Metabolic rerouting of valine and isoleucine oxidation increases survival in zebrafish models of disorders of propionyl-CoA metabolism

Abstract

Branched-chain amino acid (BCAA) oxidation is a multistep process leading to the formation of acetyl-CoA and propionyl-CoA. The syndromes associated with disturbed BCAA oxidation are clinically and biochemically heterogenous. While the common organic acidemias, propionic (PA) and methylmalonic acidemia (MMA), arise from deficient activity of propionyl-CoA carboxylase and methylmalonyl-CoA mutase and are life-threatening conditions with limited treatment options, isobutyryl-CoA dehydrogenase (IBD), and 2-methylbutyryl-CoA dehydrogenase (2-MBD) deficiencies manifest as biochemical traits, with no associated symptoms or consistent metabolic phenotypes. To assess whether the proximal interruption of valine and isoleucine oxidation might represent an approach to treat MMA and PA, we investigated the effects of loss of function of acad8 (encoding IBD) and acadsb (encoding 2-MBD), singly and doubly, on biochemical and morphological findings of zebrafish models of pccb-related propionic acidemia (PA) and mmut methylmalonic acidemia (MMA). Although acad8^−/−;acadsb^−/− double mutants showed growth failure and early mortality, the proximal interruption of valine and isoleucine oxidation in double (pccb/acad8, pccb/acadsb, mmut/acad8, mmut/acadsb) and triple (pccb/acad8/acadsb, mmut/acad8/acadsb) homozygous mutants improved pccb^−/− and mmut^−/− survival and reduced propionate-derived toxic metabolites, supporting the rationale for pursuing modulation of IBD and 2-MBD activity as a strategy to reduce the metabolic load and improve clinical outcomes in PA and MMA.

Introduction

Oxidation of the branched chain amino acids (leucine, isoleucine and valine) is a multistep process occurring mainly in the mitochondrial matrix and eventually leading to the formation of succinyl-CoA and acetyl-CoA, important Krebs cycle intermediates (Fig. 1A). Transamination of BCAAs by branched-chain amino acid transferase (BCAT1/2) yields branched-chain ketoacids which then undergo irreversible oxidative decarboxylation by the branched-chain alpha-ketoacid dehydrogenase complex. While early steps are shared by all three BCAAs, the oxidative pathways subsequently diverge to include acyl-CoA dehydrogenase family 8 (IBD encoded by ACAD8) and short/branched acyl-CoA dehydrogenase (2-MBD encoded by ACADSB), which are specific to the valine or isoleucine catabolism, respectively (Fig. 1A). Ultimately, propionyl-CoA derived from isoleucine and valine is converted to succinyl-CoA via the actions of propionyl-CoA carboxylase (PCC) [1], methylmalonyl-CoA epimerase (MCEE), and finally, the cobalamin-dependent enzyme, methylmalonyl-CoA mutase (MMUT) (Fig. 1A) [2].

Figure 1.

Zebrafish pccb^−/− and mmut^−/− recapitulate human propionic acidemia (PA) and methylmalonic acidemia (MMA) phenotypes. (A) Oxidation of BCAA amino acids in mitochondria. Under nutritionally replete conditions, some essential amino acids—valine, isoleucine, methionine, threonine—, can be oxidized to form propionyl-CoA. acad8 and acadsb encoding isobutyryl-CoA (C4) and 2-methylbutyryl-CoA (C5) dehydrogenases, respectively, catalyze corresponding intermediate steps in the oxidation of valine and isoleucine. Propionyl-CoA carboxylase composed of subunits pcca and pccb converts propionyl-CoA to D-methylmalonyl-CoA. D-methylmalonyl-CoA is converted to L-methylmalonyl-CoA by methylmalonyl-CoA epimerase encoded by mcee. In the final step, methylmalonyl-CoA mutase (mmut) converts L-methylmalonyl-CoA to succinyl-CoA, which then enters the Krebs cycle. (B) Genomic structure of the first five exons of zebrafish pccb in chromosome 2 (ch2) has 16 exons in total. Relative exon-intron distances are depicted based on zebrafish genome sequence (sequence ID: NC_007113.7). Exon 1 contains the mitochondrial leader (MTL) sequence (green) and CRISPR/Cas9 target found in exon 3 (red). DNA sequencing revealed that a 17 bp insertion (yellow highlighted red) after CRISPR/Cas9 mutagenesis, and this insertion causes a frame shift (fs) which results in the non-specific addition of 44 amino acids (yellow highlighted in red) before the stop codon. Single guide RNA (sgRNA) is shown in grey highlight with protospacer adjacent motif (PAM) site underlined. (C) Genomic organization of the first four exons of zebrafish mmut in ch20 which has 13 exons in total. Relative zebrafish exon-intron distances are depicted (sequence ID: NC_007131.7). Exon 2 has a MTL sequence and was selected for knock-out zebrafish mmut by using zinc finger nucleases (ZFNs), the cleavage sites are underlined. A 10 bp deletion (yellow highlighted red) was confirmed after Sanger sequencing in exon 2 which results in a premature early stop codon. (D) Lateral view of pccb^−/− at 15 days post fertilization (dpf). No discernable phenotypic differences are found in pccb^+/− or pccb^+/+ fish, while significant developmental delay in pccb^−/−. All scale bars represent 1 mm in size. (E) Time course observation of survival ratio among pccb^+/+ (blue), pccb^+/− (red), and pccb^−/− (green) during early developing lava stages from 7 dpf to 22 dpf. A significant mortality was observed around 15 dpf with uniform lethality of pccb^−/− before 19 dpf. The total genotyped number (n) are put on the top of each bar (F) showing mortality in mmut^+/+ (blue), mmut^+/− (red), and mmut^−/− (green) from 11 dpf, with dramatic lethality at 19 dpf. (G) Total 2-methylcitrate (2-MC), a biomarker for propionic acidemia in human, in 10 samples of WT (pccb^+/−) and pccb^−/− at 15 dpf. 2-MC was elevated in pccb^−/− while no accumulation was noted in both pccb^+/− (WT) or pccb^+/+ (data not shown). (H) Measurement of methylmalonic acid (MMA), a biomarker for methylmalonic acidemia in humans, in WT (mmut^+/−) (blue) and mmut^−/− (red) at day15. MMA was elevated in in mmut^−/− while no accumulation was noted in mmut^+/− (WT) or mmut^+/+ (data not shown). E, eye; h, heart; sb, swim bladder. Unpaired t-test used for total 2-MC and MMA analysis. ****(P < 0.0001).

