HCV antigen design:
Epitope prediction, evaluation and enhancement
Report for the first internship program of Martijn Vlaming
Master in Biomedical sciences
Under supervision of Prof. dr. Toos Daemen
Dept Medical Microbiology
Section Tumor Virology and Cancer Immunotherapy
University Medical Center Groningen June 2015
Hepatitis C virus (HCV) causes a persistent infection in approximately two-third of the people who contact the virus, resulting in almost 130–150 million infected people worldwide being approximately 3% of the world population. HCV is characterized by the lack of a proofreading mechanism and specifically targeting the liver in which it can cause cirrhosis and/or fibrosis, liver failure and hepatocellular carcinoma. Consequently it is the leading cause for liver transplantation in the western world. Despite improved clinical treatment options, a strategy that not only clears the virus but also induces a memory response against it is warranted. In order to provoke a specific immunological reaction towards HCV infected cells, previous studies examined the HCV genome and pointed out specific epitopes located in the HCV polyprotein which are processed by MHC I. Recent literature also suggests that MHC II epitopes in close proximity or overlapping with MHCI epitopes could play a role in the elicited immunological response. For this reason we identified and compared strong HCV epitopes for MHC I as well as MHC II by theoretical algorithms. Good animal HCV models are lacking because mice do not support HCV infections naturally. Therefore we also wanted to further optimize an existing in vitro T-cell activation model to test the immunogenicity of predicted HCV epitopes.
Recently, P.P. Ip et al. developed an alphavirus-based human HPV vaccine which also contains a helper epitope and an ER targeting signal which strongly enhances the immunogenicity of the vaccine.
Remarkably these constructs do not strengthen the elicited response when inserted in a HCV vaccine construct. For this reason we wanted to examine the immunogenicity increasing properties of the helper epitope and the ER targeting signal as encoded by the HPV vaccine, in combination with a rSFV-HCV vaccine. In this study we identified several strong immunogenic HCV epitopes by theoretical epitope prediction and tested the immunogenicity in an in vitro model. This model was able to reveal promising immunogenic differences between the predicted epitopes. This model requires further optimization before being reliable enough for an intermediate step between animal models and clinical trials. Also, helper sequences do not appear to be universal applicable for enhancing the immunogenicity of rSFV vaccine constructs.
Since the discovery that viruses can be involved in the process of carcinogenesis six human viruses have officially been ascribed as tumor-associated (Rous P., 1911, Martin D and Gutkind JS., 2008).
Due to elicited inflammation and/or persistent expression of oncoproteins in the host, these viruses are responsible for 10-15% of all global cancers (Parkin DM., 2002). Hepatitis virus (A, B and C) infections are accounted for almost half (4,9%) of these cancers worldwide (Parkin DM., 2002). Hepatitis C virus (HCV) was first described as non-A, non-B hepatitis virus in 1989 (Houghton M., 2009). Persistent infection occurs in approximately two-third of the people who contact the virus, resulting in almost 130–150 million infected people worldwide (approximately 3% of the world population) and 350.000 to 500.000 deaths each year (WHO, Gravitz L, 2011). 15-45% of the infected people are able to naturally clear the virus within 6 months after infection. The lack of a proofreading mechanism in the HCV and the enormous production of virus particles in the infected patient (1012 particles per day) can result in quasispecies within a host and rapid adaption to the new environment (Myrmel H et al., 2009, Thomas HC et al., 1999). This process yielded at least seven major genotypes and many more subtypes around the world (Smith DB et al., 2014). Spreading of HCV occurs mainly among intravenous drug users which share needles, through contaminated blood products or through the parenteral route (Maheshwari A et al., 2010). HCV is characterized specifically targeting the liver in which it can cause cirrhosis and/or fibrosis, liver failure and hepatocellular carcinoma. Consequently it is the leading cause for liver transplantation in the western world (Mukherjee S et al., 2008). Also, the lack early specific symptoms results in patients unaware of the infection and rapid spreading of the virus.
HCV is an enveloped , positive, single-stranded RNA virus in the Flaviviridae family. It contains a 9.6- kb genome with 5’ and 3’ untranslated region and one open reading frame which codes for one viral polyprotein. Upon posttranslational cleavage, this polyprotein codes three structural proteins (E1, E2 and core protein), six non-structural proteins and one membrane protein. Structural proteins E1 and E2 make up de the virus envelope. The six non-structural proteins (NS2, NS3, NS4A, NS4B, NS5A and NS5B) play important roles in processing the virus life cycle. Membrane protein P7 functions as a ion channel. (T. Seng-Lai., 2006) The linear organization of the HCV genome is displayed in figure 1.
Figure 1. Linear organization HCV (adapted from Bartenschlager R et al., 2010)
After 2 decades of research no effective and completely safe medicine/vaccine has been produced.
Current treatment consist of a combination of peg-IFN-α and ribavirin, which leads to a curative rate for about 50% (Fried MW et al., 2002). Unfortunately this treatment comes with side effects which is often the reason of treatment discontinuation (Sulkowski MS et al., 2011). Fortunately antiviral treatment has greatly improved in the past years. Telaprevir and boceprevir, direct-acting antiviral agents (DAAs), are now being used in combination with the standard-of-care treatment which is associated with significantly improved rates of constant virologic response (Jacobson IM et al., 2011).
However, due to the high mutation rate of HCV, combinations of antiviral drugs are required which could lead to drug-drug interactions and additional side effects (Lutchman G et al., 2007, Burger D et al., 2013). The quest for the holy grail of HCV treatment is still ongoing.
Vaccination against HCV
Despite improved clinical treatment options, a strategy that not only clears the virus but also induces a memory response against it is warranted. Therapeutic vaccination could facilitate the activation of both humoral and cellular immune responses. Prophylactic vaccines are aimed at inducing virus- neutralizing antibodies against structural components of the virus. Therapeutic vaccines are aimed at the induction of virus-specific CD8+ cytotoxic t lymphocytes (CTL) against nonstructural proteins of the virus. Several epitopes of HCV have been identified to be recognized by CTLs and CD4+ T cells in patients with an HCV infection including the core, NS5A/B region and the NS3 region (Lauer GM et al., 2004, Thimme R et al., 2001). To target these epitopes in the process of developing a therapeutic vaccine several approaches have been tested, including peptide- or protein-based vaccines, DNA vaccines, viral vector vaccines, recombinant yeast-based vaccines and vaccines based on dendritic cells. These methods all come with their advantages as well as their disadvantages, which resulted in none of these therapeutic vaccines reaching phase 3 clinical trials yet (Ip PP et al., 2012).
