SorLA in Interleukin-6 Signaling and Turnover

ABSTRACT Interleukin-6 (IL-6) is a multifunctional cytokine with important functions in various physiologic processes. Mice lacking IL-6 exhibit multiple phenotypic abnormalities, such as an inadequate immune and acute-phase response, and elevated levels of circulating IL-6 have been found to accompany several pathological conditions. IL-6 binds the nonsignaling IL-6 receptor (IL-6R), which is expressed as a transmembrane, as well as a secreted circulating protein, before it engages homodimeric gp130 for signaling. Complex formation between IL-6 and the membrane-bound IL-6 receptor gives rise to classic cis signaling, whereas complex formation between IL-6 and the soluble IL-6R results in trans signaling. Here, we report that the endocytic receptor SorLA targets IL-6 and IL-6R. We present evidence that SorLA mediates efficient cellular uptake of both IL-6 and the circulating IL-6R in astrocytes. We further show that SorLA interacts with the membrane-bound IL-6R at the cell surface and thereby downregulates IL-6 cis signaling. Finally, we find that the SorLA ectodomain, released from the cell membrane upon enzymatic cleavage of full-length SorLA, may act as an IL-6 carrier protein that stabilizes IL-6 and its capacity for trans signaling.

many in vivo functions of IL-6 can be blocked by specific inhibition of the trans signaling pathway (15).
Since they are both prerequisites for signaling, deficiency in either IL-6 or IL-6R result in very similar phenotypes characterized by an impaired immune response and a defective acute-phase response (7,16). Elevated levels of circulating IL-6 and sIL-6R, on the other hand, frequently accompany pathophysiologic conditions and constitute potential drug targets in the treatment of diseases like rheumatoid arthritis, asthma, and multiple sclerosis (3,12,(17)(18)(19).
We show here that the endocytic receptor SorLA may impact the cellular uptake as well as the signaling of IL-6. Human SorLA is one of the five type 1 receptors that constitute the human Vps10p-domain (Vps10p-D) receptor family (20,21). It is expressed in most regions of the nervous system but is also found in a number of nonneuronal tissues and cell types such as the liver, kidney, pancreas, and cells of the immune system, e.g., monocytes and macrophages (20,(22)(23)(24). Apart from an N-terminal Vps10p-D comprising a unique ligand-binding ten-bladed ␤-propeller supported by two minor domains (25,26), the luminal part of SorLA also contains a small ␤-propeller domain with an associated epidermal growth factor class B-like motif and a cluster of 11 low-density lipoprotein receptor class A repeats (21). SorLA's cytoplasmic tail is short but contains several motifs for the binding of adaptor proteins, such as AP-1 and -2, GGA1 to -3, and the retromer complex, that are involved in endocytosis, basolateral sorting, and trafficking between Golgi and endosomal compartments (27)(28)(29)(30)(31)(32).
In agreement with the above findings, SorLA displays a variety of different functions. It affects APP processing and A␤ amyloid generation (38), and mutations in SorLA constitute a risk factor for familiar and sporadic forms of Alzheimer's disease (39). It is further implicated in the cellular release of lipoprotein lipase (LPL) (31) and is linked to obesity and glucose tolerance (40,41). It also influences receptor tyrosine kinase RET signaling via interaction with GDNF and its primary receptor GFR␣1 (34). Furthermore, SorLA is subject to cleavage and cellular shedding of its luminal part (sSorLA) (42), and sSorLA in circulation is implicated in the migration of smooth muscle cells and may serve as a biomarker for atherosclerosis, coronary stenosis, and diabetic retinopathy (43)(44)(45)(46)(47). Likewise, elevated levels of sSorLA in cerebrospinal fluid are a potential marker of progressive Alzheimer's disease (48).
We have recently demonstrated that SorLA targets the heterodimeric cytokine CLC:CLF-1 (via the CLF-1 subunit) and its primary receptor, the CNTFR, and that it modulates the cellular response to CLC:CLF-1 by mediating CLF-1-dependent endocytosis and downregulation of the CNTFR (33). In the present study, we show that SorLA targets IL-6 and the IL-6R, and we investigate the functional implications of these interactions in cells. Our study reveals that full-length SorLA conveys the cellular uptake of both IL-6 and sIL-6R (individually or in complex) and also interacts with the membrane-bound IL-6R blocking its binding of IL-6. The findings further suggest that, whereas full-length SorLA may downregulate IL-6 cis signaling, sSorLA may stabilize IL-6 and its capacity for trans signaling.

