A Wnt-Independent LGR4–EGFR Signaling Axis in Cancer Metastasis

LGR4/5/6 proteins form a subclass of the glycoprotein hormone receptor family that has crucial functions in embryonic development, adult tissue homeostasis, and various diseases (1). Mice null for Lgr4 or Lgr5 die during embryogenesis (2) or at the neonatal stage (3), although Lgr6 null mice are viable, likely due to overlapping functions provided by Lgr4/5 (4). Nonsense mutations of LGR4 in humans are associated with multiple diseases in development including osteoporosis (5). LGR4 is essential for stem cell maintenance (6), whereas LGR5 and LGR6 are selectively expressed on stem cells in various adult tissues (4, 7–10). LGR4/5/6 also play important roles in cancer initiation and progression. LGR4 is highly expressed in multiple types of cancer and is associated with poor patient outcome (11–14). LGR4 promotes tumorigenesis and metastasis and modulates cancer stem cells, whereas LGR5 and LGR6 mark cancer stem cells and progenitor cells that contribute to tumor initiation (9, 15–18).

The biological functions of LGR4/5/6 have been primarily attributed to their roles in potentiating Wnt signaling, which requires the binding of their four R-spondin ligands (RSPO1–RSPO4; refs. 19–22). The resulting RSPO–LGR complexes bind to the transmembrane E3 ubiquitin ligases RNF43/ZNRF3 of the Wnt receptors Frizzled and coreceptors LRP5/6, causing autoubiquitination and membrane clearance of these E3 ligases, thereby stabilizing Wnt receptors and amplifying Wnt signaling (23, 24). In addition, RSPO–LGR complexes can recruit the scaffold protein IQGAP1 to promote LRP5/6 phosphorylation and regulate actin dynamics to potentiate Wnt signaling (25, 26).

We have previously reported that LGR4 has critical functions in mammary gland development and breast cancer initiation and progression (13, 27). While studying LGR4 in breast cancer metastasis, we found evidence that LGR4 may control more than Wnt signaling to promote metastasis. First, LGR4 can promote breast cancer cell migration and invasion without ligand stimulation (Wnt and RSPO; ref. 13). Second, emerging evidence has suggested that Wnt signaling may have only marginal or insignificant effects on metastasis. Despite being potent in inducing mammary tumors in mice, transgenic expression of Wnt ligands or stabilized β-catenin rarely leads to metastatic tumors (28–32). Also, Wnt pathway inhibitors failed to inhibit breast cancer metastasis in a mouse model of breast cancer (33). These observations suggest that LGR4/5/6 may exert their biological functions by regulating more than Wnt signaling; however, the direct evidence for Wnt-independent functions of LGR4/5/6 in development or cancer has still been missing. Here, we demonstrate a Wnt-independent role of LGR4 in promoting breast cancer metastasis. We show that blockade of Wnt signaling by either Wnt inhibitors or genetic depletion of Wntless (which encodes the essential receptor for Wnt intracellular trafficking and production) cannot abolish the effect of LGR4 on breast cancer cell metastatic ability. LGR4 mutants that are uncoupled from Wnt signaling can still promote breast cancer cell migration and invasion in vitro and breast cancer metastasis in vivo, as potently as wild-type (WT) LGR4. Mechanistically, LGR4 enhances EGFR signaling to promote breast cancer metastasis. LGR4 interacts with EGFR and prevents EGFR ubiquitination and degradation, thereby enhancing EGFR signaling.

MDA-MB-231, MDA-MB-468, HCC1954, MCF7, T-47D, and HEK293T cells were purchased from ATCC. SUM159 and SUM149 cells were provided by X. H.-F. Zhang; BT549, H1299, and PC3 cells were provided by X. Lin. MDA-MB-231, MDA-MB-468, MCF7, BT549, PC3, and HEK293T cells were cultured in DMEM (Corning, 10-013-CV) with 10% FBS and 1% penicillin/streptomycin. SUM159 and SUM149 cells were cultured in Ham's F12 nutrient mixture (Gibco, 11765054) with 5% FBS, 10 mmol/L HEPES, 1 μg/mL hydrocortisone, 5 μg/mL insulin, and 1% penicillin/streptomycin. HCC1954, T-47D, and H1299 cells were cultured in RPMI-1640 (Corning, 10–040-CV) with 10% FBS and 1% penicillin/streptomycin. All cell lines were tested Mycoplasma negative with MycoAlert Mycoplasma Detection kit (Lonza) before the experiment and passed within 20 passages after thawing.

LGR4 D137F and LGR4 D161F-mutant constructs were generated using the QuikChange Multi Site-Directed Mutagenesis Kit (Stratagene, 200515). Lentiviral expression plasmids for LGR4 WT, D137F, and D161F were constructed by subcloning into FUCGW vector by TA cloning (Invitrogen, 450641). Lentiviral expression plasmids for EGFR and LGR4 were constructed by subcloning into pBobi vector by In-Fusion (TaKaRa, 639642). pcDNA-Wnt3a was purchased from Addgene (35908).

For CRISPR constructs, HEK293T cells were transfected with the lentiviral transfer vector together with psPAX2 and pMD.2G. For pCMV-Tet-On, pWPT-Fluc-RFP (34), FUCGW, or pBobi constructs, HEK293T cells were transfected with the lentiviral transfer vector, and pCMV-VSVG, pMDLg/pRRE, pRSV-Rev. Viral supernatants were collected 48 hours after transfection, filtered through a 0.45-μm filter (Millipore, SLHV033RS). Target cells were infected by spin-down for 90 minutes at 2,000 rpm at 35°C in virus-containing medium supplemented with 2 μg/mL polybrene (Santa Cruz, sc-134220) and then incubated for additional 2 to 3 hours at 37°C before replacing virus-containing medium with the desired culture medium.

