Comparative metabolomics in the Pah^enu2 classical PKU mouse identifies cerebral energy pathway disruption and oxidative stress

Abstract

Classical phenylketonuria (PKU, OMIM 261600) owes to hepatic deficiency of phenylalanine hydroxylase (PAH) that enzymatically converts phenylalanine (Phe) to tyrosine (Tyr). PKU neurologic phenotypes include impaired brain development, decreased myelination, early onset mental retardation, seizures, and late-onset features (neuropsychiatric, Parkinsonism). Phe over-representation is systemic; however, tissue response to hyperphenylalaninemia is not consistent. To characterize hyperphenylalaninemia tissue response, metabolomics was applied to Pah^enu2 classical PKU mouse blood, liver, and brain. In blood and liver over-represented analytes were principally Phe, Phe catabolites, and Phe-related analytes (Phe-conjugates, Phe-containing dipeptides). In addition to Phe and Phe-related analytes, the metabolomic profile of Pah^enu2 brain tissue evidenced oxidative stress responses and energy dysregulation. Glutathione and homocarnosine anti-oxidative responses are apparent Pah^enu2 brain. Oxidative stress in Pah^enu2 brain was further evidenced by increased reactive oxygen species. Pah^enu2 brain presents an increased NADH/NAD ratio suggesting respiratory chain complex 1 dysfunction. Respirometry in Pah^enu2 brain mitochondria functionally confirmed reduced respiratory chain activity with an attenuated response to pyruvate substrate. Glycolysis pathway analytes are over-represented in Pah^enu2 brain tissue. PKU pathologies owe to liver metabolic deficiency; yet, Pah^enu2 liver tissue shows neither energy disruption nor anti-oxidative response. Unique aspects of metabolomic homeostasis in PKU brain tissue along with increased reactive oxygen species and respiratory chain deficit provide insight to neurologic disease mechanisms. While some elements of assumed, long standing PKU neuropathology are enforced by metabolomic data (e.g. reduced tryptophan and serotonin representation), energy dysregulation and tissue oxidative stress expand mechanisms underlying neuropathology.

1. Introduction

PAH deficient PKU (OMIM 261600) is the paradigm treatable metabolic disease being the principal motivation for prospective newborn screening [1–3]. PKU clinical management is primarily focused on neurologic phenotypes; however, less prominent bone, vascular, and optic phenotypes are recognized [4–6]. Biochemically, blood hyperphenylalaninemia defines PKU; however, Phe over-representation extends to essentially all tissues [7]. Since the 1960’s and continuing to this day, dietary substrate reduction (i.e. Phe restriction) has served as the primary interventional means [8,9]. Cofactor therapy (Sapropterin) and alternative-pathway enzyme therapy (Palynziq) are among contemporary Phe reduction means [10,11]. Independent of the interventional strategy, treatment monitoring exclusively relies upon plasma amino acid assessment [9].

Characterization of PKU neurologic pathophysiology remains incomplete. The long accepted, albeit poorly supported, neuropathology mechanism centers upon the blood:brain barrier, the large neutral amino acid transporter (LAT1, SLC7A5 gene product), and high blood Phe concentration precipitating imbalanced cerebral amino acid transport to reduce Tyr and tryptophan (Trp) concentrations [12]. A consequence of Tyr and Trp under-representation is paucities of dopaminergic and serotonergic neurotransmitters [13,14]. A linear pathway relates Phe, Tyr, and dopamine. As PAH deficiency blocks Tyr synthesis, secondary dopamine deficiency is intuitive. Serotonergic neurotransmitter deficiency postulates Phe inhibition of tryptophan hydroxylase [15] or alternatively, imbalanced blood brain barrier amino acid transport reduces Trp availability, depriving tryptophan hydroxylase of substrate with ensuing serotonin deficiency. Shortcoming of hypotheses where neurotransmitter paucities drive PKU neuropathology lies in clinical phenotypes of genuine neurotransmitter deficiencies (e.g. tyrosine hydroxylase deficiency) where dystonia and seizures are primary presentation, not mental retardation as occurs in PKU [16]. Direct Phe toxicity is suggested as causal in microcephaly, demyelination, and other white matter disease [17]. Specifically regarding myelin, it is suggested that under the influence of hyperphenylalaninemia myelin is normally produced but the product is in some manner functionally impaired [18]. Genuine evidence supporting Phe-toxicity as a driver of white matter disease is minimal.