RESULTS
SorLA binds IL-6 and mediates its cellular uptake. Initially, we examined the binding of IL-6 to the immobilized ectodomains of IL-6R and SorLA (sIL-6R and sSorLA) using surface plasmon resonance (SPR) analysis. As demonstrated in Fig. 1A and B, IL-6 exhibited concentration dependent high-affinity binding not only to sIL-6R (K d ϳ 10 nM) but also to sSorLA (K d ϳ 29 nM). The binding to sSorLA was completely inhibited in the presence of a surplus of SorLA's own propeptide ( Fig. 1C) but was almost unchanged by neurotensin (NT) (data not shown). IL-6 also bound to the purified SorLA Vps10p-D (K d ϳ 12 nM) (Fig. 1D) in competition with the SorLA propeptide (Fig. 1E).
We next studied the binding of IL-6 to full-length SorLA in cells. Wild-type (wt) HEK293 cells and cells stably transfected with SorLA were incubated at 125 nM IL-6 (30 min, 37°C) in medium supplemented or not supplemented with 20 M SorLA propeptide or NT. The cells were fixed, permeabilized, and stained with anti-IL-6 and anti-SorLA antibodies. As shown (Fig. 1F), staining was undetectable in wt cells, whereas cells transfected with SorLA displayed a significant intracellular staining for both proteins and an almost complete colocalization of IL-6 with SorLA (Pearson's r ϭ 0.74 Ϯ 0.04 [mean Ϯ standard error of the mean {SEM}, n ϭ 9]). The uptake was unchanged at a surplus of 20 M NT (not shown) but was almost abolished in the presence of a surplus of SorLA propeptide. In cells treated with lysosomal inhibitors (leupeptin and pepstatin A), internalized IL-6 accumulated and to some extent colocalized with LAMP-1, strongly suggesting that SorLA targeted IL-6 for degradation in the lysosomes (Fig. 1G).
To further substantiate these findings, we examined the uptake of IL-6 in astrocytes, which exhibit endogenous expression of SorLA, as well as sortilin and IL-6R mRNA, but little or no detectable IL-6R protein (49). Cultured astrocytes isolated from wt mice were incubated (30 min, 37°C) with or without 125 nM IL-6 in the absence or presence of 20 M SorLA propeptide prior to staining with anti-IL-6 and anti-SorLA antibodies. As demonstrated in Fig. 2A, astrocytes incubated in the presence of IL-6 showed a substantial vesicular uptake of IL-6, which to a large extent colocalized with SorLA (Pearson's r ϭ 0.47 Ϯ 0.03 [mean Ϯ SEM, n ϭ 15]). This uptake was practically abolished in the presence of a surplus of SorLA propeptide. Thus, as determined by automated counting of 15 randomly selected cells, astrocytes incubated at 125 nM IL-6 contained 159 Ϯ 10 (mean Ϯ SEM) positive vesicles per cell, whereas 25 Ϯ 5 positive vesicles were found in cells coincubated with IL-6 and SorLA propeptide and only 6 Ϯ 2 positive vesicles were seen in controls not subjected to IL-6 ( Fig. 2B). The results in SorLA ko astrocytes (lower panel in Fig. 2A, far right column in Fig. 2B) demonstrated a comparatively reduced uptake of IL-6 (ϳ 41%) in the absence of SorLA, and practically no uptake in the presence of SorLA's propeptide, which, apart from SorLA, also inhibits sortilin. Since sortilin also binds IL-6 (50), the data strongly suggest that SorLA and sortilin account for the uptake of IL-6 in astrocytes. Taken together, these data demonstrate that IL-6 selectively binds the SorLA Vps10p-D and that SorLA conveys the cellular uptake and endocytosis of IL-6.
Soluble IL-6R targets SorLA and is internalized. Using a similar setup, we next sought to determine whether SorLA also interacts with sIL-6R. SPR analysis demonstrated ( Fig. 3A to D) that it does and that sIL-6R displays much the same binding characteristics as IL-6. Thus, sIL-6R bound SorLA with high affinity (K d ϳ 34 nM) (Fig. 3A), and the binding was completely inhibited by the SorLA propeptide (Fig. 3B), but only partly (ϳ 50%) by NT (not shown). This suggested binding to the Vps10p-D, and, in agreement, the purified Vps10p-D was found to bind sIL-6R with an affinity (K d ϳ 9 nM) comparable to that of the entire extracellular part of SorLA (Fig. 3C) and, as before, binding was completely inhibited in the presence of the SorLA propeptide (Fig. 3D).