CRISPR guide sequences were designed by http://crispr.mit.edu and cloned into lentiCRISPR v2 (Addgene 52961). Sequences were as follows: sgLGR4_1: 5′-CTGCGACGGCGACCGTCGGG-3′; sgLGR4_2: 5′-GGGCTGACGGCCGTGCCCGA-3′; sgLGR4_3: 5′-TACCCAGTGAAGCCATTCGA-3′; and sgWLS: 5′-CTACATGTCGGTGAAATGTG-3′. MDA231 or MDA468 cells were infected with lentivirus carrying CRISPR constructs or vector, and then selected in the culture medium containing 2 μg/mL puromycin (InvivoGen, ant-pr-1). Pooled MDA231 or MDA468 LGR4 knockout (KO) cells were characterized by Western blot analysis. A single colony of MDA231 WLS KO cells was picked and validated by sequencing.

Stable MDA231 or SUM159 cell lines with doxycycline (Dox; 1 μg/mL, Sigma-Aldrich, D9891)-inducible expression of LGR4 were established and selected by puromycin (2 μg/mL). Inducible expression of LGR4 was validated by Western blot analysis.

The in vitro metastatic ability of cells was measured by transwell assay. Cells were serum-starved for 16 hours before plating into Falcon cell culture inserts (Corning, 353097) for migration assay or BioCoat Matrigel invasion chambers (Corning, 354480) for invasion assay. Cells were resuspended in serum-free medium and migrated/invaded into complete culture medium with fetal bovine serum. After the non-migrated/invaded cells were removed, the inserts/chambers were washed in PBS, fixed in 100% methanol, stained with 0.5% crystal violet, washed in distilled H₂O, air-dried, and imaged. The migrated/invaded cells were counted using ImageJ (NIH). Fold change was calculated by the number of migrated/invaded cells relative to the control condition. The 2×10⁴ SUM159 cells were seeded and incubated 24 hours for migration and invasion assay. For other migration and invasion assays, 5×10⁴ cells were seeded and incubated for 6 or 8 hours for migration assays or 24 hours for invasion assays prior to fixation and quantification. At least three independent experiments were performed.

For EGFR knockdown, two separate EGFR siRNA were purchased from Thermo Fisher Scientific (siRNA ID 43833) and Santa Cruz (sc-29301). Control siRNA was purchased from Thermo Fisher Scientific (AM4611). siRNAs were transfected into cells using oligofectamine (Invitrogen, 12252011). Cells were starved in the serum-free medium 24 hours after transfection and proceeded for transwell assay. LGR4 knockdown experiments were performed as previously described (13).

Cells were transfected with Top-Flash (Addgene, 12456) and Renilla-TK (Promega, E2241) plasmids. Luciferase activity was measured using the Dual-Luciferase Reporter Assay System (Promega, E1960) 72 hours after transfection in at least duplicate in three independent experiments. Luminescence data are represented as the firefly luminescence normalized to the Renilla luminescence and then relative to the control condition.

Cells were lysed with TRIzol reagent (Invitrogen, 15596-026). Total RNA was extracted by chloroform and isopropanol precipitation. cDNA was obtained using the SuperScript III First-Strand Synthesis System (Invitrogen, 18080-051). qPCR analyses were performed with primers listed below using SsoAdvanced Universal SYBR Green Supermix (Bio-Rad, 1725270) in at least duplicate in three independent experiments. Plotted are data normalized to ACTB and relative to the control. The following primers were used:

• AXIN2: 5′-TAGGTTCTGGCTATGTCTTTG -3′, 5′-GTATCGTCTGCGGGTCTT-3′;

• ACTB: 5′-ACTCTTCCAGCCTTCCTTCC-3′, 5′-CAGTG ATCTCCTTCTGCATCC-3′;

• CD44: 5′-TGGCACCCGCTATGTCGAG-3′; 5′-GTAGCAGGGATTCTGTCTG-3′;

• CCND1: 5′-ATGTTCGTGGCCTCTAAGATGA-3′, 5′-CAGGTTCCACTTGAGCTTGTTC-3′;

• EGFR: 5′-TGCCCATGAGAAATTTACAGG-3′, 5′-ATGTTGCTGAGAAAGTCACTGC-3′;

• MYC: 5′-CGTCTCCACACATCAGAGCACAA-3′, 5′-TCTTGGCAGCAGGATAGTCCTT-3′;

• SOX2: 5′-TACAGCATGTCCTACTCGCAG-3′, 5′-GAGGAAGAGGTAACCACAGGG-3′.

NSG mice and athymic nude mice were purchased from Jackson Laboratories and Charles River. All mouse experiments were approved by the Baylor College of Medicine Animal Care and Use Committee.

For subcutaneous injection, 3 × 10⁶ MDA231 cells suspended in 0.1 mL serum-free DMEM were injected into the left back or right back flank of 14- or 17-week-old female NSG mice. Tumors were detectable 10 days after injection and were measured with a caliper twice a week. The growing tumor size was calculated by length × width²/2. The mice were euthanized 36 days after injection. Tumors were weighed. Intraductal injections were performed as previously described (35). 8 × 10⁴ luciferase-labeled MDA231 cells in 4 μL of 0.1% Trypan Blue in PBS containing 2% FBS were injected into the nipple of the fourth mammary gland of 9-week-old NSG mice. The bioluminescence was measured once a week. The mice were euthanized 9 weeks after injection. Tumors were weighed.

