A novel function of R-spondin1 in regulating estrogen receptor expression independent of Wnt/β-catenin signaling

R-spondin1 (Rspo1) has been featured as a Wnt agonist, serving as a potent niche factor for stem cells in many tissues. Here we unveil a novel role of Rspo1 in promoting estrogen receptor alpha (Esr1) expression, hence regulating the output of steroid hormone signaling in the mouse mammary gland. This action of Rspo1 relies on the receptor Lgr4 and intracellular cAMP-PKA signaling, yet is independent of Wnt/β-catenin signaling. These mechanisms were reinforced by genetic evidence. Luminal cells-specific knockout of Rspo1 results in decreased Esr1 expression and reduced mammary side branches. In contrast, luminal cells-specific knockout of Wnt4, while attenuating basal cell Wnt/β-catenin signaling activities, enhances Esr1 expression. Our data reveal a novel Wnt-independent role of Rspo1, in which Rspo1 acts as a bona fide GPCR activator eliciting intracellular cAMP signaling. The identification of Rspo1-ERα signaling axis may have a broad implication in estrogen-associated diseases.


Introduction
Estrogen and progesterone are the main players in mammary development and the progression of breast cancers (Hilton et al., 2018;Macias and Hinck, 2012). Both hormones act through their cognate receptors, estrogen receptor (ER) and progesterone receptor (PR) (Hilton et al., 2018). The mechanisms of ERa activity have been extensively studied (Carroll, 2016). However, the upstream regulation of ERa (Esr1) expression is much less understood.
R-spondin1 (Rspo1) has been identified as a hormone-mediated local factor, whose expression is upregulated by estrogen and progesterone (Cai et al., 2020;Cai et al., 2014). R-spondin protein family (Rspo1-4) have been reported to function as niche factors for adult stem cells in multiple organs (Greicius et al., 2018;Han et al., 2014;Planas-Paz et al., 2016;Sigal et al., 2017), and Rspo1 has been implicated as critical growth factor in many in vitro stem cell expansion systems, including intestine, stomach and liver (Barker et al., 2010;Huch et al., 2013;Kim et al., 2005;Sato et al., 2009). The role of Rspo1 in Wnt signaling has been extensively studied. Rspo1, through its interaction with its receptors Lgr4/5/6, enhances Wnt signaling by attenuating the turnover of Wnt receptors (Hao et al., 2012;Koo et al., 2012) and potentiating phosphorylation of the Wnt coreceptor Lrp (Carmon et al., 2011;de Lau et al., 2011;Glinka et al., 2011;Gong et al., 2012). In the mammary gland, Rspo1 synergizes with another niche factor, Wnt4, to promote mammary basal stem cell self-renewal (Cai et al., 2014). In line with the role of Rspo1 in MaSC regulation, Rspo1 expression is enhanced in the diestrus phase of the estrous cycle and during pregnancy (Cai et al., 2014), coinciding with the rise of progesterone level and the expansion of basal stem cells (Asselin-Labat et al., 2010;Joshi et al., 2010). Our recent study also reported the enhanced Rspo1 expression in estrus, a stage with high estrogen signaling activity (Cai et al., 2020). Another role of Rspo1 may exist besides maintaining basal stem cells.
In this study, we uncover a novel function of Rspo1 distinct from its previously reported role in stem cell regulation. We provide evidence that Rspo1 promotes ERa (Esr1) expression in luminal cells of the mammary gland. This action of Rspo1 is through activating G-protein coupled cAMP/ PKA pathway, while independent of Wnt/b-catenin signaling. Our data reveal a novel Wnt-independent role of Rspo1, and a new upstream regulatory axis for Esr1 expression.

