Activation of naı¨ve B lymphocytes via CD81, a pathogenetic mechanism for hepatitis C virus-associated B lymphocyte disorders

Domenico Rosa*†, Giulietta Saletti*†, Ennio De Gregorio*†, Francesca Zorat‡, Consuelo Comar‡, Ugo D’Oro*, Sandra Nuti*, Michael Houghton§, Vincenzo Barnaba¶, Gabriele Pozzato‡, and Sergio Abrignani*ʈ**

*Chiron Vaccines, 53100 Siena, Italy; ‡Institute of Internal Medicine, University of Trieste, 34127 Trieste, Italy; §Chiron Corporation, Emeryville, CA 94608-2916; and ¶Fondazione Andrea Cesalpino, Institute of Internal Medicine, University of Rome, 00158 Rome, Italy

Communicated by Rino Rappuoli, Chiron Corporation, Siena, Italy, October 28, 2005 (received for review September 30, 2005) Infection with hepatitis C virus (HCV), a leading cause of chronic Indeed, cross-linking of CD81 by the HCV-E2 protein or anti-CD81 liver diseases, can associate with B lymphocyte proliferative dis- mAbs leads to costimulation of human T cells (16) or inhibition of orders, such as mixed cryoglobulinemia and non-Hodgkin lym- human NK cells (17). In this study, we investigated whether phoma. The major envelope protein of HCV (HCV-E2) binds, with engagement of CD81 by recombinant HCV-E2 protein and͞or high affinity CD81, a tetraspanin expressed on several cell types. anti-CD81 mAbs had any effect on human B cell function. We Here, we show that engagement of CD81 on human B cells by a found that multimeric engagement of CD81 activates B lympho- combination of HCV-E2 and an anti-CD81 mAb triggers the JNK cytes in the absence of BCR coligation. At variance with other B cell pathway and leads to the preferential proliferation of the naı¨ve polyclonal stimuli, CD81 engagement preferentially induced pro- ؊ (CD27 ) B cell subset. In parallel, we have found that B lymphocytes liferation of naı¨ve B cells. We also found that, in the great majority from the great majority of chronic hepatitis C patients are activated of HCV-infected patients, B cells display an activated phenotype and that naı¨ve cells display a higher level of activation markers that disappears after therapeutic eradication of HCV. than memory (CD27؉) B lymphocytes. Moreover, eradication of HCV infection by IFN therapy is associated with normalization of Materials and Methods the activation-markers expression. We propose that CD81-medi- Patients. We analyzed peripheral blood B lymphocytes from 64 ated activation of B cells in vitro recapitulates the effects of HCV patients (55% males and 45% females; median age of 62.5 years) binding to B cell CD81 in vivo and that polyclonal proliferation of with chronic HCV infection, as demonstrated by the presence of naı¨veB lymphocytes is a key initiating factor for the development HCV-RNA and elevated serum levels (Ͼ40 units͞liter) of alanine of the HCV-associated B lymphocyte disorders. aminotransferase. Most of the patients (61%) were infected with HCV genotype 1; the remainder (39%) were infected by genotypes ͉ ͉ ͉ monoclonal antibody multimeric engagement B cell antigen receptor 2 and 3. A fraction of these patients (26 of 64) were diagnosed with cryoglobulinemia cryoglobulinemia, as indicated by the presence of dosable levels of cryoglobulins (Ͼ1.0%) accompanied by weakness, arthralgias, and epatitis C virus (HCV) is a positive-stranded RNA virus of the purpura (see Table 3, which is published as supporting information HFlaviviridae family (1). The HCV genome is 9.6 kb in length, on the PNAS web site). Patients were therapy-naı¨ve or had been off with one large translational ORF encoding a single polyprotein, antiviral therapy for Ͼ12 months at the time of the sampling. None which is processed by host and viral proteases into at least three of them had clinical evidence of cirrhosis. structural and seven nonstructural proteins with various enzymatic Twenty-two of 64 subjects were selected among therapy-naı¨ve activities (1). Two heavily N-glycosylated proteins E1 and E2 are patients and underwent antiviral therapy. Patients received 1.5 ␮g virion-envelope proteins and form heterodimers in vitro (2). An of peginterferon ␣-2b (PEG-Intron, Shering–Plough) per kg of estimated 170 million individuals are infected with HCV worldwide body weight s.c. once weekly and oral ribavirin (Rebetol, Shering– (3). HCV infection is associated with the development of chronic Plough) at 1,000–1,200 mg͞day (those weighing Ͼ75 kg received hepatitis, cirrhosis, and hepatocellular carcinoma (4). B cell abnor- the higher dose). Treatment lasted 6 months for patients with malities, including cryoglobulinemia (5) and an increased risk of B genotypes 2 and 3 (13 of 22) and 12 months for patients with cell non-Hodgkin lymphoma (6, 7), have been reported in a genotype 1 (9 of 22). Analyses were performed before and at the minority of HCV infections. end of therapy. In responders to therapy, as indicated by HCV- Until very recently, it was not possible to grow HCV in cell RNA negativity at PCR, an additional analysis after 12 months of culture, therefore studies of virus interaction with human cells have follow-up was performed to identify patients who achieved sus- been surrogated by the assessment of binding and entry of HCV tained virological response. The main biochemical and virological recombinant glycoproteins (8) or virus pseudotypes (9). characteristics of HCV-infected patients at the moment of enroll- We have previously reported that HCV-E2 protein binds with ment are summarized in Table 3. Twenty-one healthy blood donors high affinity to the large extracellular loop of human CD81 (CD81- LEL) and that ‘‘bona fide’’ HCV particles bind human CD81 (10). Recently, it has been demonstrated that CD81 is required for entry Conflict of interest statement: D.R., G.S., E.D.G., U.D., S.N., and M.H. are employees of and infection of human cells by in vitro-generated infectious HCV Chiron Corporation; S.A. is a consultant of Chiron Vaccines. (11). CD81 is a widely distributed cell-surface tetraspanin that Freely available online through the PNAS open access option. participates in different molecular complexes on various cell types, Abbreviations: BCR, B cell antigen receptor; CFSE, carboxyfluorescein succinimidyl ester; including B, T, and natural killer (NK) cells (12). On human B cells, HCV, hepatitis C virus; NK, natural killer; PBMC, peripheral blood mononuclear cell; PE, CD81 is known to form a costimulatory complex with CD19 and phycoerythrin; SAC, Staphylococcus aureus Cowan I. CD21 (13, 14) and that coligation of the B cell antigen receptor †D.R., G.S., and E.D.G. contributed equally to this work. (BCR) with any of the components of this costimulatory complex ʈPresent address: National Institute of Molecular Genetics, 20122 Milan, Italy. lowers the threshold required for BCR-mediated B cell prolifera- **To whom correspondence should be addressed at: National Institute of Molecular tion (15). We have proposed that HCV exploits CD81 not only to Genetics, Via Francesco Sforza 28, 20122 Milan, Italy. E-mail: [email protected]. invade hepatocytes but also to modulate the immune system. © 2005 by The National Academy of Sciences of the USA

