Sanger sequencing is a mainstay for the identification of gene mutations used in molecular diagnostic laboratories. However, in autosomal recessive disorders, failure of allele amplification can occur for a variety of reasons, leading heterozygous mutations to appear homozygous. We sought to investigate the frequency at which apparently homozygous mutations detected by Sanger sequencing in our laboratory appeared homozygous due to other molecular etiologies.
A review of 12,406 cases from 40 different genetic tests that were submitted to the Medical Genetics Laboratories at Baylor College of Medicine for Sanger sequence analysis was performed. The molecular status of apparently homozygous cases was further investigated by testing parents using various methods.
A total of 291 cases of apparent homozygosity were identified, ranging from 0 to 37% of the total per gene. One-third of the apparently homozygous cases were followed up by parental testing. Parental carrier status was confirmed in 88% of the cases. Of the cases in which parental carrier status could not be confirmed, deletions encompassing point mutations, allele dropout due to single-nucleotide polymorphisms at primer sites, and uniparental isodisomy were observed.
For individuals with autosomal recessive disorders and apparently homozygous mutations, confirmation by parental testing can rule out other causes of apparent homozygosity, including allele dropout, copy number variations, and uniparental isodisomy.
Genet Med 2012:14(10):877–882
The occurrence of true homozygous mutations is not unusual for some autosomal recessive (AR) disorders with common mutations, such as medium-chain acyl-CoA dehydrogenase deficiency, cystic fibrosis (CF), and hereditary hemochromatosis.1,2,3,4 Homozygous mutations are also more likely to be detected in consanguineous families or small populations. However, numerous publications have reported mutations that appeared to be homozygous yet had another underlying genetic cause.5,6,7,8 Therefore, whether observed homozygosity is a result of an individual carrying the same mutation on both chromosomes, or a product of two different genetic events, can be investigated by testing the parents of the affected individual.
Targeted analyses for specific common mutations such as the c.985A>G (p.K329E) mutation in medium-chain acyl-CoA dehydrogenase deficiency and the cystic fibrosis deltaF508 mutation can be performed using a variety of molecular techniques including but not limited to restriction fragment length polymorphism, allele-specific oligonucleotide, pyrosequencing, and amplification refractory mutation system. However, if there are no frequently appearing mutations in the genes of interest, Sanger sequence analysis remains the standard methodology for DNA sequence analyses in most molecular diagnostic laboratories. Regardless, all of these techniques rely on PCR-based amplification of a particular genomic region. In general, a single set of primers is used for PCR amplification of regions of interest. As a result, a single-nucleotide polymorphism (SNP) present within a primer site may disrupt the binding of that primer, and allele dropout could unknowingly occur giving the appearance of homozygosity. If a heterozygous deletion is located in the same region of amplification, only one chromosome is amplified, which would also result in the region of interest appearing homozygous. In uniparental isodisomy (UPD), two identical copies of the same chromosome are inherited from one parent. Therefore, although UPD causes true homozygosity of a point mutation, the molecular mechanism is different than if a proband inherited the same mutation from both parents, and the two mechanisms have very different recurrence risks (almost 0% in the case of UPD but 25% in the case of identity by descent).
Apparent homozygosity is a phenomenon observed in every molecular diagnostic laboratory. Nevertheless, to our knowledge, its frequency and the molecular etiology behind it have never been systematically studied. To determine the rate at which apparently homozygous mutations detected in our laboratory were truly homozygous, we reviewed 291 cases of apparent homozygosity from 40 different AR genetic diseases tested in our laboratory. All of the cases reviewed were analyzed by PCR amplification and Sanger sequencing. Genes in which targeted testing for specific mutations, such as the medium-chain acyl-CoA dehydrogenase deficiency c.985A>G (p.K329E), was performed were not included in this study.
Although further testing was recommended by the laboratory in every case, only about one-third of the cases were followed up by the referring clinician through testing of parental samples. Of those cases, we found that nearly 12% did not inherit the mutation from each parent. Our results underscore the importance of further confirmation and investigation of apparently homozygous mutations detected by Sanger sequencing.
