Search tips
Search criteria 


Logo of nihpaAbout Author manuscriptsSubmit a manuscriptHHS Public Access; Author Manuscript; Accepted for publication in peer reviewed journal;
Cancer Res. Author manuscript; available in PMC 2017 March 15.
Published in final edited form as:
PMCID: PMC4863655

Galectin-3 cleavage alters bone remodeling: Different outcomes in breast and prostate cancer skeletal metastasis


Management of bone metastasis remains clinically challenging and requires the identification of new molecular target(s) that can be therapeutically exploited to improve patient outcome. Galectin-3 (Gal-3) has been implicated as a secreted factor that alters the bone tumor microenvironment. Proteolytic cleavage of Gal-3 may also contribute to malignant cellular behaviors, but has not been addressed in cancer metastasis. Here, we report that Gal-3 modulates the osteolytic bone tumor microenvironment in the presence of RANKL. Gal-3 was localized on the osteoclast cell surface, and its suppression by RNAi or a specific antagonist markedly inhibited osteoclast differentiation markers, including TRAP, and reduced the number of mature osteoclasts. Structurally, the 158–175 amino acid sequence in the carbohydrate recognition domain (CRD) of Gal-3 was responsible for augmented osteoclastogenesis. During osteoclast maturation, Gal-3 interacted and co-localized with myosin-2A along the surface of cell-cell fusion. Pathologically, bone metastatic cancers expressed and released an intact form of Gal-3, mainly detected in breast cancer bone metastases, as well as a cleaved form, more abundant in prostate cancer bone metastases. Secreted intact Gal-3 interacted with myosin-2A, leading to osteoclastogenesis, whereas a shift to cleaved Gal-3 attenuated the enhancement in osteoclast differentiation. Thus, our studies demonstrate that Gal-3 shapes the bone tumor microenvironment through distinct roles contingent on its cleavage status, and highlight Gal-3 targeting through the CRD as a potential therapeutic strategy for mitigating osteolytic bone remodeling in the metastatic niche.

Keywords: Galectin-3, carbohydrate recognition domain (CRD), breast and prostate carcinoma, bone metastasis, tumor-microenvironment (TME)


Cancer bone metastasis remains as a debilitating clinical problem in patients with prostate, breast and lung cancers that utilize either hematogenous and/or lymphatic venues. Bone colonization by disseminated cancer cells leads to communication with the host bone microenvironment. This causes bone remodeling and skeletal-related events such as bone pain, which reduces the patient’s quality of life (1, 2). Some patients with advanced metastatic lesions experience pathological fractures or spinal cord compression. In these cases, clinical modality may include surgical intervention while considering patients’ performance status and estimated prognosis (3). Radiation therapy provides an option for localized lesion and pain control, however in the event of multiple cancer foci, systemic multidisciplinary treatments including hormone therapy, chemotherapy, and bone-targeted therapy have been utilized following bone metastasis treatment guidelines (4). Since bone metastasis is considered an incurable situation in most cases, current treatments are generally restricted to palliative management. There is an unmet need to identify treatment target(s) that orchestrate the bone tumor microenvironment (TME), and targeting such a critical molecule would be beneficial to halt and eradicate the progression of bone metastasis.

Galectin-3 (Gal-3), a lectin family protein, plays a significant role as a signaling modulator intracellularly, and a potent pro-inflammatory protein extracellularly and is involved in cell growth, homeostasis, apoptosis, adhesion, transformation, signaling induction, angiogenesis, fibrosis, cancer progression, and metastasis (5). During the bone metastatic process, Gal-3 mediates adhesion at the interface between tumor cells and bone marrow endothelial cells (6, 7). Focusing on the bone TME, prostate and breast carcinoma secrete Gal-3, suppressing osteoblast differentiation (1), which also results in a higher serum concentration of Gal-3 in patients with breast and prostate cancer metastasis as compared with normal counterparts (8, 9). Structurally, human Gal-3 is a β-galactoside-binding protein comprising 250 amino acid residues and is a chimeric gene-product composed of three distinct structural domains: a short NH2-terminal domain containing a phosphorylation site, a repeated collagen α-like sequence and a C-terminal domain containing a single carbohydrate recognition domain (CRD) composed of 140 amino acids (10). Due to its unique molecular structure, Gal-3 was reported to be a substrate for enzymatic cleavage by MMPs and prostate specific antigen PSA at Gly32-Ala33, Ala62-Tyr63, and Tyr107-Gly108 (1113). This proteolytic modification of secretory factors is named the degradome-peptidome (14). In humans, more than 500 proteases such as MMPs and PSA and their substrates, including Gal-3, are categorized in this manner (15), any of which could significantly affect the TME. Past studies showed that the cleaved forms of extracellular Gal-3 by breast and prostate carcinomas modulates angiogenesis and cell migration (12, 13) and is associated with malignancy in primary lesions (11, 16). However, the cleavage status of Gal-3 in metastatic lesions remains unclear. Therefore, we hypothesized that Gal-3 and its cleaved products could similarly affect the bone TME in breast and prostate cancer.

In the bone metastatic niche, growing cancer cells disturb the bone cellular differentiation by secretory factors affecting both osteoblasts and osteoclasts (2). Bone degradation is controlled by osteoclasts nourished in the bone TME, following which substrates released from the bone matrix enhance cancer progression in bone, resulting in pain (2). Thus, unlike other metastatic sites, anti-cancer therapeutic approaches alone are not enough to treat bone metastasis and alleviate pain (17, 18). Evidence from a clinical study showed that it is significant to control not only cancer growth but also osteoclasts in metastatic bone lesions, since osteoclast targeting therapy prolonged bone metastasis-free periods (19). Importantly, the functions of osteoclasts depends on their differentiation status regulated mainly by M-CSF (Macrophage colony-stimulating factor) and RANKL (Receptor activator of nuclear factor kappa-B ligand) (20, 21), which induce maturation into multinucleated functional osteoclasts (osteoclastogenesis). It should be noted that M-CSF stimulation induces Gal-3 expression to be localized on plasma membranes of osteoclast precursors, suggesting a role of Gal-3 in osteoclastogenesis (22). Nevertheless, the Gal-3 function of osteoclasts in bone TME is yet to be addressed and we hypothesized that Gal-3 may contribute to osteoclastogenesis in bone TME.

Here, we provide evidence that Gal-3 cleavage in the bone TME of breast and prostate cancers alter osteoclastogenesis, and suggest that targeting the CRD of Gal-3 may suppress bone remodeling induction by disseminated cancer cells.