18544–18549 ͉ PNAS ͉ December 20, 2005 ͉ vol. 102 ͉ no. 51 www.pnas.org͞cgi͞doi͞10.1073͞pnas.0509402102 Downloaded by guest on September 30, 2021 and 16 patients with chronic hepatitis B, matched for age and sex, B Cell Proliferation Assay. Triplicates of purified B lymphocytes were were used as control groups. Study protocols were approved by the seeded at 2 ϫ 105 cells per well in a 96-well flat-bottom plate ethical committee of Trieste University School of Medicine. (Costar) and stimulated with the following reagents: anti-CD81 All subjects provided written informed consent. mAbs (MG81 and N81), HCV-E2 protein, anti-CD19, anti-CD21, SAC, and anti-IgM. For inhibition experiments, MG81, HCV-E2, Clinical Tests. Alanine aminotransferase was measured by using SAC, and anti-IgM were preincubated with recombinant CD81- standard clinical tests. HCV RNA was determined by using RT- LEL for 10 min at 37°C and added to the cells. After 4 days of PCR by Amplicor HCV assay (Roche Diagnostics) and quantified culture, 1␮Ci (1 Ci ϭ 37 GBq) per well of [3H]thymidine was added, with a REAL QUANT C kit (Genedia, Munich). HCV genotypes and cells were incubated for an additional 16 h and harvested onto were determined by using the line-probe assay (Inno-Lipa HCV, filter plates (Packard). [3H]thymidine incorporation was measured Innogenetics, Ghent, Belgium). To evaluate the presence of cryo- by using a TopCount NXT ␤-counter (Packard). globulins, blood samples were allowed to clot at 37°C; serum was centrifuged for 10 min at 350 ϫ g at 4°C in graduated Wintrobe Flow Cytometry. PBMCs or purified B cells were stained in a tubes, and the volume percent of the precipitate compared with the one-step procedure incubating cells for 20 min at 4°C with the total volume of serum (cryocrit) was calculated. Rheumatoid factor following FITC-, PE- and PerCP-conjugated mAbs: anti-CD19 was measured by rate nephelometry. (clone 4G7), anti-CD27, anti-CD69 (clone L78), anti-CD71 (clone L01.1), anti-CD86 (clone 2331), and anti-CXCR3 (clone 49801). Cell Preparation and Purification. Peripheral blood mononuclear Mouse isotype-matched FITC, PE, or PerCP were used as negative cells (PBMCs) were obtained from the blood of patients and controls. All mAbs were purchased from Becton Dickinson or healthy volunteers by Ficoll-Hypaque (Amersham Pharmacia Bio- Pharmingen, except for the CXCR3 mAb, which was purchased tech) gradient centrifugation. B lymphocytes were negatively pu- fromR&DSystems. After washing, samples were acquired on a rified from PBMCs of healthy donors. Briefly, the cells were first FACSCalibur flow cytometer (Becton Dickinson), and data were stained with purified mAbs specific for CD2 (clone RPA-2.10), processed by using the program CELLQUEST (Becton Dickinson). CD14 (clone M5E2), CD16 (clone 3G8), and CD56 (clone B159) Analyses of B cells on total PBMCs were performed, gating on (BD Biosciences) at 4°C for 30 min; after washing, the cells were CD19-positive cells. incubated with magnetic beads coated with goat anti-mouse IgG antibodies (Miltenyi Biotec, Auburn, CA) according to the manu- Carboxyfluorescein Succinimidyl Ester (CFSE) Labeling. Purified naı¨ve facturer’s instructions. The purity of selected B lymphocytes was and memory B lymphocytes were resuspended at 2 ϫ 107 cells Ͼ98%, as confirmed by FACS analysis using peridinin– per ml in PBS, and CFSE (Molecular Probes) was added at a final chlorophyll–protein complex (PerCP)-anti-CD19 mAb (clone 4G7, concentration of 0.5 ␮M (19). The CFSE-labeled cells were BD Biosciences). Purified B lymphocytes were stained with phy- incubated with complete medium, anti-CD81 mAbs (MG81 and coerythrin (PE)-anti-CD27 mAb (clone L-128, BD Biosciences), N81), or SAC for 96 h, stained with PE-anti-CD27, and analyzed and naı¨ve (CD27Ϫ) and memory (CD27ϩ) subsets were sorted by by flow cytometry by using FACSCalibur (Becton Dickinson). FACSVantage SE (Becton Dickinson) with Ͼ96% purity. Human Cell viability was measured by propidium iodide incorporation. tonsils were obtained from patients undergoing routine tonsillec- tomy, and B lymphocytes were negatively isolated by using magnetic In Vitro Ig Production. For Ig measurements, 1 ϫ 106 purified naı¨ve beads coated with an anti-CD2 mAb (Dynabeads, Oxoid, Basing- and memory B lymphocytes were cultured with CD81 mAbs stoke, U.K.). More than 97% of the isolated cells expressed CD19. (MG81 and N81) or SAC. At day 10, IgM and IgG concentration in culture supernatants were measured by standard ELISA tech- Generation of mAbs Specific for Human CD81-LEL. Monoclonal anti- niques, as described in ref. 20. bodies (MG81 and N81, IgG1 isotype) were obtained in our laboratory from BALB͞c mice immunized with a recombinant Western Blot Analysis. B lymphocytes purified from human tonsils human CD81-LEL fused with thireodoxin (10) by using standard (30 ϫ 106 cells per ml) were stimulated with of MG81 (5 ␮g͞ml) and IMMUNOLOGY technique. Hybridoma supernatants were screened for the ability to N81 (5 ␮g͞ml) antibodies or goat anti-human IgM F(abЈ)2 frag- bind human CD81 and to neutralize the binding of HCV-E2 protein ments (IgM) (10 ␮g͞ml) at 37°C. At the indicated time, cells were to CD81 on CD81-transfected NIH 3T3 cell line (American Type washed with cold PBS and lysed for 1 h on ice in 300 mM NaCl, 50 Culture Collection). The F(abЈ)2 fragments were prepared by mM Tris⅐HCl (pH7.6), 1% Triton X-100, 1 mM NaF, 1 mM pepsin digestion by using the ImmunoPure F(abЈ)2 kit (Pierce) Na3OV4, 10 mM sodium pyrophosphate, 1 mM PMSF, and 1ϫ according to the manufacturer’s instructions. Retention of the EDTA-free complete protease-inhibitor mixture (Roche Diagnos- ability of the F(abЈ)2 fragment of N81 and MG81 mAbs to bind tics). Equal amounts of protein extracts were resolved on 10% CD81 on the cell surface was confirmed by flow cytometry. SDS͞PAGE gel and transferred on nitrocellulose membrane. Blot- ting was performed by using horseradish-peroxidase-conjugated Cell Cultures and Reagents. All cultures were performed in RPMI 4G10 anti-phosphotyrosine antibody (Upstate Biotechnology, medium 1640 (GIBCO͞BRL) supplemented with vitamin, strep- Lake Placid, NY), phospho-c-Jun (KM1) and total c-Jun (H-79) tomycin, and glutamine, all from GIBCO͞BRL and 10% FCS antibodies (Santa Cruz Biotechnology), total CD19 and phospho- (HyClone) and incubated at 37°C in a humidified atmosphere CD19 (Tyr-531) antibodies (Cell Signaling Technology, Beverly, containing 5% CO2. For the immunoglobulin-secretion experi- MA), phospho JNK1 and -2 (pTpY183͞185), and total JNK1 ment, complete medium containing 5% of ultralow IgG FCS (BioSource International, Camarillo, CA). The JNK inhibition (GIBCO͞BRL) was used. experiments were conducted by preincubating tonsil-derived B cells Cells were stimulated in vitro with the following reagents: anti- for 10 min at 37°C with increasing concentrations of SP600125 CD81 mAbs, N81 and MG81; F(abЈ)2 fragments of N81 and MG81; (Calbiochem) or an equivalent amount of DMSO as control. recombinant purified HCV-E2384–715 protein (96% pure) (18); anti-CD19 mAb (clone HIB19, BD Biosciences); anti-CD21 (clone Statistical Analysis. Comparison of the percentages of activation B ly4, BD Biosciences); recombinant human CD81-LEL (purity markers and CXCR3 on B cells from HCV patients before therapy, Ͼ90%) (10); formalinized Staphylococcus aureus Cowan I (SAC) at the end of therapy, and at the end of follow-up was performed (Pansorbin, Calbiochem); goat anti-human IgM F(abЈ)2 fragments by using two-tailed P values, Wilcoxson matched-pairs test specific for ␮-chains (Cappel ICN Biomedical). (INSTAT3, GraphPad).

