Research ArticleInfluenza

Antibodies to influenza nucleoprotein cross-react with human hypocretin receptor 2

See allHide authors and affiliations

Science Translational Medicine  01 Jul 2015:
Vol. 7, Issue 294, pp. 294ra105
DOI: 10.1126/scitranslmed.aab2354


The sleep disorder narcolepsy is linked to the HLA-DQB1*0602 haplotype and dysregulation of the hypocretin ligand-hypocretin receptor pathway. Narcolepsy was associated with Pandemrix vaccination (an adjuvanted, influenza pandemic vaccine) and also with infection by influenza virus during the 2009 A(H1N1) influenza pandemic. In contrast, very few cases were reported after Focetria vaccination (a differently manufactured adjuvanted influenza pandemic vaccine). We hypothesized that differences between these vaccines (which are derived from inactivated influenza viral proteins) explain the association of narcolepsy with Pandemrix-vaccinated subjects. A mimic peptide was identified from a surface-exposed region of influenza nucleoprotein A that shared protein residues in common with a fragment of the first extracellular domain of hypocretin receptor 2. A significant proportion of sera from HLA-DQB1*0602 haplotype–positive narcoleptic Finnish patients with a history of Pandemrix vaccination (vaccine-associated narcolepsy) contained antibodies to hypocretin receptor 2 compared to sera from nonnarcoleptic individuals with either 2009 A(H1N1) pandemic influenza infection or history of Focetria vaccination. Antibodies from vaccine-associated narcolepsy sera cross-reacted with both influenza nucleoprotein and hypocretin receptor 2, which was demonstrated by competitive binding using 21-mer peptide (containing the identified nucleoprotein mimic) and 55-mer recombinant peptide (first extracellular domain of hypocretin receptor 2) on cell lines expressing human hypocretin receptor 2. Mass spectrometry indicated that relative to Pandemrix, Focetria contained 72.7% less influenza nucleoprotein. In accord, no durable antibody responses to nucleoprotein were detected in sera from Focetria-vaccinated nonnarcoleptic subjects. Thus, differences in vaccine nucleoprotein content and respective immune response may explain the narcolepsy association with Pandemrix.


With the declaration of a global A(H1N1) influenza pandemic in 2009 [A(H1N1)pdm09], mass vaccination campaigns using a range of newly developed monovalent A(H1N1)pdm09 vaccines were initiated in a number of countries (1). The highest rates of pediatric and adolescent immunizations were achieved in Finland, Ireland, Norway, and Sweden (2). Almost 1 year after the authorization of A(H1N1)pdm09 vaccines, an increase in reported narcolepsy cases was associated with the European AS03-adjuvanted A(H1N1)pdm09 vaccine (Pandemrix), which was distributed to more than 30.5 million people in European Union/European Economic Area (EU/EEA countries) (2). As of January 2015, more than 1300 cases of vaccine-associated narcolepsy have been spontaneously reported to the European Medicines Agency EudraVigilance database (3). Currently, no increased risk has been reported for the MF59-adjuvanted A(H1N1)pdm09 vaccine (Focetria), for which an estimated 6.5 million doses were distributed in EU/EEA, and 25 million doses were used in Europe and Latin America (4). The possibility that an influenza virus antigen might be a trigger for vaccine-associated narcolepsy (5) is suggested by the cyclical occurrences, in China, of narcolepsy cases observed from 1996 to 2008 after seasonal influenza infections and the peak in narcolepsy cases observed after 2009 in China with the rapidly transmitted A(H1N1)pdm09 virus (6). In that retrospective study of narcoleptic patients (86% children) diagnosed at one hospital in Beijing, China, there was a threefold increase in narcolepsy 6 months after infection with A(H1N1)pdm09 virus (6).

Finland conducted the first series of studies in Europe and reported, in 2012, a 12.7-fold increased risk of narcolepsy in subjects between 4 and 19 years of age occurring within 8 months after vaccination with Pandemrix (7). In addition to increased narcolepsy risk after vaccination in children and adolescents, Finnish analyses of hospital and primary care data also demonstrated a three- to fivefold increased risk in people 20 to 64 years of age who received Pandemrix (8). Narcolepsy associations detected with Pandemrix-vaccinated children have been summarized previously (5) and include Sweden (4.2 cases per 100,000, reported in 2011), Ireland (5.8 cases per 100,000, reported in 2012), England (9.9 cases per 100,000, reported in 2013), Norway (10 cases per 100,000, reported in 2013), and France (odds ratio of 6.5, reported in 2013). The vaccine-attributable risk estimates from the UK and Finnish studies range between 1 in 16,000 doses and 1 in 50,000 doses (9), which makes vaccine-associated narcolepsy a very rare adverse event by definition (1).

In 2014, a report from England identified 75 subjects developing narcolepsy between 2008 and 2011, with disease development within 3 to 10 months in 7 children receiving seasonal influenza vaccine between 2007 and 2011, within 3 to 14 months in 11 children after Pandemrix vaccination, within 6 months in 2 children after an influenza-like illness, and about 2 years after suspected A(H1N1)pdm09 infection in a child in July 2009 (10). This study in England showed no evidence for an earlier diagnosis in vaccinated cases, suggesting that substantial media attention was less likely confounding the demonstrated association. In the Netherlands, Pandemrix was offered to children 6 months to 5 years of age, whereas Focetria was offered to older children as part of a large national vaccination campaign (11). Active case finding in the Netherlands in all 16 referral sleep centers found no cases of narcolepsy for the estimated 654,885 people (ages 5 to 50 years) who were vaccinated with Focetria (154,622 of whom were ages 5 to 19 years) (12). In the 588,750 children receiving Pandemrix, there were 20 diagnosed cases of narcolepsy, of which 3 cases occurred before Pandemrix vaccination, whereas the remaining 17 cases occurred anywhere from a few days to 3 years after Pandemrix vaccination (11). All these cases occurred in children less than 5 years of age with the exception of a 19-year-old adolescent who received Pandemrix. More recently, a retrospective epidemiological study from Germany investigated whether there was a change in incidence rates of narcolepsy before and after A(H1N1)pdm09 (13). Of the 1198 patients with an initial diagnosis of narcolepsy within the observed period, the age-standardized adjusted incidence rate significantly increased from 0.14 per 100,000 person-years in the pre-pandemic period to 0.50 per 100,000 person-years in the post-pandemic period (incidence density ratio, 3.57; 95% confidence interval, 1.94 to 7.00) only in subjects under the age of 18 years and was noted to have started in the spring of 2009. Whether this increase in incidence rate was related to vaccine-associated narcolepsy remains to be determined. However, these findings could be consistent with the role for A(H1N1)pdm09 infection in narcolepsy development as has been suggested by the studies indicated above in England and China.

In Canada, an estimated 12 million doses of another AS03-adjuvanted pandemic vaccine (Arepanrix) were administered (1). Arepanrix is manufactured in Canada using steps different from those used to make Pandemrix in Germany (5), which could result in several differences between the vaccines despite using the same adjuvant, AS03. A possible signal of vaccine-associated narcolepsy was first noted in early 2010 involving narcolepsy referral centers in Montreal (Canada), Montpellier (France), and Stanford University (United States), comprising recent-onset cases of narcolepsy in 14 subjects after vaccination and 2 nonvaccinated subjects after H1N1 infection (14). Whereas the 10 cases outside Canada were linked to Pandemrix vaccination, all 4 cases in Canada were observed in subjects receiving Arepanrix (14). In Quebec, a recent retrospective population-based study to assess narcolepsy risk after vaccination with Arepanrix reported an excess risk of about one case per million vaccine doses, predominantly in subjects less than 20 years of age (15). Although this was lower than in other countries, a clustering of cases with the first and second waves of A(H1N1)pdm09 suggests that viral infection may have had a confounding role (15), again echoing the observations made in other countries (6, 10).

