Abstract
Inborn errors of purine metabolism are rare syndromes with an array of complex phenotypes in humans. One such disorder, adenylosuccinate lyase deficiency (ASLD), is caused by a decrease in the activity of the bi-functional purine biosynthetic enzyme, adenylosuccinate lyase (ADSL). Mutations in human ADSL cause epilepsy, muscle ataxia, and autistic-like symptoms. Although the genetic basis of ASLD syndrome is known, the molecular mechanisms driving phenotypic outcome are not. Here, we characterize neuromuscular and reproductive phenotypes associated with a deficiency of adsl-1 in Caenorhabditis elegans. Characterization of the neuromuscular phenotype reveals a disruption of cholinergic transmission affecting muscular contraction. Using genetics, pharmacological supplementation, and metabolite measurements, we correlate phenotypes with distinct metabolic perturbations. The neuromuscular defect is associated with a toxic accumulation of a purine biosynthetic intermediate whereas the reproductive defect can be ameliorated by purine supplementation, indicating differing molecular mechanisms behind the phenotypes of ASLD. Because purine metabolism is highly conserved in metazoans, we suggest that similar separable metabolic perturbations result in the varied symptoms in the human disorder and that a dual-approach therapeutic strategy may be beneficial.
Author summary
Adenylosuccinate lyase deficiency is a rare metabolic disorder that is associated with epilepsy, muscle ataxia, and autistic-like symptoms in humans. This disorder arises from mutations in adenylosuccinate lyase, an enzyme involved in purine nucleotide biosynthesis. While we understand the genetic basis of this disorder, the mechanism of pathogenesis is unknown. Moreover, the linkage between phenotype and metabolic perturbation remains unclear. We report here on neuromuscular and reproductive phenotypes caused by a deficiency of adsl-1 in Caenorhabditis elegans. For each defect, we identified a specific metabolic perturbation that causes the phenotype. The neuromuscular phenotype is associated with a toxic accumulation of a purine metabolic intermediate whereas the reproductive phenotype can be alleviated by purine supplementation. Our results point to separate molecular mechanisms as causative for the phenotypes, suggesting that there may be a similar relationship between phenotype and metabolic perturbation in humans. As such, our model suggests the use of a multi-pronged approach in humans to therapeutically target the metabolic perturbation contributing to each symptom.
Introduction
Inborn errors of purine metabolism are understudied syndromes that arise from mutation of purine biosynthetic or catabolic enzymes. Although rare, these disorders are thought to be underdiagnosed because the varied clinical symptoms mimic other disorders (1). Purine disorders can have devastating health effects and often result in early death. Not only are there few therapeutic options available to patients, but the intriguing biological mechanisms linking defects in purine biosynthesis to phenotypic outcomes have also been difficult to decipher. Our aim is to use a fast, inexpensive and yet applicable model to explore the molecular links between perturbations in purine biosynthesis and organismal physiological and behavioral outcomes and to generate therapeutic strategies for these rare and understudied syndromes.
Purine nucleotides are monomers that polymerize with pyrimidine nucleotides to form nucleic acids. They also serve critical roles in cell signaling, energy storage and transfer, and metabolic regulation (2). Purines are synthesized via two biosynthetic pathways: de novo and salvage. De novo purine biosynthesis forms purine monomers from the components of intracellular amino acids and sugars. This pathway takes eleven steps to convert ribose-5-phosphate (R5P) to inosine monophosphate (IMP), the precursor for other purine monomers (Fig 1). The salvage biosynthetic pathway uses nucleic acid constituents from the diet or purine catabolism to create new purine products.
Adenylosuccinate lyase (ADSL) is an enzyme with dual functions in de novo purine biosynthesis. It catalyzes the cleavage of succinyl groups to yield fumarate twice in de novo synthesis; it converts succinylaminoimidazole carboxamide ribotide (SAICAR) to aminoimidazole carboxamide ribotide (AICAR) and succinyladenosine monophosphate (S-AMP) to adenosine monophosphate (AMP). Adenylosuccinate lyase deficiency (ASLD) is a human syndrome associated with a spectrum of symptoms including seizures, ataxia, cognitive impairment, and autistic-like behaviors (3–5). Symptoms range in severity from mild to severe and are negatively correlated with the degree of residual ADSL activity (6). In the most extreme cases, ASLD is neonatally fatal due to prenatal growth restriction, encephalopathy, and intractable seizures (6,7). This autosomal recessive neurometabolic disorder has been reported in over 50 patients since its original characterization in 1969; for these cases, over 40 separate mutations in adenylosuccinate lyase (ADSL) are associated with the disease state (8–10).
