Biomedicines, Vol. 10, Pages 3171: A Missense Variant in PDK1 Associated with Severe Neurodevelopmental Delay and Epilepsy

1. IntroductionThe pyruvate dehydrogenase complex (PDC), located in the mitochondrial matrix, is responsible for the irreversible conversion of pyruvate into acetyl-CoA upon glycolysis, fueling the tricarboxylic acid (TCA) cycle. The PDC is composed of three enzymes: pyruvate dehydrogenase or PDH (E1), dihydrolipoyl transacetylase (E2), and FAD-containing dihydrolipoamide dehydrogenase (E3) [1], and its activity is tightly regulated to reassure the energy needs of the cells are met. The key enzyme at the center of PDC activity control is PDH, whose activity is controlled by the balance between the activity of PDH kinases and phosphatases, that in turn sense the metabolic state of the cell [2,3]. PDH is a tetramer composed of 2a and 2b subunits [4], in which the a subunits are the target for phosphorylation at Ser232, Ser293, and Ser300 [5,6]. Tissue-specific pyruvate dehydrogenase kinases (PDKs) phosphorylate PDH-E1α subunits and render PDC inactive. Conversely, pyruvate dehydrogenase phosphatases (PDPs) revert that inactivation by removing the phosphate groups [7]. Four PDKs (PDK1-4) have been found in humans and, according to expression analysis in humans and rodents, they are found at highest levels in the cardiac and skeletal muscle. PDK1, PDK2, and PDK4 are also found in other organs such as the brain, pancreas, lung, or kidneys [8,9]. In the skeletal muscle, the most abundant isoforms are PDK2 and PDK4; in the brain, northern blot analysis showed that PDK2 is the most abundant isoform, followed by PDK1. Conversely, the in vitro kinetic analysis of PDKs activity showed that the isoforms with the highest kinase activity are either PDK1 or PDK3, depending on the publication, followed by PDK4 and lastly PDK2 [9,10]. Interestingly, the rate of reactivation of PDC by PDPs is dependent on the PDK that phosphorylated PDH, with the fastest dephosphorylation occurring following PDK2 activity, then PDK3, PDK4, and finally PDK1 [10]. The activity of the kinases depends on the ratio of CoA/acetyl-CoA, NAD/NADH, as well as the amount of pyruvate present, with high levels of acetyl-CoA and NAHD stimulating the activity of PDK and, consequently, inactivating PDC [2,11,12]. Furthermore, the extracellular environment also affects PDC regulation, as PDK1 transcription is directly upregulated by HIF-1 (hypoxia-inducible factor-1) during hypoxia, to favor anaerobic catabolism [13,14].Previous publications have reported that patients with dysregulated pyruvate dehydrogenase activity typically present with neurological features, such as microcephaly, developmental delay, seizures, and ataxia, that appear during early childhood. Such patients are less well suited to use sugar as a source of energy, as pyruvate is not converted to acetyl-CoA, causing an energy deficiency. Instead, pyruvate is converted to lactate by lactate dehydrogenase [15]. As neurons are a major energy consuming cell type, they are severely affected by a reduction in PDC activity [16]. To date, most of the variants affecting the PDC have been found in PDHA1 (PDH-E1α), with fewer variants found in DLAT (E2) and PDHX (E3) [17,18,19,20,21,22,23,24,25,26,27]. In two unrelated families, a hemizygous missense variant in PDK3 was associated with Charcot-Marie-Tooth disease, affecting the peripheral nervous system [28,29]. The mechanism involved hyperphosphorylation and concomitant reduced PDC activity, which impaired mitochondrial axonal transport. Furthermore, two brothers have been reported to each carry a homozygous 3-nucleotide deletion variant in PDP1 and another patient to carry a homozygous 1-nucleotide duplication with a clinical phenotype including lactic acidosis, neonatal hypotonia, and feeding difficulties [30,31,32]. Current therapies of PDC deficiency involve the administration of co-factors such as thiamine to optimize residual enzyme activity and implementing a ketogenic diet, bypassing the need for PDC activity to produce energy [26,33,34,35].

