An activation-induced IL-15 isoform is a natural antagonist for IL-15 function

Interleukin 15 (IL-15) expression induces the secretion of inflammatory cytokines, inhibits the apoptosis of activated T cells and prolongs the survival of CD8+ memory T cells. Here we identified an IL-15 isoform lacking exon-6, IL-15ΔE6, generated by alternative splicing events of activated immune cells, including macrophages and B cells. In vitro study showed that IL-15ΔE6 could antagonize IL-15-mediated T cell proliferation. The receptor binding assay revealed that IL-15ΔE6 could bind to IL-15Rα and interfere with the binding between IL-15 and IL-15Rα. Over-expression of IL-15ΔE6 in the murine EAE model ameliorated the EAE symptoms of the mice. The clinical scores were significantly lower in the mice expressing IL-15ΔE6 than the control mice and the mice expressing IL-15. The inflammation and demyelination of the EAE mice expressing IL-15ΔE6 were less severe than the control group. Furthermore, flow cytometry analysis demonstrated that IL-15ΔE6 expression reduced the percentages of inflammatory T cells in the spleen and spinal cord, and inhibited the infiltration of macrophages to the CNS. Our results demonstrated that IL-15ΔE6 could be induced during immune activation and function as a negative feedback mechanism to dampen IL-15-mediated inflammatory events.

As a potent immunomodulator, the expression of IL-15 is tightly controlled and its mRNA is distributed across various cell types and tissues, including activated monocytes/macrophages, DCs, epithelial cells, and placenta, kidney, lung, heart, and skeletal muscle 17,18 . Although IL-15 mRNA expression is widespread, measuring sufficient quantities of IL-15 in the culture supernatant is challenging due to the regulation of IL-15 protein production 18,19 . Multiple AUG sequences in the 5′ -UTR, regulatory sequence in the C terminus of IL-15 mRNA reduce IL-15 translation [19][20][21] . Alternative splicing also regulates IL-15 expression and previous work confirms that five alternative spliced forms of IL-15 mRNA have been identified, three of which encode an identical mature IL-15 protein with differences only in signal peptide coding regions 19,[21][22][23][24] . The other two isoforms were found in mouse intestinal epithelia, lacking exon 6 or a portion of exon 7, and both encode in-frame IL-15 proteins 25 . Although the lacking exon-6 and -7 isoforms are thought to inhibit IL-15-mediated cell proliferation in vitro, it is unclear whether they can be produced by immune cells other than intestinal epithelium or whether their expression is physiologically relevant in disease models.
IL-15 is clinically relevant for autoimmune diseases. Serum IL-15 level is increased in the patients with psoriasis, inflammatory bowel disease, systemic lupus erythematosus, inflammatory synovitis, asthma bronchiale, CTLL-2 assay. CTLL-2 cells were used 2 days after the last feeding. For the assay, CTLL-2 cells were washed three times in medium without IL-2. Prepare dilutions of purified IL-15, IL-15Δ E6, or commercially bought IL-15 in RPMI1640 complete medium in 96-well plates followed by addition of 5 × 10 4 CTLL-2 cells/100 μ l/well. For the blocking assay, 40 μ g/ml purified IL-15, 2.5 ug/ml commercially bought rhIL-15 or 1.25 ug/ml LPS was added into the wells containing titrated purified IL-15Δ E6 or IL-15. In the bioassays for binding activity, the mixture of IL-15Rα /IL-15 or IL-15Rα /IL-15Δ E6 were added in the medium. In a separate experiment, increasing amounts of IL-15Δ E6 were added to the mixture of IL-15Rα /IL-15. The CTLL-2 cells were then added. The plates were incubated in a humidified 37 °C, 5% CO 2 incubator a total of 24 h. 1 μ Ci/well [ 3 H] thymidine (Shanghai Institute of Physics, Chinese Academy of Sciences, Shanghai, China) was added to the culture for the last 8 h of the assay. Cells were harvested and [ 3 H] thymidine was measured on a beta plate reader (PerkinElmer Instruments, Meriden, CT).