The breadth of altered biochemical homeostasis owing to hepatic PAH deficiency remains ill defined. In both experimental and clinical studies involving PKU patients and PKU animal models, biochemical assessment is largely limited to amino acids and monoamine neurotransmitters. To appreciate systemic implications of PKU upon biochemical homeostasis, we applied untargeted metabolomics to the Pah^enu2 classical PKU mouse. Further, metabolomics was applied to blood plasma, liver tissue, and frontal cortex brain tissue. Phe, Phe catabolites, and Phe related analytes (e.g. Phe conjugates, Phe containing peptides) were detected universally across tissues. Biochemical homeostasis of Pah^enu2 brain tissue was complex and unique. Observed were altered representation of standard amino acids (e.g. Phe, Tyr, Trp, His), amino acid catabolites, amino acid conjugates, intrinsic oxidative stress response analytes (glutathione pathway, homocarnosine pathway), and energy pathway analytes (glycolysis, oxidative phosphorylation). In Pah^enu2 brain tissue, oxidative stress is demonstrated by reactive oxygen species over-representation providing clear insight for anti-oxidative response evidenced by increased glutathione and homocarnosine pathway analytes. High-resolution respirometry performed in Pah^enu2 brain tissue mitochondria demonstrated attenuated respiratory chain complex 1 activity in response to pyruvate substrate. Pah^enu2 brain tissue metabolomics identified an increased NADH/NAD ratio being further evidence of respiratory chain complex 1 dysfunction. Summation of the brain metabolomic profile, reactive oxygen species, and respirometry gives indication PKU neuropathology involves energy deficit and oxidative stress. Among neurodegenerative diseases (e.g. Parkinson’s disease, amyotrophic lateral sclerosis), energy deficit and oxidative stress are recognized pathophysiological drivers [19,20]. Data presented herein suggests similar mechanisms are participatory in PKU. Early identified, continuously treated classical PKU patients avoid severe early onset mental retardation; however, reduced IQ compared to unaffected siblings, executive function deficit, and late onset neurologic phenotypes are common. We posit oxidative stress and energy disruption impact susceptible cell populations in the brain being a central element in PKU neuropathology. Aggressive therapy supporting cerebral energetics and reducing oxidative stress may be required to ameliorate residual PKU neurologic disease.

2. Methods and materials

2.1. Pah^enu2 and control animals

Maintenance of Pah^enu2 animals utilizes an approved protocol in the Rangos Research Center animal facility at Children’s Hospital of Pittsburgh. Experimental and control animals were generated mating heterozygous females and heterozygous males. Offspring genotyping is as described [21]. Alternative homozygous genotypes (PKU enu2/enu2, control wt/wt) were selected for experimental cohorts, each consisting of six animals (3 male, 3 female). After weaning, animals were provided standard mouse chow thereby maintaining hyperphenylalaninemia (~2200 μM Phe females; ~2000 μM Phe males) among homozygous experimental animals. Wild type littermates, provide an identical diet, have blood Phe ranging from 50 to 85 μM. Experimental and control animals were used in the fed state. While cohorts contain equal male and female representation, no effort was made to determine sex-specific difference in metabolomic profiles and other metrics assessed (reactive oxygen species, respirometry). Animal sacrifice applied CO₂ asphyxiation at 3 months of age. Blood was collected by terminal draw and plasma prepared by centrifugation. Liver and brain frontal cortex tissue were dissected. For metabolomics, samples (plasma, liver, brain) were stored at −80 °C prior to utilization. For high-resolution respirometry and reactive oxygen species assessment, a portion of frontal cortex brain tissue or liver tissue was placed on ice and used immediately as described below.

2.2. Small molecule metabolomics

Metabolon Research Services provided Pah^enu2 and C57bl/6 tissue metabolomic assessment (plasma, liver tissue, brain tissue). Data analysis applied previously described means [22–26]. Both automated and visual comparison of ion features to reference standards (retention time, mass-to-charge ratio, fragmentation mass spectra) was determined. Statistical analysis applied “R” version 2.14 where p ≤ 0.05 determined significant differences.

