Constitutive Endocytosis of the Chemokine CX3CL1 Prevents Its Degradation by Cell Surface Metalloproteases*

CX3CL1, a chemokine with transmembrane and soluble species, plays a key role in inflammation by acting as both chemoattractant and adhesion molecule. CX3CL1 is the only chemokine known to undergo constitutive internalization, raising the possibility that dynamic equilibrium between the endocytic compartment and the plasma membrane critically regulates the availability and processing of CX3CL1 at the cell surface. We therefore investigated how transmembrane CX3CL1 is internalized. Inhibition of dynamin using a nonfunctional allele or of clathrin using specific small interfering RNA prevented endocytosis of the chemokine in CX3CL1-expressing human ECV-304 cells. Perusal of the cytoplasmic domain of CX3CL1 revealed two putative adaptor protein-2 (AP-2)-binding motifs. Accordingly, CX3CL1 co-localized with AP-2 at the plasma membrane. We generated a mutant allele of CX3CL1 lacking the cytoplasmic tail. Deletion of the cytosolic tail precluded internalization of the chemokine. We used site-directed mutagenesis to disrupt AP-2-binding motifs, singly or in combination, which resulted in diminished internalization of CX3CL1. Although CX3CL1 was present in both superficial and endomembrane compartments, ADAM10 (a disintegrin and metalloprotease 10) and tumor necrosis factor-converting enzyme, the two metalloproteases that cleave CX3CL1, localized predominantly to the plasmalemma. Inhibition of endocytosis using the dynamin inhibitor, Dynasore, promoted rapid metalloprotease-dependent shedding of CX3CL1 from the cell surface into the surrounding medium. These findings indicate that the cytoplasmic tail of CX3CL1 facilitates its constitutive clathrin-mediated endocytosis. Such regulation enables intracellular storage of a sizable pool of presynthesized CX3CL1 that protects the chemokine from degradation by metalloproteases at the plasma membrane.

Inflammation is marked by the migration of circulating leukocytes into sites of injury, a process that occurs via a series of coordinated interactions between leukocytes and endothelial or epithelial cells. Central to this process are chemokines, a family of low molecular weight proteins that can attract leukocytes bearing the complementary receptors. When engagement of the chemokine receptor occurs, the leukocyte becomes activated and is induced to firmly adhere to the inflamed endothelium. These initial steps culminate in diapedesis of the leuko-cyte across the endothelium and migration into the injured tissue. The local complement of chemokines elaborated is organ-specific and varies with the type of inflammation present. In addition, specific leukocyte subsets also bear distinct chemokine receptors. In this way, chemokines and chemokine receptors confer organ specificity to leukocyte migration and help to "fine-tune" the nature of the observed inflammatory response.
Among the 40 chemokines identified so far, CX 3 CL1 is one of only two that have a transmembrane structure (1,2). The chemokine domain of CX 3 CL1 binds to its complementary receptor, CX 3 CR1, through two distinct amino acid residues (3). The mucin stalk of CX 3 CL1 allows efficient presentation of the chemokine to circulating leukocytes that express CX 3 CR1, thereby allowing these leukocytes to be captured by the underlying endothelium (4,5). CX 3 CL1 also possesses a cytoplasmic tail 37 amino acids in length. However, the specific functions of the cytoplasmic tail have been left completely unexplored.
Accumulating evidence demonstrates a critical role for CX 3 CL1 in the pathogenesis of diverse inflammatory diseases, including atherosclerosis, systemic lupus erythematosus, and rejection of transplanted organs (6 -15). Cell surface expression of CX 3 CL1 is known to be regulated by proteolytic cleavage, or shedding, from the plasma membrane (16 -18). Constitutive cleavage of CX 3 CL1 occurs at low levels and is mediated by ADAM10 (a disintegrin and metalloprotease 10) (17). In response to inflammatory stimulation with lipopolysaccharide or to protein kinase C activation using phorbol 12-myristate 13-acetate, proteolytic cleavage of CX 3 CL1 is markedly enhanced. Inducible cleavage of CX 3 CL1 is mediated by tumor necrosis factor-␣ converting enzyme (TACE; ADAM17), 2 a related protease of the metzincin family (16,18).
