Rapid endocytosis of interleukin-15 by cerebral endothelia

Abstract

Interleukin (IL)-15 receptors are present in the cerebral endothelia composing the blood-brain barrier where they show robust upregulation by neuroinflammation. To determine how IL15 receptor subunits participate in the endocytosis and intracellular trafficking of IL15, we performed confocal microscopic imaging and radioactive tracer uptake assays in primary brain microvessel endothelial cells (PBMEC) and related cell lines transfected with modulatory molecules. By immunostaining and co-localization studies with organelle markers, we showed that IL15 was rapidly endocytosed via lipid rafts and was directed to diverse intracellular pathways. During the course of intracellular trafficking, Alexa dye-conjugated IL15 was partially co-localized with both the specific receptor IL15Rα and the co-receptor IL2Rγ. However, deletion of one of the receptor subunits had only a minor effect in slowing IL15 uptake when PBMEC from the receptor knockout mice were compared with those from wildtype mice. IL15 was trafficked to early, recycling, and late endosomes, in the Golgi, and to lysosomes. The diffuse distribution suggests that IL15 activates multiple endothelial signaling events.

INTRODUCTION

Interleukin (IL)-15 is a ubiquitously expressed 14 kD cytokine with diverse biological actions. It binds to its unique receptor IL15Rα as well as two co-receptors IL2Rβ and IL2Rγ. This heterotrimeric receptor complex is the high affinity form that activates Janus kinase (JAK) and Signal Transducer and Activator for Transcription (STAT) proteins (Giri et al. 1995; de Jong et al. 1996). IL15 shows limited permeation across the blood-brain barrier (BBB) in the basal state. In this study, we determined the mechanisms of endocytosis and intracellular trafficking of IL15 in cerebral microvessel endothelial cells composing the BBB.

The BBB endothelia are joined by tight junctions, underlaid by a continuous basement membrane, and reinforced by other cellular components such as astrocytic endfeet, pericytes, and microglia. The surface area of the BBB is 150 to 200°cm² g⁻¹ tissue, thus providing a large interface for blood–brain exchange (Abbott et al. 2006; 2010). Blood IL15 levels are low in the resting state but increase after inflammatory and autoimmune challenges. Tumor necrosis factor α (TNF) stimulates IL15 production in cerebral endothelia. TNF also upregulates IL15 receptors both in cerebral microvessels and cultured RBE4 endothelial cells (Pan et al. 2009). Lipopolysacchride (LPS), a prototypic inflammatory stimulus, increases the mRNA of all three IL15 receptor subunits in both cerebral microvessels and CNS parenchyma (Pan et al. 2010). Experimental autoimmune encephalomyelopathy (EAE) also increases IL15 receptor mRNA in all brain and spinal cord regions studied (Wu et al. 2010c). IL15 crosses the BBB, and its permeation can be accelerated by LPS (Pan et al. 2008). This raises the possibility that blood-borne IL15 may increase cerebral IL15 signaling through upregulated receptors during inflammation.

The beneficial effect of IL15 to CNS homeostasis is reflected by the increased susceptibility of IL15 knockout (KO) mice to EAE and the therapeutic effect of IL15 in improving EAE (Wu et al. 2010c). In a study of facial nerve regeneration, IL15Rα KO mice had dramatically reduced regeneration although IL15 KO mice did not show an apparent effect (Huang et al. 2007). Even in the resting state, IL15Rα KO mice have profound neurobehavioral deficits, including altered circadian rhythms of temperature and activity (He et al. 2009), impaired spatial learning and memory with deficits in GABA transmission (He et al. 2010), lack of normal anxiety (Wu et al. 2010a), and depressive-like behavior in association with decreased serotonin reuptake (Wu et al. 2010b). In fetal, developing, and adult mouse brain, IL15 and IL15Rα are ubiquitously and constitutively expressed in hippocampus, cerebellum, cortex, and thalamus. Both IL15 and its receptor show a greater distribution in cytoplasm than in membrane (Hanisch et al. 1997; Kurowska et al. 2002). Altogether, these results suggest that IL15 signaling plays crucial roles in both acquired and innate immunity involving the CNS.

