HLA class II “typing”: Direct sequencing of DRB, DQB, and DQA genes

HLA Class II "Typing": Direct Sequencing of DRB, DQB, and DQA Genes Pere Santamaria, Michael T. Boyce-Jacino, Alan L. Lindstrom, Jose J. Barbosa, Anthony J. Farm, and Stephen S. Rich

ABSTRACT: Routine clinical HLA class II typing is based largely on serological and cellular methods. These methods have many drawbacks that have led to the evaluation of molecular approaches to typing, including restriction fragment length polymorphism studies and oligotyping. We present here an alternative molecular approach, sequence-based typing (SBT), that allows direct determination of the sequences of all HLA class II polymorphic genes, thus providing the most detailed information currently possible in this regard. The data presented here using SBT are based on direct sequencing of polymerase chain reaction (PCR)-amplified DRB, DQB, and DQA cDNAs using a limited number of oligonucleotides. The oligonucleotides are designed to allow simultaneous determination of allelic sequences in any heterozygote as well as characterization of DRB isotypic complexity. Two types of amplification oligonucleotides (nonconserved and/or conserved) are used for DRB typing, which involves a maximum of four simultaneous cDNA/PCR/sequencing reactions. The first of these reac-

tions only uses conserved oligonucleotides and is designed to detect all the different DRB transcripts present in any given heterozygote; the other three reactions use nonconserved oligonucleotides and are designed to ensure the unambiguous interpretation of the most complex DRB heterozygote combinations. Characterization of DQA 1 and DQB I sequences can be performed by using conserved oligonucleotides and only involves one reaction per locus. We have applied SBT to 43 homozygous cell lines and to 38 different heterozygote combinations that had previously been serologically typed. In all cases we were able to determine the allelic composition at DRB1, DRB3/4/5 and/or DQB1, and DQA1 loci of these cell lines and subjects; our results, analyzed by blind protocol, were consistent with the serological phenotypes. SBT can be extended to class I and class III genes and is automatable. We believe that this strategy deserves further evaluation as a possible HLA typing method. Human Immunology 33, 69-81 (1992)

ABBREVIATIONS LCL lymphoblastoid cell lines PBMC peripheral blood mononuclear cells PCR polymerase chain reaction

RFLP SBT

restriction fragment length polymorphisms sequence-based typing

INTRODUCTION The class II genes o f the H L A - D region on the short arm of human chromosome six constitute one of the most polymorphic genetic systems known [ 1]. The class

From the Institute of Human Genetics (P.S.; M.T.B.-J.; A.L.L.; A.J.F.; S.S.R.), Division of Endocrinology and Metabolism (P.S.,'JJ.B.), Department of Medicine, and Department of Laboratory Medicine and Pathology (S.S.R.), University of Minnesota, Minneapolis, Minnesota. Address reprint requests to Dr. Pere Santamaria, Box 716 UMHC, Department of Medicine, University of Minnesota, Minneapolis, MN 55455. Received February 8, 1991; acceptedSeptember21, 1991.

Human Immunology33, 69-81 (1992) © AmericanSocietyfor Histocompatibilityand Immunogenetics,1992

II molecules (DR, DQ, and DP) are heterodimeric glycoproteins composed of two noncovalently associated chains (alpha and beta) that serve as restricting elements in nominal antigen presentation in the context of self [2] or as foreign antigens in alloresponses [3]. Allelic polymorphism o f the H L A - D r e g i o n - e n c o d e d specificities can be determined by serological methods for phenotyping, mixed lymphocyte cultures using homozygous typing cells, primed lymphocyte testing, determination of restriction fragment length polymorphisms (RFLP), and, more recently, oligotyping [1, 4, 5]. For

69 0198-8859/92/$5.00

70

several reasons that have been previously discussed [46] present efforts focus largely on the development of molecular approaches to typing, such as RFLP and oligotyping. The cloning and sequencing of several HLA-DR, -DQ, and -DP alleles has revealed that their amino acid polymorphisms are located in hypervariable regions of their N-terminal domains, encoded by the second exon of DRB1, DRB3/4/5, DQA1 and DQB1, and DPA1 and DPB1 genes [7, 8]. This information has allowed the design of allele-specific oligonucleotide primers, which have been used for characterization of the known HLA class II polymorphisms by means of their hybridization to DNA on a solid support (oligonucleotide typing) or for sequencing [5, 6, 8-13]. Oligonucleotide typing, although rapid, requires the use of a rather large number of oligonucleotides for each locus and cannot detect previously unidentified sequence polymorphisms likely to exist in non-Caucasian populations; further, this approach may not be easily applicable to and may not be practical for the analysis of class I polymorphisms. A method that provides the sequences for all class II alleles carried by any given individual maximizes the information that can be presently obtained. We present here what we believe to be such a method, sequencebased typing (SBT), which has been evaluated by sequencing DR and DQ polymorphic genes. This method combines enzymatic amplification of class II cDNA molecules and direct sequencing of the double-stranded amplified DNA. We have applied this method to determine the second exon nucleotide sequences of DRB1, DRB3, DRB4, DRB5, DQA1, and DQB1 genes of 43 homozygous cell lines and 38 different heterozygote combinations. The method is reproducible and the results obtained have correlated completely with established typing methods. This methodology has allowed us to detect sequence microheterogeneity within welldefined class II haplotypes and in Caucasians [14, 15]. The technique we present is rapid, requires the use of only a small number of oligonucleotide primers, can readily detect new sequence variants unidentifiable with more conventional approaches, is applicable to the analysis of other class II as well as class I and class III genes, and is automatable. MATERIAL AND METHODS

Cell lines and subjects. Lymphoblastoid cell lines (LCLs) representing most of the known class II haplotypes defined at the 10th International Histocompatibility Workshop [16] were provided by Dr. Miriam Segall (University of Minnesota). Thirty-eight different heterozygote combinations that had been previously sero-

P. Santamaria et al.

logically typed were also studied. The serological types of each of the samples under study were not known to the investigator performing the sequence analysis at the time the analysis was performed. These subjects included both healthy and affected (insulin-dependent diabetes and autoimmune thyroid disease) individuals. The cell lines and the different heterozygote combinations tested are shown in Table 1.

RNA preparation. Total cellular RNA was prepared from 5-50 x 106 peripheral blood mononuclear cells TABLE 1 Cell lines and heterozygote combinations tested A. 10th Workshop cell lines Cell line

SA MZ070782 KASOII *CALOGERO *WJR076 *DEM WT24 RML

SCHU WT8 *AMAI E4181324 MT14B EJ32B RSH

DEU WT51

JBAF YAR KT17 sPool0 JBUSH TISI JVM BM16 *H0301 WDV WT47 TEM

EK AMALA LBF BH CF96 BER DBB MOU BTB OLGA LUY TAB089 DKB

Class II haplotype DR1-Dwl DR1-Dw20 DRwl6-Dw21 DRw16-DwDRw 16-Dw21 DRw 16-Dw21/DR4 DRw 16-Dw21 DRw 16-Dw22 DRw 15-Dw2 DRw 15-Dw2 DRw 15-Dw2 DRwl5-Dwl2 DR4-Dw14 DRwl7-SYD3 DRw 18-DwRSH DR4-Dw4 DR4-Dw4 DR4-Dw 13 DR4-Dw 10 DR4-DKT2 DRwl 1-DB2 DRwl 1-Dw5 DRwl 1-DwTISI DRwl 1-DwJVM DRw 12-DB6 DRw 13 -Dw 19 DRw 13 -Dw 18 DRwl3-Dwl9 DRw 14-Dw9 DRw 14-Dw9 DRwl4-Dwl6 DR7-DB 1 DR7-DB 1 DR7-Dw7 DR7-Dw7 DR7-Dwll DR7-Dw7 DRw8-Dw8.1 DRw8-Dw8.2 DRw8-Dw8.3 DRw8-Dw8.3 DR9-Dw23

