RET Oligonucleotide Microarray for the Detection of RET Mutations in Multiple Endocrine Neoplasia Type 2 Syndromes¹

The human RET gene encodes a transmembrane receptor of the protein tyrosine kinase family, which is implicated in neural crest tissue development and differentiation (1, 2). The RET gene, located on chromosome 10q11.2, is composed of 21 exons and is ∼55 kb in size (3). The RET protein is composed of an extracellular domain, a transmembrane domain, and intracellular tyrosine kinase domains (3, 4). The RET gene is responsible for MEN2³ syndromes, which are inherited in an autosomal dominant fashion with high penetrance and diverse clinical manifestations. The syndromes have three subtypes: MEN2A, MEN2B, and FMTC (2, 5). MEN2A, the most frequent subtype of the MEN2 syndromes, is characterized by MTC (or C-cell hyperplasia), pheochomocytoma, and hyperparathyroidism. RET mutations in MEN2 syndromes have usually been detected by single-strand conformational polymorphism or by direct sequencing. A method for the rapid mutation analysis of a few gene sequences has been developed using an oligonucleotide microarray, which can be effectively used for sequence analysis, diagnostics for genetic diseases, and gene polymorphism studies (6). A typical DNA microarray-based method is less time consuming and is cheaper than conventional sequencing, and plays a valuable role in high throughput sequence analysis (7, 8). Oligonucleotide microarrays show high sensitivity in terms of point mutation detection. Thus, it would be very useful to use an oligonucleotide microarray for analyzing genes with more frequent point mutations, such as the RET gene in MEN2 syndromes. The predominant mutations in MEN2 syndromes are missense mutations, and these are restricted to some codons of the RET gene. In MEN2A and FMTC, most RET mutations are found in exons 10 and 11, and are located in the extracellular cysteine-rich region, which is a part of the putative ligand-binding domain (codons 609, 611, 618, 620, 630, and 634). Infrequently, noncysteine codon mutations have been reported in FMTC (at codons 768 and 804; Ref. 9). Missense mutations at codons 609, 611, 618, 620, and 634 have been identified in 98% of MEN2A families and in 85% of FMTC families. Missense mutations at codons 768 and 804 have been known to be responsible for 5–10% of the FMTC families (10). We developed an oligonucleotide microarray for rapid and simplified RET mutation detection. In our microarray system, a glass slide spotted with 67 oligonucleotides and fluorescence-labeled PCR products were used to detect RET mutations. The RET oligonucleotide microarray can find RET missense mutations at nine codons in MEN2 syndromes (MEN2A, MEN2B, and FMTC). In this study, the RET proto-oncogene was analyzed in eight Korean MEN2A families using automatic bidirectional sequencing and the RET oligonucleotide microarray.

Blood samples of five Korean MEN2A families (SNU-MEN2A-I, -II, -III, -IV, and -V) from Seoul National University Hospital and three MEN2A families (SMC-MEN2A-I, -II, and-III) from Samsung Medical Center were collected. SNU-MEN2A-I and three SMC-MEN2A families have been reported previously (11, 12). Sixty-two SNU-MEN2A family members (21 affected members, 17 gene carriers, and 24 unaffected members) were studied using the clinical approach, and 46 members of the 62 were analyzed genetically (Table 1).

Total genomic DNA was extracted using Ficoll-Paque (Amersham Pharmacia-biotech Ltd., Uppsala, Sweden) and TRI reagent (Molecular Research Center, Cincinnati, OH) following the manufacturer’s instructions.

The PCR primers for exons 10 and 11 were used as described in Ref. 13, and primers for exons 13, 14, and 16 were as described in Ref. 2. Reactions were initiated by denaturation for 5 min at 94°C in a programmable thermal cycler (Gene Amp PCR System 9700; Applied Biosystems Inc., Foster City, CA). PCR conditions consisted of 35 cycles of 94°C for 30 s, 60°C for 30 s, and 72°C for 1 min, with a final elongation of 7 min at 72°C. Each exon was amplified separately.

Fresh PCR products were ligated into PCR-TOPO vectors, and subcloned using the TA cloning system (Invitrogen, Carlsbad, CA). Bidirectional sequencing was performed using a Taq dideoxy terminator cycle sequencing kit and an ABI 377 DNA sequencer (Perkin-Elmer, Foster City, CA).

Each PCR product of exon 11 in SNU-MEN2A-I-III was digested with HinP1I restriction enzyme (NEB Inc., Beverly, MA), and the fragments obtained were separated by electrophoresis on 2% agarose gels.

