Chemokine Amplification in Mesangial Cells¹

Leukocyte migration from the blood into the tissues is controlled by a series of steps involving adhesion molecules and chemoattractants (1). Several mediators with the potential to selectively attract either polymorphonuclear or mononuclear cells have been isolated. Many of these effector molecules share a structural motif associated with the rapidly expanding family of chemoattractant cytokines termed chemokines. Chemokines are divided into two major subfamilies, based on the spacing of the first pair of N-terminal cysteine residues (2, 3). In CXC-chemokines, a single nonconserved amino acid residue separates these cysteines. In contrast, these cysteines are adjacent in CC-chemokines. This molecular subdivision generally correlates with functional activity. CXC-chemokines usually recruit neutrophils, whereas CC-chemokines tend to attract monocytes.

The potential involvement of chemokines in the pathogenesis of various infectious and inflammatory diseases has drawn considerable attention (4, 5, 6). Immunohistochemical analysis of human kidney biopsies indicated that expression of the CC-chemokine monocyte chemoattractant protein-1 (MCP-1)³ correlated with the local infiltration of macrophages in membranous nephropathy and glomerulosclerosis (6, 7). Other studies identified glomerular staining for MCP-1 in biopsies of patients with proliferative glomerulonephritis and lupus nephritis (8, 9). These data suggest a role for MCP-1 in the macrophage infiltrate of various forms of nephritis.

Chemokines are also implicated in the pathogenesis of experimental kidney diseases. In rodent models of Ab-induced glomerulonephritis. mRNA for macrophage inflammatory protein-2 (MIP-2), a CXC-chemokine that is chemotactic for neutrophils, was increased within the first hour after induction of disease, whereas mRNA for MCP-1 appeared later (10, 11, 12). In these models, the kinetic relationships among infiltrating neutrophils and monocytes correlate with the appearance of CXC- and CC-chemokines (12, 13, 14). Independent experiments using anti-chemokine Abs have also indicated a role for CXC- and CC-chemokines in the evolution of renal pathogenesis. Administration of Abs directed to the CXC-chemokines MIP-2, cytokine-inducible neutrophil chemoattractant, or IL-8 reduced acute inflammation, fibrin deposition, and glomerular damage as evidenced by reduction of proteinuria in rodent disease models (12, 14, 15). Similarly, treatment with Abs against the CC-chemokine MCP-1 reduced proteinuria, inflammation, interstitial fibrosis, and glomerular crescent formation, but the animals developed exacerbated neutrophil infiltrates (14). In comparison, treatment with a RANTES antagonist inhibited proteinuria and the numbers of infiltrating leukocytes but not fibrosis or crescent formation (14). A separate report using an Ab-mediated nephritis model in rats demonstrated that treatment with anti-MIP-1α attenuated proteinuria but not the accompanying neutrophil influx (16).

Mesangial cells are part of the renal glomerulus. They control the rate of glomerular filtration and provide support for the capillary loops (17). Mesangial cells are also important during glomerular injury; they are positioned to receive proinflammatory signals from capillary endothelium and infiltrating leukocytes. It is well established that IL-1, TNF-α, and LPS activate the in vitro synthesis of various chemokines, including MCP-1 and RANTES, in mesangial cells (6, 18, 19). This report extends these observations, demonstrating that treatment of mesangial cells with MIP-2 or KC induces the expression of MCP-1 and RANTES as well as autoinduction of MIP-2 and KC. This chemokine cascade may contribute to the spatial and temporal characteristics of renal inflammation.

We purchased 6- to 12-wk-old BALB/cJ mice of either sex or BALB that were deficient for CXC-chemokine receptor-2 (CXCR2) from The Jackson Laboratory (Bar Harbor, ME). Mice were maintained according to the guidelines of the Committee on Animals of the Harvard Medical School and those prepared by the Committee on Care and use of Laboratory Animals of the Institute of Laboratory Resources, National Research Council (Department of Health and Human Services Publication, National Institutes of Health 85-23, revised 1985).