For the tail-vein injection model, 1×10⁶ luciferase-labeled MDA231 cells suspended in 0.1 mL PBS were injected into 12-week-old female NSG mice via the tail vein. Metastasis was monitored by bioluminescence imaging (BLI) of live animals once a week. Bone metastasis was recorded when the bioluminescence signals around the hind limb area reached above 1×10⁴. Hind limb bones were extracted immediately after euthanasia for ex vivo imaging and then for X-ray imaging. Lung and liver tissues were imaged under fluorescent dissection microscope and fluorescence intensity was quantified using ImageJ. For the tail-vein injection to compare lung metastases via hematoxylin and eosin (H&E) and qPCR, 5×10⁵ luciferase-labeled MDA231 cells suspended in 0.1 mL PBS were injected into 12-week-old female NSG mice via the tail vein. Five weeks after injection, the mice were euthanized and the whole lung tissues were collected for qPCR analyses. Intra–iliac artery (IIA) injections were performed as previously described (36). 1×10⁵ luciferase-labeled MDA231 cells suspended in 0.1 mL PBS were injected into 7-week-old female nude mice via the external iliac artery. Bone metastasis was monitored by BLI twice a week.

Bioluminescence was measured by injection of 100 μL 15 mg/mL D-luciferin (LUCNA-1G, Goldbio) via the intraorbital sinus. Mice were imaged once or twice a week using IVIS Lumina II (Advanced Molecular Vision). The acquired bioluminescence signals were normalized to the means of day 0 intensity within each group.

For Wnt-C59 treatment, cells were pretreated with 100 nmol/L Wnt-C59 (R&D Systems, 5148) for 48 hours before the experiment. For Wnt signaling stimulation experiments, cells were treated with 50 ng/mL human recombinant Wnt3a (R&D Systems, 5036-WN-010) or 200 ng/mL human recombinant RSPO2 (R&D Systems, 3266-RS-025) or a combination of Wnt3a and RSPO2 for 24 hours before sample collection. For EGF stimulation, cells were treated with 50 ng/mL human recombinant EGF (Invitrogen, PHG0311) for 10 minutes. To block proteasomal and lysosomal degradation, cells were pretreated with 10 μmol/L MG132 (Sigma-Aldrich, M8699) and 10 nmol/L bafilomycin A1 (Sigma-Aldrich, B1793) 4 hours before EGF stimulation. For blockade of EGFR activation, cells were treated with 5 μmol/L erlotinib (Selleck Chemicals, S7786) for 48 hours before sample collection.

Whole-cell protein lysates were prepared using Transmembrane Protein Extraction Reagent (FIVEPhoton Biochemicals, TmPER-200) according to the manufacturer's instructions, and their protein concentration were quantified using the BCA assay (Thermo Fisher Scientific, 23225). Cell lysates were mixed with Laemmli sample buffer (Bio-Rad, 1610747) and β-mercaptoethanol (Thermo Fisher Scientific, 21985023) before boiling for 5 minutes at 95°C. For LGR4 blotting, cell lysates were mixed with Laemmli sample buffer and incubated for 1 hour at 37°C. Equal amounts of protein lysates were loaded and run in 8% to 12% SDS-PAGE gel. Gels were transferred onto nitrocellulose membrane (Thermo Fisher Scientific, 88018) at 100 V for 90 minutes at 4°C. Membranes were then blocked with 5% nonfat milk in TBST at room temperature for 1 hour, incubated with primary antibodies overnight at 4°C and secondary antibodies at room temperature for 1 hour, and scanned using the Odyssey LI-COR imaging system. Primary antibodies used in the study included LGR4 (7E7) rat antibody (37), EGFR (D38B1) XP Rabbit antibody (Cell Signaling Technology, 4276), EGFR (A-10) mouse antibody (Santa Cruz, sc-373746), P-EGFR (Y1068) XP rabbit antibody (Cell Signaling Technology, 3777), MET mouse antibody (Cell Signaling Technology, 3127), HER2 (44E7) mouse antibody (Cell Signaling Technology, 2248), HA-tag mouse antibody (BioLegend 901501), β-catenin mouse antibody (BD Transduction Laboratories, 610153), Ub (P4D1) mouse antibody (Santa Cruz, sc-8017), CBL mouse antibody (Santa Cruz, sc-1651), and GAPDH rabbit antibody (Santa Cruz, sc-25778). Secondary antibodies used included anti-Rat IRDye 800CW (LI-COR, 926-32219), anti-mouse IRDye 680RD (LI-COR, 926-68070), and anti-Rabbit IRDye 800CW (LI-COR, 926-32211). All experiments were performed for at least three biological repeats.

Cells were lysed on ice using NP-40 lysis buffer supplemented with protease inhibitors (Sigma-Aldrich, P8340). Cell lysates were then centrifuged at 14,000 rpm for 15 minutes at 4°C. Supernatant was collected and the concentration of protein lysate was quantified by BCA. Protein lysates were incubated with either EGFR (Ab-13) Mouse antibody (Thermo Fisher Scientific, MS-609-P1) or LGR4 (8D2) Rat antibody (Q. Liu), and Protein G Sepharose beads (GE Healthcare, 17061801) overnight at 4°C. Beads were washed with NP-40 lysis buffer. Whole-cell lysates and immunoprecipitates were analyzed by Western blot analysis.

