Background
Breast cancer is the most common cancer affecting women [
1]. Gene expression profiling has identified distinct biological subtypes of breast cancer: luminal A or B, human epidermal growth factor receptor 2 (HER2) amplified, basal like, and claudin low [
2]. The luminal A and B subtypes are both estrogen receptor (ER) positive and comprise up to 70% of all breast cancers. Luminal B tumors are also HER2 positive and have a poorer prognosis [
2-
4]. The basal like and claudin low subtypes are both triple negative, lacking expression of ER, HER2 and the progesterone receptor. Treatment of luminal A tumors with tamoxifen, a selective ER modulator, has significantly reduced the mortality rate. However, not all patients respond to tamoxifen and one third of initial responders have recurrent disease within 15 years [
5]. Hormone resistance can occur through ER-dependent as well as ER-independent mechanisms, including activation of pro-proliferative signaling pathways such as HER2 and EGFR [
6], PI3K/Akt, and MAPK [
7]. Use of trastuzumab, an antibody targeting HER2, has extended the overall survival of patients with HER2 amplified tumors [
8]. However, about 40-60% of these tumors show
de novo resistance even when treatment is combined with systematic chemotherapy [
9]. Furthermore, about 70% of initial responders show progressive disease within a year. Acquired resistance can occur through overexpression of EGFR family receptors [
10] or IGF-R1 [
11], PTEN loss, or activation of PI3KCA [
12,
13]. Therefore, there is a need to identify new therapeutic targets.
Recently, aberrant activation of the Sonic Hedgehog (Shh) pathway has been implicated in breast cancer progression [
14-
26]. The hedgehog family of secreted signaling molecules includes Shh, Indian and Desert Hedgehog. Interaction of Shh with the transmembrane receptor Patched-1 (Ptch-1) relieves inhibition of the transducer Smoothened (Smo). This leads to the stabilization and nuclear translocation of the Gli family of transcription factors [
27]. The resulting activation of target gene transcription regulates various cellular processes such as cell fate determination, proliferation, and survival [
27]. A role for abnormal Shh signaling activity in breast cancer development was first reported using transgenic mouse models, where Ptch-1 haploinsufficiency or ectopic expression of Smo lead to distinct forms of mammary ductal dysplasia [
28,
29]. Furthermore, expression of Gli-1 under the mouse mammary tumor virus promoter leads to the development of hyperplastic lesions and tumors [
22]. Mutations in Shh, Ptch, and Smo are rarely identified in human breast cancer [
23]. Ptch expression is reduced in ductal carcinoma
in situ (DCIS) [
29,
30], possibly due to increased promoter methylation [
30]. In addition, ectopic expression of Smo has been identified in both DCIS and invasive breast cancer [
29]. Breast tumor growth and metastasis in mice is stimulated by Shh overexpression and is decreased by inhibiting Shh signaling [
14]. In humans, Shh overexpression occurs in breast tumor initiating cells and in invasive ductal carcinoma (IDC), where it is associated with increased metastasis and death [
14]. A progressive increase in Shh expression correlates with disease progression from low grade DCIS to IDC [
14,
15]. In addition, three studies have noted strong Gli-1 expression in stromal cells [
14,
18,
19]. Shh and Ihh secreted by breast cancer cells can signal in a paracrine manner to induce osteoclast differentiation and increase bone resorption [
24]. Furthermore, other pathways, including osteopontin and TGFβ, can also activate Gli-mediated transcription in breast cancer cells [
25,
26].
To date, analyses of the hedgehog pathway in breast cancer have focused mainly on downstream signaling events. Little is known about components of the pathway upstream of ligand production. Shh is synthesized as a precursor protein that undergoes autoprocessing to produce a ~25 kDa C-terminal fragment and a ~19 kDa N-terminal fragment (ShhN) that retains all signaling activity [
31,
32]. ShhN is modified with two lipids. Cholesterol is covalently attached to the C-terminus during the autoprocessing reaction [
33]. Cholesterol attachment contributes to long-range signaling activity, but is not essential for signaling [
34]. The N-terminus of ShhN is modified by covalent attachment of the 16-carbon fatty acid palmitate to the N-terminal cysteine [
35,
36]. Shh palmitoylation is catalyzed by Hedgehog acyltransferase (Hhat), a multipass transmembrane enzyme that belongs to the membrane bound O-acyltransferase (MBOAT) family [
36]. Multiple studies have established that palmitoylation of Shh by Hhat is critical for Shh signaling activity [
34,
37-
40]. Furthermore, Hhat activity is required for the proliferation of pancreatic cancer cells
in vivo and for the maintenance of a stem-like phenotype in lung squamous cell carcinoma [
41-
44].