Sequence-based Class II Typing

TABLE 1

71

(Continued)

B. H e t e r o z y g o t e c o m b i n a t i o n s Class lI Loci HC

DRB 1

DRB 3

DRB4

DRB 5

DQB 1

DQA 1

1/3 1/4 l/w8 2/3 2/4 2/3 5 × 6/3 6/3 *6/3 4/3 4/w12 4/1 4/3 4/2 *4/3 w l 1/3 w12/1 w12/w8 *6/7 7/3 7/4 w8/7 w8/5×6 w8/wl 1 wS/1 w8/2 9/1 6/2 6/4 w l 1/2 wl2/wll w l 1/6 2/4 6/1 6/4 6/2 w l 1/10 4/wl 1

0101/0301 0101/0401 0101/0801 1501/0301 1501/0401 1601/0301 1303/0301 1301/0301 1302/0301 0401/0301 0401/1201 0403/0101 0403/0301 0404/1501 0405/0301 1101/0301 1201/0101 1201/0801 1401/0701 0701/0301 0701/0401 0803/0701 0801/1303 0 8 0 2 / I 101 0803/0101 0803/1501 0901/0101 1301/1601 1301/0801 1101/150l 1201/1101 1101/1303 1601/0401 1301/0101 ND ND 1104/1001 0401/1103

0101 --0101 -0101 0101 0101 0301/0101 0101 0201 -0101 -0101 0101/0201 0201 0201 0201 0101 --0101 0201 ---0101 0101 0201 0201

-0101 --0101 ----0101 0101 010l 0101 010l 0101 ---0101 0101 0101 010 l ----0101 -----

---0101 0101 0201 -------0101 -----------0101 -0201 -0101 --

0501/0201 0501/0302 0501/0402 0602/0201 0602/0302 0502/0201 0301/0201 0603/0201 DQB6.5/0201 0301/0201 0301/0301 0302/0501 0302/0201 0302/0602 0201/0201 0301/0201 0301/0501 0301/0402 DQB5.4/0201 0303/0201 0201/0301 0301/0201 0402/0301 0402/0301 0301/0501 0301/0602 0303/0501 0603/0502 0603/0402 0301/0602 0301/0301 ND 0502/0301 0603/0501 0604/0301 0604/0502 ND ND

0101/0501 0101/0301 0101/0401 0102/0501 0102/0301 0102/0501 0501/0501 0103/0501 0102/0501 0301/0501 0301/0501 0301/0101 0301/0501 0301/0102 0301/0501 0501/0501 0501/0101 0501/0401 ND 0201/0501 0201/0301 0601/0201 0401/0501 0401/0501 0601/0101 0601/0102 0301/0101 0103/0102 0103/0401 0501/0102 0501/0501 ND 0102/0301 ND ND ND ND ND

0101/0201

-0101 ND ND 0201 0201

--

0101 -ND ND -0101

--

0201 -ND ND ---

The allelic composition at DRB, D Q A 1, and D Q B I loci for the sequenced haplotypes corresponded to that expected according to published sequence information from well-characterized homozygous cell lines [23] unless indicated by (*). The specific sequence information for these cell lines, including new allelic variants, has been published elsewhere [ 15]. The first column of B lists the serological DR types of every heterozygous combination. DQB6.5 and D Q B 5.4 are local designations for two of these new allelic sequences. The serological specificity 5 × 6 was given this arbitrary designation according to serological and RFLP information [24]. ND, not determined.

(PBMC) or LCLs by guanidium isothiocyanate lysis followed by cesium chloride centrifugation [17]. Alternatively, total R N A from peripheral blood ( 2 - 1 0 ml) was partially purified using a much faster protocol based on red blood cell lysis, followed by phenol extraction and ethanol precipitation [18]. Both methods for RNA preparation yield c D N A of similar quality. Care should

be taken in removing the organic solvents if the faster protocol is used, since they could interfere with the enzymatic reactions. With regard to which is the method of choice, if the investigator prefers to store the total R N A for weeks or months before the typing is done, the cesium chloride/guanidium thyocianate method, which yields cleaner R N A (no phenol or chlo-

72

roform or proteins), is preferable. However, if the typing is to be done immediately after preparing the RNA, the faster method becomes the method of choice.

HLA-DRB, DQB, and DQA transcript amplification using conserved and nonconserved oligonucleotides. Approximately 1/,g of total cellular RNA is reverse transcribed with Moloney leukemia virus reverse transcriptase (200 U, Bethesda Research Laboratories) in 50 mM tris HC1, pH 8.3, 75 mM KC1, 10 mM DTT, 3 mM MgC12, in the presence of the ribonuclease inhibitor RNAsin (10 U, Promega, Madison, WI), 75/,M each dNTP, and 10 pmols of a specific antisense primer (see below and Table 2) in a 20 /,l final volume for 30-45 minutes at 37°C. The amount of total cellular RNA used for each individual or cell line can be widely varied (i.e., 500 ng to several micrograms), with minimal effects on either the efficiency or specificity of the reactions; we have not observed significant interindividual variability with respect to either the yield of total RNA or the efficiency of the reactions. It remains to be tested, however, whether samples from leukemic patients can be processed with the same efficiency as normal samples. After the incubation period, 8 /.d of 10X PCR buffer (500 mM KCI, 100 mM tris-Cl, pH 8.3, 0.1% gelatin) and 1.6 /,l or 4 /~l of 25 mM MgC12 (depending on the primer combinations, see Table 2) are added to the reverse transcription reaction tube (1 mM and 1.5 mM final MgC12 concentrations, respectively). A sense primer (20 pmoles) (type 1 or type 2 primers, respectively, see Table 2) plus 10 pmols of the antisense primer and 2 U of Taq polymerase (Cetus Corporation, Emmeryville, CA) are also added and the final volume is adjusted to 100 ~l with distilled water. The reaction mixture is subjected to 35 cycles of 30 seconds at 94°C, 30 seconds at 37°C, or 55°C and 30 seconds at 72°C using a Perkin-Elmer Cetus Thermocycler [9-12]. The reactions for each locus are usually performed in separate microfuge tubes. However, when using conserved primers, the cDNA and PCR reactions for several loci (i.e., DRB and DQB and/or DQA) can be successfully performed simultaneously in the same tube; all the corresponding cDNA and PCR primers are added at the same concentrations as when each locus is analyzed separately and there is no need to increase the total amount of RNA of the reaction. In that case sequencing of the products corresponding to DRB and DQB and/or DQA loci, respectively, is performed in two or three separate sequencing reactions (one per locus being analyzed), each using one third of the original reaction.