Disordered BCAA oxidation can cause severe metabolic disease in humans or be associated with only biochemical phenotypes. Propionic (OMIM # 606054) [3] and methylmalonic (OMIM # 609058) [4] acidemias are debilitating, life-threatening inborn errors of metabolism arising from deleterious variants in the genes encoding propionyl-CoA carboxylase (PCCA, PCCB) and deficient activity of methylmalonyl-CoA mutase, most commonly caused by MMUT variants. In contrast, biallelic pathogenic variants in ACAD8 and ACADSB result in, respectively, IBD deficiency (OMIM #611283) and 2-MBD deficiency (OMIM #610006). Newborn screening has prospectively identified infants with IBD deficiency and 2-MBD deficiency based on elevated C4 and C5 acylcarnitine species in dried blood spots. These individuals appear to exhibit minimal, if any, clinical symptoms, suggesting that IBD and 2-MBD deficiencies are benign biochemical traits in humans, and do not require treatment [5].

Although PA and MMA can be prospectively identified via newborn screening, FDA-approved treatments are lacking. In the absence of curative therapy, careful titration of daily intake of essential propiogenic amino acids (isoleucine, valine, threonine, and methionine) has been the main approach to nutritional therapy [6]. While dietary management can achieve a reduction in the propionate load, it is often associated with complications such as protein malnutrition, amino acid imbalance, and poor growth [7]. In addition, the over-restriction of branched-chain amino acids (BCAA) can adversely affect other functions of isoleucine and valine such as the regulation of protein synthesis, hematopoiesis, and insulin signaling [8–10].

A potential strategy to treat disorders of propionyl-CoA oxidation involves reducing the formation of metabolites generated beyond the irreversible branched-chain α-ketoacid dehydrogenase (BCKDH) step, but proximal to PCC, thereby limiting further branched-chain amino acid (BCAA) oxidation. Given that IBD (ACAD8) and 2-MBD (ACADSB) deficiencies in humans are benign, the treatment of MMA and PA via inhibition of IBD and/or 2-MBD might allow the adequate intake of nutritionally essential BCAAs while mitigating excessive propionyl- and methylmalonyl-CoA accretion.

In this study, we adopted a genetic approach to examine the effects of elimination of IBD and 2-MBD on the metabolic and clinical phenotypes of PA and MMA in zebrafish. Using genome editing, we introduced deleterious variants into zebrafish orthologues of said genes—pccb, mmut, acadsb and acad8—and characterized the corresponding homozygous mutants. Very much like patients with the respective deficiencies, pccb and mmut mutants recapitulated key aspects of PA and MMA including growth retardation, lethality, and increased propionyl- and methylmalonyl-CoA derived metabolites, while acad8^−/− and acadsb^−/− fish were normal appearing with only biochemical manifestations. acad8^−/−;acadsb^−/− double mutants showed growth failure and high mortality 3 weeks post fertilization suggesting that simultaneous interruption of both valine and isoleucine oxidation could be deleterious. By crossing pccb and mmut mutant lines with those of either acad8 or acadsb, we additionally demonstrate that sequential interruption of valine or isoleucine oxidation can reduce the propionate and methylmalonate load as well as improve growth and survival of the pccb^−/− and mmut^−/− mutants. Our results support the further exploration of targeted modulation of ACAD8 and ACADSB to treat MMA and PA.

Results

pccb and mmut mutants recapitulate biochemistry and pathology of PA and MMA

Using genome editing, we generated zebrafish models of PA and MMA by introducing loss-of-function mutations into pccb and mmut, respectively. Pccb is located on chromosome 2 (Fig. 1B) and shares a high percentage of amino acid identity/similarity to its murine (92.4/98.3%) and human (86.6/94.3%) orthologues (Table S1). To avoid interfering with a potential mitochondrial target leader (MTL) sequence in exon 1, a guide for CRISPR/Cas9 was designed to target exon 3. A 17-bp insertion allele resulting in a predicted frameshift and leading to a premature stop codon (NM_212925.1, c.397_398ins ACAATGACTGGTCTAAT, p.(Arg133Asnfs45)) was identified by DNA sequencing (Fig. 1B).

A mmut knockout line was created using zinc finger nucleases (ZFNs) targeting exon 2 of mmut located on chromosome 20 (Fig. 1C). Like pccb, mmut shares a high percentage of identity/similarity of amino acids to its murine (94.8/98.7%) and human (84.1/94.7%) orthologues (Table S1). Zebrafish mmut exon 2 contains a potential MTL, therefore ZFNs were designed to avoid targeting these sequences. A recovered 10-bp deletion was predicted to result in a premature stop codon at amino acid 109 (NM_001099226.1 c.326_335del, p.(Tyr109Ter)).

We crossed both the pccb and mmut zebrafish to homozygosity, tracked survival and evaluated for gross morphological phenotypes during the embryonic and larval stages. Early studies of mortality and development showed that pcca^−/− and pccb^−/− zebrafish were indistinguishable from each other, therefore all future studies were focused exclusively on the detailed characterization of pccb^−/− animals and their unaffected clutch mates. Similar to the pccb^+/+ animals, pccb^+/− carriers had no discernable phenotype. In contrast, we observed significant developmental delay of pccb^−/− fish in the early larval stages (Fig. 1D). Increased mortality was identified in pccb^−/− animals in the beginning of the larva stage after initiation of feeding measured as % lethality between ages 7 and 22 days after fertilization (dpf). The maximum survival of pccb^−/− was 17 dpf with 100% lethality observed by 19 dpf of the early larval stages (Fig. 1E).

