Functional Analysis in Saccharomyces Cerevisiae of Naturally Occurring Amino Acid Substitutions in Human Dihydrolipoamide Dehydrogenase

Abstract

Dihydrolipoamide dehydrogenase is a common component of mammalian multienzyme complexes that decarboxylate α-ketoacids and catabolize glycine. The common function is to reoxidize a reduced lipoate component of each complex, thereby preparing that lipoate for another round of catalysis. Regions within dihydrolipoamide dehydrogenase involved in association with other proteins of the complexes are poorly defined, and despite high amino acid sequence conservation through evolution, it is unknown if dihydro-lipoamide dehydrogenases are functionally equivalent across species. To address this issue, we asked whether the human enzyme could restore function to the -ketoacid dehydrogenase complexes in a yeast strain deficient in endogenous dihydrolipoamide dehydro-genase. This dihydrolipoamide dehydrogenase null mutant will not grow on non-fermentable carbon sources. The human enzyme expressed from a CEN plasmid complemented the growth phenotype and restored full activity to the pyruvate and α-ketoglutarate dehydrogenase complexes. Human dihydrolipoamide dehydrogenases with selected amino acid substitutions were then tested in the null strain for their ability to restore function. Substitutions tested represented naturally occurring candidate mutations identified in an individual with inactive dihydrolipoamide dehydro-genase. A K37E change had full function while a P453L change resulted in reduced dihydrolipoamide dehydrogenase abundance in the mitochondria and no detectable catalytic activity.

Introduction

Dihydrolipoamide dehydrogenase [E3] (EC 1.8.1.4) is a flavo-protein with a structure that is conserved in evolution from bacteria to plants and humans (1,2). In eukaryotic cells this protein functions as a homodimer within the mitochondria where it participates as a component of the α-ketoacid dehydrogenase complexes and the glycine cleavage system (1,3,4). Catalytically, E3 oxidizes the reduced dihydrolipoate covalently bound to the acyl transferase component of the ketoacid complexes and the hydrogen carrier protein of the glycine cleavage system. Lipoate reduction occurs along with the oxidative decarboxylation of the complex-specific substrate. For lipoate reoxidation, E3 activity utilizes non-covalently bound FAD as an intermediate leading to the ultimate reduction of NAD⁺(4).

Current knowledge for a structure-function relationship in E3 derives mainly from X-ray crystallogrraphic analysis of the yeast and bacterial proteins (5,6). Additional information comes from the comparison of this information with the crystal structure and amino acid sequence of glutathione reductase (4). Based on these analyses, FAD binding is thought primarily to involve residues within the first 150 amino acids of the protein (4). Two amino acids are implicated as having a critical role in catalytic activity, H452 and E457 (7). Naturally occurring mutations are reported for an individual expressing an E3 deficiency and thus suggest other functionally important residues. Two examples of patient identified amino acid substitutions are K37E and P453L, hypothesized as causal of the observed E3 inactivity (8). Based on earlier reports (4,7), Liu et al. that the K37E substitution affected FAD-binding while the P453L change altered the catalytic site. However, confirmation of these two amino acid substitutions as causal of the enzyme dysfunction has yet to be determined.

Human proteins can function in yeast (9,10) including proteins that function in mitochondria (11). Interaction of a human protein with a mitochondrial multi-protein complex in yeast also has been shown (12,13); however, a human protein interacting with several mitochondrial multienzyme complexes has yet to be demonstrated. The ability of human E3 [hE3] to function in yeast requires correct mitochondrial targeting, protein import and processing, and finally association with three different α-ketoacid dehydrogenase multienzyme complexes. This association is not identical for all of the complexes. E3 binds to the branched chain acyltransferase component of the branched chain α-ketoacid dehydrogenase [BCKD] complex (14), and to the decarboxylase component of the α-ketoglutarate [KGD] complex (15). For E3 to associate with the pyruvate dehydrogenase [PD] complex requires a separate E3-binding protein (16,17). Based on these various protein-protein interactions required by the different complexes, full complementation of the LPD null yeast strain by hE3 was not a trivial assumption.

