- Top of page
- Materials and Methods
Purpose: Pyridoxine-dependent seizure (PDS) is a rare disorder characterized by seizures that are resistant to common anticonvulsants, and that are ultimately controlled by daily pharmacologic doses of pyridoxine (vitamin B6). Mutations of the antiquitin gene (ALDH7A1) are now recognized as the molecular basis of cases of neonatal-onset PDS.
Methods: Bidirectional DNA sequence analysis of ALDH7A1 was undertaken along with plasma pipecolic acid (PA) measurements to determine the prevalence of ALDH7A1 mutations in a cohort of 18 North American patients with PDS.
Results: In patients with neonatal-onset PDS, compound heterozygous or homozygous ALDH7A1 mutations were detected in 10 of 12 cases, and a single mutation was found in the remaining 2. In later-onset cases, mutations in ALDH7A1 were detected in three of six cases. In two patients with infantile spasms responsive to pyridoxine treatment and with good clinical outcomes, no mutations were found and PA levels were normal. In total, 13 novel mutations were identified.
Discussion: Our study advances previous findings that defects of ALDH7A1 are almost always the cause of neonatal-onset PDS and that defects in this gene are also responsible for some but not all later-onset cases. Later-onset cases of infantile spasms with good outcomes lacked evidence for antiquitin dysfunction, suggesting that this phenotype is less compelling for PDS.
Pyridoxine-dependent seizures (PDS: OMIM, 266100), also known as pyridoxine-dependent epilepsy, is a rare, autosomal recessive disease with an estimated prevalence of between 1 per 276,000 and 1 per 700,000 births (Baxter, 2001; Bennett et al., 2005; Gospe, 2006; Bok et al., 2007). Patients classically present with neonatal seizures that are unresponsive to conventional anticonvulsant therapy, but which can be controlled with pyridoxine monotherapy; less commonly, later-onset cases present through the second to third years of life. A variety of clinical seizure types are seen including myoclonic seizures, atonic seizures, partial and generalized events, and infantile spasms. Although most patients with PDS show a rapid response to pyridoxine treatment, some show a transient response to common anticonvulsants or a poor initial response to pyridoxine (Gospe, 2002). Patients with PDS display varying degrees of developmental delay, which may, in part, be independent of the timing of pyridoxine therapy (Baynes et al., 2003; Rankin et al., 2007). A related condition designated as pyridoxine-responsive seizure (PRS) includes infants and young children with seizures that respond initially to pyridoxine but lack seizure recurrence with pyridoxine withdrawal; familial cases of PRS have not been described (Baxter, 1999, 2001).
Until recently, the unequivocal diagnosis of PDS was made by withdrawing pyridoxine treatment from responsive patients, to provoke seizures, followed by the reintroduction of pyridoxine and the return of seizure control (Gospe, 1998, 2002; Baxter, 2001). Since the discovery of biomarkers for PDS—α-aminoadipic semialdehyde (AASA) and pipecolic acid (PA)—and available clinical DNA tests for the antiquitin gene, it is no longer necessary to withdraw patients from pyridoxine treatment to confirm the diagnosis (Plecko et al., 2000, 2005; Willemsen et al., 2005; Gospe, 2006).
The underlying enzymatic defect of PDS has been located at the level of AASA dehydrogenase within the cerebral lysine catabolism pathway. The significance of elevated PA in PDS was disclosed from understanding the biochemical mechanism leading to seizures in hyperprolinemia type II (Farrant et al., 2001). A defect in L-Δ1-pyrroline-5-carboxylate (P5C) dehydrogenase causes accumulation of P5C, which in turn reacts with pyridoxal-5-phosphate (PLP) by Knoevenagel condensation, resulting in PLP depletion. A similar mechanism was shown to cause PLP depletion in PDS patients with mutations in the antiquitin gene (ALDH7A1). When antiquitin function is defective, piperideine-6-carboxylic acid (P6C) accumulates and reacts with PLP, thereby causing its concomitant depletion (Mills et al., 2006). The precise mechanism by which cerebral PLP depletion leads to seizures is not fully understood but undoubtedly involves compromised neurotransmitter synthesis where PLP is required as a coenzyme. In addition, given that PLP depletion is common to both hyperprolinemia type II and PDS, it may be suggested that metabolite or adduct build-up could directly contribute to the disease phenotypes.
We undertook clinical evaluation of 18 patients with PDS and performed DNA sequence analysis of the ALDH7A1 gene to evaluate the hypothesis that the prevalence of ALDH7A1 mutations is discordant between early (neonatal) and later-onset cases of PDS.