Direct sequencing of amplified products with Taq polymerase. The reaction mixture (100/~l) is freed of unincorporated dNTPs and excess of oligonucleotides by spin-dialysis using Centricon-100 (Amicon, Danvers,

P. Santamaria et al.

MA) or Ultrafree-100 (Millipore, Bedford, MA) microconcentrators. Approximately one third of the retentate (15-20/,l) is dried down and resuspended in 15 >l of 1X Taq sequencing buffer (50 mM tris-HCl, pH 9, 10 mM MgCl2). Internal oligonucleotides are used for priming the sequencing of DQB, DRB, and DQA genes, respectively (Table 2). Primers for sequencing each strand are listed in Table 2. Only one strand is routinely sequenced for typing; sequencing of the other strand is performed in cases where a new allelic sequence is suspected. Eighty to 100 ng of primer are end-labeled with 10 pmol of gamma-32P labeled ATP (5000 Ci/mmol, 10/,Ci//,L) and 5 U ofT4 polynucleotide kinase (Promega Biotec) in a 10 /*l final volume. Ten nanograms of primer (1 /zl) are added to the sequencing mixture without extraction of unincorporated labeled ATP, boiled for 5 minutes, and then left at room temperature for 15 minutes. Eight units of recombinant Taq polymerase (Cetus Corporation) are added to the mixture. Four microliters of the annealed primer/template mixture are later added to 4 /21 of each of the following stop nucleotide mixes (ddG mix: 15/,M each dGTP, dATP, dCTP, dTTP; 45/~M ddGTP; ddA mix: 15 ~M each dGTP, dATP, dCTP, dTTP; 600 /,M ddATP; ddT mix: 15 ~M each dGTP, dATP, dCTP, dTTP; 1200/~M ddCTP; ddC mix: 15/,M each dGTP, dATP, dCTP, dTTP; 450 /,M ddCTP). The reactions are allowed to proceed for two consecutive periods of 10 minutes at 72-74°C, each followed by a quick microfuge spin to avoid volume reduction due to condensation. After the second cycle, each reaction is chased with 2 btl of a 7.5 ~M mixture of dATP, dGTP, dTTP, dCTP, and allowed to proceed for 5 minutes. After spinning down, the reactions are stopped by adding 4/,l of 95% (vol/vol) formamide/20 mM EDTA, heated to 80°C for 5 minutes, and loaded on a .4-ram thick 6% polyacrylamide/7 M urea gel. Electrophoresis is performed at 2500 V for 2 hours, the gel fixed in 5% (vol/ vol) glacial acetic acid/5% (v/v) methanol for 15 minutes, dried, and exposed to KodaK X-Omat film (Rochester, NY) for 4 to 12 hours either at room temperature or at -70°C with intensifying screen. Following the conditions described here the sequence ladders between the sequencing primer and the upstream or downstream amplification primer, respectively, can be clearly read starting from 2 to 14 bases from the sequencing primer binding site (Figs. 1 and 2).

Oligonucleotide sequences and cDNA/PCR/sequencing primer combinations for SBT. Figure 1 shows the general SBT strategy as well as the relative positions of each of the oligonucleotides used for the cDNA, PCR, and sequencing reactions on the mature DRB, DQA, and DQB mRNA molecules. The sequences of these primers, the loci they are specific for, the specific posi-

TABLE 2 Oligonucleotides used for the cDNA/PCR/sequencing reactions A. Oligonucleotides Type i

Anneal

Ori

Loci/allele

QB{105-111) RB(105-111) QA~147-1571 QB(I-7) RB(-4/3) RB(-32/-26) QA(- 10/-4) QB(gv-103) RB(97-103) QB(78-83) RB(87-94) QA(88-95) QA{19-24)

5'-GGTGGTTGAGGGCCTCTGTCC-Y 5'-GTGCTGCAGGGGCTGGGTCTT-3' 5'-GGTGAGGTTACTGATCTTGAAG-3' 5'-AGAGACTCTCCCGAGGATTTC-3' 5'-CTGGCTTTGGCTGGGGACACC- 3' 5'-TGTTCTCCAGCATGGTGTGTC- Y 5'-CTGTCCTCCGTGATGAGCCC-3' 5'-ATGGGGAGATGGTCACTGTGG-3' 5'-AGGATACACAGTCACCT17AGG-3 ' 5'-GTAGTTGTGTCTGCACAC-Y 5'-GCCGCTGCACTGTGAAGCTC-Y 5'-CACGGTTCCGGTAGCAGCGGTAG-Y 5'-TACGGTCCCTCTGGCCAG-Y

105-111 105-111 147_ 15 v 17 -4-3 -32/-26 - 10/-4 9 ~- 103 97-103 v8-83 87-94 88-95 19-24

AS AS AS S S S S AS AS AS AS AS S

DQB DRB DQA DQB DRB DRB DQA DQB DRB DQB DRB DQA DQA

1 1/3/4/5 1 1 1/3/4/5 i/3/4/5 1 1 1/3/4/5 1 1/3/4/5 1 1

Type 2 RB(7-13 )

5'-'I~FCTTGCAGCAGGATAAGTA-Y

7_ 13

S

RB(5-11)

5'-CCACGTTTCTTGGAGTACTCT-Y

5-11

S

RB(6-13) #RB(29-35a)

5'-TTFCTTGGAGCAGGTTAA ACA-Y 5'-AGATGCATCTATAACCAAGAG- Y

6-13 29-35

S S

#RB129-35b)

5'-AGATACTTCCATAACCAGGAG-3'

29-35

S

#QB(-8/-2a)

5'-CTGAGCACCCCAGTGGCTGAG-Y

-8/-2

S

#QB(-8/-2b)

5 ' - C T G A G C T C C T C A C T G G C T G A G - 3'

-8/-2

S

#QB(-8/-2c)

5'-CTGAGCACCTCGGTGGCTGAG-Y

-8/-2

S

D R B 1" 15 DRBI*16 DRB 1"03 D R B 1"08 DRBI*I 1 DRBl*12 DRBI*I3 DRBI*14 D R B 1"04 D R B 1/3/4/5 (DRBI*01) D R B 1/3/4/5 (DRBI*I4) {DRBI*03) (DRBI*13) (DRB3*01 ) (DRB3*03) DQB 1 ( D Q B 1"0601 ) ( D Q B 1"0604) DQB I {DQBI*0501) DQB 1 ( D Q B 1"0301 )

Step RT/PCR RT/PCR RT/PCR PCR/SEQ PCR/SEQ PCR PCR SEQ SEQ SEQ SEQ SEQ SEQ

PCR PCR

PCR PCR PCR

PCR

PCR PCR

Each oligonucleotide is designated with two letters (to indicate the locus-specificity) fi~llowed by two numbers in parenthesis {to indicate codons on template the primer anneals to); a "/" between codon numbers is used in primers annealing to signal peptide sequences (positions - 1, -2, etc.). "Ori" is to indicate the orientation of the primer (AS, antisense; S, sense). #, under the conditions described here, means these oligonucleotides will selectively amplify different allelic templates from different allelic combinations, that is, depending on the sequence differences between the primer and each allele's primer-binding site Isee text); the exact specificity of each of these primers is indicated in parenthesis. B. C o m b i n a t i o n s o f primers for c D N A / P C R / s e q u e n c i n g reactions Reaction type

cDNA

PCR

Routine A B C D E F

1 2 2 2 1 1

RB(105-111) RB(105-111) RB(105-111) RB(105-111) QB( 105-111 ) QA(147-157)

RB(-32/-26) RB(7-13) RB(5-11) RB(6-13) QB(1-7) QA(-10/-4)

37°C/1.5 55°C/1.5 55°C/1.5 55°C/1.5 37°C/1.5 37°C/1.5

mM mM mM mM mM mM

RB(97-103)/RB(87-94)/RB(-4/3) RB(97-103)/RB(87-94) RB(97-103)/RB(87-94) RB(97-1031/RB(87-94) QB(97-103)/QB(78-83) QA(88-95)/QA(19-24)