Rosa et al. PNAS ͉ December 20, 2005 ͉ vol. 102 ͉ no. 51 ͉ 18545 Downloaded by guest on September 30, 2021 Fig. 1. CD81 engagement induces human B cell proliferation. B lymphocytes purified from healthy donor PBMCs were cultured for 5 days in the presence of increasing concentrations of MG81 mAb (0.15 to 10 ␮g͞ml) and 5 ␮g͞ml HCV-E2 (ࡗ), N81 mAb (■), anti-CD19 (Œ), or anti-CD21(F) mAbs (A); MG81 F(abЈ)2 (0.15–10 ␮g͞ml) and 5 ␮g͞ml N81 F(abЈ)2 (B); CD81-LEL (0.03–5 ␮g͞ml) combined with MG81 plus HCV-E2 (ᮀ)(5␮g͞ml each), goat anti-human IgM F(abЈ)2 fragments (‚) (10 ␮g͞ml), or SAC (E) (1:30,000, vol͞vol) (C). Cell proliferation was assessed by [3H]thymidine incorporation. In A and B, data are shown as the mean of cpm ϫ 103Ϯ SD of triplicate wells. In C, data are expressed as the percentage of inhibition of the B cell proliferation rate calculated in the absence of CD81-LEL. Data are representative of at least five different experiments performed with independent donors.