Materials and Methods

Cell culture and reagents

Mouse osteoclast precursors, Raw 264.7, were purchased from American Type Culture Collection (ATCC, Manassas, VA) (2013), and the 2nd passage cells were used in all experiments. The cells were cultured in α-minimum essential medium supplemented with 10% fetal bovine serum, and the differentiation was induced by RANKL at 100µg/ml. Human osteoclast precursors, Poietics™, were purchased from Lonza (Walkersville, MD) (2014), and the differentiation was induced by M-CSF and RANKL following the manufacture’s protocol. The human breast cancer cell line MDA-MB-231 was a gift from Dr. Isaiah J. Fidler (University of Texas, MD Anderson Cancer Center, Houston, TX) (2005). The human breast cancer cell line T47D was a gift from Dr. Eric W. Thompson (St. Vincent’s Institute of Medical Research and University of Melbourne, Melbourne, Australia) (2000). Human prostate cancer cell lines LNCaP, DU145 and PC-3 were purchased from ATCC (2005). Customized polyclonal rabbit anti-CRD of Gal-3 antibody against amino acids 158–175 (HFNPRFNENNRRVIVCNT) was created by Genemed Biotechnologies (San Francisco, CA) based on the 3D structure of CRD (23) and purified by affinity chromatography using recombinant Gal-3. Monoclonal mouse anti-human RANKL neutralizing antibody was purchased from R&D systems (Minneapolis, MN). Polyclonal rabbit anti-Myosin-2A antibody (MYH9, D-16) was purchased from Santa Cruz Biotechnology (Santa Cruz, CA).

Human tissue specimens

Clinical samples were processed according to institution-approved protocols. Twenty-four biopsies of bone metastasis and 15 lesions each of other metastasis sites including lymph node and lung originating from breast cancer primary sites plus an additional 13 biopsies of bone metastasis and 10 lesions each of other metastasis sites including lymph node and lung originating from prostate cancer primary sites were obtained from the Wayne State University, Department of Pathology depository. A tissue array for prostate cancer metastasis was prepared at the Department of Pathology, University of Michigan which consists of 84 lesions from bone metastasis including limb, spine, rib, and 51 lesions from other metastasis sites including lymph node, lung, liver, bladder, and other soft tissues (24). All specimens were formalin-fixed and paraffin-embedded. Due to the nature of our paired sample analysis, if only one out of the two antibody measurements is available, that sample was excluded from analyses. As a result, 8 out of 54 and 22 out of 129 were removed for breast and prostate cancer, respectively.


Monoclonal rat anti-Gal-3 antibody (TIB166) against the amino terminus of Gal-3, which detects only intact Gal-3, was obtained from ATCC. Customized polyclonal rabbit anti-Gal-3 antibody (HL31) directed against the carboxy terminus of Gal-3, which detects the cleaved Gal-3 as well as intact Gal-3, was created by Bethyl Laboratories (Montgomery, Texas). Immunohistochemical staining and tissue evaluation was performed as described previously (11, 16).

Transfection and plasmid constructs

si-RNA against mouse Gal-3, Myosin-2A, and control si-RNA (Santa Cruz Biotechnology, shown as si-RNA No.1 and Silencer Select, Ambion, Austin, TX, shown as si-RNA No.2) were transfected into each cells using Lipofectamine RNAiMax™ reagent (Invitrogen, Carlsbad, CA). sh-RNA against Gal-3 was created using pSilencer 3.1-H1 neo expression vector (Ambion) following previous study (16). LnCaP transfectants secreting full-length (1–250 aa) and cleaved (33–250 aa, 63–250 aa) Gal-3 were generated using Lipofectamine® LTX and Plus™ transfection reagent (Invitrogen) and p3xFLAG-MYC-CMV-25 expression vector (Sigma-Aldrich) containing a preprotrypsin leader sequence for secretion following previous study (1, 12). Recombinant Gal-3-V5 was created using the pET30as (modified pET30a) vector containing Gal-3 sequence as described (1, 13).

TRAP stain and assay

The expression and enzyme activity of tartrate-resistant acid phosphatase (TRAP), an osteoclast differentiation marker, was estimated by using TRAP staining kit (B-Bridge International, Cupertino, CA).

In vitro osteolysis formation assay

Human osteoclast precursors were seeded on Osteoassay surface, 96-well format (Corning, Lowell, MA) at a density of 1.0 × 104 cells in differentiation medium containing M-CSF and RANKL. After 12 hours, 500 cells of hFOB, human fetal osteoblast cells and cancers were added. On the following day, 1µg of each antibody was added. On 6 day, osteolytic areas were visualized by von Kossa stain and quantitated by using Image J software. For morphological analysis, human osteoclast precursors, hFOB and cancer cells were seeded on β-TCP (beta tri-calcium phosphate) disk (3D-Biotek, North Brunswick, NJ) in the same method as above. On 6 day, cells were fixed by 1.5% paraformaldehyde with 0.5% glutaraldehyde in 0.1M phosphate buffer for 1 hour, followed by 1% OsO4 in 0.1M phosphate buffer for 1 hour. Then, samples were washed by 0.1M phosphate buffer, and dehydrated with ethanol. After immersion in tert-Butanol (2-metyl-2-propanol), samples were kept in a −20 freezer, and tert-butanol was evaporated using a vacuum evaporator. Next, ion coating was performed using gold, and then images were obtained using a Field Emission Scanning Electron Microscope, JSM-7600F (JEOL, Tokyo, Japan).

Statistical analysis

For a histological evaluation of immunohistochemistry, the paired data on each patient was categorized into one of the four possible patterns, namely (0, 0), (0, 1), (1, 0), and (1, 1) for positive (1) or negative (0) as results of stain by TIB166 and HL31. Our focus is the proportion of the pattern (0, 1) named ‘cleaved Gal-3 pattern’, and the pattern (1, 1) named ‘intact Gal-3 pattern’ out of all possible outcomes. Summary statistics were calculated within primary cancer origin and each metastasis site. Wilson’s 95% confidence intervals for the estimate of proportion were calculated. For each metastasis site, Fisher’s test was performed to test the association between the origin of cancer (breast cancer and prostate cancer) and the pattern of TIB166 and HL31; Odds ratios and 95% confidence intervals were calculated with conditional maximum likelihood estimation implemented in Fisher’s exact test. The raw p values were reported, as well as the adjusted p values for multiple testing using FDR method developed by Benjamini & Yekutieli procedure (26). The adjusted p values less than 0.05 are considered statistically significant. The statistical software R 3.0 was used for immunohistochemical analyses. Other statistical analyses were performed with EZR (Saitama Medical Center, Jichi Medical University, Saitama, Japan), which is a graphical user interface for R (27). Statistical differences were determined by two sided t-test or one-way ANOVA with Tukey HSD posthoc test.

