Abstract
To metastasize, tumor cells often need to migrate through a layer of collagen-containing scar tissue which encapsulates the tumor. A key component of scar tissue and fibrosing diseases is the monocyte-derived fibrocyte, a collagen-secreting profibrotic cell. To test the hypothesis that invasive tumor cells may block the formation of the fibrous sheath, we determined whether tumor cells secrete factors that inhibit monocyte-derived fibrocyte differentiation. We found that the human metastatic breast cancer cell line MDA-MB-231 secretes activity that inhibits human monocyte-derived fibrocyte differentiation, whereas less aggressive breast cancer cell lines secrete less of this activity. Purification indicated that Galectin-3 binding protein (LGALS3BP) is the active factor. Recombinant LGALS3BP inhibits monocyte-derived fibrocyte differentiation, and immunodepletion of LGALS3BP from MDA-MB 231 conditioned media removes the monocyte-derived fibrocyte differentiation-inhibiting activity. LGALS3BP inhibits the differentiation of monocyte-derived fibrocytes from wild-type mouse spleen cells, but not from SIGN-R1−/− mouse spleen cells, suggesting that CD209/SIGN-R1 is required for the LGALS3BP effect. Galectin-3 and galectin-1, binding partners of LGALS3BP, potentiate monocyte-derived fibrocyte differentiation. In breast cancer biopsies, increased levels of tumor cell-associated LGALS3BP were observed in regions of the tumor that were invading the surrounding stroma. These findings suggest LGALS3BP and galectin-3 as new targets to treat metastatic cancer and fibrosing diseases.
Introduction
A key component of scar tissue is the fibrocyte, a CD34+,CD45+,collagen+ cell found in healing wounds and fibrotic lesions. Monocytes are recruited to wounds or fibrotic lesions by chemokines (1, 2), and in response to wound signals such as tryptase released from mast cells, or thrombin activated during blood clotting, differentiate into monocyte-derived fibrocytes (3–5). The term “fibrocyte” has been used to refer to multiple cell types, including cells of the ear (6), CD34+,CD45+,collagen+ circulating cells (4, 7), and spindle-shaped CD34+,CD45+,collagen+ cells that differentiate in a tissue or in cell culture (4, 8). Spindle-shaped CD34+,CD45+,collagen+ fibrocytes are found in scar tissue (9, 10), and it appears likely that circulating CD34+,CD45+,collagen+ PBMC are either precursors to the fibrocytes in scar tissue (11, 12) or are fibrocytes that have left the scar tissue.
Although circulating fibrocytes can be directly purified (12–14), in this study, we cultured PBMC or monocytes to obtain monocyte-derived fibrocytes (15). Monocytes isolated from PBMC differentiate in vitro in a defined medium into monocyte-derived fibrocytes (16). Monocyte-derived fibrocytes express collagen and other extracellular matrix proteins, secrete proangiogenic factors, and activate nearby fibroblasts to proliferate and secrete collagen (3, 17–20). Increased monocyte-derived fibrocyte differentiation correlates with increased fibrosis in animal models (21, 22). Elevated circulating fibrocyte counts also associate with poor prognosis in human diseases (23).
In response to a foreign object or inflammatory environment, the immune system can initiate a desmoplastic response in which monocytes differentiate into monocyte-derived fibrocytes to form a sheath of fibrotic tissue around the foreign object (24–27). In response to some tumors, the immune system also initiates a desmoplastic response, attempting to contain the tumor (9, 28). This desmoplastic sheath is a dynamic, responsive tissue that adjusts to changing conditions in the tumor microenvironment (29, 30).
To metastasize through this desmoplastic tissue, cancer cells must find a way to remove scar tissue or to prevent scar tissue from forming (29–34). As cancer progresses toward metastasis and a more mesenchymal phenotype, it interacts with the immune system in different ways. Some tumors attempt to evade the immune system, and others act to suppress the immune system (35–39).
The MDA-MB 231 (231) cell line was isolated from metastases of a breast cancer patient (40). 231 cells behave aggressively in culture and murine models, displaying a metastatic phenotype that suggests that these cells retain the protein expression profile which allowed them to metastasize through the basement membrane of the original patient (41).
Galectin-3 binding protein (LGALS3BP), previously called Mac-2 binding protein and tumor-associated Ag 90K, is a heavily glycosylated 90-kDa protein (42). LGALS3BP binds to galectins 1, 3, and 7, fibronectin, and collagen IV, V, and VI (42–44). LGALS3BP is a member of the scavenger receptor cysteine-rich domain (SRCR) family of proteins (45). LGALS3BP is ubiquitously expressed in bodily secretions, including milk, tears, semen, and serum, usually 10 μg/ml (46). In patients with aggressive hormone-regulated cancers, including breast cancer, serum LGALS3BP concentration can be an order of magnitude higher than in normal serum (47–49). In breast milk, LGALS3BP concentration can rise and fall over the same range (∼10–100 μg/ml) depending on the length of time after the pregnancy (46). LGALS3BP is produced mostly by epithelial cells in glands (breast and tear ducts) and cancer cells (especially breast cancer cells) (50).
Higher levels of serum LGALS3BP correlate with worse outcomes in breast cancer patients (48, 49, 51, 52), whereas higher levels of LGALS3BP’s binding partner galectin-3 correlate with better outcomes for breast cancer patients (53). LGALS3BP promotes angiogenesis by increasing vascular endothelial growth factor signaling and directly signaling endothelial cells (43, 54). Mouse knockouts of LGALS3BP show higher circulating levels of TNF-α, IL-12, and IFN-γ, suggesting a role of LGALS3BP in regulating the immune system (55).
Galectin-3 is a ∼30-kDa protein expressed nearly ubiquitously in human tissues and can be secreted from cells, associated with membrane-bound carbohydrates, or located in the cytoplasm (53, 56–59). Galectin-3 is a biomarker of fibrosing diseases such as heart disease and pulmonary fibrosis (60, 61). As the disease severity increases, serum galectin-3 concentrations increase. Galectin-3 is widely expressed by immune system cells and promotes the differentiation of monocytes into macrophages (62). Galectin-3 interacts with a number of intercellular and intracellular receptors and ligands and is theorized to have roles in inflammation, host response to a virus, and wound healing (57, 62, 63).