Although the BBB permeability to IL15 is low under basal conditions, LPS can increase IL15 influx to a greater extent than would be expected from BBB disruption measured by the vascular marker albumin (Pan et al. 2008). Together with the observation of upregulated IL15 receptors in enriched microvessels from these LPS treated mice (Pan et al. 2010), we hypothesized that IL15 crosses the BBB by a receptor-mediated transport system. Based on results from other types of cells (Hemar et al. 1995; Fehniger and Caligiuri 2001; Budagian et al. 2006), we postulated a model of receptor-mediated endocytosis of IL15 shown in figure 1. All three receptor units have an IL15-binding Sushi domain (shown in the left panel), and all are co-localized in lipid rafts, at least in T lymphoma cells (Bodnar et al. 2008). The right panel depicts the dynamics after IL15 binds to its high affinity receptor IL15Rα (step 1). After the IL2Rβ and IL2Rγ receptors with moderate affinity are recruited, the resulting heterotrimeric complex is internalized with the IL15 ligand (step 2). IL15Rα is dissociated after exiting early endosomes, enters recycling endosomes, and returns to the cell surface (3a). IL2Rβ/γ may dissociate from IL15 that is then directed to late endosomes and lysosomes for degradation (3b), or IL2Rβ/γ may continue to serve as chaperon proteins for trafficking of intact IL15 (3c). The pool of IL15 that is transported across the BBB endothelia uses cytoskeletal motors for further migration of vesicles toward the basolateral surface, eventually being exocytosed (step 4). In this study we determined the kinetics of IL15 endocytosis and the involvement of receptors using endothelial cells of the BBB and KO mice lacking IL15Rα or IL2Rγ.

Fig.1.

Model of receptor-mediated endocytosis of IL15. The various steps of receptor recruitment and cooperation in IL15 endocytosis and trafficking are described in the text.

MATERIALS & METHODS

Cells

Following an approved Institutional Animal Care and Use protocol, primary mouse brain microvessel endothelial cells (PBMEC) were isolated from brain cortices of wildtype or genetically modified mice. This includes 4–6 week old C57 (B6) mice, IL15Rα KO and matching B6129SF2/J controls, and strain IL2Rγ KO and matching B6 controls, all from Jackson Laboratories (Bar Harbor, ME). The method of obtaining PBMEC was similar to established protocols (Pan et al. 2003;Song and Pachter 2004) but included several modifications. Dounce tissue grinders with loose and tight pestles were used to homogenize cerebral cortex after removal of meninges and large vessels. Myelin was removed by centrifugation in buffer A (HBSS, 20 mM HEPES, 1× penicillin /streptomycin) containing 25 % bovine serum albumin (BSA). The pellet was filtered through a 40 µm cell strainer to remove single cells and debris. The microvessels were digested with 1mg/ml collagenase/dispase (Roche Diagnostics, Indianapolis, IN) and TLCK (Sigma, St. Louis, MO) for 5–10 min. The short microvessel fragments were collected after filtration with a 100 µm cell strainer to remove macrovessels and plated onto 8-well Permanox chamber slides (Thermo Fisher Scientific, Pittsburgh, PA). The M131 media was supplemented with 15 % platelet depleted serum (PDS) and 5 % microvessel growth supplement (MVGS, Invitrogen, Carlsbad CA). The slides were coated with a mixture of rat tail collagen (50 µg/ml; BD Biosciences, San Jose, CA), BSA (1 %) and human fibronectin (10 µg/ml; Invitrogen) in M131 media. For the first 3 days, the cells were cultured in the presence of 3 µg/ml puromycin to remove non-PBMEC (Calabria et al. 2006).

