6
Preprocalcitonin signal peptide generates a cytotoxic T lymphocyte-defined tumor epitope processed by a proteasome-independent pathway Faten El Hage , Vincent Stroobant , Isabelle Vergnon , Jean-Franc ¸ois Baurain § , Hamid Echchakir †¶ , Vladimir Lazar , Salem Chouaib , Pierre G. Coulie §†† , and Fathia Mami-Chouaib †,††‡‡ Institut National de la Sante ´ et de la Recherche me ´ dicale, Unite ´ 753, Laboratoire Immunologie des Tumeurs humaines: Interaction effecteurs cytotoxiques-syste ` me tumoral, Institut Fe ´de ´ ratif de Recherche-54, Institut Gustave Roussy, F-94805 Villejuif, France; Ludwig Institute for Cancer Research, Brussels Branch, B-1200 Brussels, Belgium; § Universite ´ Catholique de Louvain and de Duve Institute, B-1200 Brussels, Belgium; and Unite ´ de Ge ´ nomique Fonctionnelle, Institut Fe ´de ´ ratif de Recherche-54, Institut Gustave Roussy, F-94805 Villejuif, France Communicated by Jean Dausset, Centre d’Étude du Polymorphisme Humain, Paris, France, May 16, 2008 (received for review November 7, 2007) We identified an antigen recognized on a human non-small-cell lung carcinoma by a cytotoxic T lymphocyte clone derived from autologous tumor-infiltrating lymphocytes. The antigenic peptide is presented by HLA-A2 and is encoded by the CALCA gene, which codes for calcitonin and for the -calcitonin gene-related peptide. The peptide is derived from the carboxy-terminal region of the preprocalcitonin signal peptide and is processed independently of proteasomes and the transporter associated with antigen process- ing. Processing occurs within the endoplasmic reticulum of all tumoral and normal cells tested, including dendritic cells, and it involves signal peptidase and the aspartic protease, signal peptide peptidase. The CALCA gene is overexpressed in medullary thyroid carcinomas and in several lung carcinomas compared with normal tissues, leading to recognition by the T cell clone. This new epitope is, therefore, a promising candidate for cancer immunotherapy. antigen processing signal peptidase signal peptide peptidase T he analysis of tumor-reactive cytotoxic T lymphocytes (CTLs) derived from patients with various solid tumors had led to promising new treatments for malignant diseases, by either expanding the T cells in vitro before transferring them with IL-2 into patients (1) or identifying their target antigens (Ags), which can then be used in therapeutic vaccines. A large number of tumor-associated Ags recognized by CTLs has been identified mainly in malignant melanoma. Unfortunately, clinical studies indicate that, despite an increase in the frequency of antitumor CD8 T cells, the efficacy of current therapeutic vaccines remains limited (2). Current studies are focusing on a better understand- ing of the mechanisms of rare tumor regressions observed (3, 4), the activation state of anti-vaccine CD8 T cells, and their capacity to migrate to the tumor site. Much less is known about the antigenicity and susceptibility to CTL attack of human lung tumors. Most of these tumors are non-small-cell lung carcinomas (NSCLCs), a large group that includes squamous-cell, adeno-cell, and large-cell (LCC) carci- nomas. NSCLCs can be infiltrated by T cell antigen receptor (TCR) / T cells (5). The identified T cell target Ags include peptides encoded by the HER2/neu protooncogene (6), which is overexpressed in many lung tumors, and by several genes that were found to contain a point mutation in tumor cells compared with autologous normal cells. These mutated genes include elongation factor 2 (7), malic enzyme (8), -actinin-4 (9), and NFYC (10). In addition, several cancer/germ-line genes are expressed in NSCLCs (11, 12), which should lead to the presence of tumor-specific Ags at the surface of cancer cells. However, spontaneous T cell responses against MAGE-type Ags have not been observed in lung cancer patients thus far. Therefore, identification of new lung cancer Ags, in particular those shared by tumors of several patients, would help the design and immu- nological monitoring of vaccination strategies in lung cancer. Most antigenic peptides recognized by CD8 T cells originate from degradation in proteasomes of intracellular mature pro- teins and their transport, by the transporter associated with antigen processing (TAP) from the cytosol into the endoplasmic reticulum (ER) (for review, see ref. 13). The resulting peptides of 9 to 10 amino acids bind MHC class I (MHC-I) molecules and are then conveyed to the cell surface. An increasing number of epitopes recognized by tumor-reactive T cells has been reported to result from nonclassical mechanisms acting at the transcrip- tion, splicing, or translational levels (for review, see ref. 14). It is noteworthy that several tumor epitopes are poorly processed by dendritic cells (DCs), which are unique in their capacity to process Ags and to prime CD8 T cells, but which constitutively express immunoproteasomes (15, 16). In this article, we identi- fied an antigenic peptide recognized on a human LCC by an autologous CTL clone. This epitope is derived from the carboxy (C)-terminal region of the calcitonin (CT) precursor signal sequence and is processed by a proteasome-independent path- way involving signal peptidase (SP) and signal peptide peptidase (SPP). Results A CTL Clone Recognizing Autologous Lung Carcinoma Cells. Patient Heu is a now disease-free lung cancer patient 12 years after resection of the primary tumor. LCC cell line IGR-Heu was derived from a tumor resected from the patient in 1996. Mono- nuclear cells infiltrating the primary tumor were isolated and stimulated with irradiated IGR-Heu tumor cells, irradiated autologous EBV-transformed B cells, and IL-2. Responder lymphocytes were cloned by limiting dilution. Several tumor- specific CTL clones were obtained and classified into three groups on the basis of their TCRV usage (5). We previously reported that the first two groups of clones recognized an antigenic peptide encoded by a mutated -actinin-4 gene (9, 17). Here, we analyze the third group of clones, including Heu161, which expresses a V3-J1.2 TCR. CTL Heu161 lysed the autologous tumor cell line, but not autologous EBV-B cells or the NK-target K562 (Fig. 1A). The recognition of IGR-Heu by the CTL clone was inhibited by anti-HLA-A2 mAb (9). Identification of the Gene Encoding the Ag Recognized by Heu161 CTL. A cDNA library from IGR-Heu cells was cloned into expression plasmid pCEP4 (9) and divided into 264 pools of 100 recom- Author contributions: S.C., P.G.C., and F.M.-C. designed research; F.E.H., I.V., and H.E. performed research; J.-F.B. and V.L. contributed new reagents/analytic tools; F.E.H. and V.S. analyzed data; and P.G.C. and F.M.-C. wrote the paper. The authors declare no conflict of interest. Present address: Nokad SA, 91058 Evry, France. †† P.G.C. and F.M.-C. contributed equally to this work. ‡‡ To whom correspondence should be addressed. E-mail: [email protected]. © 2008 by The National Academy of Sciences of the USA www.pnas.orgcgidoi10.1073pnas.0802753105 PNAS July 22, 2008 vol. 105 no. 29 10119 –10124 IMMUNOLOGY