Viral vector vaccines, in particular alphavirus vectors, possess good capacities for development of therapeutic vaccines. They are more immunogenic compared to DNA-based vaccines and are not affected by pre-existing immunity against vaccines based on Adenoviruses (Ad) or modified vaccinia virus Ankara. Vaccines based on Alphaviruses are able to induce humoral as well as cellular responses which seem to be more wide and robust than responses elicited by other approaches.
In order to provoke a specific immunological reaction towards HCV particles, previous studies examined the HCV genome and pointed out specific CTL epitopes located in the polyprotein which could elicit strong immunological responses. According to previous literature (Guevare APH., 2014, Wertheimer AM et al., 2003) CTL epitopes “CINGVCWTV” and “KLVALGINAV” (both located in the NS3 region) seem to elicit strong immunological responses in PBMC’s challenged to these epitopes.
Especially epitopes which are processed by MHC-I, 9-mer epitopes, are mentioned. Though, recent literature also mention MHC II to process identical or longer epitopes compared to MHC I and thereby greatly contributing to the elicited immunological response (Arnold PY et al., 2002). MHC-II normally
present slightly longer epitopes compared to MHC-I, normally 13-17 amino acids. Some studies also mention the importance of peptide-flanking residues (Cole DK et al., 2012). Not only the core binding sequence of the epitope is important to bind MHC II for a proper response, 5’- and 3’ ends also contribute to the specific response. Changing these non-bound residues can even lead to an altered immune response. These findings have motivated us to reconsider the choice for selecting specific CTL epitopes for optimizing the in vitro model to test the immunogenicity of HCV epitopes for future HCV vaccines. By using several epitope prediction logarithms we will identify new strong HCV epitopes for our study.
Good animal HCV models are lacking because mice do not support HCV infections naturally (Guha C et al., 2005). To bridge the gap between animal and human models, an in vitro system based on human cells has been developed (Appay et al., manuscript in preparation). Appay et al. designed a model based on human peripheral blood mononuclear cells (PBMCs) cultures to induce CD8+ T cell priming. This method relies on the induction of mature dendritic cells (mDC) which are capable of priming naïve T cells. This will make them an attractive adjuvant for a therapeutic vaccine. Numerous protocols have been described to generate mDCs in vitro, which are traditionally differentiated from monocytes. However, recent approaches rely on PBMCs to generate mDCs. By stimulation of these cells by different maturation factors, this strategy saves time and yields mDCs which are more efficient in antigen processing compared to existing approaches (Bürdek M et al., 2010). Appay et al.
stimulated the human PBMCs with several maturation factors and Melan-peptide, which yielded a specific CD8+ T cell response. This CD8 priming protocol was recently used to develop an in vitro model to test the immunogenicity of HCV epitopes and alphavirus-based vaccines (Guevare APH., 2014). It was shown that this model can be used to test the immunogenicity of HCV epitopes through the expression of CD69+ and IFNγ+ response. It was also possible to induce a T cell response to human cells which were vaccinated with HCV vaccines. These promising results motivated us to further optimize and improve this model by use of an ELIspot assay.
Also, previous studies have already led to new developments in the production therapeutic vaccines against HCV as well as against HPV (Ip PP et al., 2014, 2015). For HPV vaccine constructs several Helper epitopes are identified which enhance the immunogenicity of the vaccine. Ip pp et al., showed that inclusion of an helper epitope and an ER targeting signal in a recombinant Semliki Forest virus (rSFV3e-sig HELPE7SH-KDEL) vaccine could significantly increase the frequencies of HPV-specific T cells. However, the same helper construct was not able to increase the immunogenicity of a previously developed HCV vaccine (rSFV4.2eNS3/4A) when included. The failure to increase the immunogenicity of the vaccine could have several causes. One possibility could be that the included helper epitope only works properly when combined with the HPV vaccine. This could explain the malfunction of the helper epitope when inserted in the HCV vaccine. To find out whether it is just the combination of the inserted helper epitope in the HCV vaccine that doesn’t work or if the HPV vaccine is required for a specific activation of the helper epitope, we also performed an immunization study in mice to examine this question.
1) Identify and compare strong HCV epitopes for MHC I as well as MHC II by theoretical algorithms.
2) Further optimize an existing in vitro T-cell activation model to test the immunogenicity of HCV epitopes.
3) Examine the immunogenicity increasing properties of a HPV helper epitope and an ER targeting signal in combination with a rSFV-HCV vaccine.
Materials & Methods
Identification of hepatitis C virus cytotoxic t lymphocyte epitopes
To strengthen the motivation for selecting a specific HCV epitope, SYFPHEITHI (http://www.syfpeithi.de/ ) was used as an epitope prediction program in combination with two other epitope prediction programs; NetMHC pan 2,8 (http://www.cbs.dtu.dk/services/NetMHCpan/) and IEDB (http://www.iedb.org/). Strong binding epitopes found in SYFPHEITHI (MHC I, HLA-A*02:01) were subsequently compared in the other two programs. In SYFPHEITHI, epitopes which scored higher than 20 were assumed to be strong binders. In NetMHC pan 2,8 and IEDB, a score lower than 0,5 was considered a strong binder.
To further identify the binding capacity of the specific HCV epitopes, MHC II epitopes overlapping or in close proximity to CTL epitopes were also examined with IEDB (HLA-DRB1*15:01). Binding capacity was determined for epitopes which contain the complete CTL epitope (or more) or for epitopes which have an 5’ or an 3’ end overlap with the specific CTL epitope. An overlap is considered an overlap when it contains the complete CTL epitope minus 1, when it starts at a maximum of 11 amino acids at the 5’end or when it ends at a maximum of 11 amino acids at the 3’end (see figure 2). Comparing the binding capacities of both MHC I and MHC II epitopes may give a better insight in the possible strength of the immunological response.
Figure 2. Example of MHC II binding epitopes. Highlighted in red is a MHC I epitope.
7 Figure 3. Timetable of the in vitro HCV model for testing the immunogenicity of HCV epitopes with an ELIspot.
Peptides, antibodies and reagents
The long peptide arrays from NS3 and NS5B (strain H77) were purchased from BEI Resources and reconstituted in 100% DMSO to a concentration of 20 mg/ml. Further dilutions to 1mg/ml and 0,1mg/ml were performed in PBS. Inflammatory human cytokines TNF-α , FLt3L, IL-1β, PGE2 and IL-7 were purchased from Preprotech. PGE2 was acquired from Merck Calbiochem.
PBMC isolation from whole blood
5 heparin tubes with blood (HLA A2/DR 15) were received from transplantation immunology (UMCG).
Blood was diluted with PBS and gently pipetted on top of 15 ml Ficoll-paque plus in three 50 ml tubes.