All 67 of the oligonucleotides with 12 carbon spacers were synthesized by MWG-Biotech (Ebersberg, Germany). They were 5′ modified with amino residues for Schiff’s base reaction with an aldehyde group on a glass slide. Ten pmol/μl of oligonucleotide in micro spotting solution (TeleChem International Inc., Sunnyvale, CA) was printed on the aldehyde-coated glass slide (26 × 76 × 1 mm; CEL Associates Inc., Houston, TX) using a pin microarrayer (Cartesian Microsys 5100; Cartesian Technologies Inc., Irvine, CA). A total of 67 oligonucleotides were printed on each glass slide in a 3.7 mm × 7.6 mm. Spot spacing was 300 μm, and spot size was 130 μm. After printing, the RET microarray was dried at room temperature, at least overnight, and then stored at 4°C. All 67 of the oligonucleotide sequences are shown in Table 2.

In the λ exonuclease-based method, PCR was carried out using 5′-phosphate modified forward primer and the original reverse primer in a volume of 25 μl containing the following: 100 ng of genomic DNA, 10 pmol of each primer, dNTP at 40 μm each, 20 μm Cy5-dCTP (Amersham Pharmacia-biotech Ltd., Buckinghamshire, United Kingdom), and 0.5 units of Taq polymerase (Intron-biotechnology, Seoul, Korea). After purification, the PCR product was digested with 5 units of λ exonuclease (Amersham Pharmacia-biotech Ltd.) to obtain single-stranded DNA. Asymmetric PCR, which contains a reverse primer and 20 μm of Cy5-dCTP, was performed using ethanol-precipitated PCR product without fluorescence-labeled dNTP.

In the UDG-based approach, PCR reactions were carried out in a volume of 25 μl containing 100 ng of genomic DNA, 10 pmol of each primer, dNTP at 40 μm each, 16 μm of dUTP, 20 μm of Cy5-dCTP, and 0.5 units of Taq polymerase. After the PCR amplification, the Cy5-labeled PCR product was purified using a purification kit (Qiagen Inc., Valencia, CA) and fragmented by adding 2 units of UDG (MBI Fermentas Inc., Hanover, MD) at 37°C for 4 h. Enzyme inactivation was performed at 95°C for 5 min in 50 mm of NaOH and 5 mm EDTA (14). For the DNase I-based system, PCR amplification containing dNTP at 40 μm each and 20 μm of Cy5-dCTP was performed. The PCR products were then purified and digested with 0.25 units of DNase I (Takara, Shiga, Japan) at 25°C for 10 min. Enzyme was inactivated at 95°C for 10 min.

The RET microarray was rinsed with 0.2% SDS three times and denatured at 95°C for 3 min. It was then rinsed with sodium borohydride (Sigma Chemical Co., St. Louis, MO) and 0.2% SDS. Finally, it was washed in distilled water. Prepared PCR products of each exon were mixed, resuspended in prewarmed 1 × UniHyb Solution (TeleChem International Inc., Sunnyvale, CA) at 3μℓ, and hybridized in a saturated vapor tube at 60°C for 3 h. The hybridized microarray was rinsed at room temperature in a buffer of 2 × SSC and 0.2% SDS. Both the hybridization and the washing steps were performed in the dark.

The RET oligonucleotide microarray was scanned using a ScanArray Lite (Packard Instrument Co., Meriden, CT) and analyzed using Quantitative Microarray analysis software (QuantArray, version 2.0). To establish a cutoff, all of the data analysis was carried out using a SigmaPlot (SPSS Inc., San Rafael, CA), and means and SDs were calculated.

Each oligonucleotide was printed four times horizontally, in an upward direction starting from position 1, and the printing was divided into three groups. Positions 1–25 were printed first, then positions 26–50, and finally, positions 51–67. Positions 1 and 50 were positive controls for exons 10 and 11, respectively. The other oligonucleotides were as follows: 2–33 for exon 10 (codons 609, 611, 618, and 620), 34–49 for exon 11 (codons 630 and 634), 51–58 for exon 13 (codon 768), 59–65 for exon 14 (codon 804), and 66 and 67 for exon 16 (codon 918).