Preprocalcitonin signal peptide generates a cytotoxic T ... · Preprocalcitonin signal peptide generates a cytotoxic ... Jean-Franc¸ois Baurain§, Hamid Echchakir†¶, Vladimir

Embed Size (px)

Citation preview

Page 1: Preprocalcitonin signal peptide generates a cytotoxic T ... · Preprocalcitonin signal peptide generates a cytotoxic ... Jean-Franc¸ois Baurain§, Hamid Echchakir†¶, Vladimir

Preprocalcitonin signal peptide generates a cytotoxicT lymphocyte-defined tumor epitope processed by aproteasome-independent pathwayFaten El Hage†, Vincent Stroobant‡, Isabelle Vergnon†, Jean-Francois Baurain§, Hamid Echchakir†¶, Vladimir Lazar�,Salem Chouaib†, Pierre G. Coulie§††, and Fathia Mami-Chouaib†,††‡‡

†Institut National de la Sante et de la Recherche medicale, Unite 753, Laboratoire Immunologie des Tumeurs humaines: Interaction effecteurscytotoxiques-systeme tumoral, Institut Federatif de Recherche-54, Institut Gustave Roussy, F-94805 Villejuif, France; ‡Ludwig Institute for Cancer Research,Brussels Branch, B-1200 Brussels, Belgium; §Universite Catholique de Louvain and de Duve Institute, B-1200 Brussels, Belgium; and �Unite de GenomiqueFonctionnelle, Institut Federatif de Recherche-54, Institut Gustave Roussy, F-94805 Villejuif, France

Communicated by Jean Dausset, Centre d’Étude du Polymorphisme Humain, Paris, France, May 16, 2008 (received for review November 7, 2007)

We identified an antigen recognized on a human non-small-celllung carcinoma by a cytotoxic T lymphocyte clone derived fromautologous tumor-infiltrating lymphocytes. The antigenic peptideis presented by HLA-A2 and is encoded by the CALCA gene, whichcodes for calcitonin and for the �-calcitonin gene-related peptide.The peptide is derived from the carboxy-terminal region of thepreprocalcitonin signal peptide and is processed independently ofproteasomes and the transporter associated with antigen process-ing. Processing occurs within the endoplasmic reticulum of alltumoral and normal cells tested, including dendritic cells, and itinvolves signal peptidase and the aspartic protease, signal peptidepeptidase. The CALCA gene is overexpressed in medullary thyroidcarcinomas and in several lung carcinomas compared with normaltissues, leading to recognition by the T cell clone. This new epitopeis, therefore, a promising candidate for cancer immunotherapy.

antigen processing � signal peptidase � signal peptide peptidase

The analysis of tumor-reactive cytotoxic T lymphocytes(CTLs) derived from patients with various solid tumors had

led to promising new treatments for malignant diseases, by eitherexpanding the T cells in vitro before transferring them with IL-2into patients (1) or identifying their target antigens (Ags), whichcan then be used in therapeutic vaccines. A large number oftumor-associated Ags recognized by CTLs has been identifiedmainly in malignant melanoma. Unfortunately, clinical studiesindicate that, despite an increase in the frequency of antitumorCD8 T cells, the efficacy of current therapeutic vaccines remainslimited (2). Current studies are focusing on a better understand-ing of the mechanisms of rare tumor regressions observed (3, 4),the activation state of anti-vaccine CD8 T cells, and theircapacity to migrate to the tumor site.

Much less is known about the antigenicity and susceptibility toCTL attack of human lung tumors. Most of these tumors arenon-small-cell lung carcinomas (NSCLCs), a large group thatincludes squamous-cell, adeno-cell, and large-cell (LCC) carci-nomas. NSCLCs can be infiltrated by T cell antigen receptor(TCR) �/� T cells (5). The identified T cell target Ags includepeptides encoded by the HER2/neu protooncogene (6), which isoverexpressed in many lung tumors, and by several genes thatwere found to contain a point mutation in tumor cells comparedwith autologous normal cells. These mutated genes includeelongation factor 2 (7), malic enzyme (8), �-actinin-4 (9), andNFYC (10). In addition, several cancer/germ-line genes areexpressed in NSCLCs (11, 12), which should lead to the presenceof tumor-specific Ags at the surface of cancer cells. However,spontaneous T cell responses against MAGE-type Ags have notbeen observed in lung cancer patients thus far. Therefore,identification of new lung cancer Ags, in particular those sharedby tumors of several patients, would help the design and immu-nological monitoring of vaccination strategies in lung cancer.

Most antigenic peptides recognized by CD8 T cells originatefrom degradation in proteasomes of intracellular mature pro-teins and their transport, by the transporter associated withantigen processing (TAP) from the cytosol into the endoplasmicreticulum (ER) (for review, see ref. 13). The resulting peptidesof 9 to 10 amino acids bind MHC class I (MHC-I) molecules andare then conveyed to the cell surface. An increasing number ofepitopes recognized by tumor-reactive T cells has been reportedto result from nonclassical mechanisms acting at the transcrip-tion, splicing, or translational levels (for review, see ref. 14). Itis noteworthy that several tumor epitopes are poorly processedby dendritic cells (DCs), which are unique in their capacity toprocess Ags and to prime CD8 T cells, but which constitutivelyexpress immunoproteasomes (15, 16). In this article, we identi-fied an antigenic peptide recognized on a human LCC by anautologous CTL clone. This epitope is derived from the carboxy(C)-terminal region of the calcitonin (CT) precursor signalsequence and is processed by a proteasome-independent path-way involving signal peptidase (SP) and signal peptide peptidase(SPP).