The interaction between sIL-6R and full-length SorLA was confirmed in wt and SorLA-transfected HEK293 cells. The cells were incubated (30 min, 37°C) at 250 nM sIL-6R and subsequently stained with anti-IL-6R and anti-SorLA antibodies. As apparent from Fig. 3E, the SorLA transfectants, unlike wt HEK293, displayed a heavy intracellular sIL-6R stain colocalizing with SorLA (Pearson's r ϭ 0.72 Ϯ 0.03 [mean Ϯ SEM, n ϭ 17]), signifying a considerable SorLA mediated uptake of the soluble receptor. In accordance with the SPR analysis, coincubation with the SorLA propeptide completely blocked the cellular uptake of sIL-6R. The uptake of sIL-6R was finally examined in cultured astrocytes isolated from wt and SorLA ko mice. It appears (Fig. 4A) that whereas SorLA-deficient cells did not present any internalization of sIL-6R, wt astrocytes exhibited a significant vesicular uptake largely colocalizing with SorLA (Pearson's r ϭ 0.40 Ϯ 0.06 [mean Ϯ SEM, n ϭ 9]). Again, this uptake was minimized in the presence of a surplus of SorLA propeptide. Quantification of the sIL-6R containing vesicles revealed that SorLA ko cells-and cells not exposed to sIL-6R-contained 5 Ϯ 1 and 2 Ϯ 1 (means Ϯ SEM, n ϭ 9) sIL-6Rpositive vesicles per cell, respectively, as opposed to wt astrocytes that contained 25 Ϯ 4 positive vesicles per cell upon incubation with sIL-6R and as little as 3 Ϯ 1 in cultures supplemented with SorLA propeptide (Fig. 4B). It follows that sIL-6R, similarly to IL-6, targets the SorLA Vps10p-D and is internalized by SorLA in cells.

Membrane-bound IL-6R and SorLA form a complex on the cell surface.
To determine whether SorLA also targets the full-length transmembrane IL-6R, crosslinking experiments were performed on transfected HEK293 cells. Transfectants expressing either IL-6R, SorLA, or both in combination were treated with the thiolcleavable and membrane-impermeable chemical cross-linker DTSSP. After 45 min, the reaction was stopped, the cells were lysed, and cross-linked adducts were subsequently immunoprecipitated (anti-SorLA Ig) and subjected to reducing SDS-PAGE and immunoblot analysis with anti-SorLA and anti-IL-6R Ig. As apparent from Fig. 5A, IL-6R was undetectable (not precipitated) in SorLA single transfectants, and only a weak band  SorLA in Interleukin-6 Signaling and Turnover Molecular and Cellular Biology resulted from IL-6R transfectants, whereas a comparatively much stronger band was produced by adducts precipitated from IL-6R/SorLA double transfectants. Coprecipitation was also seen in the absence of cross-linker (Fig. 5B). The data evidently indicate that membrane-bound IL-6R and SorLA, when coexpressed, interact on the cell surface membrane. Functional implications of the SorLA-IL-6R interaction. The findings discussed above suggest that SorLA may impact the functions of the full-length transmembrane IL-6R. Since IL-6R early on was reported to internalize poorly on its own (51, 52), we speculated whether SorLA could facilitate its endocytosis and trafficking. To that end, HEK293 cells, either IL-6R single transfectants or IL-6R/SorLA double transfectants, were incubated with anti-IL-6R and anti-SorLA antibodies (2 h, 4°C). The cells were then washed and reincubated in unsupplemented warm (37°C) medium, followed by fixation either immediately (zero time) or after 25 min, and eventually stained using matching   rapid and efficient internalization of both receptors. Notably, however, internalization of full-length IL-6R appeared equally effective in IL-6R single transfectants (Fig. 6B), whereas an IL-6R construct without the cytoplasmic tail (IL-6RΔtail) presented only a minor degree of endocytosis (Fig. 6C), even upon coexpression with SorLA (Fig. 6D) or in the presence of 25 nM IL-6 ( Fig. 6C), which has been reported to facilitate rapid gp130-mediated downregulation of IL-6R (53). Thus, IL-6R has the capacity to internalize on its own, and this capacity relies on its cytoplasmic domain and is unaffected by SorLA and ligand binding.
