Geranylgeranyl transferase inhibition stimulates ... - The FASEB Journal

©2005 FASEB

The FASEB Journal express article 10.1096/fj.04-3482fje. Published online June 29, 2005.

Geranylgeranyl transferase inhibition stimulates antimelanoma immune response through MHC class I and costimulatory molecule expression Anne-Françoise Tilkin-Mariamé, Carine Cormary, Nathalie Ferro, Guillaume Sarrabayrouse, Isabelle Lajoie-Mazenc, Jean-Charles Faye, and Gilles Favre INSERM U563 CPTP, Département Innovation Thérapeutique et Oncologie Moléculaire, Institut Claudius Regaud, 20-24 rue du Pt St Pierre, Toulouse F-31052; Université Paul Sabatier, Toulouse F-31062 Corresponding author: Anne-Françoise Tilkin-Mariamé, INSERM U563 CPTP, Département Innovation Thérapeutique et Oncologie Moléculaire, Institut Claudius Regaud, 20-24 rue du Pt St Pierre, Toulouse F-31052; Université Paul Sabatier, Toulouse F-31062. E-mail: [email protected] ABSTRACT Defective antitumor immune responses are frequent consequences of defects in the expression of major histocompatibility complex (MHC) class I and costimulatory molecules. We demonstrated that statins, inhibitors of HMGCoA reductase, enhance mIFN-γ induced expression of MHC class I antigens on murine B16F10 melanoma. GGTI-298, a geranylgeranyl transferase I inhibitor, but not FTI-277, a farnesyl transferase inhibitor, mimics this effect of statins. This effect is related to peptide transporter protein TAP1 up-regulation. Simultaneously, GGTI-298 induces the expression of CD80 and CD86 costimulatory molecules. C3 exoenzyme, which selectively inactivates Rho proteins, phenocopies the effects of GGTI-298, indicating a role for Rho proteins in these events. Furthermore, the treatment of B16F10 cells with GGTI-298 or C3 exoenzyme associated with mIFN-γ induces in vivo tumor growth slowing down in immunocompetent but not in nu/nu syngeneic mice. Both in vivo injections and in vitro restimulation of splenocytes with GGTI-298- and mIFN-γ-treated B16F10 cells induces an enhancement of specific CD8 T lymphocytes labeled by TRP-2/H-2Kb tetramers. Finally, these effects are not limited to mouse models since they were also reproduced in two human melanoma cell lines. These observations indicate that protein geranylgeranylation as well as Rho protein are critical for costimulatory and IFN-γ-dependent MHC class I molecule expression in melanoma. Key words: cancer • immunity • antigen presentation • costimulation • Rho proteins

I

mmunotherapeutic strategies are based on the existence of tumor-associated antigens (TAAs) against which the host is able to mount a specific immune response. CD8 T lymphocytes recognize and kill tumor cells because most of them express peptide/major histocompatibility complex (MHC) class I complexes containing peptides derived from TAAs (1). However, Page 1 of 20 (page number not for citation purposes)