We screened for the presence of RET mutations in all five of the SNU-MEN2A families and three SMC-MEN2A families. Forty-six of 62 members of five SNU-MEN2A families were sequenced. Screenings of the RET gene were performed by bidirectional sequencing in exons 10, 11, 13, and 14. RET mutations were found in five SNU-MEN2A families and three SMC-MEN2A families. SNU-MEN2A-I had a C634W mutation in exon 11 (codon 634, TGC → TGG, Cys → Trp). SNU-MEN2A-II and SNU-MEN2A-I had a C634R mutation (codon 634, TGC → CGC, Cys → Arg) in exon 11. SNU-MEN2A-II had a C618S mutation (codon 618, TGC → AGC, Cys → Ser) in exon 10. In the case of three SNU-MEN2A families (SNU-MEN2A-I-III), each PCR product was digested with Hin Pl I to determine the cosegregation of these mutations. All of the samples produced a new Hin Pl I restriction site at codon 634. The SNU-MEN2A-V family appeared to be a wild type by direct sequencing. After the RET oligonucleotide microarray analysis, cloning and sequencing were performed, and the C634Y mutation (codon 634, TGC → TAC, Cys → Tyr) was confirmed (see “Discussion”; Table 1). The SMC-MEN2A-I, -II, and -III families had a C634R (codon 634, TGC → CGC, Cys → Arg), C618R (codon 618, TGC → CGC, Cys → Arg), and C634Y (codon 634, TGC → TAC, Cys → Tyr), respectively.

Twenty-three of 62 members from five SNU-MEN2A families and 13 members from three SMC-MEN2A families were available for the RET oligonucleotide microarray analysis. Twenty-two of 23 SNU-MEN2A family members had been automatically sequenced previously. One (SNU-MEN2A-V) sample was analyzed using the RET oligonucleotide microarray, and confirmed by cloning and sequencing. DNase I- based hybridization results of 22 previously sequenced samples were consistent with the sequencing data. SNU-MEN2A-I showed specific signals in all eight of the positions (positions 9, 17, 25, 33, 41, 49, 58, and 65) exhibiting wild type, two positive controls (positions 1 and 50), and an additional signal at position 48. This signal at position 48 indicated the presence of a C634W mutation in codon 634 (Fig. 1,A). SNU-MEN2A-II and -III families showed an additional signal at position 42 of C634R (Fig. 1 B). SNU-MEN2A-IV had an additional signal at position 19 indicating the C618S mutation. In SNU-MEN2A-V, we found an additional signal at position 46, indicating the C634Y mutation. SNU-MEN2A family members with no RET mutation showed only signals at wild type and positive control positions. SMC-MEN2A-I, -II, and -III had an additional signal at positions 42, 18, and 46, which indicate C634R, C618R, and C634Y, respectively.

The signal intensities of the oligonucleotides that were spotted in quadruplicate were averaged, and these averages values at each position were regarded as the real signal values. All of the signals from the wild types and the positive controls were then excluded. In the case for the presence of a mutation, a signal showing the strongest intensity among the remainder was also excluded, and the remaining values were defined as background signals. The mean (BA) and SD (BSD) of the background signals were calculated, and BA + 2.58 BSD was established as the cutoff level. We regarded any signals over the cutoff level to be significant signals. Data are presented graphically in Fig. 2, which also shows a horizontal solid line showing BA + 2.58 × BSD (cutoff value) in SNU-MEN2A-I.

PHEO is more common in patients with C634 mutations than in those with C618 mutations (9). Korean MEN2A patients with MTC and C634 mutations (SNU-MEN2A-I, -II, -III, -V, SMC-MEN2A-I, and -III) had ∼80% (12 of 15) frequency of PHEO. However, SNU-MEN2A-IV and SMC-MEN2A-II with the MTC and the C618 mutation showed a PHEO occurrence of 15% (2 of 13). It was reported that PHEO is shown in 50% of MEN2A (9). The frequency of PHEO in Korean MEN2A patients was also 50% (14 of 28). SNU-MEN2A-IV had 1 of 11 PHEO patients, but SNU-MEN2A-I, SNU-MEN2A-II, SNU-MEN2A-III, and SNU-MEN2A-V had 2 of 5, 2 of 2, 2 of 2, and 1 of 1 patients with PHEO, respectively. SMC-MEN2A-I, -II, and -III had 3 of 3, 1 of 2, and 2 of 2 PHEO incidence. None of eight Korean MEN2A families had hyperparathyroidism.

Many cancer-causing genes show different mutation types such as frameshift, nonsense, missense, and an alternative splicing. To identify mutation types in cancer patients, many tools such as single-strand conformational polymorphism, protein truncation test, and gel-based sequencing have been used. Because these tools are relatively time-consuming and labor intensive, a rapid and high throughput system, such as that using the oligonucleotide chip, has been developed. Although the oligonucleotide chip can efficiently detect point mutations, it is not easily applied to some genes with multi-bp frameshift mutations. Thus, the oligonucleotide microarray would be very effective as a mutation detection tool in those genes dominated by point mutations. (Because the predominant RET mutations are missense mutations of some restricted codons in MEN2 syndromes, the oligonucleotide microarray method can efficiently detect mutations in this case.)