The recombinant mouse chemokines MIP-2, KC, MIP-1α, and MIP-1β were purchased from R&D Systems (Minneapolis, MN). Eotaxin was a gift of Millennium Pharmaceuticals, Inc. (Cambridge, MA). Because some lots of chemokine contained traces of endotoxin, all chemokine reagents were treated with Detoxi-Gel (Pierce, Rockford, IL) before use to eliminate potential endotoxin contamination. Murine TNF-α and IFN-γ were procured from Genzyme (Cambridge, MA), whereas mouse platelet-derived growth factor-BB (PDGF-BB) was purchased from Life Technologies (Grand Island, NY). LPS and polymyxin B were obtained from Sigma (St. Louis, MO). Mouse IL-1β, 2H5 anti-MCP-1 mAb, and biotinylated monoclonal 4E2 anti-MCP-1 ELISA reagent were purchased from PharMingen (San Diego, CA). The 5F11 anti-MCP-1 Ab was prepared as described elsewhere (20). Anti-KC antisera and anti-KC mAb were obtained from R&D Systems. BSA was purchased from United States Biochemicals (Cleveland, OH).

Mesangial cells were obtained by outgrowth from mouse glomeruli obtained from four to ten mice. The method of Kreisberg et al. (21) was used for mesangial cell isolation. Briefly, kidneys were coarsely minced with scissors; tissue fragments were passed through a no. 60 mesh sieve (Curtin Matheson Scientific, Houston, TX) with a sterile rubber stopper and rinsed intermittently with 2% FCS/HBSS. The suspension was then sequentially passed through no. 100 and no. 200 sieves. Glomeruli were digested with 0.1% collagenase type IV (Sigma) and 0.1% trypsin (Life Technologies) for 30 min at 37°C before plating in 6-well tissue culture plates. Mesangial cells were cultured in DMEM with 20% heat-inactivated FCS in a 37°C humidified 10% CO₂ incubator. Cellular outgrowth was observed 10–14 days after seeding. Mesangial cultures were fed biweekly and transferred at confluence. Cells were passaged in d-valine-substituted media to eliminate fibroblasts. After five passages the cells were apparently homogeneous as assessed by phase contrast and light microscopy and by their staining characteristics with anti-muscle actin Ab. Contamination with monocytes was excluded by the absence of reactivity with anti-Mac-1 (M1/70, Boehringer Mannheim, Indianapolis, IN) and anti-Ia mAb (34-5-3, a gift from Dr. S. Abromson-Leeman, Harvard Medical School, Boston, MA) as assessed by direct and indirect immunofluorescence, respectively. Confluent mesangial cell cultures were washed with serum-free medium before the addition of chemokines or cytokines, which were added in serum-free medium.

Renal cortical cells were isolated with minor modifications of an established technique (22). Briefly, eight mouse kidneys were coarsely minced with scissors and tissue fragments were sequentially passed through no. 60, no. 100, and no. 200 sieves. The cell suspension was incubated with a mixture of 0.1% collagenase and 0.1% trypsin for 15 min at 37°C. Dissociated cells were washed in HBSS and seeded into a 75-cm² tissue culture flask. The cultures were maintained in DMEM supplemented with 10% FCS for 2–3 wk before use.

To induce leukocyte infiltration into the peritoneum, four to five mice were injected i.p. with 1 ml of 9% sodium casein (Sigma) in endotoxin-free PBS (23). After 16–24 h, animals were given a second casein injection. Three h later, peritoneal cells were harvested in 5–10 ml of HBSS without calcium chloride containing 0.5 mM EDTA. These cell populations contained 60–80% neutrophils, 20–30% macrophages, and <3% lymphocytes.

Limulus amebocyte lysate assays for the semiquantitation of endotoxin were conducted according to the manufacturer’s protocol (E-toxate, Sigma).