Subcellular fractionation of HEK293 transfectants further indicated that SorLA has little or no influence on the cellular localization and trafficking of IL-6R in general. Thus, the subcellular distribution of IL-6R (Fig. 6E, top panel) was virtually unchanged upon coexpression with SorLA (middle panel), as well as upon coexpression with a SorLA construct lacking the cytoplasmic tail (SorLAΔtail) the subcellular localization of which differs distinctly from that of wt SorLA (bottom panel).
Additional experiments were set up to determine whether SorLA influences the cellular turnover/half-life of the IL-6R. IL-6R single transfectants and IL-6R/SorLA double transfectants were biolabeled prior to washing and reincubation in unsupplemented medium at zero time. At given time points (0, 180, and 360 min), IL-6R was immunoprecipitated from cell lysates and analyzed by reducing SDS-PAGE and autoradiography. The result (Fig. 7A) shows similar half-lives of IL-6R in the two cell lines. Likewise, Western blot detection of unlabeled receptors released to the medium in a corresponding experiment (Fig. 7B) demonstrated that the cellular release of IL-6R was similar in single and double transfectants.
It can be concluded from the above that although IL-6R and SorLA appear to interact on the cell membrane, SorLA has no major impact on the sorting, endocytosis, or shedding of IL-6R. Thus, the findings are in good agreement with previous reports demonstrating that basolateral sorting of IL-6R in polarized cells is mediated by the IL-6R cytoplasmic domain and that monoclonal antibodies decrease the cellular uptake of IL-6 by blocking its binding to the receptor (54,55).
SorLA downregulates IL-6 cis signaling. Since signaling is another function that may be affected by complex formation between SorLA and IL-6R on the cell membrane, we next assessed the phosphorylation of STAT3 (pSTAT3) upon IL-6 cis and trans signaling in BA/F3 and HEK293 cells. IL-6 cis signaling was initially tested in BA/F3 cells, which has no endogenous expression of gp130, IL-6R, or SorLA. Cells transfected with either gp130/IL-6R or gp130/IL-6R/SorLA were incubated with or without 5 nM IL-6 for 15 min, and their response in terms of pSTAT3 was subsequently determined by Western blotting. As apparent from Fig. 8A, cells expressing SorLA showed a significantly lower level (ϳ28%) of pSTAT3 than cells without SorLA. Similar experiments were then performed in HEK293 cells, which have a minor endogenous expression of both gp130 and SorLA. The cells were transfected with either IL-6R, IL-6R/SorLA, or IL-6R/ SorLAΔtail and stimulated with 5 nM IL-6 as described above. In agreement with the result in BA/F3 cells, HEK293 cells coexpressing IL-6R and SorLA presented a lower (ϳ10%) content of pSTAT3 than the IL-6R single transfectants (Fig. 8B), and an even lower level (ϳ 33%) was seen in cells transfected with SorLAΔtail which, in contrast to wt SorLA, accumulates on the surface membrane (Fig. 8C). In both cases the difference was significant (P Ͻ 0.05), as determined by Wilcoxon signed-rank test. In contrast, transfection with SorLA in BA/F3 and HEK293 cells (not expressing IL-6R) did not significantly alter the response to stimulation with a combination of IL-6 and sIL-6R ( Fig.  8D and E). An SPR analysis was finally performed to determine whether SorLA might inhibits the binding of IL-6 to the IL-6R. As shown in Fig. 8F, the binding of IL-6 was completely abolished when IL-6R had been subjected to a saturating concentration of sSorLA.
Taken together the results suggest that expression of SorLA downregulates IL-6 cis signaling, presumably by interacting with the membrane-bound IL-6R and thereby hampering formation of the signaling complex by blocking the association between IL-6 and the membrane-bound IL-6R. On the other hand, SorLA does not affect IL-6 trans signaling, which involves the soluble form of the IL-6R.