ResultsA CTL Clone Recognizing Autologous Lung Carcinoma Cells. PatientHeu is a now disease-free lung cancer patient 12 years afterresection of the primary tumor. LCC cell line IGR-Heu wasderived from a tumor resected from the patient in 1996. Mono-nuclear cells infiltrating the primary tumor were isolated andstimulated with irradiated IGR-Heu tumor cells, irradiatedautologous EBV-transformed B cells, and IL-2. Responderlymphocytes were cloned by limiting dilution. Several tumor-specific CTL clones were obtained and classified into threegroups on the basis of their TCRV� usage (5). We previouslyreported that the first two groups of clones recognized anantigenic peptide encoded by a mutated �-actinin-4 gene (9, 17).Here, we analyze the third group of clones, including Heu161,which expresses a V�3-J�1.2 TCR. CTL Heu161 lysed theautologous tumor cell line, but not autologous EBV-B cells orthe NK-target K562 (Fig. 1A). The recognition of IGR-Heu bythe CTL clone was inhibited by anti-HLA-A2 mAb (9).

Identification of the Gene Encoding the Ag Recognized by Heu161 CTL.A cDNA library from IGR-Heu cells was cloned into expressionplasmid pCEP4 (9) and divided into 264 pools of �100 recom-

Author contributions: S.C., P.G.C., and F.M.-C. designed research; F.E.H., I.V., and H.E.performed research; J.-F.B. and V.L. contributed new reagents/analytic tools; F.E.H. and V.S.analyzed data; and P.G.C. and F.M.-C. wrote the paper.

The authors declare no conflict of interest.

¶Present address: Nokad SA, 91058 Evry, France.

††P.G.C. and F.M.-C. contributed equally to this work.

‡‡To whom correspondence should be addressed. E-mail: [email protected].

© 2008 by The National Academy of Sciences of the USA

www.pnas.org�cgi�doi�10.1073�pnas.0802753105 PNAS � July 22, 2008 � vol. 105 � no. 29 � 10119–10124

IMM

UN

OLO

GY

Page 2: Preprocalcitonin signal peptide generates a cytotoxic T ... · Preprocalcitonin signal peptide generates a cytotoxic ... Jean-Franc¸ois Baurain§, Hamid Echchakir†¶, Vladimir

binant clones. DNA prepared from each pool was transfectedinto 293-EBNA cells, together with an HLA-A*0201 construct.CTL Heu161 was added to the transfectants, and then TNF� wasmeasured. A large proportion (85 of 264) of cDNA pools provedpositive, suggesting that a surprisingly high frequency of �0.4%of cDNA clones encoded the Ag. One pool of cDNA wassubcloned, and a cDNA clone named 150 was isolated (Fig. 1B).

cDNA 150 was 956 bp long and contained a polyadenylationsignal and a poly(A) tail. Its sequence corresponded to that ofgene CALCA, which codes for both the calcium-lowering hor-mone CT and the CT gene-related peptide � (�-CGRP). Aprimary transcript is spliced into either CT or �-CGRP mRNAthrough tissue-specific alternative RNA processing (18). cDNA150 contains the complete CT coding sequence, spanning exons2, 3, and 4 of gene CALCA (Fig. 2A). However, its 5� end differsfrom that of the CT cDNA sequences present in databanks by thepresence of an intronic sequence of 213 nucleotides.

Identification of the Antigenic Peptide. The region coding for theantigenic peptide was identified with truncated cDNA fragmentscloned into expression plasmids and cotransfected with theHLA-A2 construct into 293-EBNA cells. As shown in Fig. 2B, afragment encoding the first 41 residues of preproCT (ppCT)transferred the expression of the Ag, whereas a fragmentencoding the first 35 residues did not. We then prepared a seriesof CT cDNA fragments truncated at their 5� end and engineeredto contain an initiation codon and a Kozak consensus sequence.Screening with the CTL clone indicated that the antigenicpeptide was contained within residues 9–47 (Fig. 2B). Furthertrimming narrowed down the peptide-encoding region to resi-dues 9–38 (Fig. 2B). Among a set of overlapping peptidescovering this region, two were recognized by clone Heu161,VLLQAGSLHA and LVLLQAGSLHA, which are identical,but with an additional Leu in the latter peptide. As shown in Fig.3A, both peptides sensitized HLA-A2 melanoma cells to recog-nition by Heu161, with half-maximal effects obtained with �10nM of peptide. In a lysis assay, the decamer was slightly more

efficient than the 11-mer by a factor of �3 (Fig. 3B). Weconcluded that the optimal peptide recognized by Heu161 wasVLLQAGSLHA or ppCT16–25. It contains one of the consensusHLA-A2 peptide-binding motifs, Leu, Ile, or Met in position 2,but fails to contain the peptide-binding motif, Leu or Val inposition 10, and has therefore a moderate ability to bind toHLA-A2 (data not shown). This peptide corresponds exactly tothe C-terminal part of the ppCT signal peptide (19).

Processing of the Antigenic Peptide. This localization of the peptidein the protein suggested that it could be processed in the ERindependently of proteasomes and TAP. To examine the in-volvement of proteasomes, IGR-Heu cells were treated withspecific proteasome inhibitor epoxomicin (Fig. 4A). Epoxomicin

Fig. 1. CTL clone Heu161 recognizes an Ag expressed by IGR-Heu autologoustumor cells. (A) Cytotoxic activity of CTL Heu161 toward tumor cells IGR-Heu,autologous Heu-EBV B cells, and K562. Cytotoxicity was measured by 51Cr-release assay at indicated E/T ratios. (B) Identification of a cDNA clone encod-ing the Ag recognized by the CTL clone. Heu161 (3,000 cells) was stimulatedfor 24 h by 293-EBNA (30,000 cells) cotransfected with vectors pCEP4 contain-ing cDNA clone 150 and pcDNA3.1 containing HLA-A2. Control stimulator cellsincluded IGR-Heu and 293-EBNA transfected with cDNA 150 or HLA-A2 alone.The concentration of TNF� released in medium was measured. Data arerepresentative of five independent experiments.

