IL-15 Trans-Presentation Is an Autonomous, Antigen-Independent Process

Interleukin-15, a 14–15 kDa member of the four–α-helix bundle family of cytokines that was described nearly simultaneously by the Waldmann and Grabstein laboratories, signals through a heterotrimeric receptor (1, 2). IL-15 and closely related IL-2 in their heterotrimeric receptors use cytokine-specific receptor α-chains, which bind the ligand with different affinities, IL-15Rα (CD215) for IL-15 (K_d, 50 pM) and IL-2Rα (CD25) for IL-2 (K_d, 10 nM). IL-2 and IL-15 share the IL-2/IL-15R β-chain (CD122), and with IL-4, IL-7, IL-9, and IL-21, they share the common γ_c chain (CD132) (3–5). Both IL-15 and IL-2 stimulate proliferation of T cells, induce the generation of CTLs, and stimulate the expansion of NK cells (6–8). In many adaptive immune responses, IL-2 and IL-15 have distinct roles (6, 7). In contrast to IL-2, IL-15 inhibits IL-2–mediated activation-induced cell death, does not activate functional regulatory T (T_reg) cells, and does not cause a major capillary leak syndrome in mice or in patients (9). IL-15 promotes the maintenance of CD44^hi CD8⁺ T cells and the renewal of virus-specific memory CD8⁺ T cells (10, 11). Furthermore, IL-15 is critical in the development of tissue memory phenotype CD103⁺ CD28⁺ CD8⁺ T cells (12). IL-15 also plays an essential role in the homeostasis of naive and memory CD4⁺ T cells. Naive CD4⁺ T cells expanding in the absence of IL-15 showed impaired Ag-induced activation (13). Ag-specific CD4⁺ memory cells were reported to be dependent on IL-15 for their basal homeostatic proliferation and long-term survival (14). It was shown that CD4 effector T cell differentiation was controlled by IL-15 that is expressed and presented in trans (15). IL-15 is also a mediator of CD4⁺ help for CD8⁺ T cell longevity (16). Continuous i.v. or s.c. IL-15 infusion efficiently expanded NK and CD8⁺ cell populations in patients with cancer; in the case of bolus infusion, IL-6–mediated dose-limiting toxicity (“cytokine storm”) was reported in some cases (17).

Lodolce et al. showed that IL-15–induced proliferation of T cells did not require expression of IL-15Rα on the T cell membrane, hinting at unconventional ligand delivery of IL-15 (18). Dubois et al. demonstrated a prolonged effect of IL-15 but not of IL-2 on the survival of T cells after cytokine withdrawal from the media (19). They revealed that surface-bound IL-15 is associated with the expression of IL-15Rα. In addition, endosomal recycling was associated with the expression of the IL-15Rα/IL-15 complex. IL-15 recycling was shown to be dependent on IL-15Rα but not on the β-chain. Critically, the IL-15Rα/IL-15 complex stimulated proliferation of neighboring CTLs in trans; this process was coined “IL-15 trans-presentation” (TP) (19). In particular, IL-15Rα/IL-15 complexes on the surfaces of activated monocytes induced trans-proliferation of βγ_c-expressing neighboring cells. This novel method of ligand delivery is enabled by the high ligand-binding affinity of IL-15Rα, thanks to which a stable IL-15Rα/IL-15 complex can form intracellularly and shuttle to the cell membrane. T cells expressing only the IL-2/15Rβ and γ_c subunits have a lower affinity (K_d, 1 nM) for IL-15 than for IL-15Rα, and because soluble IL-15 does not exist at such high concentrations in vivo, these cells require an alternative, directed ligand delivery method. For this reason, IL-15 is primarily presented by the IL-15Rα to responding cells (19–22). One way of presenting is when the IL-15Rα/IL-15 complex is proteolytically cleaved at the extracellular domain, creating soluble IL-15Rα/IL-15, which may induce IL-15 signaling in responding cells (23–25). During IL-15 TP, the presenting cell, with the IL-15Rα/IL-15 complex on its surface, comes in direct contact with a responding cell expressing the IL-2/15Rβγ_c heterodimer. Next, the whole IL-15Rαβγ_c/IL-15 TP complex may be internalized by the responding cell with part of the presenting cell’s membrane, inducing signaling by IL-2/15Rβ and γ_c (26). There are three main signaling pathways used: JAK/STAT, RAS/RAF/MAPK, and PI3K/Akt/mammalian target of rapamycin (mTOR). The JAK/STAT pathway has a crucial impact on determining the fate of Th cells, which can differentiate into multiple subsets, such as Th1, Th2, Th17, and T_reg cells (27). The Ras/Raf/MAPK and PI3K/Akt/mTOR pathways both contribute to the survival, proliferation, and maintenance of T cells, the former by strong proliferative signals and the latter with increased expression of antiapoptotic proteins (Bcl-2, Bcl-xL) while decreasing proapoptotic protein (Puma, Bim) expression. The PI3K/Akt/mTOR pathway has a key role in maintaining the balance between T_reg cells and other CD4⁺ T cells (28, 29).