There are competing hypotheses about the etiology of ASLD symptoms. Severity of symptoms has been positively correlated with the level of accumulation of two succinylnucleosides, SAICAr and S-Ado, in the urine and cerebrospinal fluid (9,11). These nucleosides are the dephosphorylated forms of the ADSL substrates SAICAR and S-AMP, respectively, and their accumulation in body fluids is the only biochemical marker of the disorder. Previous findings associated a lower ratio of S-Ado/SAICAr with more severe symptoms, and it was hypothesized that S-Ado is protective while SAICAr is toxic (9,11). Recent findings indicate that this ratio is not predictive of phenotype severity, but correlates to the patient’s development and age during sample collection (6). Dephosphorylation of SAICAR to SAICAr has also been proposed to be a detoxification mechanism to reduce the toxic accumulation of SAICAR in affected cells (12). Thus, questions remain about the role of ADSL substrates in disease etiology.
It is also hypothesized the blockage of purine biosynthesis specifically contributes to ASLD symptoms. Deficiency of ADSL is expected to result in decreased concentrations of purine products, particularly adenine nucleotides, due to the dual function of this enzyme in the biosynthesis of AMP. However, no deficit in purines has been detected in patients; measurements of purine levels in kidney, liver, and muscle cells of ASLD patients are normal (13). Residual activity in patients likely contributes to the conservation of purine levels. Measurements of ADSL enzyme activity indicate that 3% residual activity is sufficient to convert S-AMP to AMP; although metabolic flux is greatly hindered (13). Moreover, a reduction in ADSL activity can be circumvented via supply of purines through the salvage pathway and dietary intake (14). In this case, affected cells and tissues would be dependent on high activities of the salvage enzymes to maintain purine levels. It remains possible that a deficit in the ability to synthesize purines de novo at a specific developmental stage contributes to phenotypic outcome, but evidence in support of these hypotheses to explain ADSL phenotypes is still lacking.
The pathological mechanisms causing the disorder also remain unknown (15–17). We are interested in probing the mechanism behind the disorder using Caenorhabditis elegans, an established organism for studying metabolism and associated metabolic disorders (18). The purine metabolic pathways are highly conserved across all eukaryotes, including C. elegans (19). This level of conservation indicates the functionality of C. elegans as a model for errors of purine metabolism. In addition to metabolic conservation, C. elegans provides a well-characterized nervous system that is essential for studying symptomatic aspects of ASLD. By using a model with a simple and fully identified neural network (20), the nervous system function can be studied under conditions of ADSL depletion. Thus, C. elegans has physiological benefits that other models, such as mammalian cell culture and yeast (21,22), do not provide.
We report here on the development of C. elegans as a model for ASLD using a mutant allele and RNAi knockdown of the adsl-1 gene. Extensive analysis of locomotive and reproductive phenotypes gives insight into which biological processes are disrupted by a decrease in ADSL function. We find that altered cholinergic synaptic transmission impacts muscle function in mutants. We examine metabolite levels in control and adsl-1 animals and use pharmacological and metabolite supplementation to associate substrate accumulation and purine production with the different phenotypes of ASLD in C. elegans. We propose a similar linkage between metabolic perturbation and phenotype in humans due to the high level of conservation of de novo purine biosynthesis.
Results
We used the adsl-1(tm3328) mutant and RNAi of adsl-1 in the RNAi hypersensitive strain eri-1(mg366) (23,24) to model adenolyosuccinate lyase deficiency. The tm3328 allele is a 792 bp deletion that removes over half of the adsl-1 coding sequence, including the N-terminus. RNAi of adsl-1 results in efficient yet incomplete knockdown of message levels to approximately 20% of controls (Fig 2A). We observed reproductive, developmental and locomotion defects in both adsl-1(tm3328) and adsl-1(RNAi) animals.
Disruption of adsl-1 function results in reproductive defects and embryonic lethality
Neither adsl-1(tm3328) mutants nor animals exposed to RNAi of adsl-1 for their whole life cycle are capable of producing progeny (n>100). Compared to N2 strain control animals, the gonad arms of adsl-1(tm3328) adults appear deformed, severely shrunken, and lack any indication of mature germ cell production (S1 Fig), indicating a requirement for adsl-1 in normal gonadogenesis. To reveal processes that may require adsl-1 function acutely, we also exposed fertile egg-laying adult animals in their first day of egg-laying to RNAi. Within 24 hours, these animals display an array of phenotypes. We observed both germ cells in the proximal gonad and oocytes in double file as opposed to single file in the proximal gonad, indicating abnormal progression of oogenesis (Fig 2). Deterioration of gonad arms was also evident (Fig 2, S2 Fig). We conclude that adsl-1 is required for normal development of the gonad and is required acutely for maintenance of normal oogenesis.