Fig. 2. Identification of the gene segment encoding the epitope recognizedby Heu161 CTL. (A) Representation of cDNA 150 compared to the CT and�-CGRP gene and transcripts. Numbered boxes represent exons. Arrows indi-cate forward (O) and reverse (R) primers used in RT-PCR. (B) Minigenes used toidentify the region coding for the antigenic peptide. A series of truncatedconstructs were cotransfected into 293-EBNA cells with HLA-A2. The corre-sponding encoded sequences are shown. Recognition by Heu161 was assessedby using TNF� assay as in Fig. 1B. Statistical analyses were performed by usinga Mann–Whitney U test (P � 0.01). Data are representative of three indepen-dent experiments.

Fig. 3. Identification of the peptide recognized by clone Heu161. (A) CTLstimulation with purified synthetic peptides. Peptides were loaded on allo-geneic HLA-A2 MZ2-MEL.3.1 melanoma cells for 1 h at room temperaturebefore addition of Heu161 at 1/10 E/T ratio. TNF� release was measured 24 hlater. (B) Cytotoxicity of Heu161 toward peptide-pulsed cells. 51Cr-labeledHeu-EBV B cells were incubated over 1 h with the indicated concentrations ofpeptides before addition of CTL at 10:1 E/T ratio. Data are representative offour independent experiments.

10120 � www.pnas.org�cgi�doi�10.1073�pnas.0802753105 El Hage et al.

Page 3: Preprocalcitonin signal peptide generates a cytotoxic T ... · Preprocalcitonin signal peptide generates a cytotoxic ... Jean-Franc¸ois Baurain§, Hamid Echchakir†¶, Vladimir

had no effect on recognition by anti-ppCT CTL (Fig. 4A). Incontrast, it strongly inhibited stimulation of another autologousCTL clone, Heu127, which recognizes a mutated �-actinin-4peptide (9). This finding was expected because �-actinin-4 is acytosolic protein that is degraded, at least in part, in proteasomes(20). These results suggest that the processing of the ppCT16–25peptide does not require proteasomal activity. The involvementof TAP was tested by cotransfecting into 293-EBNA cellsconstructs coding for the antigenic peptide, HLA-A2, and theimmediate-early protein ICP47 of herpes simplex virus type 1,which binds to and inhibits human TAP (21). As shown in Fig.4B, cotransfecting ICP47 had no detectable effect on recognitionof the transfectants by anti-ppCT CTL, whereas it stronglyinhibited that by the anti-�-actinin-4 CTL. These results stronglysuggest that the processing of the ppCT16–25 epitope is TAP-independent.

Because the C terminus of antigenic peptide VLLQAGSLHAcorresponded to the C terminus of the ppCT signal sequence(19), it was expected to be generated by type I SP, which cuts offsignal peptides from secretory proteins on the luminal side of theER membrane (22). SP involvement was tested by using theserine protease inhibitor dichloroisocoumarin (DCI) (23). Re-markably, preincubation of IGR-Heu cells with DCI renderedthem resistant to lysis by the anti-ppCT CTL (Fig. 5A). The sametreatment had only a moderate effect on recognition by theHeu127 clone (Fig. 5A), and this outcome probably resultedfrom a slight decrease in MHC-I expression on DCI-treatedtumor cells (data not shown). These results are compatible withinvolvement of SP in processing the ppCT16–25 peptide. Aftercleavage by SP, some of the signal sequences inserted in the ERmembrane in a type II or loop-like orientation can be furthercleaved by the intramembrane protease SPP (reviewed in ref.24). We therefore specifically knock down SPP expression in

IGR-Heu with two distinct siRNA. siRNA-S1 and siRNA-S2specifically inhibit SPP expression at both RNA (Fig. 5B) andprotein (data not shown) levels. The down-regulation of SPPresulted in a strong decrease in the sensitivity of the tumor cellsto lysis by the anti-ppCT, but not by the anti-�-actinin-4 CTL(Fig. 5C). Similar inhibition was observed when tumor cells wereused to stimulate production of IFN� by CTL (Fig. 5D). Partial

Fig. 4. Processing of the ppCT16–25 peptide is proteasome- and TAP-independent. (A) IGR-Heu cells were incubated in the presence or absence ofthe proteasome inhibitor epoxomicin (10 �M), and then Heu161 cells wereadded at 1/10 E/T ratio. The autologous Heu127 clone was included as apositive control. TNF� released in medium after 24 h of culture was measured.(B) 293-EBNA cells were cotransfected with pCEP4 containing either cDNA 150(Upper) or the mutated �-actinin-4 cDNA (Lower), with HLA-A2 construct, andwith various amounts of vector pBJi-neo containing IPC47 cDNA. Heu161(Upper) or Heu127 (Lower) were then added at 1/10 E/T ratio. TNF� releasedafter 24 h of culture was measured. Controls included 293-EBNA cells trans-fected with HLA-A2 or pBJi-neo-IPC47 alone and incubation of transfectantswith either ppCT or �-actinin-4 peptides. Data correspond to one of fourindependent experiments.

Fig. 5. Processing of the ppCT16–25 epitope involves SP and SPP. (A) Process-ing of the ppCT16–25 epitope is SP-dependent. IGR-Heu cells were incubatedwith the SP inhibitor DCI (250 �M) before addition of anti-ppCT (Left) oranti-�-actinin-4 (Right) CTL. (B) Processing of the ppCT16–25 epitope involvesSPP. Analysis of SPP mRNA expression by real-time RT-PCR analysis. Total RNAextracted from IGR-Heu, electroporated or not with siRNA targeting SPP(siRNA-S1 and siRNA-S2), was reverse-transcribed and quantified by TaqMan.(C) Effect of SPP knockdown on tumor cell recognition. (Left) Lytic activity ofHeu161 and Heu127 against IGR-Heu, electroporated or not with siRNA-S1 orcontrol siRNA, determined by 51Cr-release assay at 10/1 E/T ratio. (Right)Cytotoxicity of Heu161 against IGR-Heu, electroporated or not with siRNA-S1,siRNA-S2, or control siRNA, determined by 51Cr-release assay at indicated E/Tratios. (D) Production of IFN� by Heu161 and Heu127 clones stimulated withtumor cells electroporated or not with siRNA-S1 or control siRNA. (E) TheppCT16–25 peptide is located at the C terminus of the signal sequence of the CThormone precursor. The optimal peptide recognized by Heu161 is boxed.Arrows indicate the SP and the approximate SPP cleavage sites. The n, h, andc regions in the ppCT signal peptide were predicted by using SignalP 3.0software. Data are representative of three independent experiments.

