The p53-independent tumoricidal activity of an adenoviral vector encoding a p27-p16 fusion tumor suppressor gene

We describe here that DE1-adenovirus vectors (AV) expressing a p27-p16 fusion molecule, termed W9, induce tumor cell apoptosis when overexpressed in a wide range of tumor cell types. However, in primary human cells derived from a variety of normal tissues, AV-W9 induced minimal apoptosis. In tumor cells AV-W9 demonstrated 5- to 50-fold greater tumoricidal activity than either of the parental molecules p16 and p27. In these studies, AV-W9 elicited apoptosis independent of the p53 and Rb status of the tumor cells. In several murine tumor models AV-W9 demonstrated p53-independent antitumor activity. It completely prevented tumor formation in two ex vivo models, whereas the parental molecules resulted in partial protection. Furthermore, AV-W9 induced tumor regression or suppressed tumor growth when introduced intratumorally into preestablished tumors in mice. This effect may be mediated through tumor cell apoptosis or antiangiogenic activity of AV-W9. Thus, this novel chimeric molecule is more potent and capable of killing a broader spectrum of tumors than the parental p16 and p27 molecules independent of the tumor cell p53 and phenotype and represents a powerful new therapeutic agent for cancer gene therapy.

Epitope mapping of human anti-adeno-associated virus type 2 neutralizing antibodies: implications for gene therapy and virus structure

Recombinant adeno-associated virus type 2 (AAV) is a common vector used in human gene therapy protocols. We characterized the humoral immune response to AAV and observed that 80% of normal human subjects have anti-AAV antibodies and that 18% have neutralizing antibodies. To analyze the effect of neutralizing antibodies on AAV readministration, we attempted to deliver recombinant AAV expressing human factor IX (AAV-hFIX) intraportally into the livers of mice which had been preexposed to AAV and shown to harbor a neutralizing antibody response. While all naive control mice expressed hFIX following administration of AAV-hFIX, none of the mice with preexisting immunity expressed hFIX, even after transient immunosuppression at the time of the second administration with anti-CD4 or anti-CD40L antibodies. This suggests that preexisting immunity to AAV, as measured by a neutralizing antibody response, may limit AAV-mediated gene delivery. Using human sera in an enzyme-linked immunosorbent assay for AAV and a capsid peptide scan library to block antibody binding, we mapped seven regions of the AAV capsid containing immunogenic epitopes. Using pools of these peptides to inhibit the binding of neutralizing antibodies, we have identified a subset of six peptides which potentially reconstitute a single neutralizing epitope. This information may allow the design of reverse genetic approaches to circumvent the preexisting immunity that can be encountered in some individuals.

Determination of glutamic acid decarboxylase 65 peptides presented by the type I diabetes-associated HLA-DQ8 class II molecule identifies an immunogenic peptide motif

Particular HLA class II allelic sequences are associated with susceptibility to type I diabetes. To understand the mechanism, knowledge of the molecular nature of the specific TCR/peptide/class II interactions involved in the disease process is required. To this end, we have introduced the diabetes-associated human class II HLA-DQ8 allele (DQA1*0301/DQB1*0302) as a transgene into mice and analyzed T cell responses restricted by this molecule to an important Ag in human diabetes, human glutamic acid decarboxylase 65. Hybridomas were used to determine the particular peptides from this Ag presented by HLA-DQ8 to T cells and to map the core minimal epitopes required for T cell stimulation. Analysis of these core epitopes reveals a motif and relevant features for peptides that are immunogenic to T cells when presented by HLA-DQ8. The major immunogenic epitopes of glutamic acid decarboxylase 65 do not contain a negatively charged residue that binds in the P9 pocket of the HLA-DQ8 molecule. PBMC from HLA-DQ8+ diabetic and nondiabetic individuals respond to these peptides, confirming that the mouse model is a useful tool to define epitopes of autoantigens that are processed by human APC and recognized by human T cells.

T cell responses to a human cartilage autoantigen in the context of rheumatoid arthritis-associated and nonassociated HLA-DR4 alleles

OBJECTIVE

To analyze the CD4+ T cell responses to the human cartilage antigen glycoprotein-39 (HCgp-39) in the context of rheumatoid arthritis (RA)-associated (DRalphabeta1*0401) and nonassociated (DRalphabeta1*0402) HLA class II molecules.