In this paper, we show that 231 cells secrete LGALS3BP, which in turn inhibits monocyte-derived fibrocyte differentiation, and that conversely galectin-3 promotes monocyte-derived fibrocyte differentiation. LGALS3BP and galectin-3 are new modulators of fibrosis in the tumor microenvironment. In addition, the effects of LGALS3BP and galectin-3 on monocyte-derived fibrocytes show these proteins are active signaling molecules in cancer and fibrosis, respectively, and not passive biomarkers.
Materials and Methods
PBMC isolation and culture
Human blood was collected from adult volunteers who gave written consent and with specific approval from the Texas A&M University human subjects Institutional Review Board. PBMC were isolated and cultured as described previously (64, 65). Protein-free medium (PFM) was Fibrolife basal medium (Lifeline, Walkersville, MD) supplemented with 10 mM HEPES (Sigma-Aldrich, St. Louis, MO), 1× nonessential amino acids (Sigma-Aldrich), 1 mM sodium pyruvate (Sigma-Aldrich), 2 mM glutamine (Lonza, Basel, Switzerland), 100 U/ml penicillin, and 100 μg/ml streptomycin (Lonza). Serum-free medium (SFM) was PFM further supplemented with 10 μg/ml recombinant human insulin (Sigma-Aldrich), 5 μg/ml recombinant human transferrin (Sigma-Aldrich), and 550 μg/ml filter-sterilized human albumin (65). PBMC were cultured in SFM with the indicated concentrations of conditioned medium (CM), recombinant human galectin-1 and galectin-3 (PeproTech, Rocky Hill, NJ), or recombinant human galectin-3 binding protein (R&D Systems, Minneapolis, MN) for 5 d, after which PBMC were stained, and monocyte-derived fibrocytes were identified by morphology as 40- to 100-μm-long, spindle-shaped cells tapered at both ends with oval nuclei (65). The same phenotypic criteria were used to identify the 25- to 60-μm-long mouse monocyte-derived fibrocytes. Adhered cells and macrophages were counted as described previously (65). Human monocytes were purified, tested for purity, and cultured as described previously (8, 65). For immunohistochemistry, PBMC were fixed and stained for CD209 (BioLegend, San Diego, CA) as described previously (8).
Tumor cell lines and CM
231 (40), 435 (66), and DCIS.com (67) cells were gifts from Dr. W. Porter, Texas A&M University (College Station, TX). HT-29 (68), SW480 (69), DKOB8 (70), and HCT (71) cells were gifts from Dr. R. Chapkin (Texas A&M University, College Station, TX). MCF-7 (72), ADR-RES (73), OVCAR-8 (74, 75), SNU 398 (76), HEP-G2 (77), SW 1088 (78), U87 MG (79), and PANC-1 (80) cells were gifts from Dr. D. Wallis (Texas A&M University). Mono mac-1 (81) and Mono mac-6 (82) were from the DSMZ (Liebniz Institute: German Collection of Microogranisms and Cell Culture, Braunschwieg, Germany), and U-937 (83), HL-60 (84), THP-1 (85), and HEK-293 (86) cells were from the American Type Culture Collection (Manassas, VA). The MDA-MB 435 cell line has previously been classed as a breast cancer cell line but is currently classed as a melanoma cell line (66, 87). Each tumor cell line was tested for mycoplasma contamination using a PCR detection kit (MDBioproducts, St. Paul, MN) following the manufacturer’s instructions, and all work was done with cell lines containing undetectable levels of mycoplasma.
Tumor cell lines were grown in DMEM supplemented with 10% FCS (Genesee Scientific, San Diego, CA) in 75-cm2 flasks (BD Biosciences, Franklin Lakes, NJ) until 70% confluent. Adhered cells were washed three times with PBS and were then incubated with 10 ml PFM. After 24 or 48 h, the CM was collected and clarified by centrifugation at 300 × g for 10 min. CM from 231 cells was further clarified by centrifugation at 1,000 × g for 10 min, followed by clarification at 200,000 × g for 1 h. The supernatant was then concentrated with a 100-kDa centrifugal filter (EMD Millipore, Billerica, MD), and buffer was exchanged with 20 mM sodium phosphate (pH 7.2). Proteins were visualized by silver stain on 4–20% SDS-PAGE gels (Bio-Rad, Hercules, CA), and protein concentration was assessed by absorbance at 280 nm (Synergy MX, Bio-Tek, Winooski, VT).
Protein purification and identification
A total of 300 ml 231 CM was clarified by ultracentrifugation, concentrated, and buffer-exchanged as described above and resuspended in 1 ml. This was loaded on a 5-ml MonoQ anion exchange column on an AKTA chromatography system (GE Healthcare, Piscataway, NJ). The column was washed with 6 ml 20 mM NaPO4 (pH 7.4) buffer (first six fractions), and bound proteins were then eluted with a 24-ml gradient of 0–0.5 M NaCl in 20 mM NaPO4 (pH 7.4), collecting 0.5-ml fractions. Serial doubling dilutions of fractions were then mixed with PBMC, and their effect on monocyte-derived fibrocyte differentiation was measured as described previously (15). Trypsin digestion of samples, purification of peptides with Zip tips (EMD Millipore), and mass spectrometry were performed by the University of Utah Mass Spectrometry Core Facility, and peptides with MASCOT scores > 40 and mass errors < 3 ppm were used for protein identification.
Flow cytometry
PBMC were placed on an ultralow attachment plate (Corning, Corning, NY) and incubated with 231 CM at the indicated concentrations for the indicated time. These cells were then removed from the plate using ice-cold 5 mM EDTA in PBS (Rockland, Limerick, PA) with gentle pipetting. PBMC were collected by centrifugation at 300 × g for 10 min, resuspended in 100 μl ice-cold PBS, and analyzed for viability using propidium iodide (Sigma-Aldrich) and forward/side scatter via flow cytometry (Accuri-BD, Franklin Lakes, NJ) as described previously (88, 89).