Results ment of CD81, in the absence of exogenous cytokines and BCR Multimeric Engagement of CD81 Induces B Cell Activation and Prolif- engagement, induces activation, proliferation, and increased ex- eration. It is well documented that CD81 cross-linking by HCV-E2 pression of CXCR3 molecules on B lymphocytes and that the protein or by anti-CD81 mAbs inhibits NK cells (17) and costimu- epitopes involved in HCV binding to target cells are required for lates T lymphocytes (16). Therefore, we assessed the effect of CD81 this process. cross-linking on human B lymphocytes. Engagement of CD81 on freshly purified B cells from healthy individuals through HCV-E2 CD81 Engagement Preferentially Activates Naı¨ve B Cell Proliferation. coated on plastic or anti-CD81 mAb cross-linked by anti-mouse B lymphocytes can be schematically divided into naı¨ve and memory antibodies did not affect expression of activation markers, matu- subsets on the basis of CD27 expression (21). To better assess the ration, or proliferation (data not shown). However, when B lym- effect of CD81-mediated activation on these B cell subsets, periph- phocytes were cultured for 5 days with a combination of two eral blood B lymphocytes were separated as CD27ϩ and CD27Ϫ anti-CD81 mAbs, MG81 and N81, in soluble form, we detected a cells, labeled with CFSE, and activated by CD81, IgM, or SAC. robust B cell proliferation (Fig. 1A). Both MG81 and N81 are We first assessed the proliferative response through CFSE la- directed against the recombinant CD81-LEL protein; however, beling. Fig. 2 shows that naı¨ve and memory B lymphocytes dis- only N81 is capable of neutralizing HCV-E2 binding to CD81. No played a similar proliferation rate when activated by SAC. In stimulation of proliferation was detected when B cells were treated contrast, CD81 engagement resulted in the preferential division of with MG81 or N81 alone (data not shown) or combined with mAbs naı¨ve B cells (45% vs. 1% control), with a minor proliferative effect directed against the other components of the B cell coreceptor, in the memory B cell subset (15% vs. 9% control) (Fig. 2 Top). In CD19 and CD21 (Fig. 1A). The stimulation of B cell proliferation a control experiment, we found no differences in the cell death rate was preserved, although to a lesser extent, when N81 was replaced of memory and naı¨ve B cells activated through CD81 engagement by a recombinant form of the HCV-E2 glycoprotein, whereas (data not shown). HCV-E2 alone did not have any effect (Fig. 1A). We found that B CD81 multimeric engagement and SAC stimulation induced the Ј cells stimulated with F(ab )2 MG81 and N81 mAbs proliferate as differentiation of the majority of naı¨ve cells into memory B efficiently as cells stimulated with intact immunoglobulins, exclud- lymphocytes. In addition, both CD81 and SAC treatments resulted ing the involvement of Fc-mediated activities (Fig. 1B). To further in the up-regulation of activation molecule CD71 in the naı¨ve and validate that B cell proliferation was mediated by CD81, recombi- memory B cell subsets (Fig. 2 Middle). nant CD81-LEL protein was combined with HCV-E2 plus MG81, We then measured the concentration of immunoglobulins in SAC, or anti-IgM. The proliferation induced by CD81 engagement was strongly inhibited (50% inhibition at 0.25 ␮g͞ml of CD81- LEL), whereas CD81-LEL did not affect SAC or anti-IgM stimu- Table 1. In vitro analysis of B lymphocytes from healthy donors lation (Fig. 1C). The same results were obtained when HCV-E2 was stimulated by multimeric CD81 engagement replaced by N81 or when CD81-LEL was substituted with an Marker Control* CD81† SAC‡ IgM§ anti-HCV-E2 mAb, which is known to interfere with the binding of HCV-E2 to CD81 (data not shown). CD69 6 Ϯ 181Ϯ 557Ϯ 276Ϯ 10 CD81 engagement resulted in an increased percentage of B cells CD71 17 Ϯ 777Ϯ 756Ϯ 379Ϯ 6 expressing the early activation marker CD69, the transferrin recep- CD86 7 Ϯ 271Ϯ 749Ϯ 475Ϯ 2 tor CD71, and the costimulatory molecule CD86. Similar results CXCR3 8 Ϯ 427Ϯ 44Ϯ 29Ϯ 6 were obtained when well characterized polyclonal B cell stimuli Data are shown as percentage of CD19 cells expressing the markers indi- (anti-IgM or SAC) were used (Table 1). Table 1 also shows that cated on the left (mean Ϯ SD). B lymphocytes from healthy donors (n ϭ 10) CXCR3, which is expressed in a small fraction of resting B cells, is stimulated in vitro for 24 h by using: *, complete medium; †, MG81 mAb (5 up-regulated by anti-CD81 mAbs but not by anti-IgM and SAC. ␮g͞ml) ϩ N81 mAb (5 ␮g͞ml); ‡, SAC (1:30,000 vol͞vol); §, Goat anti-human Taken together, these results demonstrate that multimeric engage- IgM F(abЈ)2 (10 ␮g͞ml).