In addition to proteolytic cleavage, surface expression of CX 3 CL1 is also regulated by subcellular trafficking. We recently demonstrated that cell surface CX 3 CL1 rapidly recycles to and from a specialized endocytic compartment, raising the possibility that the intracellular pool serves as a storage depot and that dynamic equilibrium between the endocytic compartment and the plasma membrane determines the availability and processing of transmembrane CX 3 CL1 (19). In the current study, we explored whether the unique cytoplasmic tail of CX 3 CL1 is important for this novel mode of regulation of the chemokine and whether it affects susceptibility of the chemokine to surface proteases. Our data suggest that plasmalemmal CX 3 CL1 undergoes constitutive clathrin-mediated endocytosis (CME), facilitating storage of an intracellular pool of chemokine that is protected from cell surface metalloproteases.
Small interfering RNA (siRNA) directed against the heavy chain of clathrin (chc-2) has been described previously and was synthesized as Option C siRNA by Dharmacon RNA Technologies (Lafayette, CO) (24). Control nontargeting siRNA was also from Dharmacon. The cells were electroporated with siRNA by nucleofection (Amaxa, Gaithersburg, MD) on Days 0 and 2. After the second electroporation, the cells were plated on glass coverslips or in tissue culture wells and were analyzed on Day 4 (25).
Immunofluorescence Staining-ECV-CX 3 CL1 cells were grown on glass coverslips, fixed using 4% paraformaldehyde, washed, permeabilized using 0.1% Triton, and incubated with blocking 5% donkey serum at room temperature for 1 h. The cells were incubated with anti-CX 3 CL1 Ab (2.5 g/ml), anti-HA Ab (10 g/ml), anti-AP-2 Ab (5 g/ml), anti-clathrin Ab (3 g/ml), anti-ADAM10 Ab (10 g/ml), and/or anti-TACE Ab (2.7 g/ml) at room temperature for 1 h. To label cell surface rather than total CX 3 CL1, the cells were incubated with the corresponding primary Ab prior to permeabilization. After washing, the cells were incubated with the appropriate Cy2-, Cy3-, or Cy5-conjugated secondary Ab. In some experiments, the cells were transfected with R-pre cDNA construct, to clearly label the plasma membrane, 48 h prior to the performance of experiments.
Immunofluorescent labeling of cells was examined using a Leica DMIRE2 microscope and OpenLab software (Improvision Inc., Lexington, MA). Co-localization was determined using OpenLab 4.0.2 co-localization software. The degree of co-localization was measured using the Pearson's correlation coefficient or the co-localization coefficient, as previously described (19). Cell surface immunofluorescence intensity was measured using MetaMorph imaging software (Universal Imaging Corporation, Westchester, PA) (19). In other experiments, the cells were examined using a spinning disc DMIRE2 confocal microscope with a Hamamatsu backthinned EM-CCD camera. Images were acquired using the appropriate excitation and emission filters and a 100ϫ oil immersion objective (26). Optical sections of 0.2-0.5 m were obtained, and deconvolution, when appropriate, was performed using Volocity software (Improvision). Co-localization was examined using Volocity co-localization software (20).
Immunoblotting-SDS-PAGE and immunoblotting were performed using anti-clathrin Ab (0.3 g/ml) as previously described (19,20,25). Anti-actin Ab was used to control for protein loading. Immunoreactive bands were visualized by enhanced chemiluminescence (Amersham Biosciences) recorded on x-ray film.
