Abstract
Deficiency of the E1α-subunit of the pyruvate dehydrogenase (PDH) complex is an X-linked inborn error of metabolism and one of the major causes of lactic acidosis in children. Although most heterozygous females manifest symptoms of the disease, it is often difficult to establish the diagnosis as results based on measurement of total PDH activity, and E1α-immunoreactive protein in patient fibroblasts may be ambiguous because of the variability in the pattern of X chromosome inactivation. We report the development of a set of monoclonal antibodies (MAbs) specific to four subunits of the PDH complex that can be used for detection of PDH E1α deficiency. We also show that anti-E1α and anti-E2 MAbs, when used in immunocytochemical analysis, can detect mosaicism in cell cultures from female patients in which as few as 2–5% of cells express the deficiency. This immunocytochemical approach, which is fast, reliable, and quantitative, will be particularly useful in identifying females with PDH E1α-subunit deficiency as a precursor to mutation analysis.
Keywords
The pyruvate dehydrogenase (PDH) complex is central to aerobic carbohydrate metabolism. It is localized to the matrix space of mitochondria, where it catalyzes the irreversible oxidative decarboxylation of pyruvate entering the organelle to produce acetyl-CoA, NADH, and CO2 (Reed et al. 1992). The PDH complex contains three primary enzyme components: pyruvate dehydrogenase (E1), dihydrolipoamide transacetylase (E2), and dihydrolipoamide dehydrogenase (E3) (Jilka et al. 1986; Patel and Roche 1990; Reed et al. 1992). The E1 component is a heterotetramer of two α- and two β-subunits. PDH activity is regulated by specific E1 kinase and phospho-E1-phosphatase enzymes (Gudi et al. 1995; Bowker-Kinley et al. 1998; Bowker-Kinley and Popov 1999; Wu et al. 1999; Baker et al. 2000; Steussy et al. 2001), which respectively inactivate and activate the complex by phosphorylation and dephosphorylation of three serine residues in the E1α-subunit (Sugden et al. 1978,1979; Sugden and Simister 1980; Korotchkina et al. 1995). There is an additional component, the E3-binding protein, (E3BP), which is required for proper interaction of the E2 and E3 components (Patel and Roche 1990). The overall complex contains 12 copies of E3, 30 copies of E1, 60 copies of E2, and 12 copies of E3BP. In total, the molecular weight of the PDH complex is more than 8 million Da (Patel and Roche 1990; De Vivo 1998).
PDH complex deficiency is a relatively common cause of severe metabolic and neurodegenerative disease in infants and young children. There is a wide variety of clinical presentations, ranging from severe lactic acidosis in newborns to a progressive neurodegenerative disease with prolonged survival (Brown et al. 1994; Cross et al. 1994; Fujii et al. 1994; Patel and Harris 1995; Patel et al. 1995; Lissens et al. 2000). The vast majority of these deficiencies (>95%) involve the E1α-subunit, although mutations in other subunits and in PDH phosphatase have also been described (Brown et al. 1994; Hong et al. 1996). Although the gene for the E1α-subunit is located on the X chromosome, defects in this subunit occur in approximately equal frequency in males and females. Complete PDH E1α deficiency has not been observed in males and is presumed to be lethal at an early stage of development. By contrast, mutations that abolish E1α protein synthesis are commonly found in heterozygous female patients who are mosaic for two populations of cells as a consequence of X chromosome inactivation. In these females, some cells have completely normal PDH function whereas others are deficient (Brown et al. 1989b,1990). The highly variable pattern of X inactivation in different individuals, and in different tissues in the same individual, contributes significantly to the widely different clinical presentations in female patients. Although the majority of these heterozygous females present with obvious PDH deficiency, a small number of females carry an E1α-subunit mutation but have mild symptoms or no symptoms at all, presumably because of an X-inactivation pattern favoring expression of the normal X chromosome (Brown and Brown 1994; Fujii et al. 1994; Matthews et al. 1994).
