Blockade of the Short Form of Prolactin Receptor Induces FOXO3a/EIF-4EBP1–Mediated Cell Death in Uterine Cancer

Human prolactin (PRL) acts primarily to regulate the normal functions of the female reproductive system (1), but it is also involved in multiple processes during tumor pathogenesis, including angiogenesis and regulation of the immune system (2). Levels of circulating PRL are elevated in gynecologic malignancies (3, 4). Extrapituitary PRL plays key regulatory roles during the development and progression of endometriosis (1), as the production of PRL by the endometrium is elevated during the normal menstrual cycle (5). Substantially elevated levels of PRL and its receptor (PRLR) have been reported in serum samples from patients with uterine cancer (2), suggesting that PRL/PRLR signaling may have potentially important roles in malignant conditions (3) and as a possible marker for uterine cancer (4).

Although some antibodies targeting PRLR have been shown to reduce tumor malignancy by blocking autocrine/endocrine PRL activities (6, 7), the full-length PRLR gene product was not detected in endometriosis tissues (8). Our results from screening an array of human uterine cancer cells indicated that transcriptionally spliced isoforms of PRLR products might be responsible for mediating activities of the tumoral PRL/PRLR axis. However, the expression and functional mechanisms of different PRLR isoforms remain uncharacterized. PRL/PRLR axis is reportedly involved in multiple signaling pathways (e.g., activation of p59^fyn/p120^jak2; refs. 9–11, Stat family members and JAK2; refs. 12–14, GRB2 signaling cascade, and regulation of transcription factors such as c-Myc, Jun, and T-cell factors; refs. 15–17). This diversity is partly due to the wide variety of PRLR isoforms, which in turn leads to the regulation of different downstream signaling cascades. RefSeq data from the UCSC Genome Browser predicted nine isoforms among the transcripts encoded by the PRLR gene. Eight of the nine isoforms are transcribed into cell-associated PRLR isoforms, while the other is a noncoding transcript variant. Structurally, the extracellular ligand-binding domains are highly conserved and retain PRL-binding activity, while the membrane-proximal region, including the transmembrane domains and intracellular domain, varies between isoforms; this variation contributes to the diversity in PRL signaling activities (18). High expression of variable PRLR isoforms has been reported to be involved in cancer cell survival in gynecologic (19) malignancies (20).

Among the PRLR transcribed isoforms, three have been characterized in animals and humans as transmembrane receptors: the long form (LF, ∼100 kDa), intermediate form (IM, 65–70 kDa), and short form (SF, 45–50 kDa). PRLR_LF is transcribed from exons 3–10 (21) and PRLR_IM (65–70 kDa) from an alternative splicing deletion of exon 10 (22). The two types of the short form of PRLR (PRLR_SF) are produced via alternative splicing of exons 10 and 11 during transcription of the PRLR gene (23, 24). The PRLR_SF isoform is functionally different from PRLR_LF because of their involvement with distinct downstream factors in mediating PRL signaling in cancer cells (25). Despite the importance of PRL/PRLR signaling in the pathogenesis of uterine cancer, our knowledge of the biological roles of this complex is quite limited, especially regarding our ability to effectively target the PRL/PRLR axis in tumors.

We hypothesized that the overexpression of PRLR_SF in uterine cancer cells contributes, in part, to the oncogenic activity of the PRL/PRLR axis. To block the oncogenic signaling of the PRL/PRLR axis in uterine cancer models, we utilized G129R (19), a human PRL antagonist containing a steric Gly129-to-Arg mutation. The hormonal activity of PRL in lactation initiation is tightly regulated by PTEN and the PI3K–Akt pathway during mammary development (26), and PTEN negatively regulates the PI3K–Akt signaling pathway during the pathogenesis of uterine cancer (27). Given the biological roles of PTEN in uterine cancer (28), we included both PTEN wild-type (WT) Hec-1A and PTEN-mutated (Mut) Ishikawa uterine cancer cells in this study. Ishikawa cell is a well-differentiated human endometrial adenocarcinoma cell line (29), while the human endometrial cancer-one (HEC1A) cell was derived from a moderately differentiated adenocarcinoma of human endometrium, whose histologic feature is close to papillary adenocarcinoma (30–32). Here, we report a new mechanism for the blockade of PRL/PRLR_SF by G129R in inhibiting tumor growth of uterine cancer through initiating cell death mediated by FOXO3a/EIF-4EBP1.

Uterine cancer cell lines, including Ishikawa and Hec1A, were obtained from ATCC and authenticated by the Characterized Cell Line Core at The University of Texas MD Anderson Cancer Center (Houston, TX). Authentication was performed by the short tandem repeat method using the Power Plex 16HS Kit (Promega). G129R was supplied by Oncolix, Inc. Scramble siRNA (sc-siRNA SIC001 and SIC002), siRNAs against EIF-4EBP1 (siRNA1: SASI_Hs02_00336903 and siRNA2: SASI_Hs01_00077259), and siRNAs against FOXO3a (siRNA1: SASI_Hs01_00161590 and siRNA2: SASI_Hs01_00161591) were purchased from Sigma-Aldrich.

