Zhou et al: HLA Matching using DNA Melting Curves18-10-18
High Resolution DNA Melting Curve Analysis and the Identification of Shared HLA Alleles Between Siblings
L. Zhou, J. Vandersteen, L. Wang, T. Fulder, R.Palais, C. Wittwer,
Running title: HLA Matching by DNA Melting Curve Analysis
Department of Pathology,
University of Utah,
Salt Lake CityUT
Abstract:
Introduction:
HLA must match in different kinds of transplantation.
Existing methods of identifying HLA genotypes suffer in: cost, speed, throughput
Special case of siblings and identity by descent
Discuss melting curve analysis of homo and heteroduplexes.
D6 facilitates accurate DNA melting curve analysis
Application of this melting curve methodology to HLA genotyping
Materials and methods
DNA Samples
Cell pellets from American Society for Histocompatibility & Immunogegetics (ASHI) were the source of genomic DNA for seven characterized HLA genotypes. These cell lines and their genotypes are shown in Table 1 (genotype names follow the international convention (REF)). The sequence of each haplotype is available in the IMG/HLA sequence database ( DNA was extracted using the DNA isolation kit from PUREGENE (Gentra Systems).
Genomic DNA samples from the CEPH/Pedigree Utah family 1331 (REF) were obtained from the Coriell Institute. There are 17 individuals among three generations in this family including four internal grandparents, two parents and eleven children. The pedigree of family 1331 is shown in Figure 1.
Amplification of the HLA-A gene and melting curve analysis
Exons 2 and 3 of the HLA-A gene were amplified using a nested PCR strategy in which an initial PCR specifically amplified a large (948bp) fragment of the HLA-A gene followed by secondary amplifications of that product using internal primers. The primers used in the first PCR hybridized to HLA-A intron 1(forward primer 5’ GAAAC(C/G)GCCTCTG(C/T)GGGGAGAAGCAA, REF) and intron 4 (reverse primer 5’- TGTTGGTCCCAATTGTCTCCCCTC, REF) . In the secondary PCRs the forward primer 5’AGCCGCGCC(G/T)GGAAGAGGGTCG and reverse primer 5’GGCCGGGGTCACTCACCG were used to amplify a 340bp segment of HLA-A exon 2. The forward primers 5’CCC(G/A)GGTTGGTCGGGGC and reverse primers 5’ATCAG(G/T)GAGGCGCCCCGTG were used to amplify a 366bp fragment of HLA-A exon 3. All PCRs were performed in glass capillaries using the Roche LighterCycler. The initial PCR contained 0.5 µM forward and reverse primers, 50ng genomic DNA in a buffer of 3mM MgCl2, 50mM Tris-HCl pH 8.3, 0.2mM dNTP, 500µg/ml BSA and 0.2mM of dye D6 in 10µl. Cycling conditions were 94°C for 20 s followed by 40 cycles of 94°C 1 s, 62°C for 0 s, 72°C for 1 min. The secondary, nested PCRs contained 0.25 µM forward and reverse primer, 1/10,000 of the first PCR product in the PCR buffer containing 2mM Mg Cl2. Cycling conditions were 94°C for 5 s, 25 cycles of 94°C 1 s, 65°C for 0 s, 72°C for 8s. A melt was performed on LightCycler to examine the PCR productfollowed by the PCR which is 94°C 0s, 60°C 0s and melted from 60°C to 94°C at a rate of 0.3°C/s.
The glass capillaries containing the secondary amplification were transferred to the high resolution melting instrument HR-1(REF), and a melt was performed. The sample was heated from 60°C to 94°C at a rate of 0.3°C/s and fluorescence (450 excitation/470 emission) and temperature measurements were acquired every 40 ms. A sentence here about the normalizations done to the data.
To perform melting curve analysis of reannealed PCR products from different genomic DNA samples, 25ng of each of the two genomic DNAs were mixed before the firs PCR, and the analysis performed as described.
The nested PCR products were sequenced using an ABI 3700 at the University of Utah Core facility using the amplification primers. Sequencher version 4.0 was used for the sequence analysis.