Protein interaction assay. The CheckMate ™ /Flexi ® Vector Mammalian Two-Hybrid System (Promega, Fitchburg, WI) was used to validate interactions between IL-15Δ E6 and IL-15Rα and compare the affinity of IL-15/IL-15Rα with IL-15Δ E6/IL-15Rα . The coding regions of IL-15 or IL-15Δ E6 were transferred into pFN10A (ACT) vector and the coding regions of IL-15Rα was transferred into pFN11A (BIND) vector. These constructs and Pgl4.31[luc2P/GAL4UAS/Hygro] vector were purified and combined and transfected into CHO cell line. The pACT and pBIND vectors were transfected together as negative control. The pACT-MyoD and pBINDId control vectors were used together as positive control. The bioluminescent signal from this mammalian two-hybrid experiment was measured by microplate reader (Winooski, VT). Renilla luciferase activity expressed from the Scientific RepoRts | 6:25822 | DOI: 10.1038/srep25822 BIND-type vectors can be used to normalize transfection efficiency. The binding ratio was calculated as follows: firefly luciferase reading/Renilla luciferase reading. The coding region of IL-15 or IL-15Δ E6 was amplified by RT-PCR and inserted into minicircle (MC) plasmid (pMC.EF1, SBI, Mountain View, CA). Plasmid DNA was purified by Maxi-prep Kit (Axygen, Union City, CA). For hydrodynamic gene transfer, the mice were injected i.v. with 25 μ g of the recombinant plasmid in a total of 1 ml saline solution within 5s using a 23-gauge needle three days before disease induction.
Western blotting. Murine splenocytes lysate or purified proteins were applied to 10-20% SDS polyacrylamide gel and electro-transferred to PVDF membrane (Millipore, Billerica, MA). The membrane was blocked with 5% non-fat milk in PBS at 4 °C overnight. The PVDF membrane was then incubated with rabbit anti-mouse-IL-15 polyclonal Ab followed by secondary Abs goat-anti-rabbit IgG (ABcam, Cambridge, MA) conjugated with HRP, and developed by ECL-advanced western detection kit (Cell Signaling Technology, Danvers, MA). EAE model. Female mice (7-8 weeks old) received an HGT injection 3 days before EAE induction and were divided into 3 groups as follows: mice received HGT injection of plasmid DNA encoding IL-15 gene; mice received HGT injection of plasmid DNA encoding IL-15Δ E6 gene; mice received HGT injection of minicircle plasmid. Subsequently, the mice were injected subcutaneously with 300 μ g MOG (Sangon Biotech, Shanghai, China) in complete Freund's adjuvant (Sigma-Aldrich, St. Louis, MO) mixed with 4 mg/ml Mycobacterium tuberculosis (BD Difco, San Jose, CA). 500 ng pertussis toxin (List Lab, Campbell, CA) was injected intraperitoneally on days 0 and 2. Clinical signs were assessed daily as follows: 0, no clinical signs; 0.5, partially limp tail; 1, paralyzed tail; 2, mild hind-limb paresis; 3, severe hind-limb paresis; 4, hind-limb paralysis; 5, hind limbs paralyzed and partial forelimb weakness. The clinical scores were evaluated and recorded in a double blinded fashion.
Histology and immunohistochemistry. Brain and spinal cord tissues fixed in 10% phosphate-buffered formalin (pH 7.4) were dehydrated in 100% ethanol and stained with hematoxylin and eosin to identify areas of demyelination and the number of parenchymal inflammatory lesions. To determine the number of immune cells per section, the number of immune cells in three randomly chosen lesions was counted using ImageJ software. The resulting mean value of immune cells/mm2 was then multiplied by the total lesion area for that section to yield the total number of immune cells/section.