Endocytosis Assays-Endocytosis assays were performed as previously described (19). Briefly, ECV-CX 3 CL1, ECV-CX 3 CL1-360, ECV-CX 3 CL1-Y362A, ECV-CX 3 CL1-Y392A, or ECV-CX 3 CL1-Y362A-Y392A cells were grown on glass coverslips and incubated with anti-CX 3 CL1 Ab (2.5 g/ml) and/or rhodamine-conjugated transferrin (30 g/ml) at 37 or 4°C for 1-2 h (19). The cells were washed, fixed with 4% paraformaldehyde, and permeabilized with 0.1% Triton. The cells were incubated with Cy3-conjugated anti-goat IgG or Cy5-conjugated anti-goat IgG, washed, and mounted on glass slides using DAKO fluorescent mounting medium (DAKO Corporation, Carpinteria, CA). In some experiments, the cells were transfected with dominant negative dynamin-1 (K44A) cDNA and EGFP at a ratio of 10:1 prior to performing endocytosis assays. In other experiments, the cells were preincubated with Dynasore (160 M) for 30 min at 37°C or were transfected with siRNA directed against clathrin prior to performing endocytosis assays (21,25). CX 3 CL1 Shedding Assays-ECV-CX 3 CL1 cells were plated at a density of 1 ϫ 10 5 cells/well in a 6-well tissue culture plate and grown to confluence for 48 h. The cells were incubated with Dynasore (160 M) for 30 min prior to collecting conditioned medium (21). In some experiments, the cells were incubated with the metalloprotease inhibitor, GM6001 (20 M), for 4 h prior to incubation with Dynasore (16,20). The cells were washed once with phosphate-buffered saline and lysed by adding 0.5 ml of radioimmune precipitation assay buffer. Protease inhibitor mixture (Sigma-Aldrich) was added to both harvested medium and cell lysis buffer to prevent protein degradation. Supernatants and cell lysates were cleared by centrifugation at 13,000 rpm at 4°C for 10 min. The amount of soluble CX 3 CL1 in conditioned medium was detected using a CX 3 CL1 ELISA kit (R & D Systems) according to the manufacturer's specifications, using fresh samples (18). The data are the averages Ϯ S.E. of three independent experiments and expressed as the amount of soluble CX 3 CL1 released into the medium relative to the total amount of cell-associated CX 3 CL1 in the corresponding well. In other experiments, ELISA was performed using conditioned medium harvested from cells expressing CX 3 CL1, CX 3 CL1-Y362A, CX 3 CL1-Y392A, and CX 3 CL1-Y362A-Y392A.
Statistical Analyses-The data were analyzed by comparing the mean values using the Student's t test. A value of p Ͻ 0.05 was considered significant.

Cell Surface CX 3 CL1 Undergoes Temperature-sensitive
Internalization-We first examined the subcellular distribution of full-length CX 3 CL1. In accordance with our previous observations, full-length CX 3 CL1 was expressed on the plasma membrane and within an intracellular compartment ( Fig. 1A) (19). When cells were incubated at 37°C with Ab directed to the exofacial domain, CX 3 CL1 accumulated within a juxtanuclear endosomal compartment in 81 of 127 cells counted (64%; Fig. 1, B and D) (19). Labeling of the intracellular pool was inhibited when the experiment was performed at 4°C (Fig. 1, C and D). In this instance, mainly the superficial CX 3 CL1 was labeled, and intracellular CX 3 CL1 was detected in only 26 of 163 cells (16%; Fig. 1D). These data are consistent with the notion that endocytosis, a temperature-sensitive process, is needed to deliver the Ab to the intracellular pool.
Endocytosis of Plasmalemmal CX 3 CL1 Is Dynamindependent-To determine whether internalization of CX 3 CL1 requires dynamin, ECV-CX 3 CL1 cells were transfected with cDNA expression plasmid encoding a dominant negative allele of dynamin-1 (K44A) (22,(27)(28)(29). The cells were co-transfected with EGFP to readily allow identification of transfected cells (Fig. 2, A and C). The effects of K44A were verified by incubating cells with rhodamine-conjugated transferrin at 37°C (24, 30 -32). As expected, dynamin inhibition prevented rapid internalization of transferrin receptor (Fig. 2, A and B). FIGURE 1. CX 3 CL1 undergoes temperature-sensitive internalization from the cell surface. A, ECV-CX 3 CL1 cells were fixed, permeabilized, and incubated with anti-CX 3 CL1 Ab (2.5 g/ml), followed by Cy3-conjugated anti-goat-IgG. The cells were examined using a Leica DMIRE2 spinning disc confocal microscope with a Hamamatsu backthinned EM-CCD camera. The images were acquired using the appropriate excitation and emission filters and a 100ϫ oil immersion objective. B, ECV-CX 3 CL1 cells were incubated with anti-CX 3 CL1 Ab (2.5 g/ml) at 37°C for 1 h, then fixed, permeabilized, and incubated with Cy3-conjugated anti-goat IgG. The cells were examined using a Leica DMIRE2 spinning disc confocal microscope at 100ϫ magnification. Optical sections of 0.2-0.5 m were obtained. The lower panel depicts x versus z optical section. C, cells were incubated with anti-CX 3 CL1 Ab (2.5 g/ml) at 4°C for 1 h, then fixed, permeabilized, and incubated with Cy3-conjugated anti-goat IgG. The cells were examined using a Leica DMIRE2 spinning disc confocal microscope at 100ϫ magnification. We next incubated cells with an Ab directed to the exofacial domain of CX 3 CL1 at 37°C. CX 3 CL1 accumulated within a juxtanuclear endosomal compartment in cells expressing native dynamin-1 (Fig. 2D) (19). However, the presence of nonfunctional dynamin-1 (Fig. 2C) prevented rapid labeling of the juxtanuclear endocytic pool (Fig. 2, C and D). In this instance, predominantly the superficial CX 3 CL1 was labeled (Fig. 2D), implying that dynamin is required to deliver the Ab to the intracellular pool.
