Characterization of Two cDNA Clones for Pyruvate Dehydrogenase E& Subunit and Its Regulation in Tricarboxylic Acid Cycle-deficient Fibroblast”

Two distinct types of cDNA clones encoding for the pyruvate dehydrogenase (PDH) E1@ subunit were isolated from a human liver Xgtll cDNA library and characterized. These cDNA clones have identical nucleotide sequences for PDH EIB protein coding region but differ in their lengths and in the sequences of their 3’-untranslated regions. The smaller cDNA had an unusual polyadenylation signal within its protein coding region. The cDNA-deduced protein of PDH EIj3 subunit revealed a precursor protein of 359 amino acid residues (Mr 39,223) and a mature protein of 329 residues (M= 35,894), respectively. Both cDNAs shared high amino acid sequence similarity with that isolated from human foreskin (Koike, K. K., Ohta, S., Urata, Y., Kagawa, Y., and Koike, M. (1988) Proc. Natl. Acad. Sci. U. S. A. 85,41-45) except for three regions of frameshift mutation. These changes led to dramatic alterations in the local net charges and predicted protein conformation. One of the different sequences in the protein coding region of liver cDNA (nucleotide position 452-752) reported here was confirmed by sequencing the region after amplification of cDNA prepared from human skin fibroblasts by the polymerase chain reaction. Southern blot analysis verified simple patterns of hybridization with E1j3 cDNA, indicating that the PDH El/3 subunit gene is not a member of a multigene family. The mechanisms of differential expression of the PDH Elcw and EIB subunits were also studied in established fibroblast cell lines obtained from patients with Leigh’s syndrome and other forms of congenital lactic acidosis. In Northern blot analyses for PDH E1cu and El/3 subunits, no apparent differences were observed between two Leigh’s syndrome and the control fibroblasts studied: one species of PDH Ela! mRNA and three species of EIB mRNA were observed in all the cell lines examined. However, in one tricarboxylic acid cycle deficient fibroblast cell line, which has one-tenth of the normal enzyme activity, the levels of immunoreactive PDH Elcr and El@ subunits were markedly decreased as assessed by immunoblot analyses. These data indicated a regulatory mutation caused by either inefficient translation of E1a! and E# mRNAs into protein or rapid degradation of both subunits upon translation. In contrast, the PDH Elan and E1fi subunits in two fibroblast cell lines from Leigh’s syndrome patients appeared to be normal as judged by 1) enzyme activity, 2) mRNA Northern blot, 3) genomic DNA Southern blot, and 4) immunoblot analyses

Two distinct types of cDNA clones encoding for the pyruvate dehydrogenase (PDH) E1@ subunit were isolated from a human liver Xgtll cDNA library and characterized.
These cDNA clones have identical nucleotide sequences for PDH EIB protein coding region but differ in their lengths and in the sequences of their 3'-untranslated regions.
The smaller cDNA had an unusual polyadenylation signal within its protein coding region.
These changes led to dramatic alterations in the local net charges and predicted protein conformation.
One of the different sequences in the protein coding region of liver cDNA (nucleotide position 452-752) reported here was confirmed by sequencing the region after amplification of cDNA prepared from human skin fibroblasts by the polymerase chain reaction.
Southern blot analysis verified simple patterns of hybridization with E1j3 cDNA, indicating that the PDH El/3 subunit gene is not a member of a multigene family. The mechanisms of differential expression of the PDH Elcw and EIB subunits were also studied in established fibroblast cell lines obtained from patients with Leigh's syndrome and other forms of congenital lactic acidosis. In Northern blot analyses for PDH E1cu and El/3 subunits, no apparent differences were observed between two Leigh's syndrome and the control fibroblasts studied: one species of PDH Ela! mRNA and three species of EIB mRNA were observed in all the cell lines examined.
However, in one tricarboxylic acid cycle deficient fibroblast cell line, which has one-tenth of the normal enzyme activity, the levels of immunoreactive PDH Elcr and El@ subunits were markedly decreased as assessed by immunoblot analyses. These data indicated a regulatory mutation caused by either inefficient translation of E1a! and E# mRNAs into protein or rapid degradation of both subunits upon translation.