Alternative G H I J K L M

2 2 1 1 2 2 2

RB(105-111) RB(105-111) RB( 105-111 ) QB( 105-111 ) Q B ( 1 0 5 - 1 1 l) QB{105-111) QB(105-111)

RB(29-35a) RB(29-35b) RB(-4/3) QB(-8/-2a) QB(-8/-2a) QB(-8/-2b) QB(-8/-2c)

55°C/1.0 55°C/1.0 37°C/1.5 37°C/1.5 55°C/1.0 55°C/1.0 55°C/1.0

mM mM mM mM mM mM mM

RB(97-103)/RB(87-94) RB(97-103)/RB(87-94) RB(97-103)/RB(87-94) QB(1-7)/QB(78-83) QB(97-103)/QB(78-83) QB(97-103)/QB(78-83) QB/97-103)/QB(78-83)

A.T./MgCI2

Sequencing

The first six combinations are those designed to be used for routine typing and the rest represent alternative combinations that we have tested and that may be useful in specific situations. For instance, reaction I for DRB may be useful in hypothetical situations where homozygosity at DRB loci is not expected according to the rest of the haplotype. (') For sequencing, we include two alternative sequencing primers per locus. A.T., annealing temperature in PCR.

74

P. Santamaria et al.

RTIPCR I

PCR

~d i

I

anti-sense

primer

sense

Type

1 primer

PCR sense

Type

2 primer

Seq p r i m e r Exon

(sense

1 1

or a n t l - s e n s e ) Exon

2

~

3'

mm--4~ RB (-32/- 26)

~(105/lll) RB(7/3), (6/13), (5/11) ~-mi--4~ RB (29/35)

Zxon

Exon

2

ml-4P -QB (¢7) QB (-8/q2)

/

gxon

DRB mRNA

I I

RB (I05/Iii) R~ (10B/ill) .9~J'---']RB (97103) ~-, I RB(87/94)

~

3'

I

~xon 3

[ SEQ

DQB mRNA

QB (I05"/iii) I QB(97/I03) "' IQB(78/~3) Exon

RT/PCR

/3'

I SEQ

DQA mRNA

w

QA(-10/-4) QA(19/24)

FIGURE 1 Schematic of cDNA/PCR/Seq experiments for DRB, DQA1, and DQB1 genes as well as the relative primer binding sites of each of the oligonucleotides used here on DRB, DQB, and DQA transcripts. Each primer pair as well as each type 2 primer is used in a different reaction (RT/PCR) (see also Table 2). The primers that can be used to sequence the products of these reactions are shown below (Seq).

tions (codons) to which they anneal, and the reaction(s) they are used in (reverse transcription, PCR, or sequencing) are indicated in Table 2A. In Table 2B we show the specific combinations of primers that can be used for the c D N A / P C R / s e q u e n c i n g reactions for each locus. RESULTS

Sequence-basedtyping of DR and DQ polymorphic genes in homozygous typing cells. Homozygous LCLs from the panel of the Tenth International Histocompatibility Workshop (Table 1) [16] were used as an initial test of the methodology. In total, these cell lines were representative of most o f the known D R and D Q haplotypes at the time the analysis was conducted.

,QA(88 s

QA(147/-157) I R T / P ~ I sEQ

Total cellular R N A isolated from homozygous LCLs was reverse-transcribed and the resultant cDNAs amplified using conserved oligonucleotides specific for D R B 1 / D R B 3 / D R B 4 / D R B 5 or DQB1 or DQA1 genes, as described in Materials and Methods (reactions A, E, and F of Table 2B). Such conserved oligonucleotides (which we call type 1 primers) anneal to regions of conserved D N A sequences. These conserved primers, as opposed to nonconserved primers (which we call type 2 primers, see below), were designed to amplify all known alleles at DRB, DQA1, and DQB1 loci and, thus, all possible combinations of these alleles in any given heterozygote. Sequencing of these amplified templates was performed using a type 1 nested primer (annealing to a conserved region of the cDNAs, internal to the sequence recognized by the amplification primers): (1) antisense sequencing primers: RB(97103); RB(87-94); QB(97-103); QB(78-83); QA(88-95); and (2) sense sequencing primers: RB(-4/-3); QB(1-7); QA(19-24). N o anomalous amplification products or sequencing ladders were detectable upon direct sequence analysis of amplified DRB, D Q B 1, and D Q A 1 cDNAs from the 43 homozygous cell lines tested (Table 1). The number of ladders generated for each cell line was always that expected according to the specificity of the

DRBI*0301

DRBI* 1501

~'DNA'P'~ (~05- i 1 ~ )CDNk=R~ [Z05- i II )-~DNA=RB (105_ 111 ) PCR=R~ (5-11) PCR=P.B (7 - 13) PCR=RB (-32/-26)

o oo

DOB1*0501 cDllUk,.QS[105-111) PCI;~QB(-S/-2)a Seq.,QB(97-103) DQ).I * 0 2 0 1 DQAI*0301

DORl*0301

DQBI*0302 DQBI*0201

DQBI*0302

oDImm~/k{14 T- 15T) ~N),aQ~ ( 1i 7- 15T) oJ~mmQB(105-111 )¢DNRmQB( 105" 111 ) l'Cm.OA (-101-*)

&eq-gA(co- .5)

T T C G A A G

G A T A G A T

T T C G T T C

p(:11-(~l~ (-1 O/- ¢)

~(18*)s)

1'Cam~;~l (1- 7 )

P C ~ l l (1-7)

seq-Qm (70-e3)

seq-Qm( ?e- s])

T T T G T C G

Codon 42

T

A

A

T

C o d o n 50

A C

T

C

T

codon

A A C A

(;4

C o d o n 42 C~on62

C T C A

.--4b,

Codon

)

55/56

G A G G C G C C G C G G

G C G G G C C C C G C G G C G C

Codon 76 A C

C

A

C

C

T

C

C o d o n 64

Codon

83

COdl Codon Codon

93

76 Codon

LS

S$

LS

71

SS

F I G U R E 2 Sequence-based typing for DRB, DQB, and D Q A loci. Lanes are read from left to right as G-A-T-C. The left panel shows an example of DRB typing using the strategy shown in Fig. 4, using the primer combinations of reactions A, B, and C of Table 2B, also shown at the top of each ladder. Each of the first two ladders corresponds to a different DRB1 allele, selectively amplified from a complex heterozygote (four sequences, third ladder of left panel) with type 2 primers. The positions where the two different DRB 1 ladders differ are shown on the left side of this panel. For the other ladders, positions where there is more than one band are indicated on the side of the ladder and the templates they correspond to are indicated at the top of the figure. The middle panel shows a D Q typing example (DQA and DQB) for two related subjects (a homozygote and a heterozygote). For the D Q A heterozygote, the arrow points at a position where codon 56 of D Q A 1"0201 corresponds to codon 57 of D Q A 1"0301, due to a deletion; because of the deletion there are two bands at almost all the positions of the ladder. The right panel corresponds to a D Q B 1"0501 ladder from a DQB 1"0501/DQB I*0301 heterozygote selectively amplified with a type 2 primer (QB(-8/-2)b). Numbers corresponding to nonpolymorphic codons are arbitrarily shown on the side of each panel to help read the ladders. ** are to indicate nonspecific termination (DQA ladder, middle panel) or a compression (third panel). Faint bands present at every position of the ladder represent nonspecific termination and should be ignored. To read unambiguously the last 5 0 - 6 0 base pairs of the ladder it is necessary to run the sequencing gel for an additional hour, approximately.