Immulon II 96-well microtiter plates (Dynatech Laboratories, Chantilly, VA) were coated with 50 μl containing 10 μg/ml purified anti-MCP-1 mAb (5F11) or 2 μg/ml anti-KC mAb (48415) in 50 mM of sodium carbonate-bicarbonate buffer (pH 9) at room temperature for 5 h. After removing the excess capture Ab, the wells were filled with 200 μl of 3% BSA in PBS and incubated at 4°C overnight to saturate excess binding sites. After three washes with PBS, serial dilutions of the experimental samples diluted in 3% BSA/PBS were added to the plates for 3 h at 37°C. After three washes, 50 μl of biotinylated detector anti-MCP-1 mAb (4E2) or purified anti-KC antiserum in 3% BSA/PBS was added to the wells and incubated for 2 h at room temperature. After three additional washes, the plates were incubated with alkaline phosphatase-conjugated avidin for 2 h. After three final washes, plates were developed with p-nitrophenylphosphate (Sigma). Titrations of purified recombinant mouse MCP-1 (PharMingen) or KC (R&D Systems) were included in each experiment for the preparation of standardization curves (20).

A total of 100 μl of culture supernatant was twice passed over a 50-μl heparin agarose column (Sigma) in a micropipet tip. The heparin beads were washed with 50 mM NaCl/10 mM HEPES (pH 7.2) buffer. The beads were then boiled for 5 min in 30 μl of sample buffer containing 5% 2-ME/2% SDS. Electrophoresis of a 20-μl protein sample was performed in 15% SDS-polyacrylamide minigels made with a MiniProtean II gel assembly kit (Bio-Rad, Hercules, CA) and electrophoresed at 150 V for 1.25–1.5 h. Proteins separated by SDS-PAGE were transferred to a 0.22-μm pore size nitrocellulose sheet (Bio-Rad). Sheets were blocked for 2 h at room temperature with 3% BSA/PBS and washed three times with PBS. Nitrocellulose sheets were incubated with 2H5 anti-MCP-1 mAb (10–20 μg/ml) in 3% BSA/PBS for 2 h at room temperature. The sheets were then washed three times for 10 min each with 3% BSA/PBS and reacted for 2 h at room temperature with a 1/1000 dilution of alkaline phosphatase-conjugated secondary Abs (Kirkegaard and Perry Laboratories, Gaithersburg, MD) in 3% BSA/PBS. The nitrocellulose sheet was finally washed three times for 5 min each with PBS; bound Abs were visualized by incubation with nitro blue tetrazolium/bromochloroindolyl phosphate (Kirkegaard and Perry Laboratories) at room temperature. The reaction was stopped by vigorous washing with PBS.

RNA was isolated from cell suspensions according to the manufacturer’s protocol using an RNA Isolation Kit (Stratagene, La Jolla, CA). The RNA was further cleansed by precipitation followed by washing with isopropanol and 75% ethanol, respectively. RNA was finally suspended in 50 μl of diethyl pyrocarbonate-treated water.

Assays for chemokine mRNA were conducted with multiprobe templates according to the manufacturer’s protocol (RiboQuant assay kit, PharMingen). The assay kits can simultaneously detect mRNA for each of the following mouse chemokines: lymphotactin, RANTES, eotaxin, MIP-1β, MIP-1α, MIP-2, IFN-γ-inducible protein-10 (IP-10), MCP-1, TCA3, and mRNA for the L32 and GAPDH housekeeping genes.

Cell migration was evaluated in 48-well Boyden microchambers separated with a 14-μm pore size polycarbonate filter (Neuroprobe, Cabin John, MD) as described previously (24). Mesangial cells (4 × 10⁶/ml) were suspended in endotoxin-depleted DMEM with 1% BSA. A total of 50 μl of cells was added to the upper Boyden chamber, and 30 μl containing the indicated concentration of chemokine was added to the lower chamber.