18546 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.0509402102 Rosa et al. Downloaded by guest on September 30, 2021 Fig. 3. CD81-mediated signaling in human primary B cells. (A) Tonsil-derived B cells were stimulated with 5 ␮g͞ml MG81 plus 5 ␮g͞ml N81 (CD81) or with 10 ␮g͞ml anti-IgM (IgM). After the indicated time, equal amounts of protein extracts where subjected to Western blot analysis using anti-phospho-tyrosine Fig. 2. CD81 engagement induces the preferential expansion of naı¨ve B 4G10 (pTyr), anti-phospho c-Jun (pJun), anti-phospho JNK1 and -2 (pJNK1 and lymphocytes. Purified naive (CD27Ϫ) or memory (CD27ϩ) B cells were incu- pJNK2), and anti-total JNK1 (JNK1) antibodies. Protein standards are ex- bated with complete medium (Control), MG81 and N81 mAbs (5 ␮g͞ml each) pressed on the left in kDa. The asterisk indicates an unidentified protein (CD81), or SAC (1:30,000, vol͞vol). (Top) Proliferation was measured as CFSE cross-reacting with the anti-phospho JNK1 and -2 antibody. (B) Tonsil-derived dilution (x axis) by flow cytometry after 4 days of stimulation. For each sample, B cells were stimulated as described above for the indicated times, and protein 5 ϫ 105 events were acquired. Numbers indicate the percentage of dividing extracts were subjected to Western blot analysis using anti-phospho-CD19 cells. (Middle) Surface expression of CD27 (y axis) and CD71 (x axis) after 3-day (pCD19) and anti-total CD19 (CD19) antibodies. (C) Tonsil-derived B cells were ␮ ͞ ␮ ͞ ␮ ͞ stimulation was measured by flow cytometry. For each sample, 1 ϫ 105 events stimulated with 5 g ml MG81, 5 g ml N81, or 5 g ml both MG81 and N81 were acquired, and numbers in quadrants represent percentages of CD27- and mAbs. After the indicated time, Western blot analysis was performed by using CD71-positive cells. (Bottom) IgM and IgG secretion in cell-culture superna- pJun, anti total c-Jun (Jun), pCD19, and CD19 antibodies. ctr, untreated cells. tants was measured 10 days after stimulation by ELISA. Data are shown as mean Ϯ SD of duplicate wells and are representative of at least five experi- ments performed with independent donors. that CD81-mediated activation of B cells is independent from the B cell receptor and coreceptor complexes. The activation of the JNK pathway, monitored by phospho c-Jun accumulation, was culture supernatants from naı¨ve and memory B cells stimulated specific for multimeric CD81 engagement (MG81 plus N81) and through CD81 or SAC. As expected, activated naı¨ve B cells did not did not occur when MG81 and N81 mAbs were provided alone (Fig. produce IgG but a low (in the case of SAC) or very low (in the case 3C). In contrast, the transient dephosphorylation of CD19 was of CD81) amount of IgM, whereas memory B cells produced a high observed in response to both multimeric engagement and N81 amount of IgM and IgG in response to SAC activation and a low alone (Fig. 3C). amount of IgM and IgG in response to the modest CD81 activation (Fig. 2 Bottom). Chronic HCV Infection Associates with B Lymphocyte Activation. We In summary, these results demonstrate that, although CD81 reasoned that polyclonal B cell activation obtained by multimeric engagement up-regulates activation markers in both the naı¨ve and CD81 engagement in vitro could mimic what occurs in HCV- the memory B cell subsets, it promotes a preferential polyclonal infected patients when whole HCV particles bind CD81 on the B expansion of naı¨ve B lymphocytes. cell surface in vivo. To test this hypothesis, we compared the phenotype of peripheral blood B cells from 64 chronic hepatitis C

CD81-Mediated Activation of B Cells Triggers JNK Activity. We next patients with (n ϭ 26) or without (n ϭ 38) cryoglobulinemia with IMMUNOLOGY assessed the intracellular signaling events induced by multimeric 21 healthy controls and 16 patients with chronic hepatitis B. We CD81 engagement on B cells. We found that, in contrast to BCR found that B cells from the great majority of HCV patients (54 of stimulation by anti-IgM, multimeric CD81 engagement did not 64) expressed elevated levels of the activation markers CD69, result in a significant increase in total tyrosine phosphorylation CD71, and CD86 and of the chemokine receptor CXCR3, whereas levels in purified tonsil-derived B cells (Fig. 3A). Despite the B cells from all 16 patients with chronic hepatitis B had the same extensive difference observed in the tyrosine phosphorylation levels of activation markers and CXCR3 as healthy controls (Table pattern, both CD81 and BCR engagement activated the JNK 2). The up-regulation of the activation markers and CXCR3 did not pathway, as shown by the increased phosphorylation of JNK1 and associate with a particular HCV genotype, RNA levels (Table 3), -2 and by the accumulation of the activated form of c-Jun (pJun) or the presence of cryoglobulinemia (Table 2). In contrast with 2–4 h after treatment (Fig. 3A). When B cells from the same previous observations showing that the frequency of B cells is experiment were monitored for activation markers and prolifera- higher in HCV patients (23), we found the same percentage of B tion, we obtained the same results shown in Table 1 and Fig. 1 for lymphocytes (CD19ϩ cells) in healthy controls and HCV- and PBMC-derived B cells (data not shown). In addition, JNK inhibitor hepatitis B virus (HBV)-infected patients (data not shown). We SP600125 (22) abolished both c-jun activation and proliferation of then analyzed the relative expression of activation markers and tonsil B cells induced by CD81 engagement (see Fig. 6, which is CXCR3 on naı¨ve (CD27Ϫ) and memory (CD27ϩ) subsets from published as supporting information on the PNAS web site). peripheral B cells of healthy controls and HCV patients. We found Because CD81 is part of the coreceptor complex with CD19, we more activated naı¨ve and memory B cells in HCV-infected patients assessed whether multimeric CD81 engagement could signal than in healthy controls; however, the difference in the frequency through the activation of CD19, which occurs through phosphor- of B cells expressing activation markers and CXCR3 between ylation of its cytoplasmic domain. BCR engagement by anti-IgM healthy controls and HCV patients was more pronounced in the stimulated CD19 phosphorylation, whereas CD81 engagement naı¨ve B cell subset (Fig. 4). Because CD27-negative cells up- resulted in a transient inhibition of CD19 phosphorylation (Fig. regulate CD27 a few days after activation, the presence of a sizeable 3B). The absence of a significant change in tyrosine phosphoryla- fraction of CD27-negative cells that expressed activation markers ex tion levels, together with the inhibition of CD19 activity, suggests vivo suggests that a continuous activation of naı¨ve B cells occurs

