The Juxtamembrane Domain of Butyrophilin BTN3A1 Controls Phosphoantigen-Mediated Activation of Vγ9Vδ2 T Cells

This information is current as Cassie-Marie Peigné, Alexandra Léger, Marie-Claude of September 28, 2021. Gesnel, Fabienne Konczak, Daniel Olive, Marc Bonneville, Richard Breathnach and Emmanuel Scotet J Immunol published online 1 May 2017 http://www.jimmunol.org/content/early/2017/04/29/jimmun

ol.1601910 Downloaded from

Supplementary http://www.jimmunol.org/content/suppl/2017/04/29/jimmunol.160191 Material 0.DCSupplemental http://www.jimmunol.org/

Why The JI? Submit online.

• Rapid Reviews! 30 days* from submission to initial decision

• No Triage! Every submission reviewed by practicing scientists

• Fast Publication! 4 weeks from acceptance to publication by guest on September 28, 2021

*average

Subscription Information about subscribing to The Journal of Immunology is online at: http://jimmunol.org/subscription Permissions Submit copyright permission requests at: http://www.aai.org/About/Publications/JI/copyright.html Email Alerts Receive free email-alerts when new articles cite this article. Sign up at: http://jimmunol.org/alerts

The Journal of Immunology is published twice each month by The American Association of Immunologists, Inc., 1451 Rockville Pike, Suite 650, Rockville, MD 20852 Copyright © 2017 by The American Association of Immunologists, Inc. All rights reserved. Print ISSN: 0022-1767 Online ISSN: 1550-6606. Published May 1, 2017, doi:10.4049/jimmunol.1601910 The Journal of Immunology

The Juxtamembrane Domain of Butyrophilin BTN3A1 Controls Phosphoantigen-Mediated Activation of Human Vg9Vd2 T Cells

Cassie-Marie Peigne´,*,†,1 Alexandra Le´ger,* Marie-Claude Gesnel,* Fabienne Konczak,* Daniel Olive,‡ Marc Bonneville,*,† Richard Breathnach,*,2 and Emmanuel Scotet*,†,2

Vg9Vd2 T lymphocytes are the major human peripheral gd T cell subset, with broad reactivity against stressed human cells, including tumor cells. Vg9Vd2 T cells are specifically activated by small phosphorylated metabolites called phosphoantigens (PAg). Stress-induced changes in target cell PAg levels are specifically detected by butyrophilin (BTN)3A1, using its intracellular B30.2 domain. This leads to the activation of Vg9Vd2 T cells. In this study, we show that changes in the juxtamembrane domain of

BTN3A1, but not its transmembrane domain, induce a markedly enhanced or reduced gd T cell reactivity. There is thus a specific Downloaded from requirement for BTN3A1’s juxtamembrane domain for correct gd T cell–related function. This work identified, as being of particular importance, a juxtamembrane domain region of BTN3A molecules identified as a possible dimerization interface and that is located close to the start of the B30.2 domain. The Journal of Immunology, 2017, 198: 000–000.

n healthy human adults, ∼5% of peripheral blood lymphocytes PAg levels artificially, thus enhancing Vg9Vd2 T cell activation.

are accounted for by the major peripheral gd T cell subset, The mechanisms underlying the antigenic activation of Vg9Vd2 http://www.jimmunol.org/ I which carries a TCR composed of a Vd2 chain systematically T cells induced by PAg have only recently begun to be clari- paired to a Vg9 chain (1–4). Human Vg9Vd2 T cells recognize fied. The ubiquitous butyrophilin (BTN) CD277/BTN3A a wide array of infected or transformed cells and display rapid were shown to be indispensable for the PAg-induced activation and broad functional responses following activation (5). They are of Vg9Vd2 T cells (10–12). In , the BTN3A subfamily strongly and specifically activated by small nonpeptidic phos- contains three members: BTN3A1, -A2, and -A3. Their extra- phorylated compounds called phosphoantigens (PAg). These are cellular domains are composed of two Ig-like domains (IgV and metabolites of the mammalian mevalonate or the microbial IgC domains). Both BTN3A1 and BTN3A3, but not BTN3A2, deoxyxylulose phosphate pathways (6–8). Enhanced tumor cell rec- contain an intracellular B30.2 domain (13), and it is specifically