In quantifying the effect of dynamin inhibition on internalization of CX 3 CL1 from the cell surface, we recognized that delineating the plasma membrane using traditional markers such as cholera toxin or wheat germ agglutinin is not entirely accurate, because these markers themselves can undergo endocytosis. We, therefore, co-transfected cells with the recently described K-Ras R-pre cDNA construct (R-pre), whose product, by virtue of its surface charge, localizes exclusively to the plasma membrane (Fig. 2E) (23). The cells were superficially labeled with anti-CX 3 CL1 Ab at 37°C, and the degree of CX 3 CL1 internalization was measured by determining the ratio of CX 3 CL1 that co-localized with R-pre (i.e. cell surface CX 3 CL1) to the total amount of CX 3 CL1 present within the cell (Fig. 2, E-H). In cells expressing native dynamin-1, 53 Ϯ 7% of anti-CX 3 CL1 Ab was detected at the plasma membrane, and the remaining 47% was internalized (Fig. 2H). In cells expressing nonfunctional dynamin-1, endocytosis of cell surface CX 3 CL1 was markedly impaired, with 88 Ϯ 7% of anti-CX 3 CL1 Ab seen at the plasma membrane, and only 12% within the cell ( Fig. 2H; p Ͻ 0.0001 versus control). These data indicate that endocytosis of cell surface CX 3 CL1 is a dynamin-dependent process.
The involvement of clathrin in internalization of CX 3 CL1 was functionally tested using siRNA specifically targeting clathrin (24,25). Clathrin siRNA effectively decreased clathrin expression (Fig. 3, E and G), whereas nontargeting control siRNA did not (Fig. 3, E and H). As predicted, the cells in which The cells were incubated with anti-CX 3 CL1 Ab at 37°C for 1 h, then fixed, permeabilized, and incubated with an Alexa488-conjugated secondary Ab. To identify cells expressing the K44A mutant allele, the cells were incubated with anti-HA Ab followed by a Cy5-conjugated secondary Ab. Using Volocity TM co-localization software, the percentage of total CX 3 CL1 signal that co-localized with plasma membrane R-pre-RFP was calculated for at least 20 cells from three separate experiments. *, p Ͻ 0.0001. clathrin was knocked down (Fig.  3G) did not internalize transferrin receptor from the cell surface (Fig.  3J). In contrast, both untreated cells and cells transfected with nontargeting siRNA efficiently internalized transferrin receptor (Fig. 3, compare F and I with H and K, respectively). When CME was disrupted using clathrin siRNA, anti-CX 3 CL1 Ab labeled only superficial CX 3 CL1 but was not delivered to the juxtanuclear endocytic compartment (Fig. 3M). In untreated cells and in cells transfected with control nontargeting siRNA, plasma membrane CX 3 CL1 was internalized (Fig. 3, L and N). In the presence of nontargeting siRNA, intracellular accumulation of CX 3 CL1 was detected in 58 of 79 cells examined (74%). When clathrin expression was knocked down, intracellular CX 3 CL1 was only seen in 21 of 132 cells examined (15%). Taken together, these data demonstrate that endocytosis of CX 3 CL1 from the cell surface is clathrin-dependent.