The role of Hhat in breast cancer has not yet been examined. In this study, we demonstrate that Hhat is required for the proliferation of ER positive, HER2 positive, and tamoxifen resistant breast cancer cells. Increased Hhat expression resulted in increased cell proliferation, while Hhat depletion reduced proliferation of ER positive cells. Hhat inhibition with RU-SKI 43, a selective small molecule inhibitor of Hhat recently identified by our group [
45], also reduced the growth of ER positive cells. Furthermore, Hhat depletion or inhibition led to a significant decrease in HER2 positive and tamoxifen resistant cell proliferation. None of the cell lines we tested responded to inhibition of Smo, and only a subset responded to Shh depletion, indicating that non-canonical Shh signaling pathways were operative. Taken together, these data suggest that Hhat may serve as an important therapeutic target in ER positive, HER2 amplified, and hormone resistant breast cancers.
Discussion
In this study, we used genetic and pharmacologic methods to establish Hhat as a critical regulator of breast cancer cell growth. Hhat depletion or treatment with the selective Hhat inhibitor RU-SKI 43 reduced both anchorage-dependent and anchorage-independent proliferation of ER positive cells (Figures
1,
2 and
3). Hhat knockdown or inhibition also reduced the growth of HER2 positive and tamoxifen resistant cells (Figures
1,
7 and
9). Inhibition of breast cancer cell growth by RU-SKI 43 was dose dependent and was rescued by Hhat overexpression (Figure
3). Treatment with C2, a compound that is structurally similar to RU-SKI 43 but does not inhibit Hhat activity [
45], had no effect on proliferation (Figure
3). We have previously demonstrated that the inhibitory effect of RU-SKI 43 is selective for Hhat, as this compound does not inhibit palmitoylation of H-Ras and Fyn, myristoylation of c-Src, or fatty acylation of Wnt3a by Porcupine, another member of the MBOAT family [
45]. Overexpressing increasing amounts of Hhat, but not Porcupine, decreases the inhibitory effect of RU-SKI 43 on Shh palmitoylation [
45]. Moreover, overexpression of Hhat reduced the inhibitory effect of RU-SKI 43 on breast cancer cell proliferation (Figure
3H). It is possible that breast tumors that overexpress Hhat due to gene amplification might require higher doses of Hhat inhibitor. However, our finding that RU-SKI 43 inhibits the growth of T47D cells, which express relatively high levels of Hhat compared to other cell lines (Figure
1A), suggests that Hhat inhibition is a viable approach to reducing breast cancer cell growth. Taken together, these data suggest that the primary target of RU-SKI 43 is Hhat, and provide the first identification of Hhat as a novel target in breast cancer.
Hhat was identified as the palmitoyl acyltransferase for Shh and the hedgehog family of proteins [
36,
40], and Hhat inhibition has been shown to block Shh signaling [
45]. Thus, it was important to monitor expression of Shh and hedgehog signaling pathway components in breast cancer cells. There is general agreement between the findings reported here and in four other studies [
17,
19-
21] that examined expression levels of Shh pathway components in four of the same cell lines (T47D, MCF7, MDA-MB-231, and BT474) that we analyzed: 1) Shh is expressed in MCF7, T47D, and MDA-MB-2312) Ptch-1 and 2 are expressed in all four cell lines, and 3) Smo is expressed in T47D and BT474 but not in MCF7 and MDA-MB-231 cells. However, in contrast to other studies, we did not detect Ihh, Dhh, Gli-1 or Gli-2 expression in MCF7 or T47D cells (Figure
5, Additional file
3: Figure S3). Differences in Gli expression among the four studies may be due to differences in culture methods or confluence state of cells.