immunosurveillance is often poorly protective against tumorigenicity in vivo, because several mechanisms allow tumor escape (2, 3). Defective death receptor signaling or Fas-ligand expression may contribute to the proliferation of tumor cells or interfere with T cell responses (4). The lack of costimulatory molecule expression by tumor cells and the fact that they generally grow in a noninflammatory microenvironment favor their development in the absence of dendritic cell activation and maturation (5, 6). Furthermore, some tumor cells produce cytokines or chemokines that affect immune cell function (7). But the best documented and the most frequent tumor escape mechanism is the loss or the down-regulation of the expression of tumorderived peptide/MHC class I complexes (8, 9). This can be achieved by a decrease in either the synthesis of MHC class I or the processing of antigens (10). Such deficiencies in the expression of peptide/MHC class I complexes have been described for various murine and human tumors and correlated with tumor progression (11). In vivo murine experiments have demonstrated that deficiency in the expression of peptide transporter protein 1 (TAP1) by tumor cells results in evasion of immune surveillance and increased tumorigenicity (12). Consequently, the development of drugs, which could enhance expression of peptide/MHC class I complexes on tumor cells, would be of definite interest for immunotherapeutic strategies. IFN-γ is secreted by natural killer (NK), CD4, and CD8 T cells after stimulation and is known to enhance the expression of MHC class I and MHC class II antigens. While hIFN-γ has been widely used in clinical protocols, it has shown only limited benefit (13). Recently, statins, inhibitors of 3-hydroxy-3-methylglutaryl CoA (HMG CoA) reductase, have been described as direct inhibitors of MHC class II induction by IFN-γ and therefore are able to act as repressors of MHC class II-mediated T cell activation. However, the involvement of the isoprenoid pathway in this inhibitory effect of statins was not evaluated (14). Here, we have tested the hypothesis that statins could be general modulators of IFN-γ activities and therefore could also be able to modulate the expression of MHC class I antigens induced by IFN-γ. To determine the checkpoint activity of statins on the mevalonate pathway for their effect on IFN-γ activity, we examined the possibility that the effects of statins occur through farnesyl pyrophosphate or geranylgeranyl pyrophosphate depletion impairing prenylation of proteins involved in IFN-γ functions. Indeed, statins inhibit isoprenoid pathways by preventing the reduction of HMG CoA to mevalonate, which is the precursor of isopentenyl pyrophosphate, then successively converted to geranyl pyrophosphate, farnesyl pyrophosphate, and geranylgeranyl pyrophosphate (15). These last two isoprenoid compounds are covalently bound to proteins through the activity of farnesyl and geranylgeranyl transferases, respectively, to numerous proteins, including monomeric GTPases of the Ras superfamily including Rho proteins. This posttranslational protein modification is required for protein functions (16). We have therefore tested the capacity of statins, of the farnesyl transferase inhibitor FTI-277 and of the geranylgeranyl transferase I inhibitor GGTI298, to favor tumor immunogenicity by enhancing membrane expression of MHC class I. The murine B16F10 melanoma cell line, which has deficient MHC class I (H-2 class I) antigen surface expression that can be corrected by mIFN-γ administration, was chosen for this study. B16F10 cells are often used in murine immunotherapeutical studies because they are highly aggressive, metastatic, and demonstrate weak immunogenicity. The underlying mechanism of their deficient MHC class I expression has been recently analyzed. It is attributable to downregulation or loss of the expression and function of multiple components of the MHC class I antigen-processing pathway, including the peptide transporters associated with antigen

Page 2 of 20 (page number not for citation purposes)

processing (TAP1 and TAP2); the proteasome subunits LMP2, LMP7, and LMP10; PA28α and β; and the chaperone tapasin (17). We also tested two human melanoma cell lines, the LB1319MEL cell line, which spontaneously expresses high amounts of HLA class I antigens, and the BB74-MEL cell line, which in contrast expresses low amounts of these molecules. We now present evidence that IFN-γ-induced expression of MHC class I antigens on melanoma tumor cells is increased either by statins or by GGTI-298, but not by FTI-277, and that Rho protein function is involved. According to the two-signal model, optimal T cell activation requires specific antigen recognition by lymphocyte T cell receptors (TCR) and an additional signal through the costimulatory molecules (18). Interestingly, we also demonstrated that GGTI298 and inhibition of Rho function (by C3 or TAT-C3 treatment) were able to induce the expression of CD80 and CD86 costimulatory molecules. Our data identify Rho proteins as new targets for anticancer therapy to stimulate the immune response associated with the expression of MHC class I and costimulatory molecules in melanoma cells. MATERIALS AND METHODS Flow cytometry analysis Membrane expressions were analyzed using FITC-conjugated antibodies specific for H-2Kb (Caltag Laboratories, Burlingame, CA); HLA-A, B, C, (HLA class I); mouse and human CD80, CD86; mouse CD3, mouse and human CD4 and CD8 and PE-conjugated anti-mouse CD27 (BD Biosciences, San Jose, CA) antibodies. Stainings were also performed with PECy5-conjugated TRP-2/H-2Kb tetramer (with the TRP-2181-188 epitope: SVYDFFVWL) (ProImmune, Oxford, UK). All the stainings were performed according to the manufacturer’s instructions and were analyzed on a FACS Calibur (Becton-Dickinson, Franklin Lakes, NJ). To evaluate membrane antigen expression, we determined the mean fluorescence intensities (MFI) and/or the index of specific fluorescence (ISF). The ISF was calculated with the following formula: 100 × (MFI with the specific antibody – MFI with the isotype control/MFI with the isotype control). Data on 1 × 104 cells were collected for analysis, and all experiments were performed at least 3 times. Mice Six- to eight-week-old female C57BL/6JRj or C57BL/6JRj-nu/nu mice were purchased from Elevage Janvier (Le Genest-St-Isle, France) and were maintained in a specific pathogen-free animal facility. Tumor cell lines The B16F10 murine melanoma cell line was used. It is a MHC class I loss metastatic variant of B16, a spontaneous melanoma of C57BL/6 (H-2b). This tumor cell line was maintained in vitro by serial passaging in RPMI 1640 medium (Gibco-BRL, Gaithersburg, MD) supplemented with 10% heat-inactivated fetal calf serum (FCS) (Sigma, St. Louis, MO). To establish s.c. tumors, after extensive washing 1 × 105 cells were injected, in 0.1 ml PBS, into the flank region of syngeneic C57BL/6JRj (H-2b) immunocompetent or athymic nu/nu mice. Tumor growth was determined by measuring tumor diameter in two dimensions every 3 days. All in vivo experiments were performed 3 times with groups of 5 mice. The statistical significance was