The mmut^−/− fish demonstrated a phenotype similar to the pccb^−/− mutants. Significant developmental delay was observed in the early larval stages, while mmut^+/− fish had no discernable phenotype and were indistinguishable from mmut^+/+ zebrafish (Fig. S2A). Lethality in mmut^−/− animals was observed starting on 10 dpf and with a virtually 100% mortality by age 19 dpf (Fig. 1F). Histological studies of the mmut^−/− liver samples revealed eosinophilic inclusions (Fig. S2B, C and B’,C′) reminiscent of mega-mitochondria represented by enlarged and curved vesicles (Fig. S2D,E and D’,E’) previously noted in MMA mouse models and in the livers of MMA patients [11, 12].

Next, we confirmed the presence of biochemical features of propionic acidemia in the pccb^−/− zebrafish model by measuring 2-methylcitrate (2-MC) and propionylcarnitine (C3) at 15 dpf. The results showed a 5-fold elevation of 2-MC in pccb^−/− (P < 0.0001) and more than 10-fold elevation of C3 (P < 0.0001) compared to wild type (WT) animals (Fig. 1G and Fig. 3C). Similarly, we measured methylmalonic acid, an MMA specific biomarker, and C3 in mmut^−/− mutant zebrafish. Significant elevations of both metabolites were detected in mmut^−/− fish (P < 0.0001) compared to controls (Fig. 1H and Fig. 3F).

Figure 3.

Propionate reduction by metabolic rerouting increases pccb^−/− and mmut^−/− survival. (A) Percentage of increased survival ratio of pccb^−/− at 17 dpf. The pccb^−/−;acadsb^−/− (b) is from pccb^−/−;acad8^+/−;acadsb^−/− incross (b), and pccb^−/−;acad8^−/− from pccb^−/−;acad8^−/−;acadsb^+/− incross (c), and compared to pccb^−/− from pccb^+/− incross (a) at 17 dpf. I-III indicate the percentage of survival among single, double, and triple mutants. The total genotyped number (n) are put on the top of each bar (B, C) reduction of 2-MC and C3-acylcarnitine accumulation in double and triple mutants at 15 dpf. Measurement of 2-MC and C3-acylcarnitine (propionylcarnitine) among WT, single, double, and triple in pccb^−/−. Reduction of total 2-MC and C3-acylcarnitine among double (pccb^−/−;acad8^−/− and pccb^−/−; acadsb^−/−), and triple (pccb^−/−;acad8^−/−;acadsb^−/−) compared to pccb^−/−. (D) Percentage of increased survival ratio of mmut^−/− at 19 dpf. The mmut^−/−;acadsb^−/− is from mmut^−/−;acad8^+/−;acadsb^−/− incross (e), and mmut^−/−;acad8^−/− from mmut^−/−;acad8^−/−;acadsb^+/− incross (f), and compared to mmut^−/− from mmut^+/− incross (d) at 19 dpf. IV-VI indicate the percentage of survival among single, double, and triple mmut^−/−. n represents fluorescent PCR genotyped total number of zebrafish. (E, F) reduction of MMA and C3-acylcarnitine among double (mmut^−/−;acad8^−/− and mmut ^−/−; acadsb^−/−), and triple (mmut ^−/−; acad8^−/−;acadsb^−/−) mutants compared to mmut^−/− at 15 dpf. Increased survival of pccb^−/− and mmut^−/− are associated with a reduction of toxic metabolite concentrations. Chi-squared analysis (asterisk in panels A and D) revealed significantly higher percentage of pccb^−/− and mmut^−/− surviving homozygous animals on the double-mutant acad8^−/− and acadsb^−/− backgrounds. p values: ns, not significant; * < 0.05; ** < 0.005; **** < 0.0001. WT in B and C are pccb^+/− and in E and F are mmut^+/−.

Collectively, the genomic, phenotypic, histological, and biochemical data presented here demonstrate that pccb^−/− and mmut^−/− zebrafish recapitulated key features of PA and MMA seen in humans including a high lethality in early larva stage, developmental delay and increased levels of toxic disease-defining metabolites.

acad8^−/− and acadsb^−/− zebrafish show no discernable phenotype while acad8^−/−;acadsb^−/− double mutants demonstrate lethality

We hypothesized that blocking enzymes involved in valine and isoleucine catabolism pathway upstream of pcc and mmut (Fig. S2A) may reduce the generation of toxic metabolites and increase the survival of pccb^−/− and mmut^−/− mutants (Fig. 2). We generated acad8 and acadsb mutant lines and analyzed phenotypic and biochemical outcomes. In humans, IBD encoded by ACAD8 converts isobutyryl-CoA to methylacrylyl-CoA (Fig. 1A).

Figure 2.

acad8^−/−;acadsb^−/− fish have a discernable phenotype with increased metabolite accumulation. (A) Lateral view of acad8^−/− and acadsb^−/− fish at 6 months. No discernable phenotype in acad8^−/− and acadsb^−/− was noted. (B) C4-acylcarnitine and C5-acylcranitine measurements in liver at 6 month old adult zebrafish. In spite of no phenotypic differences, C5-acylcarnitine was elevated in acadsb^−/−. (C) Lateral view of acad8 and acadsb double mutants at 29 dpf. No discernable phenotypes were found in both acad8^+/−;acadsb^−/− and acad8^−/−;acadsb^+/− but significant developmental delay in acad8^−/−;acadsb^−/− compound mutants was noted. Scale represents 1 mm in size. (D) Mortality of acad8 ^−/−;acadsb^−/− compound mutant from two different incrosses; acad8^+/−;acadsb^−/− and acad8^−/−;acadsb^+/−. Lethality of acad8^−/−;acadsb^−/− from two different crosses begins equally at 29 dpf with significant loss at 45 dpf. (E) Elevation of C4-acylcarnitine levels in acad8^−/−;acadsb^−/− at 29 dpf. C4 acylcarnitine levels corresponding to the expected inability to process isobutyryl-CoA (the substrate for acad8) were significantly elevated in acad8^−/−;acadsb^−/− zebrafish, but very mildly increased in acad8^−/−. (F) Elevation of C5-acylcarnitine levels in acadsb^−/− and acad8 ^−/−;acadsb ^−/− at 29 dpf. C5 acylcarnitine levels corresponding to the expected inability to process 2-methylbutyryl-CoA (the substrate for acadsb) and were elevated in acadsb^−/− and even higher in acad8^−/−; acadsb^−/− fish. (G, H) Predicted zebrafish binding pocket model. IBD can accommodate its native substrate, isobutyryl-CoA, but due to its relatively small size, it is unable to bind a bulkier molecule of 2-methylbutyryl-CoA. The circle denotes the overhang of the additional carbon present in 2-methylbutyryl-CoA outside of the binding pocket. Based on this predictive model, a larger binding pocket of zebrafish 2MBD is predicted to accommodate both butyryl-CoA (purple) and 2-methylbutyryl-CoA. ns, not significant; **(P < 0.005); ***(P < 0.001); ****(P < 0.0001). WT line in a, B, E, and F are normal TAB5 strain.

