Journal of Neuroscience Research 2656-73 (1990)

Multiple Sclerosis and HLA Class I1 Susceptibility and Resistance Genes D.G. Haegert, M. Michaud, C. Schwab, and G.S. Francis Department of Pathology. Montreal General Hospital (D.G.H., M.M., C.S.), and Department of Neurology. Montreal Neurological Hospital (G.S.F.). Montreal. Quebec, Canada

Probes to the HLA class I1 genes DRP, DQP, and DQa et al., 1984). Population studies have shown that MS is were used to study DNA from unrelated Caucasian associated with various HLA alleles, but particularly multiple sclerosis (MS) patients by sequential restric- with the HLA class TI alleles DR2(15) and tion fragment length polymorphism (RFLP) analysis DQwl(DQw6) (Tiwari and Terasaki, 1985; Francis et in Taql restriction enzyme digests. Comparison of 104 al., 1986) in most but not all Caucasian patient groups patients and 108 controls, who were not matched for (see, for example, LMarrosu et al., 1988); HLA-DRw15 DR type, has identified for the first time a linked series or DR2( 15) is a subdivision of DR2, previously termed of allele-specific RFLPs or allogenotypes which form DR2 long ("Nomenclature for Factors of the HLA Sysan extended haplotype that is preferentially associated tem. 1987"; 1988). These alleles have usually been with MS. These allogenotypes include DRwlS or identified by serology, mixed lymphocyte culture. and DR2(15); DQPlb, which corresponds at the DNA level cellular typing. but more recent studies have used gene to the DQwl(DQw6) serotype; a DQAl allogenotype probes to DR and DQ (Jacobson et al., 1986; Cohen et termed DQalb; and a 2.2 kb DX (DQA2) allogenotype al., 1984; Marcadet et al., 1985; Olerup et al., 1987; termed DXaU (DQA2U). Haegert et al., 1989); these probcs detect restriction fragThe role of HLA class I1 genes in susceptibility to tnent length polyrnorphisms (RFLPs) that are allele-speMS was found to be complex. First, 23 of 104 MS cific and are termed allogenotypes (Bidwell, 1988; Copatients showed DR-DQ linkages which were not ob- hen et al., 1984: Tilanus et al., 1986; Trowsdale et al., served in our control population. We suggest these 1985). Allogenotypic analysis has the potential advananomalous associations may be important in the tage of identifying genes, e.g., DXu(DQA2) and pathogenesis of M S . Second, homozygosity of a 2.0 DXP(DQB2) for which corresponding cell surface antikb DX (DQA2) gene, termed DXaJJ (DQA2L), gens have not been identified (reviewed by Bidwell, showed a strong negative association with MS. DXaL 1988; Bidwcll et al., 1988a). Moreover DR and/or DQ (DQA2L) is in strong linkage disequilibrium with genetic analysis may identify polymorphisms that are DR1, S(wl1) 7, and a subset of DR4, all of which also more closely linked to disease than HLA phenotypes showed a negative association with M S . Since DXaL defined by serology or mixed lymphocyte culture (Awad (DQA2L) does not code for any known product, DR1, et al., 1988). S(11), 4, and 7 become candidates for disease resisIn a preliminary study we investigated HLA class I1 tance genes. Third, in EcoRl and EcoRV digests of allogenotypes in 34 unrelated Caucasian MS patients and DNA from both controls and patients homozygous for concluded that the only class I1 allogenotype associated DQPlb a number of different RFLP patterns were with MS is DQP l b (Haegert et al., 1989), which correidentified and these RFLPs were associated with ei- sponds at the molecular level to the DQwI(DQw6) sether relapsing-remitting or progressive MS. This sug- rotype. We have now extended this earlier study and gests there may be HLA sequence differences between performed sequential HLA-DRP, -DQP and -DQa individuals bearing a particular class I1 allele and RFLP analysis on a total of 104 unrelated Caucasian MS these may correlate with the clinical course of MS. patients. Our more extensive data indicate for the first time that it is not DQPlb alone but a series of closely Key words: MS, HLA class I1 RFLPs, major histocompatibility complex INTRODUCTION Multiple sclerosis (MS) is a demyelinating disease with a genetically determined susceptibility (McFarland 0 1990 Wiley-Liss, Inc.