In many cases, the diagnosis of PDH deficiency is easily established by enzyme assay in patients' cells or tissue (commonly by measuring 14CO2 production from [1-14C]-labeled pyruvate), with subsequent definition of the basic defect by Western blotting with antibodies against subunits of the complex and/or by direct cDNA or genomic sequencing (Old and De Vivo 1989; Matthews et al. 1994; Brown et al. 1997; Otero et al. 1998). However, two aspects of PDH E1α deficiency can make it difficult to establish a genetic diagnosis. The first is related to the fact that, in a number of male patients with clear biochemical evidence of a PDH E1α defect, no mutation can be detected in the gene (Brown et al. 1997). The second is that a significant proportion of female patients with clinical features suggestive of PDH E1α deficiency have levels of enzyme activity and E1α-immunoreactive protein within the normal range (Brown et al. 1989a,1994, 1997; Brown and Brown 1993; Brown and Brown 1994). In these females it is unclear whether genetic analysis is warranted and, even if it is undertaken, the simplest approach of direct sequencing of PDH E1α cDNA is precluded if only the normal sequence will be represented in the fibroblast mRNAs.
In these female patients, a rapid, simple, and sensitive screening test for mosaicism would provide support for further genetic studies with the expectation that an E1α gene mutation would be identified in positive cases. Here we describe the development of such a diagnostic approach, based on immunocytochemical staining of cell cultures, to identify PDH E1α defects. We have tested this approach by examining well-characterized female patients with PDH E1α deficiency and have determined the sensitivity of the method using a model system in which cells with normal and mutant enzyme are mixed in various proportions. In patients with PDH E1α mutations that reduce the amount of immunoreactive protein, mosaicism with as few as 2–5% of deficient cells can be detected unequivocally by this method.
Materials and Methods
Cell Lines
MRC5 fibroblasts were obtained from the American Type Culture Collection (Manassas, VA). Patient fibroblasts (Table 1) were obtained from skin biopsies of patients with confirmed defects in expression of the E1α-subunit. Cells were grown in high-glucose Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum and 50 μg/ml uridine. Cell lines BRI-T22/2 and BRI-T22/4 were derived from the fibroblast culture BRI, established from a female patient heterozygous for a PDH E1α frameshift mutation. Fibroblasts were transfected with the T22 plasmid containing a mutant variant of the SV40 T antigen (Brown et al. 1997) and clones of stably transformed cells that express only the mutant (BRI-T22/2) or the normal (BRI-T22/4) E1α gene were isolated. Cells were grown to confluency on 150-mm2 plates, washed with calcium- and magnesium-free PBS (CMF-PBS), and then trypsinized for collection into 15 ml tubes. The two cell lines were each diluted to a concentration of 105 cells/ml and mixed in different proportions to give experimental mosaic samples with from 1% to 99% mutant.
Cell lines used in the study; correlation of PDH activity measurements and calculated % of normal cells by ICC
Unpublished.
Monoclonal Antibodies
The MAbs used in this study were developed in the University of Oregon Monoclonal Antibody Facility (Eugene, OR). Anti-E2, E2/E3bp, and E1β MAbs were generated respectively by immunizing mice with purified porcine PDH complex. The anti-E1α MAb was raised against purified Histagged human PDH E1α-subunit overexpressed in E. coli. Antibodies were screened first for binding to purified porcine PDH (Sigma; St Louis, MO) or pure E1α-subunit (for anti-E1α) and then for specific binding to a single subunit by denaturing Western blots of both pure porcine PDH (or pure E1α-subunit for anti-E1α) and human mitochondria.
Production of PDH E1α-subunit in E. coli
The PDH E1α was derived from the plasmid PDH 1C (Dahl et al. 1987). Oligodeoxynucleotides were synthesized by Gibco BRL (Rockville, MD). Polymerase chain reactions (PCRs) were performed using primers ML1 and ML2 (5′ to 3′ ML1: cgg gat ccg ttt gca aat gat gct aca ttt g; ML2: cgg gat cct taa ctg act gac tta aac ttg atc c) to introduce BamH I restriction sites into the 5′ and 3′ ends of the sequences coding for human liver mature PDH E1α. The amplified DNA fragments were subcloned into the expression vector pET-15b (Novagen; Madison, WI) and the resulting plasmid was used for the overexpression and purification of the protein in an E. coli strain BL21(DE3) as described in the Novagen Handbook protocol.
