Expression and Therapeutic Targeting of Dopamine Receptor-1 (D1R) in Breast Cancer

Abstract

Patients with advanced breast cancer often fail to respond to treatment, creating a need to develop novel biomarkers and effective therapeutics. Dopamine (DA) is a catecholamine which binds to five G-protein-coupled receptors. We discovered expression of DA type-1 receptors (D1R) in breast cancer, thereby identifying these receptors as novel therapeutic targets in this disease. Strong to moderate immunoreactive D1R expression was found in 30% of 751 primary breast carcinomas, and was associated with larger tumors, higher tumor grades, node metastasis, and shorter patient survival. DA and D1R agonists, signaling through the cGMP/protein kinase G (PKG) pathway, suppressed cell viability, inhibited invasion, and induced apoptosis in multiple breast cancer cell lines. Fenoldopam, a peripheral D1R agonist which does not penetrate the brain, dramatically suppressed tumor growth in two mouse models with D1R-expressing xenografts by increasing both necrosis and apoptosis. D1R-expressing primary tumors and metastases in mice were detected by fluorescence imaging. In conclusion, D1R overexpression is associated with advanced breast cancer and poor prognosis. Activation of the D1R/cGMP/PKG pathway induces apoptosis in vitro and causes tumor shrinkage in vivo. Fenoldopam, which is FDA-approved to treat renal hypertension, could be repurposed as a novel therapeutic agent for patients with D1R-expressing tumors.

INTRODUCTION

Dopamine (DA) is a catecholamine that acts as a major neurotransmitter in the brain and as a circulating hormone in the periphery, where it is produced by sympathetic nerves, adrenal medulla, and the GI tract.¹ DA receptors (DAR) are expressed in peripheral tissues such as kidney, gut, and coronary arteries,^1,2 and their dysregulation is associated with hypertension, gut motility disorders, and metabolic dysfunctions.^2–4 After discovering expression of functional DAR in human adipocytes and breast adipose tissue,⁵ we questioned whether these receptors are also expressed in breast cancer, and if so, what are their functions.

DA binds to five G-protein-coupled membrane receptors, grouped by structure, pharmacology and function into D1R-like (D1R and D5R) and D2R-like (D2R, D3R and D4R) receptors. According to the original classification, D1R-like are coupled to Gαs proteins, activate adenylate cyclase (AC), increase cAMP, and stimulate protein kinase A (PKA), while D2R-like are coupled to Gαi/o proteins, inhibit AC, suppress cAMP, and inhibit PKA.⁶ Such classification is oversimplified, however, as evidence indicates that DAR can couple to other G-proteins and can activate alternative signaling pathways such as the guanylate cyclase (GC)/cGMP/protein kinase G (PKG) pathway.^7,8

Several studies reported that activation of D1R-like in striatal neurons,^9,10 coronary arteries,¹¹ and adipocytes,⁵ increases cGMP, which is generated from GTP by two distinct guanylate cyclases: particulate (pGC) and soluble (sGC). The pGCs are activated by atrial natriuretic peptides, while the cytosolic sGCs are the main targets of nitric oxide (NO);¹² sGC can be directly stimulated by YC-1 and Riociguat.¹³ Elevated cAMP or cGMP are rapidly hydrolyzed by phosphodiesterases (PDEs), a superfamily with 11 members that differ in substrate specificity and catalytic properties.¹⁴ Viagra (Sildenafil), Cialis (Tadalafil), and Levitra (Vardenafil), used to treat erectile dysfunction, selectively inhibit PDE5 which hydrolyzes cGMP,¹⁵ resulting in sustained cGMP elevation which can lead to apoptosis.¹⁶ Cialis has the longest half-life of the above PDE5 inhibitors.¹⁷

Dopaminergic drugs are widely used to treat Parkinson’s disease, schizophrenia, addiction, and hyperprolactinemia. Fenoldopam (Fen) is a high affinity (Kd=2.3 nM) peripheral D1R agonist which does not penetrate the brain. Fen is FDA-approved to treat renal hypertension,¹⁸ while causing only a small drop in blood pressure in normotensive patients.¹⁹ Given its short half-life in the circulation,²⁰ Fen is commonly administered by infusion.

Our objectives were to: 1) explore D1R expression in breast cancer cell lines and primary carcinomas, and examine for correlations with tumor attributes and disease outcome; 2) determine whether D1R activation induces apoptosis in breast cancer cells, and investigate the mechanism involved; 3) examine if Fen suppresses tumor growth in mouse xenograft models; and 4) develop an imaging approach for visualizing D1R-expressing tumors and metastases.