In contrast, the PDH Elan and E1fi subunits in two fibroblast cell lines from Leigh's syndrome patients appeared to be normal as judged by 1) enzyme activity, 2) mRNA Northern blot, 3) genomic DNA Southern blot, and 4) immunoblot analyses * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 USC. Section 1734 solely to indicate this fact. $ To whom correspondence should be addressed.
indicating that the lactic acidosis seen in these patients did not result from a single defect in either of these Ela and El/3 subunits of the PDH complex. (2-4). Because of the central role of the PDH complex in glucose metabolism and energy production, this enzyme complex has been extensively studied in a variety of pathological conditions as to structure, function, subunit interaction, and regulation of enzyme activity (1,4,5). Defects in any one of the components may result in congenital lactic acidosis, which manifests symptoms varying from mild to severe ataxia (6-8), or in a form designated as Leigh's syndrome (subacute necrotizing encephalomyelopathy), which is characterized by mental and growth retardation with occasional premature death (9, 10). These diseases usually follow an autosomal recessive pattern of inheritance (5, 11). It has been also suggested that 2,3-butanediol, found in the serum of human alcoholics both in the presence (12-14) and the absence of ingested ethanol (15), may result from the reduction of acetoin, a reaction product of the PDH E1 subunit (16)(17)(18).
Despite numerous studies of the catalytic properties and the regulation of the PDH E~(Y subunit, EiP subunit has not been well characterized regarding its structure and functional interaction with the PDH Eia! subunit. Recent data suggested that a deficiency of the Eio and El/3 subunits may be responsible for certain forms of lactic acidosis. (19). Here we report the complete nucleotide and deduced protein sequences of two cDNAs encoding for the human liver PDH El/3 subunit. We also present the evidence of differential regulation of the PDH E1~ and EIP subunits in cultured fibroblast cell lines including two cell lines from Leigh's syndrome patients and one cell line from a patient deficient in tricarboxylic acid cycle. was added (11). Cells were gently mixed and immediately frozen in liquid nitrogen and stored overnight at -80 "C. The next morning cells were frozen and thawed twice, immediately prior to assay.
PDH Assay-PDH activity in crude cell extracts was assayed by following acetyl coenzyme A production at 37 "C as the method described by Sorbi and Blass (24) using acetyl coenzyme A:arylamine N-acetyltransferase (EC 2.3.1.5). Arylamine N-acetyltransferase used in the PDH assays was prepared from pigeon liver acetone powder (Sigma).
Extraction and acetone precipitation were done as described by Tabor et al. (25) Fig. 2) and primer 2, 5'-TCTAAGCAGTGGCCCACAGGT-3' (antisense of nucleotide position 752-732 in Fig. 2). cDNA amplification was performed by using the GeneAmp DNA amplification reagent kit (Perkin-Elmer-Cetus Instruments). The reaction mixture (100 ~1 in volume) contained 200 ng of human skin fibroblast cDNAs, 1.0 jiM of two primers, 200 pM of each dNTP, and 2.5 unit of TuqI DNA polymerase. Fifty cycles of denaturation (94 "C, 60 s), annealing (50 "C, 90 s), and extension (72 "C, 120 s) were carried out in an automatic DNA thermal cycler (Perkin-Elmer-Cetus Instruments). Amplified DNAs were separated on a 1.0% agarose gel electrophoresis. The amplified DNA band equivalent to 301 base pairs on agarose gel was excised, electroeluted, and subsequently subcloned into a SmaI sites of Ml3 mp18 and mp19 sequencing vectors. The correct sequence of PDH E&l cDNA prepared from human fibroblasts was subsequently confirmed by the DNA sequencing using dideoxynucleotide-chain termination method (23).