Animals exposed to adsl-1(RNAi) starting in the mid-fourth larval stage produce early offspring that can be phenotypically examined. We observed a high degree of embryonic lethality (18%) in these offspring (Fig 3A). Thus, not only is oogenesis hindered when adsl-1 function is decreased, but embryonic development is disrupted as well. We also examined the adsl-1(tm3328) mutant strain for evidence of developmental lethality. The sterility of the adsl-1(tm3328) strain requires the strain to be maintained using a balancer chromosome. Because the hT2 balancer is homozygous lethal, a genotypic ratio of one adsl-1(tm3228) homozygote for every two balanced heterozygotes should segregate from the balanced heterozygote. However, only 16% of the progeny of balanced heterozygotes were homozygous adsl-1(tm3328/tm3328) mutants (Fig 3B). This altered genotypic ratio of one homozygote for every 5.3 heterozygotes indicates that 62% of the adsl-1(tm3328/tm3328) population is missing. We conclude that there is a developmental lethality for the homozygous mutants, similar to the embryonic lethality of adsl-1(RNAi).
Disruption of adsl-1 function results in neuromuscular defects
adsl-1(tm3328) and adsl-1(RNAi) animals are noticeably sluggish compared to control animals. We quantified crawling speed of adsl-1(tm3328), demonstrating that they have severely slowed locomotion (Fig 4A). Upon transfer to liquid, C. elegans will continually thrash for over 90 minutes before alternating to periods of inactivity (25). We also manually counted the thrashing rate during this active period as an indication of body wall muscle function. Thrashing rate is reduced for both adsl-1(tm3328) and adsl-1(RNAi) animals; mutants and RNAi animals exhibit a 77% and 22% reduction in thrashing rate, respectively (Fig 4B). The decreased phenotypic severity of adsl-1(RNAi) likely reflects the incomplete knockdown by RNAi (Fig 2A).
adsl-1(tm3328) animals appear uncoordinated in addition to their sluggish movement. Thus, we measured additional parameters of thrashing animals using ImageJ. The adsl-1(tm3328) animals have a 78% reduction in the average speed at which their body bends, consistent with manual counts of thrashing rates (Fig 4C). Control N2 animals exhibit an undulatory pattern of locomotion (26,27) with a normally distributed angle of bending intensity around an average of 37.8 while adsl-1(tm3328) animals display a clearly distinct distribution of bend intensities (Fig 4D). Mutant animals bend with less intensity relative to N2 controls during the majority of contractions. Despite the preference for these small bends, adsl-1(tm3328) are capable of contractions comparable to and beyond that of N2; a small proportion of mutant bends exceed the typical range of bending for N2 controls. Body bends of minimal or maximal intensity in adsl-1(tm3328) deviate greatly from the undulatory bending required for coordinated movement in C. elegans (28,29). We conclude that both pace and quality of muscle contractions is altered in adsl-1 mutant animals.
Disruption of adsl-1 function affects cholinergic signaling
We next considered the question of how disruption of adsl-1 function might result in the observed muscle contraction phenotypes. We investigated the hypotheses that hindered locomotion was caused by either disruption of the cholinergic synaptic transmission, which is required for potentiating action potential firing in body wall muscle, or the reduced response of the muscle cells to this signal (30). We assessed the functionality of pre-synaptic neurons and post-synaptic muscle tissue in the neuromuscular junction of cholinergic body wall muscles using levamisole and aldicarb. Levamisole is a cholinergic receptor agonist that stimulates body wall muscles to the point of paralysis (31–33). Because levamisole only affects postsynaptic function, resistance to levamisole is indicative of altered function in the muscle. adsl-1(tm3328) displayed mild resistance to levamisole over a five hour period of exposure. Following 24 hours of continual exposure, no difference was observed between adsl-1(tm3328) and N2 controls (Fig 5A). Resistance to aldicarb has been shown for mutants in both pre-synaptic and post-synaptic tissue (34). adsl-1(tm3328) displayed a strong resistance to aldicarb over a five hour period of exposure (Fig 5B). Following 24 hours of continual exposure, 25% of the adsl-1(tm3328) animals resisted paralysis (Fig 5B). We conclude that adsl-1 mutants exhibit a stronger resistance to aldicarb than levamisole. The mild resistance to levamisole may indicate that muscle response is suboptimal. However, the resistance to aldicarb indicates that neural transmission is significantly affected by loss of adsl-1 activity.