After centrifugation, the interphase was isolated and further purified by centrifugation. 106 cells were dissolved in FCS and freezing solution (DMSO and FCS) and stored overnight at -80°c. The next day they were transferred into liquid nitrogen.
PBMC isolation from buffy coat
50 ml of buffy coat (HLA-A*02:01/DRB1*15:01) was ordered at Sanquin Bloodbank, Groningen, Netherlands. Blood was diluted with RPMI medium (12,5 : 35) and gently added on top of 15 ml ficol containing tubes. After centrifugation, the interphase (PBMC ring) was isolated and further purified by centrifugation. Cells were dissolved in 8 ml RPMI and counted. Next, 500x106 cells were divided over 20 vials (25x10e6 each) and stored overnight in -80°c. The following day they were transferred into liquid nitrogen.
PBMCs culture and stimulation
On day 0 PBMCs were recovered from liquid nitrogen (or fresh) and subsequently dissolved in AIM-V medium (life technologies) supplemented with Flt-3L to a final concentration of 50ng/ml. Cells were seeded 2x106/200µl per well in a 48 well plate and incubated overnight at 37°C / 5% CO2. The next day (after 24 hours, day 1) a cocktail of cytokines, TNF-α (2000U/ml), IL-1β (20ng/ml), PGE2 (2uM) and IL-7 (1ng/ml) in AIM-V media was added (200 µl/well) to aid the DC maturation. Subsequently the long peptide mixtures were added (final concentration of 1µg/ml). Cultures were incubated overnight at 37°C / 5% CO2. After 48 hours (day 2) FCS was added to the cultures to a final concentration of 10%
(40 µl). Time schedule is displayed in figure 3. Total incubation time for the culture was 10 days. Cells were collected for an ELIspot assay.
IFN-gamma ELIspot assay
On day 9 of the PBMC culture, a 96 well millipore multiscreen HTS filter plate ( Millipore, Merck) was coated with IFN-γ capture antibody (5µl/ml, mAB-1-DK, Mab Tech). Dilutions were performed in PBS.
The filter plate was stored overnight at 4°C. On day 10, the coating antibody was aspirated and the plate washed 4 times with PBS. Next, the plate was blocked by adding 100µl of X-VIVO medium (LONZA) and subsequently incubating one hour at 37°C / 5% CO2. PBMC’s were harvested, counted and diluted (with AIM-V medium) to 4x10e6/ml or 2x10e6/ml. X-VIVO medium was aspirated from the filter plates and PBMC’s were seeded in duplicates or triplicates. Next, PBMC’s were re-stimulated with the long peptides. 1,5 µl of peptide (1mg/ml) was diluted in 48,5µl X-VIVO medium and added to the wells (final concentration of 10µg/ml). PHA was used as a positive control. The filter plate was incubated for 3 days at 37°C / 5% CO2. On day 13, the contents of the filter plate were aspirated and the wells were washed thoroughly with PBS/Tween. Next the plate was incubated with 7-B6-1 biotin (0,3µg/ml) for two hours at room temperature in the dark. The plate was washed again and incubated for one hour with Extravidin-ALP (1:1000 in PBS) at room temperature in the dark. The plate was washed again and developed using one tablet of BCIP/NBT ALP substrate in 10 ml of milliQ.
Incubation time ranged from 1-20 minutes and colorimetric reaction was stopped by washing the plate with water. Plate was dried overnight and examined the next day using the AID ELIspot/flurospot reader. Results were assessed using AID ELIspot 6.0 iSpot software.
Intracellular cytokine (IFN-gamma) staining assay
On day 0 PBMCs were recovered from liquid nitrogen and re-suspended in AIM-V medium (life technologies) supplemented with Flt-3L to a final concentration of 50ng/ml. Cells were seeded 2x106/200µl per well on a 48 well plate and incubated overnight at 37°C / 5% CO2. The next day (after 24 hours, day 1) a cocktail of cytokines, TNF-α (2000U/ml), IL-1β (20ng/ml), PGE2 (2uM) and IL-7 (1ng/ml) in AIM-V media was added (200 µl/well) to aid the DC maturation. Next, the long peptide mixtures were prepared to a final concentration of 0,1 mg/ml and 4µl was added to the assigned wells.
Cultures were incubated overnight at 37°C / 5% CO2. After 48 hours (day 2) FCS was added to the cultures to a final concentration of 10% (40 µl). On day 8 cells were harvested (centrifuged at 350g) and re-seeded them with fresh medium (containing 10% FCS) for culturing without the cytokine and peptide mixture for 2 days. On day 10 cells were harvested and transferred into 5 ml tubes (Greiner Bion-one) (1x10e6 cells/250µl). Next, half the cells was re-stimulated with the same peptide as the first stimulation ( 2µl from the 1mg/ml stock). The positive control sample received 1 µl of a cell stimulation cocktail (eBioscience). AIM-V/10%FCS was added to all samples to a final volume of 0,5 ml. Sample were centrifuged at 1500rpm at room temperature for 5 min and subsequently incubated for one hour at 37°c/5% CO2. After one hour, warm Brefeldin A (BFA) was added to each tube in a final concentration of 10µg/ml. Next, all samples were incubated for four hours at 37°c/5% CO2. After 4 hours, half of the samples were stored in the fridge until the next day. The other half of the samples were incubated 30 minutes on ice in the dark with 1ml PBS containing 1000x VIVID (Life Technologies). Then, the samples were incubated with 1µl CD4 PeCy7 (BD) and 5µl CD8 Fitch (BD) for 20 minutes at room temperature in the dark for surface staining. For the fixation 200 µl of warm 4%
PFA/sample was added and subsequently incubated for 10 min. After incubation, cells were re- suspended in FACsbuffer en kept at 4°c overnight. The next day the first half of the sample were also subjected to Vivid staining, surface staining and fixation. Afterwards all the samples were washed with 1 ml P/W 9 (eBioscience) and then incubated in PBS-S/milk (5%) at 4°c for 30 minutes. Next, INFg PerCP-C5.5 (Biolegend) and PBS-S/milk (5%) were added for IFN-gamma staining. After 30 min incubation at 4°c in the dark the samples were washed twice with P/W 9 and re-suspended in facsbuffer for FACS analysis. Data were analyzed using Flowjo software.
A total of 10 female C57BL/60LaHsd mice (8-10 weeks of age) were obtained from Harlan CPB, Zest.
All animal experiments were approved by the local Animal experimentation Ethical committee (the Institutional Animal Care and Use Committee of the University Medical Center of Groningen.
Virus particles rSFV4.2eNS3/4A, rSFV3e-sig HELPE7SH-KDEL, rSFV3eE7SH and rSFVeOVA were produced by Ip PP. On day 0 (prime day) the virus particles were diluted with PBS and subsequently activated by adding 80mM CaCl2 (1/40 of the total volume) and α-chymo (1/20 of the total volume).