Rspo1 induces ERa expression and promotes ERa signaling
To investigate the potential role of Rspo1 in luminal cells, we isolated primary luminal cells (Lin -, CD24 + , CD29 lo ) by FACS (fluorescence-activated cell sorting), and cultured them in 3D Matrigel in the presence of RSPO1 (0.5 mg/ml) (Figure 1-figure supplement 1a). Transcriptome and Gene ontology (GO) analysis identified enrichment of various features, including estrogen receptor activity (Figure 1a and b). qPCR analysis verified that the expression of ERa signaling target genes, including Pgr (progesterone receptor, PR), Ctsd1 (Cathepsin D1) (Meneses-Morales et al., 2014), and Wisp2 (Zhang et al., 2012b) are enhanced in the presence of RSPO1 (Figure 1-figure supplement  1b).
To further investigate how Rspo1 regulates ERa signaling, we isolated ER + luminal cells (Lin -, CD24 + , CD29 lo , Sca1 + ) and ERluminal cells (Lin -, CD24 + , CD29 lo , Sca1 -) based on Sca1 expression (Figure 1c), and cultured them in 3D. RSPO1 treatment resulted in the upregulation of ERa targets, Pgr, Ctsd1 and Wisp2 in ER + luminal cells, indicating the further activation of ERa signaling ( Figure 1d). Interestingly, the expression of ERa itself (Esr1) is also enhanced (Figure 1e). In contrast, ERluminal cells did not respond to RSPO1 stimulation ( Figure 1-figure supplement 1c). Estrogen (Estradiol-E2, E2) is one of the few known upstream regulator of Esr1 (Chu et al., 2007;Kanaya et al., 2019). Thus, E2 (1 mM) was used as control to show the extent of Esr1 activation. We found that in this ER + luminal cell culture system, RSPO1 elevated the expression of Esr1 and its target Pgr to a level comparable with E2 treatment (compare Figure 1d-e with Figure 1f). The upregulation of ERa protein by RSPO1 was confirmed by Western blot analysis (Figure 1g). This role of RSPO1 was further validated in mouse mammary Eph4 cells. RSPO1 upregulates the expression of Esr1 and ERa signaling targets Pgr and Greb1 (growth regulation by estrogen in breast cancer 1) in a dose-depending manner (Figure 1-figure supplement 2a-c).
To investigate whether Rspo1 regulates Esr1 transcription, we utilized a luciferase reporter driven by the proximal promoter (promoter A) of human ESR1 (Tanimoto et al., 1999). We found that RSPO1 can induce luciferase expression in a dose-dependent manner, while the control reporter lacking ESR1 promoter was not activated in any conditions ( Figure 1h). Together, these data suggest that Rspo1 enhances Esr1 transcription.

Rspo1-induced ERa expression is dependent on Lgr4
To investigate the mechanisms through which Rspo1 regulates Esr1, we first examined which receptor of Rspo1 is involved. qPCR analysis indicated that all three Lgr receptors, Lgr4/5/6 are expressed in basal cells, but only Lgr4 is expressed in luminal cells (Figure 2a), suggesting that Rspo1 may rely on Lgr4 to signal in luminal cells in the context of Esr1 induction. Within the luminal compartment, Lgr4 was evenly distributed in ER + (Sca1 + ) and ER -(Sca1 -) luminal cells (Figure 2a). In situ hybridization validated the expression pattern of Lgr4 in both basal and luminal layers (Figure 2b). We next investigated whether Lgr4 mediates Rspo1's action on Esr1 expression. We generated Lgr4 shRNA and validated its knockdown efficacy in primary luminal cells by qPCR analysis (Figure 2c). Lgr4 knockdown suppressed the upregulation of Esr1 induced by RSPO1 (Figure 2d). In an ESR1luciferase reporter assay using T47D (a human breast cancer cell line), LGR4 knockdown also inhibited the luciferase activities induced by RSPO1 (Figure 2e). The effect was validated using two different shRNAs (

ERa induction by Rspo1 is independent of Wnt/b-catenin signaling
As Rspo1 is known for amplifying Wnt/b-catenin signaling, we investigated whether Wnt ligands have a synergistic influence on Esr1 expression. We first examined Wnt4, a major Wnt ligand in the mammary gland that can activate Wnt/b-catenin signaling (Cai et al., 2014;Rajaram et al., 2015). The activation of Axin2 expression indicated that Wnt/b-catenin signaling was activated in primary luminal cell culture in the presence of Wnt4 (Figure 3a). Wnt4+RSPO1 combination further stimulated Axin2 expression ( Figure 3a). Intriguingly, addition of Wnt4 alone was ineffective in activating Esr1 expression in these cells (Figure 3b), and Wnt4+RSPO1 combination was unable to further increase Esr1 level compared to RSPO1 alone ( Figure 3b). These results suggest that canonical Wnt signaling may not be involved in this regulatory axis. Furthermore, we used either Wnt3a or a GSK3 inhibitor CHIR99021 (CHIR) to stimulate Wnt/b-catenin signaling in primary luminal cell culture. Although Wnt-signaling activators markedly increased the expression levels of its target gene Axin2 stimulating effect to Axin2, Wnt3a and CHIR treatment suppressed Esr1 expression (Figure 3d), an observation in line with a previous report, in which Wnt/b-catenin signaling represses the expression of luminal differentiation genes, mainly Esr1 (Lindley et al., 2015). Inhibition of the Frizzled receptor using its soluble CRD domain (FzCRD) (Hsieh et al., 1999) or stimulating b-catenin degradation using XAV939 (Huang et al., 2009) effectively suppressed Axin2 expression induced by Wnt3a ( Figure 3c), still, they could not suppress Esr1 upregulation by Rspo1 ( Figure 3d). To further verify, we used HEK293T cells transiently expressing ESR1-luciferase reporter and cultured them in the presence of RSPO1 or RSPO1 in combination with XAV939 or IWP2. Consistently, inhibition of WNT signaling did not affect ESR1 promoter activities induced by RSPO1 (Figure 3e and f). Together, these data suggest that Rspo1 induces ERa expression independent of Wnt/b-catenin signaling.