determined by Student’s t tests. LB1319-MEL and BB74-MEL are human melanoma cell lines kindly provided by T. Boon (Ludwig Institute Cancer Research, Brussels). These tumor cell lines were also maintained in vitro by serial passaging in RPMI 1640 medium (Gibco-BRL) plus 10% FCS (Sigma). In vitro treatment of tumor cells B16F10 cells (1×105) or LB1319-MEL or BB74-MEL cells (2×105) were cultivated for 4 days in six-well plates (VWR, France) in 2 ml of RPMI 1640 medium plus 10% FCS. Different substances were added alone or combined: murine (PharMingen, San Diego, CA) or human (Roche, Indianapolis, IN) IFN-γ (25 or 50 IU/ml), Lovastatine (30 μM) (Merck, Whitehouse Station, NJ), Atorvastatine (10 μM) (Merck), mevalonolactone (800 μM) (Sigma), FTI-277 (5 or 10 μM) (Calbiochem, San Diego, CA), GGTI-298 (5, 10, 15, or 20 μM) (Calbiochem), or an inhibitor of Rho protein function, the C3 exoenzyme (7.5 or 15 μg/ml). This inhibitor was also used as a recombinant permeant form TAT-C3 (7.5 μg/ml). pTAT C3 was kindly provided by J. Bertoglio (INSERM U 461, France). The recombinant protein was produced in E. coli and purified as described previously (19). Western blot analysis Protein extracts were prepared by the standard procedure and then separated (20 or 30 μg protein/lane) on SDS-PAGE gels. Proteins were blotted onto PVDF membranes. The filters were incubated at 4°C overnight with primary antibodies against TAP1 (Santa Cruz Biotechnology, Santa Cruz, CA), LMP2, or LMP7 (Abcam, Cambridge, UK). The membranes (Bio-Rad, Hercules, CA) were then incubated with horseradish peroxidase (HRP)-labeled secondary antibodies (Jackson ImmunoResearch, West Grove, PA) for 1 h at room temperature and then detected by chemilumiscence. In vitro restimulation of splenocytes with B16F10-treated cells Spleens were harvested from the four groups of immunocompetent mice 7 days after s.c. injection of 105 B16F10 untreated or treated by mIFN-γ and/or GGTI-298 cells. Spleens were minced, and single-cell suspensions were prepared. The splenocytes were restimulated in vitro for 7 days by incubating 2 × 106 splenocytes with 2 × 104 irradiated (120 Gy) and treated tumor cells in 24-well plates (Nunc, Dutscher, France). These stimulations were performed at 37°C in a 5% CO2 humidified atmosphere in RPMI/10% FCS. After 3 days of culture 50 IU/ml hIL-2r (Proleukin, Chiron, Suresnes, France) were added. RESULTS Statins and GGTI-298, but not FTI-277, enhance mIFN-γ-induced MHC class I expression through a mechanism involving TAP1 overexpression Statins, competitive inhibitors of HMG CoA reductase, have been recently reported to act as direct inhibitors of IFN-γ-induced MHC-II expression on human endothelial cells and monocytemacrophages (14), suggesting that compounds of the mevalonate pathway could act as modulators of IFN-γ functions. Here we tested the role of statins on IFN-γ induction of MHC