Ishikawa and Hec1A human uterine cancer cells were cultured in minimum essential medium and McCoy medium supplemented with 15% FBS and 0.5% gentamicin. All cells were routinely tested to confirm the absence of Mycoplasma using the MycoAlert Kit (Lonza). For the hormone-depleted conditions, cells were cultured in the same medium containing charcoalstripped FBS and 0.5% gentamicin. The IC₅₀, cytotoxicity, and proliferation of treated Ishikawa and Hec1A cells were determined under these hormone-depleted conditions.

After approval by the Investigational Review Board for the Protection of Human Subjects at The University of Texas MD Anderson Cancer Center (Houston, TX, IRB PA15-0441), archived clinical specimens of human uterine cancer and normal uterine tissues were obtained. Ninety paraffin-embedded uterine tumor or normal uterine specimens were subjected to PRLR expression analysis.

Age-matched (4- to 6-week-old) female athymic nude mice were purchased from Taconic Biosciences. All mouse studies were approved and supervised by the MD Anderson Cancer Center Institutional Animal Care and Use Committee (IACUC). To establish intrauterine orthotopic tumors, mice were surgically implanted with Ishikawa or Hec1A cells (both 4 × 10⁶ cells per 25 μL Hank's Balanced Salt Solution). Briefly, mice were anesthetized via isoflurane (Baxter) inhalation, and a 0.5-cm incision was surgically created in the right lower flank to optimize exposure of the right uterine horn. The distal portion of the horn was identified and pulled to the incision for exposure. A near single-cell suspension of 25 μL or less was injected into the lumen of the uterine horn. The injection site was closely monitored during and following injection to ensure that no spillage into the peritoneal cavity occurred (33). The incision was then closed with absorbable suture and staples. Mice were monitored daily for any postoperative adverse effects.

Twenty-one days after inoculation, mice were randomized into one of four treatment groups (n ≥ 7 mice/group): (i) control (mannitol, 100 μg/mouse intraperitoneally daily; n = 7, 1 mouse died 12 days after injection with no tumor); (ii) G129R (100 μg/mouse intraperitoneally daily; n = 8); (iii) paclitaxel (75 μg/mouse intraperitoneally weekly; n = 8); or (iv) G129R + paclitaxel (same dosages and routes as monotherapy; n = 7, 1 mouse died 16 days after injection with no tumor). Mannitol was used as the control for these experiments because it is the principal excipient in the G129R formulation (34). Mice were treated for approximately 28 days and were monitored daily and weighed weekly. When mice in any single group became moribund, all mice in all four groups were euthanized. The body weight, tumor burden, and number/location of nodules in each mouse were assessed through intraperitoneal dissection.

The Student t test and ANOVA (one-way ANOVA) were used to identify statistically significant differences between groups in the cell-based assays and animal studies. For the studies comparing gene/isoform levels between groups from The Cancer Genome Atlas (TCGA) and Genotype-Tissue Expression (GTEx) databases, we used the Wilcoxon rank-sum test, and the analyses were performed using the R language (2016 version; R Foundation for Statistical Computing, https://www.R-project.org/). Continuous variables were compared with two-sample t tests (between two groups) or with ANOVA (for all groups). A P value of less than 0.05 from a two-tailed statistical test was considered statistically significant. All cell-based assays, including the EdU incorporation assay and the cell-cycle analysis, were repeated at least three times.

IHC, reverse phase protein array (RPPA), cell viability, cell cycle, and proliferation (EdU incorporation) assays, immunofluorescence imaging, multiphoton confocal microscopy, and nascent protein synthesis—AHA assay are presented in the Supplementary Materials and Methods.