After 30 minutes the activation was stopped by adding aprotinin to a final concentration of 1 mg/ml.
Final dilutions were performed in PBS and the immunization itself was executed by intramuscular injections in both legs (2 x 106 particles/mouse). Two weeks after the first immunization the mice were immunized again similar to the first immunization.
One week after immunization 5 to 10 drops of blood were extracted from each mouse to perform an dextramerstaining to examine the percentages of HCV and HPV specific CD8+ T cells. All blood samples were diluted in 1 ml heparine/PBS (1:500) and centrifuged at 1500g for 5 minutes. Next all samples were re-suspended in 1 ml of ACK buffer, transferred to 4 ml of ACK buffer and incubated for 11 minutes at room temperature to lyse red blood cells. The reaction was stopped by adding 15 ml of b-IMDM/5%FCS. Next, samples were incubated for 10 minutes at room temperature with 2 µl of NS3- PE (HCV dextramer) and 5µl of RAHINYVTF-PE (HPV dextramer). After 10 minutes 0,8µl of CD8-PE- Cy7 was added to the samples and followed by an additional 20 minutes incubation at 4°c. Samples were washed twice and re-suspended in FACS buffer for FACS analysis. Data were analyzed using Flowjo software.
Hepa 1-6 stimulator cells were cultured in the presence of 50U/ml recombinant IFN-gamma for 48 hours. Three weeks after the first immunization the mice were sacrificed and spleens were harvested.
Splenocytes were isolated and cultured with Hepa 1-6 (100 Gy-irradiated, level 25) cells into 96-well round bottom plates at E:S ratio 25:1. Recombinant IL-2 (5 U/ml) was added into the CTL culture on day 3 and 5 of the culture. After 7 days of culture we harvested the EL4 target cells (EL4 and EL4
pulsed with NS3 603-611 GAVQNEVTL (3P1)). Target cells were cultured in the presence 50U/ml recombinant IFN-gamma for 48 hours before co-culturing with the effector cells. Before co-culturing, target cells were pulsed with 51Cr (90µCi/2 x 106)(PerkinElmer) in the presence or absence of 3P1 peptide (1mg/ml) for 1 hour at 37°c with 5%CO2. Target cells were added to the 96-well plates in 6 copies and cultured for 4 hours at 37°c with 5%CO2. After 4 hours the supernatant of the cultures was harvested and analyzed with a RiaStar manual gamma counter (Packard, Meriden, CT). The cytotoxicity was calculated using the formula: ((experimental release – spontaneous release) / (maximal release – spontaneous release)) x 100%.
Identification of hepatitis C virus cytotoxic T lymphocyte epitopes
After analyses of nonstructural regions of the HCV genome (genotype 1 H77) with MHC class I epitope prediction algorithms SYFPEITHI, NetMHCpan 2.8 and IEDB, 16 epitopes with predicted good binding capacities with HLA-A*02:01 were identified (see figure 4). These epitopes were found in nonstructural proteins of the HCV genome. The first round of analysis was based on the most common CTL epitope prediction program, SYFPHEITHI. Epitopes which scored higher than 20 were assumed to be strong binders. By comparing the binding capacities of the epitopes found in STFPHEITHI with the binding capacities found in NetMHCpan 2.8 and IEDB (score <0,5 is a strong binder), several epitopes which theoretically possess very strong binding capacities with MHC I were identified. To expand the understanding of the binding capacities of the CTL epitopes, MHC II epitope binding capacity was also examined with the IEDB algorithm (<10 strong binding). As mentioned before, increasing evidence is emerging which suggest that peptide flanking residues appear to play an important role in the immunological response (Cole DK et al., 2012). This is why the neighboring sequences were also examined for the presence of MHC II epitopes. Binding capacity was determined for epitopes which contain the complete CTL epitope (or more) or for epitopes which have an 5’ or an 3’ end overlap with the specific CTL epitope. For the MHC II analysis the most abundant MHC II type in the western world was used: HLA-DRB1*15:01 (Allele Query Form IMGT/HLA). MHC II epitopes were scored from best to worst binding capacity (see figure 4). As can been seen in figure 4, most of the strong binding MHC I epitopes also possess a strong MHC II in the same sequence (overlap) or at the 5’ or an 3’ end (flanking residues). This corresponds with existing literature (Arnold PY et al., 2002, Cole DK et al., 2012). We identified several epitopes which possess good binding capacities for MHC I as well as for MHC II. Bei Resources supplemented us with a peptide array of HCV, H77, NS2 to NS5B (beiresources.org). The peptide array gave us the possibility to pick the peptides which possess the best sequence for binding MHC I as well as MHC II for our specific HLA type. Figure 5 displays the long peptides we eventually chose to use in our in vitro study.
igure 4. Sequences epitopes which bind MHC I (HLA-A*02:01) determined by SYPHEITHI (20> strong binding), NetMHCpan 2,8 (<0,5 strong binding) and IEDB (<0,5 strong binding). MHC II (HLA-DRB1*15:01) binding was also determined by IEDB. Results display the range of binding capacity from the strongest to the weakest binder (<10 strong binding). Binding capacity is determined for epitopes which consists the complete CTL epitope (or more) or for epitopes which have an 5’ or an 3’ end overlap with the specific CTL epitope. An overlap is consisted an overlap when it consist the complete CTL epitope minus 1, when it starts at a maximum of 11 amino acids at the 5’end or when it ends at a maximum of 11 amino acids at the 3’end. Strong MHC II binders are highlighted in grey.
Figure 5. Strong epitopes determined by epitope prediction programs SYPHEITHI, NetMHCpan 2,8 and IEDB. Highlighted in red are the predicted strong CTL epitopes.
Optimize an in vitro HCV model for testing the immunogenicity of HCV epitopes.
The first ELIspot assay’s we performed were teased by high background of IFN-gamma producing cells. This has led to several trial and error modifications of the existing protocol. Eventually we were able to scale back the background by washing more thoroughly, reducing the developing time of the millipore multiscreen filter plate and replacing the medium on day 8 of the culture. Actual differences between the different stimulations performed now became visible. Figure 6 displays four different groups in first stimulation round (Nothing, NS3 7, NS3 81 or NS5B 66/67). On day ten of the culture (after harvesting) each sample was re-stimulated with different long peptides, as can been seen in figure 6. An increased number of spots was clearly visible in previously mentioned high potential peptides such as NS3 1069 (NS3 7 in Bei-resources array) when also re-stimulated with NS3 1069.
Also NS3 1543 (NS3 81) clearly displayed a high number of spots when re-stimulated with NS3 1543.
