HLA high-resolution typing by next-generation sequencing in Pandemrix-induced narcolepsy

The incidence of narcolepsy type 1 (NT1) increased in Sweden following the 2009–2010 mass-vaccination with the influenza Pandemrix-vaccine. NT1 has been associated with Human leukocyte antigen (HLA) DQB1*06:02 but full high-resolution HLA-typing of all loci in vaccine-induced NT1 remains to be done. Therefore, here we performed HLA typing by sequencing HLA-DRB3, DRB4, DRB5, DRB1, DQA1, DQB1, DPA1 and DPB1 in 31 vaccine-associated NT1 patients and 66 of their first-degree relatives (FDR), and compared these data to 636 Swedish general population controls (GP). Previously reported disease-related alleles in the HLA-DRB5*01:01:01-DRB1*15:01:01-DQA1*01:02:01-DQB1*06:02:01 extended haplotype were increased in NT1 patients (34/62 haplotypes, 54.8%) compared to GP (194/1272 haplotypes, 15.3%, p = 6.17E-16). Indeed, this extended haplotype was found in 30/31 patients (96.8%) and 178/636 GP (28.0%). In total, 15 alleles, four extended haplotypes, and six genotypes were found to be increased or decreased in frequency among NT1 patients compared to GP. Among subjects with the HLA-DRB5*01:01:01-DRB1*15:01:01-DQA1*01:02-DQB1*06:02 haplotype, a second DRB4*01:03:01-DRB1*04:01:01-DQA1*03:02//*03:03:01-DQB1*03:01:01 haplotype (p = 2.02E-2), but not homozygosity for DRB1*15:01:01-DQB1*06:02:01 (p = 7.49E-1) conferred association to NT1. Alleles with increased frequency in DQA1*01:02:01 (p = 1.07E-2) and DQA1*03:02//*03:03:01 (p = 3.26E-2), as well as with decreased frequency in DRB3*01:01:02 (p = 8.09E-3), DRB1*03:01:01 (p = 1.40E-2), and DQB1*02:01:01 (p = 1.40E-2) were found among patients compared to their FDR. High-resolution HLA sequencing in Pandemrix-associated NT1 confirmed the strong association with the DQB1*06:02:01-containing haplotype but also revealed an increased association to the not previously reported extended HLA-DRB4*01:03:01-DRB1*04:01:01-DQA1*03:02//*03:03:01-DQB1*03:01:01 haplotype. High-resolution HLA typing should prove useful in dissecting the immunological mechanisms of vaccination-associated NT1.

Introduction Narcolepsy type 1 (NT1) is a chronic disease characterized by excessive daytime sleepiness and disturbed nocturnal sleep [1,2]. The underlying cause behind NT1 is a specific degeneration of the hypocretin producing neurons in the lateral hypothalamus [3,4]. The cell loss, together with the near-complete association to Human leukocyte antigen (HLA) DQB1 � 06:02 [5], suggests an autoimmune etiology. Historically, it has been difficult to fulfill criteria for autoimmunity in narcolepsy through the identification of autoantigens or disease-specific autoantibodies [6] as well as a disease-related appearance of hypothalamic immune-cell infiltration [2]. Recent studies have identified autoreactive T cells against hypocretin neurons and peptides in NT1 [7][8][9]. The HLA-DQB1 � 06:02 allele is present in about 20% of the general European population; it is more common in northern than in southern countries [5]. Consequently, HLA-DQB1 � 06:02 has been considered as a necessary but not sufficient genetic factor for NT1, since this allele is common in the population, and the incidence of the disease is low. In HLA [5,[10][11][12][13][14][15].

Study populations
In collaboration with the Swedish Narcolepsy Association, families with one member affected by Pandemrix-induced NT1 were informed about the study and gave consent to participate. The legally authorized representatives, parents or guardians of under aged study participants gave informed consent. In total, 31 patients and 66 FDR, geographically scattered all over Sweden, agreed to donate blood samples at their local Health Center. To the best of our knowledge, the patients had cataplexies and were treated accordingly.
The Swedish population controls (n = 448) subjected to high-resolution HLA typing were reported previously [27] and added to LifeGene (https://www.lifegene.se/For-scientists/ About-LifeGene/) population controls (n = 188), representing the control group (GP, n = 636) in the present study. The LifeGene controls were sequenced simultaneously with the NT1 families. The present study is part of the AMINA-Study (Autoimmune Multiple Sclerosis and Narcolepsy), a Swedish research project focused on neurological autoimmune diseases. The study was approved by the Regional Ethical Review Board in Lund, Sweden (dnr 2015/257). All study participants gave written consent to participate.

DNA extraction
DNA from all participants (n = 97) in the NT1 families were extracted using a QIAmp Blood Maxi Kit (Qiagen) according to the manufacturer's instructions. Two 8 mL CPT Mononuclear Cell Preparation Tubes (ref 362 761, Becton Dickinson, Franklin Lakes, New Jersey, U.S.) of whole blood were collected, and DNA was extracted from the remaining erythrocyte layer in the tubes following Peripheral Blood Mononuclear cell (PBMC) isolation. CPT tubes were centrifuged for 5 min at 2000 rpm to free the erythrocyte layer underneath the gel plug. The blood clots were lysed in 0.5 mL Qiagen protease and 6 mL DNA purification AL lysis buffer according to the manufacturer's instructions. Incubation for 10 min in a 70˚C water-bath was followed by the addition of 5 mL 99.5% ethanol. The reaction mixture was applied into QIAamp maxi columns; DNA was bound to the filter during 3 min centrifugation at 4000 rpm. The eluate was discarded and the columns were washed in 5 mL washing buffers AW1 followed by AW2. DNA was eluted in two steps by the addition of 2 × 600 μL Buffer AE and later pooled. The DNA concentration was measured using a Nanodrop device (Thermo Fisher Scientific, Waltham, MA, USA). All samples were diluted in nuclease-free water to a concentration of 50 ng/μL DNA.