In this study, we present a girl with dysmorphic facial features, severe developmental delay, and seizures. Using trio whole genome sequencing (WGS), we found a de novo missense variant in PDK1, suggested to be the cause of the disease. The expression of the variant in zebrafish leads to a reduced number of cells with detectable mitochondrial membrane potential, a reduced muscle activity, and overall defective neurodevelopment. The findings support a role for mutant PDK1 in this patient and suggest PDK1 as a novel disease-causing gene.

3. Results 3.1. Clinical Description

The female proband was the first child to healthy, non-consanguineous parents. In retrospect, the mother recalled frequent jerks or hiccups from the fetus from early pregnancy. The routine ultrasound during pregnancy was normal. She was delivered vaginally with vacuum extraction at gestation week 41 + 1, with a birth weight and head circumference at -1SD and length at +1SD. Due to irritability, hypertonic posture, and tachypnea she was admitted to the NICU at one day of age for suspected infection. She recovered and the blood-samples did not indicate an ongoing infection. Shortly after, at two months of age, she developed feeding difficulties and frequent myoclonic jerks. The EEG initially showed a burst-suppression pattern, evolving into hypsarrhythmia already at 2 months of age. She was also noted to have a special appearance with long fingers, a prominent nasal tip, narrow face, and low-set ears. She developed a therapy resistant epilepsy within her first two months of life, with multiple seizure types including myoclonia, absences, and tonic seizures. She had a continuously pathological EEG, however, not filling the LGS EEG criteria. Her development was very slow, she was continuously irritable, and her body tone altered between severely hypertonic and hypotonic. She was on ketogenic diet from age 6 months and for the rest of her life. Concomitant with the introduction of the diet, the number of seizures were reduced. When she was about 2 years old, she was started on Q10 orally and thereafter became relatively seizure free. However, she continued to be severely affected throughout her life and did not reach any developmental milestones beyond the level of 4–6 weeks. Her head growth haltered around 7 months. MRI of the brain at 10 days, 6 months, and at 3 years of age had a normal appearance, although the latter showed some minor dilatation of her ventricles compared to the earlier examinations. At 2 years and 8 months, she was treated in the PICU for a severe pneumonia with multiorgan failure. At 3 years of age, she developed pubic hair and sweat with adult odor, although the basic endocrine laboratory tests were normal. She passed away from a second septic episode during the evaluation of this suspected precocious puberty. Extensive metabolic screening, including clinical trio WGS and analysis with an epilepsy gene panel did not reveal any cause. Muscle biopsy indicated normal ATP-production and normal function of the complexes in the respiratory chain. Muscle histology was normal in clinical routine histological stainings, however, during the autopsy, skeletal muscle investigations by electron microscopy showed the presence of electron dense inclusions (1–2 μm in size) of unknown origin.

3.2. De Novo Missense Variant in PDK1The research analysis of the entire WGS data identified a heterozygous de novo missense variant (c.1139G > A, p.Gly380Asp (OMIM 602524, NM_002610)) in pyruvate dehydrogenase kinase 1 (PDK1) on chromosome 2q31.1. The variant had not previously been reported in GnomAD [47], ExAC [48], or 1000G [49] and had a combined annotation-dependent depletion (CADD) score of 33. The affected amino acid was conserved across species according to PHAST, GERP, and phyloP. The missense substitution was predicted to be deleterious by SIFT and probably damaging by Polyphen. The combined rank score by the Karolinska Clinical Genome Analysis software was 26, indicating a high likelihood of clinical relevance [37].The Sanger sequencing of the region that was flanking the variant was performed and it confirmed the presence of the heterozygous variant in the patient but not in the parents (Figure 1A). 3.3. Lower Enzymatic Activity of PDH in Cultured Patient Fibroblasts

The activity of PDH in patient fibroblasts was measured in an enzymatic assay that revealed a slightly lower enzymatic activity 0.58 nmol/mg protein/min (ref 0.60–0.90 nmol/mg protein/min). The conditions were standard and did not involve stressing the cells.