HEK293T cells were seeded on poly-D-lysine coated chamber slide (Corning, 354632) and transfected with empty vector, or constructs expressing LGR4 WT, D137F, D161F. Forty-eight hours after transfection, cells were incubated in culture medium with 5 μg/mL LGR4 (8D2) antibody at 37°C for 1 hour, fixed in 3.2% paraformaldehyde in PBS at room temperature for 15 minutes, permeabilized in 3% BSA/0.1% saponin (Sigma-Aldrich, 47036) in PBS at room temperature for 30 minutes, and then incubated in the secondary antibody anti-Rat Alexa Fluor 488 (Invitrogen, A11006) at room temperature for 1 hour. Afterward, the cells were stained with DAPI (Thermo Fisher Scientific, 62248) and mounted.

The Phospho Explorer antibody microarray, which was designed and manufactured by Full Moon Biosystems, Inc., contains 1,318 antibodies in duplicate. The array experiments were carried out according to the manufacturer's instructions.

The immunoprecipitation-mass spectrometry (IP-MS) experiments were performed as previously described (38). For each IP experiment, 1 mg protein lysate was incubated with 5 μg EGFR (Ab-13) antibody for 2 hours at 4°C and cleared by ultracentrifugation (100,000 × g, 15 minutes). The supernatant was then incubated with 30 μL 50% protein A-Sepharose slurry (GE Healthcare, 17-0780-01) for 1 hour at 4°C. The bead-bound complexes were washed 4 times with NETN buffer (20 mmol/L Tris pH7.5, 1 mmol/L EDTA, 0.5% NP-40, 150 mmol/L NaCl) and eluted with 20 μL 2× Laemmli buffer and heated at 95°C for 10 minutes. IP samples were resolved on a NuPAGE 10% Bis-Tris gel (Life Technologies, WG1201BX10) in 1× MOPS running buffer; the gel was cut into 3 molecular weight regions plus the IgG heavy and light chain bands. Each band was in-gel digested overnight with 100 ng of trypsin that cleaves peptide chains at the C-termini of lysine or arginine in 20 μL of 50 mmol/L NH₄HCO₃ at 37°C. Peptides were extracted with 350 μL of 100% acetonitrile and 20 μL of 2% formic acid, and dried in a Savant Speed-Vac. Digested peptides were dissolved in 10 μL of loading solution (5% methanol containing 0.1% formic acid) and subjected to LC-MS/MS assay as described previously (39).

EGFR-associated proteins were identified as those that showed a more than 10-fold increase in abundance (ratio of treated/nontreated), identified with more than two unique and strict peptides with more than 10⁶ iBAQ after EGF treatment in both MDA231 and MDA468 cell lines or in more than two gastric cell lines; additionally, proteins that appeared in the list in HeLa cells (38) were also included if they showed a more than 2-fold increase in abundance after EGF treatment.

Ten breast cancer tissue microarrays (TMA) containing 270 breast tumor specimens arrayed in duplicate from our Breast Center Tissue Bank were stained by IHC for LGR4 using 7E7 Rat antibody (37). The stained TMAs were scored for both proportion scores (0–100) and intensity scores (0–3). Valid scores from 235 tumors were used for the analysis. Kruskal–Wallis test and post hoc pairwise two-sample ranking tests were used to compare LGR4 levels between different subsets of breast cancer.

All statistical analyses in our study were conducted using GraphPad Prism, Winstat add-in for Excel, or SAS 9.4. For data comparison between two experimental groups, two-tailed t test without equal variation assumption was used. For data comparison with at least three groups, one-way ANOVA or Kruskal–Wallis test was performed first to assess the overall difference among groups. If differences existed, Dunnett test or LSD test was performed to assess the significance of differences between two groups. Two-way repeated-measures ANOVA or mixed-effects analysis and uncorrected Fisher LSD test were used to assess the difference between data sets with time-series measurements, including growth curves of cell proliferation, metastasis BLI signals, or tumor sizes. The log-rank test was used for survival analyses of mice. P values less than 0.05 were considered as statistically significant.

All animal studies were approved by the Institutional Animal Care and Use Committee of Baylor College of Medicine.

LGR4 mRNA is more highly expressed in triple-negative breast cancer (TNBC) than in the other two subtypes (ER⁺ and HER2⁺) based on The Cancer Genome Atlas (TCGA; Supplementary Fig. S1A). IHC staining for LGR4 protein in TMAs of 223 cases of breast cancer in our Breast Center Tissue Bank also detected higher levels of LGR4 in TNBC than the other two subtypes (Supplementary Fig S1B). Therefore, we focused our work on this subtype of breast cancer. We first took a bioinformatic approach and found that high LGR4 expression was associated with poor metastasis-free survival in TNBC patients in the EMC-MSK data set (Fig. 1A; refs. 40, 41); however, a well-validated Wnt signature (42) failed to correlate with metastasis-free survival in this same TNBC cohort (Fig. 1B). These bioinformatic data suggest divergent functions of LGR4 versus Wnt signaling in TNBC progression. Furthermore, we found that LGR4 mRNA has only very modest correlation with the Wnt signature in TCGA breast cancers in general and in basal-like breast cancer (a molecular subtype overlapping with TNBC) specifically, and that many LGR4-high cases do not even exhibit increased Wnt signaling (Fig. 1C and D). Based on these premises, we hypothesized that LGR4 engages a Wnt-independent mechanism to promote breast cancer metastasis.