Reduction of adsl-1 function alters intermediate metabolite levels but has no effect on global purine levels
To investigate the hypotheses that changes in ADSL substrate or purine levels are causative of phenotypes, we quantified metabolite levels in adsl-1(RNAi) animals using LC-MS. We specifically measured the levels of both ADSL substrates, SAICAR and S-AMP, in six biological replicate samples of adsl-1(RNAi) and control eri-1 animals. There were no detectable peaks for SAICAR in any of the control RNAi replicates (Fig 6A), indicating that the amount of SAICAR is typically below the threshold for metabolite detection via our methods. In all six replicates of adsl-1(RNAi), SAICAR was easily detected (Fig 6A), indicating that there is an increase in SAICAR levels when adsl-1 is knocked down. Global levels of S-AMP are also increased in adsl-1(RNAi) compared to the control (Fig 6B). This data suggests that knockdown of adsl-1 leads to the accumulation of ADSL substrates, similar to substrate accumulation shown in humans.
We also measured global levels of purine monophosphate metabolites in adsl-1(RNAi) and control animals. Interestingly, none of these metabolites showed statistically significant changes upon adsl-1 knockdown. AMP is the only metabolite that shows a downward trend (Fig 6C) with an average 37% decrease in the adsl-1 samples, relative to controls. Levels of IMP do not exhibit any difference in adsl-1(RNAi) compared to the control (Fig 6D). Levels of XMP are more variable than that of AMP or IMP, but do not display any relative difference when comparing adsl-1(RNAi) to controls (Fig 6E). The relative levels of GMP have the largest variance of the examined metabolites for adsl-1 RNAi, but did not exhibit a statistically significant difference from the control (Fig 6F). Overall, this data indicates that there is no significant decrease in purine metabolite levels caused by a knockdown of adsl-1.
Reduced de novo synthesis contributes to the reproductive phenotype
To investigate the potential toxic effects of intermediate metabolite accumulation and the blockage of de novo purine production as causative of phenotypes, we examined the effect of both supplementation with purines and inhibition of substrate production on phenotypic outcome. Even though we detect no global deficit in purine levels, we investigated whether decreased purine production is functionally contributing to the reproductive phenotype by supplementing with purine products. This supplementation strategy would allow the purine salvage pathway to more efficiently compensate for the blockage of de novo biosynthesis. To block substrate accumulation, we used methotrexate, an antimetabolite that inhibits de novo purine biosynthesis upstream of ADSL (35,36).
Supplementation of cultures with purine products restored fertility in adsl-1(RNAi) animals. Fertility was restored to 90% of animals upon adenosine supplementation and 80% of animals upon guanosine supplementation (Fig 7A). Fecundity was also restored by supplementation with purines. Supplementation with adenosine restored fecundity to 65% of control levels and supplementation with guanosine restored fecundity to 62% of control levels (Fig 7B). Supplementation of cultures with methotrexate had no effect on the fecundity or fertility of adsl-1 RNAi animals (Fig 7B); evidence for the uptake and inhibitory effect of methotrexate is shown below. Thus, the sterility phenotype is linked to a deficit in de novo purine synthesis, and we detected no role for substrate accumulation in the fertility phenotype.
Substrate buildup contributes to the neuromuscular phenotype
We also examined the effect of methotrexate and purine supplementation on the phenotypic outcome of adsl-1(tm3328) and adsl-1(RNAi) animals using thrashing assays. Both adsl-1(tm3328) and adsl-1(RNAi) displayed improved locomotion upon methotrexate supplementation. The supplemented mutants displayed a 212% increase in thrashing rate compared to the control mutants, but are only restored to ~45% of the N2 control (Fig 8A). The attenuation of the milder phenotype of adsl-1(RNAi) is more robust than that of the mutants; these animals thrash at a rate indistinguishable from the empty vector control (Fig 8B). We then used LC-MS to quantify the effects of methotrexate supplementation on adsl-1(RNAi) animals. As expected, methotrexate supplementation results in a decrease in SAICAR levels in adsl-1(RNAi) animals (Fig 8C). In contrast, the minor increase in S-AMP observed in adsl-1(RNAi) is not significantly affected by methotrexate supplementation (Fig 8D). Thus, methotrexate supplementation effectively decreases the accumulation of SAICAR, the first ADSL substrate in the de novo pathway. We also investigated whether a deficit in purine production is functionally contributing to the neuromuscular phenotype, similar to the reproductive phenotype. Supplementation with adenosine, sufficient to restore fertility and fecundity, had no effect on thrashing rate for adsl-1(tm3328) or the N2 control (Fig 8E). We conclude from these data that SAICAR accumulation likely affects neuromuscular function of adsl-1.