3.4. PDK1G380D Has No Relevant Effect on the Growth of Zebrafish EmbryosTo understand how the expression of PDK1G380D affects development, the zebrafish model was used. Zebrafish have one orthologue for PDK1 sharing over 77% similarity at the amino acid level with the human protein and high conservation surrounding the amino acid affected by the variant (Figure 1B). The similarity in the protein sequence suggests that the function is also conserved between species. Rescue experiments in zebrafish embryos were performed by overexpressing wild type or mutant human PDK1 (PDK1wt and PDK1G380D) in embryos with reduced endogenous Pdk1 by using a translation blocking morpholino (pdk1-MO) [39]. Firstly, we investigated whether the presence of PDK1G380D would affect the early stages of embryonic development. The injection of pdk1-MO + PDK1G380D did not result in any visible defects in embryo development when compared to the other experimental groups (control, pdk1-MO and pdk1-MO + PDK1wt, Figure 2). Even though the larvae in the experimental groups were statistically smaller than those in the control group, the larger difference between control and pdk1-MO + PDK1wt is 161 μm, which corresponds to a 5% difference in body length. This finding, combined with a normal appearance of the larvae, suggests that the expression of mutant PDK1 does not have a major impact in the overall embryo growth during early development. 3.5. Mutant PDK1 Affects the Membrane Potential of Mitochondria and Muscle FunctionPDK1 phosphorylates the E1a component of PDH, a key component in providing the substrates necessary for the TCA cycle in the mitochondria. In order to investigate the mitochondrial network in the presence of PDK1G380D, we labeled the zebrafish embryos with a Mitotracker labelling assay. Briefly, viable mitochondria tightly regulate their membrane potential within physiological levels, which is important for ATP production and storage and, consequently, cell viability. Mitotracker, a membrane potential-dependent mitochondria dye, was therefore used to quantify the metabolically active mitochondria. Surprisingly, the quantification of positively labeled cells revealed a lower number of cells containing mitochondria with membrane potential in embryos injected with pdk1-MO and pdk1-MO + PDK1G380D when compared to the other experimental conditions for the same area in the tail of two-days-old embryos (area analyzed: 300 × 350 µm width × height, pFigure 3A, square). The ultrastructure of the mitochondria was also compared using electron microscopy and we found no difference in the total number nor the structure of the mitochondria between the experimental groups (Figure S1).The membrane potential in the mitochondrion is vital for the homeostasis of this organelle [50]. Therefore, the reduced number of cells with labeled mitochondria seen in the embryos expressing PDK1G380D suggests that the energy conversion may be affected. Since the energy supply is essential for skeletal muscle function, we assessed muscle function in zebrafish embryos expressing mutant PDK1. At early developmental stages, typically 1 day post fertilization, the zebrafish embryos present with spontaneous muscle activity driven by the spinal cord and independent of signals originating from the brain [51]. Spontaneous activity was recorded (Video S1), quantified, and revealed that embryos co-injected with pdk1-MO + PDK1G380D mRNA move significantly less times for the duration of the recording (5 min) when compared with the other experimental groups (pFigure 3B). Importantly, the reduced spontaneous activity was not due to defects in the skeletal muscle, as the tissue structure was intact in the embryos from all the experimental groups (Figure S2). These finding indicate that the loss of PDK1 results in fewer mitotracker positive cells and that this phenotype can be rescued by wildtype PDK1 albeit not mutated PDK1. Additionally, in the embryos injected with pdk1-MO + PDK1G380D, we detected reduced coiling as compared to both morpholino and wildtype. 3.6. PDK1 Is Important for Neuronal Differentiation ProgressionThe patient described here presented with severe symptoms affecting the central nervous system. To assess if a similar presentation is seen in zebrafish, we immunolabeled the neurons of embryos at early developmental stages (three days post fertilization) and found that the number of neuronal connections in the tectum was lower for the pdk1-MO + PDK1G380D group than in the other experimental groups (pFigure 3C). A reduction in the number of neuronal connections can be due to developmental delay or degeneration of the neurons. To assess this, we quantified the apoptosis in the different experimental groups and found that the increase in apoptosis in the experimental groups compared to the control did not fully explain the phenotype (Figure 3D). This increase may, to an extent, be a result of some toxicity due to the injection of the morpholino and mRNAs in the embryos and not necessarily their role in apoptosis. The findings suggest that the reduction in the number of neuronal connections at the developmental stage analyzed is mostly due to a delayed neuronal development, which may include a failure in axonal outgrowth. 3.7. Variant in PDK1 Impairs Phosphorylation of PDH-E1αPDK1 phosphorylates the E1a subunit of PDH. To further investigate the suspected hypomorphic nature of PDK1G380D, we used PS10, a compound that blocks phosphorylation of PDH-E1α [42,43], in combination with pdk1-MO and the human mRNA. The quantification of the number of cells containing mitochondria with membrane potential in the injected embryos showed that PDKwt did not rescue the phenotype in the presence of PS10, as expected, while the number of cells with labeled mitochondria was unchanged when the PS10 was co-injected with either pdk1-MO or pdk1-MO + PDKG380D (pFigure 3E). To confirm the reduction in phosphorylation by mutant PDK1 in the patient, we quantified the amount of phosphorylated PDH, PDH-E1α pSer232, in fibroblasts using Western blot (Figure 4). When compared to control samples, we found a reduction in the ratio of phosphorylated protein over total PDH-E1α. These results support the hypothesis that PDK1G380D is less able to phosphorylate PDH-E1α. 4. Discussion