ognition by Vg9Vd2 T cells correlates with upregulation of in- the BTN3A1 isoform on-target cells that plays a mandatory role in by guest on September 28, 2021 tracellular PAg levels resulting from increased cell metabolism Vg9Vd2 T cell recognition (10). It has been proposed that direct (9). Aminobisphosphonate (NBP) treatment of cells can be used to loading of PAg onto the BTN3A IgV ectodomain exposed on the mimic this phenomenon. These pharmacological agents block surface of target cells creates a complex that interacts with metabolism of the PAg isopentenyl pyrophosphate and so increase Vg9Vd2 TCR to trigger antigenic activation (14). If so, BTN3A would act as a classic Ag-presenting molecule. However, several recent publications have shown that PAg act intracellularly, *Centre de Recherche en Cance´rologie et Immunologie Nantes-Angers, INSERM, through binding to the internal B30.2 domain of BTN3A1, and CNRS, Universite´ d’Angers, Universite´ de Nantes, 44035 Nantes, France; thus do not behave as classic Ag (15–18). It is internal sensing of †Labex-Immunotherapy Graft Oncology, Nantes 44000, France; and ‡CNRS, INSERM, Institut Paoli-Calmettes, Centre de Recherche en Cance´rologiedeMar- changes in target cell PAg levels by BTN3A1 that represents an seille, Immunite´ et E´ quipe Cancer, Aix-Marseille Universite´, 13284 Marseille, important step in Vg9Vd2 T cell recognition. This led to prop- France osition of an inside–out signaling model, where PAg binding 1 Current address: Okinawa Institute of Science and Technology, Structural Cellular to the BTN3A1 intracellular B30.2 domain leads to a change Biology Unit Okinawa, Okinawa, Japan. in the presentation of the BTN3A1 extracellular domain, leading 2R.B. and E.S. contributed equally as cosenior authors. in turn to recognition by Vg9Vd2 T cells (12). The importance ORCIDs: 0000-0002-1170-0323 (C.-M.P.); 0000-0003-1299-4113 (D.O.). of the extracellular domain and intracellular B30.2 domain of Received for publication November 8, 2016. Accepted for publication March 30, BTN3A1 for its activity has now been established. In this study 2017. we have used chimeric BTN3A molecules to investigate the con- This work was supported by funding from INSERM, CNRS, Universite´ de Nantes, Re´gion Pays de la Loire (to C.-M.P.), Agence Nationale de la Recherche Project tributions of the BTN3A1 transmembrane and juxtamembrane GDSTRESS, and by Ligue Nationale contre le Cancer and Investissements d’Avenir (JTM) domains to the antigenic activation of human Vg9Vd2 (Agence Nationale de la Recherche–Programme Laboratoires d’Excellence Immu- T cells. We provide evidence that the BTN3A1 JTM domain notherapy Graft Oncology) Grant ANR-11-LABX-0016-01. is involved in controlling Vg9Vd2 T cell activation induced by Address correspondence and reprint requests to Dr. Emmanuel Scotet, INSERM UMR1232, CNRS ERL, Centre de Recherche en Cance´rologie et Immunologie target cells. Nantes-Angers, IRS_UN, 8 quai Moncousu, 44007 Nantes cedex 1, France. E-mail address: [email protected] Materials and Methods The online version of this article contains supplemental material. Reagents and flow cytometry Abbreviations used in this article: BTN, butyrophilin; JTM, juxtamembrane; NBP, aminobisphosphonate; PAg, phosphoantigen; sh, short hairpin. L-Glutamine, leucoagglutinin, streptomycin, penicillin, monensin, DMEM, and RPMI 1640 media were from Sigma-Aldrich. Recombinant human IL-2 Copyright Ó 2017 by The American Association of Immunologists, Inc. 0022-1767/17/$30.00 (Proleukin) was from Chiron. Zoledronic acid monohydrate (zoledronate

www.jimmunol.org/cgi/doi/10.4049/jimmunol.1601910 2 BTN3A JUXTADOMAINS CONTRIBUTE TO Vg9Vd2 T CELL ACTIVATION

[Zometa]) was obtained from Novartis Pharma. Mevastatin was purchased coculture them with allogeneic human Vg9Vd2 T cells and com- from Sigma-Aldrich. mAbs were from Beckman Coulter (anti-Vd2, no. pare their specific activation (CD107a expression). To avoid any IMMU389) or BD Biosciences (CD107a, no. H4A3). Mouse anti-CD277 complication due to the variability of the activation levels in the mAbs (clones 20.1 and 103.2) were generated, produced, and purified as described (19). Flow cytometry data were collected on a FACSCalibur experiments, each figure has its own internal controls necessary for cytometer (BD Biosciences) and analyzed with FlowJo software (Tree comparison of the activity of BTN3A1 derivatives. Star). Gatings were performed on human T cells expressing Vd2+ TCR chains. The transmembrane region of BTN3A1 is not specifically required for Vg9Vd2 T cell activation Human Vg9Vd2 T and HEK 293 cells To determine whether the transmembrane region of BTN3A1 spe- PBMC of human healthy donors were isolated from blood samples obtained from the Etablissement Franc¸ais du Sang (PLERNTS-2016-08). For spe- cifically contributes to the antigenic activation of Vg9Vd2 T cells, cific expansions of Vg9Vd2 T cells, fresh peripheral blood lymphocytes chimeric BTN3A1 molecules containing the transmembrane re- (1 3 106 cells/ml) were stimulated with NBP (25 mM) in complete RPMI gions from three other human type I glycoproteins (BTN3A3, 1640 medium supplemented with 10% FCS and recombinant human IL-2 BTN1A1, or FGFR3) in place of the BTN3A1 sequence were (20 ng/ml). After 4 d, cultures were supplemented with recombinant hu- + engineered and expressed at the cell surface of HEK sh284 cells man IL-2 (60 ng/ml). Specific expansion of Vd2 T cells was checked 2 wk postactivation. Resting Vg9Vd2 T cell lines (Vd2+ T cell purity of .85%) (Fig. 1A). Following mAb or NBP sensitization of target cells, all of were then used in functional assays. The HEK 293FT cell line was cultured these chimeric molecules mediated Vg9Vd2 T cell activation, with in complete DMEM medium with 10% FCS. HEK short hairpin (sh)284 no significant difference relative to wild-type BTN3A1 (Fig. 1B). cells, which do not detectably express any BTN3A isoform following These results indicate that there is no specific contribution of the shRNA mediated knockdown, were cultured as previously described (10). Cell cultures were routinely tested for Mycoplasma contamination. BTN3A1 transmembrane region to the development of this gd T Downloaded from cell reactivity. BTN3A domain shuffling and generation of chimeric BTN3A molecules Specific requirement for the BTN3A1 JTM region for for Vg9Vd2 T cell activation Full-length cDNAs for human BTN3A1 (LIFESEQ3294566) and BTN3A3 (BC015815) isoforms were from Open Biosystems. cDNAs silently mutated Significant homology of the intracellular regions of BTN family