Other methods (Tissue microarray, Immunofluorescence, Quantitative Real-Time PCR, Proteomics analysis, Co-immunoprecipitation, Co-culture) were described in Supplementary Materials and Methods.


Gal-3 involvement in osteoclastogenesis of bone tumor microenvironment

Initially, in order to establish the relationship of Gal-3 expression and bone metastasis, we performed immunohistochemistry with anti-Gal-3 antibody in the bone metastasis specimens obtained from prostate and breast cancer patients. Significantly, we observed the Gal-3 positive immunoreactivity in matured osteoclast lining on the bone matrix (Fig. S1-A). To find the difference of Gal-3 expression between bone metastatic lesion and primary bone tumor, we next examined Gal-3 expression in osteosarcoma, which is a major malignant tumor arising in bone, and in giant cell tumor, which is a benign osteoclast producing tumor (28). In the tissue microarray of osteosarcoma, Gal-3 positive cells appeared to congregate near the matured osteoclasts (Fig. S1-B). Similarly, in the giant cell tumor tissue array, immunoreactions with Gal-3 antibody were detected in the individual cells close to giant cells, which are excessively multi-nucleated osteoclast nourished by stromal cells (Fig. S1-C), and some of the Gal-3 positive cells were found to be located at the part of giant cells (Fig. S1-C’). Collectively, these findings gave evidence of Gal-3 role(s) on osteoclast differentiation in bone TME.

Gal-3 is essential for cell fusion during osteoclast differentiation

Next, we wanted to explore Gal-3 expression and its localization in osteoclast precursor cells during osteoclastogenesis. Then, we firstly examined Gal-3 expression in osteoclast precursor cells of Raw 264.7 (mouse), which also share the characteristics of monocyte/macrophage and Poietics™ (human), and detected in both precursors (Fig. S2-A, B). RANKL is a pivotal regulator controlling osteoclastogenesis in bone microenvironment and binds to the receptor i.e. RANK on osteoclast precursors, which leads to maturation of osteoclasts (20, 21). Since osteoclast differentiation induced by RANKL is a well-established experiment in vitro, we employed RANKL for our culture system. Then, we investigated the endogenous Gal-3 expression and its secretion into conditioned medium during RANKL-stimulated differentiation, and found a plentiful amount of endogenous Gal-3 in both lysate and medium (Fig. S3-A). Furthermore, we examined the difference of Gal-3 localization during osteoclast differentiation, in which Raw 264.7 cells were induced by RANKL, and Poietics™ cells were treated with RANKL plus M-CSF. Gal-3 was mainly localized on the outer surface of osteoclasts regardless of maturation level (Fig. 1–A). Next, we designed a gain of function study with recombinant Gal-3 to observe Gal-3 role in osteoclastogenesis. Since it was previously reported that the addition of recombinant Gal-3 inhibited the formation of osteoclasts (29), we firstly examined the inhibitory effect. However, we did not find the Gal-3 inhibitory effect on osteoclast differentiation in our hands. Also, recombinant Gal-3 alone was not sufficient to promote osteoclastogenesis (Fig. S3-B). Next, we addressed loss of function studies with siRNA-Gal-3 during osteoclast differentiation. Raw 264.7 cells were transiently transfected with si-Gal-3, followed by RANKL treatment, then we examined the expressional status and enzymatic activity of TRAP, a typical marker for osteoclast differentiation. The results showed that the number of TRAP-positive differentiated osteoclasts and its enzymatic activity were significantly reduced in si-Gal-3 transfected cells, indicating the contribution of Gal-3 on osteoclast differentiation (Fig. S3-C, 1-B). Since the carbohydrate recognition domain (CRD) of Gal-3 is responsible for multi-functionality of Gal-3 (10), we examined whether the CRD is required for the Gal-3 function on osteoclast differentiation, lactose, a CRD binding carbohydrate was added into the culture medium during osteoclast differentiation. In cells treated with lactose, the number of TRAP positive osteoclast cells and its enzymatic activity were noticeably reduced, compared to the cells treated with sucrose (control disaccharide) (Fig. 1–C). To validate the responsibility of Gal-3 CRD, we additionally used a Gal-3 antagonist i.e., MCP (also known as GCS-100) (30, 31). As predicted, MCP also inhibited osteoclastogenesis in a dose-dependent manner (Fig. 1–D). Next, we created Gal-3 down-regulated Raw 264.7 cells stably transfected with sh-Gal-3 and examined how cells progresses in response to RANKL. Interestingly, we observed that these established Gal-3 knockdown clones showed a deficient phenotype for cell fusion despite of RANKL stimulation (Fig. 1–E). These results suggested that endogenous Gal-3 in osteoclast cells play a crucial role during the process of cell fusion and differentiation.

Figure 1
Gal-3 is essential for cell fusion during osteoclast differentiation. A, Gal-3 expresses mouse osteoclast cells on the cell surface. Raw 264.7 cells were stained with or without RANKL exposure. Cells were fixed at 3 days after treatment. Images were shown ...

Carbohydrate recognition domain of extracellular Gal-3 regulates osteoclast differentiation

Because Gal-3 was intensely detectable in osteoclast culturing medium (Fig. S3-A), results of the lactose-based study (Fig. 1–C) made us question whether extracellular Gal-3 participates in osteoclastogenesis. To answer this, we generated anti-CRD antibody against the 158aa–175aa motif harboring the CRD, and purified the antibody with antigen-based affinity chromatography and tested whether anti-CRD antibody binds to CRD (Fig. S4). Next we explored whether anti-CRD antibody inhibits osteoclastogenesis in comparison to anti-RANKL neutralizing antibody, which is clinically well-established for the inhibition of osteoclast differentiation. Expectedly, the results showed that anti-CRD antibody obviously reduced the number of TRAP positive multinucleated cells and its enzymatic activity (Fig. 2–A, B, C). Intriguingly, anti-CRD antibody reduced the number of TRAP-positive multinucleated cells and TRAP enzymatic activity relatively less than anti-RANKL antibody in our tested condition (Fig. 2–A, B, C). Subsequently, we validated whether anti-CRD antibody also inhibits the expression of osteoclasts’ differentiation makers, including the transcriptional regulators of C-fos, Nfat-c1 and the executors of Acp5, Ctsk, Dc-stamp (32, 33). C-fos was identified as a key regulator, and caused a linage shift toward osteoclastogenesis in macrophage (34). Nfat-c1 regulates a number of crucial osteoclast-specific genes, including Acp5 and Ctsk (35, 36). Acp5 encodes tartrate-resistant acid phosphatase (TRAP), which is a highly expressed enzyme in matured osteoclasts, known as a typical marker for osteoclast differentiation (37). Ctsk encodes cathepsin K, which is a matrix-degrading proteinase involved in bone remodeling and resorption on lacunae (38). Dc-stamp was identified as an essential regulator of osteoclast and macrophage cell fusion, especially acts at later stage during fusion process (39). Quantitative real-time PCR analysis indicated that anti-CRD antibody down-regulated mRNA levels of these genes (Fig. 2–D), suggesting that extracellular Gal-3 affects the transcription of osteoclast differentiation markers.