Immunocytochemistry
For immunocytochemical (ICC) analysis, cells were cultured on glass coverslips until they reached approximately 75% confluency. They were rinsed with CMF-PBS and fixed with 1 ml 4% paraformaldehyde for 20–30 min at 4C. After washing several times with PBS, they were either processed immediately or stored in PBS with 0.02% azide at 4C for later use. Cells were permeabilized for 15 min in acetone at −20C, washed several times with PBS, and blocked for 1 hr in 10% normal goat serum (NGS)-PBS at room temperature (RT). For labeling with single MAbs, the cells were incubated with the appropriate antibody (i.e., with 0.6 μg/ml of anti-E2-subunit antibody or with 10 μg/ml of anti E1α-subunit antibody) for 2 hr at RT or overnight at 4C. After washing, they were incubated in goat anti-mouse IgG H+L Alexa Fluor-594 (Molecular Probes; Eugene, OR) at 2 μg/ml. For double labeling, the same conditions were used for reaction with the E1α MAb and secondary labeling with anti-mouse IgG H+L Alexa Fluor-594. Then, after extensive washing, the cells were further incubated with E2 MAb conjugated directly with Alexa Fluor-488. In the last wash, Dapi was sometimes added to stain the cell nuclei, and after this the coverslips were mounted on slides with 20 μl of Mowiol solution. Cells were visualized using a Zeiss Axioplan2 microscope equipped with a X100 Neofluar objective, UV, FITC, and Texas Red filters (Chroma Technologies; Brattleboro, VT). Images were acquired using a Hamamatsu Orca camera. Subsequent image processing was done using Open lab software (Improvisions; Lexington, MA).
Western Blotting Analysis
Protein samples (1 μg/lane pure enzyme; 2 μg/lane human heart mitochondria; 10 μg/lane cell lysate from patient fibroblasts) were dissolved in SDS-PAGE tricine sample buffer (Bio Rad; Hercules, CA) containing 2% β-mercaptoethanol by incubation for 30 min at 37C. The proteins were separated on a 10% SDS-polyacrylamide gel at 100 V for ∼20 min (stacking gel) and then at 200 V for ∼40 min (separating gel). Proteins were transferred electrophoretically to 0.45-μm polyvinylidine difluoride (PVDF) membranes for 2 hr at 100 mAmp in CAPS buffer (3.3 g CAPS, 1.5 liter 10% methanol, pH 11.00) on ice. The membranes were rinsed in PBS and blocked in 5% milk-PBS + 0.02% azide overnight. The membranes were incubated in primary antibody diluted in 5% milk-PBS + 0.02% azide overnight, rinsed once, then washed four times with PBST (1 ml Tween-20/liter PBS) for 15 min. Finally, the membranes were incubated for 2 hr with horseradish peroxidase-conjugated goat anti-mouse IgG+M at 0.5 μg/ml (Jackson ImmunoResearch; West Grove, PA) diluted in 5% milk-PBS. The secondary antibody was detected using the chemiluminescent ECL Plus reagent (Amersham Pharmacia Biotech; Arlington, IL) after washing the blots three times with PBST and once with PBS.
Results
Characterization of Anti-PDH Complex Antibodies
We have developed a set of MAbs against four different subunits of the PDH complex using two different antigens. The specificity of each MAb was established by Western blotting analysis and ICC of human fibroblasts. The specificities and concentrations of these MAbs required for the different protocols described here are listed in Table 2. All except the anti-E2/E3BP MAb reacted with a single band in Western blots of mitochondria and whole cell extracts. The anti-E1α and anti-E2 MAbs were found to react with PDH in ICC, and they form the basis of the PDH mosaicism test.
PDH-subunit Detection by Western Blotting
Western blots of cell lysate proteins isolated from several of the cell lines listed in Table 1 are presented in Figure 1. For these experiments, the four MAbs listed in Table 2 (i.e., 13G2AE2BH5, 9H9AF5, 17A5E2H8, and 21A11AE7) were combined and the data are the result of a single blotting experiment. Cell line BR1-T22/4, derived from BR1 fibroblasts (see Materials and Methods) and expressing the normal E1α gene, provided a control sample (Figure 1, Lane 1). The primary fibroblast culture from female patient BR1, with 15% residual PDH activity, had normal amounts of E2 protein compared with control fibroblasts, but a much reduced level of E1α-subunit and essentially no detectable E1β. The mutation in this patient is a 20-bp deletion in exon 10, which results in complete deficiency of E1α protein. The subunit profiles of cell lines BR5 and BR6 were similar to that of BR1 (result not shown), indicating the insertion mutation in BR5 and the exon skipping mutation in BR6, respectively, also affect assembly of E1.
Monoclonal antibodies used in the study a
ICC, immunocytochemistry; WB, Western blotting.