in the sh284 target sequence were expressed under control of a human CMV members begins with the motif (D/E)V(I/T)LD (DVILD residues in http://www.jimmunol.org/ promoter as described (10). Plasmids were used for transfections in BTN3A1, aa 311–315). The final exon of BTN3A1 encodes its BTN3A-silenced HEK293FT cells (sh284; clone 30). Plasmids encoding chimeric BTN3A1 molecules carrying swapped or mutated domains were entire B30.2 domain and this exon’s first complete codon codes generated by standard procedures. Briefly, a BTN3A1 coding sequence for the starting D of the DVILD motif. We took this motif to mark containing multiple silent mutations creating unique restriction enzyme the start of their B30.2 domains. The JTM domain of each family sites flanking or within the sequences coding for the transmembrane/JTM member was defined as the sequence between the end of its domains was obtained by synthesis (Eurofins ). Restriction enzymes/T4 DNA ligase were used to replace the transmembrane or JTM transmembrane region and the (D/E)V(I/T)LD motif. Defined in domain coding sequences with corresponding sequences from other . this way, the BTN3A1 JTM covers aa 242–310 and is 69 aa long. These replacement sequences were obtained by gene synthesis. Otherwise, A computed Clustal alignment of BTN family member JTMs is for block mutation of the BTN3A1 JTM, short oligonucleotides were used shown in Fig. 2A. These sequences diverge in both sequence and by guest on September 28, 2021 to replace as above the JTM coding sequences we sought to mutate. HEK length. Only the BTN3A3 JTM is relatively homologous to the cells were transfected using jetPEI (Polyplus-transfection, Ozyme) according to the supplier’s instructions. Two days posttransfection, HEK BTN3A1 JTM (56% identity). The JTM region of many trans- cells were checked for the expression of CD277 molecules by flow membrane receptors, such as growth factor receptors (e.g., epi- cytometry. dermal growth factor receptor), has been shown to play a critical role in signaling activity (20). To check for a role played by the In vitro functional assays JTM portion of BTN3A1 in the antigenic activation of Vg9Vd2 Target HEK cells were sensitized overnight with either a CD277-specific T cells, chimeric BTN3A1 molecules containing the JTM from mAb (clone 20.1; 10 mg/ml) or a NBP (zoledronate; 30 mM unless oth- other BTN family members in place of the BTN3A1 JTM erwise stated) at the indicated concentrations, extensively washed, and cocultured with gd T cells at 37˚C in complete RPMI 1640 medium in the (Fig. 2B) were expressed at the cell surface of HEK sh284 cells. presence of 10 mM monensin and CD107a-specific mAb. When indicated, Cytometric analysis using mAb 103.2 showed an approximately cells were cultured in the presence of mevastatin (50 mM). After 4 h, cells equivalent (within a 2-fold range) cell surface expression of were harvested, stained with Vd2 TCR mAb, and analyzed for CD107a BTN3A1 and each of the chimeric molecules, with the exception expression by flow cytometry. of the ERMAP JTM hybrid (Supplemental Fig. 1). Consistent with Statistical analysis this, most of these chimeric molecules mediated Vg9Vd2 T cell Data were analyzed using GraphPad Prism v5 and v6. The statistical activation induced after mAb 20.1 sensitization about as effi- significance of differences between populations was assessed with ciently as re-expressed wild-type BTN3A1 molecules. Impor- nonparametric Mann–Whitney tests. A p value ,0.05 was considered tantly, chimeras with the BTN3A3 JTM (JTM BTN3A3) were significant. significantly more effective than wild-type BTN3A1 in activating Vg9Vd2 T cells (Fig. 2C). Following NBP sensitization, most of Results the JTM chimeras did not induce Vg9Vd2 T cell activation at Human BTN3A1-expressing target cells activate Vg9Vd2T levels over and above those measured with mock-treated cells lymphocytes following either binding of agonist mAb to the IgV (some activity was nevertheless observed with the BTN2A1 JTM ectodomain (e.g., mAb 20.1) or sensitization with NBP. This latter chimera. Once again the BTN3A3 JTM chimera proved signifi- treatment induces intracellular PAg accumulation and binding cantly more effective than did wild-type BTN3A1 in mediating to the intracellular B30.2 domain. This is sensed by gd T cells in Vg9Vd2 T cell activation. Note that cell surface expression levels a TCR-dependent manner (12). HEK sh284 cells have a knock- of the BTN1A1, BTN2A1, BTN2A2, BTNL3, and BTNL9 JTM down of all three BTN3A isoforms and are severely impaired for chimeras were equal or superior to those of the BTN3A3 JTM Vg9Vd2 T cell activation (10). The strategy we used was to chimera (Supplemental Fig. 1). Thus the low or undetectable ac- transfect HEK sh284 cells with expression vectors for mutant or tivity observed with the former chimeras cannot be attributed to wild-type BTN3A1, sensitize them (mAb 20.1 or NBP), and then insufficient cell surface expression. The results of experiments The Journal of Immunology 3