Figure 2
Carbohydrate recognition domain of extracellular Gal-3 regulates osteoclast differentiation. A–C, Raw 264.7 cells were cultured at 10000 cells/well on 96-well plate. After 8 hours, human RANKL was added with 0.5µg of isotype IgG/anti-RANKL ...

Myosin-2A is a binding partner of Gal-3 for osteoclast differentiation

To understand how extracellular Gal-3 affects osteoclast differentiation, we needed to identify the extracellular Gal-3 interacting protein in osteoclast cells. Thus, we performed immunoprecipitation studies with recombinant Galectin-3-V5 (Gal-3-V5) as a bait. We found Gal-3 interacting proteins and submitted it to mass spectrometric analysis. The Gal-3 interacting protein was identified as Myosin-2A (p=0.044) (Protein Accession No. MYH9_MOUSE, Q8VDD5), which is reported to play as a modulator in osteoclast differentiation (40) (Fig. S5-A). To confirm interaction of Gal-3 and Myosin-2A, we treated the cells with recombinant Gal-3-V5 and carried out immunoprecipitation with anti-Mysosin-2A antibody, and the result showed that exogenous recombinant Gal-3-V5 more intensely interacted with Myosin-2A after RANKL exposure (Fig. 3–A). Next, we wanted to know their endogenous interaction and performed co-immunoprecipitation using osteoclast precursor cell lysates with/without RANKL treatment. As a result, Myosin-2A was co-precipitated with Gal-3 (Fig. 3–B). The reciprocal experiment showed that Gal-3 bound with Myosin-2A was apparently increased in response to RANKL (Fig. 3–B), indicating that Gal-3/Myosin-2A interaction is enhanced during osteoclast differentiation. To visualize their interaction in cells during osteoclast differentiation, confocal immunofluorescence analysis was utilized and showed the co-localization of endogenous Gal-3 and Myosin-2A on the cell surface, but no difference of co-localization pattern following RANKL stimulation and the differentiation process (Fig. 3–C). Further, to distinguish between cytosolic and secreted Gal-3 in Myosin-2A interaction, we used exogenous recombinant Gal-3-V5 and performed immunofluorescence analysis. Interestingly, exogenous recombinant Gal-3-V5 was also co-localized with Myosin-2A at the cell surface and both proteins strongly bound on the contact region of fused cells (Fig. 3–D). The result suggested that secreted Gal-3 modulates the role of Myosin-2A in the cell fusion process during osteoclast differentiation. Indeed, Myosin-2A knockdown and overexpression studies reported that Myosin-2A inhibits osteoclast precursor fusion (40). Therefore, next we wanted to understand how the positive role of Gal-3 affects a negative regulator of Myosin 2A during osteoclast differentiation and designed the experiment combined with si-RNA-Myosin-2A and anti-Gal-3 antibody (Fig. 3–E and S5-B, C). In accordance with a previous report of Myosin-2A inhibitory effect on osteoclast differentiation (40), si-RNA-Myosin-2A enhanced the number of TRAP positive multi-nucleated cells, and its enzymatic activity (Fig. 3–E). Also, si-RNA-Myosin-2A-enhancing osteoclast differentiation was suppressed by anti-CRD antibody (Fig. 3–E). In other words, anti-CRD antibody inhibits osteoclast differentiation regardless of the expression level of Myosin-2A. These data showed the pattern of epistasis that besides Gal-3 affects osteoclast differentiation through Myosin-2A, Gal-3 is likely to play an additional role, independent of Myosin-2A, ultimately resulting in the enhancement of osteoclast differentiation.

Figure 3
Myosin-2A is a binding partner of Gal-3 for osteoclast differentiation. A, at 2 days after the addition of RANKL, purified recombinant Gal-3-V5 (1µM) was added and incubated for 1 hour, then cross linked by DTSSP. Cells were lysed with Mem-PER®. ...

Cancer-secreted Gal-3 enhances osteoclastogenesis

Given that extracellular Gal-3 regulates osteoclast differentiation in a physiological setting (Fig 2 and and3),3), cancer-secreted Gal-3 could play a crucial role in a pathophysiological setting, i.e. bone tumor microenvironment. Firstly, we checked Gal-3 secretion in conditioned media cultured with metastatic breast and prostate cancer. As reported, secreted Gal-3 was detected in two forms: intact secreted Gal-3 and cleaved Gal-3 (Fig. 4–A) (1113). Although Gal-3 cleavage by MMPs is suggested to be an active process during tumor progression (11, 12), no experimental evidence of cleaved Gal-3 functionality in osteoclastogenesis has been provided so far. Thus, before delineating the functional activity of cleaved Gal-3 on osteoclast differentiation, we initially examined whether cleaved Gal-3 is also able to interact with Myosin-2A. Immunoprecipitation assay using conditioned media above showed that the intact form of Gal-3 binds to Myosin-2A, but the cleaved form did not (Fig. 4–B), implying that cleaved Gal-3 loses the ability of intact Gal-3 to bind with Myosin-2A. Next, we generated the cells stably transfected with N-terminal deletion mutants (33–250 aa, 63–250 aa) mimicking cleaved Gal-3 and intact Gal-3 sequence (1–250 aa) mimicking secretion by cancer cells (1, 12). In addition, we wanted to avoid the endogenous Gal-3 effect and employed LNCaP, a human prostate cancer cell line in which endogenous Gal-3 was not detected under tested conditions (1). Then, we examined how cancer-secreting truncated Gal-3 affects osteoclast differentiation in co-culture system. As shown in Fig. 4–C, LNCaP parental cells suppressed the formation of TRAP positive multi-nucleated cells, compared with the control (no cancer cells), indicating that LNCaP cells showed the inhibitory effect on osteoclast cells as a net result of osteoclast differentiation. Interestingly, in co-culturing with Gal-3- (1–250 aa) transfected cancer cells, secreted Gal-3 overrode the LNCaP-inhibitory effect and recovered the formation of TRAP-positive multi-nucleated cells in osteoclast cells (Fig. 4–C). This result indicated that cancer-secreting Gal-3 can regulate osteoclast differentiation, which is different from the previous observation that recombinant Gal-3 did not affect the differentiation (Fig. S3-B). We also observed that compared with intact Gal-3, N-terminal deletion mutants did not show a dominant negative effect, competing against intact Gal-3 secreted by osteoclast cells. Instead, these mutants slightly recovered TRAP positivity, indicating that truncated mutants showed, in part, the characteristics of intact Gal-3 activity (Fig. 4–C). Thus, we suggest that a shift of intact Gal-3 to cleaved Gal-3 might result in a reduction of total Gal-3 activity and the cleavage can attenuate osteoclast differentiation.