We describe a patient with severe epilepsy where genetic analysis showed a de novo missense variant in PDK1, a gene not previously associated with disease. To investigate the role of PDK1G380D in energy metabolism and neuronal development, we used a zebrafish model. In zebrafish embryos injected with pdk1-MO + PDK1G380D, we found a reduction in the number of cells containing mitochondria with membrane potential, a reduced spontaneous movement, and affected neuronal development. Furthermore, our results suggest that the disease mechanism involved the reduction in the ability for PDKG380D to phosphorylate PDH-E1α, which we suggest affects the regulation of PDC activity and, therefore, the homeostasis of energy conversion in the mitochondria.

Variants in several genes involved in PDC activity have been found to cause disease. Most reported patients carry variants in PDHA1 [19,26,34,35,52,53,54,55,56], followed by DLAT (E2) [17,21], PDHX (E3) [18,23,25,27], and PDK3 [28,29,31] and to a smaller extent PDH-E1b [20] and PDP1 [22,30,32]. Here, we report for the first time a disease-causing variant in PDK1. Despite variants in the six genes coding for structural components of the PDC or enzymes that regulate PDH activity, the symptoms overlap to a great extent. In all the cases, a decrease in PDH activity is detected, including the variant in PDK3, which was reported as a gain-of-function variant [29]. It is, therefore, surprising that our patient presented with a de novo suspected hypomorphic variant in PDK1, as seen by a reduction in the phosphorylation of PDH-E1α in patient cells, but having a similar clinical presentation as the patients with a reduced PDC activity. In parallel, the studies have shown an increased expression of PDK1 in cancer cells, which deactivates the PDC by phosphorylation. This deactivation is suggested to result in a change in the cells’ energy conversion toward aerobic glycolysis instead of mitochondrial respiration, also known as the Warburg effect [57,58,59]. PDK1 is therefore a good target for cancer therapies, for which inhibitors have been tested. Several compounds able to inhibit PDK1 function resulted in either proliferation inhibition or apoptosis of cancer cells [6,60,61,62]. Even though we did not detect a significant increase in apoptosis in zebrafish embryos when expressing PDK1G380D compared to the other groups, we suggest that cells in both the patient and zebrafish embryos may be similarly affected by the lower phosphorylation of PDH. A decrease in the number of mitochondria with membrane potential in the zebrafish embryos in the presence of PDK1G380D suggests some level of regulation of mitochondrial function or cell viability. However, when the electron micrographs were analyzed, we did not find a significant difference in the number of mitochondria nor the presence of structural defects in this organelle. Several publications have reported on the absence of mitochondrial phenotypes, such as the number, localization, size, or structure [22,52], which is consistent with our results but that we do not fully understand. In embryos where the Pdk1 is reduced or when mutant PDK1 is expressed, we found a decrease in the spontaneous muscle activity, or coiling. Given that no structural defects were found in the skeletal muscle, we suggest that there may be a defect in energy conversion, likely from a decrease in functional mitochondria. As glycolysis results in less energy conversion than the mitochondria-driven OXPHOS [63], which, we suggest, accounts for the reduction in spontaneous skeletal muscle contraction.