To detect the expression of IL-15 and IL-15 isoform in the mice after HGT treatment, liver tissues fixed in 10% phosphate-buffered formalin (pH 7.4) were dehydrated in 100% ethanol and embedded in paraffin wax at 58 °C. The paraffin sections were stained with a polyclonal rabbit Ab to murine IL-15 (Abcam, Boston, U.K.). The IL-15 Ab was diluted in PBS (pH 7.4) and applied at concentrations of 1:1000 at 37 °C for 30 min. Endogenous peroxidase activity was blocked with 3% H 2 O 2 and methanol. Avidin and biotin pretreatment was used to prevent endogenous staining. The secondary Ab was biotinylated goat anti-rabbit IgG used at 1:2000 dilution (Abcam, Boston, U.K.). For the analysis of macrophage infiltration in CNS, brain and spinal cord tissues fixed in 10% phosphate-buffered formalin (pH 7.4) were dehydrated in 100% ethanol and embedded in paraffin wax at 58 °C. The paraffin sections were stained with a polyclonal rabbit Ab to murine F4/80 (Santa Cruz Biotechnology, Taxes) at concentrations of 1:500 followed by secondary Ab staining which was goat anti-rabbit IgG used at 1:2000 dilution. Color development was performed with the amino ethylcarbazole detection kit from Ventana Medical Systems (Beijing Biosynthesis Biotechnology, Beijing, China). Statistical analysis. One-way ANOVA was used to determine statistically significant differences among more than two experimental groups. Unpaired Student t tests were used to determine statistically significant differences between two experimental groups. Data were analyzed using GraphPad Prism 5 software for Windows (GraphPad Software, San Diego, CA). P value < 0.05 was considered statistically significant (*), less than 0.01 or 0.001 was shown as ** or ***, respectively.

Results
Exon-6 lacking IL-15 isoform can be generated by alternative splicing in activated B cells and macrophages. It was reported that the level of IL-15 transcription was increased in the immune cells after stimulation with LPS 24,48 . Interestingly, we observed a smaller product of IL-15 from RT-PCR reaction using specific primers for IL-15 and the cDNA from splenocytes stimulated with LPS for 24 h (Fig. 1a). This smaller product was not present in the unstimulated splenocytes. The smaller PCR products were sequenced and proved to be the exon-6-missing IL-15 isoform (Fig. 1b).
As LPS mainly activates B cells and macrophages, the activation-induced production of IL-15Δ E6 could be from B cells and macrophages. In order to confirm that, we stimulated mouse primary B cells, macrophages, mouse myeloma cell line SP2/0 and mouse monocyte-macrophage leukemia cell line RAW246.7 with LPS (Fig. 1c). The RT-PCR results indicated that the expression of IL-15Δ E6 could be observed in all these cells after LPS stimulation. Primary T and NK cells served as negative control in these experiments and showed no IL-15 expression (Supplemental Fig. 1). Macrophages from TLR4 deficient mice failed to express IL-15Δ E6 upon LPS stimulation, suggesting that this isoform expression was indeed activation-induced (Supplemental Fig. 1). To further determine whether this IL-15 isoform was generated by an alternative splicing and not by enzymatic process of the protein, we treated LPS-stimulated splenocytes with cycloheximide (CHX), a drug that could inhibit protein translation (Fig. 1d). After 12 h of treatment, the expression level of IL-15Δ E6 was significantly reduced together with IL-15 in the LPS-stimulated splenocytes, maintaining a similar ratio as in the control cells, suggesting that the protein synthesis was required for the generation of IL-15Δ E6. These results demonstrated that IL-15Δ E6 was generated by alternative splicing in activated B cells and macrophages.

IL-15ΔE6 suppresses IL-15-mediated T cell proliferation in vitro.
In order to determine the biological function of IL-15Δ E6, both IL-15Δ E6 and IL-15 were expressed in E. coli and purified using a Ni-NTA agarose column. CTLL-2 is an IL-2 dependent T cell line which can also respond to IL-15. We cultured the CTLL-2 cells in the presence of IL-15Δ E6 or IL-15 (Fig. 2a). The recombinant IL-15 protein promoted the growth of CTLL-2 cells in a dose-dependent manner while recombinant IL-15Δ E6 protein couldn't induce CTLL-2 cell proliferation. To further assess whether IL-15Δ E6 had an antagonistic effect on IL-15-mediated CTLL-2 cell proliferation, the recombinant IL-15Δ E6 protein were added to the CTLL-2 cells cultured with recombinant IL-15 (Fig. 2b). Addition of IL-15Δ E6 significantly down-regulated IL-15-mediated CTLL-2 proliferation in a dose-dependent manner.