Figure 1

Western blot of mitochondrial proteins from the parent strain transformed with pRS315-GAL1/10 [DMA2] and the lpd1::URA3 mutant strain bearing pRS315-GAL1/10 [MML22], YpML7 [MML22(hE3)] or YpML8 [MML22(yE3)]. DMA2 and MML22 lanes contain 30 µg protein and were exposed to film for 20 min. MML22(hE3) and MML22(yE3) lanes contain 20 µg protein and were exposed to film for 15 s.

Here we show that hE3 is able to complement a yeast strain deficient in endogenous E3 by interacting with the various yeast α-ketoacid dehydrogenase complexes. Further, the E3-null model was then used directly to test the effect of the two reported amino acid substitutions on E3 activity.

Results

To develop a yeast cell in which human E3 mutations could be tested for function, the strain MML22(lpd1::URA3) was constructed. Genotypes of all yeast strains used in these studies are shown in Table 1. The LPD1 gene of S. cerevisiae strain DMA2 was disrupted by homologous recombination with a targeting vector that substituted the URA3 gene for the middle third of the LPD1 gene. Structure of the mutant allele was confirmed by Southern blot and PCR analysis (data not shown). As seen in Figure 1, immunoblots demonstrated the wild-type parental strain (DMA2) expresses E3 while MML22 expresses no detectable endogenous yeast E3. In the transformed cells overexpression of either human or yeast E3 resulted in high levels of these proteins (note exposure time in Fig. 1). Since antihuman E3 antibody was used, cross reactivity with the yeast protein might account for some of the intensity difference observed.

MML22, has a phenotype of failure to grow on glycerol identical to that of a naturally occurring lpd1 strain that produces a dysfunctional Lpd1p (18). This phenotype is the result of a dysfunctional KGD complex (19). Transformations with the plasmids containing either the yeast or human DNA sequences for E3 rescue the cells from growth restriction on glycerol (Fig. 2). In five separate transformations, enhanced growth on glycerol was observed in transformants containing hE3 compared to those transformed with yE3.

Enzyme assays with MML22 detected no activity for either PD or KGD. Upon transformation with plasmids expressing either the human or yeast E3, PD and KGD activity was restored to MML22 (Table 2). BCKD activity was not detectable in the parent strain DMA2 so this activity was not measured in transformants. Activities of the three complexes vary among different yeast strains and with different growth conditions and BCKD is most variable (20).

Amino acid substitutions reported for the individual with E3 deficiency (8) were recreated within separate copies of the hE3 cDNA using site directed mutagenesis. Nucleotide sequence analysis confirmed that only the specific nucleotide substitution was present in each clone. When MML22 was transformed with the plasmid encoding the E3-K37E protein, the cells grew as well on glycerol as when MML22 was transformed with plasmid encoding wild-type hE3. As seen in Table 3 activity of E3 independent of its association with the dehydrogenase complexes was identical for E3-K37E and the wild-type E3. PD and KGD activities were also present in cells transformed with the plasmid encoding E3-K37E (Table 4), showing that E3-K37E can also associate and function within these multienzyme complexes. When MML22 was transformed with a plasmid encoding the E3-P453L protein, no E3 activity was detected (Table 3). These transformants did not grow on glycerol (data not shown) and since growth on glycerol requires a functional KGD complex (19) this phenotype means KGD is not functioning. PD activity was also absent in the cells transformed with the construct for E3-P453L (Table 4). Data reported in Tables 3 and 4 were from cells grown in 3% glycerol/0.3% glucose to enhance respiratory function of the mitochondria in an attempt to detect minimal activity from the altered hE3 proteins. Cells used for data in Table 2 grew in 1% glycerol/1% glucose, conditions that promote rapid growth from the higher glucose content while still forcing mitochondrial respiration. The different PD and KGD activity values reported in Tables 2 and 4 reflect these different growth conditions.