Narcolepsy is a chronic disorder presenting with excessive daytime sleepiness, and the variant with cataplexy (a transient loss of muscle tone triggered by strong emotional stimuli) is tightly associated with HLA-DQB1*0602 (16). Investigations in patients with narcolepsy-cataplexy have revealed that due to loss of hypothalamic cells (17, 18), the neuropeptide hypocretin (19) (HCRT) is deficient in the cerebrospinal fluid. This neuropeptide is also known by the name orexin (20). To date, the exact mechanism culminating in the loss of HCRT-producing cell bodies of the perifornical lateral hypothalamus in narcoleptic patients remains unclear. However, it has been demonstrated that dogs with hereditary narcolepsy have a mutation in the HCRT receptor 2 gene (21). Furthermore, murine models with a loss of function of the HCRT receptor 2 (22) or the preprohypocretin gene (23) and mice with genetic ablation of HCRT neurons (24) recapitulated the narcolepsy phenotype. Last, another idiopathic sleep disorder, excessive daytime sleepiness, has been linked by single-stranded conformational polymorphism analysis to an HCRT receptor 2 coding variant resulting in a Pro11Thr change that impaired HCRT ligand binding–related signaling (25). These findings suggest that alterations in HCRT ligand or its receptors can disrupt neurotransmission and may have key roles in sleep related disorders, including narcolepsy.

Observations from the United States, Canada, and Europe suggest that half of adult narcoleptic patients report disease onset before 15 years of age (26). A report from China demonstrated that 70% of narcoleptic patients had disease onset before age 10, whereas 15% had onset before 6 years of age (26). The population prevalence in China is 0.034%, which is similar to the reported prevalence rates in North America and Europe (26). The delay between onset and diagnosis can be greater than 10 years (27) but is reduced to only a few years when disease onset occurs in childhood (28). Here, we hypothesized that differences between the adjuvanted A(H1N1)pdm09 vaccines Pandemrix and Focetria explain the association of narcolepsy with Pandemrix-vaccinated subjects.


Differences between protein sequences from A(H1N1)pdm09 virus and the derived vaccine reassortant strains identify an influenza nucleoprotein (NP) peptide similar to human HCRT receptor 2

Alignment of protein sequences was performed on sequences retrieved in April 2013 for the A(H1N1)pdm09 strain, A/California/07/2009, and the pandemic vaccine reassortants X-179A and X-181 (indicated in the package insert for Pandemrix and Focetria, respectively), searching for potential differences in the influenza protein sequences expected to be present in the vaccine preparations [matrix, NP, hemagglutinin (HA), and neuraminidase] (29, 30). This approach identified one sequence fragment from influenza HA and two sequence fragments from influenza NP (Table 1) that were common to the A(H1N1)pdm09 strain, A/California/07/2009, and the vaccine reassortant X-179A, but differed, by one residue, in the vaccine reassortant X-181. Because natural infections can trigger autoimmune disease through molecular mimicry (5) (for example, Sydenham chorea is the result of an immune response cross-reacting with similar proteins contained in the microbe, β-hemolytic streptococcus, and the human brain), we compared these three microbial sequence fragments from influenza for similarity to fragments contained in human HCRT ligand or the HCRT receptors (components of the neurotransmission pathway dysregulated in narcolepsy) using a Smith-Waterman alignment. From these three fragments, only the sequence of influenza NP111-121, “YDKEEIRRIWR,” was found to be similar to (mimicked) a sequence from the first extracellular domain (N terminus) of HCRT receptor 234-45, “YDDEEFLRYLWR” (e value = 0.0061), and the corresponding domain of HCRT receptor 127-37, “YEDEFLRYLWR” (e value = 0.026) (Fig. 1). The surface-exposed conformation of most of these extracellular domains of HCRT receptor 2 can now be visualized (fig. S1), because the crystal structure of the receptor was recently published (31). The mimic peptide from influenza NP111-121 (containing residues in common with a peptide from HCRT receptor 234-45) was also found to be located on the surface-exposed region of the influenza NP crystal structure (figs. S1 and S2 and movie S1), which may be relevant for antibody generation as demonstrated by another B cell epitope for influenza NP (32). The conservation of this NP peptide mimic in other influenza strains from 1902 to 2013 is illustrated in fig. S3, and its alignment with HCRT receptors and surface-exposed location suggested a potential role for influenza NP in generating a cross-reactive immune response to HCRT receptors linked with narcolepsy.

Table 1. Influenza virus and vaccine reassortant strain differences in protein sequence.

Influenza sequences were retrieved from the National Center for Biotechnology Information (NCBI) Influenza Virus Resource (75) querying by the strains for A(H1N1)pdm09 virus (A/California/07/2009) and for vaccine reassortants (X-179A and X-181) with duplicate sequences removed. Each influenza protein expected to be present in the vaccine preparations (M1, NP, HA, and NA) was compared across strains, and identified three sequences where X-181 differed from X-179A by at least one residue. Sequences identified for X-179A were similar to those of A(H1N1)pdm09 virus. Residues that vary are underlined.

View this table:
Fig. 1. Alignment studies with influenza sequences and components of the HCRT neurotransmission system demonstrated a shared motif between influenza NP and HCRT receptors.

The influenza NP peptide contained isoleucine (in natural virus and X-179A vaccine reassortant) or methionine (in X-181 vaccine reassortant). Isoleucine in the figure is underlined because of its structural similarity to leucine. Adding in additional flanking residues to the N or C terminus (up to 11 residues total) of the fragments did not extend the alignment beyond the motif indicated. Subscript numbers indicate amino acid positions.

Anti-HCRT receptor 2 antibodies that cross-react with influenza NP peptide are detectable in narcoleptic patients with a history of Pandemrix vaccination

A commercially available cell line expressing either human HCRT receptor 1 or HCRT receptor 2 was used to determine by in-cell enzyme-linked immunosorbent assay (ELISA) whether immunoglobulin G (IgG) antibodies to HCRT receptors were present in the various clinical sera available for testing. Antibodies to HCRT receptor 2 were detectable in a significant proportion of day 500 postvaccination sera from HLA-DQB1*0602 haplotype–positive Finnish narcoleptic patients with a history of Pandemrix vaccination in 2009 (17 of 20 sera) compared to either day 22 or day 202 postvaccination sera from Italian subjects after Focetria vaccination in 2009 (0 of 6 sera for both time points, P < 0.001, χ2 test) (Fig. 2A) or day 18 convalescent sera from Finnish individuals with A(H1N1)pdm09 infection (5 of 20 sera, P < 0.001, χ2 test). Antibodies to HCRT receptors were also detected in sera obtained from children not known to have narcolepsy in 2004/2005 (11 of 20 sera, not significant as P value > 0.05) from Finland, where the genotypic frequency of the narcolepsy-cataplexy–associated HLA-DQB1*0602 allele is 17.1% and projected population coverage is 31.3% (table S1). Double-labeling immunofluorescence microscopy demonstrated colocalization of the staining pattern of narcoleptic patient sera (fig. 2D) with the specific membrane punctate staining seen with a commercial antibody for human HCRT receptor 2 (fig. 2E). Nonspecific background signal indicated in Fig. 2A was confirmed by microscopic studies not to be generated by punctate membrane staining specific for HCRT receptor 2 (fig. S4), and individual background signals are listed in table S2. Although narcoleptic patient sera reactivity (5 of 20 sera) was demonstrated for the cell line expressing human HCRT receptor 1 (fig. S5), differences between groups did not achieve statistical significance.

Fig. 2. Reactivity of clinical sera with cell lines expressing human HCRT receptor 2.

(A) Narcoleptic patients with a history of Pandemrix vaccination in 2009 (red color), subjects at day 22 after vaccination (light blue color) or day 202 after vaccination (dark blue color) with Focetria, individuals with A(H1N1)pdm09 infection (pink color), and Finnish children not known to have narcolepsy in 2004/2005 (yellow color) demonstrated IgG binding to HCRT receptor 2 (data are means ± SD of triplicate experiments; inter-assay variability was zero). BK indicates non–HCRT-specific in-cell ELISA background verified by microscopic studies (fig. S4). The proportion of sera from *0602-positive Pandemrix-vaccinated narcoleptic patients, 1:800 dilution, with reactivity to HCRT receptor 2 (17 of 20 sera) was found to be significantly larger than that calculated for Focetria-vaccinated subjects (0 of 6 sera, P < 0.001, χ2 test for proportions) and for individuals with A(H1N1)pdm09 infection (5 of 20, P < 0.001, χ2 test for proportions). (B) Differential interference contrast microscopy of cell line engineered to express human HCRT receptor 2. (C) Nuclear staining with DAPI (4′,6-diamidino-2-phenylindole) (blue color). (D) Punctate membrane staining with serum, 1:400 dilution, from a narcoleptic patient with a history of Pandemrix vaccination (green color). (E) Punctate membrane staining with commercial antibody to HCRT receptor 2, 1:400 dilution, consistent with recognition of the extracellular domain of membrane-bound HCRT receptor 2 (red color). (F) Merged image (yellow) from double-labeling experiment with narcoleptic patient serum (green) and commercial antibody to HCRT receptor 2 (red). ***P < 0.001.

