Pre-ANDA strategy and human factors activities to de-risk pharmaceutical companies ANDA submission of drug-device combination products: case study of a formative comparative use human factors study.

BACKGROUND: This article presents a strategy that a Drug Delivery Device Developer (DDDD) has adopted to support Abbreviated New Drug Application (ANDA) submissions of drug-device combination products. As per the related FDA guidance, a threshold analysis should be compiled. If 'other differences' between the Reference Listed Drug (RLD) and the generic drug devices are identified, a Comparative Use Human Factors (CUHF) study may be requested. METHODS: The DDDD performed task analysis and physical comparison to assess the pen injector design differences. Then, a formative CUHF study with 25 participants simulating injections using both RLD and the generic pen injectors was conducted. RESULTS: After each participant completed four simulated injections, similar type and rates of use error between the RLD (0.70) and generic (0.68) pen injectors were observed. CONCLUSION: DDDDs can support pharmaceutical companies in the ANDA submission strategy of their drug-device combination product by initiating comparative task analysis and physical comparison of the device as inputs for the threshold analysis. If 'other differences' are identified, a formative CUHF study can be performed. As shown in our case study, this approach can be leveraged to support the sample size calculation and non-inferiority margin determination for a CUHF study with the final combination product.

Formative and Validation Human Factors studies of a new disposable autoinjector for subcutaneous delivery of chronic disease therapies.

BACKGROUND: Use-related risks related to autoinjector devices have been previously identified. To minimize these problems, the identification of potential use errors is a critical task during device development. METHODS: This article presents iterative human factor studies, which aim to assess user interaction with the tested push-on-skin BD Intevia™ 1 mL Disposable Autoinjector, across a wide range of indications, and a broad user population. RESULTS: Through the different human factor studies, use errors were recorded when the participants completed the critical tasks, but their occurrence continuously decreased. First, the incidence of use errors was reduced when the participants read the IFU. In addition, the IFU updates and design change implemented contributed to improve the usability performance. During the validation study, some use errors were still observed, mainly during the first uses. Nevertheless, providing a training to the participants almost fully eliminated the remaining use errors. CONCLUSION: Thus, these results demonstrated that this new autoinjector can be safely and efficiently used for its intended uses and under the expected use conditions by all tested user groups.

Comprehensive and comparative exploration of the Atp7b-/- mouse plasma proteome.

Wilson's disease (WD), a rare genetic disease caused by mutations in the ATP7B gene, is associated with altered expression and/or function of the copper-transporting ATP7B protein, leading to massive toxic accumulation of copper in the liver and brain. The Atp7b-/- mouse, a genetic and phenotypic model of WD, was developed to provide new insights into the pathogenic mechanisms of WD. Many plasma proteins are secreted by the liver, and impairment of liver function can trigger changes to the plasma proteome. High standard proteomics workflows can identify such changes. Here, we explored the plasma proteome of the Atp7b-/- mouse using a mass spectrometry (MS)-based proteomics workflow combining unbiased discovery analysis followed by targeted quantification. Among the 367 unique plasma proteins identified, 7 proteins were confirmed as differentially abundant between Atp7b-/- mice and wild-type littermates, and were directly linked to WD pathophysiology (regeneration of liver parenchyma, plasma iron depletion, etc.). We then adapted our targeted proteomics assay to quantify human orthologues of these proteins in plasma from copper-chelator-treated WD patients. The plasma proteome changes observed in the Atp7b-/- mouse were not confirmed in these samples, except for alpha-1 antichymotrypsin, levels of which were decreased in WD patients compared to healthy individuals. Plasma ceruloplasmin was investigated in both the Atp7b-/- mouse model and human patients; it was significantly decreased in the human form of WD only. In conclusion, MS-based proteomics is a method of choice to identify proteome changes in murine models of disrupted metal homeostasis, and allows their validation in human cohorts.

A meiotic XPF-ERCC1-like complex recognizes joint molecule recombination intermediates to promote crossover formation.