Rosa et al. PNAS ͉ December 20, 2005 ͉ vol. 102 ͉ no. 51 ͉ 18547 Downloaded by guest on September 30, 2021 Table 2. Ex vivo analysis of B lymphocytes from HCV-infected patients Marker Control* HCV cryo† HCV‡ HBV§

CD27 20 Ϯ 443Ϯ 20 42 Ϯ 20 23 Ϯ 5 CD69 5 Ϯ 133Ϯ 11 35 Ϯ 12 6 Ϯ 2 CD71 19 Ϯ 652Ϯ 16 38 Ϯ 13 17 Ϯ 6 CD86 12 Ϯ 334Ϯ 14 31 Ϯ 78Ϯ 2 CXCR3 9 Ϯ 244Ϯ 16 38 Ϯ 12 9 Ϯ 3

Data are shown as percentage of CD19 cells expressing the markers indi- cated on the left (mean Ϯ SD). *, Healthy donors (n ϭ 21); †, HCV-infected patients with cryoglobulinemia (n ϭ 26); ‡, HCV-infected patients (n ϭ 38); §, HBV-infected patients (n ϭ 16); †, ‡, characteristics of patients are summarized in Table 3. HBV, hepatitis B virus.

during chronic HCV infection. Accordingly, we also found that HCV patients display a higher percentage of memory (CD27ϩ)B cells compared with healthy controls and HBV patients (Table 2). In a previous study performed on cryopreserved peripheral blood lymphocytes from patients with chronic HCV infection, activated B cells were not found (23). We therefore assessed B cell phenotype in peripheral blood from HCV patients on either freshly Fig. 5. Eradication of HCV infection associates with a decreased expression separated cells or the same samples that had been cryopreserved. of activation markers and CXCR3. Ex vivo expression of activation molecules Freshly isolated B cells displayed an increased expression of CD69, (CD69, CD71, and CD86) and the chemokine receptor CXCR3 was measured on CD71, CD86, and CXCR3, as described above. In contrast, cryo- peripheral B cells derived from 22 HCV-infected patients treated with IFN plus preserved B cells from the same samples, although viable and ribavirin therapy (for details, see Materials and Methods) and 21 healthy responding to polyclonal stimuli, did not display the activated subjects (Control). Fifteen patients responded to therapy (Responders), phenotype (see Fig. 7, which is published as supporting information whereas seven did not eradicate HCV infection (Non Responders), as demon- strated by HCV-RNA PCR. B lymphocytes from HCV-infected patients were on the PNAS web site). This finding indicates that assessment of monitored before and soon after the end of therapy. In responders, an activation markers on human B cells should be performed in freshly additional analysis was performed 12 months after the end of therapy (follow- isolated rather than cryopreserved peripheral blood lymphocytes. up). All responders achieved sustained virological response. Data is shown as the percentage of CD19-positive cells expressing the indicated markers Eradication of HCV Associates with a Decreased Number of B Cells (mean Ϯ SD). (*, P Ͻ 0.005 as compared with patients before therapy) Expressing CXCR3 and Activation Markers. To assess whether B cell activation correlates with HCV replication, we investigated the B cell phenotype in 22 patients with chronic HCV infection before expressing CD69, CD71, CD86, and CXCR3 at the end of the therapy was comparable with the levels observed in healthy controls and after IFN and ribavirin therapy. We found that, in patients with Ͻ sustained virologic response (n ϭ 15), eradication of the virus (P 0.005). Importantly, in these patients, the same expression associates with the reduction of the number of B cells expressing pattern persisted for the 12-month follow-up period (Fig. 5 and activation markers and CXCR3. In this group, the number of B cells Table 3). In contrast, in nonresponder patients (n ϭ 7), the percentage of peripheral blood B lymphocytes expressing CD69, CD71, CD86, and CXCR3 did not change significantly before and after the end of the therapy (Fig. 5 and Table 3). Altogether, these data establish a correlation between the presence of HCV virus and the up- regulation of activation markers and CXCR3 on circulating B cells. Discussion We found that engagement of the HCV receptor CD81 activates human B cells in the absence of BCR coligation. CD81-mediated activation differs from other polyclonal B cell stimuli in that it induces preferential proliferation of naı¨ve B cells, whereas anti-IgM and SAC activate naı¨ve and memory B cell proliferation equally well, and CpGs selectively activate memory B cells (24). CD81-mediated B cell activation occurred through a combina- tion of two soluble CD81 ligands, the HCV envelope protein E2 and one anti-CD81 mAb. Because a single antibody should be able to induce capping, our finding indicates that membrane reorganiza- tion is not the only key to the CD81-mediated B cell activation. The multimeric CD81 engagement required for activation of B cells is Fig. 4. HCV patients display high percentage of activated naı¨ve B cells. B in apparent contrast with the requirement of a single CD81 ligand lymphocytes from HCV patients (HCV) and healthy subjects (Control) were and a cross-linking agent in CD81-mediated inhibition of NK cells analyzed ex vivo for the surface expression of activation markers (CD69, CD71, and CD86) and CXCR3 by flow cytometry. For each sample, 2 ϫ 105 events were (17) and costimulation of T lymphocytes (16). However, because acquired. Data is shown as the percentage of naı¨veB cells (CD19ϩ͞CD27Ϫ)or CD81 associates with specific complexes on different cell lineages memory B cells (CD19ϩ͞CD27ϩ) expressing the indicated molecules (mean Ϯ (12), it is likely that CD81 promotes cell-type-restricted signaling SD). Numbers on bars represent the ratio between the level of expression of events that might or might not require multiple ligands or receptor each marker in HCV patients over control. cross-linking.

