Analysis of genetic mutations using a recombinant model of the mammalian pyruvate dehydrogenase complex.
PhD thesis, University of Glasgow.
Full text available as:
The human mitochondrial pyruvate dehydrogenase complex (PDC) is a vital metabolic assembly that controls the key committed step in aerobic carbohydrate utilisation and energy production and as such is responsible for overall glucose homeostasis in man. PDC, particularly from prokaryotic sources, has been widely studied as a model system for investigating the molecular basis of cooperativity between physically and functionally linked enzymes in a metabolic pathway and the catalytic and regulatory advantages conferred by their organisation into precisely-engineered ‘molecular-machines’. Defects in human PDC have been implicated in a wide variety of genetic, metabolic and autoimmune disorders. Over 200 PDC-linked mutations have been reported to date in the human population leading to clinical symptoms of various magnitudes and manifestations, mostly in the X-linked gene for the α subunit of the E1 component.
PDC is a vast molecular machine (Mr, 9-10 MDa) composed of multiple copies of 3 distinct enzymes: pyruvate dehydrogenase (E1), dihydrolipoamide acetyltransferase (E2), dihydrolipoamide dehydrogenase (E3) and an additional structural protein known as E3-binding protein (E3BP). Central to its structural, morphological and mechanistic framework is its large oligomeric ‘core’ comprising 60 E2 and 12 E3BP polypeptide chains, arranged as pentagonal dodecahedron, to which up to 30 α2β2 E1 enzymes and 6-12 homodimeric E3 enzymes are tethered tightly, but non-covalently, at maximal occupancy. More recently a substitutional model of PDC has been proposed where the core is formed by 48 E2 and 12 E3BP with the 12 E3BP polypeptides replacing an equivalent number of E2s. The N-termini of the E2 enzyme(s) each contain two peripherally-extended lipoyl domains that exhibit great mobility and in effect, act as ‘swinging arms’ since their attached lipoic acid cofactors must visit the active sites of all the three enzymes in strict rotation during the catalytic cycle. Similarly, E3BP is a distinct E2-related polypeptide that is primarily involved in E3 integration but also displays overlapping functions with E2 as it contains a single, highly-flexible lipoyl domain that can participate in catalysis.
In this study, a model recombinant human PDC has been reconstituted from its E1, E2:E3BP and E3 components that were all overexpressed in and purified from E. coli in high yield prior to assembling spontaneously into fully-active complex in vitro. The wild-type recombinant PDC displayed similar enzymatic activity to the native complex purified from human heart. As a prelude to examining 3 novel naturally- occurring, E2-linked mutations, the recombinant PDC model was validated initially by examining the effects of alterations in its lipoylation status on its activity and overall properties and secondly by assessing the characteristics of recombinant PDC lacking the E3BP subunit. A possible enzymatic role for E3BP in promoting formation of an S6, S8-diacetylated dihydrolipoamide intermediate on E2 and E3BP was also investigated.
In the first analysis a series of mutant E2:E3BP cores were created containing all possible combinations of active and inactive lipoyl domains. These possible combinations were ++/+, -+/+, +-/+, --/+, ++/ -, -+/-, +-/- and --/-, where ‘+’ is a lipoylated and ‘-’ is a non lipoylated outer, inner or E3BP domain in the E2:E3BP cores. Eight recombinant PDC models were successfully reconstituted from these cores and analysed for their effects on PDC activity compared to wild-type complex. The data indicate that mutation of the outer or inner lipoyl domain of E2 by replacing the lipoylatable lysine by glutamine leads to a 25-35% decrease in activity; moreover the presence of an active or inactive lipoyl domain on E3BP had no detectable influence on PDC function. However, PDC, in which only active E3BP lipoyl domains were present, retained 15% of wild-type activity. These data confirm previous reports on the functional redundancy of lipoyl domains observed in bacterial, yeast and mammalian PDCs and suggest that the E2 lipoyl domains and the lipoyl domain of E3BP can act independently as effective substrates for E1, E2 and E3.
A parallel study on recombinant PDC lacking E3BP, was also consistent with previous studies in showing that it retained partial activity in the complete absence of this subunit. E3BP-deficient patients retain relatively high levels of PDC activity (15-20%) as compared to the 3-8% observed using stoichiometric amounts of E3 in our in vitro assay. However, our data also suggested that 40-50% PDC activity could be achieved in the presence of a 100-200 fold excess of E3 that may account for slightly higher activity in the patients. Interestingly, wild-type PDC activity declined by 10-30% in the presence of a large E3 excess in agreement with the idea that the E1 binding site on E2 retains a residual affinity for E3 and is able to partially displace the rate-limiting E1 enzyme when present in high amounts.