Discussion
We have established C. elegans as an effective model for studying adenylosuccinate lyase deficiency (ASLD). C. elegans with reduced or eliminated function of ADSL have phenotypic and biochemical similarity to the human disorder. In both humans and C. elegans, individuals heterozygous for a mutation in ADSL are phenotypically normal, but homozygous individuals exhibit severe motor and developmental phenotypes (4,9,15). The locomotive defect in C. elegans mimics the muscle ataxia in human patients (15,16). Furthermore, metabolic analysis also shows similar substrate accumulation in whole animal lysates as in human patients (37). Phenotypic similarities were shown to be present in both adsl-1 RNAi and adsl-1(tm3328) homozygotes, creating different genetic techniques to model this disorder.
Our observations regarding the sterility phenotype of adsl-1 revealed disruption of both gonadogenesis and oogenesis. We found that the development of the gonad is severely hindered for both adsl-1(tm3328) and exposure to adsl-1(RNAi) during development. Additionally, normal oogenesis acutely requires the function of adsl-1. A decrease of adsl-1 function following the L4 larval stage has minimal effect on gonadogenesis but disrupts the progression of maturing germ cells. Interestingly, the reproductive phenotypes are of similar severity when comparing the mutant to adsl-1(RNAi) animals. Given that the RNAi knockdown is incomplete, we conclude that the reproductive system is quite sensitive to the level of adsl-1 activity for proper development.
Embryonic development is also sensitive to levels of adsl-1 activity. We can avoid the typical sterility of adsl-1(RNAi) by administering the RNAi following larval development. Under these conditions, there is significant embryonic lethality associated with adsl-1(RNAi). We also demonstrated a developmental defect in the adsl-1(tm3228) mutants as evidenced by the deficit in homozygous mutants from the progeny of the balanced heterozygotes. However, this defect cannot be specifically linked to embryonic development. The fluorescent marker associated with the balancer is not visible in eggs, preventing selection of homozygous mutant eggs from the progeny pool. Embryonic lethality is likely to contribute to lethality of adsl-1(tm3328) as it does for adsl-1(RNAi), but the possibility remains that post-embryonic lethality or failure of mutant oocytes to mature contributes to the deficit of tm3328 homozygotes in the balanced strain.
adsl-1 animals are slow and display an irregular pattern of movement, mimicking the phenotypic outcome for ASLD in humans. Our data suggest a flaw in the muscle activation strategy behind the sinusoidal motion of C. elegans (38). While locomotion was slowed for both adsl-1(tm3328) and adsl-1(RNAi), the phenotype was more severe in adsl-1(tm3328) mutants. The milder locomotory phenotype of adsl-1(RNAi) likely reflects more residual enzyme activity and a less stringent requirement for high ADSL activity in locomotion compared to gonadogenesis and fertility. Nevertheless, adsl-1(RNAi) is an excellent model of the human syndrome, paralleling both a level of residual gene activity and a neuromuscular phenotype.
Given the evidence for a disruption in the patterning of muscle activation during locomotion, we investigated cholinergic signaling as a possible cause for this locomotive phenotype. A moderate resistance to levamisole in the adsl-1 mutants indicates a variation in post-synaptic body wall function. Because levamisole can only stimulate the muscle tissue, we suggest that the resistance must arise from a defect in cholinergic receptors or the initiation of the contraction within the muscle itself. This assay reveals a tissue target that is involved in the locomotive phenotype. However, the more prominent resistance to aldicarb provides additional insight. Resistance to aldicarb can arise from defects in pre-synaptic acetylcholine release or from the post-synaptic cholinergic response. Because our levamisole assay exposed an issue with post-synaptic tissue, we examined the aldicarb resistance with this in mind. adsl-1 mutants paralyze much slower on aldicarb and are capable of resisting paralysis past 24 hours of exposure. Because the aldicarb paralysis curve does not resemble that of levamisole, we conclude that there are additional factors contributing to the aldicarb resistance in pre-synaptic cholinergic neurons. This finding suggests that altered neuromuscular transmission may contribute to the muscular ataxia observed in ADSL patients.