Meiotic crossover formation requires the stabilization of early recombination intermediates by a set of proteins and occurs within the environment of the chromosome axis, a structure important for the regulation of meiotic recombination events. The molecular mechanisms underlying and connecting crossover recombination and axis localization are elusive. Here, we identified the ZZS (Zip2-Zip4-Spo16) complex, required for crossover formation, which carries two distinct activities: one provided by Zip4, which acts as hub through physical interactions with components of the chromosome axis and the crossover machinery, and the other carried by Zip2 and Spo16, which preferentially bind branched DNA molecules in vitro. We found that Zip2 and Spo16 share structural similarities to the structure-specific XPF-ERCC1 nuclease, although it lacks endonuclease activity. The XPF domain of Zip2 is required for crossover formation, suggesting that, together with Spo16, it has a noncatalytic DNA recognition function. Our results suggest that the ZZS complex shepherds recombination intermediates toward crossovers as a dynamic structural module that connects recombination events to the chromosome axis. The identification of the ZZS complex improves our understanding of the various activities required for crossover implementation and is likely applicable to other organisms, including mammals.

The PHD finger protein Spp1 has distinct functions in the Set1 and the meiotic DSB formation complexes.

Histone H3K4 methylation is a feature of meiotic recombination hotspots shared by many organisms including plants and mammals. Meiotic recombination is initiated by programmed double-strand break (DSB) formation that in budding yeast takes place in gene promoters and is promoted by histone H3K4 di/trimethylation. This histone modification is recognized by Spp1, a PHD finger containing protein that belongs to the conserved histone H3K4 methyltransferase Set1 complex. During meiosis, Spp1 binds H3K4me3 and interacts with a DSB protein, Mer2, to promote DSB formation close to gene promoters. How Set1 complex- and Mer2- related functions of Spp1 are connected is not clear. Here, combining genome-wide localization analyses, biochemical approaches and the use of separation of function mutants, we show that Spp1 is present within two distinct complexes in meiotic cells, the Set1 and the Mer2 complexes. Disrupting the Spp1-Set1 interaction mildly decreases H3K4me3 levels and does not affect meiotic recombination initiation. Conversely, the Spp1-Mer2 interaction is required for normal meiotic recombination initiation, but dispensable for Set1 complex-mediated histone H3K4 methylation. Finally, we provide evidence that Spp1 preserves normal H3K4me3 levels independently of the Set1 complex. We propose a model where Spp1 works in three ways to promote recombination initiation: first by depositing histone H3K4 methylation (Set1 complex), next by "reading" and protecting histone H3K4 methylation, and finally by making the link with the chromosome axis (Mer2-Spp1 complex). This work deciphers the precise roles of Spp1 in meiotic recombination and opens perspectives to study its functions in other organisms where H3K4me3 is also present at recombination hotspots.

Bdf1 Bromodomains Are Essential for Meiosis and the Expression of Meiotic-Specific Genes.

Bromodomain and Extra-terminal motif (BET) proteins play a central role in transcription regulation and chromatin signalling pathways. They are present in unicellular eukaryotes and in this study, the role of the BET protein Bdf1 has been explored in Saccharomyces cerevisiae. Mutation of Bdf1 bromodomains revealed defects on both the formation of spores and the meiotic progression, blocking cells at the exit from prophase, before the first meiotic division. This phenotype is associated with a massive deregulation of the transcription of meiotic genes and Bdf1 bromodomains are required for appropriate expression of the key meiotic transcription factor NDT80 and almost all the Ndt80-inducible genes, including APC complex components. Bdf1 notably accumulates on the promoter of Ndt80 and its recruitment is dependent on Bdf1 bromodomains. In addition, the ectopic expression of NDT80 during meiosis partially bypasses this dependency. Finally, purification of Bdf1 partners identified two independent complexes with Bdf2 or the SWR complex, neither of which was required to complete sporulation. Taken together, our results unveil a new role for Bdf1 -working independently from its predominant protein partners Bdf2 and the SWR1 complex-as a regulator of meiosis-specific genes.

Spiked proteomic standard dataset for testing label-free quantitative software and statistical methods.

This data article describes a controlled, spiked proteomic dataset for which the "ground truth" of variant proteins is known. It is based on the LC-MS analysis of samples composed of a fixed background of yeast lysate and different spiked amounts of the UPS1 mixture of 48 recombinant proteins. It can be used to objectively evaluate bioinformatic pipelines for label-free quantitative analysis, and their ability to detect variant proteins with good sensitivity and low false discovery rate in large-scale proteomic studies. More specifically, it can be useful for tuning software tools parameters, but also testing new algorithms for label-free quantitative analysis, or for evaluation of downstream statistical methods. The raw MS files can be downloaded from ProteomeXchange with identifier PXD001819. Starting from some raw files of this dataset, we also provide here some processed data obtained through various bioinformatics tools (including MaxQuant, Skyline, MFPaQ, IRMa-hEIDI and Scaffold) in different workflows, to exemplify the use of such data in the context of software benchmarking, as discussed in details in the accompanying manuscript [1]. The experimental design used here for data processing takes advantage of the different spike levels introduced in the samples composing the dataset, and processed data are merged in a single file to facilitate the evaluation and illustration of software tools results for the detection of variant proteins with different absolute expression levels and fold change values.