Figure 1 also shows the subunit composition of the PDH complex in the cell lysate of male patient 920002 and also his mother (cell line 920226), both with a mutation R10P in the leader sequence of the E1α precursor protein (Takakubo et al. 1995). The Western blot shows very low levels of E1α and E1β immunoreactivity in the cells of the patient (who was reported to retain 28% PDH activity). His mother, who is mildly affected, has approximately 50% of the mean control PDH activity, which is at the lower end of the normal range. As shown in Figure 1, the subunit profile of 920226 is not greatly different from that of controls. These Western blotting experiments confirm that in some females PDH E1α defects are difficult to diagnose unequivocally by Western blotting.
Western blotting analysis of cultured skin fibroblasts. Approximately 30 μg/lane of total protein from the cell lysate was separated on 10% SDS-PAGE and then transferred to a PVDF membrane (A). Western blotting signals in control (BR1-T22/4) and three patients were quantitated and the levels of E2, E3-binding protein, E1α- and E1β-subunits were plotted in relation to BR1-T22/4 control fibroblasts (B). In the control all subunits were set to 100%. All lanes were standardized by using succinate dehydrogenase for equal loading.
An Immunohistochemical Method to Analyze PDH Deficiency in Females
Two of the antibodies used in the Western blots, those that recognize the E1α- and E2-subunits, also react with their respective proteins in ICC. Hence, it is possible to perform a double-labeling experiment to analyze the levels of these subunits in individual cells. Where a particular mutation in the E1α gene leads to a significant reduction in the level of immunoreactive protein, mosaicism in cell cultures of heterozygous females is readily observed, as shown in Figure 2 for patient BR1. In these micrographs, E1α protein is stained red by a secondary reaction of the primary MAb with goat anti-mouse IgG H+L Alexa Fluor-594-conjugated antibody (MPI), and E2 is stained green with an MAb labeled directly by conjugation with Alexa Fluor-488 (MPI).
To investigate the sensitivity of this ICC approach, “model patients” were generated by mixing defined numbers of cells BR1-T22/2 (with no E1α expressed) and BR1-T22/4 (with normal levels of the E1α-subunit). These experimental mosaics were double stained with the anti-E1α and anti-E2 antibodies. Figures 2A and 2D show the uniform labeling of the E2-subunit in all cells. Figures 2B and 2E show that the E1α-subunit is detectable only in a small proportion of the cells. Figures 2C and 2F are merged images in which cells labeled green are deficient and cells staining yellow have normal PDH E1α expression. Micrographs were analyzed by counting up to 1000 cells for each sample. The plot in Figure 3 shows the percentage of mutant cells in the mixture vs the percentage of cells with no detectable E1α staining based on cell counts, and demonstrates an excellent correlation between the two. Mutant cells were readily detectable even when they constituted as little as 2–5% of the population. The data obtained on samples with a low proportion of mutant cells are replotted in Figure 3B.
Mosaic expression of E1α protein in BR1 cell line fibroblasts. (A,D) Cells stained with anti-E2 antibody only (0.6 μg/ml). (B,E) Cells labeled with anti-E1α antibody only (10 μg/ml). (C,F) Merged images. Original magnifications:
Table 1 summarizes the results of analysis of a number of patient fibroblast cultures. In general, the percentage of normal cells detected by ICC is somewhat lower than expected from activity (reported in previously published reports) and/or Western blotting data (obtained in this study). It is important to note that comparisons are not based on the concurrent estimates of PDH activity, Western blotting, and ICC from the same batch of cultured cells. The clear advantage of the ICC approach compared with previously used methods of detecting PDH deficiency in heterozygous females is shown by the results for patients 920002 and 920226. In cells from the male patient 920002, all cells stain positive for E2 (Figure 4A) but the intensity of the E1α staining is significantly reduced (Figure 4B). There appears to be some assembled PDH complex based on the overlap of the punctate green and red staining (Figure 4C), consistent with the 28% residual PDH activity reported in this patient (Takakubo et al. 1995). However, much of the E1α appears as separate red spots in the merged image, indicating that it is not associated with E2. The implication is that the E1α precursor containing the mutation R10P is still targeted to mitochondria, possibly less efficiently than normal, but is poorly assembled. Not clear in the figure but evident in the fluorescent microscope is a background of E1α staining in the cytosol of many cells. Fibroblasts from the mother of this patient, with around 50% PDH activity and near-normal amounts of E1α-immunoreactive protein on Western blotting, are shown in Figures 4D–4F. Clearly, there is an approximately 50/50 mixture of normal and deficient cells, with a markedly different staining pattern compared with the cells from her son.
