RESEARCH REPOSITORY

This is the author’s final version of the work, as accepted for publication following peer review but without the publisher’s layout or pagination.

The definitive version is available at:

http://dx.doi.org/10.1016/j.jhep.2013.08.009

Nitschke, K., Barriga, A., Schmidt, J., Timm, J., Viazov, S., Kuntzen, T., Kim, A.Y.,

Lauer, G.M., Allen, T.M., Gaudieri, S., Rauch, A., Lange, C.M., Sarrazin, C., Eiermann, T., Sidney, J., Sette, A., Thimme, R., López, D. and Neumann-

Haefelin, C. (2013) HLA-B∗27 subtype specificity determines targeting and viral evolution of a hepatitis C virus-specific CD8+ T cell epitope.

Journal of Hepatology, 60 (1). pp. 22-29.

http://researchrepository.murdoch.edu.au/id/eprint/18611/

Copyright: © © 2013 European Association for the Study of the Liver

It is posted here for your personal use. No further distribution is permitted.

Accepted Manuscript

HLA-B∗27 subtype specificity determines targeting and viral evolution of a

hepatitis C virus-specific CD8+ T-cell epitope

Katja Nitschke, Alejandro Barriga, Julia Schmidt, Jörg Timm, Sergei Viazov,

Thomas Kuntzen, Arthur Y. Kim, Georg M. Lauer, Todd M. Allen, Silvana

Gaudieri, Andri Rauch, Christian M. Lange, Christoph Sarrazin, Thomas

Eiermann, John Sidney, Alessandro Sette, Robert Thimme, Daniel López,

Christoph Neumann-Haefelin

PII: S0168-8278(13)00603-X

DOI: http://dx.doi.org/10.1016/j.jhep.2013.08.009

Reference: JHEPAT 4837

To appear in: Journal of Hepatology

Received Date: 13 February 2013

Revised Date: 29 July 2013

Accepted Date: 2 August 2013

Please cite this article as: Nitschke, K., Barriga, A., Schmidt, J., Timm, J., Viazov, S., Kuntzen, T., Kim, A.Y.,

Lauer, G.M., Allen, T.M., Gaudieri, S., Rauch, A., Lange, C.M., Sarrazin, C., Eiermann, T., Sidney, J., Sette, A.,

Thimme, R., López, D., Neumann-Haefelin, C., HLA-B∗27 subtype specificity determines targeting and viral

evolution of a hepatitis C virus-specific CD8+ T-cell epitope, Journal of Hepatology (2013), doi: http://dx.doi.org/

10.1016/j.jhep.2013.08.009

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers

we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and

review of the resulting proof before it is published in its final form. Please note that during the production process

errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

1

HLA-B*27 subtype specificity determines targeting and viral evolution of a 1

hepatitis C virus-specific CD8+ T-cell epitope 2

3

Katja Nitschke1,2, Alejandro Barriga3, Julia Schmidt1, Jörg Timm4, Sergei Viazov4, 4

Thomas Kuntzen5, Arthur Y Kim6, Georg M Lauer6, Todd M. Allen5, Silvana 5

Gaudieri7,8, Andri Rauch9, Christian M Lange10, Christoph Sarrazin10, Thomas 6

Eiermann11, John Sidney12, Alessandro Sette12, Robert Thimme1, Daniel López3, 7

Christoph Neumann-Haefelin1 8

9 1 Department of Medicine II, University of Freiburg, Freiburg, Germany 10 2 Faculty of Biology, University of Freiburg, Freiburg, Germany 11 3 Instituto de Salud Carlos III, Majadahonda, Madrid, Spain 12 4 Institute of Virology, University Hospital, University of Duisburg-Essen, Essen, 13

Germany 14 5 Ragon Institute of MGH, MIT and Harvard, Charlestown, MA, USA 15 6 Division of Infectious Diseases, MGH, Boston, MA, USA 16 7 School of Anatomy, Physiology and Human Biology, University of Western 17

Australia, Perth Australia 18 8 Institute of Immunology and Infectious Disease, Murdoch University, Perth, 19

Western Australia, Australia 20 9 Department of Infectious Diseases, Bern University Hospital and University of 21

Bern, Switzerland 22 10 Department of Medicine I, J. W. Goethe University Hospital, Frankfurt, 23

Germany 24 11 Transfusion Medicine, HLA-Laboratory, University Hospital Hamburg-25

Eppendorf, Hamburg, Germany 26 12 La Jolla Institute for Allergy and Immunology, La Jolla, CA, USA 27 28 29 Corresponding author: Christoph Neumann-Haefelin, MD 30

Department of Medicine II, University of Freiburg 31 Hugstetter Strasse 55 32 79106 Freiburg, Germany 33 Tel and Fax: +49-761-270-33530 34 email: christoph.neumann-haefelin@uniklinik-freiburg.de 35 36

37

Electronic word count: 4910; Number of figures and tables: 4 / 3 38

2

List of abbreviations in the order of appearance: HLA: human leukocyte antigen; 1

HCV: hepatitis C virus; NS5B: non-structural protein 5B; PBMC: peripheral blood 2

mononuclear cell; EBV: Epstein Barr virus; B-LCL: B-lymphoblastoid cell lines. 3

4

Conflict of interest: The authors have no competing interests. 5

6

Financial support: CNH was supported by the Deutsche Forschungsgemeinschaft 7

(DFG; Emmy Noether Program, NE 1567/1-1). CNH and RT were supported by the 8

European Union (EFRE Interreg IV Oberrhein, project A30). SV was supported by the 9

German Ministry of Education and Research (BMBF, grant no. 01 KI 1008E). DL was 10

supported by the Ministerio de Ciencia e Innovación, Spain. This project was funded, 11

in part, by Federal funds from the National Institute of Allergy and Infectious Diseases 12

(NIAID), grants R01-AI067926 (TMA) and U19 AI082630 (TMA and GML), and the 13

Swiss HIV Cohort Study, supported by the Swiss National Science Foundation (grant 14