FIGURE 1. Modification of the transmembrane region of BTN3A1 does not impact the antigenic activation of human Vg9Vd2 T cells. (A) Structures of BTN3A1 molecules tested to assess the contribution of the transmembrane region to antigenic activation of Vg9Vd2 T cells. The transmembrane region of BTN3A1 was replaced by the transmembrane region of BTN3A3, BTN1A1, or FGFR3 molecules. An alignment of amino acids sequences coding for the transmembrane domain of the various molecules is shown. Residues identical to BTN3A1 are represented as dashes and differences are indicated in capital Downloaded from letters. (B) Percentage of human Vg9Vd2 T cells expressing CD107a after coculture with HEK sh284 transiently expressing the chimeric molecules described in (A). Target cells were untreated (N) or sensitized with anti-CD277 mAb () or NBP (n). Error bars represent the mean 6 SD (n $ 3). **p , 0.005 (Mann–Whitney). TM, transmembrane domain. shown in Fig. 2C used a saturating dose of zoledronic acid (30 cellular PAg levels, whereas wild-type BTN3A1 requires higher, mM) for sensitization. Dose-response experiments demonstrate NBP-induced PAg levels to trigger gd T cell activation. To test this http://www.jimmunol.org/ that the response plateau is reached between 1 and 2 mM NBP, hypothesis, we analyzed the effect of reducing PAg levels in un- with a very high response already observed at 1 mM NBP (data not treated cells using mevastatin. Statins inhibit HMG-CoA reductase shown). Thus the low or undetectable activity of chimeras (Fig. in the mevalonate pathway, an enzyme necessary for the synthesis 2C) is observed using an NBP concentration 30-fold higher than of isopentenyl pyrophosphate. We first checked the efficacy of that needed to see an almost maximal response with wild-type statin treatments by expressing wild-type BTN3A1 molecules on BTN3A1. Taken together, these results indicate that changes to the surface of HEK sh284 cells and testing their ability to activate the JTM domain in BTN3A1 can strongly impact Vg9Vd2 T cell Vg9Vd2 T cells following treatment with NBP alone or with NBP activation, which suggests that this BTN portion plays an impor- and statins. As shown in Fig. 3A, the stimulating effect of NBP tant role in controlling this activation. treatment is almost completely abrogated by statins. This effect is by guest on September 28, 2021 NBP-specific, as statins have no effect on activation induced by Spontaneous activation potential of BTN3A3 JTM chimeras the mAb 20.1. We next tested the effect of statin treatments on the Apart from its enhanced gd T cell activation potential following spontaneous activation measured with the JTM3A3 chimera. No either mAb 20.1 or NBP sensitizations, another striking feature of significant effect of statin treatment was observed (Fig. 3B, 3C), the BTN3A3 JTM chimera is its reproducible ability to activate suggesting that lowering the intracellular PAg concentration has Vg9Vd2 T cells in the absence of any sensitization. One possible little effect on its spontaneous activation capacity. Thus the in- explanation is that this chimera is sensitive to low physiological creased activation observed with the JTM of BTN3A3 does not

FIGURE 2. The JTM region of BTN3A1 reg- ulates the antigenic activation of human Vg9Vd2 T cells. (A) Clustal alignment of amino acids se- quences encoding for the JTM portion of selected butyrophilin family members. (B) Structure of BTN3A1 molecules tested to assess the contribu- tion of the JTM region to antigenic activation of Vg9Vd2 T cells. In BTN3A1 chimeras, the JTM region of BTN3A1 has been replaced by the JTM region of the following BTN family members: BTN1A1, BTN2A1, BTN2A2, ERMAP, BTNL3, BTNL9, or BTN3A3. (C) Percentage of human Vg9Vd2 T cells expressing CD107a after coculture with HEK sh284 transiently expressing the chime- ric molecules described in (A)and(B). Target cells were untreated (N)orsensitizedwithanti- CD277 mAb ()orNBP(n). Error bars represent the mean 6 SD (n $ 3). *p , 0.05 (Mann– Whitney). 4 BTN3A JUXTADOMAINS CONTRIBUTE TO Vg9Vd2 T CELL ACTIVATION

FIGURE 3. Spontaneous gd T cell acti- vation potential of BTN3A3 JTM chimeras. (A) Percentage of human Vg9Vd2 T cells expressing CD107a after coculture with HEK sh284 cells or HEK sh284 cells re- expressing full-length BTN3A1 molecules (BTN3A1 WT). Target cells were untreated (N) or sensitized with either anti-CD277 mAb or NBP, in the presence or absence of mevastatin (50 mM). Error bars represent the mean 6 SEM. (B) Percentage of human Vg9Vd2 T cells expressing CD107a after coculture with HEK sh284 or HEK sh284 transiently expressing wild-type BTN3A1 (BTN3A1 WT) or BTN3A1 with the BTN3A3 JTM (BTN3A1 JTMA3). Target cells were untreated (N) or sensitized with either anti-CD277 mAb () or NBP (n), in the presence or absence of mevastatin (50 mM). Error bars represent the mean 6 SEM. Downloaded from (C) Representative cytometry plots showing CD107a upregulation on Vg9Vd2 T cells collected from cocultures with mAb- or NBP-sensitized HEK sh284 cells transiently expressing wild-type (BTN3A1 WT) or chimeric (BTN3A1 JTMA3) BTN3A1