To further characterize these anti–HCRT receptor 2 antibodies, we initially screened sera from the 33 individuals with HCRT receptor 2–IgG reactivity by ELISA in a blocking experiment using the influenza NP peptide containing the mimic sequence identified by bioinformatics analyses. This 21-mer NP peptide (isoleucine variant) produced from solid-phase peptide synthesis demonstrated 65 to 80% inhibition in IgG reactivity to HCRT receptor 2 (fig. S6). Subsequently, three narcoleptic patients with a history of Pandemrix vaccination and greatest IgG reactivity to HCRT receptor 2 were compared to three narcoleptic patients with a history of Pandemrix vaccination and no IgG reactivity to HCRT receptor 2 in detailed blocking experiments (Fig. 3A) conducted in triplicate (source data viewable in fig. S7). In these studies, the isoleucine variant of the 21-mer NP peptide, the methionine variant of the 21-mer NP peptide, or a recombinant peptide of 55 amino acids (reflecting the properly folded extracellular domain of HCRT receptor 2) demonstrated 62 to 74% inhibition of HCRT receptor 2 binding (multiple linear regression coefficients P value for inhibition effect < 0.001), whereas no significant inhibition was demonstrated with the scrambled variant of the 21-mer NP peptide (Fig. 3A, red bars). Similar experiments performed on sera from narcoleptic patients lacking antibodies to HCRT receptor 2 (Fig. 3A, brown bars) resulted in minimal inhibition. Confirmation of these ELISA data was undertaken using immunofluorescence microscopy of cells expressing human HCRT receptor 2. Using this different platform, the specific membrane punctate staining for HCRT receptor 2 by sera from a narcoleptic patient with a history of Pandemrix vaccination (Fig. 3B) was similarly inhibited by NP and HCRT receptor 2 peptides (Fig. 3, C and D). These studies provide evidence for cross-reactive antibody that recognizes a surface component (3337) of influenza NP and the extracellular domain of human HCRT receptor 2, a component of the HCRT neurotransmission system known to be dysregulated in narcolepsy.

Fig. 3. Inhibition of narcoleptic patient serum IgG binding to HCRT receptor 2.

(A) Sera from narcoleptic patients with a history of Pandemrix vaccination and HCRT receptor 2 antibodies (red color) or without detectable HCRT receptor 2 antibodies (brown color) were preincubated with recombinant peptide comprising the extracellular domain of HCRT receptor 2, peptides for influenza NP (amino acids 106 to 126) containing isoleucine (LILY I) or methionine (LILY M), or a scrambled peptide of the NP peptide containing isoleucine (final serum dilution 1:800). HCRT receptor 2–specific ELISA signal detected previously with patient 120, 136, and 137 sera was subsequently inhibited 62 to 74% (P < 0.001, multiple regression analysis) by either HCRT receptor 2 or NP peptide preincubation confirming the existence of cross-reactive antibodies (data are means ± SD of triplicate experiments, inter-assay variability less than 20%). No significant inhibition was detected using patient 119, 140, and 154 sera. No significant inhibition was detected when using the scrambled peptide. (B to E) HCRT receptor 2–specific punctate staining by serum, 1:400 dilution, from a narcoleptic patient with a history of Pandemrix vaccination (B) was substantially reduced when preincubated with (C) recombinant HCRT receptor 2 peptide, (D) NP peptide containing isoleucine, or (E) NP peptide containing methionine. (F and G) HCRT receptor 2–specific punctate staining by commercial antibody to HCRT receptor 2, 1:400 dilution (F), was inhibited by recombinant HCRT receptor 2 peptide (G). Source data are provided in fig. S7. **P < 0.01; ***P < 0.001.

The similar blocking by the isoleucine and methionine variants of the influenza NP peptide suggested that the one residue difference in these variants may not be contributing to differential B cell receptor recognition in the immune response pathway leading to cross-reactive antibodies in vaccine-associated narcolepsy. Therefore, we investigated whether these NP peptide variants differentially bound to the major histocompatibility complex (MHC) allele product tightly associated with narcolepsy development (HLA-DQB1*0602) because this could contribute to differences in T cell help linked to subsequent antibody generation (38). Three-dimensional modeling of HLA-peptide binding (fig. S8) suggested this to be the case, and was subsequently confirmed by differential in vitro binding of NP variant peptides to the HLA-DQA1*0102:DQB1*0602 allele product using a commercially available, ProImmune MHC class II binding, assay (table S3). Peripheral blood mononuclear cells from narcoleptic patients were not available for investigation of T cell receptor recognition and T cell reactivity studies. However, functional evidence for human lymphocyte antigen (HLA) class II processing of the identified NP mimic peptide is suggested by published T cell studies on influenza NP peptides containing the mimic sequence (table S4). The next logical step, after the above-mentioned in vitro studies, was to characterize the influenza NP content in commercial influenza vaccines and the anti-NP immune response to influenza vaccination in subjects.

The lowest amounts (trace) of influenza NP are detected in Focetria

Influenza NP content was quantitated in eight commercially available inactivated seasonal influenza vaccines and three A(H1N1)pdm09 vaccines. Seasonal influenza vaccines were compared by gel electrophoresis (Fig. 4A) and Western blotting using two commercially available influenza A NP-specific monoclonal antibodies (Fig. 4, B and C, respectively). These studies demonstrated qualitatively similar amounts of NP in the trivalent/quadrivalent split-virion vaccines (Fluzone, Fluarix, and Afluria) and the trivalent subunit vaccine, Fluvirin. However, NP in the trivalent subunit vaccines Agrippal and Chiromas (Chiromas is similar to the MF59-adjuvanted Fluad) was below the level of detection of SDS–polyacrylamide gel electrophoresis (SDS-PAGE) and Western blotting (complete image of blots for Fig. 4, B and C, shown in fig. S9). Using mass spectrometry, NP (normalized to strain A HA content) was detectable in both Agrippal and Chiromas in smaller quantities when compared to other seasonal vaccines whose NP content was about 3 to 12 times greater (Table 2). The extracted ion chromatograms from this mass spectrometry analysis additionally confirmed the presence of the previously mentioned NP peptide variants in related vaccines (fig. S10)—both of which could potentially be present in influenza vaccines because of the existence of mixed populations generated during the reassortment procedure (figs. S11 to S13). Mass spectrometry of the monovalent A(H1N1)pdm09 vaccines indicated that Pandemrix contained the highest amounts of NP. Relative to Pandemrix, Arepanrix contained 28.1% less NP, and Focetria contained 72.7% less NP (Table 2), suggesting the possibility that lower NP concentrations in Focetria could have attenuated both the immune response to NP and the subsequent generation of NP antibodies capable of cross-reactivity with HCRT receptor 2.

Fig. 4. Differential NP content in inactivated seasonal influenza vaccines.