To test this hypothesis, we first investigated whether LGR4 promotes breast cancer cell migration and invasion in the absence of Wnt signaling activation using cell line–based in vitro assays. MDA-MB-231 (MDA231) and MDA-MB-468 (MDA468) exhibited relatively high levels of LGR4 among the TNBC cell lines that we examined (Supplementary Fig. S2A). Thus, we used CRISPR/Cas9 to knock out LGR4 in these two TNBC cell lines (Supplementary Fig. S2B). LGR4 KO did not affect the proliferation of MDA231 cells (Supplementary Fig. S2C) and had no significant influence on the proliferation of MDA468 within the first 48 hours of plating (Supplementary Fig. S2D). However, LGR4 KO inhibited migration and invasion of both cell lines in transwell assays (within a 24-hour window; Fig. 1E and Supplementary Fig. S2E), consistent with our previous report using shRNA knockdown of LGR4 (13). In our migration and invasion assays, we did not add Wnt or RSPO to the culture medium, which are needed for LGR proteins to potentiate Wnt signaling (19–22); therefore, Wnt signaling was already at low baseline levels in the LGR4WT cells, and thus any potential weakening may not explain the significant migration and invasion defects of these LGR4KO cells. When quantified by the Top-Flash luciferase report assay (Fig. 1F and Supplementary Fig. S2F) or qPCR analysis for Wnt target genes (Fig. 1G and Supplementary Fig. S2G), the baseline Wnt signaling activity did not even show any detectable drop in LGR4KO cells. We validated that ligand (Wnt3a plus RSPO2)-stimulated Wnt signaling was indeed compromised in these KO cells (Supplementary Fig. S2H). Next, we generated SUM159 and MDA231 cell lines with inducible overexpression of LGR4 for gain-of-function studies (Supplementary Fig. S3A). Inducible expression of LGR4 did not have any impact on the proliferation of these two cell lines (Supplementary Fig. S3B and S3C), but promoted their migration and invasion (Fig. 1H and Supplementary Fig. S3D). These experiments were also performed in the absence of any added Wnt or RSPO and did not influence the Top-Flash reporter activity or the expression level of the Wnt target gene AXIN2 (Fig. 1I and J; Supplementary Fig. S3E and S3F). We also validated that addition of Wnt3a and RSPO2 stimulated Wnt signaling more in LGR4-induced MDA231 than in noninduced cells (Supplementary Fig. S3G). Unexpectedly, the same treatment inhibited Wnt signaling in LGR4-induced SUM159 than in noninduced cells (Supplementary Fig. S3G), presumably due to cell contextual variations. Together, our experiments performed in the absence of added Wnt and RSPO ligands showed that LGR4 KO inhibited, whereas LGR4 overexpression promoted, the migration and invasion of breast cancer cells without engaging the canonical Wnt signaling.

We next investigated whether blocking Wnt signaling suppresses migration and invasion of TNBC cell lines. Wnt-C59 (43) and LGK974 (44) are small molecular inhibitors of Wnt signaling as they suppress porcupine, which is required for Wnt palmitoylation and secretion (45, 46). Wnt-C59 diminished Wnt signaling activity in MDA231 cells (Fig. 2A), as expected, but did not affect cell migration or invasion (Fig. 2B). Importantly, in an IIA injection model of bone metastasis in nude mice (36), LGK974 treatment failed to inhibit MDA231 bone metastasis (Fig. 2C), which is in line with the previous report that the Wnt inhibitor XAV939 did not reduce breast cancer metastasis in a spontaneous metastasis mouse model (33). Together, these data suggest that blocking Wnt signaling does not impede TNBC metastasis; thus, Wnt signaling may not mediate LGR4 promotion of breast cancer metastasis. To directly test whether Wnt activity is indeed dispensable for LGR4 promotion of cancer cell progression, we used three independent approaches. First, we used Wnt-C59 to inhibit Wnt signaling in MDA231 cells stably overexpressing LGR4 (MDA231-LGR4; Fig. 2D), and we found that this treatment failed to block ectopic LGR4 from promoting migration and invasion (Fig. 2E). As Wnt-C59 may not completely abolish the secretion of Wnt ligands (47, 48), we next eliminated all Wnt signaling activity by using CRISPR/Cas9 to knock out Wntless (WLS) in MDA231 cells (Fig. 2F), which encodes the receptor essential for the intracellular trafficking and production of Wnt (49, 50). We verified WLS KO by sequencing, and we confirmed Wnt signaling deactivation—WLS KO cells failed to initiate a response to Wnt3a transient transfection whereas MDA231 WT cells responded potently (Fig. 2F). WLS KO caused MDA231 cells to become more spindle-shaped but did not weaken LGR4 promotion of migration and invasion (Fig. 2G). Furthermore, we introduced a D137F or D161F mutation into LGR4 to uncouple this protein from Wnt signaling, as these conserved amino acid residues are required for LGR4/5/6 to bind an RSPO ligand to potentiate Wnt signaling (51). These two mutants were comparable to WT LGR4 in subcellular localization and expression levels (Supplementary Fig. S4A). As expected, these two mutants failed to enhance Wnt signaling in HEK293T cells (Supplementary Fig. S4B) and failed to potentiate RSPO2-stimulated Wnt signaling in MDA231 cells (Fig. 3A), but they nonetheless promoted migration and invasion of MDA231 and SUM159 as effectively as WT LGR4 in transwell assays (Fig. 3B and Supplementary Fig. S4C). Moreover, we confirmed these data in vivo: In both subcutaneous and intraductal models of primary tumor growth, MDA231 cells carrying LGR4-WT or LGR4-D137F or LGR4-D161F exhibited no discernible advantage over vector control MDA231 cells (Supplementary Fig. S4D–S4G); however, upon tail-vein injection into NSG mice, these three groups of LGR4-expressing cells all resulted in more extensive metastases in multiple organs than the vector control MDA231 cells based on bioluminescence (Fig. 3C and D). Specifically, they caused stronger tumor cell load in both the lungs and liver based on ex vivo imaging (Fig. 3E–G). The lung metastases were also confirmed by H&E staining of lung sections, and the difference was verified by qPCR for the provirus (Supplementary Fig. S4H and S4I). These LGR4-expressing cells also led to significantly worse bone metastasis-free survival than vector control MDA231 cells (Fig. 3H). Ex vivo bioluminescence and X-ray confirmed bone metastases (Fig. 3I) and more extensive spread to the bone in all three LGR4 groups (Fig. 3J). To further validate the Wnt-independent role of LGR4 in bone metastasis, we compared LGR4-D137F with LGR4-WT in the IIA model of bone metastasis and found that LGR4-D137F was as potent as LGR4-WT in promoting MDA231 bone metastasis (Fig. 3K and L). Taken together, we conclude that LGR4 promotes breast cancer cell migration, invasion, and distant metastasis without dependence on Wnt signaling.