2.3. Oroboros high-resolution respirometry

Freshly collected Pah^enu2 frontal cortex brain tissue and C57bl/6 control frontal cortex tissue was minced and homogenized in 0.2 M mannitol, 50 mM sucrose, 10 mM KCl, 1 mM EDTA, 2.5 mM bovine serum albumin, 10 mM Hepes, pH 7.4. Homogenates were cleared by centrifugation (600 ×g) and subsequently the mitochondrial fraction was pelleted by centrifugation (10,000 ×g). The mitochondria fraction was dispersed in MiR05 mitochondrial respiration buffer [27] and used immediately.

An Oroboros Oxygraph-2 K was applied to measure mitochondrial respiration [28]. Substrate additions were cytochrome C (10 μM) and malate (2 mM) combined, followed by 5 mM ADP to stimulate state 3 respiration. Next, 5 mM pyruvate and 10 mM glutamate were added to drive Complex I and 10 mM succinate to drive combined activity of Complex I + II. Finally, rotenone was applied, inhibiting complex 1 to enable determination of complex II oxygen consumption. Oxygen consumption rates were normalized to protein content.

2.4. Reactive oxygen species measurement with 2′,7′-dichlorodihydrofluorescein diacetate (DCF)

In Pah^enu2 mouse frontal cortex brain tissue, reactive oxygen species were measured utilizing the DCF protocol [29]. Sample preparation for reactive oxygen assessment utilizes the protocol described above for Oroboros high-resolution respirometry. Spectrophotometric measurement uses excitation at 485 nm and emission at 520 nm, with normalization to protein content. Samples from six Pah^enu2 and six control C57bl/6 animals were assessed. Data analysis applied Graphpad software and student t-test.

2.5. NADP/NADPH quantification

In Pah^enu2 mouse frontal cortex tissue, NADP and NADPH were quantified with the Sigma NADP/NADPH Assay Kit (catalog number MAK321) according to the manufacturer. NADP/NADPH quantification is based on a glucose dehydrogenase cycling reaction, where NADPH probe reduction creates a fluorescent product [30]. Samples from six Pah^enu2 and six control C57bl/6 animals were assessed. Data analysis applied Graphpad software and student t-test.

3. Results

3.1. Small molecule metabolomics

Table 1 provides fold-differences of selected blood plasma, liver tissue, and brain tissue analytes where experimental Pah^enu2 samples are compared to equivalent C57bl/6 control samples. Unless specifically indicated, Tables 1 Pah^enu2 tissue analytes possess statistically significant differences (p ≤ 0.05) from controls. Analyte profiles from Pah^enu2 liver and plasma, excluding PHE, PHE-conjugates, and PHE catabolites, are not overtly abnormal. Pronounced analyte misrepresentation occurs in brain tissue. Similar to plasma and liver, Pah^enu2 brain over-represents Phe, Phe catabolites, and Phe-conjugates. However, the Pah-^enu2 brain evidences cerebral oxidative stress responses (glutathione pathway, homocarnosine pathway) and energy dysregulation (glycolysis, increased NADH/NAD ratio). Fig. 1A is a general pathway schematic for Phe metabolism relevant to all tissues assessed. Fig. 1B is a schematic of metabolomic data relating to intrinsic cerebral oxidative stress responses by homocarnosine and glutathione pathways. Homocarnosine synthesis condenses His and γ-aminobutyrate. Putrescine and N-acetylputrescine are proximal analytes in cerebral γ-aminobutyrate synthesis; putrescine is over-represented. His is over-represented and chronic over-representation is evidenced by concurrent over-representation of His conjugates (N-acetylhistidine, γ-glutamylhistidine, 1-carboxyethylhistidine) and the His catabolite histamine (Table 1). Conjugates of His (acetyl, γ-glutamyl, 1-carboxyethyl) are identical to conjugates over-represented secondary to hyperphenylalaninemia (Fig. 1A, Table 1). Metabolomics did not resolve γ-aminobutyrate. Evidence is emerging for the role of homocarnosine as a neuroprotective antioxidant. Over-representation of oxidized glutathione and cysteine-glutathione disulfide provide further evidence of cerebral oxidative stress response (Table 1).