Using CRISPR/Cas9 editing, we targeted exon 3 of the zebrafish acad8 on chromosome 15 to generate a 14 bp deletion, c.305_318del p.(Leu102HisfsTer6) (Fig. S1A), and exon 4 of the zebrafish acadsb on chromosome 12 to generate a 10 bp deletion, c.365_374del p.(Ser122CysfsTer9) (Fig. S1B). Both variants are predicted to cause a frameshift resulting in the premature stop codon and a loss of the enzymatic function. Like most human subjects ascertained through newborn screen, zebrafish models acad8^−/− and acadsb^−/− had no discernable phenotype. Animals showed normal physiological growth, development, (Fig. S3A), appearance, and reproductive behavior through the adult stage (Fig. 2A).

To confirm the presence of biochemical findings characteristic of IBD and 2-MBD deficiency, we measured C4-acylcarnitine (C4) and C5-acylcarnitine (C5) levels in tissues of both acad8^−/− and acadsb^−/− 6-month-old zebrafish. Using tandem mass spectrometry (MS/MS), we performed acylcarnitine analysis on the muscle and liver tissues in WT and mutant fish. Muscle tissue tends to have the highest levels of carnitines required for energy production during exercise, while liver is capable of fixing propionate derived from food, gut bacteria and transfer through blood [3, 13]. We found that compared to WT, the C5-acylcarnitine level in acadsb^−/− animals was increased in both liver (P < 0.0001) and muscle (P < 0.0001), while C4-acylcarnitine in either tissue was normal (Fig. 2B and Fig. S3B). In acad8^−/− liver and muscle tissues, we observed no significant increase of C4 and C5.

Given previous findings suggestive of the substrate promiscuity between IBD and 2-MBD in mammalian tissues [14, 15], we hypothesized that acad8 and acadsb double knockout zebrafish may show accumulation of both, C4 and C5. To test this hypothesis, we generated compound mutants, acad8^+/−;acadsb^−/−, acad8^−/−;acadsb^+/−, and acad8^−/−;acadsb^−/−, and analyzed their phenotype and biochemical findings. We found no discernable phenotypes in acad8^+/−;acadsb^−/− and acad8^−/−;acadsb^+/− mutants. However, acad8^−/−;acadsb^−/− double mutants generated from either acad8^+/−;acadsb^−/− or acad8^−/−;acadsb^+/− incrosses presented with developmental delay, poor growth, and lethality during the juvenile period (Fig. 2C). We observed increased lethality in acad8^−/−;acadsb^−/− animals starting at 29 dpf and found no mutants surviving past 45 dpf (Fig. 2D). acad8^−/−;acadsb^−/− zebrafish had growth impairment as demonstrated by shorter standard lengths (P < 0.0001), decreased height anterior to anal fin (P < 0.0001), and delayed melanophore development (Fig. S4A-C). Swim bladder development was not different from unaffected controls by 29 dpf (Fig. S4D). Swim bladder development was also not different between acadsb^−/− and acad8^−/−;acadsb^−/− animals (Fig. S3C-F). C4- and C5-acylcarnitines measured in homogenized whole animal extracts in acad8^−/−;acadsb^−/− at 29 dpf were significantly increased (P < 0.0001) (Fig. 2E and F). These data suggest that the loss of either IDB or 2-MBD in zebrafish does not result in a discernable phenotype, but the deficiency of both enzymes is deleterious through an additive or synergistic mechanism.

In silico modeling provides insights into substrate promiscuity of the zebrafish 2-MBD and IBD

To gain an insight into reasons for the minimal C4 elevation in acad8^−/− animals and high C4 and C5 elevations in acad8^−/−;acadsb^−/− (Fig. 2E and F), we performed in-silico protein binding analysis between a catalytic pocket and its interaction with the CoA substrates using Iterative Threading ASSEmbly Refinement (I-TASSER) [16]. Substrate switching between models of the zebrafish IBD and 2-MBD helped evaluate differences in how these enzymes interact with their native and switched substrates. We found that the predicted binding pocket of 2-MBD is sufficiently large to accommodate both substrates, isobutyryl-CoA, and 2-methylbutyryl-CoA. On the other hand, the predicted binding pocket of IBD could accommodate isobutyryl-CoA, but not 2-methylbutyryl-CoA (Fig. 2G and H). This finding is consistent with observations made in mammalian tissues and provides potential explanations for why C5 is elevated in acadsb^−/−, C4 and C5 are elevated in acad8^−/−;acadsb^−/−, and may help explain the lack of significant C4 elevation in acad8^−/− [14, 15].

Interruption of valine and isoleucine oxidation improved phenotype in pccb^−/− and mmut^−/− zebrafish

To test whether interruption of valine and isoleucine oxidation, by knocking out upstream enzymes acad8 or acadsb, could reduce propionate load and improve pccb^−/− and mmut^−/− clinical and biochemical phenotypes, we generated pccb and mmut homozygous knockouts on the acad8^+/−;acadsb^−/− or acad8^−/−;acadsb^+/− backgrounds (later referred to as double mutants). We also generated pccb and mmut homozygous knockouts on the acad8^−/−;acadsb^−/− background (referred to as triple mutants) (Fig. S5). We genotyped surviving zebrafish by fluorescent PCR and analyzed pccb^−/− and mmut^−/− mutant survival from 15 dpf.