El Hage et al. PNAS � July 22, 2008 � vol. 105 � no. 29 � 10121

IMM

UN

OLO

GY

Page 4: Preprocalcitonin signal peptide generates a cytotoxic T ... · Preprocalcitonin signal peptide generates a cytotoxic ... Jean-Franc¸ois Baurain§, Hamid Echchakir†¶, Vladimir

inhibition of Heu161 reactivity correlates with partial inhibitionof SPP protein expression in siRNA-treated IGR-Heu cells. Thisfinding may be due to the relative stability of SPP homodimersin ER membrane (25). Together, these results indicate that theppCT16–25 peptide was most likely processed by SP and SPPwithin the ER (Fig. 5E).

Expression of the CT Gene Product in Tumor Samples. Expression ofthe CT transcript was tested in a panel of lung carcinoma samplesand cell lines by RT-PCR. Twenty-seven of 209 tumor samplesand 5 of 38 cell lines were positive (Table 1). Quantitative geneexpression analysis of the CT transcript was then carried out onsome of the positive samples (Table 2). Levels of CT geneexpression in the three cell lines tested, namely, LCC IGR-Heu,SCLC DMS53, and medullary thyroid carcinoma (MTC) TT,were at least 100-fold higher than those found in normal humanthyroid. It is noteworthy that the level of expression observed inthe lung carcinoma cell lines was similar to that observed in theMTC cell line (Table 2). High levels of expression of the CTtranscript also were detected in the tumor of patient Heu(Heu-T) and in several lung cancer samples (Table 2).

Next, we tested whether CTL Heu161 also could recognizeother HLA-A2 cells that overexpressed the CT gene. As shownin Fig. 6A, Heu161 efficiently lysed MTC cells TT. As expected,Heu161 did not lyse HLA-A2� DMS53 cells, but did recognizethese cells after transfection with an HLA-A2 construct (Fig.6B). Finally, mature DCs derived from blood monocytes of ahealthy HLA-A2 donor and transfected with the CT cDNAstrongly activated Heu161 CTL (Fig. 6C). We conclude thatprocessing of the peptide ppCT16–25 occurs in all cells tested,namely, NSCLC, SCLC, MTC, melanoma, 293 embryonic kid-ney cells, and DCs. Therefore, it would appear that all cellsexpressing the CT transcript at high levels can be recognized bythe CTL clone described here.

DiscussionCT and �-CGRP polypeptides are encoded by the same gene,CALCA, which includes five introns and six exons (26). Exons 1,2, 3, and 4 are joined to produce the CT mRNA in thyroid C cells,whereas exons 1, 2, 3, 5, and 6 form the �-CGRP mRNA inneuronal cells (27). Mature �-CGRP is an endogenous vasodi-latory peptide widely distributed in the body. CT is a hormoneprimarily involved in protecting the skeleton during periods of‘‘calcium stress’’ (28, 29). In humans, it is synthesized as a ppCT,which includes a signal sequence of 25 amino acids, and proCT,comprising an N-terminal region, CT (32 amino acids), and aC-terminal peptide (26). CT was known to be produced at highlevels by MTC and, more surprisingly, by some lung carcinomas(30, 31). Here, we confirmed, by using quantitative RT-PCR,that gene CALCA was expressed at high levels in several NSCLCand SCLC. It is noteworthy that CT and �-CGRP preprohor-mones share their 75 N-terminal residues encoded by CALCAexons 2 and 3 and that the peptide ppCT16–25 is also the

pp�-CGRP16–25 peptide. It is therefore likely that cells express-ing the �-CGRP, but not the CT transcripts, also can berecognized by CTL such as Heu161.

We have shown here that the signal sequence of the CT and�-CGRP preprohormones contains an antigenic peptide thatcan be specifically recognized by CTL on lung or MTC cellsexpressing the gene CALCA. Induction of CALCA gene expres-sion in other cell types, such as 293 or DC, also results in CTLrecognition. On the basis of the expression profile of its encodinggene, the ppCT peptide is a neuroendocrine differentiation Ag.Several tissue differentiation Ags have been found to be recog-nized by tumor-specific CTL on melanomas. They are encodedby genes with melanocyte-specific expression, such as tyrosinase,Melan-AMART1, Pmel17/gp100, TRP-1, and TRP-2. Other ex-amples include gene PSA in prostate and CEA in gut carcinomas.The ppCT16–25 peptide is the first differentiation Ag recognizedby CTL in lung cancer, and it is a promising candidate forimmunotherapy. Patient Heu mounted a spontaneous CTLresponse to this Ag without clinical autoimmunity. Whether thisremains true with very immunogenic vaccination modalities oradoptive transfer of a high number of specific T cells warrantscareful examination. Relevant information may come fromvaccination studies against MTC carried out with the CTpolypeptide (32).

Table 1. Expression of the CT transcript in lung tumors

Variable Tumor samples Tumor cell lines

NSCLCSCC 7/122 0/3ADC 10/61 0/7LCC 2/8 1/5Undifferentiated carcinomas 1/3 —

SCLC 3/5 4/23Neuroendocrine tumors 3/6 —Bronchioalveolar tumors 1/4 —

CT gene expression was tested by RT-PCR.

Table 2. Relative expression of CT transcript in tumor cell linesand samples

Variable Histological typeRelative expression

of CT transcript

Tumor cell linesIGR-Heu LCC 191.34DMS53 SCLC 116.97TT MTC 259.57

Tumor samplesNSCLC

1 (Heu-T) LCC 14.932 LCC 0.023 SCC 0.284 SCC 0.825 SCC 0.116 SCC 0.177 SCC 0.028 ADC 19.439 ADC 4.86

10 ADC 0.0211 ADC 0.0012 ADC 12.8213 ADC 9.9214 ADC 29.6515 ADC 1.0016 ADC 7.2617 Undifferentiated 7.5718 Undifferentiated 13.64

SCLC19 SCLC 0.1420 Neuroendocrine 2.2722 Neuroendocrine 0.7423 Bronchioalveolar 0.02

Normal tissuesPool of human lung Lung 0.00Pool of human thyroid Thyroid 1.15

Quantitative RT-PCR analysis of CT transcript in tumor cell lines and sam-ples. Normalized copy numbers of CT transcript are shown. The values of CTtranscript that are statistically elevated were shown in bold (P � 0.0001according to Mann–Whitney U test).