4.
A total of 634/636 control subjects could be assigned to two extended haplotypes according to the scheme in step 3 (S1 Table). Conflicting results were found for two individuals; in both cases, the two extended haplotypes were matched to the same HLA-DRB3, DRB4, DRB5 allele and not matched to the other. This was solved by assigning the matched HLA-DRB3, DRB4, and DRB5 alleles to the more common haplotype.
In brief, the allele with the largest deviation between groups was identified and excluded. This sequential elimination of most significant effect was repeated in a new round of calculations. The procedure was redone until no further alleles with a significant deviation between the compared groups could be identified. For this analysis, alleles with a low number of narcolepsy-patients (� 1) were not considered significant.
In brief, allelic frequencies were compared between NT1 patients to GP or FDR. Allelic/ haplotypic associations were assessed using Haplotype-score test. Odds ratios were computed following selection of a reference allele/haplotype with a comparable frequency in both patients and controls. The odds ratio for the reference allele/haplotype was 1, higher and lower values being associated with increased and decreased risk, respectively.

Results
The present analysis represents next-generation high-resolution sequencing extended to both exons 2 and 3 for all HLA-DR, DQ, and DP alleles. The analyses follow the allele location on chromosome 6, and allele frequencies are followed by haplotypes, extend haplotypes and genotypes.
Although not revealed in the RPE analysis, it was observed in the haplo.score analysis that DRB3 � 01:01:02 and non-amplified HLA-DRB3, DRB4, and DRB5 were negatively associated with NT1 patients compared to GP (S2 Table). HLA-DRB5 � 01:01:01 was observed with increased frequency for patients compared to FDR (S6 Table).

HLA-DQA1 alleles
Thirteen different alleles of HLA-DQA1 were found among NT1 patients or GP in a frequency >1% (S2 Table). Also, ten different HLA-DQA1 alleles were found only among GP in a frequency lower than 1% (S3 Table).

HLA-DPA1 and DPB1 alleles
Six different alleles of HLA-DPA1 and fifteen different alleles of DPB1 were found among patients or GP in a frequency >1% (S2 Table). A total of six different alleles of HLA-DPA1 and eleven of DPB1 were found only among GP at a frequency lower than 1% (S3 Table). RPE analysis revealed that DPA1 � 02:06:01 and DPB1 � 23:01:01 were associated with NT1 compared to GP. DPB1 � 02:01:02 and DPB1 � 04:02:01 were less common in patients compared to GP (Table 1 and S4 Table).

Discussion
High-resolution HLA sequencing was used to dissect the association between HLA and NT1 that developed in subjects vaccinated with Pandemrix. The major findings were:  [10][11][12][13].
The strengths of the present study were the high-resolution HLA typing in a group of Pandemrix-induced NT1 patients and controls both from the general population and the NT1 patients respective immediate family. Using the families, it was possible to ascertain by descent correct extended HLA haplotypes. Our study determined allelic frequencies for patients compared to FDR. Three alleles, DRB3 � 01:01:02, DRB1 � 03:01:01, and DQB1 � 02:01:01 were found at a higher frequency among the FDR compared to the NT1 probands and could have had a protective impact in the etiological process leading to NT1. Although there are few family studies of NT1, it is generally accepted that the disease is non-hereditary. The discordance found among twins suggests other factors that might influence the etiology and pathogenesis of this disease in genetically predisposed individuals [31,32].
HLA-DPA1 :01 were reported to be negatively associated with NT1 [15]. In our Swedish patients, we were able to confirm the decreased frequency of DPB1 � 04:02:01 among the NT1 patients. In addition, we reported NT1 to be negatively associated with DPB1 � 02:01:02. The previous [15] and the present report are of interest since HLA-DP associations are suggested to affect the antibody response to influenza vaccination [33]. In addition, DPB1 alleles were related to multiple sclerosis [34], a disease sharing the DRB1 � 15:01:01-DQB1 � 06:02:01 association with NT1.
The current study presents what could be interpreted as a seemingly high number of p-values. However, it should be noted first that comparisons of HLA alleles are not independent of each other. Second, results and p-value of a single allele is replicated in the haplotype and often also in the extended haplotype. For the RPE analysis, the limitation of the small sample size is recognized. It is noted that these results are based on a small sample size and need to be verified independently. Limitations of this study include the lack of follow-up from NT1 genetics to clinical diagnosis and disease severity. NT1 is a complex disease as symptoms, and response to treatments varies in unpredictable ways. HLA-DQA1-DQB1 haplotype preferences have been addressed in previous reports [14], but further extension into extended HLA haplotypes and genotypes is largely an unexplored field as spanning the HLA complex from the DRB3, DRB4, and DRB5 to the DPB1 may have implications to individual patients. As an example, HLA genetics in type 1 diabetes affects the preference of the first autoantibody (IAA or GADA) and has been proposed to be linked to different triggering mechanisms, although both eventually cause beta cell destruction and diabetes [35,36].