observed differences in stability and cellular trafficking of proteins containing such changes.     

Mechanisms of cytoskeletal regulation: modulation of aortic endothelial cell protein band 4.1 by the extracellular matrix

The bovine aortic endothelial cell (BAEC) cytoskeleton is a complex structure modulated by many stimuli including release from contact inhibition and various components of the extracellular matrix (ECM). Transduction of information from the ECM to the cell nucleus proceeds via several complex pathways including the cytoskeleton. We have demonstrated the presence of an immunoreactive isoform of the human erythrocyte cytoskeletal protein band 4.1 (4.1) in BAEC. BAEC 4.1 is similar in molecular weight to the erythroid protein by immunoblot analyses and produces a similar pattern of cysteine specific cleavage products consistent with a cluster of cysteine residues previously described in the erythroid molecule. We have also examined the effects of defined ECM proteins on the distributions of cultured BAEC 4.1 and actin filaments (AF) at confluency and following release from contact inhibition. The distribution of 4.1 in BAEC on a plasma fibronectin substrate is complex, having partial codistribution with cytoplasmic AF and a unique perinuclear staining. In contrast, on a collagen type I/III substrate, 4.1 is localized, in part, to peripheral areas of cell-cell contact distinct from the dense peripheral band staining of AF. During migration on this substrate, 4.1 had a filamentous distribution having partial codistribution with AF. Indirect immunofluorescence staining of cross-sections of bovine calf aortae revealed a cortical staining pattern in the aortic endothelial cells with staining noted on the luminal and basolateral aspects of the cells. These data suggest that, in endothelial cells, protein 4.1 is a cortical membrane protein which may function to link actin filaments to other skeletal proteins such as spectrin. These findings also suggest an active role for protein 4.1 in cytoskeletal reorganization events which can occur in response to external stimuli, such as the extracellular matrix or contact with other cells.     

The steroid-binding properties of recombinant glucocorticoid receptor: a putative role for heat shock protein hsp90

The steroid-binding domain of the human glucocorticoid receptor was expressed in Escherichia coli either as a fusion protein with protein A or under control of the T7 RNA polymerase promoter. The recombinant proteins were found to bind steroids with the normal specificity for a glucocorticoid receptor but with reduced affinity (Kd for triamcinolone acetonide approximately 70 nM). Glycerol gradient analysis of the E. coli lystate containing the recombinant protein indicated no interaction between the glucocorticoid receptor fragment and heat shock proteins. However, synthesis of the corresponding fragments of glucocorticoid receptor in vitro using rabbit reticulocyte lystate resulted in the formation of proteins that bound triamcinolone acetonide with high affinity (Kd 2nM). Glycerol gradient analysis of these proteins, with and without molybdate, indicated that the in vitro synthesised receptor fragments formed complexes with hsp90 as previously shown for the full-length rat glucocorticoid receptor. Radiosequence analysis of the recombinant steroid-binding domain expressed in E. coli and affinity labelled with dexamethasone mesylate identified binding of the steroid to Cys-638 predominantly. However, all cysteine residues within the steroid-binding domain were affinity labelled to a certain degree indicating that the recombinant protein has a structure similar to the native receptor but more open and accessible.     

Mapping long-range chromatin organization within the chicken alpha-globin gene domain using oligonucleotide DNA arrays

We have analyzed the organization of the chicken alpha-globin gene domain using DNA miniarrays and have found two novel chromatin loop attachment regions. We have found a 40-kb loop domain that includes all the alpha-globin genes in cells of erythroid origin. One of the domain borders colocalizes almost exactly with a strong MAR element and with a block of enhancer-blocking elements found earlier at the upstream end of the alpha-globin gene domain. The domain structure was found to be different in a lymphoid cell line DT40. We propose to use the technique of DNA arrays to map the nuclear matrix attachment sites that define the borders of chromosome loop domains. The technique of DNA arrays permits a large number of DNA sequences to be immobilized on a glass or nylon matrix. This may prove useful for mapping chromatin loop positions within the human genome by using a pool of chromatin loop attachment regions as a probe in a hybridization with a DNA chip containing a specific DNA region.     

Three molecular forms of a rat asialoglycoprotein receptor

It has been reported that a rat asialoglycoprotein receptor is composed of three polypeptide chains with molecular masses of 43, 54, and 64 kilodaltons (43, 54, and 64-Kd forms) and that the first has a different primary structure from the latter two forms. Incorporation of [3H]leucine into these forms showed that no precursor-product relationship is found between the 54-Kd and 64-Kd forms. The half-life of the 43-Kd form (25 h) was shorter than those of the 54-Kd and 64-Kd forms (66 and 70 h, respectively). Glycopeptides of the three forms were prepared from rat livers previously labeled in vivo with [3H]glucosamine. Gel filtration analysis of the glycopeptides before and after endo H treatment revealed that they were all resistant to endo H. Alkali treatment did not change the elution position appreciately. These results indicate that the three molecular forms contained only complex oligosaccharide chains. The receptor was prepared from rat livers previously treated with tunicamycin in vivo and subjected to SDS-PAGE. A distinct band with a molecular mass of 33 Kd was observed. The receptor was also immunoprecipitated from rat hepatocytes in primary culture previously labeled with [35S]methionine and analysed by SDS-PAGE and fluorography. In addition to the major 43-Kd form, a band with a molecular mass of 41 Kd was found and tunicamycin treatment gave rise to a 33-Kd component, which is in good agreement with the receptor purified from tunicamycin treated rats. It is suggested that the 43-Kd form is synthesized as a 33 Kd polypeptide, cotranslationally glycosylated to form the 41 Kd component and then processed to the final 43-Kd form. We also think that the 43-Kd form could bind to asialoorosomucoid-Sepharose 4B without its carbohydrate chains.     

Organization of the Human Protein 4.1 Genomic Locus: New Insights into the Tissue-Specific Alternative Splicing of the Pre-mRNA

Protein 4.1 is a globular 80-kDa component of the erythrocyte membrane skeleton that enhances spectrin–actin interaction via its internal 10-kDa domain. Previous studies have shown that protein 4.1 mRNA is expressed as multiple alternatively spliced isoforms, resulting from the inclusion or exclusion of small cassette sequences called motifs. By tissue screening for protein 4.1 isoforms, we have observed new features of an already complex pattern of alternative splicing within the spectrin/actin binding domain. In particular, we found a new 51-nt exon that is present almost exclusively in muscle tissue. In addition, we have isolated multiple genomic clones spanning over 200 kb, containing the entire erythroid and nonerythroid coding sequence of the human locus. The exon/intron structure has now been characterized; with the exception of a 17-nt motif, all of the alternatively spliced motifs correspond to individual exons. The 3′-untranslated region (UTR) has also been completely sequenced using various PCR and genomic-sequencing methods. The 3′ UTR, over 3 kb, accounts for one-half of the mature mRNA.