Soluble SorLA may stabilize IL-6 and its capacity for trans signaling. As described previously (42) and demonstrated in Fig. 7B SorLA exhibits TACE mediated shedding of its luminal part at the cell surface whereby a circulating soluble receptor (sSorLA) is generated. To examine whether sSorLA may complex with IL-6 and affect IL-6 trans signaling, sSorLA and IL-6 were coincubated for 3 h at room temperature to allow them to form a complex (sSorLA:IL-6). In parallel, sSorLA, IL-6, and sIL-6R were each incubated separately under similar conditions. BA/F3 cells expressing gp130 were then stimulated for 15 min with either IL-6 and sIL-6R (5 nM each), sSorLA:IL-6 (40 nM:5 nM) and sIL-6R (5 nM), or just sSorLA (40 nM). The cells were subsequently lysed and the level of pSTAT3 determined by Western blotting. As evident from Fig. 9A, cells subjected exclusively to sSorLA did not result in any phosphorylation of STAT3, whereas cells stimulated with IL-6 and sIL-6R (trans signaling) gave rise to an increment in pSTAT3. Interestingly, sSorLA:IL-6 in combination with sIL-6R enhanced the observed IL-6 trans signaling by almost 3-fold. In contrast, if sSorLA and IL-6 were not allowed to coincubate prior to cell stimulation, the presence of SorLA did not increase IL-6 trans signaling (Fig. 9B), indicating that sSorLA must complex with IL-6 in order to affect  signaling. Moreover, IL-6 preincubated with the luminal part of SorCS3 (sSorCS3), a SorLA-related receptor with a 10-fold lower affinity for IL-6 (not shown), did not result in enhanced IL-6 trans signaling (Fig. 9C). To confirm the above findings, similar experiments were performed using SorLA ko astrocytes, as astrocytes do not express the membrane-bound IL-6R but respond to IL-6 trans signaling (13,49). As evident from  The right panel summarizes results of several (n) experiments in which the pSTAT3 levels (measured by densitometry) were set relative to the level obtained in response to IL-6ϩIL-6R (assigned the value 1). Bars indicate the SEM, The P value was calculated using the Wilcoxon signed-rank test. (B) The same BA/F3 cells were stimulated (15 min, 37°C) as indicated, but in this case none of the reagents had been coincubated prior to stimulation. The relative pSTAT3 levels were determined as described above. The inset depicts a Western blot of a single experiment, and the histogram summarizes results of nine experiments. (C) pSTAT3 levels in the same cells and stimulated as for panel A except that sSorLA had been substituted with sSorCS3. The inset shows a Western blot of a single experiment, the histogram summarizes (as described above) results of five separate experiments. (D) pSTAT3 levels in SorLA ko astrocytes stimulated with preincubated reagents as for panel A. The columns represent mean values (Ϯ SEM, n ϭ 3) relative to the pSTAT3 level in unstimulated astrocytes (assigned value 1). Data were evaluated by using one-way ANOVA and Tukey's test. (E) SPR analysis of the binding of sIL-6R to a preformed sSorLA:IL-6 complex. Immobilized sSorLA was initially exposed to IL-6 (100 nM) prior to the injection of fresh buffer containing 100 nM sIL-6R. The subsequent increase in response units signifies the binding of sIL-6R to the preformed sSorLA:IL-6 complex. The estimated K d is indicated. Fig. 9D, sSorLA in it self had no effect, but upon coincubation and complex formation with IL-6, it significantly increases IL-6 trans signaling. It follows from these findings that sSorLA may complex with IL-6 and positively affect IL-6 trans signaling.
To elaborate on the mechanism underlying the above finding, we performed an additional SPR analysis, in which immobilized sSorLA was preexposed to IL-6 prior to sIL-6R binding. As shown in Fig. 9E, sIL-6R displayed a significant binding to the preformed sSorLA:IL-6 complex, and the K d was found to be 8 Ϯ 2 nM (mean Ϯ SEM, n ϭ 3), which is similar to the K d of ϳ10 nM measured between IL-6 and sIL-6R (Fig. 1A). Thus, these data strongly suggest that sIL-6R is able to bind sSorLA-bound IL-6 and give rise to a trimeric sSorLA:IL-6:sIL-6R complex. Since sSorLA was required to coincubate (complex) with IL-6 in order to affect signaling and, furthermore, did not seem to modify the affinity between IL-6 and sIL-6R, we speculated that sSorLA might stabilize circulating IL-6 and thereby preserve its capacity for signaling. In order to explore this possibility, HEK293 cells transfected with IL-6R were stimulated (15 min) with 5 nM preincubated (3 h at room temperature) IL-6 or with 5 nM IL-6 not subjected to preincubation. The level of pSTAT3 was then determined by Western blotting, revealing that preincubated, compared to nonpreincubated, IL-6 resulted in a 27% Ϯ 2% lower level of pSTAT3 (mean Ϯ SEM, n ϭ 9; P Ͻ 0.01 [Wilcoxon signed-rank test]), i.e., preincubation had reduced the capacity of IL-6 for signaling. When preincubated in the presence of sSorLA, however, IL-6 appeared to preserve its activity ( Fig. 9A and D). These findings strongly suggest that sSorLA may serve as a stabilizing carrier protein for extracellular and circulating IL-6 and thereby maintain its functional half-life and trans signaling capability.