Single-stranded cDNA was synthesized from RNA using the Superscript Preamplification System for first strand cDNA synthesis (Life Technologies, Gaithersburg, MD) with the following modifications. A total of 3 μg of total RNA was treated with 2 U of DNase-I (bovine pancreas; Sigma) for 15 min at room temperature in an 18-μl volume containing 1× PCR buffer and 2 mM MgCl₂. Next, it was inactivated by incubation with 2 μl of 25 mM EDTA at 65°C for 10 min. Random hexamers (3 μl) were added and annealed to the RNA according to the manufacturer’s protocol. The reverse transcription reaction was performed according to protocol for a doubled reaction, with the modification that 2.4 μl of 10× PCR buffer and 6.56 μl MgCl₂ were added. One-half of the reaction (19 μl) was removed for use as a control. This aliquot was not subjected to reverse transcription. cDNA was then synthesized in a 20-μl reaction containing 1.5 μg of total RNA as in the protocol, with a parallel control reaction. The gene-specific PCR primers are listed in Table I. PCR was conducted in a 30-μl reaction mixture with 0.6 μl of cDNA and 0.5 μl of each primer under the manufacturer’s Taq DNA polymerase conditions (Qiagen, Valencia, CA). The PCR program included preincubation at 94°C for 2 min, amplification for 27–30 cycles of PCR at 94°C for 45 s plus 55–58°C annealing for 45 s plus 72°C extension for 45 s, and a final 72°C 3-min extension. A total of 6 μl of the PCR mixtures was visualized on 3% agarose minigels. ΦX174 RF DNA/HaeIII fragments (Life Technologies) were included as m.w. standards.

ELISA methods were used to measure MCP-1 levels in the supernatants of mesangial cultures. The background level of MCP-1 in unstimulated mesangial cells was consistently <1 ng/ml. In general, detectable levels of MCP-1 were first identified in the serum free-culture medium following a 42-h incubation with ≥10 ng/ml MIP-2 or KC (Fig. 1,A). These stimuli did not induce synthesis of all cytokines, because the same supernatants did not contain detectable levels of IL-1β (data not shown). Stimulation of mesangial cells with 100 ng/ml MIP-1α or 1000 ng/ml eotaxin failed to induce MCP-1 synthesis (Fig. 1,A). To establish that the Ag detected by ELISA represented conventional MCP-1 molecules, supernatants from unstimulated and LPS-, MIP-2-, or eotaxin-treated cultures were adsorbed to heparin beads that were subjected to Western blot analysis. The molecules detected in the MIP-2- and LPS-treated culture supernatants were the same size as the glycosylated recombinant mouse MCP-1 standard (Fig. 1 B).

Because LPS is a potent stimulus of mesangial cell MCP-1 synthesis (20, 25) Limulus-negative chemokine preparations were used to induce MCP-1. In addition, the effect of the endotoxin inhibitor polymyxin B was evaluated on MCP-1 production. Addition of 10 μl/ml polymyxin B blocked LPS-induced MCP-1 synthesis but had no effect on chemokine-stimulated MCP-1 synthesis (Fig. 2,A). Similar results were obtained in RPAs (data not shown). In contrast, heat treatment (100°C for 30 min) abrogated the capacity of MIP-2 and KC but not LPS to stimulate MCP-1 synthesis (Fig. 2,A). Finally, an anti-KC mAb completely blocked KC-induced MCP-1 production (Fig. 2,B). The failure of the anti-KC reagent to inhibit IL-1β-induced MCP-1 synthesis demonstrated the specificity of Ab blocking (Fig. 2 B).

The dose response and kinetics of MIP-2- and KC-induced MCP-1 expression were examined at the RNA level using RPAs. The RNase protection data were consistent with the protein data. A total of 25–50 ng/ml MIP-2 or KC were sufficient to stimulate MCP-1 mRNA synthesis after 15 h (Fig. 3,A). RNA synthesis was detected after 3 h, but 24 h were required for peak RNA expression (Fig. 3,B). MIP-2 and KC also stimulated the synthesis of RANTES and MIP-2 RNA (Fig. 3). Upon further exposure of the autoradiographs, MIP-1α, MIP-1β, and occasionally IP-10 mRNA were detected (Fig. 3,C). When mesangial cells were stimulated with 10 ng/ml IL-1β or TNF-α, mRNA for MCP-1 and RANTES was readily induced (Fig. 3,C). In addition, treatment with IL-1β and TNF-α induced MIP-2 mRNA. IFN-γ also induced MCP-1 mRNA, but mRNA for other chemokines was not detected. Although PDGF stimulates mesangial cell proliferation (20, 26, 27), it did not induce significant transcription of chemokine RNA. As a control, treatment with KC stimulated the production of mRNA for MCP-1, RANTES, MIP-2, MIP-1α, MIP-1β, and IP-10 (Fig. 3 C). In contrast, RNA for TCA3, lymphotactin, and eotaxin was not detected following the stimulation of mesangial cells with any of the indicated stimuli (data not shown).