76

amplification primers (one DQB1 and one DQA1 ladder, one DRB ladder for DR1 and DRw8 cell lines and two DRB ladders for cells carrying haplotypes of the DRw52 and DRw53 supertypic groups). Thus, analysis of the homozygous typing cell lines showed that the type 1 primers used for cDNA synthesis, PCR, and sequencing reactions allowed for accurate amplification and sequencing of all the tested alleles at each of these loci.

P. Santamaria et al.

DQB1 Codon 67 G~r

"050~

GOC

'0604

AGC

GAT

~

"0S02

`0602

~

`05o2

Co
`0601

"0e03

"COOt

"0303

'0401

'0402

"0,~03

"0601

'o6o3

`0301

`0303

"0,I0~

`0402

`0603

"0601

`0301

`0303

`0401

"0402

`0503

"0GO1

.0.~I

"c,~3

"0401

"o402

'osoo

'osol

`0303

'0401

"0402

'05O3

"O601

CAC

TAC

eGG

`06o4

CC~

`06o2

Codon 49 OCG

OCA '0501

`0504

I '020t

`0302

~'0~2

"0602

"0603

Codon 47 Trc

TAC

T^I

Amplification and direct sequencing of DQA1 and DQB1 cDNAs in subjects of unknown HLA type. DNA sequences have been determined for HLA class II allelic specificities defined by conventional HLA typing techniques [19]. Comparisons of these sequences indicates that any given DQA1 or DQB1 allelic combination is characterized by a unique sequencing ladder. Total RNA from PBMCs from subjects carrying 38 different heterozygote combinations of haplotypes (Table 1) was tested to evaluate if the allelic composition of DQA1 and DQB1 heterozygotes could be determined correctly by direct amplification and sequencing using type 1 primers (reactions E and F of Table 2B). These subjects had been previously serologically typed but the typing information was not known to the investigator who determined the class II allelic combinations from the sequencing results. All individuals were assigned DQA1 and DQB 1 allelic sequences that were in agreement with the serological phenotypes. In all the heterozygotes tested both allelic sequences could be read clearly from the composite sequence pattern; a unique pattern is found for every particular heterozygote combination in the same way that certain RFLP banding patterns correspond to certain heterozygote allelic combinations. For instance, in a DQBI*0201/DQBI*0501 heterozygote one would find the sequence GGGG(A/ T)T(T/A)CCGGGC(A/G) at codons 45 to 49, which can only be attributed to that particular allele combination. In Fig. 2 we show the DQA1 and DQB1 ladders corresponding to a DQBI*0201, DQAI*0201/ DQBI*0302, DQAI*0301 heterozygote and to a DQBl*0302, DQAI*0301 homozygote, respectively. A systematic approach for the interpretation of DQA1 and DQB1 ladders is shown in Fig. 3; interpretation of the DQA 1 and DQB 1 heterozygous ladders of Fig. 2 is shown as an example. Once both alleles are identified in a given heterozygote, their respective sequences are compared with the sequences of all known alleles at the locus under analysis; the absence of expected bands or the presence of unexpected bands for a particular allele or allelic combination is therefore suggestive of sequence heterogeneity, that is, new alleles. For instance, substitution of the A at the second base of codon 46 (where DQBI*0201 is the only allele that has an A)

T.~

~

TAT

"0602

'oe02

"oeoa

'0502

'0sO2

`06O3

TRC

CIE:;

CIOG

~

'0201

`0302

/~Gc

TAC

GaG '030t

T~C

GGG

Codon 30 CAC

C/~C

TAC

Codon 23 `0501

'0604

`020t

'0~02

`0502

`0802

"0603

"0301

'0303

"0401

'0402

`0503

"0~>01

DQA1 Codon 68 & 60 GTGCTA

GTG GCA "0104

"0101

*0104

"0101

GTG/K::^

GTCCT^

"0102

"0103

"0401

"0601

"0501

°0102

"0103

"0401

"0601

"0102

'0103

"0401

"0601

"0501

"0102

°0103

"0601

"0501

TTC

TAC

Trc

TAC

"0102

"0103

"0401

"0601

"0501

Codon 52 c^c

cc, c

^c,c

AG^

B050~

Codon 41 AGG

Codon 3 4 CAG

GAG "0201

"0301

"0104

"0101

"0104

"0101

"0401

Codon 25 13"C "0201

"0301

TAC

FIGURE 3 Readingcharts for DQA and DQB sequencebased typing. Interpretation of any possible heterozygote can be done by reading the polymorphic codons shown in the figure. Interpretation of the DQB and DQA heterozygous ladders of Fig. 2 is shown as an example.

would strongly suggest the presence of a sequence variant of DQB 1"0201. Once detected, the sequence of the variant can be confirmed after selective amplification of the variant or by subcloning the amplified products.

Amplification and direct sequencing of DRB cDNAs from subjects of unknown HLA type. As described above, the use of type 1 primers allows the unambiguous sequencing of all heterozygote combinations of DQA1 and DQB1 alleles. Because of the isotypic complexity of DRB genes (expression of more than one DRB locus by certain haplotypes), amplification and sequencing of cDNAs from DRB heterozygotes with type 1 primers can generate up to four overlapping ladders, thus resulting in complex sequencing patterns.

Sequence-based Class II Typing

DRB cDNAs from the same individuals tested above for DQA1 and DQB1 genes were amplified and sequenced using DRB-specific type 1 primers (reaction A of Table 2B). These individuals comprised 38 different heterozygote combinations, including several examples from each of the groups of complex DRB allelic combinations that would generate up to four sequencing ladders. The DRB sequence ladders generated with type 1 primers were analyzed as described above for DQA1 and DQB 1 loci: highly polymorphic positions were analyzed first for the presence of bands unique to specific alleles or groups of alleles (i.e., DR4) and the sequences deduced and compared with the sequences of all known alleles at all loci. For example, in Fig. 2 we show the ladder generated by sequencing a complex DRB heterozygote (four overlapping ladders); the positions with two or more bands are indicated on the side of Fig. 2 and are assigned to each of the allelic types composing the complex sequencing pattern. For all but one sample the information deduced from these sequencing experiments matched the independently determined serological phenotypes of the subjects under study as well as the DQA1 and DQB1 allelic types assigned to these individuals by direct sequencing of these genes as described above. The inconsistent sample had been serologically typed as DR6/DR4 but was typed by sequence analysis as DRB1*1301/DRBl*0801. The presence of a DRBI*0801 allele instead of a DRBI*04 allele was confirmed in a repeated experiment; we thus believe that the original serological typing was in error. In all cases the DQB1, DQA1, and DRB1 templates had been equally amplified and sequenced with a similar efficiency by the use of type 1 primers. DRB3, DRB4, and DRB5 sequence ladders could be read in all but one case (a DRB3*0101 [DRw52a] sequence was not initially observed in a DRw13/DRw17 heterozygote). Since DRB3*0101 is in linkage disequilibrium with DRBI*0301, the former allele was expected to be found in the overlapping ladder as well. To rule out the possibility of an error, the investigator assigning the HLA types from the sequencing ladders repeated the typing of this individual; the DRB 3"0101 could be read in the repeated experiment. Although the results generated by the use of type 1 primers were compatible with the serological phenotypes, the exclusive use of type 1 primers will not allow the assignment of each of the specific ladders to each of the expressed loci in some of all the possible heterozygore combinations. Given below are the most complex situations that cannot be addressed by the exclusive use of type 1 primers: (1) distinction among the different DR4 allelic sequences in certain heterozygotes since they differ by only a few nucleotide base pairs and such differences could be masked by the presence of addi-