Immunodepletion of CM
For immunodepletion, rabbit polyclonal anti–galectin-3 binding protein (BIOSS, Woburn, MA), mouse monoclonal anti–galectin-3 (BioLegend), mouse IgG isotype control (Jackson ImmunoResearch Laboratories, West Grove, PA), and rabbit polyclonal anti-protein S (Sigma-Aldrich) Abs were bound to protein G–coated Dynabeads (Invitrogen, Carlsbad, CA) beads following the manufacturer’s instructions. Beads complexed with Abs were mixed 1:10 with CM at 37°C for 2 h. Beads were then removed from the CM following the manufacturer’s instructions.
Sequencing 231 LGALS3BP
Total RNA was isolated from 231 cells using a kit (Omega Biotek, Norcross, GA), and cDNA was generated using a kit (Thermo Scientific, Waltham, MA). LGALS3BP was amplified using Phusion polymerase (New England Biolabs, Ipswich, MA) with primers 5′-AACTCGAGGTCCACACCTGAGTTGG-3′ and 5′-AAACTCCTAGTCCACACCTGAGG-3′ that encompassed all known or predicted transcript variants (90) and resulted in a single band on a DNA gel. Amplified LGALS3BP was ligated into pCMV and sequenced at Lonestar Labs (Houston, TX), using the primers listed above and internal primers (5′-CGCCCTGGGCTTCTGTGG-3′ and 5′-GGTCTATCAGTCCAGACG-3′).
Isolation of mouse spleen cells
Staining of biopsies
Deidentified slides of formalin-fixed paraffin-embedded biopsies or surgical specimens from patients with confirmed infiltrative ductal carcinoma of the breast were provided by Dr. K. Hunt at the University of Texas M.D. Anderson Cancer Center. Patients signed informed consent prior to the initiation of treatment. The M.D. Anderson Institutional Review Board approved the use of all patient-derived tissues and data. Slides were deparaffinized in xylene and rehydrated through graded ethanols. Ags were retrieved by incubating sections with Ag Unmasking Solution H-3300 (Vector Laboratories, Burlingame, CA) in a steamer for 20 min. Slides were then permeabilized by incubating with 0.2% Triton X-100 in PBS for 45 min at room temperature. Slides were blocked by incubating with 1% BSA in PBS for 1 h at room temperature. Slides were then incubated with primary Ab diluted in 0.1% BSA/PBS overnight at 4°C. Primary Abs were collagen I (rabbit pAb, 1:500, Abcam number ab34710), CD45RO (mouse mAb, 1:100, BioLegend number 304202), LGALS3BP (Rabbit pAb, 1:200, GeneTex number GTX116497), and galectin-3 (Rat mAb, 1:200, BioLegend number 125401). Secondary Ab in 0.1% BSA/PBS was then added for 1 h at room temperature. Secondary Abs were donkey anti-mouse DyLight 488, donkey anti-rabbit Red X, and goat anti-rat 488 (1:500, Jackson ImmunoResearch Laboratories). Sections were DAPI stained for 10 min and mounted with Dako Fluorescent mounting medium (DakoCytomation, Carpinteria, CA). Images were captured with an Olympus BX51 microscope and Olympus DP72 camera (Olympus, Tokyo, Japan) and CellSens software (Center Valley, PA).
Statistics
Statistics were performed using Prism (Graphpad Software, San Diego, CA). Differences were assessed by two-tailed t tests or two-tailed Mann–Whitney tests. Significance was defined as p < 0.05.
Results
231 and 435 cells secrete factors that inhibit monocyte-derived fibrocyte differentiation
To determine whether factors secreted from tumors might promote or inhibit monocyte-derived fibrocyte differentiation, we examined the effect of CM from a variety of human tumor cell lines on human monocytes cultured in SFM and measured the resulting monocyte-derived fibrocyte differentiation. In our culture conditions, monocyte-derived fibrocytes express canonical fibrocyte markers (8). Human PBMC were incubated in the presence or absence of tumor cell line CM, and after 5 d, monocyte-derived fibrocytes were counted. In the absence of CM, we observed 81–1374 monocyte-derived fibrocytes per 105 PBMCs from the different donors, similar to what we have observed previously (65). Per 105 PBMC originally added to each well, 17,617 ± 5,720 (mean ± SD, n = 38) remained adhered to the plate after fixation. Of these, 638 ± 278 had a monocyte-derived fibrocyte morphology, and 6062 ± 2208 had a phenotype similar to macrophages cultured in serum and MCSF (8). Because of this variability, monocyte-derived fibrocyte numbers were thus normalized to CM-free controls. CM from 231 (40) and 435 cells (66) inhibited monocyte-derived fibrocyte differentiation in a concentration-dependent manner (Fig. 1A, Supplemental Fig. 1A), and this effect was observed for PBMC from all donors tested. The 435 CM did not affect the number of adherent cells after removing weakly adhering cells, and then fixing and staining, whereas 231 CM slightly inhibited adherent cell number only at 12.5% CM, which is well above the IC50 for monocyte-derived fibrocyte inhibition (Fig. 1B, Supplemental Fig. 1B). PBMC exposed to 231 CM or 435 CM for 5 d did not have significantly increased cell death as assessed by propidium iodide staining (Fig. 1C, Supplemental Fig. 1C) or decreased total cell number (this includes the weakly adherent cells removed before fixing and staining) as assessed by removing all cells from the assay well and counting by flow cytometry (Fig. 1D, Supplemental Fig. 1D). Some concentrations of 231 CM increased total cell numbers (Fig. 1D). This may due to factors in the CM that promote cell survival and/or cell proliferation.
231 CM inhibits monocyte-derived fibrocyte differentiation. (A) PBMC were cultured in SFM in the presence of the indicated concentrations of 231 CM for 5 d. Monocyte-derived fibrocyte counts were normalized for each donor to the SFM control. (B) Counts of total adherent PBMC per five fields of view at the indicated concentrations of 231 CM. Total propidium iodide–positive PBMC (C) and total PBMC (D) after 5 d at the indicated concentrations of 231 CM, measured by flow cytometry. (E) Total number of monocyte-derived fibrocytes from monocytes cultured at the indicated concentrations of 231 CM for 5 d. Monocytes were 16 ± 9% (mean ± SEM, n = 3) of the PBMC and 92 ± 5% in the purified fraction. (F) Total number of macrophages from PBMC cultured at the indicated concentrations of 231 CM for 5 d. Values are mean ± SEM. The absence of error bars indicates that the error was smaller than the plot symbol. (A, B, and F) n = 38, (C–E) n = 3. *p < 0.05, **p < 0.01, ***p < 0.001 compared with the control (t test).