Figure 4
Cancer-secreted Gal-3 enhances osteoclastogenesis. A, Secreted/cleaved Gal-3 in human metastatic cancers. Each cancer cells were cultured with conditioned medium for 24 hours, next, the medium was concentrated to 20-fold, and then it was subjected to ...

As the inducible effect of both intact and cleaved Gal-3 derived from cancer cells on osteoclast differentiation was shown, we next examined therapeutic efficiency of anti-CRD antibody on suppression of osteoclastogenesis in co-culture systems mimicking the interactions of metastatic cancer cells and osteoclast cells. Expectedly, the treatment of anti-CRD antibody significantly reduced the formation of TRAP-positive multinucleated cells in co-culture with Gal-3- (1–250 aa) overexpressing LNCaP transfectant (Fig. 4–D). We found similar result in bone metastatic prostate cancer PC-3 cells, and breast cancer MDA-MB-231 cells, which express endogenously both intact and cleaved Gal-3 into culture media (Fig. 4–D). It is a remarkable point that anti-CRD antibody could exert effectively suppression on osteoclast differentiation in bone TME.

Targeting CRD of Gal-3 suppresses osteolytic bone remodeling

We further investigated the effect of Gal-3 on bone destruction by osteoclasts. As shown in Fig 5–A, we co-cultured human osteoclast precursors, human osteoblast cells, and human cancer cells on artificial bone matrix; i.e. β-TCP, beta-tricalcium phosphate to mimic the in vivo condition of bone TME. In accordance with the number of TRAP positive multi-nucleated cells in Fig. 4–C, LNCaP cells suppressed the osteolytic area compared to the differentiated condition without cancer cells (Fig. 5–B). Consistent with Fig. 4–C, Gal-3- (1–250 aa) overexpressing LNCaP cells recovered osteolytic area, and deletion mutants (33–250 aa, 63–250 aa) expressing cancer cells slightly recovered (Fig. 5–B), indicating that secreted Gal-3 enhanced osteolytic lesion, overcoming LNCaP-suppressing osteolysis. As shown in TRAP staining experiments (Fig. 4–D), anti-CRD antibody markedly reduced the osteolytic area (Fig. 5–C). To further detail anti-CRD antibody-mediated therapy, scanning electron microscopy allowed us to observe the 3 dimensional morphological changes of human osteoclasts on the destruction of bone matrix. We found that anti-CRD antibody maintained the osteoclast precursors in an undifferentiated state whereas in control IgG-treated condition, osteoclast precursors typically formed an osteoclast shape (Fig. 5–D). We also observed that osteoclast precursors expanded filopodia and aggregated with each other in response to RANKL stimulation, regardless of anti-Gal-3 antibody treatment (Fig. S6). However, anti-Gal-3 CRD treatment seems to block the osteoclast cell fusion, resulting in spheroidal cell clusters instead of flattened cells by maturation (Fig. S6). Collectively, anti-Gal-3 CRD antibody is considered as a promising therapeutic agent, suppressing metastatic cancer-induced osteolytic bone remodeling.

Figure 5
Targeting CRD of Gal-3 suppresses osteolytic bone remodeling. A, Schematic representation of in vitro osteolysis assay using β-TCP. B–C, On Day 1, Poietics™ human osteoclast precursors were cultured in the 96 well-format β-TCP ...

Gal-3 expression in bone metastasis TME

Preliminary immunohistochemistry (Fig. S1) and in vitro findings of Gal-3 cleavage effect on osteoclastogenesis and its therapeutic potentiality (Fig. 4 and and5)5) led us to profile relative intact and/or cleaved Gal-3 in bone metastatic and other metastatic tissues. Since it has been reported that Gal-3 is cleaved by MMPs, enzymatic modulators in TME and Prostate-Specific Antigen (PSA), a protease known as a tumor maker elevated in prostate cancer patient serum (1113), we confirmed Gal-3 cleavage by these two enzymes (Fig. 6–A). Previously we demonstrated the promising detection power with two anti-Gal-3 antibodies to distinguish expression pattern between intact and cleaved Gal-3 (11, 16) i.e., we purified IgG from antiserum, in which TIB166 antibody recognized intact Gal-3 but did not bind to cleaved Gal-3, whereas HL31 antibody recognized intact as well as the cleaved form of Gal-3 (Fig. S7-A). After that, we obtained approximately 210 paraffin embedded tissues from metastatic lesions of breast and prostate carcinomas in human patients, including lymph node, lung and bone. We serially sectioned two slides, allowing us to stain them with each antibody. As shown in Fig. 6–B and S7-B, the staining was evaluated as 0=negative, 1=positive. To organize the staining pattern of the two antibodies (TIB166, HL31), both TIB166 and HL31 positive refers to an intact Gal-3 pattern (1, 1); TIB166 negative and HL31 positive refers to a cleaved Gal-3 pattern (0, 1); Negative Gal-3 expression (0, 0) is both TIB166 and HL31 negative (Fig. 6–B and S7-B). Then, we analyzed statistically whether the significance of Gal-3 cleavage occurs between breast and prostate cancer in each metastatic niche. In bone metastasis, the odds of prostate cancer having intact-Gal-3 pattern (1, 1) is much lower than that of breast cancer (OR 0.048, 95% CI (0.010, 0.194); Fisher’s exact p value <0.001; adjusted p value <0.001) (Fig. 6–C). In contrast, the odds of prostate cancer having cleaved Gal-3 pattern (0, 1) in bone metastasis is much higher than that of breast cancer (OR 20.594, 95% CI (4.234, 200.793); Fisher’s exact p value <0.001; adjusted p value <0.001) (Fig. 6–C). The results indicated that the shift from intact to cleaved Gal-3 occurs in the prostate cancer bone metastatic niche more than in the breast cancer bone metastatic niche. In lung metastasis, the odds of prostate cancer having cleaved Gal-3 is much higher than the odds of breast cancer (OR 13.245 95% CI (1.787, 175.937); Fisher’s exact p value=0.006; adjusted p value=0.031), however, there was no significant reduction in intact Gal-3 expression (Fig. 6–C), indicating that Gal-3 cleavage shift is not obvious in lung metastasis. In lymph node metastasis, no significant difference in intact and cleaved Gal-3 expression was observed (Fig. 6–C). Taken together, it is likely that proteolytic shift from intact to cleaved Gal-3 is a characterized event in bone metastatic niche of prostate cancer.