77

tional ladders; (2) distinction between DRB 1" 1601 and DRB1*1502 since their sequence differences will be masked by those of their linked DRB5 alleles; (3) distinction between DRBI*0301 and DRBI*0302 in specific heterozygote combinations; and finally (4) distinction between alleles that differ exclusively at codon 86 since this difference can also be masked by other ladders (DRB1*1301 and DRBI*1302 or DRBI*ll01 and DRBl*1104, or DRBI*0407 and DRBI*0403, or DRBI*0404 and DRBI*0408, or DRB3*0201 and DRB5*0202). We have thus developed a more informative strategy to deal with DRB; this strategy, which consists of the additional use of nonconserved (type 2) primers (see below), permits the clear elucidation of even the most complex combination of the four DRB sequences that might be present in an individual. These nonconserved primers, as opposed to allele-specific primers, are designed to be used in reactions performed simultaneously with the reactions using type 1 primers and aim at selectively amplifying certain ladders from the complex sequencing patterns without requiring previous typing information. Analysis of the sequence variability of the second exon of the DRB genes has allowed us to identify two regions that could be used to design nonconserved (type 2) primers: (1) condons 5-13; and (2) condons 29-35. The sequence of the former region follows a groupspecific sequence pattern, that is, a sequence shared by groups of alleles at individual loci. The latter region exhibits a scattered nucleotide polymorphism in DRB 1 and DRB3, DRB4, and DRB5 genes. We designed five different nonconserved primers annealing to these two polymorphic regions: (1) RB(7-13); (2) RB(5-11); (3) RB(6-13); (4) RB(29-35a); and (5) RB(29-35b), the latter two primers annealing to the second region of moderate polymorphism (Tables 2 and 3). Because of the different nature and distribution of mismatches between these primers and the different DRB templates, the type of templates amplified selectively by these primers will be different. Each of the first three primers will amplify up to two DRB 1 cDNAs in any given heterozygote and will not amplify any DRB3, DRB4, or DRB5 cDNAs. On the contrary, the use of primers RB(29-35a) and RB(29-35 b) will allow the selective amplification of different transcripts from DRB1, DRB3, DRB4, and/or DRB5 loci in different heterozygote combinations. Empirical studies were done to establish the specificity of each primer pair and to determine which combination of primers would give the best discriminatory results for DRB typing, which are described in Table 2. Type 2 primers were able to selectively amplify certain DRB templates in all the heterozygote combina-

TABLE

3

Contribution of nucleotide base pair mismatches between 5' amplification primers and DRB alleles to the selective amplification of allelic and/or nonallelic DRB transcripts

A. Mismatches between type 2 D R B - p r i m e r s and D R B alleles at differentloci DRB1

RB(29-35a) RB(29-35b) RB(7-13) RB(5-11) RB(6-13)

"0101-3

"1501

"1601-2 "1502

"1401-2 "0301/1301-2

"0401-8

"1303 "1101-4

"0801-3

"1201

"0701

0 4 5 6 6

2 4 0 5 4

1 5 0 5 4

4 0 7 0 8

3 3 4 4 0

2 2 4 0 8

3 3 7 0 8

5 1 8 0 8

4 3 8 12 5

DRB 1

RB(29-35a) RB(29-35b) RB(7-13) RB(5-11) RB(6-13)

DRB3/DRB4/DRB5

"0901

~1001

DRB5 a

DRB3*0101

DRB3*0201

DRB3*0301

DRB4

1 5 2 5 5

3 4 4 4 4

3 1 6 6 8

4 0 5 5 6

5 1 5 5 5

4 0 5 5 5

2 2 3 3 4

a DRB 1 gene from cell line AMAI has an additional nucleotide substitution in the first base of codon 30, in comparison with DRB 1 genes of other DR2 haplotypes

[14]. B. Selective amplification of D R B and D Q B 1 c D N A in combinations of alleles mismatched with type 2 oligonucleotides (#) Haplotypes

DRBI*I301,DRB3*0101 IDRBI*I601,DRB5*0201 DRBI*I301,DRB3*0101 IDRBI*0801 DRBI*0301,DRB3*0101 fDRBI*I601,DRB5*0201 DRBI*I303,DRB3*0101 IDRBI*0801 DRBI*I303,DRB3*0101rDRBI*0801 DRBI*I303,DRB3*01011DRBI*0301 ,DRB3*0101 DRBI*II01,DRB3*02011DRBI*I501 ,DRB5*0101 DRBI*II01,DRB3*02011DRBI*I501 ,DRBS*0101 DRBI*I201,DRB3*0201 IDRBI*II01 .DRB3*0201 DRBI*I201,DRB3*02011DRBI*II01 .DRB3*0201 DRBI*0405,DRB4*0101 'DRBI* 0301 .DRB3*0101 DRBI*I303,DRB3*0101 'DRBI*l101 ,DRB3* 0201 DRB!*I501,DRB5*0101 DRBI*I601,DRBS*0201/DRBI*0401 ,DRB4*0101 DRBI*I601,DRB5*0201/DRBI*0401 ,DRB4*0101 DRBI*I601,DRB5*0201 DRBI*I601,DRB5*0201 DRBI*I602,DRB5*0202 DRBI*0401,DRB4*0101 DQBI*0604/DQBI*0502 DQBI*0301/DQBI*0501 DQBI*0301/DQBI*0501 DQBI*0201/DQBI*0603 DQBI*0604/DQBI*0301 DQBI*0301/DQBI*0502 DQBI*0603/DQBI*0501 DQBI*0603/DQBI*0501 DQBI*0201/DQBI*0501 DQBI*0201/DQBI*0302 DQBI*0201/DQBI*0502

Selected Alleles

DRBI*I601 DRBI*0801 DRBI*I601 DRBI*0801 DRB3* 0101 DRBI*I303 DRB3* 0201 DRBS*0101 DRBI*I101 DRBI*I201 + DRB3*0201 DRBI*0405 + DRB4*0101 DRBI*II01 + DRBI*I303 DRBI*I501 DRBI*0401 + DRBS*0201 DRBI*I601 + DRBI*0401 DRBI*I601 DRB5* 0201 DRBI*I602 DRB4* 0101 DQBI*0604 DQBI*0301 DQBI*0501 DQBI*0201 DQBI*0604 DQBI*0301 DQBI*0603 DQBI*0501 DQBI* 0501 DQBI*0201/DQBI*0302 DQBI* 0201

Type 2 primer

RB (25-35a) RB (25-35a) RB (25-35a) RB (25-35a) RB (25-35b) RB (25-35a) RB (25-35a) RB (25-35b) # RB (25-35a) RB(25-35b) RB(25-35a) RB(25-35a) RB (25-35a) RB(25-35b) RB(25-35a)## RB (25-35a) RB (25-35b) RB (25-35a) RB (25-35a) QB (-8/-2a) QB (-8/-2a) QB (-8/-2b) QB (-8/-2a) QB (-8/-2c) QB-(-8/-2a) QB (-8/-2a) QB (-8/-2b) QB (- 8 / -2b) QB (-8/-2a) ### QB (-8/-2a)

# Here we only show representative examples of haplotypic combinations lacking those alleles the primers are fully matched with (see Table 3A). # DRB5*0101 and DRB3*0201 templates both have one mismatch with primer RB(29-35b). Selection of DRB5*0101 could be related to the differential positioning of the mismatch with respect to the primer. # # A weaker DRB4*0101 template was also observed. # # # Despite the presence of a mismatch between these two DQB 1 allels, primer QB(-8-2a) was not able to select either of them.

