Cloned β1,4N-acetylgalactosaminyltransferase: subcellular localization and formation of disulfide bonded species

Cloned human 1,4N-acetylgalactosaminyltransferase (GalNAcT) catalyses the synthesis of the glycosphingolipids GM2, GD2, and gangliotriosylceramide. To determine the subcellular location of this enzyme and whether it exists in intermolecular disulfide bonded species, we stably transfected Chinese hamster ovary (CHO) cells with three myc epitope-tagged forms of the GalNAcT gene: the native enzyme; the lumenal domain of GalNAcT fused to the cytoplasmic and transmembrane domains ofN-acetylglucosaminyltransferase I (GNT); and the transmembrane and lumenal domains of GalNAcT fused to the cytoplasmic domain of the Iip33 form of human invariant chain in order to retain the enzyme in the endoplasmic reticulum (ER). Immunoelectron microscopic analysis with anti-myc revealed that GalNAcT/myc was present throughout the Golgi stack, the GNT/GalNAcT/myc form was restricted primarily to the medial Golgi cisternae, and the Iip33/GalNAcT/myc form was restricted to the ER. Cells transfected with each of the three constructs contained high levels of GM2 synthase activityin vitro, but only the GalNAcT/myc form and the GNT/GalNAcT/myc forms were able to synthesize the GM2 productin vivo. The enzyme produced by all three constructs was present in the transfected cells in a disulfide bonded form having a molecular size consistent with that of a homodimer or higher aggregate.Abbreviations GSL glycosphingolipid(s) - CHO Chinese hamster ovary - GSL structures: GM2 GalNAc1,4(NeuAc2,3)Gal1,4GlcCer - GD2 GalNac1,4(NeuAc2,8NeuAc2,3)Gal1,4GlcCer - GM3 NeuAc2,3Gal1,4GlcCer - Gg₃ GalNAc1,4Gal1,4GlcCer - LacCer Gal1,4GlcCer - GlcCer glucosylceramide - PBS-BSA phosphate buffered saline pH 7.4 containing 1% bovine serum albumin - GalNAcT N-acetylgalactosaminyltransferase - GNT N-acetylglucosaminyltransferase I - Iip33 p33 form of human invariant chain - HPTLC high performance thin layer chromatography - PCR polymerase chain reaction - BFA Brefeldin A This paper is dedicated to Professor Sen-itiroh Hakomori on the occasion of his 65th birthday.     

Regulation of ganglioside biosynthesis by enzyme complex formation of glycosyltransferases

Three key regulatory enzymes in ganglioside biosynthesis, sialyltransferase I (ST1), sialyltransferase II (ST2), and N-acetylgalactosaminyltransferase I (GalNAcT), have been expressed as fusion proteins with green, yellow, or red fluorescent protein (GFP, YFP, or RFP) in F-11A cells. F-11A cells are a substrain of murine neuroblastoma F-11 cells that contain only low endogenous ST2 and GalNAcT activity. The subcellular localization of the fusion proteins has been determined by fluorescence microscopy, and the ganglioside composition of these cells was analyzed by high-performance thin-layer chromatography (HPTLC). ST2-GFP (85 kDa) shows a distinct Golgi localization, whereas ST1-YFP (85 kDa) and GalNAcT-RFP (115 kDa) are broadly distributed in ER and Golgi. Untransfected F-11A cells contain mainly GM3, whereas stable transfection with ST2 or GalNAcT results in the predominant expression of b-series complex gangliosides (BCGs). This result indicates that the expression of ST2 enhances the activity of endogenous GalNAcT and vice versa. The specificity of this reaction has been verified by in vitro activity assays with detergent-solubilized enzymes, suggesting the formation of an enzyme complex between ST2 and GalNAcT but not with ST1. Complex formation has also been verified by co-immunoprecipitation of ST2-GFP upon transient transfection with GalNAcT-HA-RFP and by GFP-to-RFP FRET signals that are confined to the Golgi. FRET analysis also suggests that ST2-GFP binds tightly to pyrene-labeled GM3 but not to ST1. We hypothesize that an ST2-GM3 complex is associated with GalNAcT, resulting in the enhanced conversion of GM3 to GD3 and BCGs in the Golgi. Taken together, our results support the concept that ganglioside biosynthesis is tightly regulated by the formation of glycosyltransferase complexes in the ER and/or Golgi.     