In a separate set of experiments, mesangial cells were treated with control medium or activated with limiting doses of KC for 18 h. The RNA from these cells was assayed for KC and MIP-2 mRNA by RT-PCR. Positive controls included PECs and primers for the housekeeping gene β-glucuronidase. KC and MIP-2 PCR products were not detected from untreated mesangial cells. In contrast, MIP-2 and KC PCR products were amplified following incubation of mesangial cells with as little as 3 ng/ml KC (Fig. 4,A). To identify KC protein, mesangial cells were incubated with various stimulae for 24 h; afterward, the stimuli were washed out and the cells were plated in serum-free media for an additional 48 h. ELISAs were used to quantitate KC protein expression. After removal of the original stimulus of 10 or 100 ng/ml KC, we detected autocrine synthesis of 7 and 38 ng/ml, respectively, of KC protein. The levels of autocrine KC synthesis are comparable with the levels of KC produced following stimulation with MIP-2 or IL-1β (Fig. 4 B).

To determine whether the mesangial cell responses to MIP-2 and KC were cell type-specific, renal cortical epithelial cells were treated with 1000 ng/ml KC, 200 ng/ml MIP-2, 10 ng/ml IL-1β, or 10 μg/ml LPS. The results shown in Fig. 5 demonstrate that little or no MCP-1 mRNA was detected following stimulation with these high concentrations of KC or MIP-2. In contrast, LPS treatment strongly stimulated MCP-1 as well as RANTES, MIP-1β, MIP-1α, MIP-2, and IP-10 RNA synthesis. MCP-1 mRNA was also induced following treatment with IL-1β.

To confirm the tissue specificity of the effects of MIP-2 and KC, peritoneal neutrophils and macrophages were treated with 100 ng/ml MIP-2, 1000 ng/ml KC, 100 ng/ml LPS, 10 ng/ml IL-lβ, or 10 ng/ml TNF-α. PECs produced MCP-1 following stimulation with LPS, IL-lβ, or TNF-α. In addition, neutrophil chemotaxis was demonstrated following KC or MIP-2 stimulation (data not shown). In contrast, these cells failed to make detectable levels of MCP-1 protein following MIP-2 or KC stimulation (Fig. 6).

The most prominent feature of chemokines is their ability to induce cell migration. To examine the chemotactic capacity of MIP-2 or KC, mesangial cells were placed in a Boyden microchamber with varying doses of MIP-2 or KC and the positive and negative control chemokines TCA3 and MIP-1β, respectively (24). MIP-2-induced migratory responses displayed a bell-shaped dose-response curve characteristic of chemokines (24, 28). Peak chemotactic responses were noted with 10 ng/ml MIP-2 and 10–100 ng/ml KC. As noted elsewhere, TCA3 was a less efficacious chemoattractant; a total of 100-1000 ng/ml TCA3 were required to stimulate optimal mesangial cell migration. As expected (24), MIP-1β failed to attract mesangial cells at all concentrations tested (Fig. 7).

To determine whether MIP-2 also possessed growth factor activity, mesangial cells were serum starved for 2 days to deplete PDGF and other growth factors present in serum; the cells were then cultured in the presence of MIP-2 or MCP-1. MIP-2 and MCP-1 did not induce the uptake of [³H]thymidine, although TCA3 and IL-1β are active in this assay system (Ref. 24 , data not shown).

Mouse leukocytes bind MIP-2 and KC through a common seven-transmembrane-spanning G protein-coupled receptor, CXCR2 (29, 30). We examined primary mesangial cells for expression of CXCR2, CXCR3, and CXCR4 by RT-PCR. Although CXCR2, CXCR3, and CXCR4 products were readily detected in control PECs, we were unable to amplify these products from mesangial cells under identical experimental conditions (Fig. 8 A).