Ag presentation (AP) is a well-studied immunological process that is indispensable for T cell activation. APCs such as B cells, macrophages, and dendritic cells (DCs) are capable of internalizing exogenous Ags, which are processed and then shuttled to the cell surface bound to MHC molecules, where they are presented to the T cell Ag receptor/CD3 complex. Upon TCR–MHC class II (MHC II) interaction, CD4, a coreceptor of the TCR, also binds to MHC II, further strengthening the intercellular protein complex. In the case of CD4⁺ T cells, secondary signals are provided by additional associations: CD28 is a receptor on the T cell that binds to either CD80 or CD86 on the presenting cell. Further proteins crucial for T cell activation, such as ICOS, 4-1BB, and OX40, also need to be expressed by the T cells and their ligands on the APC. These costimuli are only provided when the presenting cell encounters a pathogen (30, 31). In a B cell–T cell interaction, CD40 and CD40L also form a complex. Additional protein–protein interactions, such as LFA-1/ICAM-1, further stabilize the immunological synapse (IS) (32). We and others have shown that various ion channels, predominantly voltage-gated potassium (Kv)1.3 channels, are also enriched, via the PSD-95 adaptor protein, at the IS in the early phases of T cell activation (33–35).

As both IL-15 TP and AP contribute to T cell activation and involve similar cell-to-cell contacts, the question arises whether these two processes are interconnected or independent from each other. There are components in ISs, namely inhibitory receptors of the killer Ig-like receptor family expressed by NK cells, that can negatively regulate IL-15 TP occurring between B and NK cells (36). Potential effects of AP on IL-15 TP have not been investigated yet. The present studies involved biophysical measurements of IL-15 TP and AP. We have previously investigated the assembly and interactions of IL-2R and IL-15R in T cells by using Förster resonance energy transfer (FRET) and other biophysical methods (37–40). Here we used a B cell–T cell model system expressing the components of AP and IL-15 TP and studied their “trans” (intercellular) and “cis” (occurring on the same cell) interactions via biophysical techniques. We aimed to give direct evidence for the assembly of IL-15R subunits expressed by the two cells during IL-15 TP, for the first time, and to ask whether the players involved in AP and IL-15 TP move jointly or independently to the IS. We also sought an answer to whether the efficiency of IL-15R signaling is influenced by AP and, conversely, if TCR-mediated signaling is affected by IL-15 TP.

Raji B lymphocytes and Jurkat T lymphocytes were cultured in RPMI 1640 medium (Sigma-Aldrich, St. Louis, MO) supplemented with 10% (v/v) FBS and GlutaMAX (Thermo Fisher Scientific, Waltham, MA), and HEK293T cells were cultured in DMEM (Sigma-Aldrich). All cells were grown in a 5% CO₂ humidified atmosphere at 37°C. Raji and Jurkat cells stably expressing enhanced GFP (EGFP)–IL-15Rα, mCherry–IL-15Rα, IL-15Rα, and IL-2/15Rβ (referred to as Raji-EGFP-IL-15Rα, Raji-mCherry-IL-15Rα, Raji-IL-15Rα, and Jurkat-IL-2/15Rβ, respectively) were cultured with Geneticin (Merck, Darmstadt, Germany) as a selective agent.

Plasmids coding for the IL-2Rα subunits tagged at the N or C terminus with mCherry, IL-15Rα nontagged or tagged at the N terminus with mCherry or EGFP, and nontagged IL-2/IL-15Rβ were prepared as described earlier (39). Plasmids for the retroviral transfer vector pBMN-Z-IN (EGFP-IL-15Rα, mCherry-IL-15Rα, IL-15Rα, IL-2/15Rβ) were digested with the restriction enzymes BamHI and SalI and ligated into the open reading frame site of the transfer vector (replacing the LacZ).

mCherry-IL-2Rα and IL-2Rα-mCherry were transiently transfected (1 µg plasmid) into 10⁶ Raji cells with the Cell Line Nucleofector Kit V for Amaxa Nucleofector II device (Lonza Group, Basel, Switzerland) according to the manufacturer’s protocol using the program (M-013) recommended for Raji cells.

HEK293T cells, used as packaging cells, were transfected by mixing the necessary plasmids, retroviral transfer vector (17 μg) (see the Plasmid Construction section), the vesicular stomatitis virus G envelope plasmid (12 μg), and psPAX2 packaging plasmid (4 μg) with 875 µl water, 125 µl CaCl₂, and 1 ml HEPES-buffered saline (150 mM NaCl and 20 mM HEPES in 900 ml water), supplemented with 25 µM chloroquine. The supernatant was changed to fresh DMEM after 6 h. Supernatants (containing retrovirus) were collected 48 h after transfection and filtered through a 0.45-µm filter. The filtered supernatant was added to Raji or Jurkat cells with additional 10 μg/ml Polybrene (Sigma-Aldrich) and incubated for 48 h, after which the cells were immunofluorescently labeled to check the efficiency of the viral transduction.