Evidence that the entire Golgi apparatus cycles in interphase HeLa cells: sensitivity of Golgi matrix proteins to an ER exit block.

We tested whether the entire Golgi apparatus is a dynamic structure in interphase mammalian cells by assessing the response of 12 different Golgi region proteins to an endoplasmic reticulum (ER) exit block. The proteins chosen spanned the Golgi apparatus and included both Golgi glycosyltransferases and putative matrix proteins. Protein exit from ER was blocked either by microinjection of a GTP-restricted Sar1p mutant protein in the presence of a protein synthesis inhibitor, or by plasmid-encoded expression of the same dominant negative Sar1p. All Golgi region proteins examined lost juxtanuclear Golgi apparatus-like distribution as scored by conventional and confocal fluorescence microscopy in response to an ER exit block, albeit with a differential dependence on Sar1p concentration. Redistribution of GalNAcT2 was more sensitive to low Sar1p(dn) concentrations than giantin or GM130. Redistribution was most rapid for p27, COPI, and p115. Giantin, GM130, and GalNAcT2 relocated with approximately equal kinetics. Distinct ER accumulation could be demonstrated for all integral membrane proteins. ER-accumulated Golgi region proteins were functional. Photobleaching experiments indicated that Golgi-to-ER protein cycling occurred in the absence of any ER exit block. We conclude that the entire Golgi apparatus is a dynamic structure and suggest that most, if not all, Golgi region-integral membrane proteins cycle through ER in interphase cells.     

Formation of HLA-B27 homodimers and their relationship to assembly kinetics

The human HLA-B27 class I molecule exhibits a strong association with the inflammatory arthritic disorder ankylosing spondylitis and other related arthropathies. Major histocompatibility complex class I heavy chains normally associate with beta(2)-microglobulin and peptide in the endoplasmic reticulum before transit to the cell surface. However, an unusual characteristic of HLA-B27 is its ability to form heavy chain homodimers through an unpaired cysteine at position 67 in the peptide groove. Homodimers have previously been detected within the ER and at the cell surface, but their mechanism of formation and role in disease remain undefined. Here we demonstrate, in the rat C58 thymoma cell line and in human HeLa cells transfected with HLA-B27, that homodimer formation involves not only cysteine at position 67 but also the conserved structural cysteine at position 164. We also show that homodimer formation can be induced in the non-disease-associated HLA class I allele HLA-A2 by slowing its assembly rate by incubation of cells at 26 degrees C, suggesting that homodimer formation in the endoplasmic reticulum may occur as a result of the slower folding kinetics of HLA-B27. Finally, we report an association between unfolded HLA-B27 molecules and immunoglobulin-binding protein at the cell surface.     

Oligomerization, transport, and Golgi retention of Punta Toro virus glycoproteins.

We have investigated the oligomerization and intracellular transport of the membrane glycoproteins of Punta Toro virus, a member of the Phlebovirus genus of the family Bunyaviridae, which is assembled by budding in the Golgi complex. By using one- or two-dimensional sodium dodecyl sulfate-polyacrylamide gel electrophoresis, chemical cross-linking, and sucrose gradient centrifugation, we found that the majority of the G1 and G2 glycoproteins are assembled into noncovalently linked G1-G2 heterodimers. At the same time, a fraction of the G2 protein, possibly produced independently of the G1 protein, is assembled into G2 homodimers. Kinetic analysis indicates that heterodimerization occurs between newly synthesized G1 and G2 within 3 min after protein synthesis, and that the G1 and G2 glycoproteins are associated as dimeric forms both during transport and after accumulation in the Golgi complex. Analysis of a G1-truncated G2 mutant, which is also targeted to the Golgi complex, showed that these molecules also assemble into dimeric forms, which are linked by disulfide bonds. Both the G1-G2 heterodimer and the G2 homodimer were found to be able to exit from the endoplasmic reticulum. Differences in transport kinetics observed for the G1 and G2 proteins may be due to the differences in the transport efficiency between the G1-G2 heterodimer and the G2 homodimer from the endoplasmic reticulum to the Golgi complex. These and previous results (S.-Y. Chen, Y. Matsuoka, and R.W. Compans, Virology 183:351-365, 1991) suggest that Golgi retention of the G2 homodimer occurs by association with the G1-G2 heterodimer, whereas the Golgi targeting of the G1-G2 heterodimer occurs by a specific retention mechanism.     