For IL15 uptake experiments, both the rat brain microvessel endothelial cell line RBE4 and the human brain microvessel cell line HCMEC/D3 (kindly provided by Dr. Pierre-Olivier Couraud, Institut Cochin, Paris, France) was used. The cells were plated onto 24-well plates coated with collagen (50µg/ml). RBE4 cells were cultured in αMEM/Ham’s F10 (1:1; Invitrogen) supplemented with 10 % FBS (heat-inactivated; Invitrogen), βFGF (1ng/ml; Invitrogen), and G418 (30 µg/ml; Invitrogen). HCMEC/D3 cells were cultured in EBM-2 media (Lonza, Basel, Switzerland) containing all provided supplements but heparin.

Transfection experiments

To determine the respective roles of IL15Rα, Vesicle Associated Membrane Proteins (VAMP)-2, and VAMP3 in IL15 uptake, RBE4 cells were transfected with IL15Rα-GFP generated in our lab (Wu et al. 2010d), VAMP2-GFP, or VAMP3-GFP (kind gifts of Drs. William S. Trimble and Amira Klip, Program in Cell Biology, The Hospital for Sick Children, Toronto, Canada). Lipofectamine 2000 (Invitrogen) was used according to the manufacturer’s instructions. At 24 h after transfection, cells were treated with IL15-Alexa568 for 10 min and fixed for imaging on a confocal microscopy.

IL15 uptake studies

Experiments to determine the time-dependent uptake of IL15 were performed on PBMEC. Before treatment, cells were serum-starved for 1 h. Carrier-free recombinant human IL15 (PeproTech, Rocky Hill, NJ) was labeled with Alexa568 (Invitrogen) according to the manufacturer’s instructions. The final concentration of IL15-Alexa568 added to PBMEC was 290 nM. After incubation for 2, 5, 10, and 30 min, the cells were fixed by use of 4% paraformaldehyde (10 min at room temperature). As a negative control, Alexa568 (non-conjugated free dye) was incubated with PBMEC for 30 min and the data presented as time point 0. To determine potential mechanisms of IL15 endocytosis, PBMEC were pre-treated for 1 h with the inhibitors nystatin (50 µg/ml, Sigma) or chlorpromazine (10 µg/ml, Sigma). Then, IL15-Alexa568 was added for an additional 30 min and the cells were fixed.

The binding and endocytosis of ¹²⁵I-IL15 into RBE4 cells in the presence and absence of nystatin was tested as described previously (Tu et al. 2007a;2007b). IL15 was labeled by the iodogen method and used at an acid precipitation rate higher than 95%. At 2, 5, 10, and 30 min, unbound, bound, and endocytosed ¹²⁵I-IL15 were collected, and ¹²⁵I radioactivity (reflecting IL15 concentration) in each fraction was measured in a γ-counter (1470 Wallac WIZARD, Perkin Elmer, Wellesley, MA). Protein concentrations of the lysates were measured with the BCA method (Thermo Scientific). Uptake was calculated as the ratio between ¹²⁵I radioactivity in the lysate and all fractions and further normalized with the protein concentration in the lysate. All experiments were run in triplicate and repeated at least twice. Data are presented as mean± standard error. Statistical evaluation was performed by analysis of variance (ANOVA), followed by Tukey’s post-hoc multiple-comparison test where appropriate.

Transport of IL15 across the BBB of IL15Rα and IL2Rγ KO mice

To determine whether IL15Rα deletion causes a reduction of IL15 transport, two groups of young adult male mice were studied (IL15Rα KO and wild-type controls, n = 8–10 /group). Similarly, to determine whether IL2Rγ deletion causes a reduction of IL15 transport, two groups of young adult male mice were studied (IL2Rγ KO and wild-type mice, n = 8 – 10 /group). In addition to ¹²⁵I-IL15, ¹³¹I-albumin was used as the vascular control in the transport assays, as described previously (Pan et al. 2008). The linear regression correlation between brain/serum ratio of radioactivity [(cpm/g)/(cpm/µl), or µl/g] and exposure time (Kastin et al. 2001) was determined by use of the GraphPad Prism program (San Diego, CA). The influx rate is the slope of the regression line between the brain/serum ratio of radioactivity and the exposure time, and the volume of distribution is determined by the intercept. Differences between the groups were determined by the least-squares method with a built-in equation in the GraphPad Prism program. In the IL2Rγ KO study, cerebral cortex was further dissected and subjected to the capillary depletion procedure as described previously (Pan et al. 2008). This dextran density centrifugation procedure enables efficient separation of brain parenchyma and capillaries. The amount of ¹²⁵I-IL15 in the parenchymal and capillary portions was determined for each group.