Either 10⁵ (for microscopic measurement) or 6 × 10⁵ (for flow cytometric measurement) Raji cells were treated with IL-15 (100 nM) and/or Staphylococcus enterotoxin E (SEE; 1 µg in 50 µl RPMI 1640 medium) (Toxin Technology, Sarasota, FL) and incubated for 20 min at 37°C. After washing in RPMI 1640 medium, Raji cells were mixed with Jurkat cells (10⁵ or 6 × 10⁵), centrifuged at 200 × g for 1 min and incubated for 20 min at 37°C, after which the cells were placed on ice for immunofluorescent labeling.

Raji and Jurkat cells were labeled with mouse mAbs conjugated with Alexa Fluor 488 or Alexa Fluor 546 (Thermo Fisher Scientific) to detect receptors and receptor subunits. The following Abs were used for labeling: anti-Tac for IL-2Rα, Mikβ3 for IL-2/15Rβ, OKT3 for CD3ε, and L243 for MHC II (HLA DR). Cells were washed once in PBS before labeling, then incubated on ice for 30 min with 50 µg/ml mAb tagged with the proper fluorescent dye. After the labeling procedure, cells were washed twice and fixed with 2% formaldehyde at room temperature and placed in 8-well chambered coverslips (ibidi GmbH, Gräfelfing, Germany).

Associations of the receptor subunits (molecular vicinity in the range of 2–10 nm) were determined by FRET microscopy on a pixel-by-pixel basis with an LSM 880 confocal microscope (Carl Zeiss, Oberkochen, Germany) equipped with a C Plan-Apochromat 63×/1.4 numerical aperture oil immersion objective. For excitation of Alexa Fluor 488, the 488-nm line of an Ar ion laser was used, and for mCherry, a 543-nm HeNe laser was used. Signals were detected in three channels: I₁ = donor (excitation, 488 nm; emission, 500–560 nm); I₂ = transfer (excitation, 488 nm; emission, 590–700 nm); and I₃ = acceptor (excitation, 543 nm; emission, 590–700 nm). These can be expressed as follows:

where E is the mean FRET efficiency in the pixel, I_D is the donor intensity in channel 1 that would be present in the absence of FRET, and I_A is the acceptor intensity in channel 3. Spectral cross-talk/bleed-through factors S₁ = (I₂ − B₂)/(I₁ − B₁) and S₂ = (I₂ − B₂)/(I₃ − B₃) were calculated from cells labeled with only Alexa Fluor 488–Mikβ3 (donor) or expressing only mCherry–IL-15Rα (acceptor). Background correction in each channel was achieved by subtracting average autofluorescence intensities (B₁, B₂, B₃) measured on unlabeled Raji cell membranes. The α factor, which relates the acceptor signal measured in the I₂ channel to the donor signal measured in the I₁ channel arising from equal amounts of excited acceptor and donor dyes (41, 42), and furthermore the Q factor, which is the ratio of the signal intensities I₃/I₁ arising from equal amounts of excited acceptor and donor fluorophores, were assessed by using Raji cells transiently transfected with IL-2Rα–mCherry and labeled with Alexa Fluor 488–anti-Tac (mCherry tag on the C terminus; therefore, expected E = 0) by the following equation:

where ε_A488(488) ∼55,831 M⁻¹ cm⁻¹ and ε_mCh(488) ∼5,564 M⁻¹ cm⁻¹ are the extinction coefficients of Alexa Fluor 488 and mCherry at the donor excitation wavelength (488 nm), respectively, and L_D is the fluorophore-to-protein labeling ratio of the Alexa Fluor 488–anti-Tac mAb determined by absorption photometry. The FRET efficiency was calculated on a pixel-by-pixel basis as follows:

Calculations were carried out in regions of interest consisting only of pixels where the intensity values were above a threshold (dynamic threshold was set to 10% of the maximum pixel intensity in each frame) in both the first and third channels. After thresholding, regions of interest were manually adjusted to cover the synaptic area on the basis of transmission image. Calculated FRET efficiency values from the selected pixels were averaged to describe each IS. Acceptor-to-donor molecular ratios were also calculated by using the following equation (43):