10122 � www.pnas.org�cgi�doi�10.1073�pnas.0802753105 El Hage et al.

Page 5: Preprocalcitonin signal peptide generates a cytotoxic T ... · Preprocalcitonin signal peptide generates a cytotoxic ... Jean-Franc¸ois Baurain§, Hamid Echchakir†¶, Vladimir

It is possible that gene CALCA is overexpressed in someneuroendocrine tumors compared with normal thyroid C cells,which would increase the tumoral selectivity of CTL such asHeu161. Our observation by using quantitative RT-PCR thattumor cell lines express 100- to 200-fold higher levels of CALCAtranscripts than normal thyroid tissue, which can be estimated tocontain �1% of CT-producing C cells, does not support theconcept of overexpression in tumoral versus normal cells. How-ever, during the screening of the tumor cDNA library, we weresurprised by the high proportion of CT transcripts. Along thesame line, during our transfection experiments with cDNA clone150, we observed that a high level of gene expression wasrequired for recognition by Heu161. At this stage, and alsoconsidering the unusual mechanism of processing of the anti-genic peptide the efficiency of which could be low, we favor thehypothesis that high levels of CALCA gene expression are indeedrequired for recognition by anti-ppCT16–25 CTL, but that suchlevels also might be present in some normal cells.

Another interesting aspect of the ppCT16–25 epitope lies in itsprocessing. Like that of other proteins, the leader sequence ofppCT is markedly hydrophobic. It mediates binding of theprotein to the membrane of the ER, where nascent polypeptideprecursors are processed (for review, see ref. 33). It is thenimmediately cleaved by SP (22) at the Ala25–Ala26 site (19).ppCT16–25 peptide is at the C terminus of the leader sequence.Several results point to involvement of SP in its processing: (i) theexact match between the SP cleavage site and the peptide Cterminus (Fig. 3A), (ii) the processing that is independent of

proteasomes and TAP (Fig. 4 A and B) and that thereforepresumably occurs in the ER, and (iii) the effect of DCI, whichis known to inhibit SP (Fig. 5A). That SP generates the Cterminus of the antigenic peptide probably explains why mini-genes coding for peptides shorter than 9–38 did not conferantigenicity, although the antigenic peptide was much smaller(Fig. 2). We have probably defined the sequence that is necessaryand sufficient to provide the conformation required for SPactivity. Although it is well known from peptide elution ex-periments that MHC-I molecules can be loaded with peptidesderived from leader sequences (34, 35) and that some of thesepeptides can be targeted by CTL (36, 37), little is known aboutthe exact mechanisms of processing of these antigenic peptides.Our results are evidence for direct involvement of SP in thisprocessing.

After release from precursor proteins by cleavage with SP,some signal peptides with a type II orientation (i.e., thosespanning the ER membrane with the n region exposed towardthe cytosol and the c region facing the ER lumen) (24) canundergo intramembrane proteolysis and be cleaved in the centerof their h region by the presenilin-type aspartic protease SPP(38). Thereby, SPP promotes the release of signal peptidefragments from the ER membrane. After cleavage by SPP, signalpeptide fragments can be released into either the cytoplasm, tobe processed by the proteasome/TAP pathway, or the ER, wherethey follow TAP-independent processing (24). The substratespectrum of SPP is thus far limited to a variety of viral proteins,such as hepatitis C virus (39) and GB virus (40), and signalpeptides, such as preprolactin (41) and human MHC-I (42). Ourresults strongly suggest that the ppCT signal peptide is anothersubstrate of SPP and that this protease directly processes theppCT16–25 CTL epitope.

Thus far, the only known peptides loaded on HLA moleculesafter processing by SPP are derived from MHC-I. Peptidesprocessed by SPP from the N-terminal portion of the signalsequences of HLA-A, HLA-B, and HLA-C can be loaded ontononclassical HLA-E molecules (42). This loading is required forHLA-E transfer to the cell surface, where these molecules canbind the NK inhibitory receptors CD94/NKG2 and block NKactivity (43). It was shown that HLA-A, HLA-B, and HLA-Cpeptides were released into the cytosol and required furtherprocessing by proteasomes and transfer into the ER throughTAP before HLA-E loading (44). The CTL epitope identified inthe present study derives from the C terminus of the ppCT signalsequence (Fig. 5E). Therefore, it is probably released directlyinto the ER and thus does not require proteasomes and TAP forits processing. The proteasome/TAP-independent Ag processingpathway leading to CTL recognition of tumor cells seems tooperate in all of the cells we tested, including DC. It may lead tonew Ag delivery strategies.

MethodsCells and Functional Assays. The IGR-Heu cell line was derived from an LCCsample of patient Heu (9). The Heu161 clone was derived from autologoustumor-infiltrating lymphocytes (5).

Cytotoxic activity was measured by a conventional 4-h 51Cr-release assay(45). IGR-Heu, Heu-EBV, K562, TT, and DMS53 (European Collection of CellCultures) cell lines were used as targets. TNF� was measured by using theTNF-sensitive WEHI-164c13 cells (46).

Construction and Screening of the cDNA Library. The cDNA library from IGR-Heutumor cells was constructed as described previously (9). Plasmid DNA wasextracted and cotransfected, together with the expression vector pcDNA3.1(Invitrogen) containing an HLA-A*0201 cDNA, into 293-EBNA cells (30,000cells per well; Invitrogen). After 24 h, Heu161 (3,000 cells per well) was added.After another 24 h, half of the medium was collected, and its TNF� content wasmeasured.