By measuring metabolite levels for adsl-1(RNAi) animals, we have established that knockdown of adsl-1 in C. elegans also results in metabolic similarity to the human syndrome. We did not measure metabolite levels in mutant strain because the sterility and developmental lethality associated with adsl-1(tm3328) prevents us from obtaining the large population of homozygous mutants required for LC-MS analysis. In addition to the limitations of the mutant strain, adsl-1(RNAi) was chosen for metabolomics analysis because this treatment is predicted to best model the human syndrome. The accumulation of the biosynthetic intermediates SAICAR and S-Ado during knockdown of adsl-1 closely resembles the SAICAr and S-Ado accumulation observed across numerous human patients and tissue samples (3,6,11). As such, adsl-1 knockdown in C. elegans mimics the primary diagnostic biochemical markers of ASLD in humans.
Interestingly, knockdown of adsl-1 did not significantly affect any of the purine monophosphate products of de novo synthesis. The salvage biosynthesis pathways likely contribute to homeostatic mechanisms that maintain global purine levels in the absence of efficient de novo synthesis. It is likely that these animals are recycling enough purines from their diets to accommodate for the blockage of de novo biosynthesis but this model remains to be tested. Our metabolite measurements are derived from mixed-stage, whole-animal lysates. Thus, it remains possible that certain cells, tissues or developmental stages do not successfully maintain purine levels. Increased demand for purines or low activity of the salvage enzymes could alter purine levels for specific cells or developmental stages; more affected cell types could be masked by the whole-animal scale of metabolite measurements. Even with this possibility, the global maintenance of purine monophosphate levels still suggests compensation for the blockage of de novo synthesis in adsl-1(RNAi) animals. This maintenance of global purine levels is consistent with previous findings for adenine and guanine concentrations in patients and disease models with decreased ADSL function (13,21). Once again, metabolic profiling of adsl-1(RNAi) in C. elegans mimics the findings for human patients with decreased ADSL function, indicating the effectiveness of this model for studying ASLD.
Supplementation with individual purine products results in restoration of fertility. Each supplement can be converted to IMP, the central metabolite of purine synthesis, through the salvage pathways. In this way, these supplementations can overcome the blockage of IMP biosynthesis that results from the first function of ADSL, conversion of SAICAR to AICAR. However, adenosine is the only supplement that overcomes the second blockage of ADSL function, conversion of S-AMP to AMP. For this reason, it is interesting that both of the tested purine supplements are able to independently reverse the sterility of adsl-1(RNAi). We observed that adenosine supplementation is more robust than guanosine, but guanosine is still capable of restoring fertility to a significant extent. This result suggests that compensation for the second enzymatic function of ADSL is not as crucial for restoring fertility to adsl-1(RNAi) or that residual levels of ADSL-1 more easily suffice for this biochemical step.
Fertility restoration upon purine product supplementation indicates that a decrease in de novo purine production contributes to this phenotype. Furthermore, the correlation of sterility to the blockage of de novo synthesis is also predicted to be related to a potential increased demand for purines during the rapid division in gonad development when germ cells are dividing. A high demand for purines during reproductive development may cause a gonad-specific deficit of purines that is not reflected through metabolomics analysis for whole animal lysates of mixed age. Due to the severity of ASLD in humans, reproduction is not an option, so any direct correlation with the reproductive phenotype is unknown. Despite this, the linkage of a phenotype to a blockage of de novo purine formation in C. elegans indicates some of the human symptoms may have the same linkage. For this reason, one possible therapeutic approach to ASLD would be to supplement additional purines to the diets of affected individuals in combination with a block to purine biosynthesis.
Our finding that methotrexate supplementation alleviates the locomotive defect for both adsl-1(tm3328) and adsl-1(RNAi) suggests that substrate accumulation is causative of this phenotype. Metabolomics analysis of adsl-1(RNAi) specifically suggests that SAICAR accumulation is causative of this neuromuscular phenotype. Although methotrexate is capable of improving the locomotion of adsl-1(RNAi) to that of the empty vector control, methotrexate does not fully ameliorate the locomotive phenotype of adsl-1(tm3328). It is possible that SAICAR accumulates to a greater extent in adsl-1(tm3328) than adsl-1(RNAi). In this case, methotrexate supplementation may not reduce SAICAR levels enough to fully attenuate the severe locomotive phenotype of adsl-1(tm3328). Metabolomics analysis of adsl-1(tm3328) could reveal if this is the case, but is not technically feasible at this time.