Calibration plot for proteomics: A graphical tool to visually check the assumptions underlying FDR control in quantitative experiments.

In MS-based quantitative proteomics, the FDR control (i.e. the limitation of the number of proteins that are wrongly claimed as differentially abundant between several conditions) is a major postanalysis step. It is classically achieved thanks to a specific statistical procedure that computes the adjusted p-values of the putative differentially abundant proteins. Unfortunately, such adjustment is conservative only if the p-values are well-calibrated; the false discovery control being spuriously underestimated otherwise. However, well-calibration is a property that can be violated in some practical cases. To overcome this limitation, we propose a graphical method to straightforwardly and visually assess the p-value well-calibration, as well as the R codes to embed it in any pipeline. All MS data have been deposited in the ProteomeXchange with identifier PXD002370 (http://proteomecentral.proteomexchange.org/dataset/PXD002370).

Benchmarking quantitative label-free LC-MS data processing workflows using a complex spiked proteomic standard dataset.

Proteomic workflows based on nanoLC-MS/MS data-dependent-acquisition analysis have progressed tremendously in recent years. High-resolution and fast sequencing instruments have enabled the use of label-free quantitative methods, based either on spectral counting or on MS signal analysis, which appear as an attractive way to analyze differential protein expression in complex biological samples. However, the computational processing of the data for label-free quantification still remains a challenge. Here, we used a proteomic standard composed of an equimolar mixture of 48 human proteins (Sigma UPS1) spiked at different concentrations into a background of yeast cell lysate to benchmark several label-free quantitative workflows, involving different software packages developed in recent years. This experimental design allowed to finely assess their performances in terms of sensitivity and false discovery rate, by measuring the number of true and false-positive (respectively UPS1 or yeast background proteins found as differential). The spiked standard dataset has been deposited to the ProteomeXchange repository with the identifier PXD001819 and can be used to benchmark other label-free workflows, adjust software parameter settings, improve algorithms for extraction of the quantitative metrics from raw MS data, or evaluate downstream statistical methods. BIOLOGICAL SIGNIFICANCE: Bioinformatic pipelines for label-free quantitative analysis must be objectively evaluated in their ability to detect variant proteins with good sensitivity and low false discovery rate in large-scale proteomic studies. This can be done through the use of complex spiked samples, for which the "ground truth" of variant proteins is known, allowing a statistical evaluation of the performances of the data processing workflow. We provide here such a controlled standard dataset and used it to evaluate the performances of several label-free bioinformatics tools (including MaxQuant, Skyline, MFPaQ, IRMa-hEIDI and Scaffold) in different workflows, for detection of variant proteins with different absolute expression levels and fold change values. The dataset presented here can be useful for tuning software tool parameters, and also testing new algorithms for label-free quantitative analysis, or for evaluation of downstream statistical methods.

In-depth study of Mollivirus sibericum, a new 30,000-y-old giant virus infecting Acanthamoeba.

Acanthamoeba species are infected by the largest known DNA viruses. These include icosahedral Mimiviruses, amphora-shaped Pandoraviruses, and Pithovirus sibericum, the latter one isolated from 30,000-y-old permafrost. Mollivirus sibericum, a fourth type of giant virus, was isolated from the same permafrost sample. Its approximately spherical virion (0.6-µm diameter) encloses a 651-kb GC-rich genome encoding 523 proteins of which 64% are ORFans; 16% have their closest homolog in Pandoraviruses and 10% in Acanthamoeba castellanii probably through horizontal gene transfer. The Mollivirus nucleocytoplasmic replication cycle was analyzed using a combination of "omic" approaches that revealed how the virus highjacks its host machinery to actively replicate. Surprisingly, the host's ribosomal proteins are packaged in the virion. Metagenomic analysis of the permafrost sample uncovered the presence of both viruses, yet in very low amount. The fact that two different viruses retain their infectivity in prehistorical permafrost layers should be of concern in a context of global warming. Giant viruses' diversity remains to be fully explored.