AP-2-binding Motifs within the Cytoplasmic Tail of CX 3 CL1 Enable
Internalization of the Chemokine-Various adaptor complexes allow integral membrane proteins to interact with clathrin, promoting internalization of the protein cargo from the cell surface (37,38). Perusal of the cytoplasmic tail of CX 3 CL1 reveals the presence of two AP-2-binding motifs: YQSL at positions 362-365 and YVLV at positions 392-395 (39 -42). To determine whether CX 3 CL1 interacts with AP-2, ECV-CX 3 CL1 cells were labeled with anti-AP-2 Ab (Fig. 4A) and anti-CX 3 CL1 Ab (Fig. 4B). As demonstrated in Fig. 4 (C and D), there was overlap of CX 3 CL1 and AP-2. As a next approach, we generated a mutant allele of CX 3 CL1 lacking the cytoplasmic tail (CX 3 CL1-360). When the intracellular domain was deleted, the chemokine was expressed exclusively on the cell surface (Fig. 4E). The cells were incubated with Ab directed to the exofacial domain of CX 3 CL1, and internalization of the chemokine FIGURE 3. Plasmalemmal CX 3 CL1 undergoes CME. A-D, ECV-CX 3 CL1 cells were grown on coverslips, fixed, permeabilized, and incubated with anti-clathrin mAb (A; 3 g/ml) together with anti-CX 3 CL1 mAb (B; 2.5 g/ml). The cells were washed and incubated with Cy2-conjugated anti-mouse IgG to visualize clathrin, and Cy3-conjugated anti-goat IgG to visualize CX 3 CL1. The cells were examined using a spinning disc Leica DMIRE2 confocal microscope. The images were acquired using the appropriate excitation and emission filters and a 100ϫ oil immersion objective. The ruler bar is 10 m. The distribution of clathrin (A) was compared with that of CX 3 CL1 (B). C, merged image of A and B. D, magnified view of the area indicated in C. E, ECV-CX 3 CL1 cells were electroporated with siRNA targeting clathrin or with control, nontargeting siRNA on Days 0 and 2. On Day 4, the cell lysates were harvested, and immunoblotting was performed using anti-clathrin Ab (0.3 g/ml) and horseradish peroxidase-conjugated secondary Ab. To control for protein loading, blots were stripped and reprobed with anti-actin Ab and horseradish peroxidase-conjugated secondary Ab. A representative blot is depicted.   A and B, ECV-CX 3 CL1 cells were grown on coverslips, fixed, permeabilized, and incubated with anti-AP-2 mAb (5 g/ml; A) together with anti-CX 3 CL1 mAb (2.5 g/ml; B). The cells were washed and incubated with Cy3-conjugated anti-goat IgG and Cy2-conjugated anti-mouse IgG secondary Ab to detect CX 3 CL1 and AP-2, respectively. The distribution of AP-2 (A) was compared with that of CX 3 CL1 (B) using a spinning disc DMIRE2 confocal microscope with a Hamamatsu backthinned EM-CCD camera (100ϫ). The ruler bar is 10 m. C, merged image of A and B. D, magnified view of the area indicated in C. E, ECV-304 cells stably expressing mutant CX 3 CL1 lacking the intracellular domain (ECV-CX 3 CL1-360) were fixed, permeabilized, and incubated with anti-CX 3 CL1 Ab (2.5 g/ml), followed by Cy3-conjugated anti-goat-IgG. The cells were examined using a Leica DMIRE2 deconvolution microscope at 100ϫ magnification. For comparison, the inset shows ECV-304 cell expressing full-length CX 3 CL1 (ECV-CX 3 CL1) labeled with anti-CX 3 CL1 Ab followed by Cy3-conjugated anti-goat-IgG. F, ECV-CX 3 CL1-360 cells were incubated with anti-CX 3 CL1 Ab (2.5 g/ml) at 37°C for 1 h, then fixed, permeabilized, and incubated with Cy3-conjugated secondary Ab. Cells were examined as described in E. G, Ab uptake experiments were performed in cells expressing CX 3 CL1-360 or full-length CX 3 CL1 as described in F. The cells were examined using a spinning disc Leica DMIRE2 inverted microscope with a Hamamatsu ORCAER charge-coupled device camera (100ϫ) and Z-stacks obtained. Upper panel, x versus z image from cell expressing CX 3 CL1-360. Lower panel, x versus z image from cell expressing full-length CX 3 CL1. H, site-directed mutagenesis was used to generate cDNA constructs in which the critical tyrosine residue of each AP-2-binding motif was replaced by an alanine residue, separately or in combination (Y362A, Y392A, and Y362A-Y392A). H-K, ECV-304 cells were co-transfected with cDNA encoding RFP-tagged R-pre, a plasma membrane marker, together with full-length CX 3 CL1, CX 3 CL1-Y362A, CX 3 CL1-Y392A, or CX 3 CL1-Y362A-Y392A. I-L, cells were incubated with anti-CX 3 CL1 Ab (2.5 g/ml) at 37°C for 1 h, then fixed, permeabilized, and incubated with Cy5-conjugated secondary Ab to visualize uptake of the chemokine. The cells were examined at 100ϫ magnification using a Leica DMIRE2 spinning disc confocal microscope with a backthinned EM-CCD camera. The images were acquired using the appropriate excitation and emission filters. Optical sections of 0.2-0.5 m were obtained. I, cell expressing full-length CX 3 CL1. J, cell expressing CX 3 CL1-Y362A. K, cell expressing CX 3 CL1-Y392A. L, cell expressing CX 3 CL1-Y362A-Y392A. M, experiments were performed as in I-L. As described for Fig. 2 (E-H), using Volocity TM co-localization software, the proportion of total CX 3 CL1 that co-localized with the R-pre membrane marker was determined. The values represent the means Ϯ S.E. for 25-30 cells/condition from three separate experiments. *, p Ͻ 0.01 versus full-length CX 3 CL1.