Fig. 6. Recognition of allogeneic cells overexpressing CT by Heu161 CTL. (A)Cytotoxicity of CTL Heu161 against allogeneic MTC (TT) and SCLC (DMS53) celllines. IGR-Heu cells were included as control. (B) Recognition of HLA-A2-transfected DMS53 by Heu161. DMS53 cells were transfected with HLA-A2before addition of CTL clone at 1/10 E/T ratio. (C) Recognition of mature DCexpressing CT. Monocytes were isolated from the blood of an HLA-A2 healthydonor by using magnetic beads and cultured for 6 days in the presence of 100ng/ml rIL-4 and 250 ng/ml GM-CSF. After maturation by adding 20 ng/ml TNF�

for another 3 days, the DC (30,000 cells per well) were transfected with cDNAclone 150 in pCEP4, and the amount of TNF� released by Heu161 (3,000 cellsper well) was measured 24 h later. Data are representative of three indepen-dent experiments.

El Hage et al. PNAS � July 22, 2008 � vol. 105 � no. 29 � 10123

IMM

UN

OLO

GY

Page 6: Preprocalcitonin signal peptide generates a cytotoxic T ... · Preprocalcitonin signal peptide generates a cytotoxic ... Jean-Franc¸ois Baurain§, Hamid Echchakir†¶, Vladimir

Sequence Analysis and Localization of the Antigenic Peptide. cDNA clone 150was sequenced as described previously (9). To identify the antigenic peptide-encoding region, a panel of cDNA fragments was amplified from cDNA 150 byPCR. PCR products were cloned into expression plasmid pcDNA3.1 by using theEukaryotic TOPO TA cloning kit (Invitrogen) and then transferred into thepCEP4 expression vector to allow overexpression.

Chemical Reagents and RNAi. For proteasomes and SP inhibition, 106 tumorcells were resuspended in media in the presence or absence of specific inhib-itors. Briefly, cells were incubated for 2 h at 37°C either with epoxomicin or DCI(Sigma), washed, resuspended in acid buffer, and then incubated for addi-tional 3 h in the presence or absence of inhibitors. None of the inhibitors wastoxic at the given concentrations. For SPP inhibition, we used siRNA targetinghuman SPP, siRNA-S1 (5-GACAUGCCUGAAACAAUCAtt-3), and siRNA-S2 (5-UGAUUGUUUCAGGCAUGUCtg-3) (Ambion). Nontargeting siRNA was used asa negative control as described previously (45).

RT-PCR Analyses. RT-PCR were performed as described previously (47). For-ward primer O (5�-ggt gtc atg ggc ttc caa aag t) and reverse primer R (5�-atc agc

aca ttc aga agc agg a) (Fig. 2A) were used. PCR conditions were 5 min at 94°C,followed by 30 cycles consisting of 1 min at 94°C, 2 min at 63°C, 2 min at 72°C,and a final elongation step of 10 min at 72°C.

Quantitative PCR analysis was performed by using the forward primer5�-atc ttg gtc ctg ttg cag gc located at the 5� end of exon 2 and the reverseprimer 5�-tgg agc cct ctc tct ctt gct located at the 3� end of exon 3 of theCALCA gene. The Taq-man probe primer was Fam 5�-cct cct gct ggc tgc actggt g-3� Tamra. The amount of RNA samples was normalized by theamplification of RNA 18S. PCR amplifications were performed as describedpreviously (47).

ACKNOWLEDGMENTS. We thank S. Depelchin, C. Richon, and Dr. D. Grunen-wald for their help; Drs. K.-I. Hanada, D. Valmori, M. Ayyoub, C. Pinilla, P.Romero, J. Riond, J.-E. Gairin, S. Stevanovic, and P. Van Endert for helpfuldiscussions; and Dr. V. Braud for critical reading of the manuscript. This workwas supported by Institut National de la Sante et de la Recherche medicale,Institut Gustave Roussy, Association de la Recherche contre le Cancer, LigueNationale Francaise de Recherche contre le Cancer, Fondation de France, andCanceropole île de France and the Institut National du Cancer grants; andAssociation de la Recherche contre le Cancer and Institut National du Cancerfellowships (to F.E.H.).

1. Gattinoni L, Powell DJ, Jr, Rosenberg SA, Restifo NP (2006) Adoptive immunotherapyfor cancer: Building on success. Nat Rev Immunol 6:383–393.

2. Rosenberg SA, Yang JC, Restifo NP (2004) Cancer immunotherapy: Moving beyondcurrent vaccines. Nat Med 10:909–915.

3. Germeau C, et al. (2005) High frequency of antitumor T cells in the blood of melanomapatients before and after vaccination with tumor antigens. J Exp Med 201:241–248.

4. Lurquin C, et al. (2005) Contrasting frequencies of antitumor and anti-vaccine T cells inmetastases of a melanoma patient vaccinated with a MAGE tumor antigen. J Exp Med201:249–257.

5. Echchakir H, et al. (2000) Evidence for in situ expansion of diverse antitumor-specificcytotoxic T lymphocyte clones in a human large cell carcinoma of the lung. Int Immunol12:537–546.

6. Yoshino I, et al. (1994) HER2/neu-derived peptides are shared antigens among humannon-small cell lung cancer and ovarian cancer. Cancer Res 54:3387–3390.

7. Hogan KT, et al. (1998) The peptide recognized by HLA-A68.2-restricted, squamous cellcarcinoma of the lung-specific cytotoxic T lymphocytes is derived from a mutatedelongation factor 2 gene. Cancer Res 58:5144–5150.

8. Karanikas V, et al. (2001) High frequency of cytolytic T lymphocytes directed against atumor-specific mutated antigen detectable with HLA tetramers in the blood of a lungcarcinoma patient with long survival. Cancer Res 61:3718–3724.

9. Echchakir H, et al. (2001) A point mutation in the alpha-actinin-4 gene generates anantigenic peptide recognized by autologous cytolytic T lymphocytes on a human lungcarcinoma. Cancer Res 61:4078–4083.

10. Takenoyama M, et al. (2006) A point mutation in the NFYC gene generates an antigenicpeptide recognized by autologous cytolytic T lymphocytes on a human squamous celllung carcinoma. Int J Cancer 118:1992–1997.

11. Weynants P, et al. (1994) Expression of mage genes by non-small-cell lung carcinomas.Int J Cancer 56:826–829.

12. Jang SJ, et al. (2001) Activation of melanoma antigen tumor antigens occurs early inlung carcinogenesis. Cancer Res 61:7959–7963.

13. Rock KL, Goldberg AL (1999) Degradation of cell proteins and the generation of MHCclass I-presented peptides. Annu Rev Immunol 17:739–779.