RESULTS

D1R expression in breast carcinomas and cell lines

The DRD1 transcript was cloned from MDA-MB-231 cells and was found to be identical to the published sequence. Next, breast carcinomas and matched normal breast tissue from the same individuals were analyzed for DRD1 gene expression by RT-PCR (Figure 1a), and for D1R proteins by Western blotting (Figure 1b). DRD1 expression was higher in tumors than in normal breast tissue in 4/7 samples, while tumor D1R proteins were higher in 6/7 samples. Expression of D1R and D2R proteins was compared in eight breast cancer cell lines (Figure 1c). The D1R proteins were most abundant in the triple-negative MDA-MB-231, SUM159, and MDA-MB-468 cells, and were generally lower in the estrogen receptor (ER)-positive MCF7, T47D, and BT474 cells, except for the ER-positive MDA-MB-175 cells. All cell lines also expressed some D2R.

Figure 1.

D1R expression in breast tumors and cell lines. DRD1 gene (a) and protein (b) expression in tumors (T) and matching normal breast tissue (N), determined by RT-PCR and Western blotting, respectively. CN: caudate nucleus, β2M: β2-microglobulin, Kid: kidney, β-Act: β-Actin. (c) D1R and D2R expression, determined by Western blotting, in 8 breast cancer cell lines. 468: MDA-MB-468, 231: MDA-MB-231, 175: MDA-MB-175. (d) Validation of the anti-D1R rabbit mAb. Blots were probed with unmodified mAb (-IP), or with mAb pre-incubated with the immunizing peptide (+IP). (e) DRD1 gene knockdown in MDA-MB-231 cells. DRD1 gene expression was determined by qPCR in cells transiently transfected with scrambled (sc) or DRD1 shRNA (sh) sequences. Data are presented as relative changes vs. sc cells (means±SEM, n=5. *P<0.01). (f) DRD1 gene expression was determined by qPCR in HEK293T cells that were untransfected (Un), or transiently transfected with sc or sh sequences. Data presented as relative changes vs. Un cells (means±SEM. n=3. *P<0.01). Reduced D1R proteins, as determined by Western blotting in MDA-MB-231 (g), and HEK293T (h) cells, transiently transfected with DRD1 shRNA. (i) DRD1 knockdown in MDA-MB-231 cells abrogated the suppression of cell viability by Fenoldopam (Fen). Cells stably transfected with scrambled or DRD1 shRNA sequences were incubated with Fen for 4 days and cell viability was analyzed by resazurin (means±SEM, n=6. *P<0.05). Gel images and graphs in this and subsequent figures are representative of at least 3 experiments.

Given reports on lack of specificity of some antibodies against DAR,^21,22 we validated the rabbit anti-D1R monoclonal antibodies (mAb) used in this study. Antibody pre-absorption with the immunizing peptide abolished the D1R bands in cell lysates by Western blotting (Figure 1d). DRD1 knockdown in MDA-MB-231 cells (Figure 1e) and HEK293T cells (Figure 1f) by shRNA markedly decreased the D1R protein band (Figures 1g and 1h, respectively). Furthermore, Fen failed to suppress the viability of MDA-MB-231 cells with downregulated D1R (Figure 1i).

Immunohistochemistry (IHC) was used to visualize D1R in carcinomas and normal breast tissue. Figure 2a shows a carcinoma with strong D1R staining, which was eliminated by antibody pre-absorption with immunizing peptide (Figure 2b). A D1R-negative carcinoma and D1R-negative normal breast tissue are shown in Figures 2c and 2d, respectively.

Figure 2.

Immunoreactive D1R in breast carcinomas and associations with tumor attributes and patient survival. (a–d) Photomicrographs (10×) of D1R-positive carcinoma before (a) and after (b) mAb pre-absorption with the immunizing peptide. (a1 and b1), ×10 of (a) and (b). (c) D1R-negative carcinoma. (d) D1R-negative normal breast tissue. (e) Distribution of immunoreactive D1R in tissue microarrays with 751 breast carcinomas and 30 normal breast samples. Data shown are percent of total tumor number, using H-scoring; all 30 normal tissue samples were D1R-negative (H-score ≤50). Positive D1R expression in carcinomas correlates with higher tumor stage (f), grade (g), and node metastasis (h); P<0.0001. Positive D1R expression is associated with shorter patient survival (i), and recurrence-free survival (j), determined by Kaplan-Meyer analysis of 508 tumors. The numbers of surviving or recurrence-free patients with D1R+ and D1R-tumors over a 20 year follow-up period are listed in the tables below the corresponding graphs. Supplementary Table 3 lists details of the Cox proportional hazards model for predictors of mortality.

Tissue microarrays were used to score 751 ductal breast carcinomas and 30 normal breast samples for D1R by IHC. Table 1 provides details of patient demographics and clinical data. Strong to intermediate D1R staining was seen in 29% of the tumors (Figure 2e), while 56% of tumors and all normal breast tissue samples were D1R-negative. D1R staining was significantly associated with pre-menopausal age, ER-negative, progesterone receptor (PR)-negative, but Her2-overexpressing tumors (Table 2). A significant association was seen between D1R-positive tumors and higher tumor stage (Figure 2f), grade (Figure 2g), and node metastases (Figure 2h). A Kaplan-Meier analysis of 508 tumors revealed that those with D1R-negative tumors had a median survival of 12 years, as compared to 6 years for those with D1R-positive tumors (Figure 2i); recurrence-free survival was similarly shortened (Figure 2j). Supplementary Table 3 shows the statistical analyses.