Altogether, the compiled results suggest that the clinical phenotype of the patient as well as the phenotypes observed in the zebrafish are mediated by abnormalities in the PDC activity. However, since the available antibodies to detect the phosphorylated PDH do not work in zebrafish, we were not able to confirm this mechanism. This limitation needs to be taken into account. Another major issue is the variability of phenotypes observed in both morphant and PDK1wt overexpressing embryos. It is unlikely that the transient approach used in this study will be able to address such questions. Further experiments using patient cells and stable mutant animal models would be better suited to fully explain the mechanisms of PDK1-associated disorders.

The patients with variants in genes coding for components involved in PDC activity present with neurological symptoms, including Leigh syndrome, neurodevelopmental delay, ADHD, and brain malformations as shown by MRI. Nevertheless, the effect of such variants in neurodevelopment remains poorly understood. Previous publications have shown that PDH is highly expressed in neurons but not in astrocytes, which express preferentially lactate dehydrogenase, involved in an alternative energy conversion pathway than OXPHOS [64]. Another particularity in the brain is that the TCA cycle also produces glutamate and g-aminobutyric acid (GABA), which are important neurotransmitters for synapsis [65]. A detailed characterization of a brain-specific Pdha1 knockout mouse model showed a decrease in glutamate in the brain and decreased electrical activity, suggesting that the TCA activity is reduced [66]. A study in rats showed that glutamatergic neurons consume more energy than GABAergic neurons [67], which suggests that glutamatergic neurons are especially susceptible to the reduced TCA activity resulting from the low PDC activity. Moreover, disrupted glutamatergic activity and increased aerobic glycolysis have been associated with epilepsy [68,69], a common clinical presentation in patients, including the girl described in this work. We suggest, therefore, that the presence of PDK1G380D in the brain may affect neuronal differentiation and activity by disrupting energy and neurotransmitter synthesis.The current treatment options are limited and used to a large degree indiscriminate of the genetic diagnosis. These include the introduction to a ketogenic diet often coupled with the supplementation of thiamine, carnitine, and dichloroacetic acid (DCA) [16,30,33,34,54]. For a number of patients, the treatment is efficacious in reducing lactic acidosis and the number of seizures. However, this is not true for all patients, which either do not respond to the treatment, or it fails due to the normal progression of the disease or when patients develop other illnesses, such as infections [26,52], also observed in our patient. It is, therefore, essential to develop better therapies that not only are more efficacious but also target the specific pathology of each patient. A zebrafish knockout for dlat, which codes for the PDH-E2 protein, recapitulated the patients’ clinical presentations, such as increased lactate and pyruvate. Feeding the larvae with a ketogenic diet resulted in a rescue of the PDH-deficiency including improved swimming behavior, similar to wild type larvae [70], which strongly supports the use of this model to test new candidate therapies.

留言 (0)

沒有登入
gif