First, we characterized the phenotype of double (acad8^−/− or acadsb^−/−) and triple (acad8^−/−;acadsb^−/−) pccb and mmut mutants (Fig. S6). Compared to single pccb^−/− and mmut^−/− mutants, pccb^−/− and mmut^−/− double mutants had decreased lethality at 17 dpf and 19 dpf respectively (Fig. 3A and D, Fig. S7). Chi-squared analysis revealed a statistically significant improved survival of pccb^−/− and mmut^−/− mutants on the double acad8^−/− or acadsb^−/− backgrounds: the percent of surviving animals was higher at 17 dpf than pccb^−/− single mutants (Fig. 3A), and at 19 dpf for mmut^−/− single mutants (Fig. 3D). We also found that mmut^−/− triple mutants lived longer than pccb^−/− triple mutants (Fig. S8). In addition to survival, growth as demonstrated by standard length and height at the anterior of anal fin, improved in both double and triple pccb^−/− and mmut^−/− mutants, while melanophore and swim bladder scores were not different (Figs S10 and S11).

Biochemical analysis in the double and triple mutants revealed a statistically significant reduction of C3-acylcarnitine and total 2-MC in the tissues of double and triple mutants (Fig. 3B and C, E and F). There was a greater reduction of the C3-acylcarnitine and total 2-MC in pccb^−/−;acadsb^−/− than in pccb^−/−;acad8^−/− double mutants and appeared to track with improved survival at 17 dpf (Fig. 3A). Similar results were observed in the mmut^−/− mutants (Fig. 3B). Methylmalonic acid and C3 levels were lower in both double and triple mmut^−/− knockouts (Fig. 3E and F). We observed no differences in the total 2-MC levels in mmut^−/− double mutants (Fig. S9).

Collectively, these data indicate that interruption of valine and isoleucine oxidation by abrogating activity of IBD (acad8^−/−) and 2-MBD (acadsb^−/−) in pccb^−/− and mmut^−/− animals improved biochemical findings, development, and survival (Fig. 4). Furthermore, our data suggest that double pccb and mmut mutants show higher survival rates on the acadsb^−/− background than on the acad8^−/− background supporting the notion that isoleucine oxidation might contribute more to the total propionate pool than valine oxidation in vivo (Fig. 3A and B, Fig. S8B) [14].

Figure 4.

Genetic interruption in BCAA oxidation and increased survival of PA and MMA mutants in zebrafish. (A) Summary of PA and MMA mutant survival in this study. The acad8^−/−;acadsb^−/− has severe phenotype compared to acad8^−/− and acadsb^−/−. The survival of pccb^−/− and mmut^−/− are improved by acad8^−/− and acadsb^−/− background-double and triple mutants. (B-C) schematic diagrams of metabolic rerouting in PA and MMA. pccb and mmut fish have metabolite accumulation in the mitochondria matrix and lethality at early larval stages. To rescue PA and MMA pathology, we applied interruption of propionyl-CoA production by blocking the upstream catabolic pathway to reduce propiogenic load through the genetic ablation of acad8 and acadsb with subsequent mitigation of metabolite burden, and increased survival of PA and MMA zebrafish.

Evidence of dysregulated posttranslational modification in pccb^−/− and mmut^−/− mutants

Mammalian models of organic acidemias develop defects in lysine post translational modifications (PTM) including propionylation and malonyl/methylmalonylation [17, 18]. We evaluated the mitochondrial PTM status in pccb^−/− and mmut^−/− zebrafish by analyzing propionylation and methylmalonylation in our zebrafish models. We found no evidence of methylmalonylation in WT and heterozygote samples but observed methylmalonated proteins in mmut^−/− mutants. Although some bands showed a reduced pattern in the double and triple mutants, overall, we found no evidence of the global reduction of propionylation and methylmalonylation events in pccb^−/− and mmut^−/− on acad8^−/−, acadsb^−/−, and acad8^−/−;acadsb^−/− backgrounds (Fig. S12).

Discussion

Zebrafish have emerged as a powerful model organism for studying the pathophysiology of human genetic disorders, including inborn errors of metabolism(IEMs). They have been used to study maple syrup urine disease (MSUD) [19, 20], Wilson disease [21, 22], cobalamin C deficiency [23–27], cystinosis [28–30], congenital disorders of glycosylation [31, 32], lysosome storage diseases [33, 34], urea cycle diseases [35], neuronal ceroid lipofuscinosis [36–38], peroxisomal disorders [39], Niemann-Pick type C1 disease [40–42], and Leigh syndrome [43, 44]. Zebrafish have also been used to test new treatments, including lipid nanoparticle mRNA therapy [45–48]. Being a vertebrate model organism, zebrafish exhibit a high degree of sequence homology with human orthologues, including those implicated in vitamin B₁₂ transport, propionyl-CoA metabolism, and BCAA oxidation [19, 20, 23, 48]. Their genetic tractability, improved with efficient genome-editing approaches, has facilitated modeling of human IEMs.

In this study, we introduced loss-of-function alleles into the zebrafish genome to generate pccb^−/−, mmut^−/−, acad8^−/−, and acadsb^−/− lines. We then interrogated the effects of interrupted isoleucine and valine oxidation on the phenotype of pccb^−/− and mmut^−/− fish by creating double (acad8^−/− or acadsb^−/−) and triple (acad8^−/− and acadsb^−/−) pccb and mmut mutants. pccb and mmut knockout fish harboring loss-of-function alleles recapitulated key metabolic changes seen in these severe organic acidemias such as massive accumulation of 2-MC in pccb^−/− mutants, methylmalonic acid in mmut^−/− mutants, and C3 in both, consistent with the biochemical phenotypes in humans. While the direct assessment of immunoreactive mmut and pccb was attempted by Western blotting, it was unsuccessful (not presented). We observed early, but not immediate, lethality of homozygous fish during the early larval stages 7 dpf for pccb^−/− and 10 dpf for mmut^−/−, likely related to the larval feeding behavior beginning at 5 dpf, when consumption of protein-rich stock feeds results in rapid accumulation of toxic metabolites leading to the death within days. Early lethality was preceded by severe growth delay, impaired development of the swim bladder, and reduced melanophores. The exact causes of demise in the mutants remains uncertain and given the age of symptom onset, mRNA complementation, which can provide rescue of developmental phenotypes, was not attempted. The zebrafish mmut^−/− phenotype was similar to that described by Luciani et al who noted developmental delay, metabolite accumulation, and abnormal mitochondria in liver and kidney in mmut^del11/del11 zebrafish [49]. Whether similar mitochondrial changes are seen in the pccb mutants is under study, with an additional focus on cardiac pathology, which is common in patients with PA compared to those with MMA.