Sequence-based Class II Typing

tions tested. In heterozygotes carrying the alleles with which these primers are matched, the alleles were selectively amplified; in heterozygotes not carrying the alleles specifically recognized by the primers, the DRB templates that had the fewest base pair mismatches with the primers were selectively amplified in the PCR. Specific examples of the latter are shown in Table 3. In Fig. 2 we show examples of selective amplification of DRB and DQB transcripts using type 2 primers. As shown in Table 3, type 2 primers could differentially amplify DRB or DQB transcripts from combinations of allelic cDNAs that differ from each other in as few as one nucleotide substitution, provided that high stringency annealing conditions are used for the PCR (annealing at 55°C and 1 mM MgC12 concentrations). Under these conditions it was also observed that for heterozygote combinations carrying alleles that have the same number of base mismatches with a primer, the differential positioning of the mismatches within the sequence recognized by the oligonucleotide also has an influence on the stability of the primer/cDNA complex and hence on the outcome of the PCR as detected by direct sequencing (Table 3B). For instance, although DRB3*0101 or DRB3*0201 and DRB5*0101 genes all have one mismatch with primer RB(29-35b), this oligonucleotide selected the DRB5*0101 sequence in a DRw11/DR2 heterozygote. Similarly, primer QB(-8/-2a) selected the DQBI*0604 sequence in a DQB I*0604/DQB 1"0502 heterozygote (Table 3). Five nucleotides separate the two mismatches between the DQB 1"0604 allele and the QB(-8/-2a) primer, whereas only two nucleotides separate the mismatches between the DQBI*0502 and the primer. Our results indicate, however, that during amplification under conditions of high stringency the number of mismatches between a given type 2 primer and the selected allele(s) appears to be more important than the location of the mismatches or the bases involved in mispairing. Based on the results described above, we designed a strategy for DRB typing in subjects of unknown HLA types. This SBT strategy for DRB loci is shown in Fig. 4 and involves the simultaneous use of type 1 and type 2 primers in a total of four reactions; the primer combinations for each of these reactions are shown in Table 2. In Fig. 2 we show a DRB typing example using the four reactions described in Table 2 and in Fig. 4. The simultaneous use of type 1 and type 2 primers for DRB will thus permit the clear elucidation of even the most complex of all DRB heterozygote combinations with only one exception: distinction between DRB3*0201 and DRB3*0202. The two base pair differences at codon 86 between these two alleles cannot always be resolved by the use of the primers described in this paper; however, this specific limitation could be eliminated by the use of

79

RB(105-111)

RB(1OS*~ 11)

RB(-32/-26)

.... ....

1

2

3

~0~O:V~0uS hom0,V~0~s ~eE~,0Zy~0S

RB{1QS-1111

RB(7-13)

RB(5-11)

/\

/\

R S ( t Q 5 - 1 1 X) RB(6-13)

/\

4 neler0zvqous

dn

Find ladder R8(.32/.26) L ~ k for I3F~85 laOOer

in

Find ladde~ RBf-327-26) LOOkfor ORB3 ladder

in

Fred radder R6( 32/ 26j + LOOk lot [2AB4 ladder

if OR2 or ~Rw52 or Oit4 hlplOlypes, chack for the pr¢llnce of a possiblo addilio~al ORS1 laeder in 88(-32/-26): DRmI"O10x, OR81"fO01, ORBf'0701. DRBI"0gO~

FIGURE 4 DRB-sequence-based typing strategy.

a type 2 primer tailored to selectively amplify DRB3 cDNAs. DISCUSSION We have presented a rapid and accurate sequencingbased I--ILA class II typing methodology that aims at providing the most complete and detailed information currently possible in this regard. This approach combines enzymatic amplification of class II c D N A molecules using conserved (type 1) and nonconserved (type 2) oligonucleotides and direct sequencing of the resulting products. This methodology is being extended to other class II genes (DPA1 and DPB1) and to class I genes, should be applicable to class III typing, and is automatable using available systems (i.e., automated R N A / D N A extractors, robotic workstations, and automated sequencers). Complete genotyping of the entire polymorphic HLA region could thus be accomplished in a single run by using a single technique and the typings could be performed in an automated fashion at what we estimate to be a low cost. The use of PBMC-total cellular R N A instead of DNA as initial template allowed us to optimize several steps and to develop a technique that can be used routinely with minimal sample handling; total cellular R N A can be prepared from a low number of cells in less than 2 hours [18]. In addition, the use of R N A as template allowed us to place the c D N A / P C R primers outside the region of interest (exon 2) and, hence, to read longer sequences of the second exon of the polymorphic genes. Furthermore, sequencing ladders are simplified by the use of RNA since pseudogenes or nonexpressed genes (i.e., DRB2, DXB, DXA) are not amplified. Related genes expressed at very low levels (i.e., DOB,

80

DZA) are also not amplified efficiently enough to interfere in the final interpretation of results. Since DRB3/ 4/5 genes are expressed at lower levels than DRB1 genes [5], this results in fainter sequencing ladders from these loci in the reactions using type 1 primers (Fig. 2). This finding, however, turned out to be useful, rather than problematic, in discerning and identifying DRB allelic sequences from overlapping ladders generated with type 1 primers in complex heterozygotes. Since the cDNA/PCR/sequencing reactions using total cellular R N A were shown to be highly specific, there was no advantage in using p o l y - A + R N A , either in specificity or yield. Since the PCR will generate enough material for sequencing from even a few copies of m R N A , the lower number of circulating mononuclear cells in leukemic patients, often regarded as a problem in serological typing, should not represent a difficulty in that respect, although this remains to be tested. We have found that double-stranded D N A templates generated by PCR are difficult to routinely sequence directly with T7 or other D N A polymerases. Alternative strategies, such as the generation of single-stranded D N A templates by asymmetric-PCR, have been proposed for that purpose [13]. While asymmetric-PCR helps the sequence analysis in some cases, the additional amplification step [13, 20] unnecessarily complicates routine analysis. Furthermore, although these singlestranded templates can be sequenced with T7 D N A polymerase, the resulting sequence ladders are not devoid of sequencing artifacts to allow its use in routine typing, at least in our hands. The use o f T a q polymerase for sequencing allows to increase the temperature of extension to 72°C, which relaxes the template's secondary structure and thus allows to obtain cleaner ladders. The use of Taq polymerase for sequencing required, however, optimization of the chemistry of the reactions. For instance, the dideoxynucleotide/deoxynucleotide concentrations and ratios normally used for T7 D N A polymerase sequencing are not optimal for Taq polymerase; the concentrations and ratios described here were those that generated the best results. Using this protocol, Taq D N A polymerase consistently yields better results than T7 D N A polymerase and improves upon previously described methods for sequencing PCR templates [ 2 0 - 2 2 ] . It must be emphasized, however, that the success in sequencing double-strandedPCR products using the protocol presented here is dependent on the sequencing primer used and, for a given sequencing primer, on the amplification primers used to generate a particular sequencing template. The primer combinations and reaction conditions described here generate clean and reproducible sequencing ladders; Fig. 3 shows representative examples. Although most of the second exon sequences corre-