(A) SDS-PAGE lanes loaded per micrograms of HA: (1 and 2) Fluzone trivalent, 1.8 and 0.9 μg; (3 and 4) Fluzone quadrivalent, 2.4 and 1.2 μg; (5 and 6) Fluarix trivalent, 1.8 and 0.9 μg; (7 and 8) Fluarix quadrivalent, 2.4 and 1.2 μg; (9 and 10) Afluria trivalent, 0.75 and 0.375 μg—lower amount of HA was loaded for Afluria to enable resolution of clear bands; (11 to 15) NP standard, 4, 2, 1, 0.5, and 0.25; (16) Agrippal trivalent, 1.8 μg; (17) Chiromas trivalent, 1.8 μg; and (18) Fluvirin trivalent, 1.8 μg. (B) Western blotting with NP-specific monoclonal antibody, lnA245. Same vaccines and loading as in A except (9) 0.9 μg, (10) 0.45 μg, and (11 to 15) NP standard, 5, 2.5, 1.25, 0.625, and 0.3125 μg. (C) Western blotting with another NP-specific monoclonal antibody, InA108. Loading reduced for lanes 1 to 10 due to more sensitive antibody and intentionally biased against Agrippal, Chiromas, and Fluvirin as follows: (1 to 10) 0.3 and 0.15 μg, respectively; (11 to 15) NP standard, 1, 0.5, 0.25, 0.125, and 0.0625 μg; (16) Agrippal, 1.8 μg; (17) Chiromas, 1.8 μg; and (18) Fluvirin, 1.2 μg. NP content in the subunit vaccines Agrippal and Chiromas (Chiromas is similar to the MF59-adjuvanted Fluad) was below the level of detection of SDS-PAGE and Western blotting. Complete image of blots for (B) and (C) is shown in fig. S9.

Table 2. Quantification of NP in influenza vaccines by mass spectrometry.

For each vaccine, the proteins were quantified by spectral count of related strains indicated in the package insert. Because commercial vaccines indicate primarily micrograms of HA content, mass spectrometry–quantitated NP spectral counts were expressed in relation to HA spectral count. NP content in seasonal split-vaccines was about 3 to 12 times greater than that detected in seasonal subunit vaccines, with the exception of Fluvirin. Testing of Focetria by mass spectrometry demonstrated NP-to-HA ratios similar to those seen with Agrippal and Chiromas. NP content of the A(H1N1)pdm09 split-virion vaccines Arepanrix and Pandemrix was about 2.9 and 3.7 times greater, respectively, than that detected in the A(H1N1)pdm09 subunit vaccine Focetria.

View this table:

Greater titers of influenza NP antibodies are present in individuals with A(H1N1)pdm09 infection compared to subjects vaccinated with Focetria

A commercially available ELISA kit with whole influenza NP was used to assess the presence of NP antibodies in nondiluted clinical sera and at 1:160 and 1:640 dilutions (Fig. 5). Sera from HLA-DQB1*0602 haplotype–positive narcoleptic patients with a history of Pandemrix vaccination were available only for day 500 after vaccination (Fig. 5A, red circles). Sera from subjects vaccinated with Focetria reflect day 202 after vaccination (Fig. 5B, dark blue circles) and, although not directly comparable temporally to narcoleptic subjects with a history of Pandemrix vaccination, demonstrate the presence of NP antibodies in both groups. However, day 18 sera from individuals with A(H1N1)pdm09 infection (Fig. 5D, pink circle) could be matched temporally and compared to Focetria-vaccinated subjects for whom sera were available for day 22 after vaccination (Fig. 5C, light blue circles). Compared to Focetria immunization [1 of 6 sera, 17% with positive NP reduction index (NPRI), at serum dilution of 1:320], greater titers of anti-NP antibodies were associated with individuals with A(H1N1)pdm09 infection (16 of 20 sera, 80% with positive NPRI at serum dilution of 1:320, P < 0.05, χ2 squared test). Additional dilutions of sera from individuals with influenza infection demonstrated that 13 of 20 sera became negative for NP reactivity at a titer of 640, whereas the remaining 7 sera became negative at a titer of 1280 (table S5). Sera from Finnish children not known to have narcolepsy in 2004/2005 (Fig. 5E) demonstrated similar patterns to vaccinated groups except for the presence of NP antibodies in three children that were similar to levels seen in individuals with A(H1N1)pdm09 infection (Fig. 5D). Additional dilutions demonstrated that sera from two of these three children became negative for NP reactivity at a titer of 640, whereas the remaining child became negative at a titer of 1280 (table S6).

Fig. 5. Influenza A virus NP antibody inhibition test.

(A to E) Undiluted, 1:160 diluted, and 1:320 diluted serum samples demonstrate presence of NP antibodies of different titers in serum from (A) narcoleptic patients with a history of Pandemrix vaccination in 2009 (about day 500 after vaccination), (B and C) subjects vaccinated with Focetria in 2009 (day 202 after vaccination or day 22 after vaccination, respectively), (D) individuals with A(H1N1)pdm09 infection (18 days after infection), and (E) Finnish children not known to have narcolepsy in 2004/2005. NPRI indicated in the figure reflects the presence of NP-specific antibody in sera that results in inhibition of NP influenza antigen in capture ELISA kit (Virusys Corporation). Only in individuals with A(H1N1)pdm09 infection (D) the proportions of positive results using undiluted sera and sera diluted by the factor 1:320 were not statistically different (P = 0.1), indicating that infection elicited very high antibody titers. At serum dilution of 1:320, antibody concentrations in (D) individuals with A(H1N1)pdm09 infection were significantly greater (16 of 20, 80% with positive NPRI) than those from (C) time-matched Focetria vaccinated subjects (1 of 6 sera, 17% with positive NPRI, P < 0.05, χ2 squared test). *P < 0.05.

Sera from narcoleptic patients obtained closer to the time of vaccination with Pandemrix were not available, thus making impossible a comparison of the immune responses in these patients to those measured in six subjects after vaccination with Focetria in December 2009. Because sera were collected at multiple time points in these six subjects, a detailed time point analysis of immunogenicity to the trace NP contained in Focetria was conducted with sera (1:80 dilution) from baseline and postvaccination day 8, day 22, and day 202 (Fig. 6, source data viewable in table S7). Day 202 values (July 2010) served to establish individual background values not confounded by A(H1N1)pdm09 virus circulating in late 2009. These analyses suggested that subjects 2, 4, and 5 had already been exposed to influenza virus, which would confound interpretation of immune response to NP contained in Focetria. However, results from subjects 1, 3, and 6 suggest that they were not exposed to influenza virus. Antibody levels indicated that the trace NP contained in Focetria vaccine elicited a significant response to NP in subject 1 at day 8 (P < 0.01, unpaired t test) but was transient with a significant decrease in NP titers already at day 22 (P < 0.01, unpaired t test). No significant increase in antibody titers to NP was demonstrated in subject 3 or 6, but their ability to mount immune responses, in general, was confirmed by a significant increase in antibodies at day 8 (P < 0.001, unpaired t test) to membrane-bound influenza antigens (for example, HA) known to be enriched in subunit vaccines like Focetria (fig. S14). Although the sample size is small, the findings from subjects 1, 3, and 6 suggest that the trace amounts of NP in Focetria would not elicit durable NP antibody responses necessary for subsequent cross-reactivity to HCRT receptor 2.

Fig. 6. Influenza NP-associated immune response in subjects vaccinated with Focetria in 2009.

Sera (1:80 dilution) at baseline, day 8, day 22, and day 202 after Focetria vaccination were tested for antibodies to NP using a commercial ELISA kit (quadruplicate experiments, inter-assay variability less than 30%). Antibody responses (y axis) represent NPRI described in Fig. 5. Day 202 values (July 2010) served to establish background antibodies in the absence of infection, suggesting that in December 2009, subjects 2, 4, and 5 had previously been exposed to influenza A virus (subject labels capitalized to facilitate quick identification). The significant response to NP in subject 1 at day 8 (P < 0.01, unpaired Student’s t test) was transient with a decrease in NP antibodies already at day 22 (P < 0.01, unpaired Student’s t test), suggesting suboptimal amounts of NP to elicit a durable immune response to NP. Subjects 3 and 6 demonstrated no statistically significant NP antibody differences over time. Asterisks directly over time point indicate significance relative to day 1 time point. Asterisks centered over horizontal bar indicate significance between indicated time points. Subject numbers were also indicated in parentheses near asterisks to indicate significance of those time points for each indicated subject. Logarithms of the data were represented for each subject and time point along with their arithmetic means and 95% confidence intervals. Source data are provided in table S7. *P < 0.05; **P < 0.01.