MicroRNA and target protein patterns reveal physiopathological features of glioma subtypes.

Gliomas such as oligodendrogliomas (ODG) and glioblastomas (GBM) are brain tumours with different clinical outcomes. Histology-based classification of these tumour types is often difficult. Therefore the first aim of this study was to gain microRNA data that can be used as reliable signatures of oligodendrogliomas and glioblastomas. We investigated the levels of 282 microRNAs using membrane-array hybridisation and real-time PCR in ODG, GBM and control brain tissues. In comparison to these control tissues, 26 deregulated microRNAs were identified in tumours and the tissue levels of seven microRNAs (miR-21, miR-128, miR-132, miR-134, miR-155, miR-210 and miR-409-5p) appropriately discriminated oligodendrogliomas from glioblastomas. Genomic, epigenomic and host gene expression studies were conducted to investigate the mechanisms involved in these deregulations. Another aim of this study was to better understand glioma physiopathology looking for targets of deregulated microRNAs. We discovered that some targets of these microRNAs such as STAT3, PTBP1 or SIRT1 are differentially expressed in gliomas consistent with deregulation of microRNA expression. Moreover, MDH1, the target of several deregulated microRNAs, is repressed in glioblastomas, making an intramitochondrial-NAD reduction mediated by the mitochondrial aspartate-malate shuttle unlikely. Understanding the connections between microRNAs and bioenergetic pathways in gliomas may lead to identification of novel therapeutic targets.

Identification of proteomic signatures of mantle cell lymphoma, small lymphocytic lymphoma, and marginal zone lymphoma biopsies by surface enhanced laser desorption/ionization-time of flight mass spectrometry.

Mantle cell lymphoma (MCL), small lymphocytic lymphoma (SLL), and marginal zone lymphoma (MZL) are small B-cell non-Hodgkin lymphomas (NHLs) that may be difficult to distinguish. In order to identify specific proteomic biomarkers, differential proteomic analysis of these three NHLs was performed using surface enhanced laser desorption/ionization-time of flight mass spectrometry (SELDI-TOF-MS). Whole cell lysates obtained from 18 MCL, 20 SLL, and 20 MZL biopsies were applied on two different ProteinChips (cationic and anionic). Hierarchical clustering and discriminating scores combined with an innovative bio-informatics microdissection strategy allowed us to distinguish specific lymphoma proteomic signatures based on the expression of 37 protein peaks. SELDI-assisted protein purification combined with nano-liquid chromatography (LC) quadrupole-time of flight tandem mass spectrometry (Q-TOF MS/MS) was used to identify proteins overexpressed in both MCL and SLL tumors. Among them two histones, H2B and H4, were identified in MCL tumor biopsies and the signal recognition particle 9 kDa protein, SRP9, in SLL tumor biopsies.

The heterogeneity of meningioma revealed by multiparameter analysis: infiltrative and non-infiltrative clinical phenotypes.

Tumor invasion or infiltration of adjacent tissues is the source of clinical challenges in diagnosis as well as prevention and treatment. Among brain tumors, infiltration of the adjacent tissues with diverse pleiotropic mechanisms is frequently encountered in benign meningiomas. We assessed whether a multiparametric analysis of meningiomas based on data from both clinical observations and molecular analyses could provide a consistent and accurate appraisal of invasive and infiltrative phenotypes and help determine the diagnosis of these tumors. Tissue analyses of 37 meningiomas combined enzyme-linked immunosorbent assay (ELISA) and surface-enhanced laser desorption/ionization time-of-flight (SELDI-TOF) assays of two different protein biomarkers (thrombospondin 1 and a phosphorylated form of vimentin) as well as gene expression analyses with oligonucleotide micro-arrays. Up to four different clinical and molecular parameters were then examined for tumor classification. From this study, we were able to cluster 36 out of the 37 tumors into two different subsets corresponding to infiltrative/invasive and non-infiltrative tumors. In addition, meningiomas that invade brain and those that infiltrate the neighboring skull bone exhibited no distinguishable molecular features. Our multi-parameter analysis that combines clinical data, transcriptomic and molecular assays clearly reveals the heterogeneity of meningiomas and distinguishes the intrinsically infiltrative/invasive tumors from the non-infiltrative meningiomas.