METHODS

Large numbers of HCgp-39-specific T cell hybridomas were generated following immunization of HLA-DR4/human CD4 transgenic, murine major histocompatibility complex class II deficient mice with native HCgp-39. Fine epitope mapping of DRalphabeta1*0401-and DRalphabeta1*0402-restricted T cell hybridomas was performed using overlapping synthetic peptides. Antigen-specific cytokine production by lymph node T cells was evaluated after immunization with native antigen. Proliferative T cell responses of healthy human subjects were compared with the T cell responses of patients with active RA using HCgp-39 epitopes defined in HLA-DR4 transgenic mice.

RESULTS

CD4+ T cells from DRalphabeta1*0401 and DRalphabeta1*0402 transgenic mice identified completely different immunodominant peptide epitopes of HCgp-39, and this was not explained by known DR4-binding motifs or direct peptide-binding studies. DRalphabeta1*0401-restricted, antigen-specific T cells produced significantly more interferon-gamma and tumor necrosis factor a in response to HCgp-39 than did T cells from DRalphabeta1*0402 transgenic mice. Finally, HCgp-39 peptides defined in DRalphabeta1*0401 transgenic mice stimulated T cells from HLA-DR4 positive human subjects and RA patients, but not T cells from HLA-DR4 negative individuals.

CONCLUSION

T cell epitopes of HCgp-39 that were defined in HLA-DR4 transgenic mice stimulated T cells from human subjects carrying RA-associated HLA-DR4 alleles. HLA-DR4 molecules may influence the disease process in RA both by presentation of selected peptide epitopes and by promoting the production of proinflammatory cytokines in synovial joints.

Impact of chimeric immune receptor extracellular protein domains on T cell function

Chimeric immune receptors (CIR) encompass tumor- or virus-specific ligands or antibodies fused to the signaling domains of either the T cell receptor or Fc receptor. T cells expressing these receptors recapitulate the cytopathic effects mediated by the T cell receptor and allow the targeting of tumor or virus infected cells in an MHC-independent manner. With this technology, large numbers of T cells with redirected target specificity can be generated. To define the structural features of recombinant CIRs required for optimal function, a panel of five closely related CIRs with identical target specificity were generated. These receptors recognized HIVenv through the single chain Fv (scFv) of an anti-gp 120 antibody. These scFv-zeta receptors were constructed to include alternative extracellular spacer and transmembrane protein domains derived from members of the immunoglobulin supergene family. The effect of these alternative extracellular protein domains on receptor stability, antigen affinity and T cell activity was assessed. We demonstrate that modifying the extracellular protein domains of the anti-HIVenv CIRs significantly impacted receptor stability and substrate binding affinity and that these effects, and not simply the level of cell surface expression, correlated most strongly with changes in CIR-mediated killing. These studies will aid in the rationale design of recombinant CIRs for the immunotherapy of viral infections, cancer and other diseases.

Anti-tumor activity of human T cells expressing the CC49-zeta chimeric immune receptor

A chimeric immune receptor consisting of an extracellular antigen-binding domain derived from the CC49 humanized single-chain antibody, linked to the CD3zeta signaling domain of the T cell receptor, was generated (CC49-zeta). This receptor binds to TAG-72, a mucin antigen expressed by most human adenocarcinomas. CC49-zeta was expressed in CD4+ and CD8+ T cells and induced cytokine production on stimulation. Human T cells expressing CC49-zeta recognized and killed tumor cell lines and primary tumor cells expressing TAG-72. CC49-zeta T cells did not mediate bystander killing of TAG-72-negative cells. In addition, CC49-zeta T cells not only killed FasL-positive tumor cells in vitro and in vivo, but also survived in their presence, and were immunoprotective in intraperitoneal and subcutaneous murine tumor xenograft models with TAG-72-positive human tumor cells. Finally, receptor-positive T cells were still effective in killing TAG-72-positive targets in the presence of physiological levels of soluble TAG-72, and did not induce killing of TAG-72-negative cells under the same conditions. This approach is being currently being utilized in a phase I clinical trial for the treatment of colon cancer.

Sodium multiple quantum spectroscopy of articular cartilage: effects of mechanical compression

The effects of mechanical compression on the multiple quantum coherences generated from sodium ions in articular cartilage were investigated. Cartilage samples obtained from bovine patellae were studied during compression at 0.7 MPa (100 psi) for 1 hour. The double quantum filtered spectra showed marked lineshape changes in the compressed samples. Compression did not seem to influence the lineshapes of the single quantum and triple quantum filtered spectra significantly. We found that the residual quadrupolar interaction was reduced in the compressed samples. Changes in the ordering of collagen fibers may be responsible for the observed effect.