molecules, in the presence or absence of http://www.jimmunol.org/ mevastatin (50 mM). Values for the per- centage of CD107a+ Vd2 T cells are indi- cated in the quandrants. seem to be due to an increased sensitivity to PAg levels. This expressed in HEK sh284 cells (Fig. 5A). Cytometric analysis was confirmed by a comparison of NBP dose-response curves of using mAb 103.2 showed an approximately equivalent (within a wild-type BTN3A1 and BTN3A1 with the BTN3A3 JTM 2-fold range) cell surface expression of BTN3A1 and each of the (Supplemental Fig. 2). by guest on September 28, 2021 Activity of mixed BTN3A1–BTN3A3 JTM chimeras An alignment of the BTN3A1 and BTN3A3 JTMs is shown in Fig. 4A (dashes in the BTN3A3 sequence represent amino acids identical to the BTN3A1 sequence). The C-terminal ends of the two JTMs are very similar, but there is more variation elsewhere. We divided the BTN3A1 and BTN3A3 JTMs arbitrarily into three boxes (b1, b2, and b3, Fig. 4A). To search for a zone of the BTN3A3 JTM capable of conferring the enhanced T cell activa- tion potential previously observed, chimeric molecules with JTMs comprising different combinations of BTN3A1 and BTN3A3 boxes were expressed on HEK sh284 cells and tested for their capacity to activate Vg9Vd2 T cells. FACS analysis using the 103.2 mAb showed that equivalent amounts of BTN3A1 and each JTM mutant were expressed at the cell surface (data not shown). No single BTN3A3 JTM block can confer on BTN3A1 the en- hanced capacity for activating T cells. It is necessary to transfer a combination of BTN3A3 JTM blocks b1 and b3 into BTN3A1 to reproduce the behavior of the BTN3A1 JTMA3 chimera (Fig. 4B). Thus we cannot identify a discrete part of the JTM as being re- FIGURE 4. Two portions of the BTN3A1 JTM region are involved in sponsible for the increased activation observed with the JTM of regulating the antigenic activation of Vg9Vd2 T cells. (A)Aminoacidsse- BTN3A3. quence alignment of the JTM regions of BTN3A1 (top line) and BTN3A3 (bottom line). BTN3A3 residues identical to BTN3A1 are represented as Critical BTN3A1 JTM residues lie close to the B30.2 domain dashes, and differences are indicated in capital letters. Subdivision blocks are B The results with JTM chimeras indicate that the BTN3A1 JTM is indicated (b1, box 1; b2, box 2; and b3, box 3). ( ) Percentage of human Vg9Vd2 T cells expressing CD107a after coculture with HEK sh284 tran- not simply an inert linker between the B30.2 and the transmem- siently re-expressing, or not (HEK sh284), full-length BTN3A molecules. brane region/extracellular domains, but rather an important player Target cells were untreated (N) or sensitized with either anti-CD277 mAb () in signaling to Vg9Vd2 T cells. In an attempt to identify key or NBP (n).b1,b2,b3,b1plusb3,andb1plusb2correspondtoBTN3A3 residues required for Vg9Vd2 T cell activation, mutant BTN3A1 JTM blocks that were used to replace the corresponding BTN3A1 blocks in molecules in which short blocks of JTM sequence have been BTN3A molecules. Error bars represent the mean 6 SD. Data are represen- replaced by alanine/glycine sequences of the same length were tative of three independent experiments. *p , 0.05 (Mann–Whitney). The Journal of Immunology 5 Downloaded from http://www.jimmunol.org/