#33CS30_134277). 15

3

Abstract 1

Background & Aims: HLA-B*27 is associated with spontaneous HCV genotype 1 2

clearance. HLA-B*27-restricted CD8+ T-cells target three NS5B epitopes. Two of 3

these epitopes are dominantly targeted in the majority of HLA-B*27+ patients. In 4

chronic infection, viral escape occurs consistently in these two epitopes. The third 5

epitope (NS5B2820) was dominantly targeted in an acutely infected patient. This was 6

in contrast, however, to the lack of recognition and viral escape in the large majority 7

of HLA-B*27+ patients. Here, we set out to determine the host factors contributing to 8

selective targeting of this epitope. 9

Methods: Four-digit HLA class I typing and viral sequence analyses were performed 10

in 78 HLA-B*27+ patients with chronic HCV genotype 1 infection. CD8+ T-cell 11

analyses were performed in a subset of patients. In addition, HLA/peptide affinity was 12

compared for HLA-B*27:02 and 05. 13

Results: The NS5B2820 epitope is only restricted by the HLA-B*27 subtype HLA-14

B*27:02 (that is frequent in Mediterranean populations), but not by the prototype 15

HLA-B*27 subtype B*27:05. Indeed, the epitope is very dominant in HLA-B*27:02+ 16

patients and is associated with viral escape mutations at the anchor position for HLA-17

binding in 12 out of 13 HLA-B*27:02+ chronically infected patients. 18

Conclusions: The NS5B2820 epitope is immunodominant in the context of HLA-19

B*27:02, but is not restricted by other HLA-B*27 subtypes. This finding suggests an 20

important role of HLA subtypes in the restriction of HCV-specific CD8+ responses. 21

With minor HLA subtypes covering up to 39% of specific populations, these findings 22

may have important implications for the selection of epitopes for global vaccines. 23

(250/250) 24

4

Keywords: Hepatitis C virus; CD8+ T-cell response; HLA subtypes; 1

HLA-B*27; vaccine design 2

5

Introduction 1

After infection with hepatitis C virus (HCV), only a minority of patients can clear the 2

virus spontaneously, while >70% of patients develop a chronic infection. Virus-3

specific CD8+ T-cells recognize HCV epitopes that are presented by HLA class I 4

molecules on antigen-presenting cells as well as infected hepatocytes, and are 5

essential effector cells in viral clearance[1]. Specific HLA class I types, such as HLA-6

B*27 and B*57, have a protective role in HCV infection, since they are associated 7

with high rates of spontaneous viral clearance[2, 3]. We could link the protective 8

effect of HLA-B*27 to dominant HLA-B*27 restricted epitopes located in the RNA-9

dependent RNA polymerase of HCV, the NS5B protein[4]. Indeed, one of these 10

epitopes, NS5B2841 (amino acid sequence ARMILMTHF), is very immunodominant, 11

since it is targeted in nearly all HLA-B*27+ patients with acute or resolved infection. 12

Nearly all HLA-B*27+ patients with chronic infection display multiple viral escape 13

mutations in this epitope. Indeed, we could demonstrate that viral escape from this 14

epitope requires several mutations in order to bypass recognition by the CD8+ T-cell 15

response without deteriorating viral replication capacity. Most likely, this high barrier 16

to viral escape explains why most HLA-B*27+ patients successfully clear the virus in 17

the acute phase of infection, before HCV can escape from the HLA-B*27 restricted 18

CD8+ T-cell response[5]. A second HLA-B*27 restricted CD8+ T-cell epitope located 19

in NS5B (NS5B2936, amino acid sequence GRAAICGKY) is subdominantly 20

recognized and selects for viral escape mutations in approx. 50% of chronically HCV-21

infected HLA-B*27+ patients[6]. 22

Next to these two HLA-B*27 restricted epitopes, we initially identified a third HLA-23

B*27 restricted epitope located in NS5B (NS5B2820, amino acid sequence 24

ARHTPVNSW) in the same acutely HCV-infected patient of Sicilian (southern Italian) 25

origin[4]. This epitope was, however, not targeted in any patient from an Irish cohort 26

6

of HLA-B*27+ patients with recovered HCV infection, who were infected with a viral 1

inoculum that contained the respective epitope sequence. Based on a more diverse 2

expression of HLA-B*27 subtypes in southern Italy (subtypes B*27:05, B*27:02, and 3

B*27:09) compared to Ireland (HLA-B*27:05 only)[7], we hypothesized that the 4

differential targeting of this HLA-B*27 restricted CD8+ T-cell epitope may be due to a 5

limited restriction by certain HLA-B*27 subtypes. Indeed, here we demonstrate that 6

the epitope NS5B2820 is immunodominant exclusively in the context of HLA-B*27:02. 7

This HLA-B*27 subtype is predominant (>50%) in the Middle East and Northern 8

Africa but is only present in 5-10% of HLA-B*27+ Euro-Caucasians[7]. People that 9

express the ancestral and most common HLA-B*27 subtype HLA-B*27:05, however, 10

do not recognize this epitope. These findings have important implications for vaccine 11

design, since epitope targeting and immunodominance may be impacted not only by 12

individual HLA types, but also by HLA subtypes. 13

7

Patients and methods 1

Patient cohorts 2

Patients with chronic HCV genotype 1 infection from four different cohorts (Table 3

1)[8-10] were tested for HLA-B*27 expression by standard methods using molecular 4

HLA typing or HLA serotyping procedures (n=1356). A subset of patients (n=165) 5

was tested with an anti-HLA-B*27 flow cytometry antibody (One lambda, Canoga 6

Park, CA, USA) following the manufacturer’s instructions. HLA-B*27+ patients 7

identified by this approach were subjected to 4-digit molecular HLA typing. 8

Autologous viral sequences corresponding to the NS5B region that contains the 9

NS5B2820 epitope were determined as described previously[8-10]. PBMC from five 10

HLA-B*27:02+ patients with chronic HCV genotype 1 infection, PBMC from an HLA-11

B*27:02+ patient with spontaneously resolved HCV infection (anti-HCV positive, 12

HCV-RNA negative), PBMC from an HLA-B*27:02+ patient of Sicilian origin with 13

acute resolving HCV genotype 1a infection[4] (all presenting to the Freiburg 14

University Medical Center), as well as PBMC from an HLA-B*27:02+ patient with 15

acute resolving HCV genotype 1a infection recruited at the Massachusetts General 16