RESULTS
Isolation and Characterization of cDNA Clones for Human Pyruvate Dehydrogenase El/3 Subunit-In order to study the structure and regulation of human pyruvate dehydrogenase complex, we isolated cDNA clones for PDH El/3 subunit using an oligodeoxynucleotide probe. Screening more than 400,000 colonies of a human liver Xgtll cDNA library, five positive cDNAs with insert sizes ranging from 0.9 to 0.5 kb were identified and plaque purified. All of these were highly homologous to a cDNA clone recently isolated from human foreskin (20). One cDNA was used as a probe to isolate the full-length cDNA clone for PDH E$ subunit. Finally, two distinct cDNAs with insert sizes of 1.1 and 1.5 kb were isolated, subcloned into plasmid pUCl3, and designated pHLPB14 and pHLPB12, respectively, (Fig. 1).
The Nucleotide and Deduced Protein Sequences of Two cDNA Clones for El/3 Subunit-The primary structures of these cDNA clones were determined by nucleotide sequencing using the strategy given in Fig. 1. The sequence of pHLPB14 revealed that it possesses the entire E1@ protein coding region, comprising the leader sequence for the precursor protein and the mature protein coding sequences with the initiation codon ATG and the termination codon TAG. In contrast, pHLPB12 lacks 72 bases of 5'-amino-terminal sequence (Fig. 2). pHLPB14 contained 5 bases of the 5'-untranslated region followed by 1,080 bases of an open reading frame and 44 bases The nucleotide and deduced protein sequences of the full-length cDNA for PDH E,fl subunit, pHLPB14 is shown. Identical nucleotide sequence of the other cDNA clone, pHLPB12, starting at nucleotide 73, is also designated by the solid lines. Amino acid sequences for the oligodeoxynucleotide probe are denoted by asterisks (') while the termination codon is shown in the box. Bold underlines represent the potential polyadenylation signal AATAAA, 22 bases upstream from the termination codon in pHLPB14 and 383 bases downstream from the termination codon in pHLPB12. The different amino acid residues between liver and foreskin (20) clones are denoted in bold letters.
of the 3'-untranslated region including a poly(A) tail which is located 11 bases downstream from the termination codon TAG. The consensus polyadenylation signal AATAAA which follows the termination codon was not observed in the 3'untranslated region of pHLPB14, but an alternative one was identified 22 bases upstream to the termination codon TAG. In contrast to pHLPB14, the second cDNA, pHLPB12, had a 400-base 3'-untranslated region including another potential polyadenylation signal AATAAA which was found 383 bases downstream from the termination codon TAG. However, the nucleotide sequence of pHLPB12 for the El/3 protein coding region is identical with that of pHLPB14.
Thus, the deduced protein sequences of pHLPB14 revealed a precursor protein of 359-amino acid residues containing the leader sequence (20,31) and a mature protein of 329 amino acid residues with molecular weights of 39,223 and 35,894, respectively. When the deduced protein sequences of our human liver cDNA clones were compared with that of a foreskin cDNA clone (20), three frameshift mutations were detected, one in the leader sequence and two in the protein coding region. As shown in Figs. 2 and 3, the absence of one base (T) between nucleotide position 23-24 and the presence of an additional base (G) at nucleotide position 39 were found in the leader sequence region of the liver cDNA clone, pHLPB14.
In the mature protein coding region, the absence of one base at nucleotide positions 663-664 (G) and 928-929 (A) and the presence of an additional base at nucleotide positions 638 (C) and 935 (T) were identified in both of our liver cDNA clones, pHLPB14 and pHLPB12. Another single base substitution at nucleotide position 438 (A in liver clone, but G in foreskin clone) was found, but it would result in a silent mutation that would not change the amino acid composition.