Loss of luminal Rspo1 results in decreased ERa expression in vivo
To investigate the role of Rspo1 in vivo, we generated a conditional Rspo1 knockout allele in which the second Rspo1 exon is subjected to removal upon Cre recombination, resulting in frame-shift of the remaining exons ( Figure 4a, also see Figure 4-figure supplement 1a-b). Of note, Rspo1 is predominantly expressed in ER-luminal cells as described previously (Cai et al., 2014), while Esr1 is expressed in ER + luminal cells. Thus, this Rspo1-Esr1 regulation is likely achieved through a paracrine (e-f) HEK293T cells with transiently expressing ESR1-luciferase reporter were cultured in the presence of RSPO1, or in combination with Wnt inhibitors (XAV939 and IWP2). Wnt inhibitors cannot suppress ESR1-luciferase activities induced by RSPO1. Data in (a-f) are pooled from more than three independent experiments and presented as mean ± s.e.m. Student's t test. ****p<0.0001, ***p<0.001, **p<0.01, *p<0.05; ns, not significant. manner in vivo. A luminal cells-specific BAC transgenic CreER line, Keratin8-CreER (Krt8-CreER) (Zhang et al., 2012a), was used to generate luminal cells-specific Rspo1 knock-out mice (Krt8-CreER;Rspo1 fl/fl ) (Rspo1-cKO) ( Figure 4b). Tamoxifen was administered into 8-week-old nulliparous female mice, and mammary glands were examined 4 weeks later. Whole-mount carmine staining showed significantly reduced side branches in Rspo1-cKO mice when compared with the control (Rspo1 fl/fl ) (Figure 4c and d). These results are consistent with previous observation in a Rspo1 -/mammary transplantation model (Chadi et al., 2009    ( Figure 4e). Rspo1 level in cKO group was significantly reduced shown by qPCR analysis (Figure 4f). By whole-mount immunofluorescence staining, we observed the decreased ERa expression in Rspo1-cKO mammary gland (Figure 4g). Quantification indicated decreased percentage of ERa + cells (Figure 4h), likely reflecting the overall reduction of ERa level in luminal compartment. Although we could not exclude the possible switching of ER + to ERcell fate due to other indirectly reasons, we tested a more direct possibility-whether it is the reduction of ERa expression in ER+ compartment that results in loss of ER + cells. To this end, we isolated ER+ luminal cells (Lin -, CD24 + , CD29 lo , Sca1 + ), and analyzed ERa levels as well as ERa signaling activities. We found that ERa levels were reduced in this compartment as shown by qPCR ( Figure 4i) and Western analysis ( Figure 4j). Consistently, ERa signaling target genes, including Pgr, Wisp2 and Ctsd1 were declined in Rspo1-cKO group (Figure 4k). Therefore, together these results suggest that loss of Rspo1 results in reduced ERa expression and its signaling activities in luminal cells.