FRET was also calculated from the fluorescence-lifetime decay of donor molecules measured on an A1 confocal microscope (Nikon, Tokyo, Japan) equipped with a Plan-Apochromat 60×/1.27 numerical aperture water immersion objective and a time-correlated single-photon counting upgrade kit (PicoQuant, Berlin, Germany). For excitation, a 485-nm picosecond pulsed laser with a repetition rate of 20–40 MHz was used. Emission was detected via a 520/35-nm emission filter using a PMA hybrid 40 photon-counting photomultiplier (PicoQuant). Data were collected for 3–5 min. The resulting fluorescence decay curves were fitted with the SymphoTime 64 software to a multiexponential reconvolution model with two components:

where A[i] is the amplitude, τ[i] is the exponential decay time of the ith component, Bkgr_Dec is correction for decay background, Bkgr_IRF is correction for background of the instrument response function (IRF), Shift_IRF is the correction for temporal IRF displacement, A_Sum is the fluorescence intensity at time 0, and τ_AvAmp is the amplitude-weighted average lifetime. The FRET efficiency was determined by the following equation:

where τ_DA is the amplitude-weighted average lifetime of the donor in the donor-acceptor–tagged samples and τ_D is that of the donor in the absence of the acceptor but in the same environment (44).

The extent of receptor translocation was assessed from confocal images by calculating the ratio of the average pixel intensity inside the IS and in the whole cell membrane using MATLAB (MathWorks, Natick, MA) as follows:

where I_IS is the intensity of a pixel in the IS, N_IS is the number of such pixels, I_tot is the intensity of a pixel anywhere in the cell membrane (including the IS), and N_tot is the number of all pixels.

To detect the efficiency of IL-15–mediated signaling, the amount of p-STAT5 was measured by flow cytometry. Prior to forming ISs, Jurkat cells were labeled with Alexa Fluor 546–W6/32 mAbs specific for the HLA I H chain. After IS formation, cells were separated by trypsin digestion for 5 min at 37°C. Cells were fixed in 2% formaldehyde for 10 min and permeabilized with 90% methanol for 30 min. Washing (twice) was carried out in staining buffer containing 10 ml FBS (to block unspecific binding) and 0.5 g Na-azide (Merck) in 490 ml PBS. Cells were then labeled with Alexa Fluor 647–anti-p-STAT5 mAb or with the IgG₁ κ-isotype control (both from BD, Franklin Lakes, NJ) (10 µl with 40 µl staining buffer, 50 min, at room temperature). Measurements were carried out on a NovoCyte 3000 flow cytometer (Agilent, Santa Clara, CA) with 488 nm (for EGFP), 561 nm (for Alexa Fluor 546), and 640 nm (for Alexa Fluor 647) excitations; EGFP (from the Raji-EGFP-IL-15Rα), Alexa Fluor 546, and Alexa Fluor 647 signals were detected via 530/30-nm, 586/20-nm, and 660/20-nm emissions filters, respectively. Jurkat cells were gated on the basis of their Alexa Fluor 546–positive and EGFP-negative signal; p-STAT5 level was determined from the Alexa Fluor 647 signal. The p-STAT signal was fitted to a two-component log-normal distribution defining a responding and a nonresponding population relative to the basal phosphorylation level. In the case of the IL-15– and double-treated samples, only the responding populations displaying an increase relative to the basal p-STAT5 signal were considered for the calculation of the mean intensity.

TCR-mediated signaling was monitored via flow cytometric detection of p-CD3ζ in Jurkat cells in a procedure similar to that for p-STAT5. Cells were labeled with allophycocyanin-tagged anti-p-CD3ζ mAb or IgG2b κ-isotype control mAb (both from Thermo Fisher Scientific). Allophycocyanin was excited at 640 nm and detected at 660/20 nm. In the analysis, T cells and T cell–B cell conjugates were selected on the basis of their Alexa Fluor 546–W6/32–positive signal; p-CD3ζ level was determined from the allophycocyanin signal.

Jurkat and Raji cells with or without the different treatments were centrifuged at 12,800 × g at 4°C for 15 min followed by lysis of the pellet in lysis buffer (50 mM Tris, 1 mM EDTA, 0.1% 2-ME, 0.5% Triton X-100, 1 mM PMSF) containing a protease inhibitor mixture (Sigma-Aldrich) at a 1:100 dilution ratio in PhosStop (Roche, Basel, Switzerland) and sodium-orthovanadate (Sigma-Aldrich) at 1:50 dilution and sonicated with five to seven strokes with 40% cycle intensity (Branson Sonifier 450; Branson Ultrasonics, Brookfield, CT). The lysed cells were centrifuged at 12,800 × g at 4°C for 15 min. The supernatant was collected, and its protein concentration was measured with the Bradford assay at 595 nm (Synergy plate reader; BioTek, Winooski, VT). Each sample was measured with three technical parallels, normalized using BSA standard (stock, 0.5 mg/ml; Sigma-Aldrich). The samples were diluted to 2 mg/ml concentration, mixed with equal volumes of 2× SDS denaturation buffer (125 mM Tris-HCl, pH 6.8, containing 4% SDS, 20% glycerol, 10% 2-ME, 0.02% bromophenol blue dye), and boiled at 99°C for 10 min.