This was consistent with the findings in the theoretical analysis. Furthermore, epitopes originating from the NS5B part of the HCV genome were also able to elicit a response. Especially NS5B 66/77, re- stimulated with NS5B 67, showed an increased number of spots compared to the background.
Next it was examined if the strong IFN-gamma response elicited by some HCV epitopes is due to the presence of an MHC I or an MHC II epitope. The ELIspot assay (see figure 8) showed that the IFN- gamma response mediated by NS3 81 was mainly based on a CD 4 response (T-helper) because the re-stimulation with NS3 82 (only a CTL epitope) resulted in a significant lower response compared to re-stimulation with NS3 81 (CTL and T-helper epitope). As can been seen in figure 8 , IFN-gamma responses evoked by NS3 82 are mostly mediated by a CD8 (CTL) response because re-stimulation with NS3 82 or NS3 81 yields almost the same IFN-gamma response. Also, in the first stimulation with NS3 82 only a CTL epitope was present. Re-stimulation with NS3 81 or NS3 82, which both contain the CTL epitope, increased the response. Interestingly, when stimulated with NS3 80 (does not contain a CTL or T-helper epitope) the IFN-gamma response was also increased. The response is also higher when re-stimulated with NS3 80 compared to re-stimulation with NS3 81.
Figure 7. NS3 long peptides which contain MHC I and/or MHC II epitopes. CTL epitopes are highlighted in red. T-helper epitopes are highlighted in green. NS3 82 only contains a strong CTL epitope. NS3 81 contains both CTL as well as several strong T- helper epitopes. NS3 80 contains none of both (only predicted weak binders).
Peptide Length Sequence
80 17 WYELTPAETTVRLRAYM
81 18 AETTVRLRAYMNTPGLPV
82 18 RAYMNTPGLPVCQDHLEF
Figure 6. The immunogenicity of HCV epitopes NS3 7, NS3 81, NS5B 66 and NS5B 67. ELIspot results in number of spots / 2x105 cells. PBMC’s were cultured for 10 days in the presence of a cytokine mixture (TNF-α, IL-1β, PGE2 and IL-7) in AIM-V media. On day one the PBMC’s were stimulated with a specific long peptide (1µg/ml per aliquot). On day 10, after harvesting, PBMC’s were re-stimulated again with a specific long peptide (10µg/ml per aliquot) and incubated in Millipore plates for 3 days before developing. Striped bar samples are re-stimulated with the same long peptide as the first stimulation. These samples are expected to yield the strongest response.
Figure 8. The immunogenicity of HCV epitopes NS380, NS3 81 and NS3 82. ELIspot results in number of spots / 2x105 cells. PBMC’s were cultured for 10 days in the presence of a cytokine mixture (TNF-α, IL-1β, PGE2 and IL-7) in AIM-V media. On day one the PBMC’s were stimulated with a specific long peptide (1µg/ml per aliquot).
Medium with cytokines was replaced on day 8 for fresh medium without cytokines. On day 10, after harvesting, PBMC’s were re-stimulated again with a specific long peptide (10µg/ml per aliquot) and incubated in Millipore plates for 3 days before developing. Striped bar samples are re-stimulated with the same long peptide as the first stimulation. Re-stimulation with NS3 82, 82, 80 - (-) stands for stimulation and re-stimulation with the same peptide minus the spot number were no re-stimulation was performed. Assay was carried out in 6-fold. Highest and lowest results were excluded from analysis. Raw data is displayed in figure S3.
Testing the immunogenicity of HCV epitopes which contain MHC I and/or MHC II epitopes.
For an alternative method to assess the IFN-γ production of PBMC’s when stimulated with specific peptides, an intracellular cytokine (IFN-gamma) staining assay was performed. Intracellular cytokine staining is a powerful technique for classification of cytokine production by several subsets T- lymphocytes. Due to the fact that it allows analyzation of individual cells in a mixed population it is a very powerful and specific tool. A 10 day cell culture, identical to the IFN-gamma ELISPOT assay, was performed. Identical to the ELIspot assay, the medium was refreshed on day eight of the culture to minimize background IFN-gamma production. Stimulation was performed with 3 different long peptides; NS3 80, NS381 and NS3 82. Re-stimulation was performed by stimulation with the same peptide (+) or no re-stimulation at all (-). Results are displayed in figure 9. Figure 9a displays the percentage IFN-gamma producing cells of the CD8 population and figure 9b displays percentage IFN- gamma producing of the total population of 1x106 cells. We added this last group to make sure that IFN-gamma producing cells which lost the CD8 antibody (possibly due to washing) were also detected. As can be seen in figure 9a and 9b, some small differences in percentages IFN-gamma producing CD8 cells are observed between the different groups. Both figure seem to display the same trend; long peptide 80 and 81 appear to evoke an increased IFN-gamma response when re-stimulated with the same peptide. This matches with the results found in the ELIspot assay’s. However, apparently part of the cells lost the CD8 antibody stain because IFN-gamma positive cells were also
Figure 10. Immunization scheme for the MicroCTL. A total of 10 mice in 4 groups. Each mouse received twice a dose of 2x106 particles.
detected in the non-CD8 positive group (see figure S4). Also, long peptide 82 does not perform as expected because re-stimulation with long peptide 82 does not yield an increased percentage of IFN-
gamma producing cells in none of the populations examined. This does not corresponds with the results found in the ELIspot assay’s.
Determination of HCV and HPV specific CD8 cells after HCV/HPV immunization
As can be seen in figure 10 and 11 we essentially performed a double vaccination study; for HCV as well as for HPV. The mice were divided into 4 groups (see figure 10). Group 1 was immunized with the
“normal” HCV vaccine in combination with the HPV vaccine with the helper epitope sig help and the KDEL sequence included. Group 2 was immunized with the normal HCV vaccine in combination with the HPV vaccine without the helper epitope. Group 3 was immunized with the HPV vaccine with the helper epitope in combination with OVA as a positive control. As a negative control we added group 4 were one mouse was injected with PBS. A dextramerstaining was performed on two different time points; one week after the first immunization (day 7) and three weeks after the first immunization (day 21, first day of micro CTL). With these dextramerstaining data we were able to examine the percentage of HCV or HPV specific CD8 cells in the blood and spleens of our mice.
Group Immunization Dose Number of
1. rSFV4.2eNS3/4A 2x10e6 3
rSFV3e-sig HELPE7SH-KDEL 2x10e6
2. rSFV4.2eNS3/4A 2x10e6 3
3. rSFV3e-sigHELPE7SH-KDEL 2x10e6 3
4. PBS 1
Figure 9. Immunogenicity of HCV epitopes which contain MHC I and/or MHC II epitopes by an intracellular cytokine (IFN-gamma) staining assay. Figure 9a displays % IFN-gamma producing cells within the CD8 population. Figure 9b displays % IFN-gamma producing of the total population of 1x106 cells. Samples were stimulated with long peptides 80 (No CTL or Th epitope), 81 (CTL and Th epitope) or 82 (only Th epitope). Re- stimulation was performed by stimulation with the same peptide (+) or no re-stimulation at all (-).