Western blots of mitochondrial proteins from the cells transformed to express E3-K37E demonstrated the presence of E3-K37E (data not shown). Thus, return of PD and KGD activity was due to the presence of the protein in the mitochondria. E3-P453L protein is also present in mitochondria of transformants although apparently at a lower concentration (Fig. 3). It is possible that the single amino acid substitution could alter antibody affinity but is unlikely since a polyclonal antiserum was used. Northern blots showed that mRNA levels for the E3-P453L were equal to those found in transformants made with wild-type hE3 constructs (data not shown).

Discussion

Growth of yeast on non-fermentable substrates depends on a functional tricarboxylic acid cycle. Therefore KGD activity is necessary for this growth (19). As an E3 null mutant, MML22 lacks KGD activity and will not grow on non-fermentable carbon sources. Transforming MML22 with plasmids encoding either human or yeast E3 restores growth on glycerol and implies a return of KGD activity. As reported by Jentoft et al. (4), yeast and human E3 have 55% identity and 73% similarity with respect to amino acid sequence. The human protein was capable of rescuing the yeast E3 null mutant suggesting that those regions of E3 most critical for function have been conserved. Growth with hE3 appears more robust than for yE3. Reasons for this observation are not clear since enzyme analysis of α-ketoacid dehydrogenases in the transformed cells showed similar restoration of KGD activity by either yeast or human E3. It is possible that overexpression of yE3 from the CEN plasmid causes some inhibition of proper association with KGD. This hypothesis is testable by changing the promoter used in the plasmid to one that produces less E3, but these experiments are beyond the scope of the current study.

Table 1

Genotypes of strains

Table 2

Activity of ketoacid dehydrogenase complexes in parent strain and LPD1::URA3 mutant strain before and after transformation

Table 3

Activity of dihydrolipoamide dehydrogenase in MML22 transformed with wild-type or mutant human E3

Table 4

Activity of ketoacid dehydrogenase complexes in cells transformed with wild type or mutant E3

PD activity was also restored demonstrating that the minor changes in E3 amino acid sequence across species results in little or no hindrance of hE3 function within the various yeast α-ketoacid dehydrogenase complexes. These results demonstrate that the E3 null yeast can be used to evaluate the function of altered E3 proteins by growth and enzyme activity assays. An ability to grow on glycerol without PD activity would reflect interaction of an altered E3 only with KGD. No growth on glycerol but PD activity would imply the reverse. Of special interest are putative changes in humans that lead to dysfunctional E3. The two substitutions modeled here were reported to exist in such an individual (8). Parental cells were not available to establish whether the substitutions were in separate alleles since complete nucleotide sequence for each allele in the proband was not reported. The possibility exists that both changes occurred in a single allele and the nucleotide change in the other allele remains to be determined. We did not create the double mutant for testing in this system.

Each amino acid change was created and tested as the only change in the E3 protein. The E3-K37E protein behaved exactly like the wild-type human E3. Alignment analysis (Fig. 1; 4) shows that K is conserved at this position in the human, porcine and yeast E3 peptide sequences. It appears therefore that changing from a positive to negative charge at this site is not important for catalysis or interaction with the complexes. A possible explanation would be that hydrophilicity is important but not charge. This region is thought to be important for FAD binding. However, since FAD is necessary for catalytic activity our results suggest that this amino acid substitution does not interfere with FAD binding. Flavin binding has not been directly assessed with either construct. Based on our results, it appears that the E3-K37E substitution does not result in a dysfunctional E3. Alternatively, the substitution could alter some aspect of expression or stability in humans that is not evident in the yeast model system.

Figure 2

Growth on SGly-leu plate (8 days) of transformed yeast: parent strain with host vector pRS315-GAL1/10 [DMA2], lpd1::URA3 mutant with pRS315-GAL1/10 [MML22], YpML7 [MML22(hE3)] or YpML8 [MML22(yE3)].

Figure 3

Western blot of mitochondrial proteins from MML22 transformed with either YpML7 [MML22(hE3)] or YpMLl 1 [MML22(P453L)].