Given the wide variety of known PRLR isoforms, we examined the expression of PRLR gene transcripts (ENST00000342362.5 and ENST00000382002.5) across all human tissues using the GTEx database (35). The PRLR gene transcripts are expressed predominantly in the adrenal gland, cervix uteri, pituitary gland, and uterus (Fig. 1A). As predicted by the UCSC Genome Browser, we identified nine RefSeq isoforms of the PRLR gene (Supplementary Table S1). Next, we queried TCGA database to examine the expression of these nine isoforms across gynecologic cancers, including 309 ovarian serous cystadenocarcinomas (OV) and 176 uterine corpus endometrial carcinomas (UCEC). Of the 176 UCECs, 106 had endometrioid endometrial adenocarcinoma (EEA), 12 had mixed serous and endometrioid endometrial adenocarcinoma, and 58 had high-grade serous endometrial adenocarcinoma. Of the nine predicted PRLR isoforms, measurements for eight were available in TCGA RNASeqV2 data for the 106 patients with UCEC-EEA; the only one not available was uc021xxl.1 (Supplementary Table S1). Through analysis by Wilcoxon rank-sum test, we found significant differences among three PRLR isoforms expressed in patients with OV or UCEC: PRLR_SF (uc003jjl.3; P < 2.2e-16), PRLR_LF (uc003jjm.2; P = 0.00244), and PRLR cytoplasmic domain (uc010iuw.1; P = 1.504e-14). Patients with UCEC had higher levels of all three of these PRLR isoforms than patients with OV. Expression of the remaining five isoforms of PRLR, including dominant-negative transcripts 3 (NM_001204316) and 4 (NM_001204317), PRLR_IM, secreted, and nonsense PRLR transcripts, did not differ significantly between patients with OV and UCEC (Fig. 1B).

In addition, we checked the transcript levels of PRLR_SF (uc003jjl.3) in a cohort of 106 patients with UCEC-EEA; we found a wide range of PRLR_SF expression in all stages of UCEC-EEA (Fig. 1C). When we combined stages I and II together versus stages III and IV and ran a Wilcoxon rank-sum test to compare the PRLR_SF (uc003jjl.3) level between these groups, we found a significant difference in the uc003jjl.3 transcript levels (P = 0.008; Fig. 1D). Among these 106 patients with EEA, 13 had histologic grade 1 (G1) disease, 19 had G2 disease, and 74 had G3 disease. G2 tumors expressed marginally higher levels of PRLR_SF uc003jjl.3 (P = 0.056 based on a Wilcoxon test) than the other two histologic grades (G1 and G3; Supplementary Fig. S1A). We then checked the level of PRLR_SF uc003jjl.3 by histologic subtype. The expression of PRLR_SF uc003jjl.3 did not differ significantly in any subtype from that in the endometrial adenocarcinoma group (Supplementary Fig. S1B).

Given that PTEN mutation is known to be prevalent in uterine/endometrial cancer and associated with metastatic behavior (28), we determined the association of PTEN status and the expression of PRLR in an array of human uterine cancer cell lines with both PTEN^mut and PTEN^WT backgrounds (ref. 36; Ishikawa, SPEC, AN3CA, Hec1A, SKUT, and KLE) using an antibody against all three PRLR isoforms (PRLR_LF, PRLR_IM, and PRLR_SF; Supplementary Fig. S2A). Unlike the PRLR_LF or PRLR_IM isoforms, PRLR_SF (∼40 kDa) was expressed consistently across all cell lines regardless of their PTEN status (Supplementary Fig. S2B). We next used antibodies specifically against PRLR_LF, PRLR_IM, and PRLR_SF to measure the expression of these PRLR isoforms in Ishikawa (PTEN^mut) and Hec1A (PTEN^WT) cells treated with PRL or its antagonist, G129R. Neither PRL nor G129R changed the level of PRLR_SF in either the Ishikawa or the Hec1A cells (Supplementary Fig. S2C). Therefore, we focused on PRLR_SF for this study.

To further investigate the unique expression of PRLR in a clinical context, we compared the levels of PRLR expression between tumor samples from a cohort of 3 patients with UCEC-EEA and a set of three normal uterine tissue samples from age-matched women without uterine cancer. IHC staining revealed significantly higher levels of PRLR expressed in the uterine/endometrial glands from UCEC-EEA tumors than in the cancer-free uterine tissues (Fig. 1E; Supplementary Table S3). When we compared their clinical pathology index, the UCEC-EEA samples had larger fractions of positive cells (11%–50% positive cells) than did normal uterine tissue (0%–10% positive cells). Furthermore, two of the three UCEC-EEA samples and none of the normal uterine tissue samples had a high-intensity score (2 on a scale of 0–2). The whole-tissue images and PRLR protein density are shown in Supplementary Fig. S1C.

To address the therapeutic potential of the PRL antagonist G129R in uterine cancer, we examined the effects of G129R in two different orthotopic uterine cancer models, Ishikawa (PTEN^mut) and Hec1A (PTEN^WT), to determine whether the PTEN status (28) affects the antitumor impact of antagonizing PRL in uterine cancer. On the basis of our previous in vivo dose-finding results for G129R in orthotopic mouse models (37), we chose 100 μg/day as the dose for the in vivo therapeutic experiments and followed the dosing schema shown in Fig. 2A. In both Ishikawa and Hec1A models, G129R monotherapy reduced tumor weights by 50% when compared with controls (**, P < 0.001; Fig. 2B and C, top). Besides the tumor burden, the average number of total tumor nodules collected from each mouse was also significantly lower in G129R treatment groups than in controls, in both models (**, P < 0.001; Fig. 2B and C, middle). In both models, G129R had no obvious toxicities, as shown by the stability of mouse body weights in each group (ns, nonsignificant; Fig. 2B and C, bottom).