231 CM inhibits monocyte-derived fibrocyte differentiation. (A) PBMC were cultured in SFM in the presence of the indicated concentrations of 231 CM for 5 d. Monocyte-derived fibrocyte counts were normalized for each donor to the SFM control. (B) Counts of total adherent PBMC per five fields of view at the indicated concentrations of 231 CM. Total propidium iodide–positive PBMC (C) and total PBMC (D) after 5 d at the indicated concentrations of 231 CM, measured by flow cytometry. (E) Total number of monocyte-derived fibrocytes from monocytes cultured at the indicated concentrations of 231 CM for 5 d. Monocytes were 16 ± 9% (mean ± SEM, n = 3) of the PBMC and 92 ± 5% in the purified fraction. (F) Total number of macrophages from PBMC cultured at the indicated concentrations of 231 CM for 5 d. Values are mean ± SEM. The absence of error bars indicates that the error was smaller than the plot symbol. (A, B, and F) n = 38, (C–E) n = 3. *p < 0.05, **p < 0.01, ***p < 0.001 compared with the control (t test).
Monocyte-derived fibrocytes differentiate from monocytes (3, 16, 17). Cells in the PBMC population include T cells, B cells, NK cells, and other cells in addition to monocytes (15). To determine whether the CM effect on monocyte-derived fibrocyte differentiation is a direct effect on monocytes or is mediated by the other cells in a PBMC population, 231 or 435 CM was added to purified human monocytes. 231 CM and 435 CM inhibited monocyte-derived fibrocyte differentiation from purified monocytes (Fig. 1E, Supplemental Fig. 1E). For the inhibition of monocyte-derived fibrocyte differentiation from PBMC, the IC50 of 231 CM was 0.33 ± 0.05% (mean ± SEM, n = 37, Hill coefficient 1.06 ± 0.13) and that of 435 CM was 0.45 ± 0.17% (n = 7, Hill coefficient 0.73 ± 0.20). When added to monocytes, the 231 and 435 IC50s were 1.2 ± 0.2% (n = 5, Hill coefficient 1.08 ± 0.15) and 1.9 ± 0.7% (n = 3, Hill coefficient 1.70 ± 0.60), respectively. The difference in IC50s for 231 CM between PBMCs and monocytes was significant with p < 0.05 (t test); the difference for 435 CM was not significant. Because purifying monocytes and thus removing other cells from the PBMC population modestly increased the 231 CM IC50, these data suggest that either the monocyte purification procedure modestly reduced the ability of monocytes to respond to the factor(s) in CM that inhibits monocyte-derived fibrocyte differentiation or else the presence of the other cells in the PBMC population somewhat potentiates the ability of monocytes to respond to the factor(s). In either case, the data indicate that monocytes can respond direct to the factor(s).
In addition to differentiating into monocyte-derived fibrocytes, monocytes can also differentiate into macrophages. To determine whether the CMs that inhibit monocyte-derived fibrocyte differentiation also affect macrophage differentiation from monocytes, we counted the total number of adhered macrophages, as assessed by morphology. 231 CM caused a decrease in macrophage differentiation at 12.5% CM but did not affect the number of macrophages at lower concentrations, which inhibited monocyte-derived fibrocyte differentiation, and 435 did not significantly inhibit macrophage differentiation (Fig. 1F, Supplemental Fig. 1F). Taken together, the data indicate that factors in 231 and 435 CM affect monocytes to strongly inhibit monocyte-derived fibrocyte differentiation, while having no effect, or a relatively modest effect, on cell death, total cell numbers, numbers of macrophages, and the numbers of adherent cells.
Some but not all human cancer cell lines also secrete a monocyte-derived fibrocyte inhibitory activity
To determine whether other tumor types might secrete factors that inhibit monocyte-derived fibrocyte differentiation, we exposed PBMC to CM from cancer cell lines derived from human breast, skin, colon, liver, pancreas, brain, ovary, and leukocyte tissue. We defined units of activity as the inverse of the CM’s IC50 for monocyte-derived fibrocyte inhibition. Of the cell lines tested, CM from OVCAR-8, U-87 MG, 231, 435, MCF-7, and DCIS.com significantly inhibited monocyte-derived fibrocyte differentiation (Fig. 2). OVCAR-8 is an ovarian cancer cell line derived from a metastatic site (74, 75). U87-mg is derived from a glioblastoma (79). 231 is derived from a breast cancer metastasis (40). 435 has an uncertain origin: originally the cell line was listed as a breast cancer line by the American Type Culture Collection (66). Currently, the cell line is listed as a melanoma cell line (87). MCF-7 is a breast cancer cell line derived from a pleural effusion (72). DCIS.com is derived from a normal breast tissue cell line (MCF10A) passaged through a mouse and forms a nonmetastatic ductal carcinoma in situ when injected into mice (67). Each cell line whose CM inhibited monocyte-derived fibrocyte differentiation was isolated from hormone-secreting tissues (breast and ovarian), with the exception of the U-87 MG and 435 cell lines. No colon, liver, pancreatic, or leukemia CM significantly inhibited monocyte-derived fibrocyte differentiation (Fig. 2A), and 0.4–10% SW480 colon cancer CM modestly potentiated monocyte-derived fibrocyte differentiation (Fig. 2B).