Proteins were separated on 15% SDS-polyacrylamide gels and blotted onto a PVDF membrane (Merck) using a semidry blotting method. Membranes were blocked with 5% nonfat dry milk in TBST for 1 h at room temperature. Primary Abs for p-CD3ζ (Cell Signaling Technology, Danvers, MA) and GAPDH (BioLegend, San Diego, CA) were diluted in 0.5% milk in 1× TBST with a dilution ratio of 1:5000 and incubated overnight at 4°C. The membranes were washed four times with 1× TBST for 15 min at room temperature, followed by incubation with HRP-labeled secondary Abs (Advansta, Menlo Park, CA) at 1:15,000 dilution for 1 h. The membranes were washed three times with 1× TBST for 15 min at room temperature. The targeted protein bands were visualized using an ECL kit (Advansta). Experiments on solitary Jurkat and Raji cells were performed on two replicates, and experiments on Raji-Jurkat mixtures were performed on five replicates. p-CD3ζ levels determined by densitometry were normalized to GAPDH.

Student t tests were performed with GraphPad Prism 8 software (GraphPad Software, La Jolla, CA). Changes with p < 0.05 were considered statistically significant.

To study IL-15 TP and AP, we used the model system depicted in (Fig. 1. As a presenter cell, we used Raji B cells modified with viral transduction to express tagged (with mCherry or EGFP on the N terminus) or nontagged IL-15Rα subunits. As responding cells, Jurkat T cells were also virally transduced to express nontagged IL-2/15Rβ, whereas the γ_c subunit was expressed endogenously. We measured the expression levels of the receptors with flow cytometry. As shown in Supplemental Fig. 1, on Raji B cells, MHC II has the highest expression level, followed by IL-15Rα with ∼50× lower expression. On the Jurkat T cells, the CD3 expression is comparable to that of IL-15Rα on the Raji cells, which is three times higher than that of the IL-2/15Rβ. Finally, the γ_c has expression two orders of magnitude lower than that of IL-2/15Rβ.

When Raji cells were treated with either SEE superantigen binding to MHC II or by IL-15 binding to IL-15Rα, they readily formed ISs with Jurkat cells expressing the TCR/CD3 complex and the IL-2/15Rβ and γ_c receptor subunits after being centrifuged together (Fig. 2A). Raji cell–Jurkat cell conjugates were very rarely formed when neither ligand was present.

In vitro studies previously suggested the existence of IL-15 TP; however, these experiments focused on its effects, such as proliferation of responding cells induced by IL-15Rα/IL-15–expressing presenter cells. To directly observe the assembly of the IL-15Rα/IL-15–IL-2/15Rβ complex on the molecular level and to study the effect of AP on the aforementioned process, we used ratiometric FRET experiments. In all FRET measurements, we labeled the lower expressed protein with the donor dye and the higher expressed one with the acceptor dye, thus increasing the probability that a donor has an associated acceptor partner. FRET efficiencies were evaluated on a pixel-by-pixel basis (Fig. 2A). IL-15 treatment alone resulted in E = 8.7 ± 2.4% FRET efficiency between Alexa Fluor 488–Mikβ3 mAb-tagged IL-2/15Rβ and mCherry–IL-15Rα, indicating the assembly of the IL-15 TP receptor/ligand complex at the IS. Double treatment with IL-15 and SEE superantigen further increased E to 10 ± 0.2%. With SEE treatment alone, the low E = 1.3 ± 2.7% value indicated that there was no significant association between the subunits at the IS in the absence of IL-15; random FRET could occur due to the high N_A/N_D acceptor-to-donor molecular ratios (Fig. 2C). In conclusion, we have provided direct biophysical evidence for IL-15 TP; AP could not induce the assembly of the IL-15 TP complex alone, but enhanced its formation when IL-15 was also present.

As a positive control for FRET, we used Raji cells transiently expressing IL-2Rα tagged with Alexa Fluor 488–anti-Tac mAb and mCherry at the N terminus (Fig. 1), resulting in an average FRET efficiency of E = 13 ± 3%; the negative control, IL-2Rα tagged with Alexa Fluor 488–anti-Tac mAb at the N terminus and with mCherry at the intracellular C terminus, yielded E = 0.8 ± 1.2% (Fig. 2A and 2B).

To verify the existence of AP in our model system, we measured the proximity of CD3 and MHC II at the IS with FLIM-FRET. On the one hand, when treated with IL-15 alone, there was no substantial association (E = 1.5 ± 3%) between the two proteins (Fig. 3A, Supplemental Fig. 3A), despite the high N_A/N_D ratio at the IS. On the other hand, with SEE treatment, there was a significant increase in the FRET efficiency (E = 5.3 ± 2.4%), although the N_A/N_D dropped to half of what was measured with the IL-15 treatment. The double treatment seemingly had no additional effect (E = 5.2 ± 3.1%) relative to SEE alone, hence leading to the conclusion that Ag alone is necessary and sufficient for efficient AP with no influence by IL-15 TP.