HCRT neurotransmission within the brain is dysregulated in human narcolepsy. We found an increased frequency of antibodies to HCRT receptor 2 in narcoleptic patients with a history of Pandemrix vaccination that were cross-reactive with influenza NP. Because an increased risk of narcolepsy has been confirmed with only one adjuvanted A(H1N1)pdm09 vaccine (Pandemrix), we examined whether differences in NP antigen content could explain some of the findings presented in this study. Because the A(H1N1)pdm09 vaccines Focetria, Pandemrix, and Arepanrix were manufactured only in 2009/2010, we initially tested non-expired split-virion and subunit seasonal influenza vaccines to control for any NP artifacts related to degradation in the expired A(H1N1)pdm09 vaccines. Mass spectrometry detected the lowest NP/HA ratios (for strain A influenza) in the seasonal influenza subunit vaccines Agrippal and Chiromas. The similar NP/HA ratio in these vaccines compared to the A(H1N1)pdm09 vaccine Focetria is consistent with the similar manufacturing process used to make these three subunit vaccines. Comparing A(H1N1)pdm09 vaccines, mass spectrometry demonstrated greater quantities of NP in Pandemrix when compared to Arepanrix, in alignment with two other studies (39, 40), and trace quantities of NP in Focetria. Although the sample size is small, detailed time point analyses of the immune response in three non–influenza-exposed subjects immunized with Focetria in 2009 suggest that the trace amounts of NP in Focetria would not elicit durable NP antibody responses necessary for subsequent cross-reactivity to HCRT receptor 2. Thus, the differences in vaccine NP content and respective immune response may explain the association of narcolepsy with Pandemrix-vaccinated subjects. Notably, not all subunit vaccines contained low amounts of NP. For example another subunit vaccine, Fluvirin, had quantities of NP similar to some of the split-virion vaccines tested. These results with Fluvirin stand in contrast to two earlier studies—one not using an NP standard (29) and the other unable to discriminate NP and neuraminidase on gel electrophoresis (30). The broad array of vaccines tested indicate that the amount of influenza NP varies widely both between, and among, split-virion and subunit vaccines. Regarding differences in the incidences of vaccine-associated narcolepsy seen with Arepanrix in Canada or with Pandemrix in Europe, other studies have suggested differences in either amount of structurally altered viral NP (40) or amount of deamidation of viral HA (39).

The HLA-DQB1*0602 allele, tightly linked to narcolepsy-cataplexy, also has been associated with increased antibody responses to NP (40). If a highly conserved NP peptide were triggering pathogenic autoimmunity through antibody cross-reactivity, this could explain the observation of the seasonal “sleepy sickness” variant of encephalitis lethargica that followed the 1918 Spanish flu (5, 41, 42) and the cyclical associations of narcolepsy with seasonal influenza infection in China (6)—all of these earlier influenza strains contained the NP mimic peptide identified in this study. It is worth mentioning that whereas studies outside China have not reported an increase in narcolepsy cases in unvaccinated subjects, there may be a confounding factor worth considering (5). Beijing is one of the most populous cities of the world whose high residential density makes residents highly susceptible to influenza (43). This close proximity for transmission of virus and the limited uptake of vaccines [1.36 million of the 70 million population, 1.94%, received vaccine (44)] may have led to an increased and accelerated exposure of narcolepsy-susceptible subjects to circulating A(H1N1) infection in contrast to European populations not reporting an increase in narcolepsy cases in unvaccinated subjects. Finally, the influenza NP mimic peptide identified in this study aligns with residues 34 to 45 of HCRT receptor 2 close to the N-terminal region of the canine HCRT receptor 2, in which a single amino acid substitution of glutamic acid to lysine at residue 54 abolishes HCRT binding in Dachshunds afflicted with narcolepsy (45). Both HCRT receptors 1 and 2 are highly conserved across mammalian species and, compared to one another, have an overall sequence identity of 64% with conserved sequences in the extracellular N-terminal domain (20).

Our findings lead us to propose a mechanism for influenza infection/pandemic vaccine-associated narcolepsy as follows: In subjects with genetic susceptibility to narcolepsy, the presentation of NP antigen during infection or after immunization with adjuvanted influenza vaccines containing increased amounts of NP may generate high titers of NP antibodies that can persist in the systemic circulation for months. During this time period, either the high titers of NP antibodies or inflammation related to an unrelated infection (for example, streptococcus) may alter the blood-brain barrier (46), allowing NP antibodies to cross-react with neural tissue expressing HCRT receptors [the direct role of adjuvants alone in triggering autoimmune disease in genetically susceptible subjects is less likely based on their inability to trigger or exacerbate disease activity in genetically susceptible patients with established or active autoimmune disease (5), respectively]. Binding of the HCRT receptor by cross-reacting NP antibody might modulate signaling to the cells responsible for producing the ligand (HCRT) and whose disappearance has been demonstrated in pathological investigations of narcoleptic brains (18). HCRT receptor 2 dysregulation through binding of cross-reactive autoantibodies finds precedent in another anti-receptor–mediated neurological autoimmune disease in which autoantibodies target the acetylcholine receptor (AChR), myasthenia gravis (47).

In contrast to our findings, a 2006 radioligand-based study using recombinant HCRT and HCRT receptor proteins to test serum from narcoleptic patients detected autoantibodies in 5% of patients with narcolepsy-cataplexy (HCRT receptor 2 > HCRT > HCRT receptor 1) compared to 3% in controls (48). However, serum in that study was obtained from patients decades after disease onset (average age of 45 years), and as the authors suggested, the cell-free system used for radioligand detection (48) of receptors may have prevented the correct conformational presentation of the receptors necessary for antibody detection. The cell-based assay used in this study detected HCRT receptor 2 autoantibodies in 85.0% of narcolepsy patients with a history of Pandemrix vaccination compared to 34.7% in the nonnarcolepsy control groups tested (Focetria-vaccinated subjects, 0%; individuals with A(H1N1)pdm09 infection, 25%; and Finnish children not known to have narcolepsy in 2004/2005, 55%). The increased detection by the cell-based assay finds parallels to a report for myasthenia gravis in which a cell-based assay engineered for tight packing of AChRs identified AChR antibodies in 60% of patients previously negative for AChR by radioimmunoprecipitation assay (49), emphasizing the importance of the physiological conformation of the receptor. The association between being narcoleptic and having antibodies to HCRT receptor 2 detected in this study was statistically significant (P < 0.001, Fisher test). Although there was about a 50% probability of vaccine-associated narcolepsy in the 33 individuals with antibodies to HCRT receptor 2 (17 of 33, 51.6%), there was a very low probability of vaccine-associated narcolepsy among the 33 individuals without antibodies to HCRT receptor 2 (3 of 33, 9%). This suggests that the absence of antibodies to HCRT receptor 2 may be associated with a low risk of vaccine-associated narcolepsy with influenza NP exposure.

A study using fixed sections of rat brain (reflecting natural tissue physiology) demonstrated that sera from 27% of patients with narcolepsy (60% with a history of Pandemrix vaccination) contained antibodies staining distinct cell populations in the rat brain that were also detected, albeit in lower numbers, from other sleep disorders and some controls (50). These populations included the following: (i) cell bodies in the zona incerta-lateral hypothalamic region/arcuate nucleus (but not HCRT neurons) and in varicose-type nerve terminals; (ii) GABAergic [γ-aminobutyric acid (GABA)–releasing] hippocampal and neocortical interneurons; and (iii) cell bodies, dendritic processes, and axons in the globus pallidus, amygdala, and piriform cortex. Passive transfer to rats of IgG purified from sera of two 19-year-old narcolepsy patients (both with a history of Pandemrix vaccination) resulted in interference of the fine regulation of sleep (not observed with IgG purified from 37-year-old Pandemrix-vaccinated healthy control). The tissue studies are consistent with our discovery of antibodies against HCRT receptor 2 in narcoleptic patients because HRCT receptors map to other regions of the brain that receive projections from the HCRT-producing cells in the hypothalamus (51).