The main objective of the thesis was to apply our recombinant PDC model to conduct a detailed investigation of the molecular defects underlying three novel E2 mutations/ deletions identified at the genetic level by Dr. Garry Brown, University of Oxford in PDC-deficient patients under his care. These were two separate ‘in frame’ 3-bp deletions encoding glutamate-35 and valine-455 in the mature E2 protein and a phenylalanine-490 to leucine (F490L) substitution located near the active site of the enzyme. Full length copies of these mutant enzymes were generated by site-directed mutagenesis as well an outer lipoyl domain-GST construct housing the 35E deletion.
In the case of 35E mutant, our data from lipoylation assays, circular dichroism, tryptophan fluorescence, non-denaturing gel electrophoresis, size exclusion chromatography and cross-linking analyses indicate that the mutant lipoyl domain is misfolded and displays a pronounced tendency to form dimers or higher order aggregates, presumably via inappropriate exposure of hydrophobic surfaces. As a result, the bulk of the E2:E3BP core does not assemble properly as evidenced by its unusual subunit composition and the presence of several abnormal species including non-specific aggregates detected by analytical ultracentrifugation. Mutant E2, E2:E3BP core and reconstituted PDC showed low activity (10-20%) as compared to wild-type controls, indicating a small proportion of active core can still form under these conditions. In this study, replacement of glu-35 by aspartate or glutamine had only minor/negligible effects on E2:E3BP core assembly and reconstituted PDC activity indicating that the size or charge of amino acid at this position is not critical for normal folding and assembly.
In the valine-455 deletion study, the patient was reported to contain no immunologically-detectable E3BP, despite the apparent absence of any mutation in the PDX1 gene. Therefore, in this study, it was hypothesised that this E2-based mutation might be responsible for preventing integration of E3BP into the E2 core assembly, thereby promoting its rapid degradation. Although, this deletion resulted in reduced E2 and PDC activity (50%), it was not found to prevent E3BP integration when the mutant core was produced at either 30 C or 37 C. As this patient has no detectable levels of E3BP protein despite the presence of mRNA for this component as detected by RT-PCR studies in Oxford, the precise molecular basis for this defect remains unclear at present. The patient is currently undergoing a complete medical and clinical re-evaluation.
Recombinant mutant PDC containing the F490L mutation also loses about 50% enzymatic activity as compared to the normal or wild-type PDC confirming the role of this mutation in PDC deficiency in this patient. This phenylalanine located near to the catalytic site of E2 has been found to be responsible for substrate specificity and its substitution could be directly responsible for decreased enzymatic activity in this case. No major structural changes were observed in this mutant core.
In summary, our recombinant PDC model has proved to be of considerable benefit in enabling us to gain a more informed insight into the molecular mechanisms of pathogenesis underlying these rare E2-linked mutations, particularly in the case of the E35-E2 mutant. In the absence of the recombinant model, such detailed investigations would have proved impossible owing to the lack of access to human tissue from individual patients.
As a corollary to the main aim of the thesis, a preliminary attempt was made to create an equivalent recombinant OGDC model. OGDC is also a mitochondrial assembly that is involved in the TCA cycle and is increasingly implicated in the aetiology of various neurodegenerative diseases linked to oxidative stress including Alzheimer’s and Parkinson’s disease.
The basic organisation of the OGDC is directed by the self-assembly of 24 copies of dihydrolipoyl succinyltransferase (E2o) to form a cubic core, to which multiple copies of 2-oxoglutarate dehydrogenase (E1o) and dihydrolipoamide dehydrogenase (E3) bind non-covalently. The mammalian E2o is unusual in lacking any obvious E3 or E1o binding domain. In this study, E2o and E3 were successfully overexpressed and purified. Initially it was confirmed that E2o and E3 do not interact with each other on gel filtration although stable association of all 3 constituent enzymes occurs in the native complex. Full-length E1o was also cloned successfully although it proved impossible to achieve detectable expression in our E. coli BL21 host system.
Previous studies employing subunit-specific proteolysis have identified the extreme N-terminal segment of E1o as a key region involved in the maintenance of complex stability and integrity and is required for E2 and E3 binding. To investigate this region in more detail, three N-terminal E1o fragments of decreasing size were overexpressed, one in His-tag form (193 amino acids) and two as E1o-GST fusion proteins (166 and 83 amino acids). In co-expression, purification and gel filtration studies, it was found that all these N-terminal truncates of E1o appeared capable of interacting with E2o although problems were encountered with rapid degradation and unambiguous identification in some cases. However, Western blotting revealed conclusively that even the shortest N-terminal E1o fragment (83 amino acids) was able to enter into a stable association with E2o. Owing to time constraints and difficulties with rapid degradation and/or solubility of the E1o truncates, it remains to be determined whether this N-terminal region of E1o can also interact with E2o in a post-translational fashion and whether it is directly involved in mediating E3 binding. However, this type of approach should continue to provide additional insights in the unique subunit organisation of OGDC and is an important step towards creating a recombinant model of OGDC. This will be invaluable for future studies on an important metabolic assembly that has been increasingly implicated in disorders linked to oxidative stress and neurodegeneration.
Actions (login required)