Hospital were isolated as described previously[4]. For control experiments, PBMC 17

from an HLA-B*27:05+ patient with acute resolving HCV genotype 1a infection, 18

spontaneously resolved infection, and chronic HCV genotype 1a infection, 19

respectively, recruited at the Freiburg University Medical Center were used. A cohort 20

of six HLA-B*27+ Irish women with resolved HCV genotype 1b infection has been 21

described previously[4]. All patients gave informed consent; the study was performed 22

in agreement with federal guidelines and the local ethics committees. 23

24

Generation of peptide-specific cell lines, intracellular interferon-gamma staining, and 25

HLA-B*27/peptide stability assays are described in the supplementary methods. 26

8

Results 1

The NS5B2820 epitope is vigorously recognized in a Sicilian HLA-B*27+ patient 2

with acute HCV infection, but not in Irish and German HLA-B*27+ patients. 3

We previously identified three HLA-B*27 restricted HCV NS5B-specific CD8+ T-cell 4

epitopes in a Sicilian patient with acute HCV genotype 1a infection[4]. Two of these 5

epitopes, NS5B2841 (amino acid sequence ARMILMTHF) and NS5B2936 (amino acid 6

sequence GRAAICGKY) are frequently targeted in HLA-B*27+ patients with resolved 7

or chronic HCV genotype 1 infection. In addition, viral escape mutations are selected 8

in these two epitopes in >90% and ~50%, respectively, of HLA-B*27+ patients with 9

chronic HCV genotype 1 infection[6, 11]. The third HLA-B*27 restricted epitope, 10

NS5B2820 (amino acid sequence ARHTPVNSW), displayed the strongest interferon-11

gamma response in the acutely infected patient (Fig. 1A). Strikingly, however, we did 12

not observe any CD8+ T-cell response against this epitope in 6 Irish HLA-B*27+ 13

patients with resolved HCV genotype 1b infection (Fig. 1B). Importantly, these 14

patients were infected with a single source HCV strain harbouring the wild-type 15

NS5B2820 epitope sequence, indicating that sequence mismatches in the epitope 16

region cannot explain the lack of recognition. In addition, we were also unable to 17

detect any NS5B2820-specific CD8+ T-cell response in 8 HLA-B*27+ German patients 18

with chronic genotype 1b infection (data not shown). Sequence analyses revealed 19

that 6 of the 8 HLA-B*27+ patients had wild-type viral sequences in this epitope 20

region, indicating the absence of substantial immune pressure (Fig. 1C). 21

Irish HLA-B*27+ individuals primarily express the ancestral and prototype HLA-B*27 22

subtype B*27:05. This subtype is also predominant in German cohorts (reported at 23

>90%). In contrast, in Mediterranean people such as Sicilians the subtype HLA-24

B*27:02 is also frequently expressed. We thus speculated that the expression of 25

different HLA-B*27 subtypes may impact the recognition of the NS5B2820 epitope. 26

9

1

Epitope NS5B2820 is restricted by the HLA-B*27 subtype B*27:02 and is not 2

targeted in the context of the prototype subtype B*27:05. 3

Interestingly, HLA-B*27 subtype analysis showed that the Sicilian patient with acute 4

HCV infection is positive for HLA-B*27:02 and not the prototype B*27:05. In order to 5

test the hypothesis that the expression of HLA-B*27:02 may be responsible for the 6

targeting and unusual dominance of the NS5B2820 epitope, we analyzed the CD8+ T-7

cell response to the three HLA-B*27-restrcited NS5B epitopes in a second HLA-8

B*27:02+ patient with acute resolving HCV genotype 1a infection. Importantly, this 9

patient showed a similar pattern of recognition of the three HLA-B*27-restricted NS5B 10

epitopes, with the NS5B2820 epitope being the dominant target (supplementary Figure 11

1). Next, we tested the ability of HLA-B*27:02 or B*27:05 expressing antigen-12

presenting cells to present the NS5B2820 peptide to an epitope-specific T-cell line 13

derived from the Sicilian patient. When we loaded HLA-B*27:02 expressing EBV-14

transformed B-cell lines with peptide and co-cultured these cells with an NS5B2820-15

specific T-cell line, we indeed observed a strong interferon-gamma response (Fig. 2A, 16

upper panel). When we used HLA-B*27:05 expressing EBV-transformed B-cell lines, 17

however, we did not observe any interferon-gamma production (Fig. 2A, lower panel), 18

even when loading B-cells with 10-fold peptide concentrations (data not shown). The 19

same result was obtained when the B-cell lines were not externally loaded with the 20

epitope peptide, but an endogenous antigen processing experiment was performed 21

using HCV-expressing vaccinia constructs (Fig. 2A). Equivalent results were also 22

obtained when NS5B2820-specific T-cell lines were used that had been generated 23

from HLA-B*27:02+ patients with spontaneously resolved HCV infection or chronic 24

HCV genotype 1a infection, respectively (supplementary Figure 2). These data 25

10

clearly demonstrate that epitope NS5B2820 is restricted by HLA-B*27:02, but not by 1

the HLA-B*27 prototype B*27:05. 2

We also stimulated PBMC from three HLA-B*27:05+ patients with acute resolving, 3

spontaneously resolved, and chronic HCV genotype 1a infection, respectively, with 4

the NS5B2820 epitope peptide for two weeks. However, we were unable to detect 5

epitope-specific CD8+ T-cell responses even when using 10-fold peptide 6

concentrations in cell culture and/ or read-out procedures, further indicating that HLA-7

B*27:05+ patients do not target this epitope (data not shown). 8

HLA-B*27:02 is the only HLA-B*27 subtype that has been shown to accept 9

tryptophan (W) at the C-terminus of antigen ligands[12]. The NS5B2820 epitope has 10

indeed a tryptophan at its C-terminus. We therefore hypothesized that the NS5B2820 11

epitope peptide may bind specifically to HLA-B*27:02 molecules, but not or only 12

weakly to HLA-B*27:05 molecules. HLA binding analyses indeed showed that the 13

NS5B2820 peptide has a high affinity for HLA-B*27:02, while it displayed an inferior 14

binding affinity to HLA-B*27:05, when compared to other known HLA-B*27 restricted 15

viral CD8+ T-cell epitopes, including the HCV-specific NS5B2841 and NS5B2936 16

epitopes, and also the HIV-specific gag p24 epitope KK10 (Fig. 2B). The relatively 17

weak binding of NS5B2820 to HLA-B*27:05 was confirmed in a competition assay 18

using radiolabeled standard peptide (Table 2). In silico modeling of the binding of the 19