Genetic evidence supports that Esr1 regulation is independent of luminal Wnt4
To investigate whether Wnt/b-catenin signaling affects Esr1 in vivo, we also generated a Wnt4 conditional knockout mouse. In this model, the second Wnt4 exon is flanked by flox, and is removed upon Cre recombination, which leads to frame shift of the remaining exons (Figure 5a, Figure 5- figure  supplement 1a-b). We generated inducible, luminal cells-specific Wnt4 knock-out mice (Krt8-CreER; Wnt4 fl/fl ) (Figure 5b). Tamoxifen was administered into 8-week-old nulliparous female mice, and mammary glands were examined 4 weeks later. Loss of Wnt4 resulted in reduced side branching (Figure 5c and d), consistent with a previous report using MMTV-Cre;Wnt4 fl/fl model (Rajaram et al., 2015).
To address whether loss of Wnt4 affects Esr1, we isolated ER + luminal populations from both Wnt4-cKO (Krt8-CreER;Wnt4 fl/fl ) and control (Krt8-CreER;Wnt4 fl/+ ) mammary gland. qPCR and Western analyses both indicated that loss of Wnt4 increases ERa level in ER + luminal cells (Figure 5f-g), as well as ERa signaling activities showed by increased target gene expression ( Figure 5h). These were in contrast to the reduced Esr1 level and ERa signaling activity observed in Rspo1-cKO mice (Figure 4i-k). These were consistent with the in vitro results that Wnt3a and CHIR treatment suppressed Esr1 expression (Figure 3d), and consistent with the previous report, in which Wnt-controlled transcriptional regulator LBH repress luminal genes, mainly Esr1 (Lindley et al., 2015). The successful deletion of Wnt4 in cKO group was validated by significantly reduced Wnt4 level in ER + luminal cells (Figure 5e), as well as reduced expression of Wnt/b-catenin signaling targets Axin2 and Lgr5 in basal cells (Figure 5i). Together, in vivo genetic evidence supports that Esr1 regulation is independent of luminal Wnt4.

Rspo1 relies on cAMP-PKA pathway to induce Esr1 expression
To further investigate the downstream mechanisms through which Rspo1/Lgr4 regulate Esr1, we conducted an inhibitor-based screen. HEK293T cells with transiently expressing ESR1-luciferase reporter were cultured in the presence of RSPO1, and screened for molecules that could suppress luciferase activity using a GPCR inhibitor library (Figure 6a, Figure 6-figure supplement 1, Figure 6-source data 1). Amongst over 250 inhibitors, the cAMP inhibitor Bupivacaine HCl (Bup), effectively suppressed ESR1-luciferase activities induced by RSPO1 (Figure 6a). Considering that the major downstream effector of cAMP in mammalian cells is Protein Kinase A (PKA), we examined the effect of inhibition of PKA. Consistently, H89, an inhibitor of PKA effectively repressed ESR1-luciferase activities induced by RSPO1 (Figure 6a). The inhibitory effects of Bup and H89 were further examined in primary luminal cell culture. Both inhibitors suppressed Esr1 expression stimulated by Rspo1 as shown by qPCR (Figure 6b), but were ineffective on Axin2 expression (Figure 6c). Considering the cAMP-PKA pathway can also be activated by estrogen and ERa (Castoria et al., 2008), we further examined whether Esr1 induction by RSPO1 involves ERa. We found that the ERa inhibitor ICI (ICI182, 780) does not affect ESR1 promoter activities that are induced by RSPO1 (Figure 6d), suggesting that Esr1-induction by Rspo1 does not involve ERa.
The transcription factor CREB (cAMP response element binding protein) is the best-characterized nuclear protein that mediates stimulation of transcription by cAMP. CREB binds to the conserved consensus cAMP response element (CRE, sequence TGACATCA) (Rosenberg et al., 2002). A CRE was found at the proximal promoter of ESR1 (À991 to À984 bp). Therefore, we examined whether this CRE is responsible for induction of ESR1 by RSPO1. While RSPO1 induced the wild type promoter-luciferase in a dose-dependent manner, it could not activate the reporter with CRE mutations Source data 1. ESR1-luciferase activities induced by RSPO1 in combination with a GPCR inhibitor library. Figure supplement 1. ESR1-luciferase activities induced by RSPO1 in combination with a GPCR inhibitor library HEK293 cells with transiently expressing ESR1-luciferase reporter were cultured in the presence of RSPO1, and screened for the molecules that could suppress the luciferase activities using a GPCR inhibitor library (Selleck, L2200).
(TGcCAgCA) (Figure 6e), rendering its specificity. In addition, we co-expressed ESR1-luciferase with RSPO1-FL (full length), RSPO1-N137Q (a mutated form with compromising secretion) (Chang et al., 2016), RSPO1 F110A/F106A (unable to bind LGR4) (Wang et al., 2013), and RSPO1 R66A/Q71A (binds to LGRs but is unable to amplify Wnt signaling) (Xie et al., 2013). RSPO1-FL and RSPO1 R66A/Q71A were able to activate the luciferase activities, but RSPO1-N137Q and RSPO1 F110A/ F106A could not (Figure 6f), suggesting the secretion of Rspo1 and its association to LGR4 are critical for Esr1 transcription. This is in line with the paracrine mechanism we propose. Next, we attempted to directly measure the change of cAMP level upon Rspo1 treatment in mammary Eph4 cells, using Forskolin (FSR) as a positive control. We observed a dose dependent increase of cAMP level in relation to increasing RSPO1 stimulation (Figure 6g). Together, these results suggest that Rspo1 signals through cAMP-PKA-CREB axis to promote Esr1 transcription (illustrated in Figure 6h).