The findings reported in this study should be viewed as a step in understanding the mechanism of vaccine-associated narcolepsy and will benefit from additional confirmatory studies. Future studies could include comparative studies of the different vaccines and their adjuvants and additional in vitro studies with serum, peripheral blood mononuclear cells, and cerebrospinal fluid that may be banked in countries with patients diagnosed with vaccine-associated narcolepsy (Finland, Sweden, Norway, Ireland, France, United Kingdom, Netherlands, and Germany). We would like to emphasize some limitations and related observations of this study that merit attention and future investigations:

1) Clear delineation of CD4 T cells that recognize HLA-DQB1*0602–associated NP peptide and provide help for B cells leading to antibodies cross-reacting with HCRT receptor 2 would be desirable—including the influence of the isoleucine and methionine variants of the NP peptides identified in this study. Unfortunately, such T cells matched from the patients studied here for antibody to HCRT receptor 2 were not available. However, what is known is that endogenously expressed or exogenously applied NP of influenza A is efficiently presented by class I and class II MHCs and is capable of expanding both CD8+- and CD4+-specific effector T lymphocytes (52, 53). Cell-mediated immunity to NP is traditionally linked to class I (infected cells) but can involve cross-presentation of exogenous antigen to class I MHC (5355). The antibody response is class II–restricted, but NP protein immunization in C57BL/6 mice has been demonstrated to be responsible for accelerated viral clearance (56).

2) HCRT receptor 2 antibodies are detected in normal-appearing individuals. Peripheral blood mononuclear samples from individuals without narcolepsy were not available and thus precluded retrospective testing for the HLA-DQB1*0602 haplotype. Antibodies to HCRT receptor 2 were detected in some Finnish individuals with A(H1N1)pdm09 infection or Finnish children not known to have narcolepsy in 2004/2005. It is possible that some of the Finnish children not known to have narcolepsy in 2004/2005 may have had subclinical infection or have received the seasonal influenza vaccine in those years. The influenza vaccine composition recommended by the World Health Organization for 2004/2005 based on influenza viruses circulating worldwide at that time were A/New Caledonia/20/1999 (H1N1-like), A/Fujian/411/2002 (H3N2-like), A/Wellington/1/2004 (H3N2-like), and A/California/7/2004 (H3N2-like) (57). Those influenza strains are identical to the A(H1N1)pdm09 strain with regard to the NP mimic peptide identified in this study, because it is a highly conserved influenza virus epitope. Given the observed time lag between disease onset and disease diagnosis in both narcolepsy and vaccine-associated narcolepsy (7, 27, 28), the presence of autoantibodies in some of the Finnish children not known to have narcolepsy in 2004/2005 (whose HLA haplotype is unknown) could be reflecting the first steps to disease development (10) in some individuals likely to be carrying the narcolepsy susceptibility allele because the projected coverage for the HLA-DQB1*0602 allele in Finland is 31.3% of the population (58). The increased detection of these antibodies in both children vaccinated with Pandemrix and those not known to have narcolepsy in 2004/2005 compared to adults with A(H1N1)pdm09 infection is intriguing because it parallels the observed increased fold risk of narcolepsy in Pandemrix-vaccinated children compared to Pandemrix-vaccinated adults. One potential explanation, although speculative, for this increased autoantibody generation could be related to “germ theory.” The naïve and “primal” immune response in children, unlike the precisely regulated and specific immune response that has been refined in adults through repetitive infections in life, may trigger autoimmunity more readily when reacting to a microbe containing antigens resembling normal tissue (for example, molecular mimicry). The progression from autoimmunity to autoimmune disease development would then depend on other host factors including genetic susceptibility.

Alternatively, natural IgG autoantibodies have also been detected in the sera of normal subjects (59), and it is well known that up to 20% or more of otherwise healthy people can have anti-nuclear antibodies associated with autoimmune diseases including systemic lupus erythematosus (60, 61). Healthy individuals have also been demonstrated to have antibodies against thyroid-stimulating hormone receptor seen in Graves hyperthyroidism (62), thyroglobulin associated with Hashimoto’s thyroiditis (62), or immunity to myelin proteins that classically are associated with demyelinating diseases like multiple sclerosis (6365). Thus, there are numerous precedents for finding autoantibodies associated with autoimmune disease in otherwise healthy-appearing individuals.

3) Sera from age-matched, HLA-DQB1*0602–positive, subjects vaccinated with Focetria, from subjects not receiving Pandemrix but developing narcolepsy, or from subjects receiving Pandemrix but not developing narcolepsy would have been ideal. Unfortunately, sera from these described control groups were not available for this investigation nor were peripheral blood mononuclear samples from the subjects vaccinated with Focetria to assess HLA-DQB1*0602 haplotype positivity. One needs to remember that the Pandemrix association with narcolepsy was identified in 2010, which led to a delay in the collection of sera samples close to the time of Pandemrix vaccination in 2009 (for example, day 500 postvaccination sera) or from various matched controls. Prospectively vaccinating subjects for the purposes of this study was impossible because the A(H1N1)pdm09 vaccines were no longer manufactured and could raise ethical concerns given the accumulating evidence related to vaccine-associated narcolepsy. Fortunately, a few sources of banked sera from Focetria- and Pandemrix-vaccinated subjects in 2009 were available for the current influenza-related study. Note that in addition to increased narcolepsy risk after vaccination in children and adolescents, Finnish analyses of hospital and primary care data also demonstrated a three- to fivefold increased risk in people 20 to 64 years of age who received Pandemrix (8). This age group matches that of the Focetria-vaccinated subjects included in this study who did not develop narcolepsy.

4) The induction of experimental autoimmune narcolepsy in an HLA-DQB1*0602 transgenic mouse with the NP mimic peptide identified in this study would provide additional evidence for the hypothesized mechanism of narcolepsy induction. Back translation from clinical studies will likely require additional engineering to ensure that the appropriate human T cell repertoire is present in humanized mice. Indeed, a recent report indicated that Pandemrix immunization in mice transgenic for DQB1*0602 did not elicit narcolepsy. The investigators suggested that the T cell receptor repertoire (not humanized) in such mice may have been inadequate for elicitation of narcolepsy after immunization with Pandemrix (40).

Cross-reactive epitopes found in vaccines may potentially lead to rare adverse events in genetically susceptible individuals. However, it is not prudent to conclude that influenza vaccination should be avoided, especially in medical risk groups. Here, the influenza vaccine NP epitope cross-reactive with HCRT receptor 2 is also present in the natural virus from which the high-yielding vaccine reassortants are derived using classical reassortment (for example, X-179A and X-181). Furthermore, acquiring influenza infection is associated with the induction of NP antibodies that we demonstrated occur in greater titers compared to time-matched vaccination (for example, Focetria). Thus, in individuals genetically predisposed to narcolepsy development, exposure to influenza infection may potentially lead to stronger NP antibody cross-reactivity with HCRT receptors than immunization with influenza vaccines. The benefits of influenza vaccination currently far outweigh the risks of complications like vaccine-associated narcolepsy and those related to natural infection. Nonetheless, reduction of adverse events will come from a deeper understanding (66) of the composition of vaccines themselves and to even more in-depth analysis of the adaptive and innate immune responses to both natural infection and to vaccines that prevent such infections.


Study design

There was no prespecified effect size in this retrospective study. Sample sizes were limited by the number of sera available from the original retrospective study and by the amount of reagents (for example, peptides) available to run the inhibition experiments. The maximum number of sera was analyzed retrospectively compatible with these limitations. Post hoc power analysis shows that, in each instance when statistical significance was reported, the probability of rejecting the null hypothesis exceeded 90%. No samples were excluded from the study. Assay endpoints were prospectively defined. Replicates for experiments depended on the assay used and included both sampling and experimental replicates. Results were substantiated by repetition using different platforms. The prespecified hypothesis defining the research objectives of this study was that differences between the adjuvanted A(H1N1)pdm09 vaccines, Pandemrix and Focetria, explain the association of narcolepsy with Pandemrix-vaccinated subjects. Clinical sera from narcoleptic patients with a history of Pandemrix vaccination, individuals with A(H1N1)pdm09 infection, and Finnish children not known to have narcolepsy in 2004/2005 were provided by National Institute for Health and Welfare (THL) (Helsinki, Finland). Clinical sera from normal subjects vaccinated with Focetria were provided by Azienda Ospedaliera University of Siena. All sera were deidentified (assigned randomization numbers) and processed at the same time in controlled laboratory experiments. Investigators performing assays (ELISA or microscopy) on samples were blinded to the identity of samples and recorded the outcome by randomization number assigned to each sample.