Increased phosphorylation of vimentin in noninfiltrative meningiomas.

BACKGROUND: Tissue invasion or tissue infiltration are clinical behaviors of a poor-prognosis subset of meningiomas. We carried out proteomic analyses of tissue extracts to discover new markers to accurately distinguish between infiltrative and noninfiltrative meningiomas. METHODOLOGY/PRINCIPAL FINDINGS: Protein lysates of 64 different tissue samples (including two brain-invasive and 32 infiltrative tumors) were submitted to SELDI-TOF mass spectrometric analysis. Mass profiles were used to build up both unsupervised and supervised hierarchical clustering. One marker was found at high levels in noninvasive and noninfiltrative tumors and appeared to be a discriminative marker for clustering infiltrative and/or invasive meningiomas versus noninvasive meningiomas in two distinct subsets. Sensitivity and specificity were 86.7% and 100%, respectively. This marker was purified and identified as a multiphosphorylated form of vimentin, a cytoskeletal protein expressed in meningiomas. CONCLUSIONS/SIGNIFICANCE: Specific forms of vimentin can be surrogate molecular indicators of the invasive/infiltrative phenotype in tumors.

AT_CHLORO, a comprehensive chloroplast proteome database with subplastidial localization and curated information on envelope proteins.

Recent advances in the proteomics field have allowed a series of high throughput experiments to be conducted on chloroplast samples, and the data are available in several public databases. However, the accurate localization of many chloroplast proteins often remains hypothetical. This is especially true for envelope proteins. We went a step further into the knowledge of the chloroplast proteome by focusing, in the same set of experiments, on the localization of proteins in the stroma, the thylakoids, and envelope membranes. LC-MS/MS-based analyses first allowed building the AT_CHLORO database (http://www.grenoble.prabi.fr/protehome/grenoble-plant-proteomics/), a comprehensive repertoire of the 1323 proteins, identified by 10,654 unique peptide sequences, present in highly purified chloroplasts and their subfractions prepared from Arabidopsis thaliana leaves. This database also provides extensive proteomics information (peptide sequences and molecular weight, chromatographic retention times, MS/MS spectra, and spectral count) for a unique chloroplast protein accurate mass and time tag database gathering identified peptides with their respective and precise analytical coordinates, molecular weight, and retention time. We assessed the partitioning of each protein in the three chloroplast compartments by using a semiquantitative proteomics approach (spectral count). These data together with an in-depth investigation of the literature were compiled to provide accurate subplastidial localization of previously known and newly identified proteins. A unique knowledge base containing extensive information on the proteins identified in envelope fractions was thus obtained, allowing new insights into this membrane system to be revealed. Altogether, the data we obtained provide unexpected information about plastidial or subplastidial localization of some proteins that were not suspected to be associated to this membrane system. The spectral counting-based strategy was further validated as the compartmentation of well known pathways (for instance, photosynthesis and amino acid, fatty acid, or glycerolipid biosynthesis) within chloroplasts could be dissected. It also allowed revisiting the compartmentation of the chloroplast metabolism and functions.

A versatile method for deciphering plant membrane proteomes.

Proteomics is a very powerful approach to link the information contained in sequenced genomes, like that of Arabidopsis, to the functional knowledge provided by studies of plant cell compartments. This article summarizes the different steps of a versatile strategy that has been developed to decipher plant membrane proteomes. Initiated with envelope membranes from spinach chloroplasts, this strategy has been adapted to thylakoids, and further extended to a series of membranes from the model plant Arabidopsis: chloroplast envelope membranes, plasma membrane, and mitochondrial membranes. The first step is the preparation of highly purified membrane fractions from plant tissues. The second step in the strategy is the fractionation of membrane proteins on the basis of their physico-chemical properties. Chloroform/methanol extraction and washing of membranes with NaOH, NaCl or any other agent led to the simplification of the protein content of the fraction to be analysed. The next step is the genuine proteomic step, i.e. the separation of proteins by 1D-gel electrophoresis followed by in-gel proteolytic digestion of the polypeptides, analysis of the proteolytic peptides using mass spectrometry, and protein identification by searching through databases. The last step is the validation of the procedure by checking the subcellular location. The results obtained by using this strategy demonstrate that a combination of different proteomics approaches, together with bioinformatics, indeed provide a better understanding of the biochemical machinery of the different plant membranes at the molecular level.