When assessing potential interactions between the IL-2/15Rβ subunit and CD3 during AP and IL-15 TP, we found a low association or no association at all. First, we determined the FRET efficiency on lone T cells, which had no treatment or contact with B cells, and received E = 2.3 ± 3.4% (Fig. 3B, Supplemental Fig. 3B). Next, when measuring at the IS, IL-15–treated samples gave E = 0.5 ± 3.1%, SEE alone resulted in E = 3 ± 4.3%, and the double treatment yielded E = 2.3 ± 3.4%. The slightly elevated E upon SEE treatment is most likely caused by the increased N_A/N_D ratio (∼5), which stands out for this sample; this is due to ligand-induced translocation of CD3 but not IL-2/15Rβ to the IS. On the basis of these results, we can conclude that there is no significant association between IL-2/15Rβ and CD3 either on resting cells or at the IS.

Our finding that AP enhanced the association of IL-15Rα and IL-2/15Rβ subunits and our previous result that MHC II was associated with IL-15Rα in FT7.10 T cells (37, 38) bring up the question whether IL-15Rα and MHC II interact with each other in B cells as well. We assessed the cis-interaction between IL-15Rα and MHC II in the IS by FLIM-FRET. Lifetime measurements for selected cells are shown in (Fig. 3E, raw lifetime decay curves in Supplemental Fig. 2, fits to the curves in (Fig. 3D, and average FRET efficiencies in (Fig. 3C. First, we measured FRET on lone B cells and got E = 8 ± 0.1%, suggesting that these molecules were associated with each other in the absence of ligands (Fig. 3C). Next, we measured FRET at the IS. IL-15– and double-treated samples gave similar results, E = 8.1 ± 2.4% and E = 9.3 ± 1.8%, respectively; however, the SEE-treated sample resulted in a significantly higher value of E = 13 ± 2.1%. The N_A/N_D values did not differ throughout the different treatments; hence, this factor cannot explain the higher E for the SEE treatment. We also calculated the FRET efficiencies outside the synaptic region of the cell membrane (Fig. 3C, bottom panels). The results showed a trend similar to that which we observed in the IS: With IL-15 treatment, E = 6 ± 3.5%, a significantly elevated E = 10.2 ± 3.5% in case of the SEE treatment and E = 7.9 ± 2.7% upon double treatment. These results indicate that IL-15Rα and MHC II exist in a complex in lone as well as synapse-engaged cells. When MHC II binds Ag, a conformational change may ensue, affecting the proximity of the epitopes and thereby altering the distance between donor and acceptor molecules.

Enrichment at the IS of the proteins partaking in AP and IL-15 TP was quantified by I_IS,norm, the ratio of the average pixel intensity at the IS relative to that in the whole cell as described in the Materials and Methods section. IL-15Rα shows a constant 1.5–1.9-fold enrichment at the IS, regardless of the treatment (Fig. 4A). The IL-2/15Rβ subunit, unlike IL-15Rα, was only enriched at the IS when IL-15 was presented by IL-15Rα (Fig. 4B): When IL-15 TP occurred, I_IS,norm increased to 4.2–4.4-fold. MHC II behaved similarly to the IL-15Rα subunit because it had a small but constant enrichment for all treatments (Fig. 4C): for IL-15 treatment, 1.4-fold; for SEE treatment, 1.6-fold; and for double treatment, 1.7-fold. CD3, like IL-2/15Rβ, translocated to the synaptic region only when MHC II presented the Ag (Fig. 4D). In these cases, the enrichment was 2.4–2.7-fold. The translocation results further support the FRET association studies, which concluded that IL-15Rα and MHC II exist in a complex and translocate together to the IS when either one presents a ligand in trans and interacts with its partner on the T cell. The relatively mild enrichment of these two presenter proteins can be explained by their expression levels, which are higher than those of their responding partners on the T cell; thus, only a fraction of IL-15Rα and MHC II could find a trans-interacting partner at the IS. In contrast, IL-2/15Rβ and CD3 are independent from each other and are enriched at the IS only when their own ligand is presented.

To study if IL-15 TP could indeed initiate signal transduction in the responding cell and to determine whether AP had any effect on it, we measured STAT5 phosphorylation levels in resting Jurkat cells and after exposure to differently treated Raji cells (Fig. 5A). We found that Jurkat cells on their own already had nonzero basal p-STAT5 levels, which were slightly increased upon SEE treatment (Fig. 5B). Interaction with IL-15–treated Raji cells largely increased the amount of p-STAT5, but there was no significant increase due to additional SEE treatment. For the latter two treatments, we detected a responding and a nonresponding population and fitted a two-component log-normal distribution to their histograms (red curves in (Fig. 5C). To compare the p-STAT5 levels, we used the means of the responding populations corrected by the basal p-STAT5 values of lone Jurkat cells (Fig. 5D). These results show that IL-15 TP can initiate signal transduction on its own and that AP has no significant effect on this process.