18548 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.0509402102 Rosa et al. Downloaded by guest on September 30, 2021 CD81 has very short cytoplasmic regions without any recogniz- likely in combination with still unidentified genetic͞environmental able signaling motif and is believed to signal through interaction factors, could evolve into frank lymphoproliferative disorders, such with associated partners. CD19 directly binds to CD81, represent- as type II cryoglobulinemia (5) or non-Hodgkin lymphoma (6, 7). ing an obvious candidate to exert such a function (25). Our data, We found that the chemokine receptor CXCR3 is up-regulated showing that CD19 is not activated during CD81 engagement, on B lymphocytes activated by CD81, whereas CXCR3 is expressed argue against this hypothesis. Other proteins interacting with CD81 at low level on B cells activated by SAC or BCR engagement. This on the B cell surface, namely integrin ␣4␤1 and MHC-II, have been finding correlates with an increased expression of CXCR3 on B identified (26, 27), but more work is required to assess whether cells derived from HCV-infected patients compared with those of these proteins play any role in CD81-mediated B cell activation. We healthy controls. The ligands of CXCR3 (CXCL9, CXCL10, and have found that CD81 engagement activates the JNK pathway CXCL11) are all produced at high levels within hepatitis-C-infected through a mechanism that, at variance with the BCR-dependent liver (29–31). B cells’ migration to the liver in a CXCR3-dependent JNK activation, is not associated with general tyrosine phosphor- fashion together with locally produced cytokines could explain the ylation, as detected by Western blot. The finding that the CD81 B lymphocyte follicles frequently observed in the livers of patients signaling pathway diverges from the BCR pathway upstream of with chronic HCV infection (32). Recently, in agreement with our JNK activation might reveal new targets for the treatment of observations, an independent study has shown that CXCR3 is lymphoproliferative diseases resulting from uncontrolled B cells up-regulated on B cells from HCV-infected patients compared with stimulation. those of healthy controls (33). We have found that chronic HCV infection associates in 85% of It is possible that the HCV–CD81 interaction plays a central role cases with a high percentage of activated B cells and that thera- for virus adaptation to the human host. We speculate that the peutic eradication of HCV coincides with the reduction of this evolutionary pressure on the compact genome of HCV virus has activated B cell phenotype. Our data demonstrate that polyclonal determined the selection of an envelope protein capable of multiple B cell activation is a general feature of chronic HCV infections, functions during infection. These activities include not only binding because B cell activation is observed regardless of the presence of and entry into target cells but also the modulation of the immune mixed cryoglobulinemia. It is tempting to speculate that the poly- system required to establish a chronic infection. The interaction clonal B cell activation obtained by multimeric CD81 engagement with the widely distributed tetraspanin CD81, a promiscuous in vitro mimics what occurs in HCV-infected patients when the virus molecule forming different receptor complexes on different cell binds CD81 on the B cell surface. Notably, our model does not types, may be a multifunctional strategy for HCV to achieve require the assumption that B cells are infected by HCV. Indeed, multiple goals through the same envelope protein. Indeed, CD81 the surface contact between HCV and CD81 could activate B cells engagement by the HCV envelope protein inhibits NK cells (17), in the absence of infection. Our finding that CD81 engagement by costimulates T lymphocytes (16), and activates B cells (this article). HCV envelope protein preferentially activates naı¨ve B lymphocyte Interestingly, it has been recently reported that the engagement of proliferation may be relevant to explain the high rate of autoanti- B cells by purified E2 induced double-strand DNA breaks specif- bodies found in HCV infections (28) and the association of HCV ically in the variable region of Ig (VH) gene locus, leading to with cryoglobulinemia (5). It is possible that the polyclonal-antigen- hypermutation in the VH genes of B cells (34). Altogether, one can independent activation of naı¨ve B cells induced by HCV engage- depict a scenario where interaction between HCV and CD81 ment of CD81 is the first step toward the polyclonal expansion of facilitates the establishment of chronic infection, creating an escape autoreactive clones of naı¨ve B cells that, by becoming memory cells, mechanism for the virus through the inhibition of the innate can be more readily activated in a bystander mode to produce immune response and the dilution of the adaptive immune response autoantibodies (24). In a few cases, this B cell autoreactivity, most resulting from polyclonal expansion.