The effect of CM from cancer cell lines on monocyte-derived fibrocyte differentiation. (A) CM from 25 different cancer cell lines show different levels of potentiation and inhibition for monocyte-derived fibrocyte differentiation. Activity units are the dilution of CM that inhibited monocyte-derived fibrocyte differentiation to 50% of the control value. Higher activity units indicate more potent inhibition of monocyte-derived fibrocyte differentiation. n = 3 for Mono mac-1, Mono mac-6, U937, HL-60, THP-1, DK0B8, HT-29, HCT-15, HCT (P21+/P53+), HCT (P21−/P53+), HCT (P21+/P53−), HCT (P21−/P53−), HEK-293, K562; n = 6 for ADR-RES, OVCAR-8, SNU-398, HEP-G2, U-87 MG, PANC-1, SW-1088, SW480, n = 7 for DCIS.com, n = 10 for MCF-7 and 435, and n = 38 for 231. A, cancers derived from breast tissue; B, skin; C, leukocyte; D, colon; E, embryonic kidney; F, ovarian; G, liver; H, brain; and I, pancreas. The absence of error bars indicates that the error was smaller than the plot symbol. #Inconsistent inhibition of monocyte-derived fibrocyte differentiation (variability of response of PBMC from different donors). *p < 0.05, **p < 0.01, ***p < 0.001 compared with the control (t test). (B) CM from SW480 cells potentiated monocyte-derived fibrocyte differentiation. Values are mean ± SEM, n = 6. *p < 0.05, **p < 0.01, ***p < 0.001 compared with the control (t test).
The effect of CM from cancer cell lines on monocyte-derived fibrocyte differentiation. (A) CM from 25 different cancer cell lines show different levels of potentiation and inhibition for monocyte-derived fibrocyte differentiation. Activity units are the dilution of CM that inhibited monocyte-derived fibrocyte differentiation to 50% of the control value. Higher activity units indicate more potent inhibition of monocyte-derived fibrocyte differentiation. n = 3 for Mono mac-1, Mono mac-6, U937, HL-60, THP-1, DK0B8, HT-29, HCT-15, HCT (P21+/P53+), HCT (P21−/P53+), HCT (P21+/P53−), HCT (P21−/P53−), HEK-293, K562; n = 6 for ADR-RES, OVCAR-8, SNU-398, HEP-G2, U-87 MG, PANC-1, SW-1088, SW480, n = 7 for DCIS.com, n = 10 for MCF-7 and 435, and n = 38 for 231. A, cancers derived from breast tissue; B, skin; C, leukocyte; D, colon; E, embryonic kidney; F, ovarian; G, liver; H, brain; and I, pancreas. The absence of error bars indicates that the error was smaller than the plot symbol. #Inconsistent inhibition of monocyte-derived fibrocyte differentiation (variability of response of PBMC from different donors). *p < 0.05, **p < 0.01, ***p < 0.001 compared with the control (t test). (B) CM from SW480 cells potentiated monocyte-derived fibrocyte differentiation. Values are mean ± SEM, n = 6. *p < 0.05, **p < 0.01, ***p < 0.001 compared with the control (t test).
The 231 monocyte–derived fibrocyte inhibitor is a protein and can be concentrated and purified
To determine whether the monocyte-derived fibrocyte inhibition activity in 231 CM is due to a protein, we exposed conditioned media to trypsin, heat, and freeze-thaw cycles. All three treatments strongly decreased the ability of 231 CM to inhibit monocyte-derived fibrocyte differentiation (Supplemental Fig. 2). The 231 CM retained ∼80% activity after clarification by ultracentrifugation and was largely retained by a 100-kDa filter (Fig. 3A). The components of 231 CM that passed through a 100-kDa filter potentiated monocyte-derived fibrocyte differentiation (Supplemental Fig. 3A). This potentiating activity was retained by a 10-kDa filter (Supplemental Fig. 3B). Taken together, these results suggest that the 231 CM activity, which inhibits monocyte-derived fibrocyte differentiation, is a protein.
231 CM’s monocyte-derived fibrocyte inhibitory activity is >100 kDa. The indicated fractions were assessed for their ability to inhibit monocyte-derived fibrocyte differentiation. The activity units were normalized to the value for the CM. Values are mean ± SEM, n = 24. #Fibrocyte potentiation.
231 CM’s monocyte-derived fibrocyte inhibitory activity is >100 kDa. The indicated fractions were assessed for their ability to inhibit monocyte-derived fibrocyte differentiation. The activity units were normalized to the value for the CM. Values are mean ± SEM, n = 24. #Fibrocyte potentiation.
To purify the 231 CM monocyte-derived fibrocyte differentiation inhibitor, the 100-kDa retentate was fractionated by anion exchange chromatography, and fractions were assayed for activity (Fig. 4A). There was little activity in the flow-through, some activity throughout the elution, and a large peak of activity in fractions 27 and 28, corresponding to a NaCl concentration of 375 mM. The IC50 of both of these fractions occurred at a ∼4,096-fold dilution, with a Hill coefficient of 4.6 ± 1.7 (mean ± SEM, n = 6). Silver-stained gels showed prominent bands at ∼85 and ∼39 kDa in fraction 27 (Fig. 4B). Tryptic fragments of proteins in fraction 27 were analyzed by mass spectrometry. In decreasing order of the number of identified peptides, the identified proteins were human galectin-3 binding protein, desmoplakin, plakoglobin, desmoglein type 1, pentraxin-3, adult intestinal phosphatase, and dermcidin. However, after purifying peptides with a Ziptip, the only identified peptides corresponded to LGALS3BP (GI:5031863).
Anion exchange chromatography of the partially purified factor. (A) A 100-kDa concentrated 231 CM, produced as in Fig. 3, was fractionated on an anion exchange column. Fractions were assayed as in Fig. 1, and the resulting monocyte-derived fibrocyte inhibition was measured by activity units, as in Fig. 2. (B) SDS-PAGE gel of the fractions, silver stained.
Anion exchange chromatography of the partially purified factor. (A) A 100-kDa concentrated 231 CM, produced as in Fig. 3, was fractionated on an anion exchange column. Fractions were assayed as in Fig. 1, and the resulting monocyte-derived fibrocyte inhibition was measured by activity units, as in Fig. 2. (B) SDS-PAGE gel of the fractions, silver stained.