NS5B2820 peptide in the HLA-B*27 peptide binding groove[13, 14] indicated that the 20

C-terminal tryptophan of NS5B2820 may indeed better embed in the F pocket of HLA-21

B*27:02 compared to the F pocket of HLA-B*27:05 (Fig. 3). Of note, in addition to the 22

relatively small differences observed by in silico modeling, additional conformational 23

changes may further impact NS5B2820 binding to the different HLA-B*27 subtype 24

molecules[15]. Taken together, these in vitro and in silico results further support the 25

11

conclusion that the NS5B2820 epitope is only presented by HLA-B*27:02 and not by 1

HLA-B*27:05. 2

3

Escape mutations in the NS5B2820 epitope are reproducibly selected in HLA-4

B*27:02+ patients. 5

Since we demonstrated the epitope NS5B2820 to be specifically restricted by the 6

subtype B*27:02, we aimed to analyze this epitope in a larger cohort of HLA-7

B*27:02+ patients. In order to include a sufficient number of HLA-B*27:02+ patients, 8

we combined data from four large cohorts of patients with chronic HCV genotype 1 9

infection, including a total of 1521 patients. 91 (6.0%) of these patients were HLA-10

B*27+ (Table 1). In a total of 78 HLA-B*27+ patients, 4-digit typing of HLA-B*27 as 11

well as viral sequence data for the NS5B region corresponding to the NS5B2820 12

epitope could be obtained. Importantly, in 12 out of 13 (92.3%) HLA-B*27:02+ 13

patients, we observed a sequence variation at amino acid position 2 of the epitope, 14

with a substitution of arginine (R) by lysine (K) (Table 3). In contrast, patients that 15

expressed the HLA-B*27 prototype B*27:05 rarely displayed sequence variations 16

compared to the consensus sequence (6/ 59 patients, 10.2%; P<0.0001, Fisher’s 17

exact test). Only a single HLA-B*27:05+ patient (1.7%) displayed a mutation at 18

position 2 of the epitope, suggesting that the few mutations occurring in HLA-19

B*27:05+ patients were not driven by NS5B2820-specific CD8+ T-cells. Patients 20

expressing rare HLA-B*27 subtypes such as B*27:03, B*27:06, B*27:10, and B*27:14 21

showed similar results compared to HLA-B*27:05+ patients, with only 1 of 6 patients 22

displaying a variant viral sequence. 23

Our findings indicate that NS5B2820-specific CD8+ T-cells are strong drivers of viral 24

evolution in HLA-B*27:02+ patients, resulting in the viral escape mutation R2821K at 25

amino acid position 2 of the epitope. Arginine (R) at this position functions as the 26

12

main HLA-B*27 binding anchor, thus, mutational escape at this position is likely to 1

have a strong impact on epitope presentation. Indeed, the R2821K mutant peptide 2

displayed only marginal binding to the HLA-B*27:02 molecule as compared to the 3

wild-type peptide (Fig. 4A). We next tested recognition of the consensus as well as 4

variant peptide by epitope-specific cell lines generated from the patient with acute 5

infection. As shown in Fig. 4B (upper panel), the R2821K mutation completely 6

abolished recognition of the epitope peptide, demonstrating that this viral variation 7

does indeed represent a viral escape mutation. 8

We also tested PBMC from five HLA-B*27:02+ patients with chronic HCV genotype 1 9

infection and PBMC from one HLA-B*27:02+ patient with spontaneously resolved 10

HCV infection for NS5B2820-specific CD8+ T-cell responses after two weeks of 11

peptide-specific cell culture. In one out of the five chronically infected patients as well 12

as in the patient with resolved infection, we were successful in generating epitope-13

specific CD8+ T-cell lines with strong interferon-gamma production in response to the 14

consensus, but not the R2821K variant peptide (Fig. 4B, middle and lower panel). It 15

is important to note, that the frequency of epitope-specific responses observed here 16

(one response in five chronically infected patients) is similar to that observed for other 17

immunodominant HCV-specific CD8+ T-cell epitopes in the setting of chronic 18

infection[16]. We further confirmed that mutation R2821K represents an escape 19

variant by performing peptide titration assays for the NS5B2820-specific T-cell lines 20

(Fig. 4C). Of note, the partial cross-recognition of the mutant peptide in the 21

chronically infected patient in this assay (Fig. 4C) in contrast to the lack of cross-22

recognition after loading of the mutant peptide on HLA-B*27:02+ B-cells with 23

subsequent washing of the B-cells prior to co-culture with the epitope-specific T-cell 24

line (Fig. 4B) indicates that the variant peptide binds to HLA-B*27:02 with low affinity, 25

resulting in a high drop-off rate. 26

13

It has been described that certain viral mutant peptides may have antagonistic 1

capacities and compete with their respective wild-type peptide for HLA and/or TCR 2

binding[17]. Thus, we performed competition assays titrating increasing doses of the 3

mutant peptide in addition to the wild-type peptide, however, we observed no 4

antagonistic effect (Fig. 4D). 5

In summary, we clearly show that the epitope NS5B2820 is specifically restricted by 6

HLA-B*27:02, but not by the prototype HLA-B*27 subtype B*27:05. In the context of 7

HLA-B*27:02, viral escape mutations are selected in the vast majority of patients. 8

These results suggest that HLA class I subtypes have an important impact in 9

directing the HCV-specific CD8+ T-cell response. 10

14

Discussion 1

We have previously identified three HLA-B*27 restricted CD8+ T-cell epitopes 2

located in the HCV NS5B protein[4]. Two of these epitopes, NS5B2841 and NS5B2936, 3

were frequently targeted in HLA-B*27+ patients with HCV genotype 1 infection and 4

also reproducibly induced viral escape. In the current study, we analyzed a third HLA-5