We developed the RET oligonucleotide microarray covering nine codons in which predominant RET missense mutations have been reported in MEN2 syndromes. Seven or eight oligonucleotides per codon were designed for MEN2A and FMTC. Because >95% of MEN2B patients had a predominant mutation type in codon 918 (Met → Thr), we designed only two oligonucleotides of the wild and the mutant one (2). All of the oligonucleotides were printed four times to obtain the accurate signal value.

After preliminary experiments, we found that some G-A mismatches often exhibited nonspecific signals (positions 4, 20, and 44). It was reported that G-T and G-A mismatches slightly destabilize a duplex, whereas A-A, T-T, C-T, and C-A mismatches cause significant destabilization (15). Whereas it was also reported that a high washing temperature enables a clear distinction to be made between the perfectly matched duplex and the mismatched duplex (15), the signal intensities in our work were reduced without sufficient distinction between specific signals and nonspecific signals throughout room temperature and 60°C. Thus, we reduced the length of the oligonucleotides from 20 mer to 16 mer in three cases (positions 4, 20, and 44) exhibiting a G-A mismatch to increase the destabilizing effect. These 16 mer oligonucleotides were analyzed with perfectly matched oligonucleotides. In another report, the length of oligonucleotides was reduced to 14 bases to increase specificity (16). However, an excessively short oligonucleotide could result in hybridization with unrelated DNA (17).

In sample preparations for the hybridization, we incorporated Cy5-labeled dCTP directly to the synthesizing strand during PCR without any extra step. We attempted a multiplex PCR for the sake of simplicity, but found that individual PCRs produced the better result. Four approaches were used to prepare samples for hybridization to determine which provided the best hybridization result, namely, asymmetrical PCR-based, λ exonuclease-based, UDG-based, and DNase I-based preparations. Asymmetric PCR-based and λ exonuclease-based preparations were investigated to obtain single-stranded DNA for effective hybridization. UDG and DNase I digestion were introduced to fragment PCR products, which decrease the interference of hairpin structures on hybridization with oligonucleotides (6). Unpredictable signals were shown in the asymmetric PCR, although all of the specific signals were present. The λ exonuclease-based preparation generally showed reinforced signals at positions 2–17 (codon 609 and 611). When localizing positions 2–17, specific signals were shown with more than three times the intensity of the background signals. However, the mean value of the background signals in these areas (positions 2–17) was similar to the specific signal value of the other codons in terms of the entire area of the RET oligonucleotide microarray. Therefore, this approach required a more detailed analysis, as such comparing each signal within an individual codon. In the UDG-based hybridization approach, SNU-MEN2A-I, -II, -III, and -V families with a mutation in exon 11 showed a specific signal in exon 11 area. However, in the case of the exon 10 hybridization, the UDG-based method often showed nonspecific signals at positions 26–28, regardless of the ratio of dUTP:dTTP. The DNase I-based approach showed relatively low incidence of the nonspecific signals. Specific signals in DNase I-based method were usually twice as strong as those obtained using the other methods and showed relatively similar signal intensity in the entire region of the RET microarray. The DNase I-based preparation was straightforward enough to allow the whole experiment (from PCR amplification to the scanning of the hybridized microarray) to be finished within a day. Therefore, we concluded that DNase I-based sample preparation was the most efficient and preferable approach for the RET oligonucleotide microarray. The hybridization buffer was considered to minimize the difference between the G:C and A:T bp (18). SNU-MEN2A-V was first analyzed by RET oligonucleotide microarray and then reexamined by direct sequencing. To confirm the signal at position 46 (C634Y type) of the RET oligonucleotide microarray in SNU-MEN2A-V, direct sequencing was performed. As a result, codon 634 seemed to be of the wild type. We could not confirm the presence of the signal at position 46 (C634Y mutation, TGC → TAC, exon 11) until we analyzed exon 11 by cloning and sequencing. If we had relied on the direct sequencing approach, we could have missed the C634Y mutation in the SNU-MEN2A-V patient. We examined the incidence of false positive and false negative three times using 13 samples from three SMC-MEN2A families under a blind status. One of 39 showed false negative (codon 618, position 18) and another case of 39 showed false positive (codon 620, position 27). One SMC-MEN2A-II member with C618R showed no distinctive signal at position 18, and the wild-type signal at codon 618 was also weak in this case. We spotted wild-type oligonucelotides in each codon. When we could not see the specific signals over the cutoff level even at wild-type oligonucleotides, our experiment was repeated. By this procedure, we could reduce false positive and false negative. Nonspecific signals might be caused by some problems in PCR amplification or sample preparations for hybridization such as enzyme digestion. In a false positive case, one sample with C634R showed a nonspecific signal at position 27 along with a specific signal at position 42 (codon 620). In this case, signals were generally strong, and wild-type oligonucleotide at codon 620 showed signals twice as strong as other wild-type signals. This might be affected by the DNase I enzyme digestion. We spotted 67 type oligonucletides covering nine codons (codons 609, 611, 618, 620, 630, 634, 768, 804, and 918) of five exons (exons 10, 11 13, 14, and 16). Because codons 609 and 618 are contiguous to codons 611 and 620, respectively, these codons share some parts of oligonucleotides each other. DNA fragments randomly digested by DNase I may affect the specificity of hybridization at nearby codon and cause either a false-positive or false-negative result.