Immunodepletion of LGALS3BP from CM removes most of the monocyte-derived fibrocyte inhibitory activity
To determine whether the LGALS3BP detected in CM affects monocyte-derived fibrocyte differentiation, we immunodepleted LGALS3BP from 231 and 435 CM. Immunodepletion with a control Ab had little effect on the ability of CM to inhibit monocyte-derived fibrocyte differentiation (Fig. 5, Supplemental Fig. 3). Immunodepletion of LGALS3BP from 231 CM (Fig. 5) increased the IC50 by 8.6 ± 1.3-fold (mean ± SEM, n = 7, p < 0.001, t test). Similarly, immunodepletion of LGALS3BP from 435 CM (Supplemental Fig. 3) increased the IC50 by 21 ± 11-fold (mean ± SEM, n = 7, p < 0.001, t test). These results suggest that LGALS3BP is a significant component of the 231 and 435 CM monocyte–derived fibrocyte inhibitory activity.
Immunodepletion of LGALS3BP decreases 231 CM’s monocyte-derived fibrocyte inhibitory activity. 231 CM was immunodepleted with anti-LGALS3BP or isotype control Abs. Values are mean ± SEM, n = 7.
Immunodepletion of LGALS3BP decreases 231 CM’s monocyte-derived fibrocyte inhibitory activity. 231 CM was immunodepleted with anti-LGALS3BP or isotype control Abs. Values are mean ± SEM, n = 7.
Recombinant LGALS3BP inhibits monocyte-derived fibrocyte differentiation
To test the hypothesis that LGALS3BP inhibits monocyte-derived fibrocyte differentiation, we incubated PBMC with recombinant human LGALS3BP. LGALS3BP significantly inhibited monocyte-derived fibrocyte differentiation with an IC50 of 0.22 ± 0.05 μg/ml (Hill coefficient 6.5 ± 2.2) (Fig. 6) but, unlike 231 CM (Fig. 6A), did not completely inhibit monocyte-derived fibrocyte differentiation. Taken together with the immunodepletion assays, the data indicate that LGALS3BP does inhibit monocyte-derived fibrocyte differentiation.
Recombinant LGALS3BP inhibits monocyte-derived fibrocyte differentiation. Recombinant LGALS3BP was added to PBMC, and monocyte-derived fibrocyte differentiation was assessed as in Fig. 1. Values are mean ± SEM, n = 8. ***p < 0.001 compared with the control (t test).
Recombinant LGALS3BP inhibits monocyte-derived fibrocyte differentiation. Recombinant LGALS3BP was added to PBMC, and monocyte-derived fibrocyte differentiation was assessed as in Fig. 1. Values are mean ± SEM, n = 8. ***p < 0.001 compared with the control (t test).
The 231 LGALS3BP mRNA encodes a canonical LGALS3BP
LGALS3BP has an experimentally verified transcript variant and several predicted transcript variants (90). To determine whether 231 cells secreted a truncated or alternatively spliced variant of LGALS3BP, mRNA was isolated from 231 and converted to cDNA. Using primers that encompass all known or predicted transcript variants (90), LGALS3BP cDNA was amplified via PCR and sequenced. The sequence encoded the four LGALS3BP peptides detected by mass spectrometry and was identical to human LGALS3BP from oral squamous carcinoma cells (93) and is identical to the sequence of the recombinant full-length human LGALS3BP (R&D datasheet).
LGALS3BP produced in cancer cells has a higher mass than recombinant LGALS3BP
To determine the concentration of LGALS3BP in CM, Western blots of CM and known amounts of recombinant LGALS3BP were stained for LGALS3BP. 231, 435, and OVCAR-8 CMs showed high concentrations of LGALS3BP compared with MCF-7, DCIS.com, and U87-mg CMs (Fig. 7). 435, 231, and OVCAR-8 cells accumulated ∼1 μg/ml LGALS3BP in their CMs. U-87 MG CM had very little LGALS3BP, suggesting that glioma cells may inhibit monocyte-derived fibrocyte differentiation by other means. LGALS3BP has a predicted mass of ∼60 kDa, but LGALS3BP isolated from the serum of cancer patients has a mass of ∼90 kDa (50) and the LGALS3BP we observed in cancer cell CM has a mass of ∼85 kDa. Although LGALS3BP in cancer CM appears as a single band, the recombinant LGALS3BP from Chinese hamster ovary (CHO) cells has several bands. R&D Systems is the only manufacturer of recombinant LGALS3BP produced in eukaryotic cells.
High concentrations of LGALS3BP are present in 321 and 435 CM. Western blot of equal volumes of the indicated concentrations (in nanograms per milliliter) of recombinant LGALS3BP and CM from the indicated cell types was stained with anti-LGALS3BP Abs. Blot image is representative of three separate experiments.
High concentrations of LGALS3BP are present in 321 and 435 CM. Western blot of equal volumes of the indicated concentrations (in nanograms per milliliter) of recombinant LGALS3BP and CM from the indicated cell types was stained with anti-LGALS3BP Abs. Blot image is representative of three separate experiments.
231 and 435 CM inhibit monocyte-derived fibrocyte differentiation using a dendritic cell–specific intercellular adhesion molecule 3-grabbing nonintegrin –dependent mechanism
The C-type lectin receptor dendritic cell–specific intercellular adhesion molecule 3-grabbing nonintegrin (DC-SIGN)/CD209 is expressed on monocytes and is upregulated as monocytes differentiate into dendritic cells (94). IgG is capable of inducing a pro- or anti-inflammatory macrophage phenotype by interacting with monocytes (95). IgG that is glycosylated with N-linked sialic-acid glycans binds DC-SIGN and causes macrophages to secrete IL-10, reducing inflammation (95, 96). Both glycosylated IgG and LGALS3BP bind human DC-SIGN (44, 95). To determine whether the DC-SIGN receptor might mediate the effect of LGALS3BP on monocyte-derived fibrocyte differentiation, we added 231 CM and 435 CM to PBMC from wild-type and CD209−/− (SIGN-R1−/−) mice. 231 CM and 435 CM inhibited monocyte-derived fibrocyte differentiation from wild-type C57BL/6 mouse spleen cells but potentiated monocyte-derived fibrocyte differentiation from CD209 knockout mouse spleen cells (Fig. 8) in a similar manner to 231 CM components that passed through a 100-kDa filter (Supplemental Fig. 3). Recombinant LGALS3BP, 231 CM, and 435 CM each upregulated CD209 expression on PBMC and decreased the number of fibronectin-positive monocyte-derived fibrocytes (Supplemental Fig. 4).