P. Santamaria et al.

sponding to the class II alleles defined by more conventional typing methods have been determined, it is generally assumed that these sequences are identical among haplotypes that type the same by other methods. If this were always true, one could argue that obtaining the entire sequence for each allele would be unnecessary. However, our sequence analysis of homozygous cell lines representative of well-defined class II haplotypes has revealed the existence of sequence heterogeneity that can be differentially recognized by alloreactive Tcell clones [14]. Sequence analysis of individuals of known HLA types with this methodology has also proven the existence of sequence heterogeneity and haplotypic diversity at the population level [15]. It is therefore likely that a more extensive search for this sequence allelism will identify new alleles, especially in less well-studied non-Caucasian populations. Many such new alleles could be missed by the other typing techniques, especially since some of these methods, such as oligotyping, only "see" a small portion of the molecule and thus do not ensure complete sequence identity between the allele the primer is designed to recognize and the allele being recognized by the primer in a given individual. Direct sequencing of class II genes as presented here may thus represent a useful alternative to other molecular typing methods. This methodology could be of special interest in typing unrelated bone marrow transplantation donor/recipient pairs where single amino acid differences may trigger rejection or graft versus host disease responses [23]. In summary, for typing purposes, four simultaneous c D N A - P C R reactions are performed in separate tubes for DRB, one reaction for D Q B 1, and one reaction for D Q A 1 gene products. For DRB the first reaction uses a type 1 primer and aims at amplifying all the different DRB transcripts present in any given heterozygote (A in Table 2B). Three additional reactions, using type 2 primers, are run simultaneously with the former reaction to amplify selectively DRB1 over D R B 3 / 4 / 5 transcripts (B, C, and D in Table 2B). Using this protocol, class II gene sequence information can be currently obtained in a single run in 1 6 24 hours. ACKNOWLEDGMENTS We especially thank Dr. F. H. Bach for continuous support and many helpful suggestions in writing this manuscript. We also acknowledge Drs. J. P. Houchins, H. T. Orr, and M. Segall for helpful discussions, critical revision of the manuscript, and for providing the workshop cell lines (M. Segall). M. K. Bryan and C. Lewis are acknowledged for their excellent technical assistance. PS was supported by a postdoctoral research grant award by the Ministry of Education and Science of Spain and by a fellowship from the Juvenile Diabetes Foundation International. This work was supported in part by

Sequence-based Class II Typing

grants from the National Institutes of Health (SSR), American Diabetes Association-Minnesota Affiliate (SSR), the Minnesota Medical Foundation (SSR), the Juvenile Diabetes Foundation (JJB), and General Clinical Research Center RR00400 (NIH JJB). This is publication number 181 from the University of Minnesota Diabetes Genetics Research Group.

REFERENCES 1. Bach FH: Class II genes and products: the HLA-D region. Immunol Today 6:89, 1985. 2. Erb P, Feldman M: The role of macrophages in the generation of T-helper cells. II. The genetic control of the macrophage T cell interaction for helper cell interaction with soluble antigens. J Exp Med 142:460, 1975. 3. Bach FH, van Rood JJ: The major histocompatibility complex-genetics and biology. N Engl J Med 295:806, 1976. 4. Bidwell J: DNA RFLP analysis and genotyping of HLADR and DQ antigens. Immunol Today 9:18, 1988. 5. Tiercy JM, Gorski J, Jeannet M, Mach B: Identification and distribution of three serologically undetected alleles of HLA-DR by oligonucleotide DNA typing analysis. Proc Natl Acad Sci USA 85:198, 1988. 6. Erlich HA, Bugawan TL: HLA class II gene polymorphism: DNA typing, evolution, and relationship to disease susceptibility. In Erlich HA (ed.): PCR Techniques. New York, Stockton Press, 1989. 7. Marsh SGE, Bodmer JG: HLA-DRB nucleotide sequences, 1990. Immunogenetics 31 : 141, 1990. 8. Todd JA, Bell JI, McDevitt HO: HLA-DQB gene contributes to susceptibility and resistance to insulin-dependent diabetes mellitus. Nature 329:599, 1987. 9. Saiki RK, Scharff S, Faloona F, Mullis KB, Horn GT, Erlich HA, Arnheim N: Enzymatic amplification of betaglobin genomic sequences and restriction site analysis for diagnosis of sickle cell anemia. Science 230:1350, 1985. 10. Mullis KB, Faloona FA: Specific synthesis of DNA in vitro via a polymerse-catalyzed chain reaction. Methods Enzymol 155:335, 1987. 11. Saiki RK, Bugawan TL, Horn GT, Mullis KB, Erlich HA: Analysis ofenzymatically amplified beta-globin and HLADQalpha DNA with allele-specific oligonucleotide probes. Nature 324:163, 1986. 12. Scharf SJ, Horn GT, Erlich HA: Direct cloning and se-

8l

quence analysis of enzymatically amplified genomic sequences. Science 233:1076, 1986. 13. Gyllenstein UB, Erlich HA: Generation of singlestranded DNA by the polymerase chain reaction and its application to direct sequencing of the HLA-DQA locus. Proc Natl Acad Sci USA 85:7652, 1988. 14. Santamaria P, Boyce-Jacino MT, Lindstrom AL, Rich SS, Faras AJ, Barbosa JJ: Alloreactive T cells can distinguish between the same human class II MHC products on different B cell lines. J Immunol 146:1822, 1991. 15. Santamaria P, Boyce-Jacino MT, Lindstrom AL, Barbosa JJ, Faras AJ, Rich SS: Detection of novel sequence heterogeneity and haplotypic diversity of HLA class II genes. Immunogenetics 33:374, 1991. 16. Yang SY, Milford E, Hammerling V, Dupont B: Description of the reference panel of B-lymphoblastoid cell lines for factors of the HLA system. In: Dupont B (ed): Immunobiology of HLA, Vol. I: Histocompatibility Testing 1987. New York: Springer-Verlag, 1989. 17. Chirgwin JJ, Przbyla AE, MacDonald LJ, Rutter WJ: Isolation of biologically active ribonucleic acid from sources enriched in ribonuclease. Biochemistry 18:5249, 1979. 18. Gouuh NW: Rapid and quantitative preparation of cytoplasmic RNA from small numbers of cells. Anal Biochem 173:93, 1988. 19. WHO Nomenclature Committee. Nomenclature for factors of the HLA system, 1990. Tissue Antigens 137:97, 1991. 20. Stoflet ES, Koeberl DD, Sarkar G, Sommer S: Genomic amplification with transcript sequencing. Science 239:491, 1988. 2 i. Wong C, Dowling CE, Saiki RK, Higuchi RG, Erlich HA, Kazazian H: Characterization of beta-thalassaemia mutations using direct genomic sequencing of amplified single copy DNA. Nature 330:384, 1987. 22. Wrischnik LA, Higuchi RH, Stoneking M, Erlich HA, Arnheim N, Wilson AC: Length mutations in human mitochondrial DNA: direct sequencing of enzymatically amplified DNA. Nucleic Acid Res 15:529, 1987. 23. Fleischhauer K, Kernan NA, O'Reilly RJ, Dupont B, Yang SY: Bone-marrow-allograft rejection by T lymphocytes recognizing a single amino acid difference in HLAB44. Engl Med J 323:1818, 1991. 24. Noreen H, Santamaria P, Davidson M, Rich SS, Segall M: Serological, RFLP and sequence analysis of a new HLADR antigen: DR5 x 6. Hum Immunol 30:168, 1991.