Received July 5 , 1989: rcviscd Novemhcr 27, 1989; accepted November 28, 1989. Addrcss reprint rcquests to Dr. D.G. Haegert. Department of Pathology, Montreal General Hospital, 16SO Cedar Avcnue, Mnntreal. Qc., Canada H3G I A4.

HLA Susceptibility Genes in MS

linked HLA class IT allogenotypes that form an extended class I1 haplotype which is present in more than 50% of MS patients; these allogenotypes include DR2( 15), DQPl b, DQaI b: and an allogenotypc that cross-hybridizes with DQa and is termed DXa (DQA2) upper (DXaU or DQA2U). Further, our results indicate that the contribution of the HLA class I1 gene region to MS is complex. Approximately 20% of all MS patients showed anomalous DR-DQ linkages, which suggests that DRDQ gene interaction may be important in the pathogenesis of MS. Some HLA class 11 allogenotypes showed a negative correlation with MS in DR-unmatched patients and controls; these gcnes may be protective against MS, Analysis of DNA from hornozygous DQPlb+ MS patients with the restriction enzymes EcoRl and EcoRV have identified RFLPs which apparently correlate with either a progressive or a relapsing course of disease.

MATERIALS AND METHODS Patients and Controls One hundred and four unrelated Caucasian MS patients followed at the Montreal Neurological Institute MS clinic were selected at random and thesc paticnts formed the DR-unmatched patient group. RFLP analysis has been performed previously on 34 of these patients (Haegert et al., 1989). All patients had definite MS according to the criteria of Poser et al. (1983). Eighty-five of these patients were classifiable into one of two clinical subgroups: rclapsing-remitting MS (n = 37) or progressive MS ( n = 48). DR-unmatched controls consisted of 108 adult Caucasians who were not age matched to the MS patients but had no clinical evidence of MS or other “autoimmune disease. All patients and controls were DR typed by RFLP analysis by using the DRP probe in Taql digests; the accuracy of DR-RFLP typing vs. DR serotyping has been previously established by Bidwell (1988) and by ourselves (Haegert ct al., 1989). Individuals among the total DR-unmatched patient and control groups shown to be DR2(15) positive by KFLP analysis were selected to form the DR2( 15) patient and control groups. In addition, 19 DR2(15) individuals who had been previously serotyped as potential organ transplant donors were included in the DR2( 15) control group. ”

Data Analysis The x2 (chi-square) test was used to analyze differences between patient and control groups. When multiple comparisons were performed by using class I1 alIele-specific RFLPs or allogenotypcs (Table 11) we adjusted the a level (a = 0.05) by performing each test at an a’ = a divided by the number of tests, i.e., 0.05 + 7 = 0.007.

DQ

DR

B2 A2 B1 A1

B1 B2*B3 B4 A

67

Fig 1 4 diagram of the named HLA-DR and -DQ genes The D R p probe hybridiLes to DRB 1-4 gcncs, thc DQP probc hybndue, to DQB 1 and DQB2, and thc D Q a probe hybridize7 to DQAl and DQA2 gcnes *. DRB peudogene

DNA Isolation, Restriction Enzyme Digestion, and Southern Blotting DNA was isolated as previoudy described from MS patients and controls by standard techniques including protcinase K digestion of blood samples, phenolchloroform extraction, and ethanol precipitation (Haegert et al., 1989). DNA samples (10 pg) were then digested to completion with Taq I , EcoRl , or EcoRV, electrophorcsed in a 0.8% agarose gel, depurinated, denatured, neutralized (Haegert and Smith, 1987), and then transferred to a nylon membrane (Hybond N: Aniersham) by the method of Southern (1975). DNA Probes, Hybridization, and Autoradiography The Southern blots of Taql-digested DNA were hybridized sequentially as described previously with: 1) an exon-specific 5 17bp fragment of a full-length HLADRP cDNA probe, termed pRTVl (Bidwell and Jarrold, 1986); 2) a 627 bp fragment of an HLA-DQP gene. termed pl I P l (Larhammar ct al., 1982); and 3) a 900 bp DQa probe termed pDCHl (Auffray et al., 1984; for further details see Bidwell et al., 1987; Haegert et al., 1989). DNA probes were labelled with 32P dCTP by using the random priming method of Feinberg and Vogelstein (1983). After hybridization, the membranes were washed as previously dcscribed (Haegert and Smith, 1987) and autoradiography was performed for from 3-8 days. Southern blots of EcoR 1- and EcoRV-digested DNA were probed with thc DQP probe.