Figure 11. Determination of HCV and HPV specific CD8 cells after immunization. Figure 11a displays the HCV specific CD8 cells in mice treated with rSFV4.2eNS3/4A in combination with rSFV3e-sig HELPE7SH-KDEL or rSFV4.2eNS3/4A and rSFV3eE7SH. Figure 11b displays HPV specific CD8 cells for the same groups with rSFV3e-sig HELPE7SH-KDEL and rSFVeOVA included. On day 7 (one week after the first immunization) blood from each mouse was examined for HCV and HPV specific CD8 cells. On day 21 (first day of micro CTL) spleen cells were used for the same examination.
As can been seen in figure 11a and 11b, HCV and HPV specific CD8 cells are higher after the booster immunization. The corresponds with previous literature (Ip PP et al., 2015). However, the second staining after the booster was performed on mouse spleen cells instead of PBMC’s extracted from blood. Mice vaccinated with rSFV4.2eNS3/4A in combination with rSFV3e-sig HELPE7SH-KDEL did not significantly differ in percentages HCV specific CD8 cells when compared to mice vaccinated with rSFV4.2eNS3/4A and rSFV3eE7SH (figure 11a). Only mouse 1 shows a substantial increase in HCV specific CD8 cells (Table S1). This indicates that the sigHELP and the KDEL sequence (when inserted in rSFV3eE7SH ) are not able to increase the HCV specific CD8 cells when administered in combination with rSFV4.2eNS3/4A. As can been seen in figure 11b, the helper construct and the KDEL sequence did increase the %HPV specific CD8 cells as expected. However, this increase is not as high as previous literature describes (Ip PP et al., 2015). This lower response could be the result of the combination vaccination. The elicited immune response is possibly shared between the HCV as well as the HPV construct.
Immunogenicity increasing properties of an HPV helper epitope and an ER targeting signal in combination with a rSFV-HCV vaccine.
Due to an infection in the C3 cell line (HPV target cells) the day before the 51Cr release assay we were not able to examine cytotoxicity. Instead EL4 cells and EL4 cells pulsed with NS3 603-611
GAVQNEVTL (3P1) were used as target cells. The quantification of the cytotoxicity is displayed in figure 12a and 12b. Both figures show the same data displayed in a different way. As can been seen in both figures, after stimulation with a specific NS3 peptide (3P1) some small difference in cytotoxicity are observed between rSFV4.2eNS3/4A in combination with rSFV3e-sig HELPE7SH-KDEL (green) and rSFV4.2eNS3/4A in combination with rSFV3eE7SH (in EL4 + 3p1). However, these variances in
cytotoxicity appear not to be significant. This indicates that the helper constructs sig HELP and the KDEL sequence are not able to significantly increase the cytotoxicity of the splenocytes when incorporated in the HPV vaccine construct. As can been seen in figure 12b, there is a significant increase in cytotoxicity when we compare the specific (black filled bars) with the non-specific (normal bars) samples.
Figure 12. No immunogenicity increasing properties of an HPV helper epitope and an ER targeting signal in combination with a rSFV-HCV vaccine. MicroCTL results in percentages cytotoxicity. E:T ratio: 25:1. Study was carried out in six-fold (highest and lowest values were excluded from examination). As target cells EL4 cells or EL4 cells pulsed with NS3 603-611 GAVQNEVTL (3P1) were used. Figure 12a separates the samples by EL4 or EL4 pulsed with the NS3 peptide. Figure 12b displays each group with EL4 and the pulsed EL4 next to each other to compare the specific with the non-specific response. Black filled bars represent EL4 cells pulsed with NS3 603-611
GAVQNEVTL (3P1) as target cells. Data displayed in table S2.
In this study we had three different research questions to examine; 1) Identify and compare strong HCV epitopes for MHC I as well as MHC II by theoretical algorithms, 2) Further optimize an existing in vitro T-cell activation model to test the immunogenicity of HCV epitopes and 3) Examine the immunogenicity-increasing properties of an HPV helper epitope and an ER targeting signal in combination with a rSFV-HCV vaccine. Despite three different study objectives rather than one main target, we were able to examine these questions separately and combine them in this paper. Some yielded straightforward data and conclusions and others need future examination.
Identify and compare strong HCV epitopes for MHC I as well as MHC II by theoretical algorithms The first part of this study was mainly focused on theoretical analysis/identification of possible HCV epitopes which could elicit protective effects when inserted in a HCV vaccine construct. Already several epitopes in different parts of the conserved HCV genome have been described as strong immunogenic (Pishraft Sabet L et al., 2014, McKiernan SM et al., 2004). However, many studies are emphasizing different strong immunogenic epitopes in one of the seven different genotypes of HCV and in different parts of the HCV genome while much is still unclear about how these epitopes will actually perform in vivo. Also, many of these epitopes are specific for a certain haplotypes which are distributed over different parts in the world. Increasing evidence for HCV evolution, due to the lack of proofreading mechanisms, is also adding to this this problem (Timm J et al., 2015). This means that it will be essential to perform epitope prediction analysis when developing personalized vaccine constructs. We started this analysis with haplotype “HLA-A*02:01/DRB1*15:01” common in the western world (European Bioinformatics Institute).
In our analysis we used three different epitope prediction programs to increase the accuracy. As can be seen in figure 4, some epitopes only become visible as strong immunogenic when analyzed with the three programs together. Recent literature suggests the importance of CTL epitopes overlapping with T-helper epitopes or at close proximity of each other (Arnold PY et al., 2002, Cole DK et al., 2012). This could result in presentation of both the MHC I and MHC II epitope by the same dendritic cells which in turn will lead to both a CD8+ as well as a CD4+ T-cell response (Ip PP et al., manuscript in preparation). As absence of virus-specific CD4+ T-cells is commonly associated with development of chronic HCV infection (Timm J et al., 2014), we also included an analysis of MHC II epitopes in our study. As can be seen in figure 4 many epitopes which were found in the MHC I analysis also show overlap with MHC II epitopes determined by IEDB. However, not all predicted epitopes show this consistency. This could indicate that other mechanism influence the immunogenicity of certain epitopes as well. In the end, epitope prediction programs are just theoretical tools. These programs are not able to predict how the epitopes will actual behave. A previous study showed that only 1 out of the 22 predicted CTL epitopes induced a strong CD8+ t-cell response (Ip PP et al., manuscript in preparation). However, by combining three different MHC I prediction programs as well as one MHC II prediction program, the results gave us an strong indication about the immunogenicity of the predicted
epitopes. Future studies should possibly also examine the proteasomal cleavages sites as well as perform an stabilization assay to further narrow the search.
