The molecular basis of Sanfilippo syndrome type B.

The Sanfilippo syndrome type B is a lysosomal storage disorder caused by deficiency of alpha-N-acetylglucosaminidase; it is characterized by profound mental deterioration in childhood and death in the second decade. For understanding the molecular genetics of the disease and for future development of DNA-based therapy, we have cloned the cDNA and gene encoding alpha-N-acetylglucosaminidase. Cloning started with purification of the bovine enzyme and use of a conserved oligonucleotide sequence to probe a human cDNA library. The cDNA sequence was found to encode a protein of 743 amino acids, with a 20- to 23-aa signal peptide immediately preceding the amino terminus of the tissue enzyme and with six potential N-glycosylation sites. The 8.5-kb gene (NAGLU), interrupted by 5 introns, was localized to the 5'-flanking sequence of a known gene, EDH17B, on chromosome 17q21. Five mutations were identified in cells of patients with Sanfilippo syndrome type B: 503del10, R297X, R626X, R643H, and R674H. The occurrence of a frameshift and a nonsense mutation in homozygous form confirms the identity of the NAGLU gene.

Characterization of a second secreted IgE isoform and identification of an asymmetric pathway of IgE assembly.

A number of alternatively spliced epsilon transcripts have been detected in IgE-producing B cells, in addition to the mRNAs encoding the classical membrane and secreted IgE heavy (H) chains. In a recent study, we examined the protein products of three of these alternatively spliced isoforms and found that they are intracellularly retained and degraded because of their inability to assemble into complete IgE molecules. We have now similarly examined a more recently described epsilon mRNA species that is generated by splicing between a donor splice site immediately upstream of the stop codon in the H-chain constant region exon 4 (CH4) and an acceptor site located in the 3' part of the second membrane exon. We show that this isoform is efficiently secreted by both plasma cells and B lymphocytes and therefore represents a second secreted IgE isoform (epsilon S2). The epsilon S2 H chain is only six amino acids longer than the classical secreted Ig H chain (epsilon S1) and contains a C-terminal cysteine, which is a characteristic sequence feature of mu and alpha H chains. However, unlike IgM and IgA, the epsilon S2 C-terminal cysteine (Cys-554) does not induce polymerization of H2L2 molecules (where L is light chain), but rather creates a disulfide bond between the two H chains that increases the rate of association into covalently bound H2L2 monomers. This C-terminal cysteine also does not function as an intracellular retention element because the epsilon S2 isoform was secreted in amounts equal to that of the epsilon S1, both in B lymphocytes and in plasma cells. The epsilon S2 H chains secreted by B lymphocytes differed from the epsilon S1 H chains in the extent of glycosylation. Interestingly, a difference in glycosylation between B-lymphocytes and plasma cells was also noted for both isoforms. The presence of the Cys-554 also allowed the identification of a distinctive asymmetric pathway of IgE assembly, common to both types of epsilon H chains.

BEN/SC1/DM-GRASP, a homophilic adhesion molecule, is required for in vitro myeloid colony formation by avian hemopoietic progenitors.

BEN/SC1/DM-GRASP is a membrane glycoprotein of the immunoglobulin superfamily isolated in the chick by several groups, including ours. Its expression is strictly developmentally regulated in several cell types of the nervous and hemopoietic systems and in certain epithelia. Each of these cell types expresses isoforms of BEN which differ by their level of N-glycosylation and by the presence or absence of the HNK-1 carbohydrate epitope. In the present work, the influence of glycosylation on BEN homophilic binding properties was investigated by two in vitro assays. First, each BEN isoform was covalently coupled to microspheres carrying different fluorescent dyes and an aggregation test was performed. We found that homophilic aggregates form indifferently between the same or different BEN isoforms, showing that glycosylation does not affect BEN homophilic binding properties. This was confirmed in the second test, where the BEN-coated microspheres bound to the neurites of BEN- expressing neurons, irrespective of the isoform considered. The transient expression of the BEN antigen on hemopoietic progenitors prompted us to see whether it might play a role in their proliferation and differentiation. When added to hemopoietic progenitor cells in an in vitro colony formation assay anti-BEN immunoglobulin strongly inhibited myeloid, but not erythroid, colony formation although both types of precursors express the molecule.