Discussion
In this study, we uncover a novel function and signaling mechanism of Rspo1 in promoting ERa expression. This action of Rspo1 is dependent on Lgr4 and G-protein coupled cAMP/PKA pathway, but independent of Wnt/b-catenin signaling. In vivo, the biological significance of this regulatory axis is first revealed in the mammary gland homeostasis. Luminal cells-specific deletion of Rspo1 results decreased ERa expression and reduced side branching.

A novel Wnt-independent role of Rspo1
Rspo1 has been known as a stem cell growth factor in many adult tissues with prominent biological and therapeutic significance. The action of Rspo1 on stem cells is through strongly potentiating Wnt signaling (Carmon et al., 2011;de Lau et al., 2011;Glinka et al., 2011;Gong et al., 2012;Hao et al., 2012;Koo et al., 2012). Here we unveil a new role of Rspo1 in promoting Esr1 transcription in hormone receptor-positive luminal cells. This Wnt/b-catenin-independent action of Rspo1 relies on the Lgr4 receptor and intracellular cAMP/PKA signaling. Knockdown of Lgr4 counteracts Rspo1's augmenting effect on ERa transcription in vitro and in vivo, while modulation of Wnt input or b-catenin activity does not affect Esr1 level induced by Rspo1. Same holds true in vivo when using Wnt4-cKO mouse model. Deletion of Wnt4 in luminal cells increased Esr1 level. To our best knowledge, this is the first report of Wnt/b-catenin independent function of Rspo1 in physiological condition. It adds to previous reported Wnt/b-catenin independent role of Rspo1 in antagonizing colon cancer metastasis, in which LGR5 directly binds to TGFb receptors for the activation of TGFb signaling (Zhou et al., 2017).

Rspo1 activates G-protein coupled cAMP signaling in regulating Esr1
cAMP is a well-known intracellular mediator of protein hormones including FSH (follicle-stimulating hormone), LH (luteinizing hormone), and TSH (thyroid stimulating hormone), which bind to LGR1, LGR2 and LGR3 respectively (de Lau et al., 2014). These known hormone receptors belong to the class-A LGRs. Class-B LGRs, including LRG4-6, are reported to promote phosphorylation of Lrp5/6 and stabilization of b-catenin without the G-protein-coupled cAMP production (Carmon et al., 2011;de Lau et al., 2011). There have been a few reports that suggest differently, in that Lgr4 activates cAMP/PKA signaling in bone (Luo et al., 2009), and in the male reproductive system (Li et al., 2010). Independently, our data demonstrate that Rspo1/Lgr4 relies on the cAMP/PKA axis to maintain proper Esr1 expression during mammary development. This action is highly likely cell type specific. In vivo, either conditional KO of Rspo1 or Lgr4 hypomorphic mutant leads to reduced Esr1 expression. The latter is in line with previous reports in the male reproductive system, in which deficiency of Lgr4 results in reduced Esr1 in the efferent ducts and epididymis (Hoshii et al., 2007;Li et al., 2010). The current study, for the first time, demonstrates that Rspo1 can activate cAMP/ PKA signaling.