To elucidate the Wnt-independent mechanisms by which LGR4 regulates breast cancer metastasis, we used Phospho Explorer Antibody Array (Full Moon Biosystems) to screen for the proteins and phosphoproteins that were altered by shRNA knockdown of LGR4 in MDA231 cells. We also mined the TCGA data set (385 breast cancer samples) to find proteins that were associated with LGR4 expression. These two screens converged on 13 proteins (Fig. 4A; Supplementary Fig. S5A; Data Set S1). All five phosphorylated proteins, including p-EGFR, p-Raf, p-Src, p-Fyn, and p-p70S6K, were members of the EGFR signaling pathway. We confirmed by immunoblotting that phospho-EGFR levels were downregulated in MDA231-LGR4KO and MDA468-LGR4KO cells in the absence of exogenous EGF (Supplementary Fig. S5B). We also found that upon EGF stimulation, these KO cell lines exhibited reduced levels of phospho-EGFR and total EGFR protein (Fig. 4B and Supplementary Fig. S5B). Conversely, LGR4 overexpression enhanced both phospho-EGFR and total EGFR protein levels in MDA231 and SUM159 (Fig. 4C and Supplementary Fig. S5C) as well as in the ER⁺ breast cancer cell line T47-D (Supplementary Fig. S5D and S5E). Our finding of LGR4-regulated EGFR level was not limited to breast cancer, as we found that LGR4 overexpression enhanced EGFR protein levels in the lung cancer cell line H1299 (Supplementary Fig. S5F) and the prostate cancer cell line PC3 (Supplementary Fig. S5G). Furthermore, we confirmed that LGR4 also regulates EGFR signaling pathway. The levels of phospho-Src and phospho-FAK were decreased in MDA468-LGR4KO cells, whereas the levels of total Src and FAK remained unchanged (Fig. 4D); and the converse effect was observed in H1299 cells overexpressing LGR4 (Supplementary Fig. S5H), which is consistent with a previous report that Lgr4 deficiency inhibited Src–FAK signaling in mouse PyMT mammary tumors (13). To determine whether LGR4 affects other receptor tyrosine kinases, we examined MET and HER2, which is a close family member of EGFR, also known to be important in breast cancer. We found that LGR4 overexpression had no impact on the level of the receptor tyrosine kinase MET in MDA231 cells (Supplementary Fig. S5I) but enhanced the protein level of HER2 in the HER2⁺ breast cancer cell line HCC1954 (Fig. 4E). Together, these data suggest that LGR4 regulates the level and activity of EGFR and its family members. In agreement, EGFR was previously reported to be downregulated in tissues of Lgr4 KO mice (52, 53).

To define the importance of EGFR in mediating LGR4-enhanced cell metastatic ability, we overexpressed EGFR in MDA231-LGR4KO and WT cells (Supplementary Fig. S6A). We found that this overexpression restored the potency of migration and invasion to MDA231-LGR4KO cells, although this ectopic EGFR caused only modest impact on migration and invasion of WT cells, probably due to higher baseline levels of EGFR in these WT cells than in KO cells (Fig. 5A). Next, we used two separate EGFR siRNAs to knock down EGFR in the MDA231-LGR4 and vector control cells (Supplementary Fig. S6B). This knockdown attenuated LGR4-induced cell migration and invasion (Fig. 5B). (Of note, LGR4 overexpression in MDA231 parental cells had a stronger impact on migration and invasion than EGFR overexpression probably due to additional molecules regulated by LGR4 that may also play a role in cancer progression.) Besides this genetic approach, we also used a pharmacological method. The small-molecule EGFR inhibitor erlotinib attenuated EGFR phosphorylation (Supplementary Fig. S6C), and blocked LGR4-induced increase of migration and invasion in transwell assays (Fig. 5C). Collectively, these results indicate that EGFR activity is required for LGR4 to promote breast cancer progression.