Figure 6
Breast and prostate cancer expresses intact/cleaved Gal-3 in bone metastasis. A, Cleavage of Gal-3 by MMP and PSA. 50ng of recombinant Gal-3 was incubated for 1 hour at 37°C with proMMP-2 or proMMP-9 activated by APMA. PSA was used in EDTA(100mM)/PBS ...


The data presented here demonstrated that cancer-secreted Gal-3 plays a crucial role in osteoclast differentiation in the presence of RANKL. RANKL is a pivotal regulator of osteoclastogenesis produced by osteoblasts, cancer cells and activated immune cells in the bone marrow (41, 42). During osteoclast maturation, extracellular Gal-3 mediates osteoclast cell fusion, consistently earlier observations reported that cytosolic Gal-3 is implicated as an inducer of placental cell fusion (43). Cell fusion is required for osteoclast differentiation, and comprised of three steps, i.e. 1) cell-cell adhesion, 2) cytoskeletal rearrangement, and 3) fusion (44, 45). Previously, a numbers of proteins were reported to be involved in cell fusion during osteoclast differentiation, including CD200, DC-STAMP and E-cadherin (39, 46, 47). Adding to this list, Gal-3 was found to be located extracellularly at the cell-cell contact site where it binds with Myosin-2A. Myosin-2A is a cytoskeletal protein and classified as a non-muscle myosin, associated with cellular morphology (48). In osteoclast precursors, proteolytic inactivation of Myosin-2A on the inner cell surface induced cell fusion, leading to osteoclastogenesis, indicating that Myosin-2A is a suppressor of osteoclast differentiation (40). Although it is not determined how extracellular Gal-3 nullifies intracellular Myosin-2A function(s), a recent study shows that extracellular Gal-3 internalizes by binding with CD44 and β1-integrin (49). Similarly, cancer-secreted Gal-3 might internalize into osteoclast cells, whereby and thereby interact with Myosin-2A. The qPCR results showed that extracellular Gal-3 affects transcriptional activities of osteoclast differentiation markers. Gal-3/Myosin-2A interaction might enhance the down-stream pathway of RANKL/RANK along with the fusion process. In addition, Galectins generally bind to glycosylated receptors via conservative CRD motif. Likewise, Gal-3 interacts with Integrin αM (CD11b) and Integrin β2 (CD18) on the cell surface of monocytes/macrophages/osteoclast precursors (50). Thus, Gal-3-activating integrins possibly cross-talk with RANKL-mediated signaling pathways and transcription. Previously, we documented that secreted Gal-3 exerts an inhibitory effect on osteoblast differentiation through CRD-mediated Notch receptor binding (1). Taken together, cancer-secreted Gal-3 may exhibit dual properties in bone remodeling: 1) secreted Gal-3 enhances osteoclast fusion and 2) suppresses osteoblast differentiation, leading effectively to osteolysis. Our immunohistochemistry data provides a new insight into the degradomic-peptidomic regulation in bone metastatic lesions by the profile of intact and cleaved Gal-3. Considering that Gal-3 cleavage events attenuate osteoclastogenesis, the level of Gal-3 secretion and its cleavage products could alter bone TME, and may be a cause of different bone metastasis lesions in clinical observations: osteolytic (mainly breast cancer, expressing intact Gal-3), osteosclerotic (mainly prostate cancer, expressing cleaved Gal-3), and mixed lesion (51. 52). The differences of Gal-3 cleavage between breast and prostate bone metastasis may be affected by the activity of tumor-derived MMPs and/or PSA (11, 16). In addition, tissue inhibitors of metalloproteinase (TIMPs) are expressed and α1-antichymotrypsin (ACT), a PSA inhibitor may be generated in the host bone microenvironment (53, 54). Thus, the rate of these proteases/protease inhibitors could determine the fate of Gal-3 cleavage, ultimately creating a unique tumor-specific bone TME.

Denosumab, a humanized monoclonal antibody against RANKL has been used as a first line therapeutic drug to reduce the risk of bone destruction due to skeletal metastasis (55), but is not an ultimate treatment because the efficacy is generally limited to osteoclasts. Our data suggest a novel therapeutic modality for both bone remodeling and metastatic cancer cells. Anti-CRD antibody and/or Gal-3 antagonist reduce the bone destruction by suppressing osteoclast differentiation. In addition, Gal-3 is closely associated with cancer cell’s proliferation, apoptosis, chemotherapeutic resistance and preferable adhesion to bone marrow endothelium, therefore Gal-3-targeting therapy may be advantageous in the quest to halt bone metastasis and progression. Meanwhile, it is important to note that prostate cancer patients do not have any antibodies which recognize the CRD (Fig. S8), supporting a previous study in colon cancer patients (56). Therefore, therapy using anti-CRD of Gal-3 may be efficacious.

In conclusion, as depicted in Fig. 7, Gal-3 accelerates/attenuates the osteolytic vicious cycle in breast and prostate bone metastasis. The degradomic-peptidomic alteration of Gal-3, producing its cleaved form should be considered as an integral part of bone TME. We propose a concept of ‘anti-osteoclast fusion’ through Gal-3 targeting as therapeutic modality to promote potentially more effective bone metastasis therapy and improve patients’ quality of life.

Figure 7
A proposed model for the functional role of Gal-3 in bone metastatic niche. (Left) in breast cancer bone metastasis, cancer cells secrete intact Gal-3, then Gal-3/Myosin-2A interaction accelerates osteoclast cell fusion, producing mature osteoclasts and ...

Supplementary Material


The authors thank Zhi Mei (Wayne state university, Dept. of Chemistry, Electron Microscopy Core) for scanning electron microscopic analysis, Linda Mayernik (Wayne State University, School of Medicine, Microscopy Imaging & Cytometry Resources Core) and Paul Stemmer (Wayne State University, Proteomics Core) for mass spectrometric analyses.

Financial Support: This work was supported by NIH/National Cancer Institute R37CA46120 (A. Raz), NIH Center Grant P30 ES06639 (G. Bepler) and NIH Shared Instrumentation Grant S10 OD 010700 (P. Stemmer).


There are no potential conflicts to declare.