Two soluble glycosyltransferases glycosylate less efficiently in vivo than their membrane bound counterparts

Many Golgi glycosyltransferases are type II membrane proteins which are cleaved to produce soluble forms that are released from cells. Cho and Cummings recently reported that a soluble form of alpha1, 3- galactosyltransferase was comparable to its membrane bound counterpart in its ability to galactosylate newly synthesized glycoproteins (Cho,S.K. and Cummings,R.D. (1997) J. Biol. Chem., 272, 13622-13628). To test the generality of their findings, we compared the activities of the full length and soluble forms of two such glycosyltransferases, ss1,4 N-Acetylgalactosaminyltransferase (GM2/GD2/ GA2 synthase; GalNAcT) and beta galactoside alpha2,6 sialyltransferase (alpha2,6-ST; ST6Gal I), for production of their glycoconjugate products in vivo . Unlike the full length form of GalNAcT which produced ganglioside GM2 in transfected cells, soluble GalNAcT did not produce detectable GM2 in vivo even though it possessed in vitro GalNAcT activity comparable to that of full length GalNAcT. When compared with cells expressing full length alpha2,6-ST, cells expressing a soluble form of alpha2,6-ST contained 3-fold higher alpha2,6-ST mRNA levels and secreted 7-fold greater alpha2,6-ST activity as measured in vitro , but in striking contrast contained 2- to 4-fold less of the alpha2,6-linked sialic acid moiety in cellular glycoproteins in vivo . In summary these results suggest that unlike alpha1,3-galactosyltransferase the soluble forms of these two glycosyltransferases are less efficient at glycosylation of membrane proteins and lipids in vivo than their membrane bound counterparts.      

Spatial Organization and Stoichiometry of N-Terminal Domain-Mediated Glycosyltransferase Complexes in Golgi Membranes Determined by Fret Microscopy

The functional link between glycolipid glycosyltransferases (GT) relies on the ability of these proteins to form organized molecular complexes. The organization, stoichiometry and composition of these complexes may impact their sorting properties, sub-Golgi localization, and may determine relative efficiency of GT in different glycolipid biosynthetic pathways. In this work, by using Förster resonance energy transfer microscopy in live CHO-K1 cells, we investigated homo- and hetero-complex formation by different GT as well as their spatial organization and molecular stoichiometry on Golgi membranes. We find that GalNAcT and GalT2 Ntd are able to form hetero-complexes in a 1:2 molar ratio at the trans-Golgi network and that GalT2 but not GalNAcT forms homo-complexes. Also, GalNAcT/GalT2 complexes exhibit a stable behavior reflected by its clustered lateral organization. These results reveals that particular topological organization of GTs may have functional implications in determining the composition of glycolipids in cellular membranes.     

The C-terminal sequence of LMADS1 is essential for the formation of homodimers for B function proteins

LMADS1, a lily (Lilium longiflorum) AP3 orthologue, contains the complete consensus sequence of the paleoAP3 (YGSHDLRLA) and PI-derived (YEFRVQPSQPNLH) motifs in the C-terminal region of the protein. Interestingly, through yeast two-hybrid analysis, LMADS1 was found to be capable of forming homodimers. These results indicated that LMADS1 represents an ancestral form of the B function protein, which retains the ability to form homodimers in regulating petal and stamen development in lily. To explore the involvement of the conserved motifs in the C-terminal region of LMADS1 in forming homodimers, truncated forms of LMADS1 were generated, and their ability to form homodimers was analyzed using yeast two-hybrid and electrophoretic mobility shift assay. The ability of LMADS1 to form homodimers decreased once the C-terminal paleoAP3 motif was deleted. When both paleoAP3 and PI-derived motifs were deleted, the ability of LMADS1 to form homodimers was completely abolished. This result indicated that although the paleoAP3 motif promotes the formation of LMADS1 homodimers, the PI-derived motif is essential. Deletion analysis indicated that two amino acids, RV, of the 5 final amino acids, YEFRV, in the PI-derived motif are essential for the formation of homodimers. Further, point mutation analysis indicated that amino acid Val was absolutely necessary, whereas residue Arg played a less important role in the formation of homodimers. Furthermore, Arabidopsis AP3 was able to form homodimers once its C-terminal region was replaced by that of LMADS1. This result indicated that the C-terminal region of LMADS1 is responsible and essential for homodimer formation of the ancestral form of the B function protein.     