In contrast the E3-P453L protein was catalytically inactive. Western blot analysis showed this protein was present in the mitochondria suggesting that import and processing had occurred. Assuming antigenic determinants are the same for wild-type and P453L E3, it appears that E3-P453L may be less stable. In support of this idea steady state mRNA levels appear at least equal to those for the wild-type transcript. Alternatively, translation efficiency could be decreased especially since the engineered change in the cDNA for E3-P453L resulted in use of a less favored yeast codon, and thus could be an artifact of the yeast model. Our data do not distinguish between the possible mechanisms for the apparently reduced amount of the abnormal protein.

Loss of catalytic activity for E3-P453L could be the result of an inability to form the homodimer, or alter the binding site for lipoamide. H452 is thought to be important for binding the lipoamide substrate, which occurs at the interface of the homodimer (4). Since leucine is an important residue in protein helix formation and proline breaks helical structure, it follows that a structural change in the binding site region occurs in E3-P453L thus accounting for the observed loss of activity.

In conclusion, our findings show that one of the two putative mutations reported for the individual with an apparent loss of E3 activity is truly catalytically inactive (8). It remains a possibility that the E3-P453L protein could have a dominant negative effect on the catalytically normal E3-K37E protein. Alternatively, both changes could be in the same allele and the change in the other allele remains to be uncovered. Double transformations of yeast are possible but demonstrating that two different proteins are interacting in the cell especially as a homodimer inside mitochondria is more difficult. However, the data presented demonstrate the validity of E3 null yeast as a model for testing mutations identified in humans that alter the amino acid composition of E3. Further, any engineered changes in the human or yeast gene can be tested. Thus this yeast strain provides a valuable tool for analysis of the E3 proteins.

Materials and Methods

Materials

Restriction endonucleases and DNA-modifying enzymes were purchased from Promega, Boehringer-Mannheim or US Biochemical. [1-¹⁴C] Pyruvic acid and sodium salt (10–30 mCi mmol) was purchased from Amersham, and [1-¹⁴C] α-ketoglutaric acid (52 mCi/mmol) from DuPont. Taq DNA polymerase was purchased from Boehringer-Mannheim.

Strains and growth media

S.cerevisiae strains used here were derived from SJR336 (gift from Dr Sue Jinks-Robertson, Emory University). Genotypes for constructed strains are described in Table 1.

pBluescript-E3 was a gift from Dr Mulchand S. Patel (SUNY Buffalo; 21), pRS315-GAL1/10 a gift from Dr C. Glover (University of Georgia), pYME3 a gift from Dr Lester Reed (University of Texas at Austin; 22), YEplac195 a gift from Dr R. Daniel Gietz (University of Manitoba, Canada) and pSR244 a gift from Dr Sue Jinks-Robertson (Emory University).

Bacterial cultures were grown in LB broth or on plates (1.5% agar) with 100 µg ampicillin/ml. Yeast were grown in SD-leu (2% glucose), SGly-leu (3% glycerol) selective media (23), YPGly-Glu (1% yeast extract, 2% peptone, with varying amounts of glycerol and glucose as noted in the figure legends) or YPD (1% yeast extract, 2% peptone, 2% glucose). Plates for growth of yeast strains were made with 2% agar. Large volume yeast cultures (200 ml) were grown in baffled 1 l flasks.

Construction of strains by gene disruption

Strain DMA2 was constructed by making gal80::HIS3 disruptions in strain SJR336. The NcoI-SmaI gal80::HIS3 fragment of pSR244 was used for transformation. pSR244 is a pGEM3Zf(+) backbone with the HindIII GAL80 gene fragment disrupted by replacing the internal 600 bp BglII GAL80 fragment with the 1.7 kb BamHI fragment of the HIS3 gene. Strain MML22 was constructed by making an lpd1::URA3 disruption in DMA2. The KpnI lpd1::URA3 fragment of pMLe3::URA3 was used for transformation. pYME3 was used to construct pMLe3::URA3 by replacing the 840 bp HincII fragment of LPD1 with the 1.2 kb URA3 SmaI fragment from YEplac195 (24). pYME3 is pGEM7Zf(+) with cDNA corresponding to nucleotides 65–1503 of the LPD1 gene as insert.