FIGURE 5. Critical BTN3A1 JTM residues lie close to the B30.2 domain. (A) Fourteen short blocks of BTN3A1 JTM sequence were replaced by alanine/glycine sequences (a and g residues) of the same length. These chimeric BTN3A1 molecules were transiently expressed in HEK sh284 cells and tested to assess the contribution of each block within the JTM region of BTN3A1 to antigenic activation of Vg9Vd2 T cells. (B) Percentage of human Vg9Vd2 T cells expressing CD107a after coculture with HEK sh284 transiently expressing chimeric molecules described in (A). Target cells were untreated (N) or sensitized with anti-CD277 mAb () or NBP (n).*p , 0.05 (Mann–Whitney). (C) Representative cytometry plots showing CD107a upregulation on Vg9Vd2 T cells from cocultures with mAb- or NBP-sensitized HEK sh284 cells (HEK sh284) or HEK sh284 cells transiently expressing wild-type (BTN3A1 WT) or mutated (Mut 13 and Mut 14) BTN3A1 molecules. Values for the percentage of CD107a + Vd2 T cells are indicated in the quandrants. by guest on September 28, 2021 chimeric molecules, with the exception of the Mut2 mutant study, we show the results of experiments designed to assess the (Supplemental Fig. 3). Consistent with this analysis, BTN3A1 importance of the remaining domains, namely the transmem- mutants mediated Vg9Vd2 T cell activation following mAb 20.1 brane and JTM domains. Whereas the transmembrane domains of sensitization about as efficiently as re-expressed wild-type single-pass transmembrane receptors might be involved in regu- BTN3A1 molecules (Fig. 5B). Many of the mutations (2, 6–9, lation of signal transduction in a number of different ways 11, 12) had little or no effect on NBP-mediated Vg9Vd2 T cell [reviewed by Bugge et al. (22)], we found no evidence that this activation, whereas others (1, 3–5, 10) reduced activation slightly. was the case for BTN3A1. Thus replacing its transmembrane The only very marked differences were observed with mutants 13 domain by that of several other transmembrane receptors had no and 14, which completely abrogated the NBP response, without significant effect on the ability of BTN3A1 to trigger Vg9Vd2 significantly altering the mAb 20.1 response (Fig. 5B, 5C). NBP T cell activation. In marked contrast, replacing the BTN3A1 JTM dose-response curves for selected mutants showed that their ac- by an unrelated JTM from BTN family members BTN1A1, tivity relative to BTN3A1 was not dependent on the NBP dose BTN2A2, ERMAP, BTNL3, or BTNL9 strongly abrogated its (Supplemental Fig. 4). Note that expression levels of all the mu- activation capability acquired following an NBP-induced increase tants tested were higher than those of the Mut2 mutant, whereas in intracellular PAg levels. This demonstrates that the JTM is not a this latter mutant mediates Vg9Vd2 T cell activation at least as simple linker between the B30.2 and transmembrane/extracellular well as wild-type BTN3A1 (Supplemental Fig. 3). Indeed, no domains of BTN3A1, but an active player in driving its gd T cell correlation between Vg9Vd2 T cell activation and cell surface activatory function. These observations are consistent with the expression levels of BTN3A1 molecules was observed. In par- report by Hsiao et al. (21) showing that PAg binding to the B30.2 ticular, the lack of activity of mutants Mut13 and Mut14 cannot be domain induces changes in the nuclear magnetic resonance attributed to their insufficient expression levels. Interestingly, spectrum of residues in the JTM of BTN3A1. Replacing the these mutations alter amino acids at the C-terminal end of the BTN3A1 JTM with the JTM from BTN2A1 or the close BTN3A1 JTM, close to the B30.2 domain. The enhanced gd T cell acti- relative BTN3A3 led, respectively, to reduced or enhanced vation potential conferred by the BTN3A3 JTM was once more Vg9Vd2 T cell activation capabilities. Although the JTMs of other confirmed in this series of experiments. BTN family members have no significant homology with the BTN3A1 JTM (except for the closely related BTN3A2 and BTN3A3 proteins), all BTN family members whose JTMs were Discussion tested have an overall structure similar to BTN3A1: extracellular The importance of both the extracellular and the intracellular Ig-like domains, transmembrane region, intracellular JTM, and B30.2 domains of BTN3A1 in mediating human Vg9Vd2 T cell B30.2 domain. They might therefore use the mechanism proposed activation has been clearly established (10–12, 14, 15, 21). In this for BTN3A1 to transmit a signal to the outside of the cell: binding 6 BTN3A JUXTADOMAINS CONTRIBUTE TO Vg9Vd2 T CELL ACTIVATION of an intracellular ligand to the B30.2 domain, followed by biology other than Vg9Vd2 T cell activation. Whatever may turn transmission of the signal via the JTM to the extracellular domain. out to be the case, the work described in this study highlights the Retaining some activity with the BTN2A1 JTM chimera, or get- key role played by the JTM in transmitting to the BTN3A1’s ting enhanced activity with the BTN3A3 JTM chimera, could be extracellular domain the signal generated by PAg binding to its reflecting this phenomenon. The enhancing effect of the BTN3A3 intracellular domain. JTM, following either NBP or mAb 20.1 sensitization, fits well with our previous observations (15) that BTN3A3 molecules with Acknowledgments the R351H mutation (and thus able to bind PAg) are significantly We thank the staff of the Cytometry (Cytocell) and Cellular and Tissular better activators of Vg9Vd2 T cells than are wild-type BTN3A1 Imaging Core Facilities (MicroPICell) of Nantes University (Structure molecules. Furthermore, the BTN3A3 JTM chimera activated Fe´de´rative de Recherche Franc¸ois Bonamy) for help and expert technical Vg9Vd2 T cells somewhat in the absence of any sensitization. We assistance. have never observed this spontaneous activity with wild-type BTN3A1 molecules despite the numerous experiments we have Disclosures carried out using them. The higher overall activity of BTN3A1 M.B. is a founding scientist of the company Innate Pharma SA and was a with the BTN3A3 JTM thus comes with a price: inappropriate consultant for this company until October 2013 and is currently vice pres- activation of Vg9Vd2 T cells. The BTN3A1 JTM is under a ident of the Institut Me´rieux in charge of scientific and medical affairs. The double constraint. It must allow efficient activation of Vg9Vd2 remaining authors have no financial conflicts of interest. T cells, when BTN3A1 is presented by stressed cells, while avoiding provoking recognition when BTN3A1 is presented by normal References Downloaded from cells. 