To further confirm the antagonistic effect of IL-15Δ E6, commercially bought IL-15 standard was used in CTLL-2 culture and recombinant IL-15 or IL-15Δ E6 protein were added to the culture (Fig. 2c). As expected, IL-15 protein slightly promoted CTLL-2 proliferation. However, IL-15Δ E6 significantly reduced the proliferation of CTLL-2 cells supported by IL-15 standard. Since IL-15Δ E6 was identified in the LPS stimulated splenocytes, we also tested whether IL-15Δ E6 could exert inhibitory effect on splenocyte proliferation stimulated with LPS (Fig. 2d). LPS was used to stimulate murine splenocytes together with recombinant IL-15 or IL-15Δ E6. We found that IL-15Δ E6 effectively suppressed LPS-induced splenocyte proliferation in a dose-dependent manner. These results demonstrated in vitro that IL-15Δ E6 induced by stimulation could antagonize the function of IL-15 in promoting T cell proliferation.  The IL-15Rα coding region was inserted into the pBIND vector. The pGL4.31 vector containing five GAL4 binding sites was used as a reporter for interactions between proteins. To validate whether the binding between IL-15Δ E6 and IL15Rα could impact the binding efficacy between IL-15 and IL-15Rα , we also transferred the coding regions of IL-15 and IL-15Δ E6 into the pACT vector without VP-16 coding region respectively. We then transfected these purified vector constructs into CHO cell line. The control and experimental groups were set up according to Fig. 3a. The supernatant of cell lysis was collected 48 hours after transfection and read after adding the substrate of firefly luciferase and Renilla luciferase respectively. The result showed that IL-15Δ E6 could interact with IL-15Rα (Fig. 3b). It slightly inhibited the binding of IL-15 to IL-15Rα , but the difference did not reach statistical significance (The IL-15 group compared with the IL-15 plus IL-15Δ E6 inhibition group). Also in the presence of IL-15, the binding efficacy between IL-15Δ E6 and IL-15Rα was down-regulated (The IL-15Δ E6 group compared with the IL-15Δ E6 plus IL-15 inhibition group, p < 0.05). The extent of inhibition with IL-15 was a bit more than the inhibition with IL-15Δ E6, suggesting the affinity between IL-15Δ E6 and IL-15Rα might be slightly lower than that between IL-15 and IL-15Rα .

IL
To further demonstrate the biological function of IL-15Δ E6 on the formation of IL-15/IL-15Rα superagonist, we performed proliferation assays with the CTLL-2 in the presence of IL-15/IL-15Rα superagonist or IL-15Δ E6/IL-15Rα mixture (Fig. 3c). The IL-15/IL-15Rα superagonist significantly promoted the proliferation of CTLL-2 cells. However, the mixture of IL-15Rα and IL-15Δ E6 did not have any effect on the proliferation. Adding increasing amounts of IL-15Δ E6 to the mixture of IL-15Rα and IL-15 dose-dependently reduced the proliferation of CTLL-2 cells (Fig. 3d), suggesting that IL-15Δ E6 could compete for the binding of IL-15Rα and interfere with the generation of the IL-15Rα /IL-15 superagonist.

IL-15ΔE6 expression significantly reduces the severity of EAE.
In order to dissect the in vivo function of IL-15Δ E6, we overexpressed IL-15Δ E6 and IL-15 by hydrodynamic gene transfer (HGT) method in a murine EAE model. The IL-15 and IL-15Δ E6 expressions were quite sustainable. Liver tissues from mice that received HGT injection of IL-15 and IL-15Δ E6 plasmids showed a significantly higher and comparable levels of IL-15 detected by antibody that could recognize the common regions of IL-15 and IL-15Δ E6 (Fig. 4a). The clinical scores of the control mice reached the peak (3.5-4) around day 14 post immunization (Fig. 4b).
Mice received IL-15 HGT had earlier onset of the disease and increased severity compared with the control mice (p < 0.05). In marked contrast to IL-15 and the control group, over-expression of IL-15Δ E6 significantly reduced the severity of EAE as mice showed lower clinical scores and better movement compared to the control group (p < 0.01) and the IL-15 group (p < 0.001). None of the mice overexpressing IL-15Δ E6 had severe EAE symptoms (defined as a clinical score ≥ 3), only mild weakness of hind legs and tail atony was observed. As the injuries of myelin sheath and the inflammation in the CNS are the primary causes of EAE, we first validated whether the reduced EAE symptoms by IL-15Δ E6 were associated with lessened damage in CNS. As expected and consistent with the clinical score results, the HE staining of spinal cord and brain showed that the inflammatory lesions were fewer in the IL-15Δ E6 group compared with IL-15 group and the control group (Fig. 4c,d). IL-15 significantly increased the number of infiltrating immune cells in the CNS, while IL-15Δ E6 expression dramatically reduced it. These results demonstrated that IL-15 could aggravate EAE while IL-15Δ E6 reduced the severity of EAE, possibly by antagonizing the effect of IL-15.