Table 1.

Pah^enu2 metabolomics: plasma, liver, brain.

	Plasma	Liver	Brain
Phe and Related
Phenylalanine	16.78	4.10	6.05
N-acetylphenylalanine	37.72	8.99	26.98
N-formylphenylalanine	9.44	3.99	7.38
γ-glutamylphenylalanine	36.46	42.70	20.75
1-carboxyethylphenylalanine	13.90	8.72	11.47
Phenylpyruvate	283.06	21.27	ND
Phenyllactate	104.15	96.25	101.42
Phenylacetate	23.15	ND	ND
2-hydroxyphenylacetate	37.16	20.34	ND
Tyr and Related
Tyrosine	0.74	0.96 ns	0.62
N-acetyltyrosine	1.51	0.94 ns	1.37
Dopamine	ND	ND	0.83 ns
Trp and Related
Tryptophan	0.84 ns	0.95 ns	0.65
Serotonin	0.32	0.15	0.44
Kynurenine	0.58	0.67	0.27
Glutathione
Glutathione reduced	ND	0.82 ns	0.64 ns
Glutathione oxidized	0.43 ns	1.12 ns	1.21
Cysteine-glutathione disulfide	0.41 ns	1.03 ns	1.28
Cysteinylglycine	ND	0.68	0.86 ns
5-oxoproline	0.82 ns	1.00 ns	0.99 ns
Sulfocysteine	1.39 ns	ND	ND
Homocarnosine Pathway
Histidine	1.13 ns	0.92 ns	1.54
N-acetylhistidine	1.26 ns	1.12 ns	3.11
1-carboxyethylhistidine	ND	0.59	1.62
γ-glutamylhistidine	1.06 ns	1.19 ns	1.62
Histamine	1.15 ns	1.32 ns	2.09
Homocarnosine	ND	ND	1.65
Putrescine	1.00 ns	1.01 ns	1.26
N-acetylputrescine	1.27 ns	1.24 ns	1.01 ns
γ-aminobutyrate	ND	ND	ND
Glycolysis and Respiratory Complex 1
Glucose 6-phosphate	0.68	0.75	1.60
Fructose 1,6-diphosphate	ND	1.26 ns	1.90
Dihydroxyacetone phosphate	ND	1.55 ns	1.75
2,3-diphosphoglycerate	ND	ND	ND
2-phosphoglycerate	0.38	1.32 ns	ND
3-phosphoglycerate	0.52	1.47 ns	1.45
Phosphoenolpyruvate	0.40	1.25 ns	1.69
Pyruvate	1.07 ns	1.01 ns	0.93 ns
Lactate	1.00 ns	0.92 ns	1.01 ns
NADH	ND	1.06 ns	2.99
NAD	ND	0.92 ns	0.64
Lipid Related Analytes
Malonate#	1.41 ns	1.16 ns	1.24
Acetyl CoA	ND	0.79 ns	1.09 ns
3-hydroxy-3-methylglutarate	1.45	1.37 ns	0.97 ns
Cholesterol+	1.79 ns	1.08 ns	0.95 ns
Docosahexaenoate*	1.43 ns	0.95 ns	1.09 ns
Eicosapentaenoate*	1.10 ns	1.18 ns	1.44 ns

Fig. 1.

A.Phe metabolism in Pah^enu2blood, liver, and brain tissue.

+analyte not observed in liver tissue; #analyte not observed in brain tissue

B. Intrinsic cerebral anti-oxidative responses. Over-represented analytes are highlighted. Pathway analytes not identified by metabolomics or showing normal representation are in black font. Left side: Homocarnosine synthesis pathway. Chronic His over representation is demonstrated by conjugation reactions (similar to Phe conjugates see Fig. 1A) and the catabolite histamine. Right side: Glutathione pathway. ROS = reactive oxygen species

C. Cerebral energy dysregulation. Over-represented analytes are highlighted. Pathway analytes not identified by metabolomics or showing normal representation are in black font. Upper Area: Pah^enu2 frontal cortex over-represented glycolysis analytes. Lower Area: NADH over-representation with increased NADH/NAD ratio suggests respiratory chain complex 1 deficit, confirmed in Fig. 2 respirometry.