1. Hayday, A. C. 2000. gd Cells: a right time and a right place for a conserved third By systematically mutating short blocks of JTM amino acids, way of protection. Annu. Rev. Immunol. 18: 975–1026. we have identified several JTM zones whose mutation decreases 2. Hayday, A. C. 2009. gd T cells and the lymphoid stress-surveillance response. Immunity 31: 184–196. BTN3A1’s ability to activate Vg9Vd2 T cells after NBP sensi- 3. Bonneville, M., R. L. O’Brien, and W. K. Born. 2010. gd T cell effector func- tization, without affecting its activation ability after sensitization tions: a blend of innate programming and acquired plasticity. Nat. Rev. Immunol. 10: 467–478. with the agonist mAb 20.1. Among these, the JTM zone lying http://www.jimmunol.org/ 4. Chien, Y. H., C. Meyer, and M. Bonneville. 2014. gd T cells: first line of defense closest to the B30.2 domain and comprised of BTN3A1 aa and beyond. Annu. Rev. Immunol. 32: 121–155. 299310 (YNEWKKALFKPA) is of particular interest. Sebestyen 5. Bonneville, M., and E. Scotet. 2006. Human Vg9Vd2 T cells: promising new et al. (23) have shown, using flow cytometry fluorescence energy leads for immunotherapy of infections and tumors. Curr. Opin. Immunol. 18: 539–546. transfer experiments, that BTN3A1 exists as a dimer on the cell 6. Constant, P., F. Davodeau, M. A. Peyrat, Y. Poquet, G. Puzo, M. Bonneville, and surface of cancer cells and the BTN3A1 residues identified J. J. Fournie´. 1994. Stimulation of human gamma delta T cells by nonpeptidic mycobacterial ligands. Science 264: 267–270. above have been observed to constitute a short symmetrical di- 7. Tanaka, Y., C. T. Morita, Y. Tanaka, E. Nieves, M. B. Brenner, and B. R. Bloom. mer interface in the BTN3A1 crystal lattice (15). The forma- 1995. Natural and synthetic non-peptide antigens recognized by human gd tion of this dimer interface may thus be necessary to ensure T cells. Nature 375: 155–158. 8. Hintz, M., A. Reichenberg, B. Altincicek, U. Bahr, R. M. Gschwind, by guest on September 28, 2021 BTN3A1-mediated gd T cell reactivity in vivo. Is binding of A. K. Kollas, E. Beck, J. Wiesner, M. Eberl, and H. Jomaa. 2001. Identification other proteins to the BTN3A1 JTM involved in the activation of (E)-4-hydroxy-3-methyl-but-2-enyl pyrophosphate as a major activator for process? Target cell recognition by Vg9Vd2 TCRs depends on human gd T cells in Escherichia coli. FEBS Lett. 509: 317–322. 9. Gober, H. J., M. Kistowska, L. Angman, P. Jeno¨, L. Mori, and G. De Libero. Rho GTPase activity (23). Using biolayer interferometry, this 2003. Human T cell receptor gd cells recognize endogenous mevalonate me- study demonstrated RhoB binding to the full-length BTN3A1 tabolites in tumor cells. J. Exp. Med. 197: 163–168. 10. Harly, C., Y. Guillaume, S. Nedellec, C. M. Peigne´,H.Mo¨nkko¨nen, intracellular domain (B30.2 plus JTM), whereas binding was J. Mo¨nkko¨nen, J. Li, J. Kuball, E. J. Adams, S. Netzer, et al. 2012. Key impli- significantly reduced to the B30.2 domain alone. Although in- cation of CD277/butyrophilin-3 (BTN3A) in cellular stress sensing by a major teractions were abolished in the presence of soluble PAg, the human gd T-cell subset. Blood 120: 2269–2279. 11. Wang, H., O. Henry, M. D. Distefano, Y. C. Wang, J. Ra¨ikko¨nen, J. Mo¨nkko¨nen, precise role of RhoB in Vg9Vd2 T cell activation is not yet fully Y. Tanaka, and C. T. Morita. 2013. Butyrophilin 3A1 plays an essential role in understood. The cytoskeletal adaptor periplakin has prenyl pyrophosphate stimulation of human Vg2Vd2 T cells. J. Immunol. 191: been recently shown to bind to the BTN3A1 JTM (17), although 1029–1042. 12. Harly, C., C. M. Peigne´, and E. Scotet. 2015. Molecules and mechanisms im- periplakin knockdown using different small interfering RNAs plicated in the peculiar antigenic activation process of human Vg9Vd2 T cells. had no easily interpretable effect on BTN3A1 activation of Front. Immunol. 5: 657. 13. Rhodes, D. A., M. Stammers, G. Malcherek, S. Beck, and J. Trowsdale. 2001. Vg9Vd2 T cells. In this study, the main evidence for a functional The cluster of BTN genes in the extended major histocompatibility complex. contribution of this interaction relies on a correlation between Genomics 71: 351–362. loss of periplakin binding and loss of activation potential, pro- 14. Vavassori, S., A. Kumar, G. S. Wan, G. S. Ramanjaneyulu, M. Cavallari, S. El Daker, T. Beddoe, A. Theodossis, N. K. Williams, E. Gostick, et al. 2013. voked by a deletion of either the VKLLEEL JTM stretch, Butyrophilin 3A1 binds phosphorylated antigens and stimulates human gd encoded by the exon 5 of BTN3A1, or just of its di-leucine T cells. Nat. Immunol. 14: 908–916. motif. The results of our chimera experiments confirm that the 15. Sandstrom, A., C. M. Peigne´,A.Le´ger, J. E. Crooks, F. Konczak, M. C. Gesnel, R. Breathnach, M. Bonneville, E. Scotet, and E. J. Adams. 2014. The intracel- deletion of this stretch impacts the activation of Vg9Vd2 T cells. lular B30.2 domain of butyrophilin 3A1 binds phosphoantigens to mediate ac- However, extensive mutations of this stretch to AAGAAEL, tivation of human Vg9Vd2 T cells. Immunity 40: 490–500. 16. Rian˜o, F., M. M. Karunakaran, L. Starick, J. Li, C. J. Scholz, V. Kunzmann, which eliminates the di-leucine motif (and so might be expected D. Olive, S. Amslinger, and T. Herrmann. 2014. Vg9Vd2 TCR-activation by to eliminate periplakin interaction), have essentially no effect on phosphorylated antigens requires butyrophilin 3 A1 (BTN3A1) and additional BTN3A1’s activation potential. Insofar as periplakin does not genes on human 6. Eur. J. Immunol. 44: 2571–2576. 17. Rhodes, D. A., H. C. Chen, A. J. Price, A. H. Keeble, M. S. Davey, L. C. James, bind to BTN3A3, whereas BTN3A3 with the R351H mutation M. Eberl, and J. Trowsdale. 2015. Activation of human gd T cells by cytosolic (thus able to bind PAg) activates Vg9Vd2 T cells very efficiently, interactions of BTN3A1 with soluble phosphoantigens and the cytoskeletal it seems unlikely that periplakin is strictly required for the adaptor periplakin. J. Immunol. 194: 2390–2398. 18. Wang, H., and C. T. Morita. 2015. Sensor function for butyrophilin 3A1 in prenyl Vg9Vd2 T cell activation process. It remains possible that sev- pyrophosphate stimulation of human Vg2Vd2 T cells. J. Immunol. 195: 4583– eral different proteins, including RhoB and periplakin, interact 4594. 19. Compte, E., P. Pontarotti, Y. Collette, M. Lopez, and D. Olive. 2004. Frontline: with the BTN3A1 JTM and play redundant roles. Alternatively, characterization of BT3 molecules belonging to the B7 family expressed on the JTM interactants may be involved in aspects of BTN3A1’s immune cells. Eur. J. Immunol. 34: 2089–2099. The Journal of Immunology 7