The correlation between SAICAR accumulation and the locomotive defect is particularly interesting due to phenotypic similarity to muscular ataxia in humans. Because of this correlation, our data suggests that the motor control of humans may be improved by blocking SAICAR accumulation in a similar manner. The high conservation of purine biosynthesis indicates that a therapeutic approach using de novo synthesis inhibition could alleviate symptoms in humans. While these results expand on the relevance of SAICAR accumulation to phenotype, this study provides a crucial role in understanding the linkage between metabolic disturbance and disorder phenotype.
Materials and methods
C. elegans culture and strains
Strains were maintained on OP50 Escherichia coli as food under standard conditions at 20° C (39). We used the following strains; N2, eri-1(mg366), and adsl-1(tm3328). The N2 and GR1373 eri-1(mg366) strains were obtained from the Caenorhabditis Genetics Center (CGC). adsl-1(tm3328) was obtained from the National BioResource Project in Tokyo, Japan and outcrossed three times against N2 The outcrossed, balanced strain was named HV854. This allele is homozygous sterile and was balanced with hT2, a balancer for the first and third chromosomes of C. elegans; this balancer causes pharyngeal expression of GFP (40). Non-GFP homozygous adsl-1(tm3328) animals were used in phenotypic analysis.
RNAi
The adsl-1 RNAi clone was from the C. elegans RNAi Library (Source BioScience, Nottingham, UK). RNAi feeding assays were carried out as described (23). Unless otherwise noted, we transferred mid-L4 eri-1 animals to RNAi plates and examined their progeny in assays. E. coli strain HT115 carrying the empty RNAi feeding vector (EV) L4440 was used as a control.
Metabolite Supplementation
We prepared filter-sterilized stock solutions of 22 mM methotrexate (Sigma) in DMSO and stock solutions of 117 mM adenosine (Sigma) in H2O with 10% 1 M NaOH and 150 mM guanosine (Sigma) in H2O with 25% 1 M NaOH. We added these solutions to OP50 seeded NGM plates to a final concentration of 22 μM methotrexate and 10 mM adenosine, guanosine. Following supplementation, we incubated the plates at room temperature for 1–2 days before use.
Metabolomics
LC-MS metabolomics analysis was done with the Metabolomics Core Facility at Penn State. ~50 μL of animals were collected in ddH2O, flash frozen in liquid nitrogen and stored at 80°C. 15 μL samples were extracted in 1 mL of 3:3:2 acetonitrile:isopropanol:H2O with 1 μM chlorpropamide as internal standard. Samples were homogenized using a Precellys(TM) 24 homogenizer. Extracts from samples were dried under vacuum, resuspended in HPLC Optima Water (Thermo Scientific) and divided into two fractions, one for LC-MS and one for BCA protein analysis. Samples were analyzed by LC-MS using a modified version of an ion pairing reversed phase negative ion electrospray ionization method (41).Samples were separated on a Supelco (Bellefonte, PA) Titan C18 column (100 x 2.1 mm 1.9 μm particle size) using a water-methanol gradient with tributylamine added to the aqueous mobile phase. The LC-MS platform consisted of Dionex Ultimate 3000 quaternary HPLC pump, 3000 column compartment, 3000 autosampler, and an Exactive plus orbitrap mass spectrometer controlled by Xcalibur 2.2 software (all from ThermoFisher Scientific, San Jose, CA). The HPLC column was maintained at 30°C and a flow rate of 200 uL/min. Solvent A was 3% aqueous methanol with 10 mM tributylamine and 15 mM acetic acid; solvent B was methanol. The gradient was 0 min., 0% B; 5 min., 20% B; 7.5 min., 20% B; 13 min., 55% B; 15.5 min., 95% B, 18.5 min., 95% B; 19 min., 0% B; 25 min 0% B. The orbitrap was operated in negative ion mode at maximum resolution (140,000) and scanned from m/z 85 to m/z 1000. Metabolite levels were corrected to protein concentrations determined by BCA assay (Thermo Fisher).
Quantitative RT-PCR
Mid-L4 eri-1 animals were placed on RNAi plates and RNA was isolated from mixed stage worms in the next generation using TRIZOL reagent (Invitrogen). 1 μg of RNA was converted to cDNA using the qScript cDNA Synthesis Kit (Quanta Biosciences). cDNA was diluted 1:10 and used for quantitative PCR using SYBR Green and Applied Biosciences RT-PCR machine. Three primer sets, cdc-42, tba-1, and pmp-3, were used as expression controls.
cdc-42 F: ctgctggacaggaagattacg; R: ctgggacattctcgaatgaag
tba-1 F: gtacactccactgatctctgctgaca; R: ctctgtacaagaggcaaacagccatg
pmp-3 F: gttcccgtgttcatcactcat; R: acaccgtcgagaagctgtaga
adsl-1 F: acagacaatggccgatcc; R: tgttggtttcaattccttggc
Results represent the average of two biological replicates each assayed in duplicate technical replicates.