Table 1.

Patient demographics & tumor data

Patient characteristics	Patients No. (%)
Age
≤55 yrs	386 (51%)
>55 yrs	365 (49%)
Tumor grade
grade I	121 (16%)
grade II	380 (51%)
grade III	250 (33%)
Tumor stage
T1	148 (20%)
T2	377 (50%)
T3	148 (20%)
T4	78 (10%)
Lymph node stage
N0	272 (36%)
N1	398 (53%)
N2	71 (10%)
N3	10 (1%)
Metastasis stage
M0	746 (99%)
M1	5 (1%)
Hormone status
ER positive	463 (62%)
PR positive	424 (56%)
Her2 overexpressing	161 (21%)
Triple negative	143 (19%)
D1R stain results
negative	420 (56%)
weak positive	113(15%)
intermediate positive	163 (22%)
strong positive	55 (7%)

Table 2.

Correlation of D1R staining & tumor characteristics

Clinical data	No. of patients	D1R Positive No. (%)	P-value (chi2)
All	751	331 (44%)
Age
≤55 yrs	386	187 (48%)	0.001
>55 yrs	365	144 (39%)
Tumor size
≤2 cm	148	46 (31%)
2–5cm	377	159 (42%)
> 5cm	226	126 (56%)	<0.0001
Lymph node stage
N0	272	124 (46%)
N1	398	169 (42%)
N2	71	28 (39%)
N3	10	10 (100%)	<0.0001
Tumor grade
grade I	121	40 (33%)
grade II	380	161 (42%)
grade III	250	130 (52%)	0.0004
ER positive
Yes	463	182 (39%)
No	288	150 (52%)	<0.0001
PR positive
Yes	424	173(41%)
No	327	161 (49%)	0.02
Her2 overexpressing
Yes	161	84 (52%)	0.01
No	590	249 (42%)
Triple negative
Yes	143	72 (50%)	0.06
No	608	264 (43%)

Suppression of cell viability and induction of apoptosis

DA and three D1R agonists markedly suppressed the viability of MDA-MB-231 and MDA-MB-468 cells, whereas cabergoline, a D2R agonist, had no effect (Figure 3a). DA and Fen caused a 25–50% suppression of BT-20 and SUM159 cell viability, were less effective in T47D cells, and were ineffective in MCF7 cells (Figure 3b).

Figure 3.

DA and D1R agonists, but not a D2R agonist, reduced the viability of multiple breast cancer cells. (a) MDA-MB-231 and MDA-MB-468 cells were incubated for 4 days with increasing concentrations of DA, cabergoline (Cab), a D2R agonist, or three D1R agonists: SKF38393 (SKF), A68930 (A6) or Fenoldopam (Fen). Cell viability was determined by a resazurin assay (means ± SEM, n=6. *P<0.05). (b), BT-20, SUM159, T47D and MCF7 cells were incubated with increasing concentrations of DA or Fen. Experimental details and statistical analyses are the same as in (a).

Reduced cell viability could be due to decreased cell proliferation and/or increased apoptosis. Since BrdU incorporation was unaltered by D1R activation (Figure 4d), we focused on apoptosis. Treatment with DA or with the D1R agonist SKF38393 increased the percent of apoptotic cells 2-fold, as determined by flow cytometry (Figures 4a and 4b). Furthermore, Fen increased TUNEL-stained apoptotic cells 4-fold (Figure 4c), and induced cleavage of caspase 9 (Figure 4e).

Figure 4.

DA and D1R agonists induce apoptosis and inhibit cell invasion. (a and b) Induction of apoptosis by DA and SKF38393 (SKF), incubated with MDA-MB-231 cells for 48 hrs, as determined Annexin V/Propidium Iodide staining followed by flow cytometry. (c) Treatment of MDA-MB-231 and BT-20 cells with Fen for 48 hrs increased apoptosis, as determined by TUNEL staining. Results are presented as percent of TUNEL-positive cells (*P<0.05). (d) DA and SKF did not affect cell proliferation. MDA-MB-231 cells were incubated with the ligands for 48 hrs and BrdU incorporation was determined by an ELISA (means ± SEM, n=4. *P<0.05). (e) Cleavage of caspase 9 in MDA-MB-231 and MDA-MB-468 cells treated with 1 nM Fen for the indicated times. The upper band is pro-caspase 9, while the lower two bands are its cleavage products.

Inhibition of cell invasion

We next used Boyden chambers to examine whether D1R activation affects FBS-induced cell invasion. As demonstrated in Figure 5a, incubation of MDA-MB-231 cells with DA or Fen inhibited FBS-stimulated cell invasion by 70%, with lower effects seen in BT-20 cells. Photomicrographs of invaded MDA-MB-231 cells are shown in Figure 5b.