The computer analysis of melting curves and of Tms
The software MultiPrimer was designed to choose the best primers. It also has ability to calculate the melting temperature of the amplicons. The model of melting temperature was calculated as follows: The melting temperature, TM, at which half of the strands are in the double-helical state and half in random-coil state is predicted using the nearest-neighbor thermodynamic model and experimental parameters reported in SantaLucia Jr. et. al. (REF) where a phase transition is predicted by 0 = dG = dH – TdS so TM = dH/dS = (∑ΔH (tetrads + ends))/(∑ΔS (tetrads + ends) +RlnCT +0.368N ln[Na+ ]) where dH and dS are the contributions to free energy in kcal/mol and to entropy in cal/Kelvin*mol from the internal conguration of the oligonucleotide (interior tetrads and end pairs), R is the gas constant (1.987 cal/Kelvin*mol), CT is the difference of the larger and half the smaller oligonucleotide strand concentrations (target and probe), N is one half of the total number of phosphates in the duplex (e.g. for an 8-bp duplex without terminal phosphates, N=7). The module TM Calculator.vi estimates the melting temperature by calling Thermo-DynamicOligos.vi which in turn refers to lookup tables of contributions to ΔH and ΔS between double-helical and random-coil states. dHdSinit.vi contains the values for A-T or G-C terminal ends. ThermodynamicPairs.vi contains the data for internal tetrads: 16 configurations with two (neighboring) matched pairs ([1]), 64 with one matched pair and one unmatched pair (16 for each unmatched pair, G-T ([2]), G-A ([3]), A-C ([4]), C-T ([5]), and 32 with AA, CC, GG, TT mismatches ([7])), and 32 tetrad configurations with one match adjacent to a dangling end" ([8]).
Results
Sentence fragments:
HLA class I genes are so similar over of the length of their coding exons that it is difficult to design PCR primers which amplify only the HLA-A gene and not the related class I genes [ ].
Nucleotide mismatchs and the Tm’s of HLA-A PCR fragments
As a first step towards using melting curve analysis to differentiate the HLA alleles of siblings, we asked whether this analysis could separate the HLA genotypes of seven arbitrarily chosen known haplotypes (Table 1) into identical groups. The software MultiPrimer (Material and Methods) was used to predict the melting temperature of a PCR fragment generated from HLA-A exon 2 of the haplotypes present. Despite sequence differences of up to 6% (see values above the diagonal in Table 2) the predicted melting temperature are all very close (less than 1ºC difference between the highest and lowest values). However when the computer analysis is performed on the hybrids (heteroduplexes) formed between PCR fragments from different haplotypes the predicted melting temperatures can be as almost 4°C lower than that of either pure (homoduplex) haplotype alone.
Melting curve analysis of mixed samples to differentiate genotypes
The analysis shown in Table 2 suggests that donor and recipient of different genotype may have the same or similar melting temperature and melting curve. To determine whether the donor and recipient have the same haplotypes, it is necessary compare the melting curve of donor, recipient and their mixed sample. If donor and recipient have the same genotype, the melting temperature and melting curve of their mixed sample will be the same as donor and recipient alone. If the donor and recipient have different genotypes, the melting temperatures may be the same or close so that they may have the same or similar melting curves but the melting curve of the mixed sample will be significantly different. A clear example of this is provided by the two homozygous samples BM15 (0101) and BM16 (0201). In this case there are 15 nucleotide mismatch differences spread over the length of the HLA-A exon 2 PCR fragment (Table 2). The calculated melting temperature of BM15 (0101) is 93.68°C and that of BM16 (0202) is 93.71°C, and thus their melting curves are very similar (Fig 2g). The melting curve of the mixed sample is shifted to the left because of the15 mismatches present in the heterohybrids generated by reannealing of the PCR fragments from HLA-A exon 2 (see Figure 2g)
Identify the alleles by melting Curve
Both the recipient and donor can be homozygous (two alleles the same) or heterozygous (two alleles different). There are 7 possible combinations when a recipient and donor are mixed (Table 3). After PCR, the product will be melted at 94ºC and reannealed, the different alleles will reanneal at random and the ratio of heteroduplexes will depend on the how many alleles are in the combination. The percentage of heteroduplexes in mixed samples are different in the 7 possibilities. This is shown for the HLA-A exon2 PCR fragments from mixed samples in Table 3. The melting curve of mixed samples is dependent upon the number of nucleotide mismatches, how the mismatches are spread over the amplicon and the number of different alleles. The more heterozygous in the sample, the lower the melting temperature will be. Comparison of the melting curve of donor, recipient and their mixed sample will easily identify alleles from donor and recipient. Identity of the melting curves of donor, recipient (either homozygous or heterozygous) and their mixture suggests that the donor and recipient have the same genotype. Differences in any of these melting curves suggests that there are different genotypes. This is demonstrated for the 7 possible cases involving the melting curves of HLA-A exon2 (Fig. 2).