Immunocytochemistry and Western blot analysis

Fixed PBMEC obtained from uptake experiments were washed with phosphate-buffered saline (PBS), permeabilized for 30 min in PBS/0.1% Triton X-100, and blocked for 1 h in 10% normal donkey serum in PBS. Primary antibodies were diluted in PBS/0.1% Triton X-100/1% BSA. For co-localization experiments we used combinations of goat anti-IL15Rα (1:200; sc-1524; Santa Cruz Biotechnology, Santa Cruz, CA) with either mouse anti-Lamp1 (1:300; ABR, Golden, CO), rabbit anti-EEA (1:200; ABR), mouse anti-α-tubulin (1:700; ZYMED, Carlsbad, CA), or rabbit anti-βCOP (1:150; ABR). Combinations of rabbit anti-IL2Rγ (1:200; sc-668; Santa Cruz) were with goat anti-rab7 (1:300; sc-6563, Santa Cruz), goat anti-IL15Rα, or mouse anti-Lamp1. Fixed cells were incubated with primary antibodies overnight at 4°C. Then cells were washed three times with PBS for 5 min and incubated with secondary antibody combinations of anti-mouse Alexa488 and anti-goat Alexa647, anti-mouse Alexa488 and anti-rabbit Alexa647, anti-rabbit Alexa488 and anti-goat Alexa647 (all 1:1000) for 1 h. After a thorough wash, PBMEC were mounted with Vectashield mounting media containing DAPI (Vector Laboratories, Burlingame, CA).

Microscopy

Immunofluorescence was captured on either an Olympus FV1000 inverted laser scanning microscope or a Zeiss Axioplan 2 fluorescence microscope connected to a Photometrics CoolSnap HQ CCD camera. Cells were visualized with a Planachromat 63×/1.4NA oil objective. Images were analyzed using the co-localization module of Imaris software (Bitplane AG, Zurich, Switzerland). After automatic thresholding of the 2-channel images, the Pearson’s correlation coefficient (r) was determined and used to evaluate co-localization.

RESULTS

1. Rapid endocytosis of IL15 and its diffuse intracellular fate in cerebral microvascular endothelial cells

IL15-Alexa568 appeared intracellularly within 2–5 min of incubation with PBMEC (Fig. 2A). The fluorescence showed a vesicular pattern of distribution in perinuclear regions. The free Alexa568 dye was not internalized, serving as a negative control. Consistent with the findings in PBMEC, IL15 also showed rapid endocytosis in the RBE4 cerebral endothelial cell line. The association of ¹²⁵I-IL15 in RBE4 cells resembles a two-site kinetic model. Most of the cellular uptake occurred in the first 10 min (Fig.2B). By contrast, the residual cell surface binding was lower than the amount endocytosed and did not change significantly over time. The uptake was temperature-dependent and much slower at 23 °C (data not shown).

Fig.2.

Kinetics of IL15 endocytosis in cerebral endothelia. (A) In PBMEC from wildtype mice, a vesicular pattern of IL15-Alexa568 was seen at 5, 10, and 30 min in perinuclear regions, although there was no apparent accumulation of fluorescence at 2 min. Bar: 10µm. (B) In RBE4 cells, the uptake of ¹²⁵I-IL15 showed biphasic association kinetics, with an initial phase of rapid endocytosis followed by a slow increase leading to a plateau. n = 3 /time point.