Cloning and detection of gene disruption by PCR

For cloning of the full-length coding region of LPD1, primers sense (5′-GGAATTCACAATGTTAAGAATC-3′) and antisense (5′-GACTAGTTTTCAACAATGAATAGC-3′) were used to amplify the yeast LPD1 gene and the product was cut with EcoRI and SpeI to ligate into EcoRI and XbaI cut pGEM3 vector, creating pML3E3. The LPD1 gene disruption with URA3 was confirmed by PCR amplification using an LPD1 specific primer with a URA3 specific primer [i.e. LPD1 sense with antisense URA3 (5′-GATTTTTCCATGGAGGGCAC-3′) and E3 antisense with URA3 sense (5′-TGTCAGATCCTGTAGAGACC-3′)].

Expression vectors and transformation

pRS315-GAL1/10 has a GAL1/10 expression cassette inserted into the EcoRI and BamHI sites in the multicloning region of the CEN plasmid pRS315 (25). Constructed plasmids for the human and yeast E3 contain the entire open reading frames of the previously reported cDNA clones (21,22). YpML7 was constructed by inserting an EcoRI (partial digest) and XhoI human cDNA fragment for E3 (from pBluescript-E3) into the SalI and EcoRI sites of the multicloning region in pRS315-GAL1/10. YpML8 was constructed by inserting the EcoRI-SalI yeast LPD1 insert from pML3E3 into the EcoRI and SalI sites of the multicloning region in pRS315-GAL1/10. YpML10 was constructed as YpML7, using pMLK37E as the source of EcoRI/XhoI insert. YpML11 was constructed by cutting pMLP453Lmat with HindIII/ApaI and ligating into HindIII/ApaI cut YpML7. All transformations were done by the lithium acetate method (26). Because pRS315-GAL1/10 vector bears the LEU2 gene, selection for transformants was done by plating on SD-leu.

Isolation of mitochondria and α-ketoacid dehydrogenase assays

Mitochondria were isolated from a cell extract produced by glass bead disruption of cells grown to stationary phase in 150–200 ml YPGlyGlu (27). The buffer was yeast busting buffer [YBB] (0.25 M mannitol, 1 mM EDTA and 50 mM Tris-HCl pH 7.4). Mitochondrial pellets were resuspended in 30 mM K_XPO₄, pH 7.5 and samples for mitochondrial protein determination were incubated at 95°C in 1% SDS for 5 min prior to the addition of BCA protein assay reagent (Pierce). Based on this protein analysis, remaining mitochondria were diluted to a final concentration of 1.5–2.0 mg protein/ml for enzyme assays. The final concentration of cofactors were 0.1 mM thiamin pyrophosphate (TPP), 1.0 mM MgCl₂, 2.5 mM NAD⁺, 0.13 mM CoA, 0.32 mM dithiothreitol (DTT) for PD complex activity; and 0.2 mM TPP, 0.2 mM MgCl₂, 0.5 mM NAD⁺, 1.3 mM CoA, 1.65 mM DTT for KGD complex activity. All reactions were incubated in 30 mM K_XPO₄ pH 7.5. Mitochondria (100 µl) and cofactors (50 µl) were added to a 1.5 ml tube with the lid cut off. This reaction tube was placed in a 1.5 ×7 cm scintillation vial containing 750 µl β-phenylethylamine as the CO₂ trap, and the vial sealed with a serum cap. After a 10 min preincubation at 37°C, the reaction was started with injection of 100 µl substrate, containing 0.1 µCi[1-¹⁴C]-substrate. The final concentration of substrate for all assays was 0.1 mM. Reactions were incubated at 37°C for 20–30 min, and terminated with the addition of 100 µl 15% TCA to release CO₂. This CO₂ was collected for 1 h at 37°C. The reaction tube was discarded, EcoLume (ICN) was added to the scintillation vial, and ¹⁴C counts representing released CO₂ were assessed by liquid scintillation counting. Since yeast PD is not regulated by phosphorylation as in mammalian tissue, conditions to fully activate the complex are not necessary during enzyme assays (28).