14. Mayrand SM, Green WR (1998) Non-traditionally derived CTL epitopes: Exceptions thatprove the rules? Immunol Today 19:551–556.

15. Morel S, et al. (2000) Processing of some antigens by the standard proteasome but notby the immunoproteasome results in poor presentation by dendritic cells. Immunity12:107–117.

16. Chapatte L, et al. (2006) Processing of tumor-associated antigen by the proteasomes ofdendritic cells controls in vivo T-cell responses. Cancer Res 66:5461–5468.

17. Echchakir H, et al. (2002) Cytotoxic T lymphocytes directed against a tumor-specificmutated antigen display similar HLA tetramer binding but distinct functional avidityand tissue distribution. Proc Natl Acad Sci USA 99:9358–9363.

18. Amara SG, et al. (1982) Alternative RNA processing in calcitonin gene expressiongenerates mRNAs encoding different polypeptide products. Nature 298:240–244.

19. Le Moullec JM, et al. (1984) The complete sequence of human preprocalcitonin. FEBSLett 167:93–97.

20. Goldberg AL, Rock KL (1992) Proteolysis, proteasomes and antigen presentation.Nature 357:375–379.

21. Banks TA, et al. (1994) Vaccination with the immediate-early protein ICP47 of herpessimplex virus-type 1 (HSV-1) induces virus-specific lymphoproliferation, but fails toprotect against lethal challenge. Virology 200:236–245.

22. Dalbey RE, Lively MO, Bron S, van Dijl JM (1997) The chemistry and enzymology of thetype I signal peptidases. Protein Sci 6:1129–1138.

23. Rusbridge NM, Beynon RJ (1990) 3,4-Dichloroisocoumarin, a serine protease inhibitor,inactivates glycogen phosphorylase b. FEBS Lett 268:133–136.

24. Martoglio B, Dobberstein B (1998) Signal sequences: More than just greasy peptides.Trends Cell Biol 8:410–415.

25. Nyborg AC, et al. (2004) Signal peptide peptidase forms a homodimer that is labeledby an active site-directed gamma-secretase inhibitor. J Biol Chem 279:15153–15160.

26. Rosenfeld MG, et al. (1983) Production of a novel neuropeptide encoded by thecalcitonin gene via tissue-specific RNA processing. Nature 304:129–135.

27. Morris HR, et al. (1984) Isolation and characterization of human calcitonin gene-related peptide. Nature 308:746–748.

28. Stevenson JC, et al. (1979) A physiological role for calcitonin: Protection of thematernal skeleton. Lancet 2:769–770.

29. Austin LA, Heath H III (1981) Calcitonin: Physiology and pathophysiology. N Engl J Med304:269–278.

30. Coombes RC, Hillyard C, Greenberg PB, MacIntyre I (1974) Plasma-immunoreactive-calcitonin in patients with non-thyroid tumours. Lancet 1:1080–1083.

31. Milhaud G, et al. (1974) Letter: Hypersecretion of calcitonin in neoplastic conditions.Lancet 1:462–463.

32. Schott M, et al. (2002) Calcitonin-specific antitumor immunity in medullary thyroidcarcinoma following dendritic cell vaccination. Cancer Immunol Immunother 51:663–668.

33. Johnson AE, van Waes MA (1999) The translocon: A dynamic gateway at the ERmembrane. Annu Rev Cell Dev Biol 15:799–842.

34. Wei ML, Cresswell P (1992) HLA-A2 molecules in an antigen-processing mutant cellcontain signal sequence-derived peptides. Nature 356:443–446.

35. Henderson RA, et al. (1992) HLA-A2.1-associated peptides from a mutant cell line: asecond pathway of antigen presentation. Science 255:1264–1266.

36. Aldrich CJ, et al. (1994) Identification of a Tap-dependent leader peptide recognizedby alloreactive T cells specific for a class Ib antigen. Cell 79:649–658.

37. Hombach J, Pircher H, Tonegawa S, Zinkernagel RM (1995) Strictly transporter ofantigen presentation (TAP)-dependent presentation of an immunodominant cyto-toxic T lymphocyte epitope in the signal sequence of a virus protein. J Exp Med182:1615–1619.

38. Weihofen A, et al. (2002) Identification of signal peptide peptidase, a presenilin-typeaspartic protease. Science 296:2215–2218.

39. McLauchlan J, Lemberg MK, Hope G, Martoglio B (2002) Intramembrane proteolysispromotes trafficking of hepatitis C virus core protein to lipid droplets. EMBO J21:3980–3988.

40. Targett-Adams P, et al. (2006) Signal peptide peptidase cleavage of GB virus B coreprotein is required for productive infection in vivo. J Biol Chem 281:29221–29227.

41. Martoglio B, Graf R, Dobberstein B (1997) Signal peptide fragments of preprolactin andHIV-1 p-gp160 interact with calmodulin. EMBO J 16:6636–6645.

42. Lemberg MK, et al. (2001) Intramembrane proteolysis of signal peptides: An essentialstep in the generation of HLA-E epitopes. J Immunol 167:6441–6446.

43. Braud VM, et al. (1998) HLA-E binds to natural killer cell receptors CD94/NKG2A, B andC. Nature 391:795–799.

44. Bland FA, et al. (2003) Requirement of the proteasome for the trimming of signalpeptide-derived epitopes presented by the nonclassical major histocompatibility com-plex class I molecule HLA-E. J Biol Chem 278:33747–33752.

45. Le Floc’h A, et al. (2007) {alpha}E{beta}7 integrin interaction with E-cadherin promotesantitumor CTL activity by triggering lytic granule polarization and exocytosis. J ExpMed 204:559–570.

46. Espevik T, Nissen-Meyer J (1986) A highly sensitive cell line, WEHI 164 clone 13, formeasuring cytotoxic factor/tumor necrosis factor from human monocytes. J ImmunolMethods 95:99–105.

47. Lazar V, et al. (1999) Expression of the Na�/I- symporter gene in human thyroid tumors:a comparison study with other thyroid-specific genes. J Clin Endocrinol Metab84:3228–3234.

10124 � www.pnas.org�cgi�doi�10.1073�pnas.0802753105 El Hage et al.