Discussion
Here we describe the ICC approach to detect the mosaicism in cell cultures from heterozygous female patients with PDH E1α deficiency. This approach to the diagnosis is applicable only to those patients in whom the mutation results in absence of immunoreactive protein or a significantly reduced level. To date, 76 different mutations of the PDH E1α-subunit have been described. Of these, 49% are missense/nonsense and 51% insertion/deletion mutations. Although there are approximately equal numbers of male and female patients, the distribution of the mutations is not even. In females, 65% of mutations are insertions/deletions, in males 72% are missense/nonsense (Lissens et al. 2000). Information about the level of PDH E1α-immunoreactive protein is available for 20 mutations in males and 17 in females. In the males there is reduced immunoreactive protein in 70% overall, 64% in the missense/nonsense group, and 78% in the insertion/deletion group. In the females, the overall frequency of reduced protein is 30%, made up of 20% in the missense/nonsense group and 58% in the insertion/deletion group.
Correlation between percentage of mutant cells in the experimental mosaic samples (expected % mutant cells) vs the percentage of cells in the mixtures with no immunoreactive E1α protein (observed % mutant cells).
The difference between males and females in terms of the overall frequency of reduced immunoreactive PDH E1α protein is quite striking because there appears to be a bias towards normal levels of protein in the females. This is even more significant given that they have a higher frequency of insertion/deletion mutations, which are more likely (on both theoretical grounds and directly from observations in male patients) to interfere with production of the protein product. The explanation for this is almost certainly the mosaicism that is a characteristic of these heterozygous females. Residual enzyme activity and the level of immunoreactive protein will vary depending on the proportion of the cells expressing the normal gene. If the majority of cells in the fibroblast culture are expressing the normal X chromosome, activity and protein content, averaged over many cells, will be close to normal, making detection difficult. Although mosaicism may obscure the diagnosis when a population of cells is pooled for analysis, it can also be exploited as a powerful diagnostic tool because individual cells in heterozygous females are either completely normal or deficient. Detection of mosaicism in a female suspected of having PDH deficiency is a clear indication that the E1α is responsible.
ICC analysis of cell lines 920002
Here we describe a two-MAb ICC approach that enables ready identification of E1α defects in individual cells, even when the proportion of mutant cells is as low as 1–2%. The basis of the assay is a two-color fluorescence staining of patients' fibroblasts using one MAb to E1α and a second MAb to E2 as a control. Western blotting analysis of cell lines described here (and several others not reported) show that E2 is present in E1α-deficient cells, often at normal levels, making this a useful control. However, other mitochondrial proteins for which good MAbs are available can also be used, e.g., porin or cytochrome c oxidase-subunit I. The ICC approach was validated using experimentally derived mixtures of normal and E1α-deficient cells, and an excellent fit between expected and observed percentages of deficient cells was obtained. In patient fibroblast cultures, PDH E1α mosaicism could be clearly demonstrated in female patients and easily quantitated. In the case of a mother of one defined patient (cell lines 920226 and 920002) with PDH E1α deficiency, it was equivocal whether she carried the mutation in her somatic cells based on activity or Western blotting data. The ICC analysis of her fibroblasts provided a clear diagnosis: she is a carrier of the PDH E1α mutation detected in her son.
To date, we have limited our studies to fibroblast cultures derived from patients. It is important to note that the X inactivation pattern (and therefore cellular mosaicism) in skin fibroblasts may not reflect that found in other tissues. This would account for the fact that some females with PDH E1α have severe neurological disease even though the results of enzyme analysis in cultured fibroblasts is within the normal range (Matthews et al. 1994). It is highly unlikely, on theoretical grounds, that there should be no deficient cells at all in the fibroblast cultures from such females, and therefore the ability to detect a minimal proportion against an overwhelming background of normal cells is of great diagnostic importance. The only case in which this approach will fail to detect mosaicism is when the mutation does not alter the amount of E1α-immunoreactive protein. At present, the proportion of female patients with mutations of this type cannot be assessed because many of the published reports do not take the contribution of mosaicism into account. On the basis of the spectrum of PDH E1α mutations identified thus far, it is expected that this method will be informative in the majority of female patients. However, further studies will provide a definitive answer to this.
In summary, we describe an ICC approach to detecting mosaicism of defects that affect the amount of PDH E1α protein in heterozygous females, which should prove an invaluable additional test in identifying patients for subsequent analysis by genetic studies to provide a definitive diagnosis and informed genetic counseling.
Footnotes
Acknowledgements
We would like to thank Dr David Thorburn for providing us with cell lines 920002 and 920226.