JOURNAL OF BIOLOGICAL CHEMISTRY 29649
was assessed (Fig. 4, F and G). In the absence of the cytosolic domain, CX 3 CL1 failed to internalize (Fig. 4, F and G), indicating that determinants within the intracellular tail direct endocytosis of the chemokine.
To refine the analysis, we used site-directed mutagenesis to mutate the tyrosine residues of AP-2-binding motifs separately and in combination while preserving the overall length of the intracellular region of CX 3 CL1 (Fig. 4H). Mutant constructs were expressed, and cells were incubated with anti-CX 3 CL1 Ab directed to the extracellular domain. In cells expressing fulllength CX 3 CL1, 63.0 Ϯ 2.0% of Ab-labeled CX 3 CL1 was detected at the cell surface, and the remainder was internalized (Fig. 4, I and M). When the two AP-2-binding motifs were disrupted, either separately or in combination, endocytosis of CX 3 CL1 was inhibited, yielding a significantly higher fraction of chemokine at the plasma membrane (Fig. 4, I-M; 83.0 Ϯ 1.5% for CX 3 CL1-Y362A; 84.0 Ϯ 1.3% for ECV-CX 3 CL1-Y392A; 84.0% Ϯ 1.7% for ECV-CX 3 CL1-Y362A-Y392A; for each, p Ͻ 0.01 versus full-length CX 3 CL1). These data demonstrate that both the proximal and distal intracellular AP-2-binding regions of CX 3 CL1 facilitate CME of the chemokine but that the effects are not additive.

Endocytosis of CX 3 CL1 Allows Storage of a Pool of Presynthesized Chemokine Protected from Degradation by Membraneassociated
Metalloproteases-Our data demonstrate that CX 3 CL1 undergoes clathrin-dependent endocytosis. The key question is what purpose endocytosis of CX 3 CL1 might serve. During an inflammatory response, CX 3 CL1 expression is upregulated at the level of transcription and translation, a process that requires several hours (2,5,(43)(44)(45). Depending on the nature of the inflammatory perturbation, this protracted response may be too slow. Rapid routing of CX 3 CL1 to the cell surface would allow expeditious recruitment of inflammatory cells. We postulated that internalization of CX 3 CL1 would allow prior synthesis and intracellular storage of the chemokine, with rapid mobilization to the plasma membrane on demand. We further reasoned that endocytosis would permit internal storage of a sizeable pool of CX 3 CL1 that is protected from degradation by metalloproteases expressed on the surface of the cell (16 -18). To test this notion, we compared subcellular distribution of CX 3 CL1 with that of TACE and ADAM10, the two metalloproteases that cleave CX 3 CL1 (16 -18). TACE and ADAM10 were detected predominantly at the plasma membrane (Fig. 5, B and C). CX 3 CL1 was detected at the plasma membrane, as well as intracellularly where it is not accessible to the metalloproteases (Figs. 1A and 5A). We reasoned that if CME indeed prevents proteolysis of CX 3 CL1 by membraneanchored metalloproteases, acute disruption of CME should augment shedding of the chemokine from the cell surface. Accordingly, treatment with the dynamin inhibitor, Dynasore, enhanced the release of soluble CX 3 CL1 into the medium ( Fig.  5D; p Ͻ 0.001). When cells were preincubated with the metalloprotease inhibitor, GM6001, which blocks both TACE and ADAM10, release of soluble CX 3 CL1 was prevented ( Fig. 5D; p Ͻ 0.05) (16,18). When proteolytic cleavage was inhibited, Dynasore treatment caused a 2-fold increase in the CX 3 CL1 fraction at the cell surface ( Fig. 5E; p Ͻ 0.01).