Case 1: Both recipient and donor are homozygous for the same allele. Two samples BM15 and E418 have identical HLA-A homozygous genotype (0101). Only homoduplexe are formed upon the reannealing of the PCR products. This is evident from the fact that the melting curve formed when the two PCR reactions are mixed is identical to either alone and both show a single melting transition (Fig. 2A).
Case 2: Both recipient and donor share the same two alleles. Two samples CF966 and EMJ share the same HLA-A heterozygous genotype (0201/0301). The heteroduplexes formed in the mixed sample is of the same type as that of the intrasample heteroduplex. However compared to case 1 there is clearly more than one melting peak evident, indicating that homo- and hetero- duplexes are formed upon reannealing of the PCR reaction (Fig. 2B).
Case 3: Recipient is homozygous for one allele shared with a homozygous donor. One sample is homozygous BM16 (0201/0201) and the other sample is heterozygous CF966 (0201/0301). They share one allele 0201. In the mixed sample, there are 6 heteroduplexes for every 10 homoduplexes (or 37.5% heteroduplexes). The mixed-sample melting curve falls between the two individual samples along (Fig. 2C).
Case 4: Recipient is homozygous, donor is heterozygous and they share no alleles. One sample is homozygous BM16 (0201/0201) and the other sample is heterozygous PMG075 (0301/3301). In the mixed sample, homoduplexes and heteroduplexes are in the ratio of 6?/10? (62.5% heteroduplex, Fig. 2D).
Case 5: Recipient and donor share one allele. The two heterozygous samples EMJ (0201/0301) and PMG075 (0301/3301) share one allele (0101). When the PCR reactions are mixed and reannealed, the 3 different alleles produce 6 homoduplexes and 10 heteroduplexes (62.5% heteroduplex). The melting curve of the mixed sample is clearly different from the other two (Fig. 2E).
Case 6: Recipient and donor are heterozygous and share no alleles. The most complicated case involves two samples CF966 (0201/0301) and LKT14 (2402/2602) with two different alleles. When the PCR reactions are mixed and reannealed, the 4 different alleles produce 4 homoduplexes for every 12 heteroduplexes (75% heteroduplex). Again the mixed sample is clearly different from the other two (Fig. 2F).
Case 7: Recipient and donor are homozygous and share no alleles. The two samples BM15 (0101/0101) and BM16 (0201/0201) with different homozygous genotypes. In this case, the two alleles have a total of 15 nucleotide differences spread over the length of the HLA-A exon 2 PCR fragment but they show similar melting curves. The melting curve of the mixed samples was significantly shifted to the left (lower melting temperature) when the PCR reactions are mixed and reannealed, the 4 different alleles produce 8 homoduplexes for every 8 heteroduplexes (50% heteroduplex) (see Fig. 2G).
Family 1331’s HLA-A’s alleles identification testing
A large family CEPH/pedigree Utah family was used to this technique for grouping alleles among siblings. Melting curve analysis of HLA-A exon 2 PCR products amplified from the 17 members of the CEPH/Pedigree Utah family 1331 clustered into six different groups (Fig. 3), which suggested that there are at least six different HLA-A genotypes in the family. In order to further split these groups into subgroups with different genotypes having similar melting curves, the samples from each group were mixed subjected to melting curve analysis after PCR. In particular one sample from each group was chosen to represent the genotype of the group and mixed (as genomic DNA) with each of the other samples in the group. In every case the mixture had the same melting curve as the parents. In contrast mixing a representative of each group with one from any other group generated melting curves distinct from either parent (data not show).
A similar analysis was performed on this family using a PCR fragment amplified from exon 3 of HLA-A and generated the same groups (data not shown)
The 17 people of CEPH/Pedigree Utah family 1331 of HLA-A exon 2 and exon 3 PCR products were sequenced, and the results confirmed the melting curve analysis, identifying the six genotypes as: HLA-A 02011/3101 (herein referred to as genotype AB) for family members 1, 4, 7,12; HLA-A 3101/2402101 (genotype BC) for family members 3,5,6,11,17; HLA-A 02011/2402101 (genotype AC) for family members 2,9,10,16, HLA-A 02011/03011 (genotype AD) for family members 13, 14; HLA-A 02011/02011 (genotype AA) for family member 8 and HLA-A 2402101/01011 (genotype CE) for family member 15.