- Top of page
- Materials and Methods
In this study, we have identified ALDH7A1 mutations in a high percentage of patients with PDS (81%), including: missense, nonsense, frame-shift, and splice-site mutations encompassing 13 mutations in this gene. Each of the nine new missense mutations (Table 2) was predicted to be deleterious from analysis by the Sort Intolerant from Tolerant (SIFT) program (Ramensky et al., 2002) and the Polymorphism Phenotype (PolyPhen) program (Ng & Henikoff, 2001). This is perhaps not a surprising finding given that disease-associated amino acid substitutions of this metabolic enzyme were located largely in highly conserved regions of the protein (Fig. 1). The single exception to this is the ALDH7A1 G255D substitution, which is significantly less conserved than all other mutated residues associated with PDS.
We hypothesized that the prevalence of ALDH7A1 mutations, with or without indirect biochemical evidence of antiquitin dysfunction, would be higher for neonatal-onset subjects, versus later-onset PDS (Tables 1 and 2). To this end we undertook genomic DNA sequencing of the ALDH7A1 gene, allelic-specific detection of an intragenic microsatellite marker to exclude an intragenic deletion (Kanno et al., 2007), and whenever possible, biochemical analyses of plasma PA levels. Although elevation of either plasma or urine AASA represents a more specific biomarker of PDS than PA (Mills et al., 2006), methods to detect and measure this particular substance are not yet readily available. Detection of elevated PA levels was able to support the causal status of the novel mutations we identified in most cases, yet this is not definitive. When patients are receiving large doses of pyridoxine, the fold-elevation observed is significantly reduced over time. For example, in the study by Plecko et al. (2005), patient 1 had PA levels of 17.6 μmol/L, which were later reduced to 5.3 μmol/L with pyridoxine treatment. We observed PA levels in the normal range of 2.6 μmol/L in one patient harboring two ALDH7A1 mutations, R238X and N167S (K3015-II.1). Ideally, incorporation of AASA testing would be more sensitive and specific, but this assay was not available for this study.
Within neonatal-onset PDS, the combination of ALDH7A1 sequencing and the measurement of plasma PA levels suggested that recessive defects of antiquitin function were causal in all 12 cases, confirming our hypothesis for this group of subjects. In later-onset PDS, ALDH7A1 mutations were identified in three of six cases. In one additional case, K3009-II.1, no mutations were detected, yet plasma PA levels were significantly elevated, suggesting antiquitin deficiency to be causal.
These three cases represent the first patients with late-onset PDS in whom ALDH7A1 mutations have been identified. In addition, within this group, we effectively excluded a causal role for loss of antiquitin function in two of the kindreds. In K3003-II.1 and K3018-II.1, no ALDH7A1 mutations were detected and plasma PA levels were in the mid-normal range. These patients (K3003-II.1 and K3018-II.1) also manifested interesting seizure types, as both siblings of pedigree K3003 (II.1 and II.2) and patient K3018-II.1 had infantile spasms. Although infantile spasms are not a typical PDS seizure type, they are occasionally observed (Baxter, 1999). These three subjects have had an unusually good outcome following the resolution of infantile spasms, with normal intellectual development resulting. Given the lack of ALDH7A1 mutations and these particular clinical features, these patients may, in fact, represent examples of PRS. If this should be the case, the two siblings in kindred K3003 would represent the first example of PRS recurring within a family. However, none of these patients have had trials of pyridoxine withdrawal, and, therefore, neither PDS nor PRS has been clinically confirmed. If a trial of pyridoxine withdrawal is eventually conducted in these patients and definite PDS is assigned, the absence of ALDH7A1 mutations along with this unique phenotype of infantile spasms with a good outcome would support genetic heterogeneity for PDS (Bennett et al., 2005).
A fourth late-onset patient, K3009-II.1 also presented with infantile spasms, and although DNA sequencing did not detect ALDH7A1 mutations, plasma PA was significantly elevated, indicative of antiquitin dysfunction and suggesting either incomplete sensitivity of DNA analysis of ALDH7A1 or another cause for the PDS. In this particular case, the poor long-term outcome is in line with cognitive outcomes for many patients with PDS (Gospe, 2002; Baxter, 2003; Rankin et al., 2007), and, indeed, as is commonly the outcome for infantile spasms. Therefore, for clinicians facing patients with infantile spasms, responsive to pharmacologic doses of pyridoxine (and, therefore, meeting at least one PDS diagnostic criterion), antiquitin deficiency cannot be presumed. The response to pyridoxine along with an elevation in plasma PA, in such cases, may suggest that another downstream protein in the cerebral lysine catabolic pathway, or an overlapping pathway, maybe involved.
Our prior segregation analysis of the chromosome 5q31 PDS locus (ALDH7A1) in six small multi-sib North American families was consistent with linkage in five of the six families. In pedigree K3001, linkage to 5q31 appeared to be excluded, as haplotype sharing was apparent between siblings discordant for PDS (Bennett et al., 2005). This appeared to confirm earlier suggestions of genetic heterogeneity for PDS (Baxter, 2003), but in our present study, K3001 was found to harbor an E399Q mutation and a W335X nonsense mutation, confirming antiquitin dysfunction as the basis of disease in K3001.