Because of apparent differences in the sequences at nucleotide position 438 (substitution of G to A) and at nucleotide position 935 (addition of T) for the El@ clones, two additional AuaII restriction enzyme sites were generated in the human liver The actual autoradiography of sequencing gels demonstrating the differences between liver and foreskin (20) cDNA clones are presented. The regions of nucleotide insertion or deletion in liver cDNAs as compared with foreskin clone are denoted by asterisks (*). Compared with foreskin cDNA, the human liver cDNAs revealed one base T deletion (-r) and one base G insertion (+G) in leader sequence (left panel), and one base G deletion (-C) and one base C insertion (+C) in the mature protein coding sequence (right panel). The directions of sequencing gel reading (from 5' to 3' end) are indicated by bold letters with arrows, and numbers in parentheses represent nucleotide position in Fig. 2. cDNA clones, pHLPB14 and pHLPB12 (Fig. l), while only one A&I site was present in the foreskin clone (20). The additional AvaII restriction endonuclease sites were confirmed by the digestion of our cDNA clones with this restriction enzyme (data not shown). Because of the differences in the nucleotide sequences of foreskin and liver, the deduced amino acid sequences for PDH E,/l subunit were different in two clones from two different tissues: 5 residues in the leader region and 11 residues in the mature protein (denoted as bold characters in Fig. 2). The actual autoradiographs of the sequencing gels demonstrated the differences between liver and foreskin cDNA clones in their leader regions (nucleotide position 19-41) and mature protein coding regions (nucleotide position 634-667). Thus, the deduced amino acid sequence for PDH E$ that we report here might result in alterations in local net charges and protein secondary structures of the leader region (amino acid position 5-15) as well as the mature protein (amino acid positions 208-226 and 304-318) from what was reported by Koike et al. (Table I). In addition to the changes described above, the deduced amino acid sequence in the leader region of liver PDH EIP contained 3 more arginine residues (5 arginine residues in liver, Fig. 2) than that of foreskin PDH EIP (20). This finding is of some interest as arginine is thought to play an important role in the protein processing of mitochondrial presequence (32). In order to verify the differences in nucleotide sequences of the PDH EIP cDNAs, human skin fibroblast cDNA spanning 301 bases (corresponded to the sequence from nucleotide position 452 to 752 of human liver cDNA in Fig. 1) was amplified by the polymerase chain reaction as described under "Materials and Methods." DNA sequencing (data not shown) of the amplified cDNA revealed identical nucleotide sequence with that of PDH E,@ cDNAs isolated from human liver (Fig.  2). Thus, the differences in the nucleotides and the subsequent translated peptide sequences might be due to cloning artifacts which occurred during the preparation of a foreskin cDNA library or simply misreading of the sequencing data. Differential Regulation of PDH E1 in Various Human Skin Fibroblasts-Recent reports suggest that there are multiple modes of regulation of the PDH E1 subunit (11,19,(33)(34)(35). To further delineate the biochemical mechanism of PDHrelated abnormalities in human subjects, several established fibroblast cell lines from patients with different clinical symptoms (Leigh's syndrome and tricarboxylic acid cycle defective) were selected. The cells were grown in in uitro tissue culture and analyzed for the enzyme activity, protein, and mRNA levels. The cell lines used in this study included one control cell line, two from patients with Leigh's syndrome, and one tricarboxylic acid cycle-deficient cell line. The levels of dichloroacetate-stimulated pyruvate dehydrogenase activity in the control fibroblast was 4.0 nmol of acetyl coenzyme A production/min/mg protein. Rather surprisingly, the enzyme activities of whole PDH complex in the fibroblasts from the two Leigh's syndrome patients (GM1503 and GM3672) were 5.5 and 3.0 nmol/min/mg protein, which are comparable to the activity of the control cell. In contrast, in the tricarboxylic acid cycle defective mutant cell, it was about 0.3 nmol/min/ mg protein, which is in agreement with an earlier report (11).
The possible mechanism of abnormality in the PDH E1 from these fibroblasts was studied. The amounts of mRNA for PDH Elcv and E,fi were measured in the various fibroblasts by the Northern blot analyses using PDH E1cr2 and PDH E,P cDNAs as probes. As shown in Fig. 4A, a single species of mRNA (4.8 kb in size) that hybridized with PDH E~LV cDNA was observed in all the fibroblasts examined. The amounts of PDH Eloc mRNA in the control and patients' fibroblasts were comparable.
When another identical gel was subjected to Northern blot hybridization with PDH E,P cDNA, one major species of mRNA (4.4 kb in size) and two minor species (1.8 and 1.1 kb in sizes) were observed (Fig. 4B). The amounts of PDH EJ mRNA in the control and patients' fibroblasts were again almost equal. The similar species and quantities of mRNAs for both PDH Elcv and EIfl subunits in various cell lines, including the tricarboxylic acid cycle defective mutant, indicated that the defects were not associated with the abnormal expressions of Ercv or E$ mRNAs.