We next determined which membrane microdomains participated in the rapid endocytosis of IL15-Alexa568. Chlorpromazine was used as an inhibitor of clathrin-dependent endocytosis, whereas nystatin was used as an inhibitor of lipid raft-dependent endocytosis. Pretreatment of PBMEC with nystatin induced an apparent reduction in the intensity of IL15-Alexa568 fluorescence when examined at 30 min. By contrast, chlorpromazine did not change it (Fig.3A). ¹²⁵I-IL15 uptake was further quantified in both RBE4 and HCMEC/D3 cerebral endothelial cells in the presence or absence of nystatin. Similar to that seen in PBMEC, nystatin significantly inhibited IL15 uptake (Fig.3B). These results indicate that lipid rafts mediate the uptake of IL15 by brain microvessel endothelial cells.

Fig.3.

Effects of endocytosis inhibitors on endothelial uptake of IL15. (A) Pre- and co-treatment of PBMEC with nystatin abolished the intracellular signal of IL15-Alexa568 studied 30 min after initiation of endocytosis. By contrast, chlorpromazine did not show the same effect. Bar: 10µm. (B) The uptake of ¹²⁵I-IL15 by both RBE4 and HCMEC/D3 cells was significantly reduced in the presence of nystatin. ***: p < 0.005; *:p<0.05.

To determine the pathways used by IL15-Alexa568 after endocytosis, PBMEC were fixed at 2, 5, 10, and 30 min after IL15-Alexa568 was added and then immunostained with several organelle markers, including early endosome antigen-1 (EEA), Rab7 for late endosomes, Lamp1 for lysosomes, β-coatomer protein (βCOP) for the Golgi complex, and tubulin. Correlation analyses shows that IL15-Alexa568 had a low level of co-localization with EEA (r = 0.28±0.035), Rab7 (r = 0.57±0.04), Lamp1 (r = 0.43±0.04), and βCOP (r = 0.26±0.07) when all time points were considered. Concomitant with partial co-localization with these organelles, most of the IL15-Alexa568 appeared to be dispersed in other intracellular vesicles. Figure 4 shows representative images taken from cells after 10 min of IL15-Alexa 568 treatment.

Fig.4.

Diffuse distribution of IL15-Alexa568 in intracellular vesicles. At 10 min after initiation of endocytosis in PBMEC, IL15 showed partial co-localization with early endosomes (EEA), late endosomes (Rab7), lysosomes (Lamp1), the Golgi (βCOP), and tubulin. A large proportion of IL15-Alexa568 did not co-localize with any of these markers. Bar: 10µm.

Together, these results indicate that endocytosed IL15 can be found in early, recycling, and late endosomes, in the Golgi, and in lysosomes. This is a process of rapid endocytosis, rapid vesicular trafficking among organelles, and diverse cellular fates of degradation, recycling, or further trafficking to secretory pathways.

2. The vesicular trafficking of IL15 involves soluble N-ethylmaleimide-sensitive-factor attachment protein receptor (SNARE) proteins

SNAREs are plasma membrane-bound small proteins mediating vesicle fusion and are involved in cellular transport. Among the SNAREs, Vesicle Associated Membrane Proteins (VAMP) are anchored in vesicular membranes. VAMP2 and VAMP3 have been detected in vesicles containing endocytosed and recycling membrane receptors (McMahon et al. 1993). To test the hypothesis that VAMP2 and VAMP3 play a role in membrane fusion events during IL15 trafficking, we overexpressed VAMP2-GFP or VAMP3-GFP in RBE4 cerebral endothelial cells. After 10 min of incubation, IL15-Alexa568 showed partial co-localization with both VAMP2-GFP and VAMP3-GFP near the cell surface (arrows, Fig.5). Quantification by Pearson’s correlation analysis showed very little IL15-Alexa568 co-localized with SNARE proteins at this time (r = 0.04±0.04 for VAMP2-GFP and r = 0.12±0.02 for VAMP3-GFP).

Fig.5.

Role of SNARE proteins in IL15 endocytosis. At 10 min after initiation of endocytosis in RBE4 cells overexpressing VAMP2-GFP or VAMP3-GFP, IL15-Alexa568 showed partial co-localization (arrows) with both of these two SNARE proteins.