An optimized strategy for ICAT quantification of membrane proteins.

The work presented here focuses on the development of a method adapting isotope labeling of proteins with ICAT to the study of highly hydrophobic proteins. Conditions for the labeling of proteins were first established using two standard soluble proteins and iodoacetamidyl-3,6-dioxaoctanediamine biotin (PEO-iodoacetyl biotin). Results demonstrated the efficiency of the labeling in the presence of high concentrations of both SDS and urea. These conditions were then used to label a highly hydrophobic mitochondrial membrane protein, the adenine nucleotide translocator ANT-1, with PEO-iodoacetyl biotin and then with the cleavable ICAT reagent. The results presented here show that labeling of proteins with cleavable ICAT is possible and may even be improved in strong denaturing buffers containing both SDS at a concentration higher than 0.5% (w/v) and urea. These results open the possibility of applying the ICAT strategy to complex samples containing very hydrophobic proteins solubilized in urea-SDS buffers. The adaptability of the developed method is demonstrated here with preliminary results obtained during the study of membrane-enriched fractions prepared from murine embryonic stem cells.

Conformation-dependent swinging of the matrix loop m2 of the mitochondrial Saccharomyces cerevisiae ADP/ATP carrier.

Structure-function relationships of the membrane-embedded Saccharomyces cerevisiae mitochondrial ADP/ATP carrier were investigated through two independent approaches, namely, limited proteolysis and cysteine labeling. Experiments were carried out in the presence of either carboxyatractyloside (CATR) or bongkrekic acid (BA), two specific inhibitors of the ADP/ATP transport that bind to two distinct conformers involved in the translocation process. The proteolysis approach allowed us to demonstrate (i) that N- and C-terminal extremities of ADP/ATP carrier are facing the intermembrane space and (ii) that the central region of the carrier corresponding to the matrix loop m2 is accessible to externally added trypsin in a conformation-sensitive manner, being cleaved at the Lys163-Gly164 and Lys178-Thr179 bonds in the carrier-CATR and the carrier-BA complexes, respectively. The cysteine labeling approach was carried out on the S161C mutant of the ADP/ATP carrier. This variant of the carrier is fully active, displaying nucleotide transport kinetic parameters and inhibitor binding properties similar to that of wild-type carrier. Alkylation experiments, carried out on mitochondria with the nonpermeable reagents eosin-5-maleimide and iodoacetamidyl-3,6-dioxaoctanediamine-biotin, showed that Cys 161 is accessible from the outside in the carrier-CATR complex, whereas it is masked in the carrier-BA complex. Taken together, our results indicate that the matrix loop m2 connecting the transmembrane helices H3 to H4 intrudes to some extent into the inner mitochondrial membrane. Its participation in the translocation of ADP/ATP is strongly suggested, based on the finding that its accessibility to reagents added outside mitochondria is modified according to the conformational state of the carrier.

Structural organization of mitochondrial human complex I: role of the ND4 and ND5 mitochondria-encoded subunits and interaction with prohibitin.

Mitochondria-encoded ND (NADH dehydrogenase) subunits, as components of the hydrophobic part of complex I, are essential for NADH:ubiquinone oxidoreductase activity. Mutations or lack of expression of these subunits have significant pathogenic consequences in humans. However, the way these events affect complex I assembly is poorly documented. To understand the effects of particular mutations in ND subunits on complex I assembly, we studied four human cell lines: ND4 non-expressing cells, ND5 non-expressing cells, and rho degrees cells that do not express any ND subunits, in comparison with normal complex I control cells. In control cells, all the seven analysed nuclear-encoded complex I subunits were found to be attached to the mitochondrial inner membrane, except for the 24 kDa subunit, which was nearly equally partitioned between the membranes and the matrix. Absence of a single ND subunit, or even all the seven ND subunits, caused no major changes in the nuclear-encoded complex I subunit content of mitochondria. However, in cells lacking ND4 or ND5, very low amounts of 24 kDa subunit were found associated with the membranes, whereas most of the other nuclear-encoded subunits remained attached. In contrast, membrane association of most of the nuclear subunits was significantly reduced in the absence of all seven ND proteins. Immunopurification detected several subcomplexes. One of these, containing the 23, 30 and 49 kDa subunits, also contained prohibitin. This is the first description of prohibitin interaction with complex I subunits and suggests that this protein might play a role in the assembly or degradation of mitochondrial complex I.