Materials and Methods
Patients and DNA
Tissue or blood samples from patients with clinical diagnoses of AR diseases were submitted to the Medical Genetics Laboratories at Baylor College of Medicine for molecular evaluation between January 2007 and December 2010. A majority of the samples analyzed were blood. Of the 291 samples of apparent homozygosity other sample types included muscle (4), skin fibroblast cultures (2), and liver (2). Four samples were received as DNA extracted from blood at the referring institution. Total genomic DNA was extracted from peripheral blood leukocytes or other tissues using a commercially available DNA isolation kit (Gentra Systems, Minneapolis, MN) according to the manufacturer’s protocols. A 260/280 absorbance measurement between 1.7 and 2, a 260/230 absorbance measurement greater than 1.5, and a concentration of a minimum of 50 ng/µl were required for further PCR analysis of DNA samples. A second extraction was performed if samples were not of sufficient quality, and also to confirm positive findings. For those samples received as extracted DNA, the original sample was run a second time. This study was approved by the institutional review board of Baylor College of Medicine.
Sequence analysis of the appropriate gene(s) was carried out using gene-specific primers to amplify coding exons and at least 50 intronic nucleotides flanking each exon, and sequencing was performed as previously described.8 DNA concentrations were consistent throughout all PCR reactions at ~50 ng. For those samples in which alternate primers were designed, primer sites outside of the first pair of primers were chosen in order to detect any SNPs that might lie in the original primer pair.
Oligonucleotide array comparative genomic hybridization analysis
Detection and analysis of copy number variation was performed using MitoMet oligonucleotide array comparative genomic hybridization (aCGH) (v2.8), a custom designed clinically validated 60K oligonucleotide array (Agilent Technologies, Santa Clara, CA) with complete coverage of the mitochondrial genome and 351 nuclear genes related to mitochondrial structure/function and metabolic diseases. The average probe spacing was ~200–300 bp per oligonucleotide probe (minimum spacing of 2–10 bp and a maximum of 500 bp between probes in targeted exonic regions). In addition, backbone coverage throughout the genome was ~400–500 kb per probe. Samples were processed as previously described using 1 µg of patient DNA.9
Detection of UPD by SNP array
The Human610-Quad SNP array, which covers 550,000 SNPs plus an additional 60,000 genetic markers per sample, was used as previously described10 per the manufacturer’s protocol (Illumina, San Diego, IL). Genotyping data was analyzed for copy-neutral absence of heterozygosity with GenomeStudio software using the CNV partition 2.3.4 algorithm (Illumina).
The results of 12,406 reports from testing 40 different AR genes related to mitochondrial or metabolic diseases were reviewed (Table 1). A total of 291 cases (~2%) with apparently homozygous point mutations or small insertions/deletions detected by Sanger sequencing were identified. The percentage of apparently homozygous cases per gene ranged from 0–37%, with a majority of genes having less than 10%. Of note, the highest percentage of apparently homozygous cases was in the ARG1 gene (13/35), in which 10 different mutations, distributed throughout the gene, were observed. Of these 13 individuals, 10 were Hispanic and 3 were from the Middle East with family histories suggesting different degrees of consanguinity. In fact, a majority of cases for the four genes that had the highest percentage of apparent homozygosity (ARG1, FAH, MMACHC, and CPS1) were from populations in which consanguinity is not uncommon.
For those genes in which more than one case of apparent homozygosity was detected, only three had less than 50% mutation diversity: POLG (7/25), ALDOB (3/10), and HADHA (1/7) (Table 1). The most frequently observed mutations in ALDOB and HADHA are the common mutations c.448G>C (p.A150P) (8 of 10 patients) and c.128G>C (p.E510Q) (seven of seven patients), respectively.
Every report with a result of an apparently homozygous mutation that was issued from our laboratory suggested parental samples be submitted for carrier confirmation. However, we received both parental samples in only ~26% (75/291) of the apparently homozygous cases. In an additional eight cases, the mothers of the probands were tested and confirmed to be carriers; however, the fathers were not tested (these cases were not counted as having parental testing performed in our analysis). Heterozygous carrier status of both parents was confirmed in 88% (66/75) of the cases.
In the remaining nine cases of apparent homozygosity, one of the parents was confirmed as a carrier of the sequencing mutation and the other was not. Therefore, these cases were further analyzed by a variety of molecular and genetic approaches including alternative primers to test for SNPs at primer sites, targeted aCGH to detect copy-number changes, and SNP analysis to detect possible UPD.