20. Kovacs, E., J. A. Zorn, Y. Huang, T. Barros, and J. Kuriyan. 2015. A structural 22. Bugge, K., K. Lindorff-Larsen, and B. B. Kragelund. 2016. Understanding perspective on the regulation of the epidermal growth factor receptor. Annu. Rev. single-pass transmembrane receptor signaling from a structural viewpoint-what Biochem. 84: 739–764. are we missing? FEBS J. 283: 4424–4451. 21. Hsiao, C. H., X. Lin, R. J. Barney, R. R. Shippy, J. Li, O. Vinogradova, 23. Sebestyen, Z., W. Scheper, A. Vyborova, S. Gu, Z. Rychnavska, M. Schiffler, D. F. Wiemer, and A. J. Wiemer. 2014. Synthesis of a phosphoantigen prodrug A. Cleven, C. Che´neau, M. van Noorden, C. M. Peigne´, et al. 2016. RhoB mediates that potently activates Vg9Vd2 T-lymphocytes. Chem. Biol. 21: 945–954. phosphoantigen recognition by Vg9Vd2 T cell receptor. Cell Rep. 15: 1973–1985. Downloaded from http://www.jimmunol.org/ by guest on September 28, 2021 Peigné et al.

Supplemental Figure S1

100

BTN3A3 80

60 BTN3A1 WT

40 R2=0.0363

CD107a (% of pos T cells) T pos of (% CD107a 20 BTN2A1 sh284 BTN2A2 BTNL9 ERMAP BTN1A1 BTNL3 0 0 50 100 150 200 250 CD277 (gMFI)

Supplemental Figure S1. Correlation analysis of BTN3A1 chimera expression and the reactivity of human Vγ9Vδ2 T cells. Linear regression analysis shows no correlation between surface expression levels of CD277/BTN3A1 wild-type (BTN3A1 WT) and seven chimeric BTN3A1 molecules (juxtamembrane region chimeras) in transiently transfected HEK 293 cells (gMFI) and the reactivity of human Vγ9Vδ2 T cells (% of CD107apos T cells). The R2 (R squared) coefficient of determination value is indicated. sh284 correspond to HEK 293 cells that do not express CD277 (n=3).! Peigné et al.

Supplemental Figure S2

60

BTN3A1 JTMA3

40

BTN3A1 WT 20 CD107a (% positive T cells) T positive (% CD107a 0 0.0 0.5 1.0 1.5 2.0 NBP (µM)

Supplemental Figure S2. Dose-response curves to compare the reactivity of human Vγ9Vδ2 T cells (% of CD107apos T cells) against zoledronic acid (NBP)-treated HEK 293 cells expressing wild-type (BTN3A1 WT, !) or chimeric (BTN3A1 JTMA3, ") BTN3A1 molecules after transfection (n=3). ! Peigné et al.

Supplemental Figure S3

100

80

60 #12 #7 #11 #2 #8 40 #6 #9 R2=0.0011 BTN3A1 WT #5 #3 #1 #4

CD107a (% of pos T cells) T pos of (% CD107a 20 #10

sh284 #14 #13 0 0 50 100 150 CD277 (gMFI)

Supplemental Figure S3. Correlation analysis of BTN3A1 juxtamembrane region mutant expression and the reactivity of human Vγ9Vδ2 T cells. Linear regression analysis shows no correlation between surface expression levels of CD277/BTN3A1 wild-type (BTN3A1 WT) and fourteen mutant molecules (juxtamembrane region mutants) in transiently transfected HEK 293 cells (gMFI) and the reactivity of human Vγ9Vδ2 T cells (% of CD107apos T cells). The R2 (R squared) coefficient of determination value is indicated. sh284 correspond to HEK 293 cells that do not express CD277 (n=3). ! Peigné et al.

Supplemental Figure S4

40

30 Mut 12 BTN3A1 WT 20 Mut 5 Mut 10 10 Mut 4 BTN3A1 JTM BTN2A1 CD107a (% positive T cells) T positive (% CD107a 0 0.1 1 10 100 NBP (µM)

Supplemental Figure S4. Dose-response curves to compare the reactivity of human Vγ9Vδ2 T cells (% of CD107apos T cells) against zoledronic acid (NBP)-treated HEK 293 expressing wild-type (BTN3A1 WT, !), chimeric (BTN3A1 JTM2A1, ") or mutant (Mut 4, !; Mut 5, #; Mut 10, $; Mut 12, !) BTN3A1 molecules after transfection (n=3). !