Characterization of hematopoietic cell expansion, oxygen uptake, and glycolysis in a controlled, stirred-tank bioreactor system

Cultures of umbilical cord blood and mobilized peripheral blood mononuclear cells were carried out in a stirred bioreactor with pH and dissolved oxygen control. Expansion of total cells and colony-forming units granulocyte-macrophage was greatly enhanced by the use of a cell-dilution feeding protocol (as compared to a cell-retention feeding protocol). The specific oxygen consumption rate (qO2) for these cultures ranged from 1.7 x 10(-8) to 1.2 x 10(-7) micromol/(cell.h). The maximum in qO2 for each culture closely corresponded with the maximum percentage of progenitor or colony-forming cells (CFCs) present in the culture. The maximum qO2 values are slightly less than those reported for hybridomas, while the lowest qO2 values are somewhat greater than those reported for mature granulocytes. Examination of the ratio of lactate production to oxygen consumption in these cultures suggests that post-progenitor cells of the granulomonocytic lineage obtain a greater portion of their energy from glycolysis than do CFCs. The different metabolic profiles of CFCs and more mature cells suggest that monitoring the uptake or production of oxygen, lactate, and other metabolites will allow estimation of the content of several cell types in culture.

T1rho-relaxation in articular cartilage: effects of enzymatic degradation

Spin-lattice relaxation in the rotating frame (T1rho) dispersion spectroscopy and imaging were used to study normal and enzymatically degraded bovine articular cartilage. Normal specimens demonstrate significant T1rho "dispersion" (approximately 60 to approximately 130 ms) in the 100 Hz to 9 kHz frequency range. Proteoglycan-degraded specimens have 33% greater T1rho values than collagen-degraded or normal samples. T1rho-weighted images reveal structure not found in conventional T1- or T2-weighted images. Our results suggest that T1rho measurements are selectively sensitive to proteoglycan content. The potential of this method in distinguishing the early degenerative changes in cartilage associated with osteoarthritis is discussed.

Identification of immunodominant T cell epitopes of human glutamic acid decarboxylase 65 by using HLA-DR(alpha1*0101,beta1*0401) transgenic mice

Glutamic acid decarboxylase isoform 2 (GAD65; EC 4.1.1.15) has been identified as a key target autoantigen of insulin-dependent diabetes mellitus (IDDM). IDDM is genetically associated with the major histocompatibility complex (MHC), and particular alleles from the HLA-DQ and HLA-DR loci contribute to disease. Among DR4 subtypes, HLA-DRB1*0401, HLA-DRB1*0402, and HLA-DRB1*0405 alleles lend susceptibility, while HLA-DRB1*0403 confers protection. We have utilized HLA-DR(alpha1*0101,beta1*0401) (hereafter referred to as DR0401), human CD4, murine class II null triple transgenic mice and recombinant GAD65 to generate T cell hybridomas, and we have used overlapping sets of peptides to map the immunodominant epitopes of this autoantigen. We have identified 10 immunogenic regions for GAD65, of which 6 are recognized by multiple hybridomas. These epitopes are also generated by human antigen-presenting cells and their presentation is DR0401 restricted, as shown by the use of typed human lymphoblastoid cell lines and antibody blocking experiments. Immunodominant GAD65 epitopes defined in transgenic mice correspond to GAD65 regions previously shown to elicit T cell responses specifically in DR0401 IDDM patients, underscoring the validity of this approach. Interestingly, although the major epitopes contain DR0401 binding motifs, one of the epitopes contains a DR0405 motif.

Delineation of the subunit composition of human proteasomes using antisera against the major histocompatibility complex-encoded LMP2 and LMP7 subunits

The products of the Lmp2 and Lmp7 genes located in the major histocompatibility complex (MHC) class II region are postulated to form part of the proteasome complex. This large, multisubunit complex forms the major, nonlysosomal proteolytic machinery for the degradation of endogenous proteins and has been suggested to play a role in the processing of antigens presented by MHC class I molecules. The role of the MHC-encoded subunits within the proteasome has however remained enigmatic. To study this role, we have raised antibodies to recombinant LMP2 and LMP7 proteins. Under denaturing conditions, the anti-LMP7 serum recognizes one subunit of proteasome, whereas the anti-LMP2 serum recognizes two subunits, which may represent different forms of LMP2. The specificity of these sera has been ascertained by a lack of reactivity in T2 cells, which lack both genes. Furthermore under native conditions the anti-LMP2 serum immunoprecipitates a complex that is similar to proteasome but lacks several subunits, including LMP7. Preclearing experiments using this serum and a monoclonal antibody (2-17) specific for the non-MHC-encoded C2 proteasome subunit demonstrate that the complexes recognized by these two sera are distinct and that four subunits are unique to the complex precipitated by the anti-LMP2 serum. Interestingly, the different forms of LMP2 are segregated between these complexes. The relationship of the two complexes is discussed.