To determine whether LGR4 regulation of EGFR depends on Wnt signaling activation or not, we first asked whether manipulation of Wnt signaling could regulate EGFR in these TNBC cells. Wnt signaling components have been reported to regulate phospho-EGFR levels in β-catenin transgenic mouse livers (54) and in pan-DVL siRNA–treated HER2⁺ breast cancer cell lines (55), but we found that neither Wnt signaling inhibition by Wnt-C59 nor Wnt signaling activation by exogenous ligands had any effect on phospho-EGFR or total EGFR protein levels in MDA231 cells (Fig. 6A and B). Next, we tested whether manipulation of Wnt signaling could affect the impact of LGR4 on EGFR. Although Wnt-C59 diminished Wnt signaling (Supplementary Fig. S7A), it failed to block LGR4-induced EGFR activation in MDA231 cells (Fig. 6C). Conversely, although recombinant RSPO2 treatment enhanced Wnt signaling as expected (Supplementary Fig. S7B), it did not affect LGR4-induced EGFR activation in these cells (Fig. 6D). Furthermore, similar to WT LGR4, both LGR4-D137F and LGR4-D161F mutants still enhanced both phospho-EGFR and total EGFR levels in MDA231 cells (Fig. 6E) and in a non–small cell lung carcinoma cell line H1299 (Supplementary Fig. S7C). Together, these data suggest that LGR4 can activate EGFR via a Wnt-independent mechanism in multiple types of cancer.

To elucidate the Wnt signaling–independent mechanism by which LGR4 regulates EGFR activation, we first examined whether LGR4 regulates EGFR transcripts and found that LGR4 did not affect EGFR mRNA levels (Supplementary Fig. S8A and S8B). Then, we studied the effect of LGR4 on EGFR protein degradation by pretreatment with both the proteasome inhibitor MG132 and the lysosome inhibitor BAF to block the two protein degradation pathways. We found that blocking protein degradation restored both pEGFR and total EGFR levels in MDA231-LGR4KO cells under EGF stimulation (Fig. 7A). Conversely, we confirmed that blocking protein degradation resulted in equal amounts of EGFR in MDA231-LGR4 cells versus vector control cells (Fig. 7B). These data suggest that LGR4 regulates EGFR degradation. Because ubiquitination is the major posttranslational modification targeting EGFR for degradation (56, 57), we next examined whether LGR4 affected EGFR ubiquitination. We found higher levels of ubiquitinated EGFR in MDA231-LGR4-KO cells under EGF treatment than in the WT cells (Fig. 7C), suggesting that LGR4 functions to suppress the ubiquitination of EGFR. CBL is the most recognized E3 ligase that mediates EGFR ubiquitination and degradation (57). Therefore, we determined whether LGR4 affects CBL targeting of EGFR. Using coimmunoprecipitation, we found that LGR4 KO enhanced the levels of EGFR-bound CBL in MDA231-EGFR cells (Fig. 7D), suggesting that LGR4 functions to impair the EGFR–CBL interaction. Moreover, we identified LGR4 in anti-EGFR immunoprecipitates prepared from MDA468, MDA231, and BGC823 (a gastric cancer cell line) by mass spectrometry (Supplementary Fig. S8C; Data Set S2). We confirmed this LGR4–EGFR interaction using coimmunoprecipitation in MDA468-LGR4 cells (whose amplified EGFR allows for more readily detection of EGFR following immunoprecipitation): an anti-LGR4 antibody immunoprecipitated both LGR4 and EGFR whereas an anti-EGFR antibody brought down both EGFR and LGR4 (Fig. 7E). Furthermore, we validated the LGR4–EGFR interaction at the endogenous levels in MDA468 cells (Fig. 7F). Taken together, these data suggest that LGR4 interacts with EGFR and maintains EGFR levels by preventing CBL-mediated EGFR ubiquitination and degradation, thereby enhancing EGFR signaling to promote breast cancer metastasis.

The importance of LGR4/5/6 in development and cancer has been increasingly recognized. All known functions are primarily attributed to their binding to their ligands, the R-Spondins, to potentiate Wnt signaling. Although LGR4/5/6 share structural homology with seven-transmembrane domain G protein–coupled receptors, their G-protein signaling activity has not been firmly established (19, 21). In fact, Dr. Mingyao Liu and colleagues, who previously reported LGR4-mediated activation of the Gαs/PKA/CREB pathway (58), detected no impact of LGR4 genetic manipulations on the Gαs/PKA/CREB pathway in MDA231 cells and MMTV-PyMT transgenic mouse mammary tumors (13). A recent report suggests an LGR4–cAMP–PKA pathway in mediating Rspo1 stimulation of ERα expression in mouse mammary gland luminal cells (59); however, we found that LGR4 knockdown enhanced ERα in breast cancer cells (Data Set S1, Full Moon Antibody Array) and that the LGR4 mRNA level was negatively correlated with the ERα protein level in breast cancer patient samples (Data Set S1, TCGA data analyses), which is in line with the clinical data that LGR4 is expressed more highly in TNBC than in other breast tumors (Supplementary Fig. S1A and S1B). LGR4 has also been suggested in one report to interact with RANKL to inhibit RANK–TRAF6 signaling during osteoclast differentiation (60), and LGR5 has been reported to interact with TGFβ receptors to activate TGFβ signaling to suppress colon cancer metastasis (61). However, these and other reports of LGR4/5/6 research have not established Wnt-independent effects of LGR4/5/6. Here, we provide compelling evidence that LGR4 promotes cancer metastasis independently of Wnt signaling. We first demonstrated that LGR4 can promote migration and invasion of cancer cell cultures without any added RSPO or Wnt ligands. As RSPOs are predominantly supplied by stromal cells in vivo (62), our in vitro experiments without the addition of exogenous RSPO and Wnt ligands indicated a Wnt-independent effect of LGR4 on cancer metastasis. Next, we proved this concept by demonstrating that neither porcupine inhibitor nor WLS KO could block LGR4 from promoting migration, invasion, and metastasis in vivo, and by showing that LGR4 mutants that are disengaged from autocrine and paracrine Wnt signaling could nevertheless enhance migration, invasion, and metastasis in vivo. Therefore, our results firmly demonstrate a Wnt-independent function of LGR4 in promoting breast cancer cell migration, invasion, and distant metastasis, which is also supported by the clinical evidence that LGR4 expression, but not Wnt signature, is associated with breast cancer metastasis (Fig. 1A and B).