B*27 restricted epitope in NS5B. This epitope seemed to be a minor CD8+ T-cell 6

target, since we could not detect CD8+ T-cell responses against this epitope in any of 7

14 HLA-B*27+ patients with resolved or chronic HCV infection. This finding was 8

somewhat in discrepancy with the very strong epitope-specific CD8+ T-cell response 9

observed in a patient with acute HCV infection. HCV genotype or subtype-diversity 10

could not explain this finding, since the epitope is conserved in geno-/subtypes 1a, 11

1b, and also 3a (supplementary table 1). When we analyzed the reason for these 12

discrepant observations, we discovered that the epitope NS5B2820 is only presented 13

by the HLA-B*27 subtype B*27:02, but not by other HLA-B*27 subtypes, including the 14

prototype HLA-B*27:05. The HLA-B*27:02 subtype-specificity of this epitope may be 15

largely mediated by a superior binding of the epitope peptide to HLA-B*27:02 and an 16

only weak binding to HLA-B*27:05. However, other mechanisms such as naïve 17

precursor frequencies may also have an additional impact[18]. 18

We found that NS5B2820 is a dominant epitope in HLA-B*27:02 patients, since the 19

vast majority (12 of 13) of HLA-B*27:02+ patients with chronic HCV infection 20

displayed a viral escape mutation at the main HLA binding anchor of the epitope. 21

Since this epitope is only restricted by the relatively rare subtype HLA-B*27:02, but 22

not by the common ‘ancestral’ B*27:05, it is by mistake considered not to be 23

immunodominant if HLA-B*27 subtypes are not taken into account. Thus, host 24

genetic diversity at the HLA-B*27 subtype level has an important impact on epitope 25

recognition and may influence protective immune responses in a similar way as has 26

15

been reported for HCV genotypes that do or do not contain specific protective 1

epitopes[11]. Of note, recognition as well as viral escape in the HLA-B*27-restricted 2

epitopes NS5B2841 and NS5B2936 do not differ between HLA-B*27:05 and HLA-3

B*27:02 (data not shown), indicating that a differential targeting of epitopes by these 4

two HLA-B*27 subtypes is not mandatory. 5

Due to a lack of large studies analyzing the impact of HLA class I alleles at high 6

resolution (4-digit HLA subtypes), it is not clear if HLA-B*27:02 has a different 7

(stronger or weaker) protective role in HCV or HIV infection compared to HLA-8

B*27:05. In this context it is important to point out, however, that the protective effect 9

of HLA-B*27 in HCV infection has been demonstrated most clearly in an Irish 10

cohort[3]. The Irish population nearly exclusively expresses the HLA-B*27 subtype 11

B*27:05, and we previously linked the protective effect of HLA-B*27 in that cohort to 12

the dominant targeting of the NS5B2841 epitope[4]. Thus, the HLA-B*27:02-specific 13

NS5B2820 epitope is most likely not involved in conferring protection in the Irish cohort. 14

Of note, differential effects on the clinical course of infection have been described for 15

other HLA subtypes. For example, the HLA subtype B*35:01, but not B*35:02/03 is 16

hazardous in HIV infection. Even more strikingly, B*58:01 is protective, but B*58:02 is 17

detrimental in HIV infection[19, 20]. Similarly, HLA-B*38 is hazardous in HCV 18

infection, while the highly related B*39 is protective in HCV infection[21]. Thus, very 19

small differences in closely related HLA class I molecules can have relevant effects 20

on both, CD8+ T-cell recognition as well as immune control. 21

Our finding may have important implications for HCV vaccine design. Indeed, 22

identification of HCV-specific CD8+ T-cell epitopes has been mainly performed in 23

North-American and European Caucasians. HLA class I subtypes were not 24

considered in most of these studies, and thus little is known about the exact 25

restriction of most HCV-specific CD8+ T-cell epitopes. HLA prototype subtypes are 26

16

defined as those HLA subtypes most prevalent in a certain population. For example, 1

HLA-A*02:01 is present in >90% of HLA-A*02+ Euro-Caucasians, while all other 2

HLA-A*02 subtypes are rare in this population. Thus, A*02:01 is the HLA-A*02 3

prototype. These prototypes do cover 96% and 87% of all HLA-A and HLA-B 4

subtypes expressed e.g. in US Caucasians[22]. When these prototypes defined in 5

US Caucasians are used to study ethnically distinct population, e.g. African-6

Americans, Asians or Hispanics living in the US, coverage rates drop to 64-80% for 7

HLA-A subtypes and 61-63% for HLA-B subtypes. For example, HLA-A*02:01 is 8

expressed by less than 50% of HLA-A*02+ individuals of Afro-American or Asian 9

origin in the US. In consequence, analysis of ethnically mixed populations using e.g. 10

only HLA-A*02:01 and HLA-B*27:05 restricted CD8+ epitopes may not be 11

appropriate. 12

Since our patient cohort consisted mainly (>90%) of patients of Caucasian ethnicity, 13

we could not perform a comprehensive analysis of the impact of HLA subtypes on 14

HCV-specific CD8+ T-cell responses in HCV infection. We performed a preliminary 15

analysis, however, in 287 patients with HCV genotype 1a infection, known 4-digit 16

HLA types and known autologous full-genome HCV sequences[9]. In this cohort, > 5 17

patients expressed non-prototype alleles of the HLA-B types B*14, B*15, B*35, B*40, 18

and B*44, respectively. Of these HLA-B types, only B*15, B*35 and B*40 displayed 19

HLA associations within known CD8+ T-cell epitopes. Interestingly, in four of the five 20

epitopes, we found evidence for HLA subtype-specific selection of viral escape 21

mutations (supplementary Fig. 3), indicating that indeed subtypes of several HLA-B 22

alleles may play an important role in restriction of HCV-specific CD8+ T-cell 23

responses. 24

While these findings clearly need to be confirmed in additional studies performed in 25

cohorts with differential ethnical backgrounds, they indicate that vaccine candidates 26

17

that have been designed considering HLA class I prototypes only may exclude 1

protective CD8+ T-cell epitopes restricted by minor HLA subtypes. In this scenario, it 2

is important to note that rare HLA types and subtypes may predominantly contribute 3

to viral control[23]. As a consequence, vaccine studies should aim at a broad 4

coverage of potential CD8+ epitopes and also include comprehensive analyses using 5

e.g. overlapping peptides in their read-outs in order to avoid a bias towards prototype 6