IL-15ΔE6 expression decreases the infiltration of macrophages in the CNS and down-regulates the percentages of inflammatory T cell subsets.
Previous research had proven that many immune cell subsets, such as Th1, Th17 and macrophages participated in the course of EAE 49,50 and regulatory T cells inhibited the disease progression [51][52][53] . To assess whether the reduced disease severity by IL-15Δ E6 expression was associated with the changes in the immune cell subsets, lymphocytes were acquired from spleen and spinal cord of EAE mice and stained with antibodies specific for different cell surface and intracellular markers (Fig. 5). By flow cytometry analysis, we found that the proportions of NK cells and macrophages were significantly reduced in the spleens of mice expressing IL-15Δ E6 compared with control group and IL-15 group (Fig. 5a). Percent of CD4 + T cells was slightly reduced in the IL-15Δ E6 group compared with the IL-15 group, while CD8 + T cells and NKT cells remained unchanged among three groups. Meanwhile, the percentages of regulatory T cells were increased by IL-15Δ E6 expression (p < 0.05).
Th1 and Th17 cells are commonly considered to be the key cell subsets in the occurrence and development of EAE. We further assessed the changes of Th1 and Th17 cells by intracellular staining (Fig. 5b,c). The percentages of CD4 + IFN-γ + T and CD4 + IL-17 + T cells were significantly down-regulated in the spleens of mice expressing IL-15Δ E6 compared with the IL-15 group (p < 0.01, p < 0.05) (Fig. 5b). The percentages of CD4 + TNF-α + and CD8 + TNF-α + T cells were also decreased in the IL-15Δ E6-expressing mice (p < 0.01). Similar results were obtained in the CNS. We found a significant increase of macrophage infiltration in the IL-15 group (p < 0.05) while the macrophage infiltration was reduced in the IL-15Δ E6 group compared with the IL-15 and the control  0.001, Fig. 5c). These results were also confirmed by immunohistochemistry analysis of macrophage infiltration in the CNS (Fig. 5d). The percentages of both CD4 + IFN-γ + and CD4 + TNF-α + T cells were significantly decreased in the 15Δ E6-expressing mice (p < 0.001, p < 0.05) (Fig. 5c). Taken together, our results demonstrated that IL-15Δ E6 expression decreased the infiltration of macrophages in the CNS and down-regulated the percentages of both CD4 + IFN-γ + and CD4 + TNF-α + T subsets in both spleen and CNS.

Discussion
IL-15 expression is tightly regulated at multiple levels due to its important immunomodulatory function 18,19 . The interferon response factor element (IRF-E) and nuclear factor NF-κ B are critical for regulation of IL-15 mRNA expression 20,54,55 . In terms of translational control, multiple upstream AUGs in the 5′ -UTR, regulatory sequence in the C terminus of IL-15 mRNA, and long signal peptides tightly regulate translation and secretion of IL-15 21,56 . In addition, IL-15 expression can be controlled by alternative splicing which has been implicated in various aspects of cytokine production and signaling 21,57 . In this way, cytokine diversity is generated without the requirement for the expression of a new gene 58,59 . In the current study, we identified an IL-15 isoform lacking exon-6, IL-15Δ E6, generated by alternative splicing events of activated immune cells, including macrophages and B cells. In vitro data confirm that IL-15Δ E6 can antagonize IL-15-mediated T cell proliferation. Receptor binding assay reveals that IL-15Δ E6 can bind to IL-15Rα and interfere with binding between IL-15 and IL-15Rα. Over-expression of IL-15Δ E6, in a murine EAE model, dramatically ameliorated EAE symptoms and reduced the percentages of inflammatory T cells in the spleen and spinal cord of the EAE mice, as well as inhibited infiltration of macrophages to the CNS. Our data show that IL-15Δ E6 could be induced during immune activation and function as a negative feedback mechanism to dampen IL-15-mediated inflammatory events.