A new hormonal regulation feed forward mechanism
Our previous studies find that hormones indirectly activate Rspo1 expression in ERluminal cells (Cai et al., 2014), and identify Areg (in ER + cells) as the intermediate paracrine factor for the hormonal regulation of Rspo1 expression (in ERcells) (Cai et al., 2020). Moreover, the elevated levels of Areg and Rspo1 are also detected in estrus, a stage with high estrogen signaling activity (Cai et al., 2020). In this study, we found that Rspo1 in turn enhances ERa expression in ER + cells. This may represent a feed forward mechanism engaging estrogen-ERa-Rspo1-ERa, highlighting the impact of local growth factors for the amplification of hormonal signaling output. This additional layer of ERa regulation by Rspo1 could be hijacked during tumor initiation or progression. Elucidating the molecular mechanisms on how estrogen engages with ERa in the mammary gland is the key for advancing current knowledge over breast cancer progression and resistance to hormone therapy.
In conclusion, our study demonstrated a novel Wnt-independent role of Rspo1, revealed a novel Rspo1-Lgr4-cAMP-ERa regulatory axis. As ERa is crucial for the development and diseases of various tissues, this new Rspo1 signaling axis may have broader implication in estrogen-associated diseases. Rspo1 flox/+ This paper Generated in our laboratory Detail refer to Figure 4 and Wnt4 flox/+ This paper Generated in our laboratory Detail refer to Figure 5 and Continued on next page

Experimental animals
Rspo1 flox/+ and Wnt4 flox/+ mice were constructed as illustrated in the text. In all conditional knockout experiments, mice were maintained on a C57BL/6 genetic background and at least three animals were analyzed for each genotype. Lgr4 LacZ/+ (Mazerbourg et al., 2004) and Krt8-CreERT2 (Zhang et al., 2012a) strains were used in this study. Nude, CD1 and BALB/c strains were purchased from B and K universal (

RNA extraction and RNA sequencing
Total RNA from day two cultured luminal cells (Lin -, CD24+, CD29 lo ) were extracted with RNAiso Plus (Takara) following manufacturer's protocol. Total mRNA concentration was determined with NanoDrop ND-1000 and RNA-seq libraries were prepared according to manufacturer's instruction (Illumina) followed by applying to sequencing on Illumina nova-seq, which was performed by ANOR-OAD (http://en.annoroad.com, Beijing). Differential gene expression analysis was carried out and genes with significant alteration were extracted and further analysed using DAVID Bioinformatics Resources. RNA-seq data can be viewed online at http://www.biosino.org/node/index, under accession number OEP000754.

Mammary gland whole mount carmine staining
The 4th pair of mammary glands were dissected and fixed for 2 hr in 4% paraformaldehyde, and then washed the tissue three times in PBS for 15 min each time. Finally, the tissues were stained in carmine alum solution (2 mg/ml carmine [Sigma; C1022], 5 mg/ml KAl(SO 4 ) 2 in H 2 O) overnight at room temperature. After the staining, the tissues were washed in de-staining solution (50% ethanol, 2% HCl) for 2 hr, and then serial dehydrated in 75%, 85%, 95%, 100%, 100% ethanol and finally stored in Histoclear (National Diagnostics; HS-200). Whole mount analyses were performed under a dissection microscope (Leica).

Mammary gland whole mount immunostaining
Whole-mount staining was performed as previously described (Rios et al., 2014), with minor modification. In brief, mammary glands were dissected into small pieces, then processed in digestion buffer (RPMI 1640 with 25 mM HEPES, 5% fetal bovine serum, 1% penicillin-streptomycin-glutamine (PSQ), 300 U/ml collagenase III (Worthington)) for 30 min at 37˚C, then fixed in 4% paraformaldehyde for 30 min at 4˚C. Tissues were incubated with primary antibodies (Krt8; 1:500; DSHB, ER; 1:200; Millipore) at 4˚C overnight, followed by washes, incubated with secondary antibodies and DAPI (Life Technologies) at 4˚C overnight. Then the tissues were incubated in 80% glycerol overnight, before dissection for 3D imaging. Confocal images were captured using Leica SP8 laser confocal scanning microscope. Representative images were shown in the figures.

Maintenance of cell lines
293T and Eph4 cell lines were cultured in DMEM high glucose (4.5 g/L) (Gibco, C11995500BT) with 1% Penicillin/Streptomycin (Gibco, 15140) and 10% Fetal bovine serum (FBS) (Hyclone). Both cell lines were cultured in tissue culture dish, kept at 37˚C with 5% CO 2 , trypsinized, and split three times a week 1:4. T47D cell line was kindly provided by Dr. Gaoxiang Ge, Institute of Biochemistry and Cell Biology and was cultured in 1640 Medium (Gibco, C11875500BT) +10 mg/ml Insulin with 1% Penicillin/Streptomycin and 10% FBS. All cell lines were routinely negatively tested for mycoplasma.