1. Nakajima K, Kho DH, Yanagawa T, Harazono Y, Gao X, Hogan V, et al. Galectin-3 inhibits osteoblast differentiation through notch signaling. Neoplasia. 2014;16:939–949. [PMC free article] [PubMed]
2. Suva LJ, Washam C, Nicholas RW, Griffin RJ. Bone metastasis: mechanisms and therapeutic opportunities. Nat Rev Endocrinol. 2011;7:208–218. [PMC free article] [PubMed]
3. Manabe J, Kawaguchi N, Matsumoto S, Tanizawa T. Surgical treatment of bone metastasis: indications and outcomes. Int J Clin Oncol. 2005;10:103–111. [PubMed]
4. Coleman R, Body JJ, Aapro M, Hadji P, Herrstedt J. Group. EGW. Bone health in cancer patients: ESMO Clinical Practice Guidelines. Ann Oncol. 2014;25:124–137. [PubMed]
5. Liu FT, Rabinovich GA. Galectins as modulators of tumour progression. Nat Rev Cancer. 2005;5:29–41. [PubMed]
6. Lehr JE, Pienta KJ. Preferential adhesion of prostate cancer cells to a human bone marrow endothelial cell line. J Natl Cancer Inst. 1998;90:118–123. [PubMed]
7. Glinskii OV, Sud S, Mossine VV, Mawhinney TP, Anthony DC, Glinsky GV, et al. Inhibition of prostate cancer bone metastasis by synthetic TF antigen mimic/galectin-3 inhibitor lactulose-L-leucine. Neoplasia. 2012;14:65–73. [PMC free article] [PubMed]
8. Balan V, Wang Y, Nangia-Makker P, Kho DH, Bajaj M, Smith D, et al. Galectin-3: a possible complementary marker to the PSA blood test. Oncotarget. 2013;4:542–549. [PMC free article] [PubMed]
9. Iurisci I, Tinari N, Natoli C, Angelucci D, Cianchetti E, Iacobelli S. Concentrations of galectin-3 in the sera of normal controls and cancer patients. Clin Cancer Res. 2000;6:1389–1393. [PubMed]
10. Nangia-Makker P, Balan V, Raz A. Regulation of tumor progression by extracellular galectin-3. Cancer Microenviron. 2008;1:43–51. [PMC free article] [PubMed]
11. Nangia-Makker P, Raz T, Tait L, Hogan V, Fridman R, Raz A. Galectin-3 cleavage: a novel surrogate marker for matrix metalloproteinase activity in growing breast cancers. Cancer Res. 2007;67:11760–11768. [PMC free article] [PubMed]
12. Nangia-Makker P, Wang Y, Raz T, Tait L, Balan V, Hogan V, et al. Cleavage of galectin-3 by matrix metalloproteases induces angiogenesis in breast cancer. Int J Cancer. 2010;127:2530–2541. [PMC free article] [PubMed]
13. Balan V, Nangia-Makker P, Kho DH, Wang Y, Raz A. Tyrosine-phosphorylated galectin-3 protein is resistant to prostate-specific antigen (PSA) cleavage. J Biol Chem. 2012;287:5192–5198. [PMC free article] [PubMed]
14. Shen Y, Liu T, Tolić N, Petritis BO, Zhao R, Moore RJ, et al. Strategy for degradomic-peptidomic analysis of human blood plasma. J Proteome Res. 2010;9:2339–2346. [PMC free article] [PubMed]
15. López-Otín C, Overall CM. Protease degradomics: a new challenge for proteomics. Nat Rev Mol Cell Biol. 2002;3:509–519. [PubMed]
16. Wang Y, Nangia-Makker P, Tait L, Balan V, Hogan V, Pienta KJ, et al. Regulation of prostate cancer progression by galectin-3. Am J Pathol. 2009;174:1515–1523. [PubMed]
17. Sturge J, Caley MP, Waxman J. Bone metastasis in prostate cancer: emerging therapeutic strategies. Nat Rev Clin Oncol. 2011;8:357–368. [PubMed]
18. Onishi T, Hayashi N, Theriault RL, Hortobagyi GN, Ueno NT. Future directions of bone-targeted therapy for metastatic breast cancer. Nat Rev Clin Oncol. 2010;7:641–651. [PubMed]
19. Smith MR, Saad F, Coleman R, Shore N, Fizazi K, Tombal B, et al. Denosumab and bone-metastasis-free survival in men with castration-resistant prostate cancer: results of a phase 3, randomised, placebo-controlled trial. Lancet. 2012;379:39–46. [PMC free article] [PubMed]
20. Edwards JR, Mundy GR. Advances in osteoclast biology: old findings and new insights from mouse models. Nat Rev Rheumatol. 2011;7:235–243. [PubMed]
21. Boyle WJ, Simonet WS, Lacey DL. Osteoclast differentiation and activation. Nature. 2003;423:337–342. [PubMed]
22. Niida S, Amizuka N, Hara F, Ozawa H, Kodama H. Expression of Mac-2 antigen in the preosteoclast and osteoclast identified in the op/op mouse injected with macrophage colony-stimulating factor. J Bone Miner Res. 1994;9:873–881. [PubMed]
23. Seetharaman J, Kanigsberg A, Slaaby R, Leffler H, Barondes SH, Rini JM. X-ray crystal structure of the human galectin-3 carbohydrate recognition domain at 2.1-A resolution. J Biol Chem. 1998;273:13047–13052. [PubMed]
24. Mehra R, Kumar-Sinha C, Shankar S, Lonigro RJ, Jing X, Philips NE, et al. Characterization of bone metastases from rapid autopsies of prostate cancer patients. Clin Cancer Res. 2011;17:3924–3932. [PMC free article] [PubMed]
25. Harazono Y, Kho DH, Balan V, Nakajima K, Hogan V, Raz A. Extracellular galectin-3 programs multidrug resistance through Na+/K+-ATPase and P-glycoprotein signaling. Oncotarget. 2015;6:19592–19604. [PMC free article] [PubMed]
26. Benjamini Y, Yekutieli D. The control of the false discovery rate in multiple testing under dependency. Ann Statist. 2001;29:1165–1188.
27. Kanda Y. Investigation of the freely-available easy-to-use software “EZR” (Easy R) for medical statistics. Bone Marrow Transplant. 2013;48:452–458. [PMC free article] [PubMed]
28. Kim Y, Nizami S, Goto H, Lee FY. Modern interpretation of giant cell tumor of bone: predominantly osteoclastogenic stromal tumor. Clin Orthop Surg. 2012;4:107–116. [PMC free article] [PubMed]