Five variants of IL-15 have been identified so far, including all combinations with an exon 2 and exon-6 deletion, a portion of exon 7 deletion and an alternative use of exon 5 or 5′ 20,23-25 . All of these variants encode the same mature IL-15 protein except for variants lacking exon 6 and a portion of exon 7. Although the exon 6-missing IL-15 isoform was first identified in mouse intestinal epithelium 25 , we reported, for the first time, that it can be generated by alternative splicing events in the activated immune cells. For many known cytokines, including IL-2 and IL-4, alternatively spliced isoforms are natural antagonists with competitive inhibitory effects on cytokine function [60][61][62] . Of note a report indicated that when macrophages were stimulated with LPS, an IL-15 isoform utilizing alternative exon 5 and exon 2 deleted-isoform were obtained 48 . This discrepancy between their results and ours may be due to the short period of stimulation in their study (6 h). We stimulated splenocytes, macrophages and B cells with LPS for 24 h to in order to observe the generation of IL-15Δ E6. This negative feedback mechanism mediated by IL-15Δ E6 may occur during late-stage of activation.
IL-15 is critical for autoimmune and inflammatory disease and previous studies show that the increased expression of IL-15 is associated with rheumatoid arthritis [63][64][65] , sarcoidosis and inflammatory bowel disease 29,66,67 . IL-15 also contributes to MS [38][39][40] . IL-15 up-regulation is observed in serum and cerebrospinal fluid of MS patients 41,42 and mice of experimental autoimmune encephalomyelitis (EAE) 2 . IL-15 contributes to the tissue damages caused by CD8 + T cells and Th17 cells in MS patients and EAE models 45,46 . In addition, IL-15 is found to actively modulate microglial reactivity, co-activating the MAPK and NF-κ B pathways, controlling additional cytokine and chemokine release 38 . Therefore, IL-15 may contribute to the disease progression of MS and EAE. Consistent with these previous reports, our results showed that mice over-expressing IL-15 had accelerated and more severe EAE (Fig. 4). Furthermore, the IL-15E6 isoform, which can negatively regulate the function of IL-15, alleviated EAE symptoms. However, IL-15 knockout mice exhibits aggravated EAE symptoms 2,44,68 . Because IL-15 is involved in NK and NKT cell development and in homeostatic maintenance of the T cells, the knockout mice data may not represent effector functions of IL-15 during EAE development. Rather, they may indicate that IL-15 is important for the development of a cell subset that could be protective against EAE. Moreover, IL-15 treatment did not significantly affect disease severity in both WT and knockout mice in the same report 2 . Differences in our and other's results in terms of IL-15 treatment in EAE may be due to the level of IL-15 proteins. The HGT method using minicircle vector enables more efficient and sustained expression over a period of weeks compared to injection of recombinant proteins that only works for a few days 69,70 . Therefore, high IL-15 expression may result in the pathological role of IL-15 we observed during EAE development.
Although IL-15 mainly function on CD8 + T and NK cells, it could also influence the differentiation and maintenance of the CD4 + T cell subsets by modulating the specific signaling pathways together with IL-15Rα 71,72 . We found that IL-15Δ E6 could interact with IL-15Rα and partially inhibit binding of IL-15 to IL-15Rα which can be presented in trans to the IL-2Rβ γ c complex and initiates a series of signaling events involving JAK/STAT, phosphoinositide 3-kinase (PI3K) and the mitogen-activated protein kinase (MAPK) signaling pathways 73 . The binding of IL-15Δ E6 to IL-15Rα may prevent its association with the IL-2Rβ /γ c to activate down-stream signaling pathways. However, how IL-15Δ E6 might interact with IL-15Rα and affect the IL-15 signaling in inflammatory conditions requires further investigation.
In conclusion, we identified an IL-15 mRNA isoform lacking exon 6, IL-15Δ E6, in activated macrophages and B cells which acts as a natural antagonist for IL-15 and may hold promise as a novel therapeutic agent for IL-15-mediated inflammatory diseases.