Major histocompatibility complex-encoded antigen processing gene polymorphism in IDDM

Susceptibility to insulin-dependent diabetes mellitus (IDDM) is greatly influenced by polymorphisms in the genes of the class II region of the human leukocyte antigen (HLA) complex. The complexity of this genetic association and the lack of a direct proof of involvement of HLA class II genes in human IDDM have continued to support speculation on a possible role of genes encoded in the close vicinity of these loci in IDDM. Because the recently discovered transporter associated with antigen processing (TAP) and large multifunctional protease (LMP) genes are encoded in the HLA class II region and are implicated in the processing of antigenic proteins for presentation by HLA class I molecules, they are additional candidates for a role in IDDM pathogenesis. We have analyzed genomic and coding sequence polymorphisms in the LMP2, TAP1, and TAP2 genes of 77 Danish IDDM patients and 102 control subjects. Although patients and control subjects did not differ in TAP1 and LMP2 alleles, we found a striking absence of the TAP2 allele B (long form) in IDDM patients. An analysis of the TAP2 alleles in individual DR types, however, revealed that this phenomenon is likely to be caused by linkage disequilibrium between the two loci. Thus, polymorphisms in the TAP and LMP genes are unlikely to be associated with IDDM.

Intrahepatic cholestasis during nicotinic acid therapy

BACKGROUND

Nicotinic acid, widely used to lower serum cholesterol levels, may rarely cause cholestatic jaundice.

SUMMARY

A 61-year-old white man with hypercholesterolemia complained of marked pruritus and became jaundiced after taking 3.0 g of crystalline nicotinic acid daily for 13 months. His total serum bilirubin level was increased at 144 mumol/L (8.4 mg/dL) and his alkaline phosphatase level was markedly elevated at 35.00 mukat/L (2100 U/L). Endoscopic retrograde cholangiopancreatography failed to demonstrate an obstructive lesion in the extrahepatic biliary system, computed tomography showed no intrahepatic dilatation, and ultrasonographic studies of the liver, gallbladder, and pancreas were normal; these factors all suggest intrahepatic cholestasis. Symptoms improved and liver function test results returned to normal within 51 days after stopping the drug.

CONCLUSIONS

Nicotinic acid-induced cholestatic jaundice may not be as rare as previously thought, and physicians should observe their patients for it.

Analyses of skin surface lipid in patients with microbially associated skin disease

The composition of the total skin surface lipid, free fatty acids and fatty acid of triglyceride was used in discriminant analyses to classify patients with severe acne vulgaris (AV), pityriasis versicolor (PV), seborrhoeic dermatitis (SD) and atopic dermatitis (AD). It was not possible to discriminate between the PV and SD patients but all the lipid groups gave a satisfactory separation between AV, PV + SD and AD patients. A similar separation was obtained when the sex of the patient was studied but not when both sex and disease were taken into account. The results suggest that there are fundamental differences in the composition of skin lipid between patients with microbially associated diseases, but the precise role of the organisms has not been ascertained.

Antigen processing gene polymorphisms in HLA-DR2 multiple sclerosis

The association between multiple sclerosis (MS) and alleles of the HLA class II genes indicates that at least one MS susceptibility gene is linked to the HLA class II region. However, the actual locus responsible has not been precisely identified. The recent cloning of new genes involved in antigen processing that map within the HLA class II region led us to investigate--using the restriction fragment length polymorphism (RFLP) technique and sequence-specific oligonucleotide analysis--whether these genes might play a role in conferring susceptibility to MS. We studied large multifunctional protease (LMP) 2 and 7 and transporter associated with antigen processing (TAP) 1 and 2 gene polymorphisms in 60 HLA-DR2 MS patients and 60 HLA-DR2 healthy subjects and found no specific or preferential RFLP patterns or coding sequence variants in the patient group. Our data do not support a role for these genes in MS susceptibility.