To determine the potential effects of IL-15 TP on AP-induced T cell signaling, we first measured the phosphorylated CD3ζ levels in the Jurkat cells exposed to differently treated Raji cells (SEE, IL-15, or SEE + IL-15) with flow cytometry (Fig. 6A). A nonzero basal level of p-CD3ζ relative to the isotype control signal was already detectable in lone Jurkat cells and in samples where only IL-15 TP occurred (Fig. 6B). Upon AP taking place alone or together with IL-15 TP (treatment with SEE or SEE + IL-15), the p-CD3ζ levels significantly increased. The signal for the double treatment was slightly lower than that for SEE treatment alone, although the difference was statistically not significant (Fig. 6C). Mechanical stress during sample preparation tearing IS-forming cells apart may leave membrane fragments of the partaking cells on each other (Supplemental Fig. 4A and 4B), which could compromise cell-by-cell determination of an IS-enriched membrane protein such as p-CD3ζ. Therefore, we also analyzed CD3ζ phosphorylation at the cell population level with Western blotting. These experiments showed a similar trend, with the strongest p-CD3ζ bands from the double- and SEE-treated Raji + Jurkat samples (Fig. 6D). Densitometry of the blots confirmed that simultaneous IL-15 TP decreased the efficiency of AP-induced TCR signaling in five of five cases, although this change was not statistically significant.

IL-15 is a pleiotropic cytokine that stimulates many cell types and functions with IL-15 disorders, playing a pathogenic role in organ-specific immune diseases (45–47). One is compelled to ask why among γ_c cytokines IL-15Rα/IL-15 alone functions predominantly via TP to neighboring cells. It is clear that TP occurs where IL-15 is required in the context of cell–cell contact (14, 48). IL-15Rα/IL-15 donor occurs in APCs (monocytes, macrophages, and DCs) that are stimulated to coordinately express the receptor and the cytokine by the actions of IFNs, anti-CD40 agonists, or TLR signals (19, 21, 49). One expresser of the βγ_c receptor chains is the CD8⁺ T cells (6). The IL-15 signal at most if not all times occurs with other signals to T cells engaged in CD40/CD40L, TCR/Ag–MHC, and CD28/CD86 interactions; that is, the IL-15 signals are coordinately provided by the other cell with receptors–counterreceptors of the interacting APCs and T cells (19, 50). This site of TP yields CD44^hi CD8⁺ memory T cells, a major functional component of immunological memory (25, 51).

In a second system, IL-15 TP is an essential mechanism for the survival and proliferation of NK cells. Accordingly, NK cells do not develop in mice lacking either IL-15 or any of the three IL-15R subunits. Here again, IL-15 is presented in trans by cells expressing the IL-15Rα/IL-15 complex, such as stromal cells and DCs to NK cells that express the βγ_c chains (52–54).

In a third arena, as suggested by Jabri and Abadie, IL-15 functions as a danger signal to regulate tissue-resident T cells and tissue destruction (55). IL-15 has an exceedingly wide cellular distribution compared with the other γ_c cytokines and can be expressed by hematopoietic and nonhematopoietic cells. IL-15Rα/IL-15 can be induced in these diverse tissues by cellular infection and sterile inflammation as in autoimmune disorders. Whatever the IL-15 system, it functions in a cell contact–dependent manner, often regulating tissue-resident T cells (56). IL-15–induced effector cytotoxic cells function locally only if there is ongoing active tissue distress that requires control of either an infectious or noninfectious stress signal. Thus, as proposed by Jabri and Abadie, TP facilitates the tissue destruction by cytotoxic T cells only when tissues provide them with a “kill me” signal. Although in select cellular infections this is of value to the host, it also underlies select autoimmune disorders such as celiac disease, type 1 diabetes, alopecia areata, inflammatory bowel disease, and sarcoidosis (55).