Further optimize an existing in vitro T-cell activation model to test the immunogenicity of HCV epitopes.
To test the immunogenicity of the predicted epitopes in the theoretical part of our study an IFN-gamma ELIspot assay performed. An existing in vitro T-cell activation model was previous described by Ana PHG et al. (2014) but needed further optimization. In this study we combined optimization of the existing protocol with testing of the immunogenicity of the predicted epitopes. An ELIspot assay is based on the principle of an enzyme-linked immunosorbent assay (ELISA) and is one of the fastest and most sensitive ex vivo antigen detecting assays available (Streeck H et al., 2009). The ELIspot was started by using epitopes which were already known to be strong IFN-gamma responders such as NS3 60 (1397 ELAAKLVALGINAVAYYR 1414) and NS3 7 (1059 TATQTFLATCINGVCWTV 1076.
This study clearly indicated that re-stimulation with the same long peptide as the first stimulation results in higher IFN-gamma response when re-stimulated with a a-specific long peptide (supplemental data S2). To evaluate our own predicted epitopes, an ELIspot was performed with long peptides NS5B 66, NS5B 67, NS3 81 and NS7. However, the performed ELIspot assays were teased by high background IFN-gamma production. When the PBMC’s were not stimulated at all, there was still production of IFN-gamma. The cytokine mixture we added on day one after start of the PBMC culture could have contributed to this production. The mixture of cytokines, in combination with the long peptides which were also added on day one, remained in the medium for 10 days. A 10 day stimulation could have been too long and excited the PBMC’s disproportionately. We decided to refresh the medium of the PBMC’s on day 8 of the culture and shorten the developing time of the ELIspot plate (until one sample gave a realistic response). By changing the medium a less immunogenic environment was created for the PBMC’s. These changes clearly removed a large part of the background IFN-gamma production (see figure 6 and 9). However, still we observed unexplainable large variations between the samples which were treated in the same way. Possibly unwanted excessive soap from washing the ELIspot plates could have influenced the interpretation of the results. During washing of ELIspot plates one should find a balance between removing all the soap and still be careful not to disturb the spots in the wells. As been indicated by ABCAM (troubleshooting for ELIspot), many factors can contribute to wrong interpretation of your results. Undoubtedly our ELIspot assay was not yet optimized enough. There is still room for future studies to improve this model.
However, the ELIspot assay’s performed gave some indication about what was happening at the immunological level of the PBMC’s. As mentioned above, MHC II epitopes also strongly influence the elicited IFN-gamma response (Arnold PY et al., 2002). It is known for example that a strong and broad CD4+ T-helper response, elicited by a MHC II epitope, is found in patients with a resolved hepatitis C virus infection (Wertheimer AM et al., 2003). To test this hypothesis, a new ELIspot assay was started based on the strong NS3 1543 (NS3 81) epitope (YMNTPGLPV). We chose three long peptides from
the Bei Resources array; NS3 1528, NS3 1534 and NS3 1541. NS3 1534 contains the strong MHC I epitope NS3 1543 (9-mer) as well as several strong MHC II epitopes (CTL + Th), see figure 7. Long peptide NS3 1541 only contains the MHC I epitope (CTL) and long peptide NS3 1528 contains none of both. To further minimize the background IFN-gamma production we further reduced the developing time of the ELIspot plate until the first spots became visible. This resulted in a significant decrease in background IFN-gamma production, as can be seen in figure 8. However, still the developing time showed differences between the plates. The ELIspot assay showed that the IFN-gamma response mediated by NS3 81 was mainly based on a CD 4 response (T-helper) because the re-stimulation with NS3 82 (only a CTL epitope) resulted in a significant lower response compared to re-stimulation with NS3 81 (CTL and T-helper epitope). As can been seen in figure 8 , IFN-gamma responses evoked by NS3 82 are mostly mediated by a CD8 (CTL) response because re-stimulation with NS3 82 or NS3 81 yields almost the same IFN-gamma response and in the first stimulation with NS3 82 only a CTL epitope was present. Re-stimulation with NS3 81 or NS3 82, which both contain the CTL epitope, increased the response. Also, when re-stimulated with NS3 81 (which contains both a CTL as well as an T-helper epitope) the response is considerably lower compared to re-stimulation with NS3 82. This could be explained by the immune response being shared on different epitopes. This could possibly yield an overall broader but lower response. Next, when stimulated with NS3 80 (does not contain a CTL or T-helper epitope) the IFN-gamma response was also increased. This does not support our previous findings from the theoretical epitope identifying study. According to SYFPHEITHI, NetMHCpan 2.8 and IEDB, there are no strong MHC I or MHC II binding epitopes present in this long peptide. Because the response is higher when re-stimulated with NS3 80 compared to re-stimulation with NS3 81, this could indicated a possible hidden strong MHC II epitope.
To assess the IFN-gamma production of PBMC’s in a different way were we could possibly also down regulate the high background IFN-gamma production and increase the reproducibility, an intracellular cytokine (IFN-gamma) staining assay was performed. According to literature, an intracellular cytokine staining assay yields the same results compared to an ELIspot assay with a slightly different threshold for positivity (Sun Y et al., 2003). An intracellular cytokine staining assay could also be performed within two days compared 4 days needed for an ELIspot assay. The intracellular cytokine staining assay is also more straight forward, does not include soap or variations in developing time which could influence the interpretation of the results. Due to the fact that the ELIspot assay yields differences in spot count between the groups on average of 100 spots/200.000 cells ((100/200.000)x100%=0,05%), it was not easy to determine the right gating for the flow cytometry data. At first there were no differences observed between the different groups. A part of the cells which were not CD8 positive did show production of IFN-gamma. Therefore examination of the complete live/single cell population for IFN-gamma production was performed (see figure 9b). These results were in line with our previous ELIspot assay data. However, only a response from long peptide 80 (no CTL or T-helper epitope) and long peptide 81 (both CTL and T-helper epitope) was observed. Long peptide 82, which does contain a CTL epitope, did not show any response in the complete population. These results do not correspond with the ELIspot data. Because the CD4 antibody was not working properly during this
study, there was no possibility to assess the CD8 and CD4 response separately. Also, examination of the complete population was necessary to identify differences between the peptide stimulations while the CD8 population did not display any significant data. Furthermore, in figure 9a, the PBS control was already as high as the highest responder in this group. These results are originating from an pilot study which clearly needs further optimization. Based on our data, the different long peptides used in this study do influence the percentages IFN-gamma producing cells. Especially the different responses elicited by MHC I or MHC II (overlapping) epitopes needs future attention.