29. Li YJ, Kukita A, Teramachi J, Nagata K, Wu Z, Akamine A, et al. A possible suppressive role of galectin-3 in upregulated osteoclastogenesis accompanying adjuvant-induced arthritis in rats. Lab Invest. 2009;89:26–37. [PubMed]
30. US National Library of Medicine. 2013 [online],
31. US National Library of Medicine. 2007 [online],
32. Nishida T, Emura K, Kubota S, Lyons KM, Takigawa M. CCN family 2/connective tissue growth factor (CCN2/CTGF) promotes osteoclastogenesis via induction of and interaction with dendritic cell-specific transmembrane protein (DC-STAMP) J Bone Miner Res. 2011;26:351–363. [PMC free article] [PubMed]
33. Maruyama K, Uematsu S, Kondo T, Takeuchi O, Martino MM, Kawasaki T, et al. Strawberry notch homologue 2 regulates osteoclast fusion by enhancing the expression of DC-STAMP. J Exp Med. 2013;210:1947–1960. [PMC free article] [PubMed]
34. Grigoriadis AE, Wang ZQ, Cecchini MG, Hofstetter W, Felix R, Fleisch HA, et al. c-Fos: a key regulator of osteoclast-macrophage lineage determination and bone remodeling. Science. 1994;266:443–448. [PubMed]
35. Takayanagi H. Induction and activation of the transcription factorNFATc1 (NFAT2) integrate RANKL signaling for terminal differentiation of osteoclasts. Dev Cell. 2002;3:889–901. [PubMed]
36. Matsumoto M. Essential role of p38 mitogen-activated protein kinase in cathepsin K gene expression during osteoclastogenesis through association of NFATc1 and PU.1. J Biol Chem. 2004;279:45969–45979. [PubMed]
37. Minkin C. Bone acid phosphatase: tartrate-resistant acid phosphatase as a marker of osteoclast function. Calcif Tissue Int. 1982;34:285–290. [PubMed]
38. Saftig P, Hunziker E, Wehmeyer O, Jones S, Boyde A, Rommerskirch W, et al. Impaired osteoclastic bone resorption leads to osteopetrosis in cathepsin-K-deficient mice. Proc Natl Acad Sci U S A. 1998;95:13453–13458. [PubMed]
39. Kukita T, Wada N, Kukita A, Kakimoto T, Sandra F, Toh K, et al. RANKL-induced DC-STAMP is essential for osteoclastogenesis. J Exp Med. 2004;200:941–946. [PMC free article] [PubMed]
40. McMichael BK, Wysolmerski RB, Lee BS. Regulated proteolysis of nonmuscle myosin IIA stimulates osteoclast fusion. J Biol Chem. 2009;284:12266–12275. [PMC free article] [PubMed]
41. Kong YY, Feige U, Sarosi I, Bolon B, Tafuri A, Morony S, et al. Activated T cells regulate bone loss and joint destruction in adjuvant arthritis through osteoprotegerin ligand. Nature. 1999;402:304–309. [PubMed]
42. Zhang J, Dai J, Qi Y, Lin DL, Smith P, Strayhorn C, et al. Osteoprotegerin inhibits prostate cancer-induced osteoclastogenesis and prevents prostate tumor growth in the bone. J Clin Invest. 2001;107:1235–1244. [PMC free article] [PubMed]
43. Dalton P, Christian HC, Redman CW, Sargent IL, Boyd CA. Membrane trafficking of CD98 and its ligand galectin 3 in BeWo cells--implication for placental cell fusion. FEBS J. 2007;274:2715–2727. [PubMed]
44. Helming L, Gordon S. Molecular mediators of macrophage fusion. Trends Cell Biol. 2009;19:514–522. [PubMed]
45. Aguilar PS, Baylies MK, Fleissner A, Helming L, Inoue N, Podbilewicz B, et al. Genetic basis of cell-cell fusion mechanisms. Trends Genet. 2013;29:427–437. [PMC free article] [PubMed]
46. Cui W, Cuartas E, Ke J, Zhang Q, Einarsson HB, Sedgwick JD, et al. CD200 and its receptor, CD200R, modulate bone mass via the differentiation of osteoclasts. Proc Natl Acad Sci U S A. 2007;104:14436–14441. [PubMed]
47. Mbalaviele G, Chen H, Boyce BF, Mundy GR, Yoneda T. The role of cadherin in the generation of multinucleated osteoclasts from mononuclear precursors in murine marrow. J Clin Invest. 1995;95:2757–2765. [PMC free article] [PubMed]
48. Vicente-Manzanares M, Ma X, Adelstein RS, Horwitz AR. Non-muscle myosin II takes centre stage in cell adhesion and migration. Nat Rev Mol Cell Biol. 2009;10:778–790. [PMC free article] [PubMed]
49. Lakshminarayan R, Wunder C, Becken U, Howes MT, Benzing C, Arumugam S, et al. Galectin-3 drives glycosphingolipid-dependent biogenesis of clathrin-independent carriers. Nat Cell Biol. 2014;16:595–606. [PubMed]
50. Dong S, Hughes RC. Macrophage surface glycoproteins binding to galectin-3 (Mac-2-antigen) Glycoconj J. 1997;14:267–274. [PubMed]
51. Roudier MP, Morrissey C, True LD, Higano CS, Vessella RL, Ott SM. Histopathological assessment of prostate cancer bone osteoblastic metastases. J Urol. 2008;180:1154–1160. [PMC free article] [PubMed]
52. Hamaoka T, Madewell JE, Podoloff DA, Hortobagyi GN, Ueno NT. Bone imaging in metastatic breast cancer. J Clin Oncol. 2004;22:2942–2953. [PubMed]
53. Kim GE, Lee JS, Choi YD, Lee KH, Lee JH, Nam JH, et al. Expression of matrix metalloproteinases and their inhibitors in different immunohistochemical-based molecular subtypes of breast cancer. BMC Cancer. 2014;14:959. [PMC free article] [PubMed]
54. Kikuchi E, Nakashima J, Ishibashi M, Ohigashi T, Oya M, Nakagawa K, et al. Usefulness of alpha1-antichymotrypsin-PSA complex for predicting bone metastases of prostate cancer. Urology. 2007;68:371–375. [PubMed]
55. Brown JE, Coleman RE. Denosumab in patients with cancer-a surgical strike against the osteoclast. Nat Rev Clin Oncol. 2012;9:110–118. [PubMed]
56. Mathews KP, Konstantinov KN, Kuwabara I, Hill PN, Hsu DK, Zuraw BL, et al. Evidence for IgG autoantibodies to galectin-3, a beta-galactoside-binding lectin (Mac-2, epsilon binding protein, or carbohydrate binding protein 35) in human serum. J Clin Immunol. 1995;15:329–337. [PubMed]