Distinct protein domains of the yeast Golgi GDP-mannose transporter mediate oligomer assembly and export from the endoplasmic reticulum

The substrates for glycan synthesis in the lumen of the Golgi are nucleotide sugars that must be transported from the cytosol by specific membrane-bound transporters. The principal nucleotide sugar used for glycosylation in the Golgi of the yeast Saccharomyces cerevisiae is GDP-mannose, whose lumenal transport is mediated by the VRG4 gene product. As the sole provider of lumenal mannose, the Vrg4 protein functions as a key regulator of glycosylation in the yeast Golgi. We have undertaken a functional analysis of Vrg4p as a model for understanding nucleotide sugar transport in the Golgi. Here, we analyzed epitope-tagged alleles of VRG4. Gel filtration chromatography and co-immunoprecipitation experiments demonstrate that the Vrg4 protein forms homodimers with specificity and high affinity. Deletion analyses identified two regions essential for Vrg4p function. Mutant Vrg4 proteins lacking the predicted C-terminal membrane-spanning domain fail to assemble into oligomers (Abe, M., Hashimoto, H., and Yoda, K. (1999) FEBS Lett. 458, 309-312) and are unstable, while proteins lacking the N-terminal cytosolic tail are stable and multimerize efficiently, but are mislocalized to the endoplasmic reticulum (ER). Fusion of the N terminus of Vrg4p to related ER membrane proteins promote their transport to the Golgi, suggesting that sequences in the N terminus supply information for ER export. The dominant negative phenotype resulting from overexpression of truncated Vrg4-DeltaN proteins provides strong genetic evidence for homodimer formation in vivo. These studies are consistent with a model in which Vrg4p oligomerizes in the ER and is subsequently transported to the Golgi via a mechanism that involves positive sorting rather than passive default.     

Structural features differentiate the mechanisms between 2S (2 state) and 3S (3 state) folding homodimers

The formation of homodimer complexes for interface stability, catalysis and regulation is intriguing. The mechanisms of homodimer complexations are even more interesting. Some homodimers form without intermediates (two-state (2S)) and others through the formation of stable intermediates (three-state (3S)). Here, we analyze 41 homodimer (25 2S and 16 3S) structures determined by X-ray crystallography to estimate structural differences between them. The analysis suggests that a combination of structural properties such as monomer length, subunit interface area, ratio of interface to interior hydrophobicity can predominately distinguish 2S and 3S homodimers. These findings are useful in the prediction of homodimer folding and binding mechanisms using structural data.     

Involvement of endoplasmic reticulum in the origins of dictyosome-like structures of guinea pig spermatocytes

Summary Dictyosome-like structures (DLS) are formed in early spermatocytes first as single saccules. These saccules occur in association with forms of endoplasmic reticulum (ER) characterized by a paucity of ribosomes and luminal content, by a constriction of the lumina, and by a tendency to fragment or form myelin figures during fixation. Nascent DLS and the unusual ER cisternae share many characteristics in common including a pattern of staining with fixatives containing tannic acid where the membranes appear thin due to the inner membrane leaflet being unstained or poorly stained. DLS also appear to form in the region conventional Golgi apparatus but always in association with ER forms that frequently occupy portions of the Golgi apparatus zone.An ability to stain with phosphotungstic acid at low pH exhibited by DLS is given also by the specialized ER forms. One possibility for DLS formation suggested by the present study is that DLS cisternae differentiate from ER membranes after which they ultimately associate into the stacked configurations that characterize mature DLS.     

A perspective on mechanisms of protein tetramer formation

Homotetrameric proteins can assemble by several different pathways, but have only been observed to use one, in which two monomers associate to form a homodimer, and then two homodimers associate to form a homotetramer. To determine why this pathway should be so uniformly dominant, we have modeled the kinetics of tetramerization for the possible pathways as a function of the rate constants for each step. We have found that competition with the other pathways, in which homotetramers can be formed either by the association of two different types of homodimers or by the successive addition of monomers to homodimers and homotrimers, can cause substantial amounts of protein to be trapped as intermediates of the assembly pathway. We propose that this could lead to undesirable consequences for an organism, and that selective pressure may have caused homotetrameric proteins to evolve to assemble by a single pathway.     