Genomic polymorphism, recombination, and linkage disequilibrium in human major histocompatibility complex-encoded antigen-processing genes

Recently, two subunits of a large cytosolic protease and two putative peptide transporter proteins were found to be encoded by genes within the class II region of the major histocompatibility complex (MHC). These genes have been suggested to be involved in the processing of antigenic proteins for presentation by MHC class I molecules. Because of the high degree of polymorphism in MHC genes, and previous evidence for both functional and polypeptide sequence polymorphism in the proteins encoded by the antigen-processing genes, we tested DNA from 27 consanguineous human cell lines for genomic polymorphism by restriction fragment length polymorphism (RFLP) analysis. These studies demonstrate a strong linkage disequilibrium between TAP1 and LMP2 RFLPs. Moreover, RFLPs, as well as a polymorphic stop codon in the telomeric TAP2 gene, appear to be in linkage disequilibrium with HLA-DR alleles and RFLPs in the HLA-DO gene. A high rate of recombination, however, seems to occur in the center of the complex, between the TAP1 and TAP2 genes.

The asparaginyl-tRNA synthetase gene encodes one of the complementing factors for thermosensitive translation in the Escherichia coli mutant strain, N4316

Escherichia coli strain N4316 is a mutant that exhibits temperature-sensitive growth at 43 degrees C and temperature-sensitive translation in vivo and in vitro. Extracts of the mutant produce an aberrant pattern of translation products of MS2 bacteriophage RNA. Previous work has shown that a protein, called 'rescue', isolated from the parental strain partly corrects the defective translation in vitro. Here we report the purification to homogeneity of a second factor from ribosomal eluates of the wild-type parental strain; the purified protein is a homodimer of 54 kDa. The partial sequence of the second protein was determined, and a recombinant plasmid was isolated based on its ability to complement the temperature-sensitive growth phenotype of the mutant at the non-permissive temperatures. The cloned gene was sequenced, mapped to the 20.9-min region of the E. coli chromosome and shown to code for a 466-amino-acid protein with a molecular mass of 52 kDa. Analysis of the DNA sequence and the correspondence to that of the partial protein sequence has identified the complementing factor as asparaginyl-tRNA synthetase. Marker rescue experiments indicate that the asnS mutation in N4316 resides within the motif 2 domain of the synthetase. A potential role of this synthetase in restoring normal protein synthesis with respect to ribosomal frameshifting, read-through of nonsense codons and protein copy number is discussed.

Acceleration of heat transfer in vial freeze-drying of pharmaceuticals. II. A fluid cushion device

A simple device for the improvement of freeze-drying efficiency is described. The device is an aluminum foil bag which contains a small amount of glycerin. The device can be either reusable or disposable. When placed on a freeze-drying tray the liquid is about 1 mm thick. When vials are placed on the device it conforms to the shape of the vial bottoms. Since both the aluminum foil and the glycerin are better heat conductors than a vacuum, the device improves heat transfer from the shelf to the vial. Drying times obtained with and without the device are compared for different sizes as well as different types of vials. In most cases the use of the device reduces the drying time by nearly a factor of two. The use of the device also increases vial-to-vial uniformity and minimizes the effect of spillage.

cDNA-derived amino acid sequence of the NADH-binding 51-kDa subunit of the bovine respiratory NADH dehydrogenase reveals striking similarities to a bacterial NAD(+)-reducing hydrogenase

A lambda gt10 bovine brain and a lambda gt11 bovine heart cDNA library were screened with oligonucleotide probes corresponding to partial protein sequences directly determined from the isolated 51-kDa subunit of the bovine respiratory-chain NADH dehydrogenase. Clones were isolated that encode a protein of 464 amino acids containing all the 11 partial tryptic peptide sequences determined from the 51-kDa subunit. The size and amino acid composition of this protein agree with those determined for the purified 51-kDa subunit. Furthermore, this protein contains a putative NADH-binding domain, a possible FMN-binding site, and a putative binding site for an iron-sulfur cluster. The above evidence indicates that the cloned protein is the 51-kDa subunit or its precursor. A search for sequence similarity with proteins in the Protein Identification Resource data base has revealed that the 51-kDa subunit has 32% amino acid sequence identity with a major portion of the alpha subunit of the soluble NAD(+)-reducing hydrogenase from Alcaligenes eutrophus. In particular, there are three segments of high sequence similarity (70-88%) between the two proteins which correspond to the three ligand-binding sites.

Acceleration of heat transfer in vial freeze-drying of pharmaceuticals. I: Corrugated aluminum quilt

The medium offering the greatest resistance to heat transfer from the freeze-dryer shelf to the moving and subliming interface within the product is the space between the flat shelf top and the concave vial bottom. The resistance to heat transfer can be greatly reduced by improving the thermal conductivity of this intervening space. Several heat transfer augmentation devices, which fill this gap, are described. The devices are inexpensive and easy to use. Experimental data show that the resistance of the intervening space is reduced appreciably and the drying time is greatly reduced.