DISCUSSION
It is well known that IL-6 binds the IL-6R and subsequently recruits homodimeric gp130 for signaling. We report here that the sorting receptor SorLA binds IL-6, as well as the transmembrane and the soluble forms of the IL-6R. Upon binding, SorLA mediates efficient cellular uptake and internalization of both IL-6 and sIL-6R. SorLA does not affect the internalization or cellular localization of IL-6R; it does, however, hamper its capacity for IL-6 binding and for the induction of IL-6 cis signaling. In contrast, our findings show that the (shed) ectodomain of SorLA may act as an IL-6 carrier protein that stabilizes circulating IL-6 and thus preserves its capacity for trans signaling.
SorLA mediates uptake and internalization of IL-6 and sIL-6R. IL-6 and sIL-6R display similar binding patterns to SorLA and to the isolated Vps10p-D. Both bind with high affinity and in both cases binding are completely abolished by SorLA's propeptide. Since the propeptide is known to prevent ligand binding to the Vps10p-D (37), without preventing the binding of ligands (e.g., LPL) that target alternative sites in the luminal part of SorLA (31,37), our data demonstrate that both IL-6 and sIL-6R interact exclusively with the ␤-propeller domain of the Vps10p-D. The fact that the tridecapeptide NT, which also targets the Vps10p-D in both SorLA and the related receptor sortilin (37,56,57), exhibits little or no inhibition of IL-6 and IL-6R binding obviously reflects that the tunnel of the ␤-propeller harbors several separate binding sites and that, unlike the propeptide which serves as a lid blocking the entrance to the tunnel, NT is simply too small to prevent other ligands from gaining access (25,33). IL-6 and IL-6R, however, may well compete for binding to SorLA, if not because they share the same site, then due to their size. In that respect it is interesting that whereas interaction between SorLA and IL-6R appeared to exclude IL-6 from binding to either of the two (Fig. 8F), binding of IL-6 to SorLA (Fig. 9E) seemed not to interfere with IL-6:sIL-6R complex formation. This of course implies that SorLA may interact not only with IL-6 and sIL-6R but also with preformed IL-6:sIL-6R complexes in circulation.
In agreement with the in vitro binding experiments our findings in cell lines and wt astrocytes demonstrate that SorLA conveys rapid and efficient uptake of IL-6 and sIL-6R (and in all probability of the IL-6:sIL-6R complex). This suggests that one of SorLA's functions is to serve as a clearance receptor for these ligands and that the level of SorLA expression might impact on the level of circulating IL-6, sIL-6R, and IL-6:sIL-6R complex. This is interesting, since a number of reports have shown that elevated levels of IL-6 and sIL-6R are associated with and may partake in the development of diseases such as rheumatoid arthritis and multiple sclerosis (3,12,(17)(18)(19). Thus, it could be speculated that SorLA (or lack of functional SorLA) also plays a role in that context.
SorLA downregulates IL-6 cis signaling. As determined by cross-linking experiments, SorLA appears to complex with the full-length transmembrane IL-6R in addition to the secreted/shed form of the receptor. SorLA has previously been reported to bind and subsequently to alter the cellular sorting of the transmembrane APP (38). We found no evidence, however, of a corresponding function in relation to IL-6R. In contrast, endocytosis, the subcellular localization, the turnover (half-life), and the shedding of the IL-6R proved unaffected by coexpression with differentially sorted SorLA constructs. Thus, in agreement with previous reports addressing IL-6R sorting and endocytosis (54,55,58), we find that IL-6R trafficking relies exclusively on sorting motifs within its own cytoplasmic domain. Likewise, SorLA trafficking was unaffected by IL-6R and, in a broader perspective, it is tempting to question the very concept that a receptor-A can "take over" the sorting of a receptor B (and why not vice versa?) if both carry functional sorting motifs. We have recently shown that SorLA does mediate the internalization and degradation of the CNTFR and GFR␣1 (33,34), but in the present context it is important to note that they are both glycosylphosphatidylinositol anchored and have no capacity for independent sorting.