For those samples in which one of the parents was negative for the apparently homozygous mutation in the proband, regions of interest were first reanalyzed with alternative primer sets. Allele dropout due to private SNPs in the primer regions was identified in two families (Table 1). In one family, the proband was initially tested for sequence changes in the ACADVL gene as a result of a positive newborn screening result suggesting very long-chain acyl-CoA dehydrogenase deficiency. This individual was reported as apparently homozygous for a frameshift mutation. Subsequent testing of both parents revealed that only the father was a carrier of the mutation. Reanalysis of the region using an alternative set of primers identified a private SNP in the mother in the region of the original primer. The proband inherited this SNP from her mother, which resulted in allele dropout at the mutation locus, making the heterozygous mutation appear homozygous.
A second male proband presented with persistent hyperbilirubinemia, and testing for both progressive familial intrahepatic cholestasis types 1 (ATP8B1) and 2 (ABCB11) was requested. An apparently homozygous missense mutation was detected in ABCB11. Sequence analysis of the ATP8B1 gene was negative. Subsequent parental testing detected the heterozygous ABCB11 mutation in the father of the proband, but not in the mother. However, the use of an alternative primer set identified a private SNP in the primer site of the mother. The child inherited the same SNP, and reanalysis of the mutation site using the alternative primer set revealed that the child was actually heterozygous for the mutation. MitoMet aCGH analysis did not detect a deletion. It is possible that the second mutation in the proband could not be detected using our current methodologies, or that the underlying cause is due to mutations in an alternative gene and this individual is simply a carrier of an ABCB11 mutation.
Heterozygous deletions encompassing a point mutation were identified in two families. In one case, the proband presented with hyperammonemia and hypoketotic hypoglycemia. Sequence analysis resulted in the identification of an apparently homozygous nonsense mutation in the SLC25A20 (CACT) gene. However, when the parents were tested for the presence of this mutation, only the father was a carrier. PCR with alternative primers reproduced the same results. MitoMet aCGH identified a large heterozygous deletion in the proband encompassing the region containing the point mutation. This deletion was confirmed in the mother of the proband. Therefore, the proband carried a heterozygous deletion and a nonsense mutation in the SLC25A20 gene in a trans configuration.8
The proband in the second case presented with an abnormal acylcarnitine profile, suggesting a primary carnitine deficiency. Sequence analysis of the SLC22A5 (OCTN2) gene revealed an apparently homozygous novel variant. Subsequent sequence analyses on the parents of the proband identified the variant in the mother, but not in the father. MitoMet targeted aCGH on the proband revealed a large heterozygous deletion encompassing the entire SLC22A5 gene. However, this deletion was not detected in the father of the proband. The results of these analyses suggest that the deletion in the proband is likely de novo; however, mistaken paternity could not be ruled out.
The two cases in which UPD was determined to be the cause of the apparent homozygosity are described elsewhere.10 Briefly, both cases presented with clinical features consistent with mitochondrial DNA depletion syndrome. In the first case, an apparently homozygous point mutation in the TYMP gene was detected. Parental testing showed that only the mother of the proband was heterozygous for this mutation. In the second case, an apparently homozygous point mutation in the DGUOK gene was identified in the proband. The mutation was confirmed in the father but not in the mother. Subsequent sequence analyses using alternative PCR primers confirmed both results and targeted MitoMet aCGH did not detect changes in the copy number in either family. Chromosomal microarray-SNP analyses were therefore performed; these revealed segmental isodisomy on chromosome 22 of ~11.3 Mb that includes the TYMP gene in the first patient, and complete isodisomy of chromosome 2, which contains the DGUOK gene, in the second.10
Three cases currently remain unresolved (genes marked with footnote cue “a” in Table 1). In two cases, apparently homozygous mutations in G6PC and PFIC2 were identified in symptomatic probands. In both cases, only the mothers were found to be carriers. Alternative primers yielded the same results, and analyses of the copy number did not detect copy-number variations in either the probands or the fathers. Both families declined chromosomal microarray-SNP analyses.
In the third case, an apparently homozygous mutation in the HADHA gene was detected in a newborn suspected of having long-chain 3-hydroxyacyl-CoA dehydrogenase deficiency. Subsequent testing of the parents of the proband detected the mutation in the father but not in the mother. Alternative primers did not detect SNPs in the primer sites of the maternal sample, and MitoMet aCGH did not detect any changes. However, there was not a sufficient amount of sample left from the proband to perform copy-number analysis and further testing was not pursued.
