Identification of FVIII gene mutations in patients with hemophilia A using new combinatorial sequencing by hybridization

BACKGROUND: Standard methods of mutation detection are time consuming in Hemophilia A (HA) rendering their application unavailable in some analysis such as prenatal diagnosis. OBJECTIVES: To evaluate the feasibility of combinatorial sequencing-by-hybridization (cSBH) as an alternative and reliable tool for mutation detection in FVIII gene. PATIENTS/METHODS: We have applied a new method of cSBH that uses two different colors for detection of multiple point mutations in the FVIII gene. The 26 exons encompassing the HA gene were analyzed in 7 newly diagnosed Italian patients and in 19 previously characterized individuals with FVIII deficiency. RESULTS: Data show that, when solution-phase TAMRA and QUASAR labeled 5-mer oligonucleotide sets mixed with unlabeled target PCR templates are co-hybridized in the presence of DNA ligase to universal 6-mer oligonucleotide probe-based arrays, a number of mutations can be successfully detected. The technique was reliable also in identifying a mutant FVIII allele in an obligate heterozygote. A novel missense mutation (Leu1843Thr) in exon 16 and three novel neutral polymorphisms are presented with an updated protocol for 2-color cSBH. CONCLUSIONS: cSBH is a reliable tool for mutation detection in FVIII gene and may represent a complementary method for the genetic screening of HA patients.

3.1 kb (exon 14) in size, spans 186 kb of genomic DNA and produces a 9030 nt mRNA. According to the UK Hemophilia Centre Doctors' Organisation (UKHCDO) Hemophilia Genetics Laboratory Network, the severity of HA in the pedigree should be determined Þ rst as this will inß uence the diagnostic strategy to be employed.
Severe Hemophiliacs should be screened for the intron 22 inversion mutation followed by the intron 1 inversion mutation. This approach identiÞ es the underlying mutation in 45-50% of severe HA patients. [3,4] The remaining severe HA pedigrees should then be analyzed further either by full mutation or linkage analysis. Mutations have been found in nearly all 26 exons of the factor VIII gene, over 400 mutations have been identiÞ ed [5,6] and de novo mutations represent approximately 30% of all cases. [7] The most common detection methods include DNA sequence analysis which requires numerous reactions and individual analysis of each exon or alternative screening methods such as singlestranded conformation polymorphism (SSCP), [8] denaturing gradient gel electrophoresis (DGGE) [8] and denaturing high performance liquid chromatography (dHPLC). [9] We applied the new combinatorial sequencingby-hybridization (cSBH) as an alternative method to the traditional Sanger dideoxy chain termination approach. [10,11] Previous works have shown that cSBH is an efÞ cient rapid and alternative method for mutation detection. [12,13] We increased the quality of results with a new cSBH method that use two different colors (TAMRA and QUASAR). The platform is an indirect method which We report development of a strategy to implement 2-color cSBH to screen a range of mutations within the FVIII gene.

Probes
Amino-modified unlabeled hexamer (6-mer) probes as well as TAMRA and QUASAR labelled pentamers

Target preparation
Prior to hybridization, the PCR nested-asymmetric products were precipitated using isopropanol, creating eight different mixes containing single-stranded DNA of the forward and eight containing the reverse strand. The mixes were performed to optimize the analysis results.

Results
The new 2-color cSBH is an indirect sequencing method in which overlapping probes of known sequence are hybridized to DNA target molecules. In cSBH, which is represented in a schematic overview in Figure 1, target DNA from a PCR product is exposed to two universal sets of short probes in the presence of DNA ligase.

Mutation Detection in Italian Haemophilic A Patients
A total of 16 clinically diagnosed Italian haemophilic A patients and one obligate carrier previously genotyped

In the panel A the letter D represents an hemizygous deletion of A in Ex12 (sample 2)(DelC aa1194). In the panel B the letter D represents an hemizygous deletion of C in Ex12 (sample 3). In the panel C the letter S represents a T (Sample 64)µ substitution of A
8519 A!G (samples Cz 36 and Cz 1).

Discussion
We demonstrated that cSBH, and the related HyChip product, is capable of identifying a range of point  The combinatorial ligation of universal probe sets coupled with informative probe pools provides a large number of long (currently 11-mers, potentially 12-to 15mers in two-probe ligation and 17-to 24-mers in three probe ligation), overlapping probes of predetermined behavior that interrogate each base. Robust statistics take advantage of this redundant dataset to provide accurate base calling in spite of array imperfections (0.5-5.0% of missing dots; high droplet size variation), some impaired probe-target hybridizations due to target-target hybridization, significant experimental noise, and the always-present statistical noise related to probe-pooling and sequence repeats. The main software improvement in the future should address target DNA behavior (e.g., primers/primer dimers, palindromes, direct and inverted repeats, and uneven amounts of target DNA) to aid in a more accurate interpretation of base scores and associated P-values. Actually, the presence of short palindromes or repeats may contribute to the number of bases with lower "scores", as seen in sample 2 [ Figure 2A]. Furthermore, detection of low-signal AT rich sequences can be improved with optimization of probe concentrations and incorporation of 1,600 AT-rich 7-mers and 8-mers in the arrays, as well as 500 AT-rich 6-mers and 7-mers in the labelled probe pools, which are currently under development.
Universal arrays use probes shorter than gene speciÞ c arrays. As a consequence, there is a higher chance to have repeated sequences longer than the length of probes (currently 11-mers in cSBH) occurring in a DNA sample. Such sequences may lead to uncertainty in the sequence assembly. For example, a stretch of 12 As can be perceived as 11, 12, 13, or more As. Also, a mutation in the middle of a repeated 20-mer or any longer segment may not be assigned to the actual copy of the repeat. This situation can be avoided by amplicon design that separates repeats into two sequencing reactions.
Short identical repeats several bases in length or longer imperfect repeats, both found in exons or immediate intronic sequences that ß ank each exon, usually do not confuse the advanced base-calling software.
There are limitations with cSBH analysis: the current software and probe set must be improved to detect DNA alterations within simple repeat regions, which is a problem common to other sequence technologies as well. [10] Another potential complication, which is common to other high resolution methods, including standard Sanger sequencing, is that all sequence changes, polymorphisms and missense changes that do not result in an obvious pathogenic mutation such as a stop codon and are of unknown clinical signiÞ cance will be detected. [10] In summary, mutation detection for the FVIII gene requires a multimodal approach that can detect the full range of mutations present in this gene. First-stage mutation screening requires a fast, efÞ cient and possibly economical approach to scan the entire gene. cSBH is a technique that is ideally suited. In this study, we showed that cSBH is capable of detecting a broad range of mutations, currently has the capacity to analyze long continuous read lengths of up to 2 kb per chip, and is able to analyze pooled PCR fragments (such as exons 1-25 in this study) on a single universal chip.