Induction of cellular immunity in chimpanzees to human tumor-associated antigen mucin by vaccination with MUC-1 cDNA-transfected Epstein-Barr virus-immortalized autologous B cells.

Aberrant glycosylation of the mucin molecule (encoded by the gene MUC-1) on human epithelial cell tumors leads to the exposure of tumor-associated epitopes recognized by patients' antibodies and cytotoxic T cells. Consequently, these epitopes could be considered targets for immunotherapy. We designed a cellular vaccine, employing, instead of tumor cells, autologous Epstein-Barr virus (EBV)-immortalized B cells as carriers of tumor-associated mucin, to take advantage of their costimulatory molecules for T-cell activation. The vaccine was tested in chimpanzees because of the identity of the human and chimpanzee MUC-1 tandem repeat sequence. EBV-immortalized B cells derived from two chimpanzees were transfected with MUC-1 cDNA, treated with glycosylation inhibitor phenyl-N-acetyl-alpha-D-galactosaminide to expose tumor-associated epitopes, irradiated, and injected subcutaneously four times at 3-week intervals. One vaccine preparation also contained cells transduced with the interleukin 2 (IL-2) cDNA and producing low levels of IL-2. Already after the first injection we found in the peripheral blood measurable frequency of cytotoxic T-cell precursors specific for underglycosylated mucin. The highest frequency observed was after the last boost, in the lymph node draining the vaccination site. Delayed-type hypersensitivity reaction to the injected immunogens was also induced, whereas no appearance of mucin-specific antibodies was seen. Long-term observation of the animals yielded no signs of adverse effects of this immunization. Autologous antigen-presenting cells, like EBV-immortalized B cells, expressing tumor-associated antigens are potentially useful immunogens for induction of cellular anti-tumor responses in vivo.

The peptide binding site of the substance P (NK-1) receptor localized by a photoreactive analogue of substance P: presence of a disulfide bond.

Substance P (SP) is a neuropeptide that mediates multiple physiological responses including transmission of painful stimuli and inflammation via an interaction with a receptor of known primary sequence. To identify the regions of the SP receptor, also termed the NK-1 receptor, involved in peptide recognition, we are using analogues of SP containing the photoreactive amino acid p-benzoyl-L-phenylalanine (Bpa). In the present study, we used radioiodinated Bpa8-SP to covalently label with high efficiency the rat SP receptor expressed in a transfected mammalian cell line. To identify the amino acid residue that serves as the site of covalent attachment, a membrane preparation of labeled receptor was subjected to partial enzymatic cleavage by trypsin. A major digestion product of 22 kDa was identified. Upon reduction with 2-mercaptoethanol the mass of this product decreased to 14 kDa. The 22-kDa tryptic fragment was purified in excellent yield by preparative SDS/PAGE under nonreducing conditions. Subcleavage with Staphylococcus aureus V8 protease and endoproteinase ArgC yielded fragments of 8.2 and 9.0 kDa, respectively. Upon reductive cleavage, the V8 protease fragment decreased to 3.0 kDa while the endoproteinase ArgC fragment decreased to 3.2 kDa. Taking into consideration enzyme specificity, molecular size, determination of the presence or absence of N-glycosylation sites, and recognition by antibodies to specific sequences of the SP receptor, the V8 protease fragment is Thr-173 to Glu-183, while the endoproteinase ArgC fragment is Val-178 to Arg-190. These two fragments share the common sequence Val-Val-Cys-Met-Ile-Glu (residues 178-183). The site of covalent attachment of radioiodinated Bpa8-SP is thus restricted to a residue within this overlap sequence. The data presented here also establish that the cysteine residue in this sequence Cys-180, which is positioned in the middle of the second extracellular loop, participates in a disulfide bond that links the first and second extracellular loops of the receptor.

Expression and signaling specificity of the IFNAR chain of the type I interferon receptor complex.