1. Houghton, M. (1996) in Virology, eds. Fields, B. N., Knipe, D. M. & Howley, P. M. 19. Lyons, A. B. & Parish, C. R. (1994) J. Immunol. Methods 171, 131–137. (Lippincott-Raven, Philadelphia), pp 1035–58. 20. Minutello, M. A., Pileri, P., Unutmaz, D., Censini, S., Kuo, G., Houghton, M., 2. Ralston, R., Thudium, K., Berger, K., Kuo, C., Gervase, B., Hall, J., Selby, M., Kuo, Brunetto, M. R., Bonino, F. & Abrignani, S. (1993) J. Exp. Med. 178, 17–25. G., Houghton, M. & Choo, Q. L. (1993) J. Virol. 67, 6753–6761. 21. Klein, U., Rajewsky, K. & Kuppers, R. (1998) J. Exp. Med. 188, 1679–1689. IMMUNOLOGY 3. World Health Organization (1997) Weekly Epidemiol. Rec. 72, pp. 65–69. 22. Bennett, B. L., Sasaki, D. T., Murray, B. W., O’Leary, E. C., Sakata, S. T., Xu, W., 4. Hoofnagle, J. H. (1997) Hepatology 26, 15S–20S. Leisten, J. C., Motiwala, A., Pierce, S., Satoh, Y., et al. (2001) Proc. Natl. Acad. Sci. 5. Agnello, V., Chung, R. T. & Kaplan, L. M. (1992) N. Engl. J. Med. 327, 1490–1495. USA 98, 13681–13686. 6. Mele, A., Pulsoni, A., Bianco, E., Musto, P., Szklo, A., Sanpaolo, M. G., Iannitto, E., 23. Ni, J., Hembrador, E., Di Bisceglie, A. M., Jacobson, I. M., Talal, A. H., Butera, De Renzo, A., Martino, B., Liso, V., et al. (2003) Blood 102, 996–999. D., Rice, C. M., Chambers, T. J. & Dustin, L. B. (2003) J. Immunol. 170, 7. Engels, E. A., Chatterjee, N., Cerhan, J. R., Davis, S., Cozen, W., Severson, R. K., 3429–3439. Whitby, D., Colt, J. S. & Hartge, P. (2004) Int. J. Cancer 111, 76–80. 24. Bernasconi, N. L., Traggiai, E. & Lanzavecchia, A. (2002) Science 298, 2199– 8. Rosa, D., Campagnoli, S., Moretto, C., Guenzi, E., Cousens, L., Chin, M., Dong, C., 2202. Weiner, A. J., Lau, J. Y., Choo, Q. L., et al. (1996) Proc. Natl. Acad. Sci. USA 93, 25. Matsumoto, A. K., Martin, D. R., Carter, R. H., Klickstein, L. B., Ahearn, J. M. & 1759–1763. Fearon, D. T. (1993) J. Exp. Med. 178, 1407–1417. 9. Cormier, E. G., Tsamis, F., Kajumo, F., Durso, R. J., Gardner, J. P. & Dragic, T. 26. Mannion, B. A., Berditchevski, F., Kraeft, S. K., Chen, L. B. & Hemler, M. E. (1996) (2004) Proc. Natl. Acad. Sci. USA 101, 7270–7274. J. Immunol. 157, 2039–2047. 10. Pileri, P., Uematsu, Y., Campagnoli, S., Galli, G., Falugi, F., Petracca, R., Weiner, A. J., 27. Schick, M. R. & Levy, S. (1993) J. Immunol. 151, 4090–4097. Houghton, M., Rosa, D., Grandi, G. & Abrignani, S. (1998) Science 282, 938–941. 28. Clifford, B. D., Donahue, D., Smith, L., Cable, E., Luttig, B., Manns, M. & 11. Lindenbach, B. D., Evans, M. J., Syder, A. J., Wolk, B., Tellinghuisen, T. L., Liu, C. C., Bonkovsky, H. L. (1995) Hepatology 21, 613–619. Maruyama, T., Hynes, R. O., Burton, D. R., McKeating, J. A. & Rice, C. M. (2005) 29. Helbig, K. J., Ruszkiewicz, A., Semendric, L., Harley, H. A., McColl, S. R. & Beard, Science 309, 623–626. M. R. (2004) Hepatology 39, 1220–1229. 12. Levy, S., Todd, S. C. & Maecker, H. T. (1998) Annu. Rev. Immunol. 16, 89–109. 30. Shields, P. L., Morland, C. M., Salmon, M., Qin, S., Hubscher, S. G. & Adams, D. H. 13. Bradbury, L. E., Kansas, G. S., Levy, S., Evans, R. L. & Tedder, T. F. (1992) (1999) J. Immunol. 163, 6236–6243. J. Immunol. 149, 2841–2850. 31. Harvey, C. E., Post, J. J., Palladinetti, P., Freeman, A. J., Ffrench, R. A., Kumar, 14. Fearon, D. T. & Carter, R. H. (1995) Annu. Rev. Immunol. 13, 127–149. R. K., Marinos, G. & Lloyd, A. R. (2003) J. Leukoc. Biol. 74, 360–369. 15. Carter, R. H. & Fearon, D. T. (1992) Science 256, 105–107. 32. Murakami, J., Shimizu, Y., Kashii, Y., Kato, T., Minemura, M., Okada, K., Nambu, 16. Wack, A., Soldaini, E., Tseng, C., Nuti, S., Klimpel, G. & Abrignani, S. (2001) Eur. S., Takahara, T., Higuchi, K., Maeda, Y., et al. (1999) Hepatology 30, 143–150. J. Immunol. 31, 166–175. 33. Butera, D., Marukian, S., Iwamaye, A. E., Hembrador, E., Chambers, T. J., Di 17. Crotta, S., Stilla, A., Wack, A., D’Andrea, A., Nuti, S., D’Oro, U., Mosca, M., Bisceglie, A. M., Charles, E. D., Talal, A. H., Jacobson, I. M., Rice, C. M. & Dustin, Filliponi, F., Brunetto, R. M., Bonino, F., et al. (2002) J. Exp. Med. 195, 35–41. L. B. (2005) Blood 106, 1175–1182. 18. Heile, J. M., Fong, Y. L., Rosa, D., Berger, K., Saletti, G., Campagnoli, S., Bensi, G., 34. Machida, K., Cheng, K. T., Pavio, N., Sung, V. M. & Lai, M. M. (2005) J. Virol. 79, Capo, S., Coates, S., Crawford, K., et al. (2000) J. Virol. 74, 6885–6892. 8079–8089.

Rosa et al. PNAS ͉ December 20, 2005 ͉ vol. 102 ͉ no. 51 ͉ 18549 Downloaded by guest on September 30, 2021