CD209 (SIGN-R1) is needed for the effect of 231 and 435 CM on mouse monocyte–derived fibrocyte differentiation. Cells were isolated from mouse spleens and incubated with the indicated concentrations of CM. Spleen cells from wild-type C57BL/6 (A) and SIGN-R1−/− knockout (B) mice were incubated with the indicated concentrations of 231 or 435 CM. After 5 d, monocyte-derived fibrocytes were counted. Values are mean ± SEM, n = 3; *p < 0.05 and ***p < 0.001 compared with the control (t test).
CD209 (SIGN-R1) is needed for the effect of 231 and 435 CM on mouse monocyte–derived fibrocyte differentiation. Cells were isolated from mouse spleens and incubated with the indicated concentrations of CM. Spleen cells from wild-type C57BL/6 (A) and SIGN-R1−/− knockout (B) mice were incubated with the indicated concentrations of 231 or 435 CM. After 5 d, monocyte-derived fibrocytes were counted. Values are mean ± SEM, n = 3; *p < 0.05 and ***p < 0.001 compared with the control (t test).
Galectin-3 and galectin-1 potentiate monocyte-derived fibrocyte differentiation
Galectin-1 and galectin-3 are binding partners for LGALS3BP (50, 97). To determine whether galectin-1 and -3 affect monocyte-derived fibrocyte differentiation, we incubated PBMC with recombinant human galectin-1 and -3 (98, 99). Galectin-1 and -3 significantly potentiated monocyte-derived fibrocyte differentiation (Supplemental Fig. 4C). To determine how a mixture of galectin-3 and LGALS3BP would influence monocyte-derived fibrocyte differentiation, recombinant galectin-3 and LGALS3BP were coincubated with PBMC. Concentrations of galectin-3 that potentiated monocyte-derived fibrocyte differentiation continued to potentiate monocyte-derived fibrocyte differentiation when mixed with concentrations of LGALS3BP that inhibited monocyte-derived fibrocyte differentiation (Fig. 6, Supplemental Fig. 4D). Galectin-3 did not potentiate monocyte-derived fibrocyte differentiation when mixed with a 3-fold higher quantity of LGALS3BP (Supplemental Fig. 4D). This suggests that the monocyte-derived fibrocyte-potentiating effect of galectin-3 competes with the monocyte-derived fibrocyte-inhibiting effect of LGALS3BP.
Increased LGALS3BP expression at the interface between breast cancer and scar tissue
To determine how tumor cells, LGALS3BP, galectin-3, and monocyte-derived fibrocytes interact in human tumors, we stained sections of human infiltrative ductal carcinomas for CD45RO, procollagen, collagen-I, galectin-3, and LGALS3BP. For all biopsies tested, the tumor cells strongly expressed LGALS3BP and procollagen. The staining intensity of LGALS3BP increased at the interface of tumor cells and stroma, particularly where tumor cells were invading through layers of collagen-rich stroma (Fig. 9). Monocyte-derived fibrocytes at the tumor edge were procollagen and collagen-1 positive, whereas the scar tissue was negative for procollagen and positive for collagen-1. “X” indicates the tumor, the arrow indicates tumor infiltration into scar tissue, and “*” indicates areas with monocyte-derived fibrocytes. Galectin-3 colocalized with the monocyte-derived fibrocyte markers CD45RO and collagen-I (Fig. 9). These data suggest that in some breast tumors LGALS3BP expression is increased at the interface between the tumor and desmoplastic tissue and that monocyte-derived fibrocytes are reduced when LGALS3BP expression is increased. Intriguingly, galectin-3, which we observed to potentiate monocyte-derived fibrocyte differentiation, colocalized with monocyte-derived fibrocytes (Fig. 9).
Breast cancer tumor sections visualized by immunofluorescence show increased LGALS3BP at the edge of tumors. Human infiltrative ductal carcinoma tumor specimens were sectioned and stained for CD45RO, collagen-1, procollagen LGALS3BP, and galectin-3. Images are H&E stain of a representative biopsy section (A and B), immunofluorescence for collagen (red) and CD45RO (green) (C), and immunofluorescence for LGALS3BP (red) and galectin-3 (green) (D), and immunofluorescence for procollagen (red) and CD45 (green) (E and F). Yellow indicates colocalization. The boxes in (A) and (E) indicate areas of magnification for (B)–(E) and (F), respectively. The image in (F) extends slightly below the white box in (E). X indicates the tumor, the arrow indicates tumor infiltration into scar tissue, and * indicates areas with monocyte-derived fibrocytes. Scale bar, 200 μm (A); scale bars, 20 μm (B–F).
Breast cancer tumor sections visualized by immunofluorescence show increased LGALS3BP at the edge of tumors. Human infiltrative ductal carcinoma tumor specimens were sectioned and stained for CD45RO, collagen-1, procollagen LGALS3BP, and galectin-3. Images are H&E stain of a representative biopsy section (A and B), immunofluorescence for collagen (red) and CD45RO (green) (C), and immunofluorescence for LGALS3BP (red) and galectin-3 (green) (D), and immunofluorescence for procollagen (red) and CD45 (green) (E and F). Yellow indicates colocalization. The boxes in (A) and (E) indicate areas of magnification for (B)–(E) and (F), respectively. The image in (F) extends slightly below the white box in (E). X indicates the tumor, the arrow indicates tumor infiltration into scar tissue, and * indicates areas with monocyte-derived fibrocytes. Scale bar, 200 μm (A); scale bars, 20 μm (B–F).