For materials and procedures indicated below, additional details are provided in the Supplementary Materials section along with descriptions for supplementary data generated by (i) structural analyses of the location of influenza NP peptide sequence analogous to HCRT receptor, (ii) prediction of peptide structure, (iii) HLA-DQB1*0602 allele population coverage determination, (iv) three-dimensional modeling of HLA-peptide binding, (v) HLA binding studies, and (vi) mass spectrometry analysis of influenza vaccines.

Influenza vaccines

Fluzone trivalent 2012–2013 and Fluzone quadrivalent 2013–2014 (Sanofi Pasteur Inc.); Fluarix trivalent 2012–2013 and Fluarix quadrivalent 2013–2014 (GlaxoSmithKline); Afluria trivalent 2012–2013 (CSL Limited); Agrippal trivalent 2012–2013 and Chiromas (MF59-adjuvanted vaccine equivalent of Fluad) trivalent 2012–2013 (Novartis Vaccines); Fluvirin trivalent 2012–2013 (Novartis Vaccines). Focetria from 2009 and 2010 (Novartis Vaccines). Arepanrix from 2010 (GlaxoSmithKline) and Pandemrix from 2009 (GlaxoSmithKline) were available to the mass spectrometry facilities of Stanford University School of Medicine as part of ongoing studies characterizing influenza vaccines.

Clinical serum samples

Informed consent was obtained after the nature and possible consequences of the studies were explained. HLA-DQB1*0602 haplotype–positive narcoleptic patients with a history of Pandemrix vaccination: Serum specimens from 20 Pandemrix-vaccinated narcoleptic patients (mean age, 12.2 ± 2.7 years) were collected during 2011 at the THL, Helsinki, Finland, and were HLA-typed and found to all be positive for HLA-DQB1*0602. Finnish children not known to have narcolepsy in 2004/2005: Serum was obtained from 20 randomly selected individuals (mean age, 9.9 ± 3.3 years) from a larger serum sample collection drawn in 2004 and 2005 that was from diagnostic leftover serum specimens from the HUSLAB diagnostic laboratory, Helsinki, Finland, which provides laboratory services in the Uusimaa region using 70 sample collection sites. HUS is the Hospital District of Helsinki and Uusimaa. The samples were made anonymous by destroying all personal identification data before shipment to THL; only the age of the subjects and the sample collection dates were retained. HLA haplotype is not known. Permission to use these anonymized serum samples was based on ethical permissions given by Helsinki-Uusimaa Health District Ethical committee [initially for polio and measles-mumps-rubella (MMR) antibody analysis in 2005 and expanded to influenza antibody analysis in 2009]. Sera samples were confirmed, on subsequent testing, to be negative for anti-H1N1pdm09 antibodies per hemagglutination inhibition test (67, 68). Individuals with A(H1N1)pdm09 infection: Convalescent serum samples were obtained from 20 patients (ages 18 to 75) with clinical and laboratory-confirmed H1N1pdm09 infection and were collected during the influenza epidemic season 2009–2010 and 2010–2011 in the Tampere city area (Finland) as part of a prospective clinical cohort study. The HLA haplotype is not known for these subjects. Subjects vaccinated with Focetria: Serum specimens from six Focetria-vaccinated subjects (mean age, 42.9 ± 12.2 years) who received one dose of Focetria administered intramuscularly in December 2010 were collected at the Azienda Ospedaliera Universitaria Senese, Siena, Italy. HLA haplotype is not known for these subjects.

Gel electrophoresis and Western blotting (influenza vaccines)

Reducing SDS-PAGE: performed using NuPAGE precast 4 to 12% bis-tris gels. Western blotting: transferred samples were incubated overnight at 4°C with influenza A NP-specific murine monoclonal antibodies diluted 1:500 (MAb InA245 and MAb InA108, HyTest Ltd.) and detected with IRDye 800CW Goat anti-Mouse IgG (H + L) diluted 1:50,000 (LI-COR Biotechnology).

ELISA for assessment of antibodies to influenza NP in clinical samples

Presence of NP antibodies in sera was ascertained using a commercial kit (Virusys Corporation) designed to test the ability of NP antibodies contained in test sera to inhibit NP antigen (full-length H1N1 NP from A/Beijing/262/1995 containing YDKEEIRRIWRQANNG) binding in a capture ELISA. The NPRI calculated for each sera was carried out with undiluted sera or various dilutions to assess spectrum of antibodies (weak to strong). For the six subjects vaccinated with Focetria, multiple time points were available for testing, so NPRI was assessed in quadruplicate and done, intentionally, at 1:80 dilution in order not to miss weak signal if present.

Single radial hemolysis assay to influenza antigen

The six subjects vaccinated with Focetria who had testing for NP antibodies were also assessed for antibodies to membrane-bound influenza antigens. Single radial hemolysis (SRH) against A/H1N1/California/07/2009 (batch 1068/09) was performed per established protocol (69).

Assays for antibody binding to human HCRT receptor

Cell-based ELISA: cloned cell lines made in the Chem-1 host expressing high levels of recombinant human HCRT receptor 1 or HCRT receptor 2 on the cell surface were purchased from Eurofins and were cultured in separate ELISA microplates (Thermo Fisher Scientific). Cell lines were tested by Eurofins and negative for mycoplasma contamination. Then, a colorimetric in-cell ELISA kit (Thermo Fisher Scientific) was used to probe either human HCRT receptor 1– or HCRT receptor 2–expressing cell lines with positive control antibodies (Abcam) or human sera (1:800 dilution) in triplicate experiments. Positive control antibodies to HCRT receptors were used at 1:50 dilution and were either a rabbit polyclonal of IgG isotype directed to the N-terminal region (within residues 2 to 35 of the first extracellular domain) of human HCRT receptor 1 (ab83925, Abcam) or a goat polyclonal of IgG isotype directed to the N-terminal region (amino acids 2 to 14 of the first extracellular domain) of human HCRT receptor 2 (ab65093, Abcam).

Antibody adsorption studies

Peptides (JPT Peptide Technology GmbH) for NP containing isoleucine or methionine (RELILYDKEEIRRIWRQANNG and RELILYDKEEMRRIWRQANNG) and a scrambled peptide for NP containing isoleucine (GENRRDWEQILEIRRNKAYLI) were synthesized and preincubated with sera to test for presence of cross-reactive antibodies to influenza NP peptides and HCRT receptors. Recombinant peptides comprising the extracellular domain of either HCRT receptor 1 (ATPGAQMGVPPGSREPSPVPPDYEDEFLRYLWRDYLYPKQYEWVLIAAYV, amino acids 5 to 54, Antibodies Online) or HCRT receptor 2 (MSGTKLEDSPPCRNWSSASELNETQEPPLNPTDYDDEEFLRYLWREYLHPKEYE, amino acids 1 to 55, Novus Biologicals) were purchased and used for similar peptide inhibition studies. One hundred–fold excess of blocking peptide was calculated for a 1:800 dilution of experimental sera based on the IgG concentration contained in human sera being 1.33 × 10-7 mol/ml. Adsorbed sera were then used for cell-based ELISA described above.


Human serum samples demonstrating antibody reactivity in the in-cell ELISA were used to probe (1:400) the same human HCRT receptor 1– or HCRT receptor 2–expressing cell lines that were additionally cultured on six-channel tissue culture slides (ibidi). Human serum antibodies were visualized with goat polyclonal secondary antibody to Human IgG-Fc DyLight 488 (1:1000, ab98619, Abcam). Positive control primary antibodies were as described previously. Positive control secondary antibodies are described in Supplementary Information. Images were acquired with a Plan Apochromat 63×/1.40 oil differential interference contrast objective mounted on a Zeiss LSM710 confocal microscope. Contrast and brightness of digital images from microscopy were slightly increased for easier viewing in printed images.