Figure 5.

Inhibition of FBS-induced cell invasion by dopamine (DA) and Fenoldopam (Fen). (a) Treatment of MDA-MB-231 and BT-20 cells with DA or Fen for 24 hrs reduced FBS-induced cell invasion through Matrigel-coated Boyden chambers. Serum-free medium served as a control, and 10% FBS served as a chemoattractant. Cells on the membrane underside, stained with Hoechst dye, were photographed and counted (means±SEM, n=3. *P<0.05). (b) Representative photographs of invaded MDA-MB-231 cells.

Signaling through the cGMP/PKG pathway

Because D1R agonists are categorized as cAMP/PKA activators, we examined their effects on cAMP. Unexpectedly, a 60 min incubation of MDA-MB-231 cells with 10 nM DA or Fen caused 25% and 50% decreases in cAMP accumulation, respectively (Figure 6a). Forskolin, a direct AC activator, induced a 6-fold increase in cAMP, indicating intact AC/cAMP machinery. In contrast, DA or Fen increased intracellular cGMP levels 2-fold, while Forskolin had no effects.

Figure 6.

Activation of the cGMP/PKG signaling pathway by DA and Fen. (a) Suppression of cAMP and stimulation of cGMP in MDA-MB-231 cells incubated with 10 nM DA or Fen for 60 min; Forskolin (10 μM) served as a positive control for cAMP (means±SEM, n=6. *P<0.05). (b) YC-1, a sGC activator, suppressed MDA-MB-231 and BT-20 cell viability (means±SEM, n=6). (c) KT5823 (KT), a PKG inhibitor, abrogated Fen-induced apoptosis. MDA-MB-231 cells were pre-incubated with 5 μM KT for 30 min, followed by incubation with 10 nM Fen for 48 hrs; apoptosis was determined by TUNEL (means±SEM, n=4. *P<0.05). (d) Cialis (1 μM), a PDE5 inhibitor, inhibited the viability of SUM159 cells, incubated with or without Fen for 4 days (means±SEM, n=6. *P<0.05). (e) SCH39166 (SCH), a D1R antagonist, abrogated DA-induced inhibition of cell viability (means±SEM, n=4. *P<0.05).

We then reasoned that if D1R activation reduces cell viability by increasing cGMP, bypassing the receptor and directly stimulating cGMP should have the same effect. Indeed, Figure 6b shows that YC-1, a direct sGC stimulator, reduced cell viability to 5–25% of controls. In addition, KT5823, a selective PKG inhibitor, prevented Fen-induced apoptosis (Figure 6c), confirming mediation of apoptosis by PKG. Cialis, which blocks PDE5 and prolongs cGMP elevation, moderately suppressed cell viability when used alone, but markedly enhanced Fen-induced apoptosis to 15% of controls (Figure 6d). Other signaling pathways, e.g., ERK1/2, Akt and CREB, were differentially activated by Fen (Supplementary Figure S1).

To verify that DA itself inhibits cell viability via D1R, cells were pretreated with the D1R antagonist SCH39166. Figure 6e confirms that D1R blockade abrogated DA-induced suppression of cell viability.

Our working model is presented in Figure 7. Activation of D1R by agonists such as Fen, stimulation of sGC by YC-1, and PDE5 blockade by Cialis, all cause augmentation of cGMP levels, PKG activation and apoptosis. A functional link between D1R and sGC, and the mechanism by which activated cGMP/PKG leads to apoptosis remain to be determined.

Figure 7.

Proposed model of D1R signaling via the cGMP/PKG pathway in breast cancer cells. Fen binds to D1R, activates sGC, increases cGMP levels, and activates PKG, leading to apoptosis. This pathway can also be activated by YC-1 or Riociguat through sGC, and by Cialis, Viagra or Levitra through PDE5 blockade; KT selectively inhibits PKG. Underlined compounds are all FDA-approved for various diseases.

Fenoldopam markedly inhibits growth of xenografts

Two mouse xenograft models were used to examine the in vivo effects of Fen on tumor growth. In one model, mice were inoculated with MDA-MB-231 cells into the inguinal mammary fat-pad. When tumor volumes reached ~250 mm³, Alzet osmotic mini-pumps containing vehicle, high-dose or low-dose Fen, were implanted subcutaneously (sc). Within one week, Fen at high-dose significantly reduced tumor volume (Figure 8a). After 3 weeks, tumor volumes in mice treated with high- and low-dose Fen were 40% and 60% of controls, respectively. None of the Fen-treated mice showed adverse physical or behavioral effects. Tumors exposed to high-dose Fen for three weeks were half the weight of controls (Figure 8d), showed a 4-fold increase in apoptosis (Figure 8e), and a 2-fold increase in necrosis (Figure 8f).

Figure 8.