Using proteomics and bioinformatics screens, we identified EGFR as a crucial downstream component of Wnt-independent LGR4 signaling. This is in line with the previous reports that LGR4 regulates EGFR activity during embryonic eyelid development (52) and in granulosa-lutein cells (53) in Lgr4 KO mice. In these studies, the authors suggested that LGR4 regulated EGFR through matrix metalloproteinase 9 (MMP9)–mediated activation of heparin-binding EGF-like growth factor (HB-EGF) and that Mmp9 expression was transactivated by β-catenin. In our current study, we demonstrated that, independently of Wnt signaling, LGR4 interacts with EGFR and prevents CBL-mediated ubiquitination and degradation of EGFR, resulting in persistent EGFR signaling activation. It will be important in the future to investigate how LGR4 interacts with EGFR and prevents its ubiquitination by CBL.

Wnt signaling activation regulates multiple cellular processes including cell proliferation and stemness (63), and hyperactive Wnt signaling has also been reported to promote metastasis of several cancers including colorectal and pancreatic cancer (64, 65). However, we found that Wnt signaling manipulation did not affect MDA231 migration and invasion in vitro and metastasis in vivo, and we further demonstrated that Wnt signaling was not involved in LGR4-mediated breast cancer metastasis—neither stimulating nor blocking Wnt signaling affected LGR4-stimulated breast cancer migration and invasion in vitro and metastasis in vivo. In agreement, a well-validated Wnt signature (42) failed to correlate with metastasis-free survival of TNBC patients in the EMC-MSK data set (Fig. 1B), whereas higher LGR4 mRNA levels were associated with worse outcome (Fig. 1A). Our data are in line with those of several other studies that also failed to detect an association of Wnt signaling with human breast cancer metastasis, and with reports of failures of Wnt inhibitors to block breast cancer metastasis in mouse models (33, 40). These findings call into question attempts to prevent or treat metastasis of human breast cancer, and possibly some other cancers as well, by addressing Wnt activity. There are several clinical trials testing Wnt inhibitors in locally advanced or metastatic breast cancer (ClinicalTrials.gov Identifier: NCT01351103 and NCT01973309), but the efficacy of these Wnt inhibitors remains to be defined. It will be important in future studies to learn why Wnt signaling may promote metastasis in certain cancers but not in others.

In summary, we used multiple cell lines and several in vivo models to demonstrate Wnt-independent activation of EGFR as a crucial mediator of LGR4 promotion of breast cancer metastasis. We elucidated a novel LGR4 oncogenic pathway that enhances EGFR signaling independently of Wnt signaling. We uncovered the interaction between LGR4 and EGFR and demonstrated that LGR4 regulates EGFR ubiquitination and degradation. Our findings do not disprove the activation and functional significance of Wnt signaling in breast and other tissues in living beings where ligands may be abundant, but they do highlight a more complex signaling network emanating from LGR4.

T. Wang reports grants from NCI during the conduct of the study. S.G. Hilsenbeck reports grants from NIH during the conduct of the study. Y. Li reports grants from NIH and Cancer Prevention & Research Institute of Texas during the conduct of the study. No disclosures were reported by the other authors.

F. Yue: Conceptualization, data curation, formal analysis, investigation, methodology, writing–original draft, writing–review and editing. W. Jiang: Data curation. A.T. Ku: Data curation. A.I.J. Young: Data curation. W. Zhang: Data curation. E.P. Souto: Data curation. Y. Gao: Data curation. Z. Yu: Data curation. Y. Wang: Data curation. C.J. Creighton: Formal analysis. C. Nagi: Formal analysis. T. Wang: Formal analysis. S.G. Hilsenbeck: Formal analysis. X.-H. Feng: Investigation. S. Huang: Investigation. C. Coarfa: Formal analysis. X.H.-F. Zhang: Conceptualization. Q. Liu: Conceptualization, resources. X. Lin: Conceptualization, data curation, investigation, writing–review and editing. Y. Li: Conceptualization, resources, supervision, funding acquisition, writing–review and editing.

The authors thank Dr. Feng Cong for generous gifts of reagents. They also thank Ms. Myra Costello and Ms. Fuli Jia from the Baylor College of Medicine Antibody–based Proteomics Core/Shared Resource for their technical assistance in performing Antibody Array experiments, as well as Dr. I-wen Song for assistance on X-ray imaging. The authors acknowledge the support from the Rolanette and Berdon Lawrence Bone Disease Program of Texas and the Baylor College of Medicine Center for Skeletal Biology and Medicine led by Dr. Brendan Lee. This work was supported in part by funds from NIH R01CA204926 (Y. Li) and DOD–CDMRP BC160240 (Q. Liu and Y. Li), resources from the Lester and Sue Smith Breast Center (P50CA186784) and Dan L. Duncan Cancer Center (P30CA125123), Cancer Prevention & Research Institute of Texas Proteomics and Metabolomics Core Facility Support Award (RP170005, S. Huang), NCI Cancer Center Support Grant (P30CA125123, S. Huang), and NIH S10 instrument award (S10OD028648–01, S. Huang) to Antibody-based Proteomics Core/Shared Resource (S. Huang).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.