RESULTS Class I1 RFLPs of DR-Unmatched Patients (n = 104) and Controls (n = 108) Genoniic DNA samples from all patients and controls who were unmatched for class I1 genes were digested with Taql and the resulting Southern blots were probed sequentially with DKp, DQP, and DQa probes to assign DR and DQ allogenotypic specificities. A diagram of the DR and DQ subregions of the HLA class I1 region I S shown in Figure 1 . Precise definition of c l a 11 ~ allogenotypes was previously established by our identi-

Haegert et al.

68

TABLE I. Typical or Classic Associations of DRB Types With DQB- and DQa-RFLPs in Taql Digests of DNA From Normal Caucasians* DV

Linkage patterns

DRP type" (6.0, 4.5) (13.0. 1.7, 1.4) (11.0. 7.5. 4.0) 113.0. 7.5, 4.0) (14.8, 6.0, 5.4, 2.5) 5(wl I ) (13, 6.5, 4.0) 7' (14.8, 7.5. 4.0,2.6) 7' (14.X, 6.1,4.0,2.5) 8 19.5) 1 2115) 17' 17' 4

d

b c

d e

f 0

h

i

DQP type

a type

la

(5.5) la (2.7) (2.8) l b (h.5) XdiorZa (4.5 &/or 2.7) 2 (5.0) 2a (2.7) 2 (5.0) 3a or 3h (2.0 or 4.6) 3 (6.0) 3b (4.6) 2 (5.0) 2h (7.2) 4 (6.0) 2b or 3a (7.2 or 2.0) 3 (6.0) 3b (4.6) Ib ( 6 . 5 )

Ib

*The nine most frequent DK-DQ linkage patterns arc shown. Parentheses after the DR and DQ types indicate the molecular weights of the RFLPs in kh. "DR17', 172, and 7* share DRP RFLPs with other DR types. These are distinguished from one another o n the basis of' linkage disequilihrium with different DQa and DQS RFLPs (see text for details).

fication of DR and DQ RFLP banding patterns in Taqldigested DNA from a series of typing cells homozygous for different class TI alleles and by our RFLP study of 54 patients who had been previously DR serotyped (Haegert et al . 1989j. The allele-specific Taq 1 RFLPs (allogenotypes) for most of the common HLA class IT genes are shown in Table I. The majority of these and their nomenclature have been described previously by Bidwell and colleagues (Bidwell, 1988; Bidwell et al., 1988,) but the molecular weights of many of the hybridization signals shown here differ from those described by Ridwell (1988) and we report for the first time linkage of DRP 17' with DQPX (DQB2) as well as with DQP2a. DRI, 2(15), 4, 5 ( w l l ) , and 7' and 8 were readily definable solely on the basis of their characteristic DRP Taql RFLPs. In Taql digests DRP17'. 13a', and 1%' share identical DRP RFLPs and DRP17*, 13a4, and 14 share a different set of identical DRP RFLPs but could readily be distinguished one from another on the basis of their linkage disequilibrium with different DQP and DQa allogenotypes (see Bidwell, 1988; Haegert et al., 1989); DRP17' and 172 are Taql RFLP-defined subsets of DRw17. DR72 and 9 share identical DRP, DQP, and DQa Taql RFLPs but were distinguished from one another by their different DQa RFLPs in Mspl digests (Haegert et al., 1989). With the DRP probe a significantly higher percentage of DR-unmatched patients than controls (x' = 14.6; P < 0.001) had hybridization bands at 13.0, 1.7, and I .4 kb (see Fig. 2, lanes 1, 4, and 7, and Table 11); these RFLPs are characteristic of the DRw15 or DR2(15) allogenotype. Two patients and two controls had DRP hybridization bands at 14.8 and 1.7 kb (not shown), i.e., were DRwl6 or DR2(16). The 2.8 kb DQBlb allogeno~