Fenoldopam Inhibits tumor growth in two mouse xenograft models. (a) Fen suppresses growth of orthotopic MDA-MB-231 xenografts. Mice were implanted with Alzet pumps delivering vehicle (Con), high Fen (400 ng/kg/min) or low Fen (133 ng/kg/min) for 3 weeks (means±SEM, n=7–8 mice. *P<0.05). (b) Mice with control- and high Fen-treated tumors, and the same tumors, pictured with an Alzet pump, removed after 3 weeks. (c) Mice with MDA-MB-231-derived tumors were iv injected with human anti-D1R antibody conjugated to Alexa-Fluor 647. In vivo fluorescence imaging after 24 hrs shows intense fluorescence of the primary tumors and metastases. Arrows indicate insets of H&E staining of the primary tumor and metastases in the axillary lymph nodes. Injection of rabbit IgG conjugated to Alexa-Fluor 647 produced no fluorescence (not shown). Fen treatment for 3 weeks decreased the weight of MDA-MB-231-derived tumors (d), increased TUNEL-positive apoptotic cells (e), and augmented necrosis (f). (g) Treatment with high Fen markedly reduced growth of SUM159-derived xenografts inoculated in the flank. One group (HF) had the pumps for 3 weeks, while another (HF:7D), had the pumps removed after 1 week (means±SEM. n=6–8 mice). All time points in the two groups were lower than controls (P<0.05). (h) SUM159-derived tumor weights 3 weeks after Alzet implantation (*P<0.05).

Another group of mice was inoculated in the flank with SUM159 cells, and high-dose Fen was delivered by Alzet pumps as above. Compared to the robust growth of control tumors, Fen treatment caused a dramatic suppression of tumor volumes (Figure 8g), as well as weights (Figure 8h), to 15% of controls. Remarkably, when pumps were removed after one week, tumor growth remained suppressed for at least 2 more weeks (Figure 8g).

In vivo imaging of D1R-expressing tumors

We also developed in vivo imaging for detecting D1R-expressing tumors. Mice with MDA-MB-231-derived tumors were intravenously injected with rabbit anti-D1R mAb (human-specific) conjugated to Alexa-Fluor 647 fluorescent dye. Figure 8c shows fluorescence imaging of two mice 24 hrs after the injection. One mouse shows intense fluorescence at the primary tumor, while another also has fluorescence in distal metastases, as confirmed histologically (Figure 8c).

DISCUSSION

This study reports three major findings: 1) substantial expression of D1R in human breast carcinomas and cell lines, 2) induction of apoptosis by D1R activation via the cGMP/PKG signaling pathway, and 3) profound suppression of tumor growth by Fen, a peripheral D1R agonist. Nearly one third of 751 breast carcinomas had strong to moderate D1R staining, while normal breast tissue samples were D1R-negative. There was a significant correlation between immunoreactive D1R and advanced disease, i.e. tumors of higher stage, grade, and lymph node metastases. Notably, positive D1R staining was associated with ER-negative, PR-negative, but with Her2/neu-overexpressing tumors, indicating that D1R-expressing tumors do not fit within the ‘triple negative’ category. Most importantly, D1R expression predicts poor prognosis, as indicated by shorter patient survival.

These data suggest that D1R should be considered a novel prognostic biomarker in advanced breast cancer. D1R expression could be detected in tumor biopsies by IHC, or by a non-invasive imaging such as positron emission tomography (PET), using isotope-labeled D1R-selective ligands.^23,24 We foresee that a considerable number of breast cancer patients could benefit from targeted D1R therapy. FDA-approved drugs such as Fenoldopam, Riociguat, a sGC activator,²⁵ and Cialis, a PDE5 inhibitor, could be repurposed for treating breast cancer patients with advanced disease who do not respond to standard hormonal therapy or chemotherapy.

Expression of DAR in peripheral tissues and in some cancers has been reported,² but there is only scant information on their expression in breast cancer. A small study reported binding of [H³]spiperone, a D2R-like antagonist, in breast tumors,²⁶ and a more recent study described D3R and D5R expression in breast cancer stem cells.²⁷ However, there are no previous reports on specific detection of D1R expression in human tumors outside the brain. Because DRD1 has no introns, our studies carefully verified lack of genomic DNA contamination and we also rigorously validated the specificity of the anti-D1R antibodies.

DA and three D1R agonists suppressed cell viability, inhibited invasion, and induced apoptosis in multiple breast cancer cell lines, while a D2R agonist was ineffective. Others reported that DA or its agonists induce apoptosis in neuroblastoma,²⁸ leukemia,²⁷ ovarian,²⁹ breast,^27,30,31 and colon³⁰ cancer cells. However, most studies did not identify which DAR was expressed in the tumor cells, and often used DA or its agonists at high pharmacological doses, raising the possibility of non-receptor-mediated toxicity. In contrast, we have confirmed expression of both D1R gene and protein in breast cancer cell lines and primary carcinomas, and demonstrated that selective D1R agonists at low nM doses suppress cell viability and induce apoptosis.