3. Role of receptors in IL15 endocytosis and intracellular trafficking

To determine whether IL15 enters PBMEC by receptor-mediated endocytosis, we performed IL15-Alexa568 endocytosis assays in three groups of cells at 2, 5, 10, and 30 min. In PBMEC from the wildtype mice, accumulation of IL15-Alexa568 in vesicles within the cytoplasm was apparent by 5 min (Fig.6, left panel). Absence of IL15Rα or IL2Rγ resulting from embryonic KO did not prevent IL15-Alexa568 from entering the cells. The patterns of cytoplasmic IL15-Alexa568 distribution were similar in the groups. Nonetheless, in PBMEC from IL15Rα KO (Fig.6, middle panel) or IL2Rγ KO mice (Fig.6, right panel), the fluorescent intensity at earlier time points appeared to be weaker, and the number of vesicles at each corresponding time point seemed to be fewer. This suggests slower accumulation of IL15-Alexa568 within PBMEC from the KO mice than in PBMEC from wildtype mice. However, no statistically significant difference was quantifiable. Altogether, these results suggest that neither receptor subunit by itself was sufficient to direct the trafficking of IL15-Alexa568.

Fig.6.

Comparison of IL15-Alexa568 uptake by PBMEC from wildtype, IL15Rα KO, and IL2Rγ KO mice. In comparison with the wildtype cells, the accumulation of fluorescence in PBMEC from the KO mice was slower and less at each time point, although the vesicular pattern was similar. DAPI nuclear staining is shown in blue. The negative control with Alexa568 free dye did not show Alexa568 (red) inside the cells after 30 min incubation (bottom row).

4. Interactions of IL15 with IL15Rα

PBMEC incubated with IL15-Alexa568 for 2, 5, 10, and 30 min were immunostained for IL15Rα. There was partial co-localization of IL15-Alexa568 with IL15Rα immunoreactivity at all time points. At the later time points, IL15-Alexa568 co-localized with IL15Rα in perinuclear regions (Fig.7A). In RBE4 cells transfected with IL15Rα-GFP, the overexpressed receptor was targeted mainly to the cell surface 24 h after transient transfection. There was partial co-localization of the receptor with IL15-Alexa568 (arrow) at early time points. IL15-Alexa568 was internalized within 5–30 min; its co-localization with IL15Rα-GFP was present in both cytoplasm and nucleus (arrows) (Fig.7B).

Fig.7.

The spatial and kinetic relationship between IL15-Alexa568 and IL15Rα during intracellular trafficking. (A) Partial co-localization of IL15-Alexa568 with IL15Rα immunoreactivity was seen at 2, 5, 10, and 30 min after initiation of endocytosis in PBMEC. The co-localization is marked by arrows. The negative control (PBMEC incubated with Alexa568 free dye for 30 min) did not show Alexa568 inside the cell. (B) In RBE4 cells overexpressing IL15Rα-GFP, IL15-Alexa568 was associated with the receptor both at the cell surface and in the cytoplasm 2 h after initiation of endocytosis. Bar: 10µm.

IL2Rγ also showed partial co-localization with IL15-Alexa568 during its course of endocytosis by PBMEC from wildtype mice (Fig.8A). The co-localization of IL15-Alexa568 and IL2Rγ immunoreactivity decreased over time, with the correlation coefficient being significantly lower at 10 min than 5 min (Fig.8B). This is opposite to that seen for IL15Rα. There was increasing co-localization of IL15-Alexa568 with IL15Rα over time, as the Pearson’s correlation coefficient was significantly higher at 30 min than 2 min (Fig.8B).

Fig.8.

Relationship between IL2Rγ and IL15-Alexa568 in PBMEC. (A) IL15-Alexa568 showed partial co-localization (arrows) with IL2Rγ immunoreactivity. The negative control in which PBMEC were incubated with Alexa568 free dye for 30 min did not show intracellular Alexa568 fluorescence. Bar: 10µm. (B) In contrast to a gradual increase of co-localization of IL15-Alexa568 with IL15Rα over time (determined from 6–10 cells /time point), there was a decrease of co-localization of IL15-Alexa568 with IL2Rγ (determined from 8–16 cells /time point by Pearson’s correlation coefficients). Data shown are means ± S.E. *: p < 0.05. ***: p < 0.005.