The IFNAR chain of the type I interferon (IFN) receptor (IFNIR) undergoes rapid ligand-dependent tyrosine phosphorylation and acts as a species-specific transducer for type I IFN action. Using the vaccinia/T7 expression system to amplify IFNAR expression, we found that human HeLa-S3 cells transiently express high levels of cell surface IFNAR chains (approximately 250,000 chains per cell). Metabolic labeling and immunoblot analysis of transfected HeLa cells show that the IFNAR chain is initially detected as 65-kDa and 98-kDa precursors, and then as the 130-kDa mature protein. Due to variation in N-glycosylation, the apparent molecular mass of the mature IFNAR chain varies from 105 to 135 kDa in different cells. IFNIR structure was characterized in various human cell lines by analyzing 125I-labeled IFN cross-linked complexes recognized by various antibodies against IFNIR subunits and JAK protein-tyrosine kinases. Precipitation of cross-linked material from Daudi cells with anti-IFNAR antibodies showed that IFNAR was present in a 240-kDa complex. Precipitation of cross-linked material from U937 cells with anti-TYK2 sera revealed a 240-kDa complex, which apparently did not contain IFNAR and was not present in IFN-resistant HEC1B cells. The tyrosine phosphorylation and down-regulation of the IFNAR chain were induced by type I IFN in several human cell lines of diverse origins but not in HEC1B cells. However, of type I IFNs, IFN-beta uniquely induced the tyrosine phosphorylation of a 105-kDa protein associated with the IFNAR chain in two lymphoblastoid cell lines (Daudi and U266), demonstrating the specificity of transmembrane signaling for IFN-beta and IFN-alpha through the IFNAR chain.

Nucleotide sequence of the cDNA encoding the proenzyme of phenol oxidase A1 of Drosophila melanogaster.

Clones encoding pro-phenol oxidase [pro-PO; zymogen of phenol oxidase (monophenol, L-dopa:oxygen oxidoreductase, EC 1.14.18.1)] A1 were isolated from a lambda gt10 library that originated from Drosophila melanogaster strain Oregon-R male adults. The 2294 bp of the cDNA included a 13-bp 5'-noncoding region, a 2070-bp encoding open reading frame of 690 amino acids, and a 211-bp 3'-noncoding region. A hydrophobic NH2-terminal sequence for a signal peptide is absent in the protein. Furthermore, there are six potential N-glycosylation sites in the sequence, but no amino sugar was detected in the purified protein by amino acid analysis, indicating the lack of an N-linked sugar chain. The potential copper-binding sites, amino acids 200-248 and 359-414, are highly homologous to the corresponding sites of hemocyanin of the tarantula Eurypelma californicum, the horseshoe crab Limulus polyphemus, and the spiny lobster Panulirus interruptus. On the basis of the phylogenetic tree constructed by the neighbor-joining method, vertebrate tyrosinases and molluscan hemocyanins constitute one family, whereas pro-POs and arthropod hemocyanins group with another family. It seems, therefore, likely that pro-PO originates from a common ancestor with arthropod hemocyanins, independently to the vertebrate and microbial tyrosinases.

Cotranslational folding and calnexin binding during glycoprotein synthesis.

To analyze cotranslational folding of influenza hemagglutinin in the endoplasmic reticulum of live cells, we used short pulses of radiolabeling followed by immunoprecipitation and analysis with a two-dimensional SDS/polyacrylamide gel system which was nonreducing in the first dimension and reducing in the second. It separated nascent glycopolypeptides of different length and oxidation state. Evidence was obtained for cotranslational disulfide formation, generation of conformational epitopes, N-linked glycosylation, and oligosaccharide-dependent binding of calnexin, a membrane-bound chaperone that binds to incompletely folded glycoproteins via partially glucose-trimmed oligosaccharides. When glycosylation or oligosaccharide trimming was inhibited, the folding pathway was perturbed, suggesting a role for N-linked oligosaccharides and calnexin during translation of hemagglutinin.

Expression cloning of a cDNA encoding a fish prolactin receptor.