To rule out any potential effects of Dynasore to enhance CX 3 CL1 cleavage by increasing metalloprotease levels at the cell surface, we examined whether mutations in the AP-2-binding motifs of CX 3 CL1 that inhibit endocytosis of the chemokine (Fig. 4, H-M) also increase shedding of CX 3 CL1. Release of soluble CX 3 CL1 from cells expressing each of the three mutant alleles was 5-fold greater than CX 3 CL1 release from cells expressing the native form of the chemokine (Fig. 5F; p Ͻ 0.0005 for CX 3 CL1-Y362A versus CX 3 CL1; p Ͻ 0.0005 for CX 3 CL1-Y392A versus CX 3 CL1; and p Ͻ 0.005 for CX 3 CL1-Y362A-Y392A versus CX 3 CL1). Collectively, these findings indicate that constitutive endocytosis of CX 3 CL1 permits the storage of an intracellular pool of the chemokine that is not subject to degradation by metalloproteases present at the cell surface.

DISCUSSION
The principal aim of this study was to investigate the subcellular traffic and processing of CX 3 CL1 and the role played by its unique intracellular domain in regulating these processes. For these studies we used CX 3 CL1-transfected ECV-304 cells. Although known to be derived from bladder epithelial carcinoma cells, ECV-304 cells have many endothelial features, including the propensity to grow in monolayers of flattened cells, the formation of moderately tight intercellular junctions, and expression of the endothelial markers, Flt-1, and von Willebrand factor (46 -49). Since the identification of CX 3 CL1, these cells have been used extensively to study both function and cleavage of the chemokine because the surface levels of CX 3 CL1 mirror those seen in cytokine-activated primary vascular endothelial cells (2, 5, 17-19, 50, 51).
Among the 40 chemokines identified to date, CX 3 CL1 is one of only two that have a transmembrane structure (1,2). The functions of the extracellular domain of CX 3 CL1 have been well studied. The chemokine domain, located within the N terminus of the extracellular region, contains a cluster of basic amino acid residues and one aromatic residue that are critical for binding of the chemokine to its receptor and for downstream receptor signaling events (3,52,53). Several groups have investigated the role of the extracellular mucin stalk, upon which the chemokine domain is anchored. Through the generation of chimeric mutants in which the chemokine domain of CX 3 CL1 is fused to the rod-like stalk of E-selectin or in which the mucin stalk of CX 3 CL1 is fused to other soluble chemokines, it has become clear that the mucin domain of CX 3 CL1 does not possess properties unique to this particular chemokine but rather serves to extend the chemokine away from the cell surface to efficiently "present" it to circulating leukocytes (4,52). Very recently, the role of the transmembrane domain of CX 3 CL1 has also been described. Using advanced energy transfer and microscopic techniques, the authors demonstrated that to promote efficient adhesion, CX 3 CL1 must cluster within the plasma membrane and that this molecular aggregation is critically dependent upon the transmembrane domain of the protein (54).
The role of the intracellular domain of CX 3 CL1 has been heretofore unexplored. Our experiments show that the cytoplasmic tail is required for efficient internalization of CX 3 CL1 from the cell surface. In keeping with these observations, we and others have previously described that in steady state, roughly half the total cellular pool of CX 3 CL1 is found intracellularly, and the remaining fraction is on the plasma membrane (19,54).
However, ours is the first report that a chemokine itself may undergo cell surface regulation in a similar fashion. Furthermore, we found that CME of CX 3 CL1, unlike endocytosis of chemokine receptors, occurred constitutively and did not require prior association with its binding partner. This represents a novel means of regulating expression of a chemokine.