The presence of D1R in breast carcinomas raises several interesting issues: a) the mechanism responsible for D1R overexpression in tumors vs normal breast tissue, b) the regulation of D1R expression in malignant cells, c) the minimal level of D1R expression necessary for responsiveness to a ligand, d) possible ligand-independent actions of D1R as is the case for Her-2, and e) the roles played by D1R during initiation, progression and/or metastatic stages of the disease.

Another relevant issue is whether circulating DA alters growth of DAR-expressing tumors. Free DA circulates at very low levels, well below its Kd values. However, in humans, most DA circulates as DA-sulfate (DA-S), at 10-fold higher basal serum levels than all free catecholamines combined.³² Sulfo-conjugation of DA, which is carried out in the gut by SULT1A3, is the major form of peripheral DA inactivation in humans, while glucuronidation predominates in rodents.³³ DA-S does not bind DAR and is biologically inactive. However, unlike the irreversible inactivation of DA by deamination, O-methylation or glucuronidation, sulfoconjugation is reversed by arylsulfatase A (ARSA), a releasable lysosomal enzyme.³⁴

Based on our discovery that human adipocytes express ARSA which converts DA-S to bioactive DA⁵, we postulate that DAR-expressing breast tumors respond to serum DA-S only if they have an active ARSA. Moreover, stimulation vs. inhibition of tumor growth by circulating DA-S/DA depends on a balance of D1R-like and D2R-like expression in each tumor. Because at low nM level Fen is highly selective for D1R, its ability to suppress D1R-expressing tumors should not be compromised by the presence of D2R-like. Notably, rodents do not have a SULT1A3 orthologue,³⁵ and have very low serum free DA levels and no DA-S. Consequently, mice carrying human cancer xenografts are not good models for assessing whether serum DA-S/DA affect tumor growth in humans. In addition, a proper response to agonists requires not only functional D1R, but a coordinated action of all the components of the cGMP pathway (sGC, cGMP, PDE5, and PKG), which may differ among cell lines.

Our data provide evidence that D1R in breast cancer cells signals via the cGMP/PKG pathway, as was reported for striatal neurons^9,10 and adipocytes.⁵ Although D1R are classified by their ability to increase cAMP, we found a decrease, rather than an increase, in cAMP levels following its activation. This decrease may be secondary to elevated cGMP, which activates cAMP-hydrolyzing PDEs, underlying a reciprocal relationship between the two cyclic nucleotides.³⁶

Striatal DA can signal through D1R to increase NO synthesis from L-arginine by activating neuronal NO synthase.^37,38 Elevated NO levels lead to sGC activation, increased cGMP accumulation and PKG activation, which often result in apoptosis.³⁹ Future studies should establish the mode of coupling of D1R to sGC, and determine whether NO is an upstream component of the D1R/sGC/cGMP/PKG signaling pathway in breast cancer.

A combination of apoptosis and necrosis likely underlies the robust in vivo effects of Fen. The increased necrosis seen in vivo could be due to inhibition of angiogenesis, as DA reduces tumor angiogenesis in rats⁴⁰ and mice⁴¹ by inhibiting vascular endothelial growth factor⁴² and its receptor⁴³ through endothelial DAR.²⁹ The long lasting effects of Fen after termination of infusion, together with a future development of slow release formulation of orally-deliverable Fen, bode well for its prospective benefits in the treatment of patients with D1R-expressing tumors.

MATERIALS AND METHODS

Cell Culture

MDA-MB-231, MDA-MB-468, MDA-MB-175-VII, BT-20, MCF-7, T47D, BT-474, and HEK293T were obtained from American Type Culture Collection (ATCC, Manassas, VA); SUM159 cells were a gift from Dr. S. Wang (University of Cincinnati). All cell lines were authenticated by the RTSF Genomics Core (Michigan State University, East Lansing, MI). Cells were cultured in DMEM, RPMI, or DMEM/F12 (Corning, Corning, NY) with 10% FBS (Atlanta Biologicals, Flowery Branch, GA) and 50 μg/ml normocin (Invivogen, San Diego, CA). Medium for SUM159 cells contained 1 μg/ml hydrocortisone (Sigma, St. Louis, MO). T47D cell medium was supplemented with 1 μM insulin (Sigma). Medium for BT-20 and MDA-MB-175-VII cells included 1% ITS+ premix (Corning). For experiments, cells were plated in growth medium, followed by starvation in treatment medium containing 5% charcoal-stripped serum (CSS, Atlanta Biologicals) and 1 mM ascorbic acid (Sigma). After 24 hrs, cells were treated as indicated.

DRD1 Cloning from Breast Cancer Cell Lines

Total RNA was isolated using RNAspin isolation kit (GE Healthcare). Oligo dT-primed cDNA was synthesized using SuperScript II (Invitrogen, Carlsbad, CA). A subset of the DRD1 transcript sequence, spanning a portion of the 5′ UTR, the entire protein coding sequence and a portion of the 3′ UTR (nucleotides 434-2316 of GenBank RefSeq NM_000794; Supplementary Table 1) was PCR amplified, using high fidelity Phusion DNA polymerase (Thermo Fisher Scientific, Pittsburgh, PA). The amplified DRD1 sequence was cloned. DRD1 sequences were successfully isolated from MDA-MB-231, MDA-MB-468, and MCF-7 cell lines, and sequenced by Genewiz. (South Plainfield, NJ). The MDA-MB-231 sequence was confirmed by a complete sequencing of both strands and shared 100% identity with the published sequence.