HLA subtypes. 7

In conclusion, we demonstrate here that HLA class I subtypes may have an important 8

role in restriction of HCV-specific CD8+ T-cell epitopes. This is highlighted in our 9

study of an HLA-B*27 restricted CD8+ T-cell epitope that is immunodominant in the 10

background of HLA-B*27:02, but not presented by other HLA-B*27 subtypes, 11

including the prototype HLA-B*27:05. 12

18

Acknowledgements 1

We would like to thank all patients who donated blood for this study. We also 2

appreciate the help of all physicians and study nurses who included patients into the 3

cohort. 4

5

19

Figure Legends 1

Figure 1: HLA-B*27 restricted epitope NS5B2820 displays a vigorous CD8+ T-cell 2

response in an HLA-B*27+ patient with acute HCV infection, but is not targeted 3

in HLA-B*27+ patients with resolved or chronic infection. A. CD8+ T-cells from 4

an HLA-B*27+ patient of Sicilian origin with acute resolving HCV genotype 1a 5

infection were tested for interferon-gamma production in response to three HLA-B*27 6

restricted epitopes located in NS5B. A sample without peptide stimulation and a 7

sample stimulated with PMA and ionomycin were included as negative and positive 8

controls, respectively. The percentage of interferon-gamma producing CD8+ T-cells 9

is indicated. B. PBMC from 6 Irish individuals with spontaneously resolved HCV 10

genotype 1b infection were stimulated for 14 days in the presence of the NS5B2820 or 11

NS5B2841 epitope peptides, respectively, and tested for interferon-gamma production 12

in response to the respective peptide. C. Autologous viral sequences corresponding 13

to the NS5B2820 epitope have been determined from 8 HLA-B*27+ patients as well as 14

9 HLA-B*27- control patients with chronic HCV genotype 1b infection. Each line 15

represents the autologous sequence from one patient; a dot indicates identity to 16

consensus. Previously published sequences corresponding to the NS5B2841 epitope 17

are shown for comparison[4]. 18

19

Figure 2: Epitope NS5B2820 is restricted by subtype HLA-B*27:02 only, and is 20

not targeted in the context of the HLA-B*27 ‘prototype’ HLA-B*27:05. A. An 21

NS5B2820 epitope-specific CD8+ T-cell line was generated from the HLA-B*27+ 22

patient with acute resolving HCV genotype 1a infection by two weeks of peptide 23

stimulation. The T-cell line was then co-cultured with EBV-transformed B-cells that 24

expressed HLA-B*27:02 or HLA-B*27:05 and were or were not loaded with peptide 25

NS5B2820. As read-out, intracellular interferon-gamma staining was performed (two 26

20

panels on the left). Alternatively to peptide loading, the B-cell lines were infected with 1

vaccinia viruses that did or did not express HCV amino acids 827-3011, including the 2

NS5B protein (two panels on the right). B. HLA-B*27:02 and B*27:05 binding 3

affinities were assessed for the NS5B2820 as well as other known HLA-B*27-restricted 4

viral CD8+ T-cell epitopes. Synthetic peptides corresponding to HIV gag p24262 5

(KK10), NS5B2820, NS5B2841, NS5B2936, and Flu NP as well as CMV pp65 as negative 6

control were titrated bound to HLA-B*27:02 and B*27:05, respectively, on RMA-S 7

TAP-deficient cells, and stabilization was measured by flow cytometry. Data are 8

means of 3-4 independent experiments. 9

10

Figure 3: NS5B2820 epitope peptide modeled into the binding groove of the HLA-11

B*27:02 molecule (left) and the HLA-B*27:05 molecule (right). Miyazawa and 12

Jernigan (MJ) energy levels were calculated for the interaction between the C-13

terminal tryptophan of the epitope peptide and its modeled interaction partners in the 14

F pocket of the HLA molecules. Analyses were performed using the MODPROPEP 15

web server and the HLA-B*27:05 (PDB ID: 1OGT) structure. A negative MJ energy 16

favours stable binding. 17

18

Figure 4: The R2821K mutation mediates viral escape from the NS5B2820 19

epitope-specific CD8+ T-cell response. A. HLA-B*27:02 binding of the wild-type 20

and R2821K mutant NS5B2820 epitope peptides were compared in an assay 21

equivalent to Fig. 2B. B. NS5B2820 epitope-specific T-cell lines generated from HLA-22

B*27:02+ patients with acute (upper panel), chronic (middle panel), or resolved 23

(lower panel) HCV genotype 1a infection were co-cultured with an HLA-B*27:02+ 24

EBV-transformed B-cell line that has been loaded with the consensus NS5B2820 25

epitope peptide or the variant epitope peptide containing the R2821K mutation. 26

21

Intracellular interferon-gamma staining was performed as read-out. C. NS5B2820-1

specific T-cell lines from an HLA-B*27:02+ patient with chronic HCV genotype 1a 2

infection (upper panel) or resolved HCV infection (lower panel) were stimulated with 3

the NS5B2820 wild-type epitope peptide or the R2821K mutant peptide in increasing 4

concentrations as indicated. D. The R2821K variant peptide does not act as an 5

antagonist for the wild-type peptide. NS5B2820-specific T-cell lines from an HLA-6