The mechanisms leading to PA and MMA complications are incompletely understood but can be in part explained by dysregulated post translational modifications (PTMs) driven by excess propionyl-CoA and methylmalonyl-CoA [17, 50]. Using anti-methylmalonyl and anti-propionyl antibodies developed to detect methylmalonylated and propionylated lysine residues [18], we showed that aberrant protein acylation seen in murine and human tissues affected with PA and MMA is also present in zebrafish. It suggests that dysregulated PTMs in organic acidemias is a relatively conserved phenomena in vertebrates although to a lesser degree in zebrafish.

Given that pccb and mmut mutants recapitulated the fundamental clinical, biochemical and molecular manifestations of PA and MMA, we pursued an in vivo approach to test the hypothesis that interruption of isoleucine and valine oxidation in zebrafish might reduce the toxic metabolite load and improve the development and survival in PA and MMA mutants. The strategy of substrate reduction, where inhibiting the upstream enzyme results in the diversion of metabolites away from the disease-implicated enzyme or accumulation of downstream toxic metabolites, has been successfully implemented in a number of inborn errors of metabolism including tyrosinemia type 1, alkaptonuria, Gaucher disease, Niemann-Pick disease type C, and acute hepatic porphyria [51–54]. Similar pharmacological and genomic strategies are being investigated in other conditions where a significant unmet need exists, such as Fabry disease, TTR-related amyloidosis, and other diseases [55, 56].

Metabolic interruption of BCAA oxidation might help with dietary protein restriction to limit the endogenous production of toxic metabolites and could represent a novel strategy to treat PA and MMA [14, 15]. Human genetics suggests that genomic or pharmacological modulation of the enzymatic activity of IBD and 2-MBD could represent a novel strategy to treat PA and MMA and is supported by recent proof-of-concept in vitro studies [14, 15]. In contrast to the severe PA and MMUT-type MMA phenotypes, biallelic mutations in ACAD8 and ACADSB in prospectively identified patients tend to present as biochemical traits with elevated C4-carnitine and C5-carnitine in the setting of normal physiological growth and development. Consistent with these observations, zebrafish acad8^−/− and acadsb^−/− mutants showed normal survival, development, and reproductive behavior. Biochemical studies revealed a more complex biochemical phenotype: C5-acylcarnitine level was increased in acadsb^−/− while acad8^−/− animals had no apparent elevation of C4-acylcarnitine. This biochemical pattern suggested the existence of an alternate pathway for isobutyryl-CoA oxidation in zebrafish, for example via 2-methylbutyryl-CoA dehydrogenase. in-silico protein binding analysis by cross comparison of both acad8 and acadsb substrates in zebrafish IBD and 2-MBD suggests that due to its larger size, the binding pocket of acadsb can accommodate both its native substrate 2-methylbutyryl-CoA and the isobutyryl-CoA, but a smaller binding pocket of acad8 can efficiently accommodate only its native substrate isobutyryl-CoA. Substrate promiscuity may, in part, explain the lack of morphological phenotypes in acad8^−/− and acadsb^−/−, and in humans, where only one of these enzymes is deficient. Our findings in zebrafish are also in agreement with the previous reports demonstrating that mammalian 2-MBD has a greater propensity to accept alternate substrates compared to IBD [15], and that other acyl-CoA dehydrogenases can compensate for the loss of ACAD8 activity [14].

The attenuated phenotype of the acad8 and acadsb mutants created an opportunity to test the hypothesis of whether the very severe phenotype of PA and MMUT-type MMA might be alleviated through abrogation of IBD activity for valine oxidation and 2-MBD activity for isoleucine oxidation as both enzymes are located upstream of PCC and MMUT. By generating double and triple pccb^−/− and mmut^−/− mutants on the acadsb^−/−, acad8^−/−, acadsb^−/−;acad8^−/− backgrounds, we demonstrated reduced mortality; improved morphological parameters; and reduced accumulation of metabolites in pccb^−/− and mmut^−/− double homozygotes on the acadsb^−/−and acad8^−/− background. These findings indicate that metabolic rerouting of BCAA oxidation and the resulting diversion of carbon can reduce propionate load. Our data also hint that PA and MMA fish on the acadsb^−/− background appear to have a longer survival compared to the acad8^−/− background. This is in line with the previous in vitro studies in patient fibroblasts showing that isoleucine oxidation by way of 2-MBD (ACADSB) is a larger contributor to the pool of propionate in tissues [15].

Our results have implications for efforts to develop an inhibitor of IBD and 2-MBD to treat disorders of propionyl-CoA oxidation. 2-methylenecyclopropaneacetic acid (MCPA) has been reported to act as an inhibitor of SBCAD (ACADSB), isovaleryl-CoA dehydrogenase, short-chain acyl-CoA dehydrogenase and medium-chain acyl-CoA dehydrogenase, but not IBD (ACAD8), highlighting difficulties developing inhibitors highly specific for each acyl-CoA dehydrogenase [14]. We explored physiological consequences of inhibiting both IBD and 2-MBD by incrossing acad8 and acadsb mutant lines. The loss of function in both enzymes in acad8^−/−;acadsb^−/− mutants resulted in high lethality, poor growth, and significant accumulation of C4 and C5-acylcarnitines (Fig. 2C-F) suggesting that simultaneous genomic knockout or simultaneous pharmacological inhibition of both enzymes may have untoward effects in humans and support a more targeted approach to modulating the specific activity of respective dehydrogenases.