Conditioned media preparation
Wnt4 conditional medium was prepared by culturing Wnt4-expressing Eph4 cells for 48 hr, followed by supernatant collect. Wnt4 conditional medium was stored at 4˚C for short-term storage (up to 1 week). For long-term usage, conditional medium was aliquoted after collection and stored at À80˚C.

RSPO1 protein purification
RSPO1-FC construct was cloned into expression vector with a C-terminal Fc tag. RSPO1-FC was transiently expressed in HEK293T cells and medium changed into CD293 medium (Gibco, 11913-019). One day after transfection, medium was collected by centrifugation and incubated with Protein A Agarose Beads (Santa Cruz, sc-2003). The bound recombinant protein was eluted using 500 ml 0.1M Glycine (pH = 3.0) and was collected in 1.5 ml tubes containing 30 ul 1 M Tris-HCl (pH = 9.5) buffer for neutralization. In total 5 tubes of elution were collected. The RSPO1 protein was subsequently purified and concentrated by Centrifugal Filter Volumes (Millipore, UFC803096).

Lentiviral vector and infection
Lgr4-shRNA was synthesized and subcloned into plko backbone with EGFP. Lentivirus was produced by transient transfection in 293 T cells. Mammary cells were isolated from 8-to 12-wk-old virgin female glands as described above, followed by sorting into luminal cells. The sorted cells were collected and cultured in a low adherent plate in EGF, ITS-supplemented DMEM/F12 with virus. At 12 hr after infection, cells were collected and resuspended in Matrigel for consequent in vitro culturing. Sequences of Lgr4-shRNA are CGTAATCAAATCTCCCTGATA and CCTCCAGAACAATCAGTTGAA.

Luciferase assay
Oligonucleotide primers (nucleotides À1133 to À1107 and À1 to À24 based on previously published sequence information for the upstream region of the ESR1 were used to generate ESR1 promoter fragments from normal placental DNA by polymerase chain reaction (PCR) (Castles et al., 1997). A 1133 bp (promoter A) of ESR1 promoter expression vector (ERP) was created by cloning this PCRgenerated product into the XhoI-HindIII sites of the promoterless luciferase reporter plasmid pGL4.17 basic respectively (Promega, Madison, WI, U.S.A.). Transfections of individual wells were performed using luciferase reporter plasmid (ERP or pGL4.17 basic vector alone), and pRL-TK Renilla luciferase control constructs as a correction for transfection efficiency, and also transfected with pcDNA3.1-RSPO1 overexpression (RSPO1-OE) plasmid (from 0.5 mg/ml to 4 mg/ml)) Cells were then harvested, the dual luciferase assays were performed using a commercial kit (Promega; E1910), Results are shown as fold activity over control activity of the promoterless pGL4.17 basic vector in each set of experiments. All transfections and assays were performed in duplicate with n ! 3 individual experiments. GPCR compound library (Selleckchem L2200) was used to for screening of inhibitors that suppress ESR1 upregulation by RSPO1. In each experiment, ESR1-lucieferase reporter cells were treated with RSPO1 for 36-48 hr.

In situ hybridization
In situ hybridization was performed using the RNAscope kit (Advanced Cell Diagnostics) following the manufacturer's instructions. Lgr4 probes were ordered from Advanced Cell Diagnostics. For in situ staining, at least three independent experiments were conducted. Representative images are shown in the figures.

AMP-Glo assay to detect intracellular cAMP levels
The intracellular cAMP concentration was measured using the cAMP-Glo assay kit (Promega, V1501) according to the manufacturer's instruction. The cAMP standard curve was generated using purified cAMP, from which the relative intracellular level of cAMP was inferred. For each drug treatment, three biological repeats were used, and each experiment was repeated 2-3 times.

Statistical analysis
One-way ANOVA or Student's t-test was performed, and the P-value was calculated in Prism on data represented by bar charts, which consisted of results from three independent experiments unless otherwise specified. For all experiments with error bars, the standard deviation (SD) was calculated to indicate the variation within each experiment. No statistical method was used to predetermine sample size. The experiments were not randomized. The investigators were not blinded to allocation during experiments and outcome assessment.