Our study addresses two key questions regarding the role of Shh in breast cancer: 1) Do Shh expressing cells exhibit an autocrine response to Shh? 2) If so, does this occur through canonical or non-canonical signaling? Here, we identify two cell lines, T47D and HCC1428, where knockdown of Shh reduced anchorage dependent and independent proliferation (Figure
5). T47D cells can also undergo increased proliferation in response to exogenous Shh, but this increase is only evident after endogenous levels of Shh are depleted (Figure
6). However, T47D and HCC1428 cells neither express Gli-1 (Figure
5) nor respond to treatment with the Smo inhibitor LDE-225 (Figure
6), indicating the presence of non-canonical Shh signaling. Others have also noted that treatment with cyclopamine, a Smo inhibitor, reduces proliferation of certain breast cancer cells, but that this does not correlate with Smo expression [
19] or inhibition [
20]. In this study, we used LDE-225 at 0.1 μM, a concentration 100x higher than IC
50 for binding of LDE-225 to Smo [
54], and found no effect on proliferation of any of the breast cancer cells (Figure
6). Taken together, these findings suggest that in breast cancer cells, canonical Smo mediated signaling is not operative, and cells that respond to Shh do so via non-canonical, Smo-independent signaling. This conclusion is supported by multiple recent studies documenting the existence of non-canonical, Smo-independent Shh signaling pathways in normal and cancer cells [
21,
41,
57-
59].
The findings presented here indicate that Hhat has regulatory roles in addition to Shh signaling. Shh depleted cells were still sensitive to Hhat inhibition and this growth defect was not rescued by addition of exogenous Shh (Figure
6). Moreover, we demonstrate a requirement for Hhat, but not Shh, for proliferation of multiple ER positive cells (Figures
1,
2 and
5), consistent with our recent report showing that Hhat can have Shh-independent functions in pancreatic cancer cells [
41]. We speculate that Hhat has substrates in addition to the hedgehog family. Studies in flies have shown that the EGF-like ligand Spitz is a substrate for Rasp, the
Drosophila melanogaster ortholog of Hhat [
60]. Although no Spitz ortholog has been identified in mammals, and none of the mammalian EGF family ligands appear to be palmitoylated by Hhat, our findings of hedgehog-independent roles of Hhat suggest that other substrates exist. We conclude that Hhat can promote breast cancer cell growth in a Shh independent manner.
All ER positive cell lines that we tested responded to Hhat depletion or inhibition by exhibiting decreased proliferation, while triple negative cell lines did not. Multiple lines of evidence argue against the possibility that Hhat operates via a direct, ER-dependent mechanism. First, despite reports that ERα is palmitoylated, ERα is unlikely to be a direct substrate for Hhat. The active site of Hhat is oriented towards the lumen of the endoplasmic reticulum. Hhat mediated palmitoylation occurs in the ER lumen and Hhat only palmitoylates secreted proteins [
36,
60]. In contrast, ERα is localized to the nucleus, cytosol and plasma membrane, and palmitoylation of ERα is thought to occur in the cytoplasm [
47]. Thus, Hhat could not topologically access ERα as a substrate as ERα does not enter the secretory pathway. Second, using
125I-iodopalmitate, a sensitive and robust probe for palmitoylated proteins, we were unable to detect incorporation of
125I-iodopalmitate into either endogenous or overexpressed ERα. Third, RU-SKI 43 treatment did not alter the localization or activation of ERα, suggesting RU-SKI 43 does not directly affect ERα function (Figure
4). Fourth, depletion or inhibition of Hhat can also inhibit the growth of HER2 positive cells that are ER negative (Figure
7, Additional file
5: Figure S5B), indicating that, in the context of HER2 amplification, Hhat can modulate cell proliferation independently of ER status.
Increased PI3K/mTOR signaling occurs in up to a quarter of breast cancers [
13] and upregulation of Akt signaling is associated with resistance to both endocrine and HER2 targeted therapies [
12,
13]. We observed that simultaneous inhibition of PI3K/mTOR and Hhat led to a greater decrease in cell proliferation than with either agent alone (Figure
8). Similarly, combined treatment with the Hhat inhibitor and tamoxifen was more effective than either drug alone (Figure
9). In addition, we noted that tamoxifen resistant cells, either through HER2 amplification (BT474) or other mechanisms (TamR), maintained sensitivity to Hhat knockdown or inhibition (Figures
1 and
9). Of note, combined treatment of the TamR cells with the Hhat inhibitor and tamoxifen was more effective than RU-SKI 43 alone (Figure
9). Since RU-SKI 43 did not alter ERα activation in TamR cells (Figure
4C), it is possible that other pathways induced during selection for tamoxifen resistance may contribute to the increased sensitivity in this clone. As with all pharmacologic approaches, we cannot exclude the possibility that off-target effects of RU-SKI 43, yet to be identified, contribute to the response in TamR cells. Taken together, these data underscore the therapeutic potential of using Hhat inhibitors alone or in combination with PI3K/mTOR inhibitors or ER modulators to treat breast cancer and circumvent or delay resistance to current treatments.