By using an expression cloning strategy, we isolated a single positive clone encoding a tilapia prolactin (PRL) receptor. Tilapia PRL188 was used to screen a freshwater tilapia kidney expression library transfected in COS cells. The tilapia PRL receptor is a mature protein of 606 amino acids. The extracellular domain is devoid of the tandem repeat units present in birds and has two pairs of cysteine residues, a Trp-Ser-Xaa-Trp-Ser motif, and two potential N-glycosylation sites. The cytoplasmic domain contains 372 amino acids, including box 1, a sequence previously shown to be important for signal transduction in mammalian species. Thus, the general structure is similar to the long form of mammalian PRL receptors; however, amino acid comparisons reveal a rather low identity (approximately 37%). Northern blot analysis shows the existence of a single transcript in osmoregulatory tissues and reproductive organs. This localization is in agreement with known functions of PRL in teleosts.

The Lutheran blood group glycoprotein, another member of the immunoglobulin superfamily, is widely expressed in human tissues and is developmentally regulated in human liver.

Glycoproteins expressing the Lutheran blood group antigens were isolated from human erythrocyte membranes and from human fetal liver. Amino acid sequence analyses allowed the design of redundant oligonucleotides that were used to generate a 459-bp, sequence-specific probe by PCR. A cDNA clone of 2400 bp was isolated from a human placental lambda gt 11 library and sequenced, and the deduced amino acid sequence was studied. The predicted mature protein is a type I membrane protein of 597 amino acids with five potential N-glycosylation sites. There are five disulfide-bonded, extracellular, immunoglobulin superfamily domains (two variable-region set and three constant-region set), a single hydrophobic, membrane-spanning domain, and a cytoplasmic domain of 59 residues. The overall structure is similar to that of the human tumor marker MUC 18 and the chicken neural adhesion molecule SC1. The extracellular domains and cytoplasmic domain contain consensus motifs for the binding of integrin and Src homology 3 domains, respectively, suggesting possible receptor and signal-transduction function. Immunostaining of human tissues demonstrated a wide distribution and provided evidence that the glycoprotein is under developmental control in liver and may also be regulated during differentiation in other tissues.

Structural conservation of ion conduction pathways in K channels and glutamate receptors.

Single channel recordings demonstrate that ion channels switch stochastically between an open and a closed pore conformation. In search of a structural explanation for this universal open/close behavior, we have uncovered a striking degree of amino acid homology across the pore-forming regions of voltage-gated K channels and glutamate receptors. This suggested that the pores of these otherwise unrelated classes of channels could be structurally conserved. Strong experimental evidence supports a hairpin structure for the pore-forming region of K channels. Consequently, we hypothesized the existence of a similar structure for the pore of glutamate receptors. In ligand-gated channels, the pore is formed by M2, the second of four putative transmembrane segments. A hairpin structure for M2 would affect the subsequent membrane topology, inverting the proposed orientation of the next segments, M3. We have tested this idea for the NR1 subunit of the N-methyl-D-aspartate receptor. Mutations that affected the glycosylation pattern of the NR1 subunit localize both extremes of the M3-M4 linker to the extracellular space. Whole cell currents and apparent agonist affinities were not affected by these mutations. Therefore it can be assumed that they represent the native transmembrane topology. The extracellular assignment of the M3-M4 linker challenged the current topology model by inverting M3. Taken together, the amino acid homology and the new topology suggest that the pore-forming M2 segment of glutamate receptors does not transverse the membrane but, rather, forms a hairpin structure, similar to that found in K channels.

Incorporation of acetylcholine receptors and Cl- channels in Xenopus oocytes injected with Torpedo electroplaque membranes.

A method was developed to transplant assembled nicotinic acetylcholine receptors (AcChoRs) and Cl- channels from the electric organ of Torpedo to the membrane of Xenopus oocytes. Membrane vesicles from Torpedo electroplaques were injected into the oocytes and, within a few hours, the oocyte membrane acquired AcChoRs and Cl- channels. The mechanism of expression of these receptors and channels is very different from that which follows the injection of mRNA, since the appearance of receptors after membrane injection does not require de novo protein synthesis or N-glycosylation. This, and other controls, indicate that the foreign receptor-bearing membranes fuse with the oocyte membrane and cause the appearance of functional receptors and channels. All this makes the Xenopus oocyte an even more powerful tool for studies of the structure and function of membrane proteins.

Metaxin, a gene contiguous to both thrombospondin 3 and glucocerebrosidase, is required for embryonic development in the mouse: implications for Gaucher disease.