B*27:02+ patient with chronic HCV genotype 1a infection (filled bars) and 7

spontaneously resolved HCV infection (white bars), respectively, were co-cultured 8

with HLA-B*27:02+ B-cell lines that had been loaded with wild-type or variant peptide 9

in different concentrations as indicated. 10

22

References 1

[1] Walker CM. Adaptive immunity to the hepatitis C virus. Adv Virus Res 2010;78:43-2 86. 3 [2] Kuniholm MH, Kovacs A, Gao X, Xue X, Marti D, Thio CL, et al. Specific human 4 leukocyte antigen class I and II alleles associated with hepatitis C virus viremia. Hepatology 5 2010;51:1514-1522. 6 [3] McKiernan SM, Hagan R, Curry M, McDonald GS, Kelly A, Nolan N, et al. Distinct 7 MHC class I and II alleles are associated with hepatitis C viral clearance, originating from a 8 single source. Hepatology 2004;40:108-114. 9 [4] Neumann-Haefelin C, McKiernan S, Ward S, Viazov S, Spangenberg HC, Killinger T, 10 et al. Dominant influence of an HLA-B27 restricted CD8+ T cell response in mediating HCV 11 clearance and evolution. Hepatology 2006;43:563-572. 12 [5] Dazert E, Neumann-Haefelin C, Bressanelli S, Fitzmaurice K, Kort J, Timm J, et al. 13 Loss of viral fitness and cross-recognition by CD8+ T cells limit HCV escape from a 14 protective HLA-B27-restricted human immune response. J Clin Invest 2009;119:376-386. 15 [6] Neumann-Haefelin C, Oniangue-Ndza C, Kuntzen T, Schmidt J, Nitschke K, Sidney J, 16 et al. Human leukocyte antigen B27 selects for rare escape mutations that significantly impair 17 hepatitis C virus replication and require compensatory mutations. Hepatology 2011;54:1157-18 1166. 19 [7] Blanco-Gelaz MA, Lopez-Vazquez A, Garcia-Fernandez S, Martinez-Borra J, 20 Gonzalez S, Lopez-Larrea C. Genetic variability, molecular evolution, and geographic 21 diversity of HLA-B27. Hum Immunol 2001;62:1042-1050. 22 [8] Gaudieri S, Rauch A, Pfafferott K, Barnes E, Cheng W, McCaughan G, et al. Hepatitis 23 C virus drug resistance and immune-driven adaptations: relevance to new antiviral therapy. 24 Hepatology 2009;49:1069-1082. 25 [9] Kuntzen T, Timm J, Berical A, Lennon N, Berlin AM, Young SK, et al. Naturally 26 occurring dominant resistance mutations to hepatitis C virus protease and polymerase 27 inhibitors in treatment-naive patients. Hepatology 2008;48:1769-1778. 28 [10] Lange CM, Roomp K, Dragan A, Nattermann J, Michalk M, Spengler U, et al. HLA 29 class I allele associations with HCV genetic variants in patients with chronic HCV genotypes 30 1a or 1b infection. J Hepatol 2010;53:1022-1028. 31 [11] Neumann-Haefelin C, Timm J, Schmidt J, Kersting N, Fitzmaurice K, Oniangue-Ndza 32 C, et al. Protective effect of human leukocyte antigen B27 in hepatitis C virus infection 33 requires the presence of a genotype-specific immunodominant CD8+ T-cell epitope. 34 Hepatology 2010;51:54-62. 35 [12] Rotzschke O, Falk K, Stevanovic S, Gnau V, Jung G, Rammensee HG. Dominant 36 aromatic/aliphatic C-terminal anchor in HLA-B*2702 and B*2705 peptide motifs. 37 Immunogenetics 1994;39:74-77. 38 [13] Kumar N, Mohanty D. MODPROPEP: a program for knowledge-based modeling of 39 protein-peptide complexes. Nucleic acids research 2007;35:W549-555. 40 [14] Sundaramurthi JC, Ramanathan VD, Hanna LE. HLA-B*27:05-specific cytotoxic T 41 lymphocyte epitopes in Indian HIV type 1C. AIDS research and human retroviruses 42 2013;29:47-53. 43 [15] Infantes S, Lorente E, Barnea E, Beer I, Barriga A, Lasala F, et al. Natural HLA-44 B*2705 protein ligands with glutamine as anchor motif: implications for HLA-B27 45 association with spondyloarthropathy. J Biol Chem 2013;288:10882-10889. 46 [16] Lauer GM, Barnes E, Lucas M, Timm J, Ouchi K, Kim AY, et al. High resolution 47 analysis of cellular immune responses in resolved and persistent hepatitis C virus infection. 48 Gastroenterology 2004;127:924-936. 49

23

[17] Bertoletti A, Sette A, Chisari FV, Penna A, Levrero M, De Carli M, et al. Natural 1 variants of cytotoxic epitopes are T-cell receptor antagonists for antiviral cytotoxic T cells. 2 Nature 1994;369:407-410. 3 [18] Schmidt J, Neumann-Haefelin C, Altay T, Gostick E, Price DA, 3 VL, Hubert E. 4 Blum1, Robert Thimme1. Immunodominance of HLA-A2 restricted hepatitis C virus-specific 5 CD8+ T cell responses is linked to naïve precursor frequency. J Virol 2011;85:5232-5236. 6 [19] Gao X, Nelson GW, Karacki P, Martin MP, Phair J, Kaslow R, et al. Effect of a single 7 amino acid change in MHC class I molecules on the rate of progression to AIDS. N Engl J 8 Med 2001;344:1668-1675. 9 [20] Kiepiela P, Leslie AJ, Honeyborne I, Ramduth D, Thobakgale C, Chetty S, et al. 10 Dominant influence of HLA-B in mediating the potential co-evolution of HIV and HLA. 11 Nature 2004;432:769-775. 12 [21] Hraber P, Kuiken C, Yusim K. Evidence for human leukocyte antigen heterozygote 13 advantage against hepatitis C virus infection. Hepatology 2007;46:1713-1721. 14 [22] Maiers M, Gragert L, Klitz W. High-resolution HLA alleles and haplotypes in the 15 United States population. Hum Immunol 2007;68:779-788. 16 [23] Trachtenberg E, Korber B, Sollars C, Kepler TB, Hraber PT, Hayes E, et al. 17 Advantage of rare HLA supertype in HIV disease progression. Nat Med 2003;9:928-935. 18 19 20

24

CD8

IFNγ

without peptide

PMA/ ionomycin

NS5B2820ARHTPVNSW

NS5B2841ARMILMTHF

NS5B2936GRAAICGKY

NS5B2820ARHTPVNSW

0

10

20

30

40

50

%IF

Nγ+

/CD

8+

NS5B2841ARMILMTHF

Fig. 1

A

B C

100 101 102 103 104100

101

102

103

104

3.78%

100 101 102 103 104100

101

102

103

104

0.25%

100 101 102 103 104100

101

102

103

104

0.52%

100 101 102 103 104100

101

102

103

104

74.48%

100 101 102 103 104100

101

102

103

104

0%

Consensus A R H T P V N S W A R M I L M T H F. . . . . . . . . V . . V . L . . .. K . . . . . . . . . . . . L P . .. K . . . . . . . . . . V . . . . .. . . . . . . . . V . . . . L P . .. . . . . . . . . V . . V . . . . .. . . . . . . . . V . . V . . . . .. . . . . . . . . . . . . . . . . .. . . . . . . . . V . . V . . S . .. . . . . . . . . . . . . . . . . .. . . . . . . . . V . . . . L . . .. . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . .. . . . . . . . . . . . V . . . . .. . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . .

HLA-B*27+

HLA-B*27-

NS5B2820 NS5B2841

1

25

Fig. 2

A without peptide + NS5B2820

B*27:02+ B-LCL

B*27:05+ B-LCL

control vaccinia virus

HCV 827-3011

CD8

IFNγ

0 102 103 104 105

0

102

103

104

105 0%

0 102 103 104 105

0

102

103

104

105 49.4%

0 102 103 104 105

0

102

103

104

105 0%

0 102 103 104 105

0

102

103

104

105 26.5%

0 102 103 104 105

0

102

103

104

105 0%

0 102 103 104 105

0

102

103

104

105 0%

0 102 103 104 105

0

102

103

104

105 0%

0 102 103 104 105

0

102

103

104

105 0.22%

B

Peptide [µM]

Flu

ore

scen

ce in

dex

1.0

1.5

2.0

Peptide [µM]0.1 1 10 100

gag p24-262

NS5B-2820

NS5B-2841

NS5B-2936

FLU-NP

CMV-pp65

HLA-B*27:02 HLA-B*27:05

1.0

1.5

2.0

0.1 1 10 1000,9

1

1,1

1,2

1,3

1,4

1,5

1,6

1,7

1,8

1,9

2

0,1 0,5 1 1,5 10 10,5 100 100,5

0,9

1

1,1

1,2

1,3

1,4

1,5

1,6

1,7

1,8

1,9

2

0,1 0,5 1 1,5 10 10,5 100 100,5

1

26

Peptide HLA-B*27:02 MJ energyresidue residue (kcal/mol)

9 W 77 N -3.0780 I -5.78143 T -3.22

Σ-12.07

Peptide HLA-B*27:05 MJ energyresidue residue (kcal/mol)

9 W 77 D -2.8480 T -3.22143 T -3.22

Σ-9.28

HLA-B*27:02 HLA-B*27:05

77 N80 I

143 T

9 W

143 T

77 D80 T

9 W

Fig. 3 1

27

Fig. 4

A B

C

% IF

Nγ+

/CD

8+ T

cel

ls

0

10

20

30

40

50

- 10 - - 10 10 NS5B2820 wild-type [µM]- - 10 100 10 100 NS5B2820 R2821K [µM]

chronicresolved

1.0

1.2

1.4

1.6

1.8

0.1 1 10 100

Peptide [µM]

Flu

ore

scen

cein

dex

wild-typeR2820K mutantCMV pp65 control

D

without peptide

+ NS5B2820consensus

+ NS5B2820R2821K

CD8

IFNγ

0 102 103 104 105

0

102

103

104

105 0%

0 102 103 104 105

0

102

103

104

105 49.4%

0 102 103 104 105

0

102

103

104

105 0.04%

0 102 103 104 105

0

102

103

104

105

0 102 103 104 105

0

102

103

104

105

0 102 103 104 105

0

102

103

104

105 0%38,1%

0 102 103 104 105

0

102

103

104

105

0 102 103 104 105

0

102

103

104

105

0 102 103 104 105

0

102

103

104

105

0%

0% 0%41%

acute

chronic

resolved

chronic

resolved

peptide concentration (µM)

% I

FNγ+/

CD

8+ T

-ce

lls

0

5

10

15

20

25

wild-type peptidemutant peptide

peptide concentration (µM)

% I

FNγ+

/CD

8+ T

-ce

lls

0

5

10

15

20

25

wild-type peptidemutant peptide

1

Table 1: HLA-B*27 frequency and subtype distribution in patients with chronic HCV genotype 1 infection

Cohort Country of n patients n HLA-B*27+ % HLA-B*27+recruitmenta (viral subtype distributionb) total B*27:02 B*27:05 others

Freiburg D 569 (1a: 179; 1b: 308; 1ns: 82) 35 6.2 31 6 24 1c

Boston USA, D, CH 557 (1a: 405; 1b: 152) 27 4.8 24 3 16 5d

Perth AUS, CH, GB 236 (1a: 185; 1b: 51) 13 5.5 13 2 11 0

Frankfurt D 159 (1a: 69; 1b: 87; 1ns: 3) 16 10.1 10 2 8 0

Total 1521 91 6.0 78 13 59 6

a Country code: D: Germany; CH: Switzerland; AUS: Australia; GB: Great Britainb ns: subtype not determined or mixed infectionc B*27:14d 2 x B*27:03, 2 x B*27:06; B*27:10

n HLA-B*27 subtypes and viral sequence available

Peptide name Epitope SequenceHLA-B*27:05 (IC50

nM)

NS5B-2820 ARHTPVNSW 338NS5B-2841 ARMILMTHF 1.1NS5B-2936 GRAAICGKY 1.0

FLU-NP SRYWAIRTR 0.44HIV gag p24-262 KRWIILGLNK 0.22

Table 2: HLA-B*27:05 binding affinity determined in a competition assay using radiolabeled standard peptide

Table 3: Autologous viral sequences corresponding to the NS5B2820 epitope

Consensus A R H T P V N S W n patients % patients

B*27:02+ . K . . . . . . . 12/ 13 92.3%. . . . . . . . . 1/ 13 7.7%

B*27:05+ . . . . . . . . . 53/ 59 89.8%. . . . . I . . . 3/ 59 5.1%. . R . . I . . . 2/ 59 3.4%. K . . . . . . . 1/59 1.7%

B*27+, other subtypes . . . . . . . . . 5/ 6 83.3%. . . . . I . . . 1/ 6 16.7%