type (x' = 18.8; P < 0.001) (Fig. 2. lanes 2. 5 . and 8) and thc 6.5 kb D Q a l b allogenotype (x' = 13.3; P < 0.001) (Fig. 2, lanes 3, 6, and 9) were present in significantly higher numbers of patients than controls (Table 11). In Taql DNA digests the DQa probe cross-hybridized with two DXa (DQA2) allelic fragments of 2.2 and 2.0 kb mobility, termed DXa upper (DXaU or DQA2Uj and DXa lower (DXaL or DQA2L) allogenotypes respectively (see Fig. 2, lanes 3, 6, 9, 12 for DXaU and/or DXaL hybridization bands). A higher percentage of DRunmatched patients than controls was homozygous or heterozygous for DXuU (x2 = 15.3, P < 0.001) whereas a higher percentage of DR-unmatched controls than patients (x2 = 19.53; P < 0.001) were homozygous for the DXaL hybridization band (Table 11). The relative risk (RR) for MS for individuals who are DR2(15)+, DQPlb ' , D Q a l b ' , D X a U + , and DXaLL+ was calculated as 3.0, 3.5, 2.8, 3.0, and 0.26 respectively (see Svejgaard ct al., 1983, for details of RR calculations). The DXaL (DQA2L) allele is in strong linkage disequilibrium with D R l , 5(wll), 7, and a subset of DR4 (Awad et al., 1988), all of which can be readily defined by RFLP analysis in Taql digests (Bidwell, 1988; Bidwell et al., 1988a; Haegert et al., 1989). This constellation of DR alleles was significantly increased among the control population (x2 = 5.9; P < 0.02).

RFLP Analysis of DR2-Matched Controls and Patients Forty-two controls and 56 patients were matched for DR2( 15) (Table TI). Ninety-six percent of the DR2+ patients but only 83% of the DR2+ controls were also DQP 1 b t ;this difference was not statistically significant. The DQPlb- but DR2 ' controls showed normal DQa linkages, but instead of the usual linkage to DQPlb showed linkage to DQPX or DQPla; the latter are rare associations but have been previously described in Caucasians (Bidwell et al., 1988b). Two of the three DR2+ DQP 1b- patients also showed linkage to DQPX (DQB2) but in addition showed anomalous DQa linkages, and the third patient showed an anomalous DQP linkage (see Table 111). Although there were differences between DR2 patients and controls for the allele-specific DXaU and DXaLL, the differences were not statistically significant. DR-DQ Linkages in Unrelated MS Patients (Table 111) One hundred and four patients were studied with the DRP, DQP, and DQa probes in Taql digcsts to search for aberrant DR-DQ associations. Approximately 80% showed a normal (classic) DR-DQ association pattern for Caucasians; normal Caucasian DR-DQ associations have been previously defined in a study of more

HLA Susceptibility Genes in MS

69

TAR1,K 11. Freauencv of HLA Class I1 WLPs in MS Patients and Controls Allogcnotypes (%)

DR2( 15)

DQPlb

DQa I b

oxauu

DXcuU"

DxcuLL

DXaLa

60* 30

61" 36

21 17

75* 49

20* 49 *

80 82

36

95 83

9 19

68 67

A. DR-unmatched patients and controls 53" MS patients (n = 104) 27 Controls (n = 107) B. DR2( 15)-matched patients and controls DR2 paticnts (n = 56)

96 83

DR2 controls (n = 42)

I00 98

31

"Homozygous or heterozygous DXaU (DQA2U) and DXwL (DQh2L) cross-hybridize with the D Q a probe and have respcctivc RFLPs of 2.2 and 2.0 kb. *Statistically significant at the 0.007 level or greatcr.

Patient 3

Patient 2

Patient 1

Patient 4 V

@!

13

CJ

Lane' 1

2

3

'

' 4

5

6

1

' 1 0 11 12

1

14.8 13.0 -

13.0

~

6.0 -

6.0 -

4.5 -

4.5 -

-3

- la -

- DXiiU - DXoL

1.7 -

1.7 -

1.4 -'

1.4

Normal DR-DQ Linkages

DX