The post-transcriptional defect in the tricarboxylic acid cycle defective mutant was further investigated.
Immunoblot analyses for PDH E1~ and E,/3 subunits were performed using protein subunits of PDH E1~ and E$. Only one immunoreactive PDH E,a band (with an apparent molecular mass of 41,000 daltons) was observed in the whole homogenates from fibroblasts used. The amounts of immunoreactive PDH Elcv subunit in fibroblasts from the control and the Leigh's syndrome patients were almost equal and easily detected whereas the amount in the tricarboxylic acid cycle defective mutant was quite low and almost undetectable (Fig. 5A). A similar observation was made with PDH EIP subunit quantified by polyclonal antibody against PDH E,P (Fig. 5B). Only one immunoreactive PDH E,P protein (with an apparent molecular mass of 36,000 daltons) was observed in whole homogenates from all fibroblasts used in this experiment. The amounts of immunoreactive PDH E,@ protein in the control and the Leigh's syndrome patients were similar and easily detected while that in the tricarboxylic acid cycle defective mutant was much lower than those of the control cells. The immunoblot data suggested that low enzyme activity observed in tricarboxylic acid cycle defective mutant was due to decreased levels of both PDH Elol and El/3 proteins. Similar immunoblot data of decreased levels of PDH EP and E3 subunits were also observed indicating the same types of defect for these subunits in this particular cell line (data not shown).
Genomic DNA Southern Blot Analysis for PDH EIP Gene-In order to determine whether PDH ErP gene has other closely related gene family members, Southern blot analysis was performed.
Total genomic DNAs, isolated from various human skin fibroblasts, were digested with restriction endonucleases, subjected to agarose gel electrophoresis and transferred to GeneScreen membrane.
The DNA band hybridization patterns with "'P-labeled EIP cDNA probe were simple and identical for all the genomic DNAs isolated. Only a few fragments were detected in all the cell lines including the control, Leigh's syndrome, and tricarboxylic acid cycle-deficient patient's fibroblasts (Fig. 6) HindIII(H) and P&I(P). DNA fragments were electrophoresed on 0.8% agarose gel, transferred to GeneScreen membrane, and then subjected to genomic Southern blot analysis using "P-labeled PDH E,@ cDNA (mixture of pHLPB14 and pHLPB12) as a probe. The fibroblast cell lines used in this experiment are represented by one-letter code: A, control (GM1654); B, Leigh's syndrome (GM1503); and C, tricarboxylic acid (TCA) cycle deficient fibroblast (GM3093). Sizes of DNA bands were estimated by 1-kb DNA ladder.
(pHLPB14) indicated that PDH E$ subunit is not the product of a multiple gene family. From the sum of the sizes of DNA fragments generated by P&I cleavage (1.4, 2.3, and 16 kb), the total length for pyruvate dehydrogenase EIP gene would be approximately 19.7 kb and may represent a single gene for PDH E$ subunit.

DISCUSSION
The pyruvate dehydrogenase E, enzyme is a tetramer consisting of two identical E,oc and two identical E,@ subunits. The enzyme is inactivated by phosphorylation and activated by dephosphorylation by a PDH-specific kinase and a PDHspecific phosphatase (2-4), respectively. Its activity is also dependent upon the concentrations of various metabolic regulators such as pyruvate, ATP/ADP, NAD/NADH', and acetyl-coenzyme A/Coenzyme A ratios (36). Although numerous studies were carried out on the function and structure of the E,(Y subunit (37,38), relatively little information on the potential role of the El/3 subunit in PDH activity is available. In this report, we described the isolation and sequences of two distinct cDNA clones for human liver PDH E$ which have identical nucleotide sequences for protein coding regions. One clone, pHLPB14, had an unusual polyadenylation signal within the protein coding sequence which is immediately followed by poly(A) tail. Similar unusual cases were recently reported for other cDNA clones for human gonadotropin psubunit (39), human factor X (40), and human lecithincholesterol acetyltransferase (41). The nucleotide sequence and deduced amino acid sequences in the two cDNAs isolated from human liver library revealed that they were highly homologous to those of a foreskin clone (ZO), except for three regions of frameshift mutations and one base substitution.