Methods
Reagents and antibodies
Lipofectamine 2000® and TRIzol® were obtained from Invitrogen (Carlsbad, CA). Polybrene was purchased from Santa Cruz Biotechnology Inc. (Santa Cruz, CA). Anti-HA antibodies, 17β-estradiol, 4-hydroxytamoxifen, and puromycin were purchased from Sigma (St. Louis, MO). Anti-actin was purchased from BD Bioscience (San Jose, CA). The ErbB2/HER2, ERα, and pSer118 ERα antibodies were purchased from Cell Signaling (Danvers, MA). LDE-225, LY2940002, and lapatinib ditosylate were purchased from Selleckchem (Houston, TX). Rapamycin was obtained from Fisher Scientific (Waltham, MA). Blasticidin S Hydrochloride was obtained from MP Biomedicals (Santa Ana, CA). 0.4% Trypan Blue Solution was purchased from Cellgro (Manassas, VA). Recombinant human Shh(C24II) was purchased from R&D Systems (Minneapolis, MN).
Plasmids
Plasmids encoding short hairpin RNA (shRNA) sequences for Shh (Clone IDTRCN0000033304), Hhat shRNA 1 (Clone ID TRCN0000035600) and Hhat shRNA 2 (Clone ID TRCN0000035601), cloned into the pLKO.1 vector, were purchased from Open Biosystems (Lafayette, CO). Control pLKO.1 vector, carrying a scrambled shRNA sequence, as well as pHRD8.2 and pCMV VSV-G plasmids, were gifts from Dr. Filippo Giancotti (Memorial Sloan Kettering Cancer Center, New York, NY). The pLenti6/V5-GW/lacZ vector was purchased from Invitrogen (Carlsbad, CA).
Cell culture
Human breast cancer cell lines were gifts from the following colleagues at Memorial Sloan Kettering Cancer Center, New York, NY: T47D, HCC1428, BT474 (Dr. Jacqueline Bromberg), MCF7 (Dr. Michael Overholtzer), BT549 and MDA-MB-231 (Dr. Alan Hall), Hs578t, CAMA-1, MDA-MB-453, and SK-BR-3 (Dr. Filippo Giancotti). Cells were grown following ATCC guidelines. TamR cells were a gift from Dr. Guangdi Wang (Xavier University of Louisiana, New Orleans, LA) and grown in ATCC-formulated Dulbecco’s Modified Eagle’s Medium, supplemented with 10% FBS and 1.0 × 10−4 M 4-hydroxytamoxifen. All cell lines were authenticated by the ATCC/Promega Cell Line Authentication Service using Short Tandem Repeat profiling analysis performed on July 1, 2014. All cell lines were scored as an exact match for the corresponding ATCC human cell line except for the TamR cell line, which was a 93% match to parental MCF7 cells.
Lentivirus production and knockdown
Endogenous Shh or Hhat were depleted using shRNA delivered to cells via a lentiviral system. Target sequences are: Shh shRNA(CTACGAGTCCAAGGCACATAT), control scrambled shRNA(CCTAAGGTTAAGTCGCCCTCG), Hhat shRNA 1 (GCCACATGGTAGTGTCTCAAA) and Hhat shRNA 2 (CGTGAGCACCATGTTCAGTTT). The shRNA-expressing lentiviruses were produced by co-transfecting confluent 293 T cells in 15 cm plates with the pLKO.1 shRNA plasmid, the HIV packaging vector pHRD8.2, and pcDNA3.1 VSV-G, using Lipofectamine2000®. Virus was collected 48 and 72 h later as follows. First, media was cleared of debris by centrifugation at 500xg for 5 min. Next, the supernatant was filtered through a 0.45 μm filter, and centrifuged at 38720 × g for 2 h at 4°C in SS-34 Rotor on RC6C centrifuge (Sorvall, Asheville, NC). Finally, the pelleted virus was resuspended in ATCC-formulated Dulbecco’s Modified Eagle’s Medium, supplemented with 10% FBS, and stored at −80°C. Transduction of cells with lentiviruses was carried out in the presence of 6 μg/ml Polybrene. Stable cell lines were produced by transducing target cells with either control scrambled, Shh, or Hhat shRNA expressing lentiviruses, followed by selection in puromycin.