When analyzing the data, we averaged the signal values of four spots and accepted the average as the real signal value, and when fluorescent debris was shown in a spot, we excluded it and calculated the signal mean value from the remaining three. The signal recognition cutoff level was set as (BA + 2.58 × BSD). (BA + 2.58 × BSD) indicates the upper limit of the 99% confidence interval, and signals more than this value were identified as specific signals.

Genotype-phenotype correlations in RET proto-oncogene and MEN2 syndromes have been well characterized, and it has been suggested that the codon 634 mutations are highly predictive of the presence of PHEO. We found that the C634 mutations (6 of 8) are prevalent in Korean MEN2A families and the correlation between C634 mutations, and that the occurrence of PHEO (14 of 28) in Korean MEN2A patients is in agreement with much of the data published previously. Six of eight Korean MEN2A families (SNU-MEN2A-I, -II, -III, -V, SMC-MEN2A-I, and SMC-MEN2A-I) showed mutations at codon 634. C634R mutation was the most common in MEN2A patients followed by C634Y in MEN2A patients (9). Three of eight Korean MEN2A families showed C634R, and two showed C634Y. Five of eight Korean MEN2A families showed C634R or C634Y, which is consistent with the data reported previously. Although there were some discrepancies, it has been suggested that the C634R mutation is consistent with a higher frequency of hyperparathyroidism (2, 3, 9). An age-related and mutation-specific hyperparathyroidism penetrance was reported in a series of patients carrying codon 634 mutations; e.g., the penetrance was 14% by age 30 and rose to 81% by age 70 (2, 3, 10). We could not find any clinical symptom of hyperparathyroidism in eight Korean MEN2A families. The age of MEN2A family members with the codon 634 mutations ranged from 6 to 67. Thus, we could not exclude the possibility that Korean MEN2A families with C634 mutations will be at risk for hyperparathyroidism in the future. It is also possible that the thyroidectomy performed in some patients with the removal of parathyroid glands had influence on the occurrence of hyperparathyroidism (2). SNU-MEN2A-IV and SMC-MEN2A-II had mutations at codon 618, and the incidences of PHEO were 1 of 11 and 1 of 2, respectively. Usually 5′ cysteine codon mutations (e.g., codon 609, 611, and 618) are associated with FMTC, and these mutations might result in a weaker activation than those nearer the transmembrane domain. Codon 634 mutations appear to have the penetrance high enough for MTC, PHEO, and hyperparathyroidism. However, 5′ cysteine codon mutations seemed to cause a reduced proportion of receptor molecules on the cell surfaces, thus resulting in only MTC (4). The low PHEO frequency in families with C618 mutations might be caused by this 5′ cysteine codon mutation.

In summary, we have developed a RET oligonucleotide microarray for the detection of RET mutations, and found RET mutations in all five SNU-MEN2A and three SMC-MEN2A families by gel-based sequencing and the RET oligonucleotide microarray. RET mutations were reliably detected by the RET microarray from the DNA of MEN2A families within a day. Because the MEN2 syndromes are inherited in an autosomal dominant fashion and show high penetrance, it is important that we test for the presence of RET mutations in MEN2A families in the early stages. Twenty-two mutation carriers in eight Korean MEN2A families were found using RET oligonucleotide microarray and direct sequencing. A close monitor is required for these mutation carriers. Because we found only five mutation types in eight MEN2A families, the international collaborations are additionally required to see whether the RET oligonucleotide microarray may be used as one of the general genetic diagnostic tools. Taken together, the RET oligonucleotide microarray may present an effective diagnostic genetic tool, which simplifies the detection of RET missense mutations.