We note several limitations of our studies. While our comparative phylogenetic analysis and experimental data suggest significant evolutionary conservation of the BCAA oxidation pathway in vertebrates, a complete biochemical overlap cannot be assumed without further experiments. As an aquatic animal, zebrafish may adopt strategies to excrete toxic compounds directly into water via diffusion and active transport through epithelium not available to land vertebrates. Potential inherent enzymatic differences may in part underlie why despite improved survival and decline in the disease-defining metabolites (Fig. 3B, C, E, F), double mutant zebrafish had relatively modest elevations of metabolites (Fig. S9). Our in silico modeling of the binding pockets of human enzymes, IBD and 2-MBD (unpublished data) suggests that both enzymes can accommodate their native and non-natives substrates. This finding indicates that acad8^−/− and acadsb^−/− zebrafish should be used with caution for screening of potential enzyme inhibitors. Furthermore, stock feeds rich in protein posed a significant barrier to ensure that all zebrafish irrespective of their genotypes consumed a similar amount of protein normalized by weight. Since C4-acylcarnitine represents an isobaric mix of butyrylcarnitine and isobutyrylcarnitine and C5-acylcarnitne represents an isobaric mix of valeryl-, isovaleryl, and 2-methylbutyrylcarnitine, future studies will need to employ methods capable of quantitative analysis of the C4 and C5 acylcarnitine species.

In summary, we demonstrate that conservation of the BCAA oxidation pathway between zebrafish and mammal vertebrate organisms continues to support the utility of zebrafish as a model to study organic acidemias. Our double and triple knockout strategy to divert BCAA metabolites via genomic abrogation of acad8 and acadsb resulted in a reduction of disease-defining metabolites and was associated with a partial rescue of the lethal pccb^−/− and mmut^−/− phenotype. Reduced survival and delayed development observed in acad8^−/−;acadsb^−/− animals suggest the essential nature of the intact BCAA oxidation pathway in vertebrates. To reduce off target effects, a future strategy of pharmacologic or genomic inhibition of IBD and 2-MBD to reduce the propionyl-CoA load may be needed to selectively target respective dehydrogenases.

Materials and methods

Zebrafish husbandry

Zebrafish were obtained from the progeny of TAB5 wildtype fish. The maintenance of zebrafish lines and subsequent studies were conducted at the NHGRI Zebrafish Core Facility in accordance with the guidelines of the NIH animal care and user committee under an approved protocol G-05-5 and G-16-3, as previously described [23].

Generation of pcca, pccb, mmut, acad8, and acadsb loss of function mutations

CompoZr Zinc Finger Nucleases (ZFNs) targeting mmut were designed by Sigma and mutant zebrafish were generated as described previously [57]. CRISPR/Cas9 was used to generate pcca, pccb, acad8, and acadsb knockout mutant lines using the procedures described previously [58]. Targeted exons were designed to avoid potential mitochondrial target sequences, and each deletion or insertion was confirmed by DNA sequencing (Table S2). Briefly, synthetic mRNAs were injected into zebrafish embryos and evaluated for somatic activity. Embryos from the injection groups with active sgRNAs were grown to adulthood. Founders were outcrossed with wild-type fish to identify germline transmitting founders [57]. After tail biopsy, DNA was extracted using 25 ul of Extraction Solution (Sigma, catalog number E7526) and 7 ul of Tissue Preparation Solution (Sigma, catalog number T3073) and incubated at room temperature for 10 min. Samples were heated at 95°C for 5 min and 25 ul of Neutralization Solution B was added to each sample. DNA was diluted 1:10 using RNAse free water. Separately, 5 ul of forward and 5 ul of reverse primers (100 nM) were mixed with 5 ul of fluorescent dye (M13 56-FAM) and 485 RNAse free 1X TE solution at pH 8.0. PCR was run using 10 uL of the primer mix and AmpliTaq Gold mix (1:100 ratio, Thermo Fisher Scientific, catalog number 4311816) and 3 uL of diluted DNA. Mutant allele designations by The Zebrafish Information Network.

(https://zfin.org/action/feature/line-designations) are provided in Table S2.

Crossing strategy

acad8^−/−;acadsb^+/− males and females were crossed to generate double knockout acad8^−/−;acadsb^−/−while acad8^+/−;acadsb^−/− males and females were crossed to generate acad8^−/−;acadsb^−/− animals. Pccb or mmut triple knockout cross with acad8 and acadsb mutants are described in detail in Fig. S5 Successfully genotyped pccb^+/−;acad8^+/−;acadsb^+/− or mmut^+/−;acad8^+/−;acadsb^+/− zebrafish were in-crossed to isolate progeny with pccb^+/−;acad8^−/−;acadsb^+/− and pccb^+/−;acad8^+/−;acadsb^−/− or mmut^+/−;acad8^−/−;acadsb^+/− and mmut^+/−;acad8^+/−;acadsb^−/− genotypes. Separately, these two populations of zebrafish were in-crossed to generate double knockouts: pccb^−/−;acad8^+/§;acadsb^−/−, pccb^−/−;acad8^−/−;acadsb^+/§, or mmut^−/−;acad8^+/§;acadsb^−/−, mmut^−/−;acad8^−/−;acadsb^+/§, and triple knockouts: pccb^−/−;acad8^−/−;acadsb^−/− or mmut^−/−;acad8^−/−;acadsb^−/−. Expected ratio for double homozygote is 18.75% and 6.25% for triple homozygote. ^+/§ represents ^+/+ or ^+/−.

Developmental scoring

Each larva stage zebrafish was imaged, then fincliped for genotyping. After genotyping, normal and mutant larva fish were carefully assessed using imaging results (Leica M205 FCA) to characterize the resultant phenotypes. Morphological measurements of zebrafish larvae were as previous reported [59]: standard length (SL, measured from head to base of caudal fin), height (HAA, measured as a distance between the dorsal fin to the point anterior to the anal fin and perpendicular to the fish axis), swim bladder score (SB, based on the number of swim lobes), and melanophore scoring (M, stages of the melanophore development). Swim bladder development was scored based on the number of swim bladder lobes (none = 0, one swim bladder lobe = 1, two swim bladder lobes = 2). Melanophore development was scored based on the recognized patterns [59]. Measured parameters were compared among groups based on the genotypes. 13 individual normal and mutants were selected after genotyping, then used for developmental counting for pccb and mmut single, double, and triple mutants compared to normal larva zebrafish at 15 dpf.