Conventional RT-PCR and Quantitative Real-Time PCR (qPCR)

cDNA was synthesized from total RNA, using RT² HT First Strand Kit (Qiagen, Germantown, Maryland). Because DRD1 does not contain introns, DNase was added during both RNA isolation and cDNA synthesis, and samples were evaluated for genomic DNA contamination by omitting reverse transcriptase. PCR amplification was done with primers for DRD1, or with intron-spanning primers for β2-Microglobulin (β2M) (Supplementary Table 1). For conventional RT-PCR, products were resolved on 1.5% agarose gel and photographed. qPCR was done using SYBR Green, and fluorometric products were detected with a StepOnePlus instrument (Applied Biosystems, Carlsbad, CA). Product purity was verified using DNA melting curve analysis and agarose gel electrophoresis. PCR efficiency was determined by the LinRegPCR program. Fold changes in gene expression were calculated from cycle threshold and efficiency measurements.

Western Blot Analysis

Cells and tissue lysates (40 μg/sample), were separated on 12% SDS gels and transferred to PDVF membranes. After overnight incubation with primary antibodies (Supplementary Table 2), followed by horseradish peroxidase (HRP)-conjugated secondary antibodies, products exposed to SuperSignal chemiluminescence reagents (Pierce, Rockford, IL) were photographed. β-actin was loading control. For antibody validation, the rabbit anti-D1R mAb was incubated for 10 minutes with the immunizing peptide, matching the human D1R C-terminus sequence (Novus, NBP1-79050PEP, Novus, Littleton, CO), before incubation with blots.

Knockdown of DRD1 Expression

A short hairpin RNA (shRNA) vector, targeted against human DRD1 (MISSION shRNA SHCLND-NM_000794; TRCN0000011334, Sigma) was used to knockdown DRD1 gene expression; scrambled vector (Sigma) was negative control. Cells were transfected using Lipofectamine 2000 (Invitrogen). Knockdown was confirmed by qPCR and Western blotting. For stable transfections, cells were selected using 0.5 μg/ml puromycin (Invivogen), and colonies were picked and expanded.

Tissue Samples

Matched frozen tissue or formalin-fixed paraffin-embedded (FFPE) slides of breast carcinomas and normal breast tissues were obtained from the University of Cincinnati Pathology Department. The institutional review board (IRB) approved the use of de-identified patient samples. Tissue microarrays (TMAs) containing FFPE samples of breast carcinomas and normal breast tissues were purchased from Lifespan Biosciences (LS-SBRCA121, n=60), Biochain (Z7020005, n=63; Z7020008, n=53), Protein Biotechnologies (TMA-1007, n=67), NCI Cancer Diagnosis Program (CDP) (stage II (n=340) or III (n=168) prognostic TMAs). 751 primary ductal carcinomas and 30 normal breast tissues were analyzed for D1R expression by IHC. Samples with known outcome (508), were used for Kaplan-Meier analysis of overall patient survival and recurrence-free survival.

Immunohistochemistry (IHC)

Antigen retrieval was done in boiling sodium citrate buffer for 12 min. Non-specific antibody binding was blocked, and slides were incubated with anti-D1R mAb (1:500), followed by anti-rabbit HRP-conjugated secondary antibody (1:500). Diaminobenzidine (DAB) was the chromogen, with hematoxylin counter-staining.

Scoring of IHC

IHC staining was scored by two investigators blinded to patient data. The histo-score (H-score) was calculated as intensity score (0=none, 1=weak, 2=mild, 3=moderate, 4=strong) multiplied by percentage of stain-positive tumor cells (≤50=negative, 51–100=weak positive, 101–200=intermediate positive, >200=strong positive).

IHC Statistics

H-score results were averaged between data from two observers and duplicate patient samples. ANOVA was performed for comparing H-score with tumor stage, grade, and node metastases. P-values <0.05 were considered significant, and those with significance were adjusted by the Bonferroni method. Data were divided into positive (H-score >50) and negative (H-score ≤50) staining groups. Pearson’s chi-square analysis was used to test independence between positive and negative staining groups across different tumor characteristics. A multivariate Cox proportional hazards model was used for predictors of mortality. Kaplan-Meier analysis was used for association between D1R-positive or D1R-negative tumors and patient survival and recurrence-free survival. All statistical analyses were done using JMP version 10 (SAS Institute, Cary, NC).