In this report, we confirmed that our nucleotide sequence of human liver PDH E,/3 cDNA (nucleotide position 452-752 in Fig. 2) was correct. The confirmation was accomplished by the sequencing of the region of difference after skin fibroblast DNA was amplified by the polymerase chain reaction. The exact mechanism for the difference in the nucleotide sequences of PDH E$ from human liver and from foreskin (20) is not known, but it could be due to mutations during gene conversions (42), cloning artifacts, or misreading of nucleotide sequences. If the previously predicted structures for foreskin cDNA is incorrect, the frameshifts observed here would result in drastic changes in the local net charges of the amino acids and probably its secondary protein structures. The significance of these alterations in amino acid composition with regard to the changes of catalytic activity awaits further biochemical characterization. Northern blot analyses of all of the fibroblasts used in our experiments indicated that there are one major species of mRNA for PDH E,cv subunit and three species of mRNA for PDH E1@ subunit. Our data for PDH E# mRNA were in contrast to the results of Koike et al. (20), who reported only one species of mRNA for El@ subunit (1.7 kb) in cultured HeLa cells. Although we do not know the reason for the apparent differences, multiple species of E$ mRNA were also observed in hamster tissues as well as in rat tissues." The simple patterns of hybridization found in the genomic DNA fragments generated by digestion with different restriction enzymes indicated that PDH E,P is not a member of a multigene family. Based on the relatively small size (about 19.7 kb in total size), it is probably derived from a single gene localized on the human chromosome 3 (43). The multiple species of EJ cDNA clones and mRNA reported here thus represent the possibility of alternative splicing of a single gene and the usage of different polyadenylation signals during the synthesis and processing of its mRNAs.
In the present study, we also attempted to explore the underlying mechanism of the deficient pyruvate metabolism in some of the well-established fibroblasts from patients who are thought to have defects in PDH El subunits. In the two cell lines from Leigh's syndrome patients, we found no abnormality in either PDH Eloc or EIP subunits as judged by the total enzyme activity, amounts of both mRNA and immunoreactive proteins. The levels of PDH enzyme activities for these fibroblasts appeared to be normal and comparable to those of the control cell lines. The defects in the fibroblasts from Leigh's syndrome patients may not be due to the defects in PDH E1 activity as claimed (10) but rather due to problems of pyruvate transport and uptake through the mitochondrial membrane and defects in other enzyme systems such as PDH phosphatase (44) or cytochrome oxidase (45). Recent reports on the same fibroblasts suggested that the defect is caused by structural abnormality in dihydrolipoamide dehydrogenase (46). The latter claim was based on a relatively lower sensitivity to inhibitory antibodies against lipoamide dehydrogenase and inefficient reconstitution between the subunits to form a catalytically active PDH complex. Our data that failed to demonstrate an abnormality in PDH Elcv and E$ subunits support their results. On the other hand, in the tricarboxylic acid cycle defective mutant, we observed the reduced levels of immunoreactive PDH E1a and E$ subunits with low enzyme activity despite the equivalent amounts of their mRNAs. The results suggested that the defects in this cell line might be due to a post-transcriptional mutation. This may include a defect in translational machinery and inefficient translation of mRNA into subunit proteins. This, in turn, may result in unreliable incorporation of PDH subunits into the mitochondria (47) leading to the rapid degradation of protein despite sufficient levels of protein translation. Alternatively, the de-" T.-L. Huh, J. P. Casazza, J.-W. Huh, R. L Veech, and B. J. Song, unpublished observations. fects might be due to abnormalities of mitochondrial structures (48) or mitochondrial carrier proteins necessary for efficient pyruvate oxidation (49). Because of the decreased levels of immunoreactive proteins and enzyme activities of all PDH subunits examined, it is more likely that the defect in this cell line could be mainly due to abnormal mitochondrial structures as suggested. However, the exact biochemical mechanisms of the defects in PDH complex enzymes in these established fibroblasts can be further elucidated by the molecular biology techniques such as expression of cloned cDNAs for PDH subunits.