Structural studies on mitochondrial NADH dehydrogenase using chemical cross-linking

The structure of bovine heart mitochondrial NADH dehydrogenase was investigated by cross-linking constituent subunits with disuccinimidyl tartrate, (ethylene glycol)yl bis(succinimidyl succinate) and dimethyl suberimidate. Cross-linked products were identified by Western blotting with monospecific antisera to nine subunits of the enzyme. Cross-links between subunits within the flavoprotein, iron-protein and hydrophobic domains of the enzyme were identified. Cross-linking between the 75 kDa iron-protein-domain subunit and the 51 kDa flavoprotein-domain subunit was modulated by the substrate NADH. Cross-linking of subunits of the iron-protein and flavoprotein domains to constituents of the hydrophobic domain was also found. This was further substantiated by photolabelling subunits of the latter region, which were in contact with the membrane lipid, with 3-(trifluoromethyl)-3-(m-[125I]iodophenyl)diazirine. One such subunit of Mr 19,000 could be cross-linked to components of the iron-protein domain.

Mitochondrial autoantigens in primary biliary cirrhosis. Association of disease-specific determinants with a subunit of complex I (NADH-ubiquinone reductase) of the inner mitochondrial membrane

Anti-mitochondrial autoantibodies (AMA) from patients with primary biliary cirrhosis (PBC) were analysed for fine specificity by immunoblotting and enzyme-linked immunosorbent assay (ELISA). Inhibition ELISA showed that complex I (NADH-ubiquinone reductase) from beef heart mitochondria completely inhibited the binding of AMA to mitochondrial inner membranes (SMP), indicating that the major mitochondrial antigens are located in complex I. Immunoblot analysis of beef heart SMP, complex I and the iron sulphur (IP) subfraction of complex I revealed several antigens, one of which (75 kDa) reacted with all PBC sera but not with the additional autoimmune sera tested. Resolution of SMP or complex I by two-dimensional electrophoresis yielded in both preparations a polypeptide of 75 kDa with an isoelectric point of 6.4, which reacted with PBC serum and with rabbit antisera raised against the 75,000 subunit of complex I. In immunoblot experiments, the antigenicity of the 75,000 polypeptide in SMP, complex I, and the IP subfraction is increased by prior reduction of the sample with mercaptoethanol. This suggests a similarity to the PBC-specific 'M-2' antigen, which is also sensitive to sulphur reagents. The data indicate that the 75 kDa polypeptide of complex I is a major mitochondrial antigen binding AMA in PBC sera, and allows us to identify the location and probable function of the PBC antigen.

Transmembrane organization of mitochondrial NADH dehydrogenase as revealed by radiochemical labelling and cross-linking

The organization of bovine heart NADH dehydrogenase in the mitochondrial inner membrane was investigated by chemical cross-linking and radiolabelling with [125I]iododiazobenzenesulphonate (IDABS). Mitochondria or submitochondrial particles were cross-linked with disulphosuccinimidyl tartrate and dimethyl suberimidate, and dimeric products containing subunits of the NADH dehydrogenase were analysed by Western blotting with subunit-specific antisera. Cross-linking of mitochondria gave rise to (49 + 30) kDa and (49 + 19) kDa dimers and an additional dimer containing the 30 kDa subunit. Cross-linking of submitochondrial particles gave rise to (75 + 51) kDa, (75 + 30) kDa and (49 + 13) kDa dimers and a further dimer containing the 30 kDa subunit. We conclude that the 49 kDa and 30 kDa subunits are transmembranous, the 19 kDa subunit is exposed on the cytoplasmic face of the membrane, whereas the 75, 51 and 13 kDa subunits are exposed on the matrix face of the membrane. Reaction of the isolated enzyme with IDABS results in labelling of 75, 49, 42, 33, 30, 13 and 10 kDa subunits. From experiments in which mitochondria or submitochondrial particles were first labelled and NADH dehydrogenase then isolated by immunoprecipitation, it was found that labelling of the 49 kDa subunit occurs predominantly from the cytoplasmic side of the membrane. On the other hand, labelling of the 75, 13 and 10 kDa subunits occurs predominantly from the matrix side of the membrane, whereas the 30 and 33 kDa subunits are heavily labelled from either side. These findings are consistent with those obtained from cross-linking.