Cell Viability Assay

Cells were plated at 5000 cells/well in 96-well plates. After starvation, cells were treated with various doses of the following: DA, YC-1 (Sigma), Fen, SKF 38393, A68930, cabergoline (all from Tocris), KT5823 (Caymen Chemical, Ann Arbor, MI), or Cialis (Selleckchem, Houston, Texas). After 4 days, cell viability was determined by fluorescence, using resazurin reduction assay (Sigma).

Invasion Assay

Cells were plated at 50,000 cells/well in serum-free medium on top of BioCoat Matrigel-coated inserts with 8 μm pore membranes (BD Biosciences San Jose, CA). The inserts were suspended over wells containing serum-free medium (Control), or medium with 10% FBS as a chemoattractant with and without the indicated doses of DA or Fen. After 24 hrs, Matrigel with non-invading cells was removed, and invading cells were stained with Hoechst fluorescent dye. Photographs were taken at 10× magnification (Zeiss Axioplan Imaging 2 microscope), and cell number per field was counted in a blinded fashion. Experiments included 3 inserts per treatment, with 6 random fields photographed per insert.

Flow Cytometry

Cells were plated at 200,000 cells/well in 6-well plates, starved, and treated for 48 hr with drugs. Apoptosis was determined using FITC Annexin V Apoptosis Detection Kit 1 (556547; BD Pharmingen), and analyzed by flow cytometry using a Cell Lab Quanta SC Flow Cytometer (Beckman Coulter). About 10,000 gated events were collected per treatment. Results were calculated using the Mod-fit program (Verity Software House).

TUNEL Assay

Cells were plated in 8-well chamber slides at 10,000 cells/well, starved, incubated with drugs for 48 hrs, and formalin-fixed. FFPE tumors from mice were sectioned onto slides. Apoptotic cells were detected by the terminal deoxynucleotidyl transferase-mediated dUTP nick-end labeling (TUNEL) method, using TACS TdT In Situ Apoptosis Detection Kit: DAB (R&D systems, Minneapolis, MN), with hematoxylin counterstain. The number of TUNEL-positive cells was determined by counting 4 fields/treatment, and apoptosis was calculated as the number of apoptotic cells/total number of cells × 100.

BrdU Incorporation

Cells were plated at 5000 cells/well in 96-well plates, starved, and treated with drugs for 48 hrs. Bromodeoxyuridine (BrdU) incorporation was determined by ELISA cell proliferation kit (Roche, Indianapolis, IN) and absorbance was determined on a KC Junior Plate Reader. Experiments included 6 replicates per treatment, and were repeated at least twice.

cGMP and cAMP Analyses

Cells were plated at 100,000 cells/well in 24-well plates, starved, and incubated with treatments for 60 min. Lysates were analyzed for cAMP or cGMP using respective colorimetric competitive ELISA kits (Cayman Chemical; cAMP: 581001, cGMP: 581021; Ann Arbor, MI). Limits of detection were 0.3 pmol/ml (cAMP) and 0.23 pmol/ml (cGMP).

Animals

Eight-week old female athymic nu/nu mice were obtained from NCI. Mice were housed 4/cage in sterile cages, kept under light/dark cycles (12 hr:12 hr), and acclimated for 7–10 days before experiments. The animal use protocol (#04-06-29-01) was approved by the University of Cincinnati Institutional Animal Care & Use Committee.

Mouse Xenograft Models

MDA-MB-231 cells (1.5 × 10⁶ cells/60μl) or SUM159 cells (1 × 10⁶cells/60μl) were suspended 1:1 in PBS/Matrigel and inoculated into the inguinal mammary fatpad or sc in the flank, respectively. Tumor dimensions were measured twice/week and tumor volume was calculated as length × width² × 0.52. Each treatment group contained 6–8 mice. When tumors were 200–250 mm³ in volume, Alzet osmotic mini-pumps (model 1004, Durect Corporation, Cupertino, CA) with a 100 μl reservoir, rated for a continuous delivery at 0.11 μl/hr for 4 weeks, were implanted sc in the dorsal neck. The pumps delivered PBS (control), high Fen (400 ng/kg/min) or low Fen (133 ng/kg/min). After 3 weeks, animals were sacrificed and tumors were weighed. Tumors were FFPE for TUNEL assay or histopathology. One group with SUM159-derived tumors had the pumps removed after one week and tumor monitoring continued for another two weeks.

In vivo Fluorescence Imaging

The rabbit anti-D1R mAb were fluorescently labeled using a SAIVI Alexa Fluor 647 Antibody Labeling Kit (Molecular Probes by Life Technologies). The labeled antibodies (100 μg proteins) were injected by tail vein into mice with MDA-MB-231-derived inguinal tumors of moderate size. Mice were imaged 24 hrs after injection using Kodak In Vivo Multispectral Imaging FX (Carestream Molecular Imaging). Rabbit IgGs were also labeled with Alexa Fluor 647 and injected into mice as a negative control; there was no specific fluorescence in the tumors.

Statistics

Student’s t-test or ANOVA were used where appropriate. P-values ≤0.05 were considered significant. All experiments were repeated at least 3 times, unless otherwise noted.