Discussion
The utility of this technique for HLA genotype testing
Two demonstrations of the ability to group a set of alleles by identity.
A large family such as this one is ideal to test HLA genotype matching.
Comparison with other HLA typing techniques
Combining this technique with sequencing will be powerful
References
Tables
Table1:
Sample Name / GenotypesBM15 / 0101:0101
E418 / 0101:0101
CF966 / 0201:0301
EMJ / 0201:0301
BM16 / 0201:0201
PMG075 / 0301:3301
LKT14 / 2402:2602
Cell lines and their respective HLA-A genotypes are indicated.
Table 2:
Haplotypes / 0101 / 0201 / 0301 / 2402 / 2602 / 33010101 / 93.68 / 15 / 7 / 16 / 10 / 15
0201 / 91.25 / 93.71 / 12 / 16 / 15 / 16
0301 / 92.57 / 91.79 / 94.11 / 18 / 13 / 12
2402 / 90.60 / 90.83 / 90.18 / 93.86 / 19 / 20
2602 / 92.20 / 91.62 / 91.62 / 90.24 / 94.05 / 12
3301 / 91.99 / 91.89 / 92.49 / 90.35 / 92.49 / 94.23
Number of nucleotide mismatches between HLA-A haplotypes and the predicted melting temperature of the reannealed hybrids. The haplotypes used in this study are indicated on the left column and top row. Numbers above the diagonal indicate the number of mismatches present in DNA hybrids formed between exon 2 PCR products (340bp) from the indicated haplotypes. Numbers on and below the diagonal (bold) indicate the predicted Tm of the indicated hybrid DNA molecule calculated using the program. The diagonal contains the predicted Tm’s of the homoduplex PCR products.
Table3:
Sample1
/ Sample2Cases /
Haplotype1
/ Haplotype2 / Haplotype1 / Haplotype2 / # of Alleles / % Heterduplex1 / 0101 / 0101 / 0101 / 0101 / 1 / 0
2 / 0201 / 0301 / 0201 / 0301 / 2 / 50
3 / 0201 / 0201 / 0201 / 0301 / 2 / 37.5
4 / 0201 / 0201 / 0301 / 3301 / 3 / 62.5
5 / 0201 / 0301 / 0301 / 3301 / 3 / 62.5
6 / 0201 / 0301 / 2402 / 2602 / 4 / 75
7 / 0101 / 0101 / 0201 / 0201 / 2 / 50
Seven possible combinations of recipient and donor are mixed.
Figures
Figure 1:
Pedigree of CEPH referenced Utah family 1331. The genotype of HLA-A of Utah family 1331 are as follows: A:02011; B:3101; C:2402101; D:03011; E:01011. Each individual is numbered. Female (circle); male (square).
Figure 2:
DNA melting curve analysis of 7 possible combinations of mixture of recipient and donor.
(A) Both recipient and donor are homozygous for the same allele. Both BM15 and E418 have the same homozygous genotype of HLA-A (0101/0101).
(B) Both recipient and donor share the same two alleles. Sample CF996 and EMJ have the same HLA-A heterozygous genotype (0101/0202).
(C) Recipient is homozygous for an allele shared with the donor. One sample BM16 is homozygous 0201/0201, another sample is heterozygous CF966 (0201/0301).
(D) Recipient and Donor share no alleles. Recipient is homozygous, donor is heterozygous and they share no alleles. One sample is homozygous BM16 (0201/0201) and the other sample is heterozygous PMG075 (0301/3301).
(E) Recipient and Donor share one allele. Two heterozygous samples EMJ (0201/0301) and PMG075 (0301/3301) are with one shared allele (0101).
(F) Recipient and Donor are heterozygous and share no alleles. Two samples CF966 (0201/0301) and LKT14 (2402/2602) are the different heterozygous genomtypes.
(G) Recipient and Donor are homozygous and share no alleles. Two samples BM15 (0101/0101) and BM16 (0201/0201) with different homozygous genotypes.
Figure 3:
Melting curve of Utah family 1331 members. There are six different melting curves representing six genotypes in HLA-A exon 2 exist among 17 family members. Fig 3a shows the full melt curve. Fig 3b shows an enlarged portion (shown in square in 3a) with the genotype designations, and individuals in parentheses.
1