Metabolite measurements

Zebrafish at 15 and 29 dpf were euthanized and tail biopsies were obtained. For methylmalonic acid (MMA), and 2-methylcitrate (2-MC) measurements, after genotyping, 10 normal or mutants were combined for each genotype, homogenized, and assayed as described [23]. The fish were homogenized with beads in 120 ul of RNAse free water over an ice bath using TissueRuptor II (Qiagen, catalog number 9002755). Samples were centrifuged at 18000 g for 10 min at 4°C and 100 ul supernatant extracts were analyzed by gas-chromatography-mass spectrometry (GC–MS/MS) (Metabolite Laboratories, Denver, CO) as previously described [23]. For tissue acylcarnitine analysis, tissues were homogenized in 200–300 μl of methanol followed by the addition of 100 μl of methanol containing isotopically labeled acylcarnitine standards (Cambridge Isotopes). Samples were vortexed, then centrifuged and the supernatant evaporated to dryness under nitrogen. Acylcarnitine species were butylated by the addition of 60 μl of 3 N HCl in n-butanol followed by incubation for 20 min at 65°C, evaporated to dryness under nitrogen, and the residue reconstituted in 100 μl of mobile phase [acetonitrile/water/formic acid (80/19.9/0.1)]. Acylcarnitines were analyzed using a Sciex 4500 tandem mass spectrometer operated in positive ion mode employing precursor ion scan for m/z 85. For identification purposes, chromatographic separation of isobaric acylcarnitine species was achieved using a Shimadzu HPLC with an Applied Biosystems AAA C18 chromatographic column (4.6 × 150 mm). Compounds were eluted with a gradient of aqueous 0.1% formic acid (mobile phase A) and 100% methanol with 0.1% formic acid (mobile phase B) from 35% B to 90% B over 40 min. Results were normalized by the corresponding sample weight. In 29 dpf zebrafish, C4-acylcarnitine (an isobaric mix of butyrylcarnitine and isobutyrylcarnitine) and C5-acylcarnitne (an isobaric mix of valeryl-, isovaleryl, and 2-methylbutyrylcarnitine) measurements for acad8^−/−, acadsb^−/−, and acad8^−/−;acadsb^−/− used four zebrafish while 10 samples of normal and mutants were measured for pccb and mmut double and triple mutants at 15 dpf, and each sample continued 20 larva zebrafish per each genotype. In acad8^−/− tissues, C4 was represented by mostly isobutyrylcarnitine and C5 is predominantly isovalerylcarnitine; in acadsb^−/− tissues, C4 was also mostly isobutyrylcarnitine but C5 was predominantly 2-methylbutyrylcarnitine.

Immunoblots

Whole fish were homogenized using sterile homogenizer tubes and pestles on ice in T-PER (ThermoFisher Scientific) supplemented with fresh protease inhibitors (cOmplete Protease Inhibitor Cocktail, EDTA-free Sigma). Homogenates were centrifuged at 16000 RCF for 10–15 min. Supernatant was collected and measured for protein content by Bradford assay. Samples were then denatured in 5x SDS loading buffer and run on SDS-PAGE gels. Resulting blots were immunoblotted with one or more of the indicated antibodies: zebrafish beta actin Antibody (NSJ bioreagents, F52293), anti-propionyl antibody (PTM Biolabs, PTM-203), and anti-methylmalonyl antibody (AP42053). Detection was performed with the Odyssey imaging system using the following secondary antibodies: IRDye 800CW Donkey anti-Rabbit IgG Secondary Antibody (LI-COR) and IRDye 680RD Donkey anti-Mouse IgG (LI-COR).

Histology

Hematoxylin and eosin (H&E) staining of nuclei and cytoplasm used mmut^−/− at 15 dpf. Larva stage zebrafish were fixed with 4% paraformaldehyde overnight at room temperature, then paraffin sectioned after dehydrated in 70% ethanol (HistoServe, Germantown, MD). Liver tissues were imaged using an Axio Imger.D2 microscope and Axion Cam HRc camera, and analyzed with Zen 2011 software (Carl Zeiss, Jena, Germany). High-resolution imaging of mitochondria used transmission electron microscopy (TEM) as previously described (NEED reference) (Electron Microscopy Core, NIH).

Statistics

Prism 9.0 (GraphPad, La Jolla, CA) was used for statistical analysis. The difference between two groups was analyzed using an unpaired t-test for normally distributed datasets and a Mann–Whitney U test for non-normally distributed datasets. For groups, analysis of variance was assessed using a one-way ANOVA followed by Tukey’s honest significance test. The association between two variables was assessed using a simple linear regression for continuous variables and logistic regression for categorical variables. Chi-squared analysis was used to assess the difference between observed (pccb^−/− and mmut^−/−) vs. expected counts (pccb^−/−;acad8^−/−, pccb^−/−;acadsb^−/− and mmut^−/−;acad8^−/−, mmut ^−/−;acadsb^−/−). p value < 0.05 was interpreted as statistically significant. Statistical significance was denoted by * < 0.05, ** < 0.005, *** < 0.001, and **** < 0.0001.

Structural Modeling of IBD and 2MBD

The protein structure prediction and structure-based functional annotation was performed by I-Tasser software (https://zhanggroup.org/I-TASSER/). Isobutyryl-CoA (ibd) and 2-methylbutyryl-CoA (2mbd) binding pockets were modeled using the previously described crystal structure of human IBD (RCSB# 1RX0) and 2MBD (RCSB# 2JIF) respectively. The predicted binding of acad8 and acadsb substrates informed how they might interact with native and switched substrates.

Conflict of interest statement: NoneNone declared.

Funding

This work was funded, in part, by the Intramural Research Program of the National Human Genome Research Institute (NHGRI) through 1ZIAHG200318–21, a grant from the Propionic Acidemia Foundation, and support from the Organic Acidemia Association.