Examine the immunogenicity-increasing properties of an HPV helper epitope and an ER targeting signal in combination with a rSFV-HCV vaccine
In the last part of this study we wanted to examine the immunogenicity increasing properties of an HPV helper epitope and an ER targeting signal in combination with a rSFV-HCV vaccine. Recent studies have shown that incorporation of several helper constructs/sequences in an rSFV vaccine expressing human papillomavirus early proteins significantly increase the vaccine immunogenicity (Oosterhuis K et al., 2012, p PP et al., 2015). Especially a sig HELP construct in combination with a KDEL sequence significantly increase the protein stability, frequencies of HPV-specific T cells and immunogenicity of a HPV vaccine developed by Ip PP et al., 2015. However, until today none of these helper epitopes are proven to be successful in combination with a HCV vaccine construct. This failure to increase the immunogenicity of the vaccine could have several causes. One possibility could be that the included helper epitope only works properly when combined with a HPV vaccine construct for activation. This could explain the malfunction of the helper epitope when inserted into a HCV vaccine construct. To find out whether it is just the combination of the inserted helper epitope in a HCV vaccine construct that doesn’t work or if the HPV vaccine construct is needed for a specific activation of the helper epitope, an immunization study in mice was performed. In this study mice were immunized against HPV as well as against HCV. As can been seen in the results of the dextramerstaining and the MicroCTL (figure 11 and 12), the percentage of HCV specific CD8 cells as well as the cytotoxicity of these cells was not significantly increased in mice immunized with rSFV4.2eNS3/4A in combination with rSFV3e-sig HELPE7SH-KDEL. This indicates that also the combination vaccine construct is not able to increase the immunogenicity. Perhaps the HCV vaccine construct by itself already elicited the strongest possible response such that there is no potential enhancement. This could explain the equal responses between the different groups observed in our study. Future studies should examine whether these helper epitopes are able to increase the immunogenicity of less strong HCV antigens.
This will prove whether the effect of an immunogenic fusion protein is specific for a particular vector vaccine or not. More studies should be performed on how exactly specific helper epitopes/constructs are able to increase the immunogenicity of specific therapeutic rSFV vector vaccines. Much is already known about how to increase vaccine potency for human papillomavirus (Lin K et al., 2010). However, for HCV vaccine constructs much is yet to be discovered. Also, increasing the percentages of HCV specific T-cells does not always yields an increased anti-tumor response. Previous studies with a HPV vaccine encoding T-helper epitopes showed that elicited percentages up to 22% E7-specific T-cells yield lower anti-tumor response compared to vaccines without T-helper epitopes yielding normal (1-
4%) percentages (Riezebos-Brilman A et al., 2007, Oosterhuis K et al., 2012). These studies show us that it is important to focus on the quality of the elicited T-cell response rather than quantity.
In this study it is shown that a combination of several theoretical algorithms can be a powerful tool to predict potential strong immunogenic HCV epitopes. Also, a combination of MHC I and the MHC II prediction tools greatly improves the accuracy of the predictions. Because of promising existing literature about overlapping and flanking MHC II epitopes, future studies a certainly needed on this promising topic. Next, the IFN-gamma ELIspot assay as well as the intracellular cytokine (IFN-gamma) staining assay both showed to be potential promising techniques for examining the immunogenicity of specific HCV epitopes. Both assays were able to reveal the differences in immunogenicity between several epitopes. However, because reproducibility was still lacking and standard deviations were higher than normal, both techniques first require further optimization before being reliable enough for an intermediate step between animal models and clinical trials. In the last part of this study we were able to point out that the immunogenicity of an existing rSFV HCV vaccine construct can’t be increased by a helper construct and ER targeting signal (sigHELP-KDEL) originating from an rSFV vaccine expressing human papillomavirus (HPV) early proteins. Apparently these helper sequences are not universal applicable for enhancing the immunogenicity of rSFV vaccine constructs.
Allele Query Form IMGT/HLA - European Bioinformatics Institute
Arnold PY, La Gruta NL, Miller T, Vignali KM, Adams PS, Woodland DL, Vignali DA. The majority of immunogenic epitopes generate CD4+ T cells that are dependent on MHC class II-bound peptide-flanking residues. J Immunol.
2002 Jul 15;169(2):739-49. Erratum in: J Immunol 2002 Oct 15;169(8):4674. PubMed PMID: 12097376
Bartenschlager R, Penin F, Lohmann V, André P. Assembly of infectious hepatitis C virus particles. Trends Microbiol. 2011 Feb;19(2):95-103. doi: 10.1016/j.tim.2010.11.005. Epub 2010 Dec 14. Review. PubMed PMID:
Bukh J. A critical role for the chimpanzee model in the study of hepatitis C. Hepatology. 2004 Jun;39(6):1469-75.
Review. PubMed PMID: 15185284
Bürdek M, Spranger S, Wilde S, Frankenberger B, Schendel DJ, Geiger C. Three-day dendritic cells for vaccine development: antigen uptake, processing and presentation. J Transl Med. 2010 Sep 28;8:90. doi: 10.1186/1479- 5876-8-90. PubMed PMID: 20920165; PubMed Central PMCID: PMC2955579
Burger D, Back D, Buggisch P, Buti M, Craxí A, Foster G, Klinker H, Larrey D, Nikitin I, Pol S, Puoti M, Romero- Gómez M, Wedemeyer H, Zeuzem S. Clinical management of drug-drug interactions in HCV therapy: challenges and solutions. J Hepatol. 2013 Apr;58(4):792-800. doi: 10.1016/j.jhep.2012.10.027. Epub 2012 Nov 5. Review.
PubMed PMID: 23137766.
Cole DK, Gallagher K, Lemercier B, Holland CJ, Junaid S, Hindley JP, Wynn KK, Gostick E, Sewell AK, Gallimore AM, Ladell K, Price DA, Gougeon ML, Godkin A. Modification of the carboxy-terminal flanking region of a universal influenza epitope alters CD4⁺ T-cell repertoire selection. Nat Commun. 2012 Feb 7;3:665. doi:
10.1038/ncomms1665. PubMed PMID: 22314361; PubMed Central PMCID: PMC3293629
Dauer M, Obermaier B, Herten J, Haerle C, Pohl K, Rothenfusser S, Schnurr M, Endres S, Eigler A. Mature dendritic cells derived from human monocytes within 48 hours: a novel strategy for dendritic cell differentiation from blood precursors. J Immunol. 2003 Apr 15;170(8):4069-76. PubMed PMID: 12682236