The overarching question is what purpose constitutive internalization of CX 3 CL1 might serve. We recognized that although experiments involving transfection of clathrin siRNA or dominant negative dynamin yield useful information regarding the mechanism of internalization of CX 3 CL1, such experiments are not ideal for determining the functional relevance of FIGURE 5. Constitutive CME of CX 3 CL1 protects the chemokine from proteolytic shedding by membrane-anchored metalloproteases. A, ECV-CX 3 CL1 cells were fixed, permeabilized, and labeled with anti-CX 3 CL1 Ab (2.5 g/ml), followed by Cy3-conjugated anti-goat IgG. The cells were examined using a spinning disc Leica DMIRE2 confocal microscope at 100ϫ magnification. Optical sections of 0.2-0.5 m were obtained. Lower panel, x versus z image. B, ECV-CX 3 CL1 cells were fixed, permeabilized, and labeled with anti-TACE Ab (2.7 g/ml), followed by Cy3-conjugated anti-mouse IgG. The cells were visualized as in A. C, ECV-CX 3 CL1 cells were fixed, permeabilized, and labeled with anti-ADAM10 Ab (10 g/ml), followed by Cy3-conjugated anti-mouse IgG. The cells were visualized as in A. D, cells were incubated with Dynasore (160 M) for 30 min at 37°C, conditioned medium was collected, and cell lysates were harvested. In some experiments, the cells were pretreated with the metalloprotease inhibitor, GM6001 (20 M for 4 h), prior to incubation with Dynasore. Protease inhibitor mixture was added to conditioned medium and to cell lysis buffer to prevent protein degradation. Supernatants and cell lysates were cleared by centrifugation. Using fresh samples, CX 3 CL1 was measured using a CX 3 CL1 ELISA Kit (R & D Systems), according to the manufacturer's specifications. Three independent experiments were performed, and the data were expressed as the means Ϯ S.E. of the normalized soluble CX 3 CL1 released into the conditioned medium in relation to the total amount of cell-associated CX 3 CL1 in the corresponding well. **, p Ͻ 0.001; *, p Ͻ 0.05. E, cells were incubated with metalloprotease inhibitor, GM6001 (20 M) for 4 h, followed by Dynasore for 30 min, and surface and total CX 3 CL1 expression was determined as in Fig. 2 E-H. As described for Fig. 2 (E-H), the fraction of total CX 3 CL1 that was present at the plasma membrane was determined for 10 cells from three separate experiments. The results are expressed as the mean values Ϯ S.E. *, p Ͻ 0.01. F, ELISA experiments were performed as described in D, using conditioned medium harvested from cells expressing the native form of CX 3 CL1, CX 3 CL1-Y362A, CX 3 CL1-Y392A, or CX 3 CL1-Y362A-Y392A. Three independent experiments were performed, and the data were expressed as the mean Ϯ S.E. of the soluble CX 3 CL1 released into the conditioned medium in relation to the total amount of cell-associated CX 3 CL1 in the corresponding well. *, p Ͻ 0.0005 versus CX 3 CL1; **, p Ͻ 0.005 versus CX 3 CL1.

Endocytosis of CX 3 CL1
OCTOBER 23, 2009 • VOLUME 284 • NUMBER 43 endocytosis of the chemokine. Because these experiments are performed over a span of 48 -96 h, changes in the level of biosynthesis of CX 3 CL1 and other compensatory proteins and pathways could certainly occur. We reasoned that acute inhibition of CME of CX 3 CL1 would yield more meaningful information about the significance of this process in living cells. We therefore used Dynasore, a cell-permeable small molecule inhibitor of dynamin GTPase activity that acutely blocks CME (21). In the presence of Dynasore, CX 3 CL1 underwent extensive cleavage at the plasma membrane. This cleavage was metalloprotease-dependent and resulted in the release of soluble chemokine into the surrounding medium. In accordance with these observations, we found that ADAM10 and TACE, the two metalloproteases known to cleave CX 3 CL1, were predominantly located at the plasma membrane but did not associate with endosomal CX 3 CL1. Our results are in keeping with those of others who have reported similar cell surface distribution of ADAM10 and TACE (62)(63)(64). Our findings would indicate that the cytoplasmic tail of CX 3 CL1 plays a key role in regulating subcellular traffic of the chemokine through constitutive CME. Regulation in such a manner would permit intracellular storage of a sizable pool of presynthesized CX 3 CL1, which could be recycled back to the cell surface as needed, while at the same time protecting the chemokine from proteolytic degradation by metalloproteases at the plasma membrane.
The relationship between CME and proteolytic shedding of transmembrane proteins by the ␣-secretases, ADAM10 and TACE, has recently been recognized, although this has never before been described for a chemokine (63,65,66). The link between CME and ␥-secretase cleavage is more firmly established. After undergoing ectodomain shedding, many ␣-secretase substrates, including Notch, amyloid precursor protein, and CD44, undergo subsequent intramembrane proteolysis by the ␥-secretase complex. The intracellular domain that is consequently released can translocate to the nucleus and induce transcription. Several groups have demonstrated that clathrinassociated endocytic machinery plays an integral role in ␥-secretase cleavage (67)(68)(69). Interestingly, CX 3 CL1 has recently been shown to also undergo ␥-secretase cleavage after initial ectodomain shedding by ADAM proteases (70). It remains to be determined how this process is regulated and what the functions of the liberated intracellular fragment of CX 3 CL1 might be.