Phenotypic Analysis
Linear Crawling Velocity
Mid-L4 hermaphrodites were aged for 1 day at 20° C prior to the assay. Individual animals were tracked as they crawled on OP50 seeded NGM plates. 30 second videos were collected on a Nikon SMZ 1500 Stereoscope using NIS-Elements software from Nikon and analyzed using ImageJ. The mean linear crawling velocity was calculated for each animal by tracing the displacement of the animal’s midpoint. The displacement of the midpoint was tracked as a vector as the animal moved in a singular direction. Once the animal changed direction, a new vector was made to track movement in that direction; this process was repeated for the length of each video. Crawling velocity was determined by dividing each vector length by the corresponding time. The velocity values from all vectors in a video were averaged and adjusted for the time-fraction of each vector within the video.
Thrashing Assay
Mid-L4 hermaphrodites of each genotype were aged for 1 day at 20° C prior to the assay. Individual animals were placed in a drop of M9 solution on the surface of an unspotted NGM plate. After 1 minute of acclimation at room temperature, thrashes, the number of body bends, were counted for 1 minute using a Nikon SMZ645 Stereoscope.
Bending Quantification
Individual animals were aged and placed in M9 solution following the same procedure as the thrashing assay. 30 second videos were collected on a Nikon SMZ1500 Stereoscope using NIS-Elements software from Nikon. Videos were analyzed using ImageJ to create ideal conditions for computer-based quantification of C. elegans locomotion. First, the video background was subtracted using the rolling ball method. The background of each video is unchanging, allowing the starting frame to be subtracted from all frames. The videos were then converted to binary by setting a threshold with the “Otsu” thresholding algorithm. Binary videos of animals were processed through the wrMTrck plugin for ImageJ. Raw data of bending angle was obtained in the BendCalc format with bendDetect set to angle; this data provides bending angle for an animal at each frame of a video.
Raw data for the rate of change in bending angle were provided on a frame by frame basis for each video. The magnitude of these values were averaged to determine the bending speed for each animal; absolute magnitude was used to combine abduction and adduction into one dataset for all types of bending. Data points for the maximum bending extent were manually selected from raw data sets for each animal. The extent of each bend was determined by recording the value at each local maximum and minimum. These turning points represent the most extreme point in a bend before movement back to the mid-line of the animal. The absolute value of each bend was used for calculation of bending extent for each animal.
Paralysis
Two days prior to the assay, we added aldicarb (Sigma) or levamisole (Sigma) to unspotted NGM plates to a final concentration of 1 mM. We allowed the plates to dry at room temperature overnight then moved them to 4° C until the time of the assay. Approximately 20 mid-L4 hermaphrodites of each genotype were aged for 1 day at 20° C prior to the assay. We placed a 10 μl spot of OP50 E. coli solution on each plate and allowed it to dry for 30 minutes, concentrating the animals in a small area. Animals of a single genotype were placed on to an aldicarb plate and scored for paralysis every 30 minutes as described (34).
Egg-laying
For the egg-laying assay, ten eri-1 animals were placed on a plate containing the desired RNAi and supplementation. For each condition, ten second generation mid-L4 animals were placed onto individual plates. The number of eggs laid by each animal was counted over a five day period.
Statistical Analysis
Two-tailed student t tests were used to determine p values when comparisons were limited between two conditions. One-way or two-way ANOVA was carried out with appropriate post-tests to determine p values between three or more experimental conditions. In LC-MS analysis, we used Welch’s two sample t test to calculate p values. We substituted all undetectable measurements with zeros to statistically compare conditions for LC-MS. In all figures: ns, not significant; *, 0.01 <p< 0.05; **, 0.001<p< 0.01; ***, p<0.001; ****, p<0.0001.
Acknowledgements
We thank A. Patterson and P. Smith in the Penn State Metabolomics Core Facility for technical assistance and advice. tm alleles were provided by the Mitani laboratory through the National Bio-Resource Project of the Ministry of Education, Culture, Sports, Science, and Technology of Japan, Japan. Other strains were provided by the Caenorhabditis Genetics Center, which is funded by National Institutes of Health Office of Research Infrastructure Programs (Grant P40 OD010440).