Discussion
In this paper, we show that several human tumor cell lines secrete activity that inhibits the differentiation of human monocytes into monocyte-derived fibrocytes. For a metastatic breast cell line and a metastatic melanoma cell line, the majority of the activity is LGALS3BP. LGALS3BP produced by cancer cells appears to act through the CD209 receptor to inhibit monocyte-derived fibrocyte differentiation. LGALS3BP’s binding partner, galectin-3, is upregulated in fibrotic tissue surrounding breast cancer tumors and promotes monocyte-derived fibrocyte differentiation at physiological concentrations. In breast cancer biopsies, LGALS3BP is concentrated at the edge of the tumor in regions with fewer monocyte-derived fibrocytes. Taken together, this suggests that LGALS3BP may inhibit monocyte-derived fibrocyte differentiation to facilitate metastasis in breast cancer and melanoma.
LGALS3BP (Mac2-BP and Ag 90K) is a secreted member of the SRCR family of proteins (50). LGALS3BP binds to galectin-1 and -3 (100), both collagen and fibronectin (101), and increases cell adhesion (97). Mouse knockouts of LGALS3BP show higher circulating levels of TNF-α, IL-12, and IFN-γ, suggesting multiple roles in regulating the immune system (102).
MCF-7 and DCIS.com cells, which are derived from non-metastatic breast cancers (67, 72), accumulate relatively low extracellular levels of LGALS3BP, whereas 231 cells, which are derived from a metastatic breast cancer (40), accumulate high extracellular levels of LGALS3BP (103–109). Patients with metastatic breast cancer tend to have abnormally high serum levels of LGALS3BP (50, 52). 435 accumulates high concentrations of extracellular LGALS3BP compared with other cancer cell lines (110), and patients with metastatic melanoma have higher serum LGALS3BP than patients with benign skin cancer (111). In patients with breast cancer (112, 113), ovarian cancer (114–117), or melanoma (118–120), serum LGALS3BP concentrations increase during the progression to metastasis (49, 121). An intriguing possibility is that LGALS3BP may play a role in metastasis by inhibiting monocyte-derived fibrocyte differentiation.
The Hill coefficient of recombinant LGALS3BP for monocyte-derived fibrocyte inhibition is 6.5, but 231 and 435 CM have Hill coefficients close to 1. Fractions of 231 CM containing LGALS3BP have a Hill coefficient of 4.6. Both 231 CM and 435 CM contain a factor (or factors) that potentiate monocyte-derived fibrocyte differentiation (Supplemental Fig. 3), suggesting that perhaps LGALS3BP competes with these factors to produce a Hill coefficient of ∼1. Recombinant LGALS3BP, 231 CM, and 435 CM all increase CD209 staining on monocytes, suggesting that LGALS3BP may cooperatively bind to monocytes by increasing CD209 expression.
LGALS3BP secreted from 231 cells has a higher apparent mass than recombinant LGALS3BP produced in CHO cells, despite the two having an identical primary structure. Because LGALS3BP is glycosylated (44), this suggests that the LGALS3BPs from the two cell lines have different glycosylations and/or other posttranslational modifications. 231 and 435 secrete ∼1 μg/ml LGALS3BP into CM, and 231 and 435 CM have an IC50 for monocyte-derived fibrocyte differentiation of ∼0.3% (Fig. 1, Supplemental Fig. 1). Taken together, these results indicate that the IC50 for LGALS3BP should be ∼0.3% × 1 μg/ml = ∼3 ng/ml. Recombinant LGALS3BP’s IC50 for monocyte-derived fibrocyte inhibition was 300 ng/ml. This then suggests that glycosylation and/or posttranslational modification of LGALS3BP may affect its ability to inhibit monocyte-derived fibrocyte differentiation. In agreement with this, LGALS3BP produced by cancer cells has at least a 4-fold higher affinity for CD209 (DC-SIGN) receptors than LGALS3BP produced by noncancer cells (44). The CD209 receptor, which we found to be necessary for the ability of LGALS3BP to inhibit monocyte-derived fibrocyte differentiation, is activated by posttranslational glycosylations (95). A reasonable possibility is thus that the posttranslational glycosylations of LGALS3BP produced in CHO and 231 cells have different affinities for CD209, resulting in the observed differences in IC50s.
Galectin-3 is expressed by monocytes and macrophages (122). Serum galectin-3 is upregulated in heart disease and other fibrosing diseases (123), and inhibitors of galectin-3 are currently in clinical trials for the treatment of fibrosing diseases (124, 125). Although serum galectin-3 is 100–900 ng/ml (98), we observed potentiation of monocyte-derived fibrocyte differentiation at 2.5–5 μg/ml. Similar concentrations of galectin-3 are necessary to induce changes in other cells (99), suggesting that either the effects of galectin-3 we and others observed in tissue culture are not physiological or that extracellular concentrations of galectin-3 in some tissues may be much higher than in the serum. If galectin-3 levels in a tissue are high enough to potentiate monocyte-derived fibrocyte differentiation, this would suggest that high levels of galectin-3 could potentiate fibrosis in part by potentiating monocyte-derived fibrocyte differentiation. Because we observed that galectin-1 also potentiates monocyte-derived fibrocyte differentiattion, high levels of galectin-1 may similarly potentiate fibrosis.
Taken together, this work elucidates a signal and receptor used by metastatic tumor cells to block a response of the innate immune system. An intriguing possibility is that blocking LGALS3BP may decrease the ability of tumor cells to inhibit the desmoplastic response and metastasize. Conversely, LGALS3BP might be useful to decrease monocyte-derived fibrocyte differentiation and thus decrease fibrosis. Because galectin-3 and galectin-1 potentiate monocyte-derived fibrocyte differentiation, these proteins might be useful to inhibit metastasis, although with the danger that they might promote fibrosis.
Acknowledgements
We thank Drs. Kelly Hunt and Cansu Karakas at the University of Texas M.D. Anderson Cancer Center for providing human breast cancer patient slides and aiding in interpretation of the immunofluorescent staining. We also thank Darrell Pilling and Nehemiah Cox for advice. We thank the volunteers who donated blood and the phlebotomy staff at the Texas A&M Beutel Student Health Center.
Footnotes
This work was supported in part by National Institutes of Health Grants T32CA009599 (to D.R.) and HL118507 (to R.H.G.).
The online version of this article contains supplemental material.
Abbreviations used in this article:
References
Disclosures
The authors have no financial conflicts of interest.