Bioinformatics analyses

Influenza sequences for the strains were retrieved from the NCBI’s Influenza Virus Resource querying by the appropriate strain names with duplicate sequences removed. Only proteins expected to be present (29, 30, 70) in the vaccine preparations were analyzed using the ssearch program from the FASTA package, an implementation of the Smith-Waterman alignment algorithm (71, 72). Significance of alignments was judged by the e value (73, 74).

Statistical analysis

Statistical inferences reported in this manuscript consist of P values of either tests of hypothesis or of log-linear multiple regression estimates. Data are reported as means ± SD, means with 95% confidence interval, or as percentages. To compare IgG binding to HCRT receptor 2 across subjects, the experimental readout (conducted in triplicates) was considered a binary indicator taking value 1 if a serum sample was found to be positive or zero otherwise (inter-assay variability was zero). The χ2 test for equality of proportions was used for pairwise comparisons between groups, and the Fisher’s exact test was used to assess the statistical association between the presence of antibodies to HCRT receptor 2 and vaccine-associated narcolepsy when pooling in one group all individuals not developing vaccine-associated narcolepsy. For peptide inhibition studies, measured ELISA data (inter-assay variability less than 20% for each subject across triplicate experiments) were analyzed on a logarithmic scale using multiple linear regression to assess differences across inhibitions as well as check exchangeability across replicates (adjusted r 2 > 97%). To compare presence of nuclear protein antibodies at various dilutions (undiluted, 1:160, 1:320, 1:640, and 1:1280), the experimental readout was a binary indicator taking value 1 if NPRI was greater than 30 and 0 otherwise (inter-assay variability was 0). We used χ2 test for equality of proportions within each group. In Focetria-vaccinated subjects, NPRI (conducted in quadruplicates) and SRH (duplicate plates with two readers measuring hemolysis) across time points on a subject-by-subject basis (inter-assay variability less than 30% for measured values less or equal to 1.1 log absorbance units) were first assessed by three-way analysis of variance (ANOVA) to detect potential differences across subjects, time points, and experiments (r 2 = 82%). Unpaired Student’s t test (two-tailed) was then used to assess differences between groups. Logarithms of the data were represented for each subject and time point along with their arithmetic means and 95% confidence intervals.


Materials and Methods

Fig. S1. Visualization of three-dimensional proteins and surface-exposed domains from published crystal structures of human HCRT receptor 2 and trimer NP structure from the 1/Wilson-Smith/1933 influenza strain.

Fig. S2. Trimer NP structure from the 1/Wilson-Smith/1933 influenza strain interrogated for structural chains, known T cell epitopes, sequence conservation, and functional determinants.

Fig. S3. The conservation of influenza NP epitopes identified in this study with other influenza strains from 1902 to 2013.

Fig. S4. Clinical serum samples with lower signal on ELISA confirmed to be background staining (1:400 dilution) based on microscopic patterns of reactivity.

Fig. S5. Microscopy of cell lines engineered to express HCRT receptor 1.

Fig. S6. Inhibition of IgG binding to HCRT receptor 2 by isoleucine variant of influenza NP peptide.

Fig. S7. Source data for blocking experiments done in triplicate on six subjects with narcolepsy and history of Pandemrix vaccination.

Fig. S8. Modeling of influenza NP peptide (amino acids 111 to 122) fit within the HLA-DQB1*0602 (allele strongly associated with narcolepsy).

Fig. S9. Western blots for influenza NP using monoclonal antibody lnA245 or monoclonal antibody lnA108.

Fig. S10. Extracted ion chromatogram (EIC) generated for the influenza NP peptides ELILYDKEEIR (isoleucine variant), ELILYDKEEMR (methionine variant), and IVVDYMMQKPGK (control sequence from influenza HA contained in all vaccines).

Fig. S11. Lineage of influenza vaccine reassortants and laboratory strains generated through crosses with high-yielding donor strains.

Fig. S12. Crosses of X-157 high-yielding donor strain with other historical influenza strains.

Fig. S13. The identity of the “X” contained in the influenza NP YDKEEXR sequence from NYMC X-157 (CY095712).

Fig. S14. Influenza HA–associated immune response in subjects vaccinated with Focetria in 2009.

Table S1. Percent of individuals with HLA-DQB1*0602 allele associated with narcolepsy.

Table S2. In-cell ELISA IgG binding value (absorbance 450 nm/615 nm) for sera with non–HCRT receptor 2–specific background staining.

Table S3. HLA-DQA1*0102:DQB1*0602 (allele tightly associated with narcolepsy-cataplexy) binding to 21- and 15-mer peptides.

Table S4. T cell studies published in the literature on the influenza NP motif “YDKEEIRRIWRQ” providing functional evidence for HLA class II processing of this epitope.

Table S5. Influenza A virus NP antibody inhibition test with sera from individuals with A(H1N1)pdm09 infection.

Table S6. Influenza A virus NP antibody inhibition test with sera from Finnish children not known to have narcolepsy in 2004/2005.

Table S7. Source data for influenza NP-associated immune response in subjects vaccinated with Focetria in 2009.

Movie S1. Influenza A virus NP indicating exposed surface location of NP mimic of HCRT receptors 2 and 1.

References (7679)


  1. Acknowledgments: We would like to thank the Finnish narcoleptic patients, Finnish controls, and Italian vaccinated subjects. We thank S. Spindeldreher and E. Oviedo-Orta for scientific discussions on immunological assays; P. Dormitzer, E. Settembre, and N. Groth for their critical review of the manuscript and influenza-related expertise; F. Berti for guidance on vaccine formulations; T. Miller for assistance with obtaining commercial vaccines; P. Stazi for contractual assistance with research collaborations; G. Tougas for guidance related to safety epidemiology; and E. L. Pastor and G. Piccini for assistance with cell culturing. Funding: There were no funding sources for this study. Award number S10RR027425 from the National Center for Research Resources was critical to the acquisition of the mass spectrometer used in this study for some of the data generated. Clinical trial registration: Focetria-vaccinated subjects, NCT01079273 (80). Author contributions: S.S.A. and L.S. designed the study, supervised the project, and wrote the paper. W.V. performed bioinformatics analyses. J.D. performed three-dimensional modeling of HLA-peptide binding. S.S.A. interrogated published structure of influenza nucleoprotein, estimated population coverage for HLA-DQB1*0602 allele, assembled lineages of vaccine reassortants, and collated published T cell studies on HLA class II processing of the identified nucleoprotein epitope. L.C. performed gel electrophoresis, Western blotting of influenza vaccines, and assay for antibody binding to hypocretin receptor with and without antibody adsorption. A.P. performed microscopy. A.K. and J.R. provided expertise and access to HLA-peptide binding studies. F.R. performed statistical analyses. R.R. provided organizational support and conceptual advice. V.N. provided access to retention lots of pandemic influenza vaccine and commercial seasonal influenza vaccines. I.J., A.V., O.V., and H.N. obtained ethical approval and provided access to Finnish clinical serum samples. F.L.P. obtained ethical approval and provided access to Italian clinical serum samples. E.M. and C.T. performed ELISA to detect nucleoprotein antibodies and SRH assay to detect membrane-bound influenza antigens. C.M.A. performed mass spectrometry analysis of influenza and pandemic vaccines. S.S.A., W.V., J.D., M.P., A.K., F.R., R.R., V.N., I.J., O.V., H.N., E.M., C.T., C.M.A., J.R., and L.S. analyzed and interpreted the data. All authors participated in the critical revision of the manuscript for important intellectual content and approved its submission for publication. Competing interests: S.S.A., J.D., L.C., M.P., A.P., A.K., F.R., R.R., and V.N. are employees of Novartis, and some hold stock in Novartis Pharma AG. W.V. is an employee of Atreca Inc. and holds stock in Atreca Inc. H.N. received honoraria for technical consultancy from GlaxoSmithKline and Pfizer for development of pneumococcal conjugate vaccines. S.S.A., L.S., and W.V. are inventors on a patent (“Avoiding narcolepsy risk in influenza vaccines”) with application number PCT/EP/2014/059672 with priority date May 10, 2013, and publication number WO/2014/180999 on November 13, 2014. Data and materials availability: All data pertaining to this study are published in the paper or included in the detailed Supplementary Materials.
View Abstract

Stay Connected to Science Translational Medicine

Navigate This Article