class I expression in tumor cells. For that purpose we used the MHC class I-negative murine melanoma cell line B16F10 for which mIFN-γ treatment restores MHC class I expression through a recently described mechanism involving antigen processing (17). B16F10 tumor cells were cultivated for 4 days in the presence of the different substances as indicated in Fig. 1. Treatment of B16F10 with Atorvastatine (10 μM) or Lovastatin (data not shown) enhanced the mIFN-γ-induced expression of H-2Kb. This effect was completely reversed by addition of mevalonolactone (800 μM), the first product of HMG CoA reductase. We then used either FTI277 or GGTI-298 to test the hypothesis that this effect occurs through an inhibition of protein prenylation. First, we controlled that these two prenyl transferase inhibitors (FTI-277 and GGTI298) prevented protein isoprenylation using our culture conditions. Indeed, as expected, FTI-277 treatment at 5 or 10 µM resulted in an inhibition of HDJ-2 farnesylation, an exclusively farnesylated protein, yet was without effect on Rap-1A geranylgeranylation, an exclusively geranylgeranylated protein. Whereas GGTI-298 at 5 or 10 µM prevented the prenylation of Rap1A, but not of HDJ-2 (results shown in supplemental data). Furthermore, mIFN-γ alone (50 IU/ml) did not modify Rap-1A or HDJ-2 prenylation (results shown in supplemental data). Then the cells were treated with 50 IU/ml of mIFN-γ in combination with FTI-277 or GGTI-298 at 5 or 10 μM. As shown in Fig. 1A, GGTI-298, but not FTI-277, enhanced the mIFN-γ-induced expression of H-2Kb to a similar extent as that noted with Atorvastatine. Note that while GGTI298 enhanced mIFN-γ-induced MHC class I expression, GGTI-298 alone did not significantly modify H-2Kb expression (Fig. 1A). Furthermore, increasing concentrations of GGTI-298 were without effect on H-2Kb expression (Fig. 1B). In contrast, H-2Kb expression increased in a dose-dependent manner in mIFN-γ-treated cells, reaching a steady-state level at 50 IU/ml. To evaluate the effects of mIFN-γ and GGTI-298 together, we combined 25 and 50 IU/ml mIFN-γ with increasing concentrations of GGTI-298. A dramatic dose-dependent increase of H-2Kb expression was noted in cells treated with mIFN-γ and GGTI-298 (Fig. 1B). These results show that inhibition of protein geranylgeranylation but not of farnesylation potentiates mIFN-γ-induced H-2Kb expression in a dose-dependent manner in the B16F10 melanoma cell line. For the next experiments, we only used GGTI-298. Statins may have additional cellular effects than GGTI-298 because they inhibit the whole mevalonate pathway, including protein farnesylation as well as dolichol, ubiquitin, and cholesterol biosynthesis. The defect of MHC class I expression in B16F10 cells can be attributed to a coordinated downregulation of multiple components of the MHC class I antigen-processing pathway, including the peptide transporter TAP1 and subunits of the multicatalytic proteasome, such as LMP2 and LMP7. Recently, Seliger et al. (17) demonstrated in this tumor model that mIFN-γ administration transcriptionally induces the expression of such antigen-processing machinery components, thereby inducing MHC class I surface expression. We determined TAP1, LMP2, and LMP7 expression after 2 days (data not shown) and after 4 days of treatment with mIFN-γ (50 IU/ml), GGTI-298 (10 μM), or a combination of both, as indicated previously. The expression of TAP1, LMP2, and LMP7 was compared with that of unknown proteins detected by the secondary antibodies in the same protein extracts. Similar results were obtained at day 2 and 4. As expected, mIFN-γ up-regulated the expression of TAP1, LMP2, and LMP7 while GGTI-298 alone had no significant effect (Fig. 1C). A twofold higher level of TAP1 expression was noted in B16F10 cells treated with the combination of GGTI-298 and mIFN-γ relative to that observed Page 5 of 20 (page number not for citation purposes)

with mIFN-γ alone, whereas the addition of GGTI-298 had no significant effect on the expression of LMP2 and LMP7 proteins (Fig. 1C). These results show that the GGTase I inhibition enhances IFN-γ-induced MHC class I expression through a mechanism involving TAP1 overexpression. GGTI-298 increases the antitumor effect of mIFN-γ treatment against B16F10 growth in immunocompetent mice but not in nu/nu mice We tested whether these in vitro combined treatments of B16F10 cells with GGTI-298 and mIFN-γ before injection could improve the antitumor response against B16F10 tumor growth in mice. B16F10 cells were treated for 4 days as indicated previously and washed extensively, and 1 × 105 viable cells were s.c. injected in parallel into syngeneic immunocompetent or athymic nu/nu C57BL/6 mice. Before in vivo injection, we verified that the H-2Kb levels were, as expected, increased after mIFN-γ treatment and overexpressed after combined treatment with mIFN-γ and GGTI-298. In immunocompetent mice, B16F10 cells treated with 50 IU/ml mIFN-γ or with 10 μM GGTI-298 alone grew slower than untreated (control) cells, but this was not statistically significant as shown by Student’s t test (P