Interestingly, interfamilial variation of PDS onset is observed for patients K3010-II.1 and K3020-II.1 (neonatal/fetal) versus patient K3017-II.1 (6 months of age), all harboring two copies of the E399Q common mutation. The same genotype was seen four times in a recent report (Plecko et al., 2007), with one possible, one probable, and two definite patient classifications. Intrafamilial variation was also present in pedigree K3008 as the onset of ALDH7A1-related PDS varied from neonatal onset (II.1) to onset at 9 months (II.2) suggesting that other genetic and environmental modifiers are influencing phenotype.
In case K3020-II.1, despite early clinical and electrographic support of a diagnosis of pyridoxine dependency, this patient has only experienced a short period of complete seizure freedom during his 17 years of life. In reviewing his complex history, many clinicians would conclude that he either does not have PDS, or alternatively has a secondary cause for his intractable epilepsy. However, K3020-II.1 is homozygous for the common E399Q mutation. As he has been treated appropriately for PDS, his intractable epilepsy must be secondary, potentially due to inherent PDS-associated cerebral dysgenesis, a finding that has been reported in many individuals with PDS (Gospe & Hecht, 1998; Baxter, 2003). This case emphasizes the utility of measuring PA or AASA levels and screening ALDH7A1 for mutations in patients with preliminary diagnoses of PDS.
Taken together, our data confirm that PDS is commonly the result of recessive mutations of the ALDH7A1 gene (Mills et al., 2006; Bok et al., 2007; Plecko et al., 2007), but that genetic heterogeneity is also present. One might posit that genetic causes other than ALDH7A1 mutations are more likely to be seen in later-onset cases. However, in the recent Japanese study, genetic heterogeneity was evident in a patient with neonatal-onset PDS (onset of symptoms, 18 days). In this patient, no ALDH7A1 mutations were found and PA levels were in the normal range. It is of interest to note that as with our three cases with infantile spasms and putative PRS in the later-onset group, mental retardation was not evident (Kanno et al., 2007).
In the initial PDS gene identification report (Mills et al., 2006), and in a subsequent study of 18 patients with PDS (Plecko et al., 2007), the E399Q mutation emerged at a high frequency of 33%. We found the E399 residue was mutated at the same high frequency of 33% (12 of 36 alleles) in our total cohort of predominantly Caucasian patients. In the limited patient subset of Hispanic or mixed Hispanic/European ancestry, the E399Q substitution represented a similar 33% of mutant alleles (2 of 6 alleles). In a recent study of ALDH7A1 mutations in Dutch patients, E399Q was again found at high frequency (Salomons et al., 2007). This common mutation may represent a possible founder effect in this latter population, yet no haplotype analysis was undertaken. Also of interest, in the recent Japanese study, no mutation of the E399 residue was present suggesting ethnic differences (Kanno et al., 2007). We found no evidence of a founder effect in our patient cohort, as no specific microsatellite genotype segregated with the E399Q mutation (Table 2, CA repeats). Irrespective, this particular residue has been clearly established as a hot-spot and critical for several reasons (Mills et al., 2006). First, it is highly conserved throughout evolution. Second, this residue in other human aldehyde dehydrogenase superfamily members was shown to interact with the ribose moiety of NAD, and that substitution of glutamine converts the rate-limiting step from deacylation to hydride transfer (Perozich et al., 1999). Third, transfection of CHO cells with the E399Q mutant produced no detectable AASA dehydrogenase activity (Mills et al., 2006).
In our patient cohort, we also identified a coding c.750g>a transversion that at first appeared to be a harmless synonymous codon substitution (V250V). However, because it was present in two patients, K3007-II.1 and K3004-II.1, both with raised PA levels, we speculated that it may represent a cryptic splice-site mutation. Recently, this mutation was confirmed to generate a mutant splice donor site that is used almost exclusively, generating a frame-shift and nonsense mediated decay of the message (Salomons et al., 2007).
In conclusion, we confirmed in this investigation that ALDH7A1 mutations underlie all cases of classic neonatal-onset PDS and a lesser proportion of later-onset cases. Within late-onset cases, three cases manifested infantile spasms, with a surprisingly good long-term outcome but lacked evidence for antiquitin dysfunction suggesting this phenotype to be less compelling for PDS. We confirm that E399Q is a common mutation within ALDH7A1 and may provide a means of triage to diagnostic molecular testing for this gene. The utility of biochemical markers combined with DNA analysis of ALDH7A1 obviate traditional means of PDS diagnosis, and enable the provision of prenatal diagnosis using this approach.