Structural basis for membrane binding specificity of the Bin/Amphiphysin/Rvs (BAR) domain of Arfaptin-2 determined by Arl1 GTPase

Membrane-sculpting BAR (Bin/Amphiphysin/Rvs) domains form a crescent-shaped homodimer that can sense and induce membrane curvature through its positively charged concave face. We have recently shown that Arfaptin-2, which was originally identified as a binding partner for the Arf and Rac1 GTPases, binds to Arl1 through its BAR domain and is recruited onto Golgi membranes. There, Arfaptin-2 induces membrane tubules. Here, we report the crystal structure of the Arfaptin-2 BAR homodimer in complex with two Arl1 molecules bound symmetrically to each side, leaving the concave face open for membrane association. The overall structure of the Arl1·Arfaptin-2 BAR complex closely resembles that of the PX-BAR domain of sorting nexin 9, suggesting similar mechanisms underlying BAR domain targeting to specific organellar membranes. The Arl1·Arfaptin-2 BAR structure suggests that one of the two Arl1 molecules competes with Rac1, which binds to the concave face of the Arfaptin-2 BAR homodimer and may hinder its membrane association.     

Visualization of APP dimerization and APP-Notch2 heterodimerization in living cells using bimolecular fluorescence complementation

We previously demonstrated that the amyloid precursor protein (APP) interacts with Notch receptors. Here, we confirmed the APP/Notch1 endogenous interaction in embryonic day 17 rat brain tissue, suggesting the interaction was not as a result of over-expression artifacts. To investigate potential homodimeric and heterodimeric interactions of APP and Notch2 (N2), we have visualized the subcellular localization of the APP/N2 complexes formed in living cells using bimolecular fluorescence complementation (BiFC) analysis. BiFC was accomplished by fusing the N-terminal fragment or the C-terminal fragment of yellow fluorescent protein (YFP) to APP, N2, and a C-terminally truncated form of N2. When expressed in COS-7 cells, these tagged proteins alone did not produce a fluorescent signal. The tagged APP homodimer produced a weak fluorescent signal, while neither full-length N2, nor a truncated N2 alone, produced a visible signal, suggesting that N2 receptors do not form homodimers. The strongest fluorescent signal was obtained with co-expression of the C-terminal fragment of YFP fused to APP and the N-terminal fragment of YFP fused to the truncated form of N2. This heterodimer localized to plasma membrane, endoplasmic reticulum (ER), Golgi and other compartments. The results were confirmed and quantified by flow cytometry. The BiFC method of specifically visualizing APP/Notch interactions can be applied to study APP and Notch signaling during development, aging and neurodegeneration.     

Inactivation of human fibroblast growth factor-1 (FGF-1) activity by interaction with copper ions involves FGF-1 dimer formation induced by copper-catalyzed oxidation.

Although the angiogenic proteins acidic fibroblast growth factor (FGF-1) and basic fibroblast growth factor (FGF-2) both interact with the transition metal copper, itself a putative modulator of angiogenesis, a role for copper in FGF function has not been established. Using nonreducing sodium dodecyl sulfate-polyacrylamide gel electrophoresis, we detect the complete conversion of recombinant forms of human FGF-1 monomer protein to FGF-1 homodimers after exposure to copper ions. In contrast, not all forms of bovine FGF-1 isolated from bovine brain or a recombinant preparation of human FGF-2 completely formed homodimers after exposure to copper ions under similar conditions. Since the copper-induced FGF-1 homodimers reverted to the monomer form in the presence of dithiothreitol, specific alkylation of cysteine residues by pyridylethylation prevented FGF-1 homodimer formation, and preformed FGF-1 homodimers could not be dissociated by the metal chelator EDTA, FGF-1 dimer formation appeared to result from the formation of intermolecular disulfide bonds by copper-induced oxidation of sulfhydryl residues. FGF-1 homodimers bound with similar apparent affinity as FGF-1 monomers to immobilized copper ions, both eluting at 60 mM imidazole. Both human FGF-1 monomer and dimer forms had a 6-fold higher apparent affinity for immobilized copper ions, as compared with human FGF-2, which eluted in the monomer form at 10 mM imidazole. Further, in contrast to FGF-1 monomers, which dissociate from immobilized heparin in 1.0 M NaCl, preformed FGF-1 homodimers had reduced apparent affinity for immobilized heparin and eluted at 0.4 M NaCl. In contrast, the apparent affinity of human FGF-2 for immobilized heparin was unaffected after exposure to copper ions. Heparin appeared to modulate the formation of copper-induced intermolecular disulfide bonds for FGF-1 but not FGF-2, since co-incubation of heparin and copper with FGF-1 monomers resulted in dimers and other oligomeric complexes. FGF-1 copper-induced homodimers failed to induce mitogenesis in [3H]thymidine incorporation assays, an effect which could be reversed by treatment with dithiothreitol, whereas FGF-2-induced mitogenic activity was relatively unaffected by pretreatment with copper. The differences between human FGF-1 and FGF-2 in protein-copper interactions may be due to differing free thiol content and arrangement between the two proteins. A recombinant human FGF-1 mutant containing the two cysteines conserved throughout the FGF family of proteins but lacking a cysteine residue (Cys 131) present in wild-type human FGF-1 but not human FGF-2 readily formed copper-induced dimers.(ABSTRACT TRUNCATED AT 400 WORDS)     