Assay of dihydrolipoamide dehydrogenase

Yeast were grown in 50 ml SD-leu for 18 h and collected by centrifugation. Cell lysates were made as described (18), using a vortex to disrupt the cells. Dihydrolipoamide dehydrogenase activity was assayed spectrophotometrically measuring the forward reaction with dihydrolipoamide and acetyl-NAD⁺ as substrate, as described (29), using 20–100 µg cell lysate protein per assay.

SDS-PAGE and Western blot analysis

Yeast mitochondrial proteins were resolved on 12.5% SDS-PAGE and transferred to nitrocellulose. Western blot analysis was done with a 1 h incubation in a 1:5000 dilution of antihuman E3 as primary antibody (gift from Dr B. Robinson, Hospital for Sick Children, Toronto, Canada). Secondary antibody incubation was for 30 min in a 1:50 000 dilution of goat anti-rabbit IgG(H+L) horseradish peroxidase conjugate (BioRad). Amersham ECL reagents and HyperFilm were used for detection.

Southern blot

Yeast genomic DNA was prepared (30) and digested overnight with NspI. After resolution of DNA fragments by electrophoresis in 0.8% agarose, the fragments were transferred to nylon membranes using a Turboblotter (Schleicher & Schuell). The NspI-EcoRI 1410 bp LPD1 fragment was labeled with ³²P using Megaprime (Amersham). This probe was allowed to hybridize overnight at 68°C after which the membrane was washed twice in 7×SSPE, 0.25% SDS for 15 min at 25°C followed by two washes in 1×SSPE, 0.75% SDS for 15 min at 37°C. Hybridizing fragments were detected with a Molecular Dynamics Phosphorimager.

Site directed mutagenesis by PCR

Mutation K37E (AAA to GAA) was made by two step PCR, starting with pBluescript-E3 as template. First, a 5′ product using a T7 sense primer with an E3 antisense primer (5′-TGTTTCATTTTCCTCAATGC-3′) and a 3′ product using an E3 sense primer (5′-GCATTGAGGAAAATGAAACA-3′) with a T3 antisense primer were amplified by PCR. These products were isolated in low melting point agarose, combined and used as template for a second PCR amplification with the E3 sense primer (5′-CATATGGATCCGGAAAAATGCAGAGC-3′) and T3 antisense primer. The product was a full length cDNA encoding the K37E mutation. The 313 bp BamHI-NcoI fragment of this product was cloned into BamHI/NcoI cut pBluescript-E3 to create pMLK37E. Nucleotide sequence was confirmed by the use of Sequenase 2.0 (Amersham).

Similarly, the P453L (CCG to CTG) mutation was made by two step PCR using the T7 sense primer with the E3 antisense primer (5′-GCCTCTGATAAGGTCAGATGTGCA-3′), and the E3 sense primer (5′-ATCTGACCTTATCAGAGGCTTTTA-3′) with a T3 antisense primer. As above the products were combined and used as template for amplification with T7 sense and T3 antisense primers. The 3′ EcoRI-XhoI 777 bp fragment encoding E3-P453L was cloned into EcoRI/XhoI cut pMLE3mat (pBluescript with cDNA for mature human E3) to create pMLP453L. In making this construct, a silent A₁₅₂₆G substitution occurred that eliminated the 3′ HindIII site. Nucleotide sequence was confirmed using Sequenase 2.0.

Acknowledgements

Special thanks to Sue Jinks-Robertson, Judy Fridovich-Keil and Grant MacGregor for their technical assistance and critical reading of this manuscript. MML was supported in part by a National Institutes of Health Predoctoral Training Grant GM08367 and the research was supported by an NIH grant DK38320.