In light of the above-described roles of IL-15 TP, it was an intriguing question whether IL-15 TP between an APC and a T cell could occur autonomously or only accompanying AP. Both IL-15 TP and AP depend critically on ligand-controlled intercellular protein–protein interactions, and, in a T cell model system, we previously demonstrated an interaction between the two presenting proteins, IL-15Rα and MHC II (37, 38). Here, we showed direct evidence for the existence of IL-15 TP by investigating the association of IL-2/15Rβ and IL-15Rα subunits in both the absence and the presence of AP. Our findings are summarized in the model shown in (Fig. 7. The IL-15R heterotrimeric receptor was only assembled when IL-15Rα bound IL-15, which could be presented to the IL-2/15Rβγ_c complex. During IL-15 TP, all IL-15R subunits translocated to the IS; however, most probably due to the difference in expression ratios, the enrichment of IL-2/15Rβ was more prominent and only occurred when IL-15 was present. By measuring STAT5 phosphorylation, our objective was to determine the efficiency of IL-15 signaling. The results showed that IL-15 TP alone could trigger the JAK/STAT pathway, and this process was independent from AP. Jurkat cells express IL-2 constitutively, which may initiate JAK/STAT signaling in an autocrine or intracrine manner, thus explaining the nonzero STAT5 basal phosphorylation in untreated lone T cells (39, 57). Interestingly, when AP occurred parallel to IL-15 TP, the association between IL-2/15Rβ and IL-15Rα subunits increased; this closer proximity could be explained by the stronger IS due to the AP complex and/or the previously shown close association of MHC II and IL-15Rα, a complex that we revisited in this study in the context of the AP and IL-15 TP. Our experiments confirmed the close molecular proximity of MHC II and IL-15Rα in the used model system in B cells; these molecules translocated jointly to the synaptic area when either IL-15 or Ag was present. Additionally, we found that if MHC II bound SEE superantigen, the FRET efficiency between IL-15Rα and MHC II increased not only at the synaptic region but also outside the IS as well as in lone B cells. SEE belongs to the zinc-coordinated family (58). This superantigen contains two binding sites for MHC II, which may induce its homodimerization; this restructuring may lead to a conformational change altering the MHC II/IL-15Rα complex in a way that brings the donor and acceptor molecules in closer proximity (59). Interestingly, when IL-15 TP happens, this phenomenon seems to be reduced. Remarkably, whereas MHC II was found to be clustered on JY B cells (60), TCR appears to be randomly distributed and monomeric (61, 62).

As with the IL-15 TP, we also provided direct evidence for the assembly of the Ag-presenting complex by measuring intercellular FRET between the CD3 coreceptor and MHC II. As expected, association of these molecules only occurred when the B cells were preloaded with the SEE Ag. We also investigated JAK/STAT signaling upon AP and found that there was a small but significant increase in STAT5 phosphorylation. This is caused by convergence between IL-15 and TCR signaling, but whereas, in the case of IL-15, STAT5 is phosphorylated by JAK1 or JAK3, during TCR signaling, phosphorylation is carried out by Lck, a lymphocyte-specific tyrosine kinase (63). Conversely, SEE-induced phosphorylation of the CD3ζ subunit (by Lck) was moderately decreased by simultaneous IL-15 TP. One could speculate that juxtaposition of IL-15Rα with the βγ_c heterodimer might interfere with the interaction between the IL-15Rα–associated MHC II and the TCR or restrict the accessibility of the CD3ζ phosphorylation site; however, further research is needed to clarify this question.

In our model system, just like for IL-15 TP, members partaking in AP had different expression levels. Therefore, not surprisingly, the lower-expressed CD3 showed a larger extent of translocation to the IS than MHC II. CD3 only translocated to the synaptic region when Ag was presented by the MHC II, similar to the behavior of the IL-2/15Rβ subunit. The Kv1.3 potassium channel is crucial in T cell activation, and its colocalization with CD3 was shown, just as was its enrichment at the IS (64, 65). We set out to investigate whether the IL-2/15Rβ subunits had a similar association with the TCR/CD3 complex; however, our FRET measurements did not find significant cis-association between IL-2/15Rβ and CD3. Taken together, cooperation between MHC II–IL-15Rα on the presenting cell and IL-2–15Rβ-CD3 on the responding cell show contrasting behavior: The former pair moves jointly to the IS, whereas the latter two seem to move independently from each other.

Our results suggest that IL-15 TP between APCs and T cells does not just happen as an accompaniment to AP but may also occur independently. We may think of IL-15 TP as an autonomous, self-sufficient process keeping up memory T cell viability as these cells encounter IL-15–presenting cells during their lifespan.

We can summarize the molecular model of the trans- and cis-interactions between the components of the IL-15 TP and the AP complex as follows (Fig. 7). IL-15 TP and AP are autonomous processes, which can occur on their own or simultaneously. If one of the ligands (IL-15 or Ag) is present, the corresponding receptors on the B and T cells translocate to the synapse and generate cell–cell interaction by forming an intercellular protein complex. On the one hand, IL-15Rα and MHC II on the B cell are associated with each other; thus, when the liganded protein moves to the IS, the other protein follows it and also gets enriched there. On the other hand, IL-2/15Rβ and CD3 on the T cell do not interact with each other; therefore, only the protein whose ligand is presented by the B cell will move to the synapse, and the other protein will not get enriched there unless its own ligand is also present. IL-15 TP can initiate signaling by the T cell alone, and even though assembly of IL-2/15Rβ and IL-15Rα is enhanced by AP, it does not influence the efficiency of IL-15 TP-induced JAK/STAT signaling. There is no synergy between TCR and IL-15 signaling: IL-15 TP slightly reduces Ag-induced CD3ζ phosphorylation.

We thank Edina Nagy for excellent technical assistance.

The authors have no financial conflicts of interest.