Although SorLA appears redundant in terms of IL-6R endocytosis and sorting, our data strongly suggest that interaction between the two may modulate induction of IL-6 cis signaling. Thus, the response to IL-6 (in terms of pSTAT3) in cells with modest or no expression of SorLA was significantly lowered upon transfection with wt SorLA and was even further reduced when the cells were transfected with SorLAΔtail. The latter indicates that signal induction is inhibited at the cell membrane where SorLA and IL-6R interact. SorLA-mediated clearance of IL-6 can hardly account for this, since SorLAΔtail is not internalized; instead, the aforementioned observation that binding of sSorLA to IL-6R prevents interaction between IL-6 and IL-6R (Fig. 8F) might offer a plausible explanation. In other words, SorLA:IL-6R complex formation at the cell membrane may hamper IL-6's access to IL-6R, its partner in signaling. It is even conceivable that SorLA could inhibit interaction with the signal transducer gp130. Thus, the luminal domain of SorLA interacts with soluble gp130-Fc (SPR analysis [data not shown]), although we were unable to cross-link and/or coprecipitate the full-length receptors in cells.
Soluble SorLA can act as an IL-6 carrier protein. Whereas the data presented above suggest that membrane-bound SorLA inhibits IL-6 cis signaling, we find that its shed luminal part (sSorLA) may serve as a carrier protein for circulating IL-6 and IL-6:sIL-6R and preserve their capacity for trans signaling. This conclusion is based on the following three findings: (i) IL-6 loses biological activity upon preincubation, but (ii) its activity appears to be protected in the presence of (and binding to) sSorLA, and (iii) binding to sSorLA does not hinder or interfere with its complex formation with sIL-6R. Furthermore, it is conceivable that sSorLA, like sIL-6R (59), may prolong the plasma half-life of IL-6 by preventing delaying its degradation (functional inactivation) in blood and tissues and its elimination via the kidneys. In this context, it is interesting that membrane-bound SorLA and sSorLA may have opposite effects, since full-length SorLA tends to remove IL-6 and downregulate its activity, whereas sSorLA seems to conserve IL-6 and IL-6 bioactivity. In any event, it is likely that SorLA can influence the level of IL-6 in circulation and perhaps even the pathophysiologic conditions accompanying elevated IL-6 levels.
Finally, our present and previous studies (33,50,60,61) demonstrate that Vps10p-D receptors, notably sortilin and SorLA, are new and significant participants in the functional regulation of cytokines and receptors of the IL-6 cytokine family. Subcellular fractionation. Subcellular fractionation of HEK293 cells by discontinuous iodixanol density gradient was performed as described elsewhere (65).
Analysis of STAT3 phosphorylation. Ba/F3 cells were seeded in 24-well plates at 1.2 ϫ 10 6 cells per well, whereas murine astrocytes were seeded in 24-well plates at 1.0 ϫ 10 5 cells per well. The cells were then starved for 3 h in unsupplemented DMEM prior to stimulation. HEK293 cells were seeded in 24-well plates and at 50 to 80% confluence starved for 3 h in unsupplemented DMEM prior to stimulation. Stimulations were performed by incubating the starved cells with the appropriate cytokines and/or soluble receptors for 15 min (37°C). The cells were then lysed at 4°C in 1% Triton X-100 lysis buffer as described above supplemented with a phosphatase inhibitor cocktail (PhosSTOP; Roche). Supernatants containing whole-cell extracts were analyzed for protein content, and the samples were subjected to Western blot analysis with antibodies specific for STAT3, pSTAT3, SorLA, and IL-6R.
Statistics. Data were evaluated either by using one-way analysis of variance (ANOVA) and Tukey's test or by using the Wilcoxon signed-rank test. For colocalization analysis, Pearson's correlation coefficient (r) was calculated using the Coloc 2 ImageJ plugin with default settings.