Hhat overexpression
The pLenti6/V5-GW/lacZ vector was purchased from Invitrogen. The lacZ gene was removed by digestion with SpeI and XhoI, and HhatHA flanked by SpeI and XhoI sites was ligated into the vector. All constructs were confirmed by DNA sequencing. Lentivirus was produced as above and stable cell lines were generated by transducing target cells with either LacZ or HhatHA expressing lentiviruses. Cells were selected in Blasticidin S.
Anchorage dependent cell proliferation
Cells were plated in 6-well plates (0.5-1 × 105 cells/well, depending on cell type). For experiments involving drug treatment, drugs were added to the media 24 h after plating and media was refreshed every 48 h. Cells were grown for up to 6 days, trypsinized and counted with a hemocytometer.
Anchorage independent cell proliferation
Cells were plated in Corning Costar Ultra-Low attachment 24-well plates (0.1-0.2 × 105 cells/well). For experiments involving drug treatment, drugs were added to the media 24 h after plating and replenished every 48 h. After 14 days, cells were pelleted, washed with PBS, and treated with 0.05% Trypsin-EDTA. The trypsin was quenched with cell culture media, 0.4% Trypan Blue Solution was added and cells were counted with a hemocytometer.
qRT-PCR
Total RNA was isolated using TRIzol extraction. cDNA was synthesized using the iScript™ cDNA Synthesis Kit (Bio-Rad Laboratories, Hercules, CA) following the manufacturer’s instructions. qRT-PCR was used to determine expression levels of Hhat, Shh, Ihh, Dhh, Patched-1, Patched-2, hHIP, Smoothened, Gli-1, Gli-2, Gli-3 and HPRT using SsoAdvanced™ SYBR® Green Supermix and the CFX Connect Real Time System (Bio-Rad Laboratories, Hercules, CA). Gene specific primers are listed in Additional file
6: Table S1. Hypoxanthine Phosphoribosyltransferase 1 (HPRT) was used as an endogenous reference, and the relative expression levels of each gene were normalized using the comparative Ct method. Gene expression was normalized to the endogenous reference given by 2
−ΔΔCT.
Immunoblot analysis
Cells were lysed in radioimmune precipitation assay (RIPA) buffer (150 mM NaCl, 50 mM Tris (pH 7.4), 1% Triton X-100, 0.5% sodium deoxycholate, 0.1% SDS, and 1 mM EDTA). Lysates in sample buffer were electrophoresed on SDS-PAGE gels, transferred to PVDF membranes, and probed with the indicated antibodies. To monitor phosphorylation of ERα Ser118, MCF7 or TamR cells were treated with either DMSO or 10 μM RU-SKI 43 for 4 h. Media was also supplemented with either ethanol or 200 nM 17β-estradiol for the last 30 min of incubation. Cells were lysed in RIPA buffer containing Halt Protease Inhibitor Cocktail and Halt Phosphatase Inhibitor Cocktail (Thermo Scientific). Lysates in sample buffer were electrophoresed on SDS-PAGE gels, transferred to PVDF membranes, and probed with the indicated antibodies.
Indirect Immunofluorescence
MCF7 cells were seeded onto coverslips in 6-well plates and cultured for an additional 24 h. Cells were treated with either DMSO or 10 μM RU-SKI 43 for 4 h. Cells were fixed with 4% paraformaldehyde for 20 min and permeabilized with 0.2% Triton X-100 for 5 min at room temperature. Cells were incubated with anti-ERα (Cell Signaling) for 1 h followed with incubation with a secondary antibody (Alexa Flour® 488-conjugated anti-mouse IgG) for 45 min. Slides were mounted with ProLong® Gold Antifade (Invitrogen). Images were collected using a Leica SP5 confocal microscope and analyzed with the Leica Application Suite software. Images were collected using the same conditions on the same day ensuring fair side-by-side comparison.
Competing interests
The authors declare that they have no competing interests.
Authors’ contributions
AM and MDR participated in conception and design of the study, development of methodology, analysis and interpretation of data, and writing of the manuscript. AM carried out the experiments. Both authors read and approved the final manuscript.