Complementary to the overexpression approach in RBE4 endothelial cells, we also used receptor knockout mice to determine the involvement of IL15Rα and IL2Rγ in ¹²⁵I-IL15 permeation across the BBB. The permeation of ¹²⁵I-IL15 in the IL15Rα KO mice (Ki = 0.26 ± 0.06 µl/g-min) did not differ significantly from that in wildtype B6.129s mice (Ki = 0.15 ± 0.08 µl/g-min). Both influx rates were significantly higher than that of the vascular permeability control ¹³¹I-albumin, showing the lack of disruption of the BBB in either strain (Fig.9A). In the IL2Rγ KO mice, the influx rate of ¹²⁵I-IL15 was 0.08 ± 0.05 µl/g-min. By contrast, the wildtype B6 control mice had a higher influx rate (Ki = 0.16 ± 0.05 µl/g-min). Because of large variation within the group, the reduction of IL15 influx in the KO group was not statistically significant. Nonetheless, the permeation of ¹²⁵I-IL15 was significantly higher than that of ¹³¹I-albumin within each group. Most of the ¹²⁵I was present in brain parenchyma and very little remained in the capillaries, indicating complete transcytosis (inset) (Fig.9B).

Fig.9.

Effects of receptor knockout on ¹²⁵I-IL15 across the BBB, determined 1 – 20 min after iv injection. (A) The permeation of ¹²⁵I-IL15 across the BBB was similar in the IL15Rα KO and wildtype mice, although the influx rate of IL15 in both groups was higher than the respective vascular permeability control ¹³¹I-albumin. (B) BBB permeability study showed that IL2Rγ KO mice appeared to show reduced influx of ¹²⁵I-IL15 into the whole brain and cerebral cortical parenchyma (inset), but the difference was not significantly different from the wildtype mice.

DISCUSSION

In this study we showed that cerebral microvascular endothelial cells can internalize IL15 rapidly via lipid rafts, and divert it to multiple intracellular routes including recycling, secretion, and degradation. Since cerebral endothelia have abundant expression of IL15 receptors, the signaling pathways activated by internalized IL15 may cause major changes of endothelial function.

To differentiate exogenous IL15 from endogenous IL15, we used ¹²⁵I-IL15 or IL15-Alexa568. The rapid plateau of IL15 endocytosis within the first 10 min is consistent with that seen in B cells in a human study. B cells take up a significant amount of IL15 within the first 5 min (Kumaki et al. 1995). By contrast, internalization of IL2 into T lymphoid cells reaches a plateau only after 60 min although the initial uptake within 5 min is also significant (Duprez and Dautry-Varsat 1986). For leptin endocytosis, ¹²⁵I-leptin uptake by HLR-15/ObRa overexpressing cells is maximum around 20–30 min (Barr et al. 1999). The different kinetics suggest differential involvement of membrane microdomains and intracellular sorting mechanisms.

Transport of small proteins across the BBB does not always involve receptors (Pan and Kastin 1999; Kastin et al. 2003). Since the mechanisms of carrier-mediated, receptor-mediated, or non-saturable permeation of a cytokine like IL15 may dictate pharmaceutical designs for drug delivery as well as understanding of the cell biology of neuroinflammation, we determined the involvement of receptors in the endocytosis of IL15. Endogenous IL15Rα showed partial co-localization with internalized IL15 at all time points studied. The relatively low level of co-localization of IL15 with endogenous IL15Rα at early time points may be related to the observation that IL15Rα recycles rapidly via recycling endosomes (Hemar et al. 1995; Dubois et al. 2002). It is also possible that IL15 endocytosis does not require IL15Rα. In cells overexpressing IL15Rα, IL15-Alexa568 was present at the cell surface, cytoplasm, and nucleus. The significant increase in perinuclear co-localization 30 min after IL15 treatment and translocation of IL15 and IL15Rα into the nucleus have been reported in non-BBB cells (Pereno et al. 2000). This is consistent with a unique function of IL15 in transcriptional regulation.