Altered representation of glycolysis and respiratory chain complex 1 analytes provide evidence of cerebral energy dysregulation. Fig. 1C provides a schematic of cerebral energy metabolism. An elevated NADH/NAD ratio suggests respiratory chain complex 1 dysregulation. Below, respirometry (Fig. 2) provides confirmatory evidence of respiratory chain complex 1 dysfunction. Glycolysis pathway analyte over-representation (Table 1) evidences glycolytic compensation for reduced cerebral oxidative energy production.

Fig. 2.

Oxygen consumption rates by respirometry using fresh mouse frontal cortex mitochondria. Black arrows indicate substrate additions; Green font indicates flux response of Complex I (CI), Complex II (CII), and combined CI + CII. Mal = malate; ADP = adenosine diphosphate; Pyr = pyruvate; Glu = glutamate; Suc = succinate; Rot = rotenone

3.2. Respirometry identifies mitochondrial dysfunction in Pah^enu2 frontal cortex

Fig. 2 provides a representative Oroboros oxygen consumption trace comparing substrate-induced respiratory activation in mitochondria prepared from frontal cortex tissue of an unmanaged Pah^enu2 mouse and a wild-type littermate. Two additional assessments (Pah^enu2 vs Control) were performed, generating identical results. Complex 1 respiration in response to pyruvate substrate is attenuated. Pah^enu2 respiration in response to glutamate is similar to control albeit starting from a lower baseline. Complex II induction by succinate substrate is similar between Pah^enu2 and wild-type controls. Upon rotenone Complex I inhibition, Pah^enu2 and wild-type mitochondria respire at similar rates, serving as further indication respiration through Complex II was not impacted in Pah^eu2 brain. These data suggest a Complex I deficit in response to pyruvate substrate to support metabolomic data (increased NADH/NAD ratio, Table 1, Fig. 1A).

Respirometry was performed with Pah^enu2 liver tissue. Substrate-induced liver mitochondria respiration through Complexes I and II was in-distinguishable between Pah^enu2 and C57bl/6 controls (data not shown). Unlike brain tissue, in the context of PAH deficiency, liver tissue has no clinical phenotype. We posit altered biochemical homeostasis induces a secondary mitochondrial deficit in one or more cell populations within affected tissue (e.g. brain) while unaffected tissue (e.g. liver) retain normal mitochondrial function. These data do not identify means by which some tissues are refractory (liver) while others are sensitive (brain, bone).

3.3. Pah^enu2 cerebral oxidative stress

Table 1 demonstrates an oxidative stress response (homocarnosine pathway, glutathione pathway) in Pah^enu2 brain tissue. Fig. 3 shows brain tissue 2′,7′-dichlorodihydrofluorescein diacetate (DCF) fluorescence from cohorts of 6 Pah^enu2 animals (3 male, 3 female) and 6 control animals (3 male, 3 female). A statistically significant increase in DCF fluorescence (p ≤ 0.001), indicating increased ROS in the Pah^enu2 brain demonstrates oxidative stress further evidenced by the anti-oxidative response (Fig. 1B, Table 1).

Fig. 3.

Reactive oxygen species in brain tissue. Brain tissue from six C57bl/6 control animals (3 male, 3 female) and six hyperphenylalaninemic Pah^enu2 animals (3 male, 3 female) were assessed. **** p≤0.001

3.4. Pah^enu2 altered cerebral NADP and NADPH representation

Glutathione reductase requires niacinamide adenine dinucleotide phosphate (NADPH) cofactor to drive anti-oxidative response by glutathione disulfide cleavage. Glucose-6-phosphate dehydrogenase, whose activity is inhibited by the Phe catabolite phenylpyruvate, is a critical source of cerebral NADPH [31]. Fig. 4 shows brain concentrations of NADP and NADPH from cohorts of 6 Pah^enu2 animals (3 male, 3 female) and 6 C57bl/6 animals (3 male, 3 female). Compared to C57bl/6, Pah^enu2 brain NADPH representation is reduced while NADP is over-represented. Table 1 shows over-representation of cysteine-glutathione disulfide the glutathione reductase substrate. Under-represented NADPH may constrain glutathione reductase activity allowing cysteine-glutathione disulfide accumulation.