CXCR2 is the only mouse chemokine receptor known to bind MIP-2 and KC with high affinity. Thus, the failure to detect CXCR2 on mesangial cells was a surprise. To independently confirm that mesangial cells express a non-CXCR2 receptor capable of responding to MIP-2 and KC, mesangial cells derived from CXCR2-deficient mice were treated with these chemokines. Mesangial cells from these genetically altered mice produced MCP-1 mRNA (Fig. 8,B) and protein (Table II) following MIP-2 or KC stimulation. In addition, RANTES and MIP-2 RNA were also detected following KC or MIP-2 stimulation of CXCR2-deficient mesangial cells. (Fig. 8 B)

Mesangial cells are often associated with the pathogenesis of infectious and immune inflammatory responses involving the kidney (6, 7, 19, 31, 32). Following in vitro stimulation with LPS, IL-1 or TNF-α mesangial cells release chemokines and other proinflammatory mediators (18, 33, 34). This report demonstrates an additional pathway for chemokine induction. Mesangial cells display a unique amplification mechanism in which the CXC-chemokines MIP-2 or KC readily induce MCP-1 and RANTES expression; mRNA for MIP-1α, MIP-1β, and IP-10 was induced with lower efficiency. Interestingly, MIP-2 and KC display autocrine activity, stimulating the production of additional MIP-2 and KC. Similar properties were reported previously for the human CXC-chemokine melanoma growth-stimulatory activity factor/gro-α, which stimulated autoinduction in HUVEC (35). The autoinduction of chemokines by mesangial cells suggests a mechanism for perpetuating inflammatory responses within the parenchyma.

The murine chemokines KC and MIP-2 share structural homology with each other (29, 36) and with the human chemokines melanoma growth-stimulatory activity factor (also termed gro-α), gro-β, and gro-γ (36, 37). Due to the profusion of human CXC-chemokines with redundant structural and functional activities, more specific homology assignments are not possible (36). KC and MIP-2 generally induced similar patterns of chemokine transcription in mesangial cells.

Only a limited number of reports have described the effects of chemokines on mesangial cells. Barnes et al. (38) demonstrated that micromolar concentrations of the human CXC-chemokine platelet factor-4 inhibited mesangial cell proliferation. Another report indicated that nanomolar concentrations of human IL-8 stimulated the adhesion of mesangial cells to fibronectin (24). In the current report, induction of migratory responses and protein synthesis was also achieved with nanomolar concentrations of MIP-2 or KC. Similarly, low concentrations of the CC-chemokine TCA4 are known to induce mesangial cell migration (39), but 10-fold higher concentrations of TCA3 are required to stimulate chemotaxis (Fig. 7). The affinity range of conventional chemokine receptors is in the nanomolar range, suggesting the physiologic relevance of the MIP-2 and KC interactions (29, 30, 40, 41).

Although it is clear that MIP-2 and KC can act on primary mesangial cells, the receptor(s) responsible for transmitting the MIP-2/KC signals remains unknown. CXCR2 is the common leukocyte receptor for MIP-2 and KC, binding these ligands with high affinity (29, 30, 41). We were unable to detect CXCR2 on mesangial cells by RT-PCR, and mesangial cells from CXCR2-deficient mice were responsive to MIP-2 and KC, indicating that additional chemokine receptors for these ligands remain to be discovered. The finding that a neutralizing anti-KC mAb that inhibits neutrophil chemotaxis also blocks mesangial cell responses (Fig. 2 B) suggests that the same KC epitope is involved in binding to both CXCR2 and the unidentified mesangial cell receptor.

The ability of MIP-2 and KC to trigger mesangial cell chemokine induction seems to be tissue-specific, because similar effects were not noted on renal cortical epithelial cells or peritoneal neutrophils and macrophages. However, analysis of additional smooth muscle-like tissues should be undertaken. In summary, these data suggest that MIP-2 and KC can stimulate a chemokine amplification cascade that may perpetuate glomerular inflammation.

We thank Michael Berman and Claire Perchonock for technical assistance.