Based on the results of our review, we believe that the identification of an apparently homozygous point mutation in an affected proband with an AR disease should be followed up by parental testing whenever possible. If testing of the parents does not confirm their carrier status, additional molecular analyses can be performed to identify the underlying molecular etiology (Figure 1). In our study, ~12% of the homozygous point mutations could not be confirmed by subsequent parental testing. The apparent homozygosity in these cases was due to SNPs at the primer sites, heterozygous deletions encompassing the point mutation, or UPD. Six cases of apparently homozygous deletions were also detected in our analysis. However, all of these cases were confirmed as homozygous using MitoMet aCGH, and were not included in the 75 cases of apparent homozygosity.
The high percentage of apparently homozygous mutations in some genes such as ARG1 is surprising considering 10 of the 13 mutations we observed were different, and spread throughout the gene. However, a majority of these cases were in families where consanguinity was highly likely. Further testing of parents (or proband) was only requested by the referring clinician in a single ARG1 case, and true homozygosity was confirmed in that case. Based on the ethnic background of the probands, apparent homozygosity for some genes, such as ABCB4, ABCB11, and ATP8B1, appeared likely to be a result of consanguinity and therefore it is not surprising that follow-up testing was not pursued.
The detection of SNPs in primer sites in apparently homozygous cases highlights a persistent problem in PCR-based sequence analysis. This problem can be minimized by continuous reassessment of the presence of SNPs in the primer sites using the constantly updated dbSNP database. Parental analyses will also identify private SNPs in families. In addition to Sanger sequencing, SNPs on primer sites may also lead to erroneous results in any testing method using PCR. In general, allele dropout due to SNPs at primer sites should always be ruled out first for any PCR-based analyses. Non-PCR-based next-generation sequencing will not have this problem as each base will be covered tens to thousands of times. However, some regions of the genome may have poor coverage, which will still require Sanger sequence analysis, and all positive next-generation sequencing results can be confirmed by a secondary method. PCR-based Sanger sequence analysis cannot detect large deletions, duplications, and other genomic rearrangements. Methods for the detection of copy number variations and UPD, such as chromosomal microarray analysis, multiplex ligation-dependent probe amplification, and SNP genotyping have become increasingly more commonly used in clinical diagnostic laboratories and should be considered as well.
A majority of the patients that are tested for AR disorders present with a particular clinical phenotype, making PCR artifacts such as allele dropout less likely. However, with the advent of testing such as screening of newborns, children may test positive for a disorder before becoming symptomatic. The positive thresholds for some of these disorders are defined such that heterozygous carriers may have borderline positive results.2,11 Therefore, confirmation of molecular results is imperative in an asymptomatic proband with an apparently homozygous mutation as the results may alter management of the child and provide information to the parents in the consideration of future pregnancies.
Determining the full molecular characteristics of familial mutations in AR disorders provides additional information in the diagnosis of the proband and parents, appropriate genetic counseling for the family, and possible prenatal diagnosis and preimplantation genetic diagnosis for future pregnancies. Not confirming the molecular basis of apparently homozygous mutations could result in false-negative carrier screens in at-risk family members who may test negative for the mutation. In fact, the results obtained from testing the parents in the apparently homozygous CACT proband who carried a point mutation over a deletion were used for prenatal testing in a subsequent pregnancy.8 The deletion, but not the point mutation, was detected in the fetus.
Parents should also be informed of what follow-up testing may reveal about parentage. Mistaken paternity rates may be as high as 10% in various populations.12,13,14 The issue of nonpaternity is a sensitive one that must be considered when apparently homozygous results in a proband cannot be resolved after all testing methods have been exhausted. Therefore, paternity should be confirmed before extensive molecular analyses are carried out.
In summary, although most cases of apparently homozygous point mutations in AR diseases reported in our laboratory are due to both parents being heterozygous carriers, 12% of those tested could not be confirmed by sequencing of the parental samples. Our results demonstrate that comprehensive mutation analysis using standard DNA sequencing in combination with additional molecular testing for allele dropout, copy-number variations, and UPD is imperative in resolving apparent mutation homozygosity.
The Department of Molecular and Human Genetics at Baylor College of Medicine derives revenue from the molecular diagnostic tests offered by the Medical Genetics Laboratory.
We thank all of the patients, families, and clinicians who have provided or facilitated sample and information collection. We also thank D. Boehning and F. Probst for their critical reviews of the manuscript.