Sec23p and a novel 105-kDa protein function as a multimeric complex to promote vesicle budding and protein transport from the endoplasmic reticulum.

A cell-free protein transport reaction has been used to monitor the purification of a functional form of the Sec23 protein, a SEC gene product required for the formation or stability of protein transport vesicles that bud from the endoplasmic reticulum (ER). Previously, we reported that Sec23p is an 84-kDa peripheral membrane protein that is released from a sedimentable fraction by vigorous mechanical agitation of yeast cells and is required for ER to Golgi transport assayed in vitro. We have purified soluble Sec23p by complementation of an in vitro ER to Golgi transport reaction reconstituted with components from sec23 mutant cells. Sec23p overproduced in yeast exists in two forms: a monomeric species and a species that behaves as a 250- to 300-kDa complex that contains Sec23p and a distinct 105-kDa polypeptide (p105). Sec23p purified from cells containing one SEC23 gene exists solely in the large multimeric form. A stable association between Sec23p and p105 is confirmed by cofractionation of the two proteins throughout the purification. p105 is a novel yeast protein involved in ER to Golgi transport. Like Sec23p, it is required for vesicle budding from the ER because p105 antiserum completely inhibits transport vesicle formation in vitro.     

A direct role for GRASP65 as a mitotically regulated Golgi stacking factor

Cell-free assays that mimic the disassembly and reassembly cycle of the Golgi apparatus during mitosis implicated GRASP65 as a mitotically regulated stacking factor. We now present evidence that GRASP65 is directly involved in stacking Golgi cisternae. GRASP65 is the major phosphorylation target in rat liver Golgi membranes of two mitotic kinases, cdc2-cyclin B and polo-like kinases, which alone will unstack Golgi membranes, generating single cisternae. Mitotic cells microinjected with antibodies to GRASP65 fail to form proper Golgi stacks after cell division. Beads coated with GRASP65 homodimers form extensive aggregates consistent with the formation of trans oligomers. These can be disaggregated using purified cdc2-cyclin B1 and polo-like kinases, and re-aggregated after dephosphorylation of GRASP65. Together, these data demonstrate that GRASP65 has the properties required to bind surfaces together in a mitotically regulated manner.     

A major part of platelet-derived growth factor purified from human platelets is a heterodimer of one A and one B chain

Platelet-derived growth factor, PDGF, purified from human platelets is a disulfide-bonded dimer consisting of two homologous polypeptide chains denoted A and B; it has not been known whether it is a heterodimer or a mixture of homodimers. We present here evidence that a major part of PDGF has a heterodimer structure. A highly homogeneous, 31-kDa PDGF was purified in the presence of protease inhibitors and shown to contain both chains by means of immunoprecipitations with peptide antisera specific for the A and B chains, respectively. The susceptibility of PDGF to mild acid treatment and its chromatographic behavior in reversed-phase high performance liquid chromatography and immobilized metal ion affinity chromatography, as compared to A and B chain homodimers, is consistent with a heterodimer structure. Analysis of PDGF purified according to our routine, large scale procedure revealed the major part to have a heterodimer structure. In addition, B chain homodimers were also found. With the demonstration that a major part of PDGF purified from human platelets occurs as a heterodimer, all three dimeric forms of PDGF have been identified. The following nomenclature to distinguish the various forms is suggested: PDGF-AA, a homodimer of A chains; PDGF-AB, a heterodimer; PDGF-BB, a homodimer of B chains; PDGF, any dimeric form of A or B chains.