Fig. 4.

Frontal cortex tissue from six Pah^enu2 and six C57bl/6 were assayed for NADP and NADPH. Fig. 4A. Quantification of NADPH. Fig. 4B. Quantification of NADP. ** p≤0.01 μM/g = micromole per gram tissue

4. Discussion

Plasma Phe concentration and the Phe/Tyr ratio are metrics applied to PKU diagnosis and following application of disease management, as the primary gauge monitoring interventional efficacy [8,9]. Measurement of primary Phe catabolites, phenylpyruvate and phenyllactate, has been performed in patients; however, these analytes are not applied in diagnostics or management monitoring. The entirety of PKU biochemical homeostasis remains poorly characterized. In the inborn error of metabolism space, metabolomics provides a natural extension to common targeted analyte assessment (e.g. acylcarnitines, amino acids). Metabolomics enables extensive characterization of biochemical homeostasis, provides pathophysiological insight, and may provide means to ascertain unrealized interventional routes [26,32,33].

Hyperphenylalaninemia is systemic in PKU, yet severe phenotypes are restricted to the brain and to a lesser extent bone. Metabolomics determined Phe, Phe catabolites, and Phe conjugates are over-represented in all tissues assessed; however, brain tissue, which realizes severe clinical phenotypes, displays a unique pattern of altered biochemical elements. Two themes are apparent in the PKU brain metabolomic profile: 1. Altered energy metabolism and 2. Anti-oxidative responses. Legacy data from the late 1960’s to mid-1970s identified aberrant PKU mitochondria function [34–37]. Contemporary investigations further a case for PKU energy dysregulation [28,38–42]. We have shown in the Pah^enu2 mouse, a brain-restricted increased NADH/NAD ratio (Table 1) suggesting respiratory chain complex 1 dysfunction where respirometry functionally confirmed reduced complex 1 activity (Fig. 2). Moreover, brain over-representation of glycolysis pathway analytes suggests reduced oxidative energy production induces glycolysis pathway compensation. Phe restriction prompts a degree of reversibility in PKU white matter disease [18]. We posit Phe management lowers oxidative stress and increases oxidative energy production enabling white matter restoration, mirroring reversible cardiac phenotypes in long chain fatty acid oxidation defects when intervention restores energy homeostasis [43].

Oxidative stress in PKU is established [44,45]. Pah^enu2 brain oxidative stress responses are apparent by glutathione and homocarnosine pathways analytes. Fig. 3 reactive oxygen species over-representation confirms cerebral oxidative stress. Glutathione is the most recognized antioxidant system, while the role of homocarnosine in oxidative neuro-protection is emerging [46,47]. Inborn errors of metabolism where oxidative stress is recognized includes peroxisomal disorders and mutase deficient methylmalonic acidemia (MMA). Respiratory complex inhibition in MMA is attributed to methylmalonate and its metabolites malonate and methylcitrate [48–53]. Dysfunction of multiple respiratory chain complexes is implicated [54–56]. Reduced glutathione is observed in MMA patients and animal models [57]. Culture studies in SHS-5Y cells suggest methylmalonate exposure reduces cellular glutathione; however, further investigation is necessary to establish in vivo cerebral effect [58]. As peroxisomal very long chain fatty acid oxidation creates hydrogen peroxide, it is logical peroxisome functional defects would precipitate oxidative stress. Mechanisms of peroxisomal disorder oxidative stress are two-fold: 1. ROS of peroxisomal origin; and 2. ROS of mitochondrial origin. Relevant to our PKU studies, in X-linked adrenoleucodystrophy hexacosanoic acid (C26:0) directly impairs oxidative phosphorylation upregulating ROS production [59,60]. Further, phytanate, the analyte over-represented in Refsum disease, is recognized to specifically inhibit respiratory chain complex 1 generating oxidative stress [61]. It would be highly relevant to compare tissue metabolomics among inborn errors for common themes (e.g. analyte patterns) related to oxidative stress to aid in ascertaining clinical significance.