We have identified a murine gene, metaxin, that spans the 6-kb interval separating the glucocerebrosidase gene (GC) from the thrombospondin 3 gene on chromosome 3E3-F1. Metaxin and GC are transcribed convergently; their major polyadenylylation sites are only 431 bp apart. On the other hand, metaxin and the thrombospondin 3 gene are transcribed divergently and share a common promoter sequence. The cDNA for metaxin encodes a 317-aa protein, without either a signal sequence or consensus for N-linked glycosylation. Metaxin protein is expressed ubiquitously in tissues of the young adult mouse, but no close homologues have been found in the DNA or protein data bases. A targeted mutation (A-->G in exon 9) was introduced into GC by homologous recombination in embryonic stem cells to establish a mouse model for a mild form of Gaucher disease. A phosphoglycerate kinase-neomycin gene cassette was also inserted into the 3'-flanking region of GC as a selectable marker, at a site later identified as the terminal exon of metaxin. Mice homozygous for the combined mutations die early in gestation. Since the same amino acid mutation in humans is associated with mild type 1 Gaucher disease, we suggest that metaxin protein is likely to be essential for embryonic development in mice. Clearly, the contiguous gene organization at this locus limits targeting strategies for the production of murine models of Gaucher disease.

Species specificity in the cell-free conversion of prion protein to protease-resistant forms: a model for the scrapie species barrier.

Scrapie is a transmissible neurodegenerative disease that appears to result from an accumulation in the brain of an abnormal protease-resistant isoform of prion protein (PrP) called PrPsc. Conversion of the normal, protease-sensitive form of PrP (PrPc) to protease-resistant forms like PrPsc has been demonstrated in a cell-free reaction composed largely of hamster PrPc and PrPsc. We now report studies of the species specificity of this cell-free reaction using mouse, hamster, and chimeric PrP molecules. Combinations of hamster PrPc with hamster PrPsc and mouse PrPc with mouse PrPsc resulted in the conversion of PrPc to protease-resistant forms. Protease-resistant PrP species were also generated in the nonhomologous reaction of hamster PrPc with mouse PrPsc, but little conversion was observed in the reciprocal reaction. Glycosylation of the PrPc precursors was not required for species specificity in the conversion reaction. The relative conversion efficiencies correlated with the relative transmissibilities of these strains of scrapie between mice and hamsters. Conversion experiments performed with chimeric mouse/hamster PrPc precursors indicated that differences between PrPc and PrPsc at residues 139, 155, and 170 affected the conversion efficiency and the size of the resultant protease-resistant PrP species. We conclude that there is species specificity in the cell-free interactions that lead to the conversion of PrPc to protease-resistant forms. This specificity may be the molecular basis for the barriers to interspecies transmission of scrapie and other transmissible spongiform encephalopathies in vivo.

In vivo assembly of rhodopsin from expressed polypeptide fragments.

Rhodopsin folding and assembly were investigated by expression of five bovine opsin gene fragments separated at points corresponding to proteolytic cleavage sites in the second or third cytoplasmic regions. The CH(1-146) and CH(147-348) gene fragments encode amino acids 1-146 and 147-348 of opsin, while the TH(1-240) and TH(241-348) gene fragments encode amino acids 1-240 and 241-348, respectively. Another gene fragment, CT(147-240), encodes amino acids 147-240. All five opsin polypeptide fragments were stably produced upon expression of the corresponding gene fragments in COS-1 cells. The singly expressed polypeptide fragments failed to form a chromophore with 11-cis-retinal, whereas coexpression of two or three complementary fragments [CH(1-146) + CH(147-348), TH(1-240) + TH(241-348), or CH(1-146) + CT(147-240) + TH(241-348)] formed pigments with spectral properties similar to wild-type rhodopsin. The NH2-terminal polypeptide in these rhodopsins showed a glycosylation pattern characteristic of wild-type COS-1 cell rhodopsin and was noncovalently associated with its complementary fragment(s). Further, the CH(1-146) + CH(147-348) rhodopsin showed substantial light-dependent activation of transducin. We conclude that the functional assembly of rhodopsin is mediated by the association of at least three protein-folding domains.