The presence of the IL2Rβ/IL2Rγ complex enables the formation of a high-affinity IL15-heterotrimeric receptor complex. In B lymphocytes, effective IL15 internalization occurs when IL2Rγ is present (Kumaki et al. 1995). We chose to study IL2Rγ rather than IL2Rβ, as the latter shows constitutive and efficient endocytosis independent of the ligand, at least in Hela and K562 human erythroleukemia cells (Subtil and Dautry-Varsat 1998). Similar to IL15Rα, IL2Rγ showed partial co-localization with IL15 in PBMEC. However, IL15-Alexa568 dissociated from IL2Rγ earlier than from IL15Rα in cerebral endothelia; this appears to be opposite from that observed in other cell types.

Fluorescent imaging in PBMEC from IL15Rα KO and IL2Rγ KO mice and studies with endocytosis inhibitors both support a role of receptors in IL15 endocytosis. The fluorescent intensity was lower in PBMEC from the KO mice, and the kinetics of intracellular accumulation appear to be slower. In T cells, both IL2Rβ and IL2Rγ localize in lipid rafts and inhibition of clathrin-dependent endocytosis does not affect receptor uptake (Subtil et al. 1994; Lamaze et al. 2001). Similar to T cells, we found that IL15Rα and IL2Rγ may initially co-localize with IL15 in lipid rafts in brain microvessel endothelial cells. Endocytosis of IL15, then, may involve at least one of these receptors, but the co-localization of IL15-Alexa568 with either of these receptor subunits was partial. Thus, although IL15Rα appears to play a role in recruiting IL15 to the cell surface and directing its endocytosis, IL15Rα alone is not sufficient to induce robust endocytosis of IL15. To make the situation more complicated, BBB endothelial cells contain multiple splice variants of IL15Rα that can respond to IL15 with JAK/STAT signaling (Wu et al. 2010d). It is possible that some remaining isoforms play different roles in IL15 trafficking and signaling.

¹²⁵I-IL15 permeation across the BBB did not show a significant reduction in the IL15Rα KO mice. A possible explanation is compensatory changes in the embryonic KO mice, as cytokines and their receptors can be redundant and pleiotropic. The lack of significant inhibition of IL15 permeation across the BBB in IL15Rα KO or IL2Rγ KO mice is reminiscent of that seen in tumor necrosis factor α (TNF) receptor KO mice. With deletion of only one receptor subunit (p55 or p75), TNF still shows basal transport across the BBB although transport across the blood-spinal cord barrier is reduced. In the TNF receptor double receptor KO mice where both receptor subtypes are absent, TNF transport is completely abolished (Pan and Kastin 2002). Overall, the lack of significant reduction of IL15 permeation in single receptor KO mice does not preclude the possibility that IL15 endocytosis can be mediated by its receptors. IL2Rγ KO mice showed a non-significant reduction of ¹²⁵I-IL15 permeation in the whole brain and cortical parenchyma, and isolated PBMEC from these mice showed less accumulation of IL15-Alexa568 than the wildtype controls. The subcellular distributions of IL15Rα and IL2Rγ, however, were not identical (Pearson’s correlation coefficient was 0.3±0.1). The results suggest that cooperation of receptors might be necessary to assure efficient trafficking of IL15 within cerebral endothelial cells.

The partial co-localization of IL15 and its receptors in different cellular compartments suggests that IL15 may be present in both active and inactive forms within the BBB endothelia. The complex intracellular IL15 network indicates that the BBB not only allows a low level of IL15 permeation from blood to the CNS, but also generates secondary signaling molecules to influence cellular function. Overall, the findings support the sequential and cooperative involvement of receptor subunits in IL15 trafficking as proposed in figure 1. The dynamic interactions between IL15 and its receptor isoforms will not only determine the fate of IL15 within the cell, but also modulate BBB endothelial functions that may be augmented during neuroinflammation.