Data presented herein does not provide mechanisms underlying either PKU cerebral respiratory chain deficit or oxidative stress. However, prior investigations and data provided herein provide a level of insight. The Phe catabolite phenylpyruvate is a characterized inhibitor of glucose-6-phosphate dehydrogenase (G6PD, UniProt O95479) and mitochondrial pyruvate carrier (UniProt Q9Y5U8, O95563) [31,62,63]. Fig. 2 respirometry applies substrates to activate respiratory chain complex 1 and complex 2. Control mitochondria respond robustly to pyruvate substrate, while Pah^enu2 mitochondria response is attenuated. Succinate driven respiratory complex 2 activation is similar between control and Pah^enu2 mitochondria. Attenuated response to pyruvate substrate suggests the PKU catabolite phenylpyruvate inhibits mitochondria pyruvate transport, subsequently limiting complex 1 response. Prior studies indicate phenylpyruvate inhibits G6PD [31]. In the pentose phosphate pathway, G6PD is rate limiting but equally important, G6PD produces (along with 6-phosphogluconate dehydrogenase, UniProt P52209) nicotinamide adenine dinucleotide phosphate (NADPH) the glutathione reductase (UniProt P78417) obligatory cofactor. Table 1 identifies glutathione disulfide in Pah^enu2 brain. Glutathione reductase, in an NADPH dependent manner, cleaves glutathione disulfide to glutathione monomers. Phenylpyruvate down-regulation of pentose phosphate pathway oxidative steps reduces NADPH production attenuating glutathione reductase participation in oxidative stress management. Concurrently inhibiting pyruvate transport and G6PD, phenylpyruvate is complicit in energy deficit and oxidative stress. Metabolomics resolved phenylpyruvate in liver and plasma demonstrating over-representation. Phenylpyruvate was not resolved in brain. Brain phenyllactate is increased 101-fold being the analyte immediately distal to phenylpyruvate; therefore, indicating phenylpyruvate over-representation in cerebral tissue. While our analyte and functional data support a cerebral PKU respiratory deficit restricted to complex 1, further investigation is required to characterize mechanisms underlying the energy deficit.

Metabolomics identified relatively consistent analyte profiles between blood and liver providing little evidence of energy dysregulation or oxidative stress response. Energy dysregulation and oxidative stress response is brain restricted. A recent analyte study in succinylsemialdehyde dehydrogenase deficiency determined relevant analyte disruption occurred in affected tissue and peripheral blood was a less informative source of pathological analytes [64]. We previously reported altered epigenome homeostasis in brain tissue of PKU patients and the Pah^enu2 mouse [21,65]. Oxidative stress induces epigenome re-patterning [66–68]. We theorize oxidative stress is a proximal step in PKU brain pathology precipitating down-stream elements including epigenome-mediated alteration of gene function (promoter methylation, micro-RNA). We view PKU neuropathology as far more complex than simple traditional models of asymmetric blood:brain barrier amino acid transport precipitating direct Phe tissue toxicity and neurotransmitter paucities. Patient presentation such as lower IQ compared to siblings and the plethora of late onset neurologic phenotypes evidence standard of care treatment is suboptimal. Characterizing the entirety of PKU pathology will enable more thorough and effective treatment regimens. We suggest cerebral energy support and oxidative stress reduction may ameliorate residual early onset and late onset PKU neurologic phenotypes.

Table 1. Assessment used terminal draw plasma, liver tissue, and frontal cortex brain tissue from six Pah^enu2 animals receiving standard diet. Control samples (plasma, liver, frontal cortex brain) used six C57bl/6 animals receiving standard diet. Fold change are statistically significant unless specifically indicated. ND not detected, the analyte was not identified; ns not significant, the fold change does not differ from controls; # acetylCoA ester of malonate is rate limiting in lipid [1] synthesis; + Cholesterol reported low in human PKU brain tissue. * Docosahexaenoate and eicosapentaenoate previously reported low in human PKU plasma.