Given that taxane-based chemotherapy is used frequently to treat uterine cancer, we also compared the therapeutic effects of paclitaxel monotherapy with the combination of G129R + paclitaxel. Results showed that the G129R + paclitaxel combination resulted in significantly lower (<50%) tumor weights than paclitaxel monotherapy in both models (*, P < 0.05; paclitaxel vs. G129R + paclitaxel; Fig. 2B and C, top). The average sizes of primary tumors at the uterine horn of the Ishikawa model in all three treated groups (paclitaxel, G129R, and G129R + paclitaxel) were comparable, but smaller than those of the control group that received mannitol (Supplementary Fig. S3). Also, groups treated with paclitaxel, G129R, or G129R + paclitaxel in combination had minimal tumor nodules (n ≤ 5), and we did not observe any statistical differences in the total numbers of tumor nodules between groups receiving treatments of paclitaxel and combination G129R + paclitaxel in either model (ns, nonsignificant; Fig. 2B and C, middle). The differences in mouse body weight between the G129R + paclitaxel combination group and the paclitaxel monotherapy group were not significant in either model (Fig. 2B and C, bottom), indicating that the G129R + paclitaxel combination treatment did not incur additional toxicity in comparison with paclitaxel treatment.

To further assess the effects of G129R on the distribution of primary and metastatic tumor nodules, we recorded and compared the numbers of nodules and their locations during necropsy. In comparison with the control mannitol treatment, G129R significantly reduced almost all metastatic tumor nodules located at the pelvic sidewall, omentum, diaphragm, pelvis, and spleen in both the Ishikawa and Hec1A models (Fig. 2D and E). Combination G129R + paclitaxel treatment also resulted in significantly fewer metastatic nodules located at the mesentery, peritoneum, pelvis sidewall, diaphragm, omentum, spleen, kidney, and porta hepatis than control groups (Fig. 2D and E). To explore the molecular mechanisms underlying the effects of blocking PRL/PRLR activity in uterine cancer, we performed RPPA analysis to compare the protein expression profiles of the lysates extracted from size-matched tumors in Ishikawa model treated with mannitol or G129R. Ingenuity Pathway Analysis (IPA) of the RPPA results showed that the PI3K/mTOR pathway and regulation of the epithelial–mesenchymal transition pathway were among the top canonical pathways that were downregulated by G129R (Fig. 2F; Supplementary Table S2).

To explore the pharmacodynamic effects of G129R treatment on uterine cancer cells, we first determined the IC₅₀ values of G129R in Ishikawa and Hec1A cells. We precultured the cells in regular medium containing FBS for 2 days and then switched to medium containing charcoalstripped FBS (hormone-depleted conditions; ref. 38). Next, cell viability was assessed by using the MTT assay, and the results showed that both Ishikawa and Hec1A cells were sensitive to G129R at a 10 μg/mL concentration (Supplementary Fig. S4A and S4B). Ishikawa and Hec1A cells were then treated with mannitol as control (10 μg/mL), PRL (0.1 μg/mL), G129R (10 μg/mL), or combined PRL + G129R (same concentrations) for 24 hours, and proliferating cells were quantified by the EdU incorporation assay. In both Ishikawa and Hec1A cells, the proliferative cell populations represented by EdU⁺ under PRL treatment were about 0.6-fold (Ishikawa, Fig. 3A) or 1.8-fold (Hec1A, Fig. 3B) higher than that in mannitol-treated cells. However, G129R treatment or combination with PRL significantly reduced cell proliferation stimulated by PRL in both Ishikawa cells (Fig. 3A; *, P < 0.05; **, P < 0.01) and Hec1A cells (Fig. 3B; *, P < 0.05; **, P < 0.01). Moreover, G129R decreased the viability of both Ishikawa and Hec1A cells cultured in hormone-depleted conditions in comparison with PRL stimulation (Supplementary Fig. S5A and S5B; *, P < 0.05; **, P < 0.001). We next examined the effects of G129R on cell-cycle progression in uterine cancer cells. G129R alone or in combination with PRL led to an increase in the sub-G₀-phase population to 20- to 30-fold higher in Ishikawa cells (Fig. 3C) and approximately 3-fold higher in Hec1A cells (Fig. 3D) cultured under hormone-depleted conditions (39).

To identify the mechanisms underlying G129R's effect, we performed another set of RPPA analyses to analyze the protein profiles of the lysates extracted from Ishikawa and Hec1A cells treated with mannitol (10 μg/mL), PRL (0.1 μg/mL), G129R (10 μ g/mL), or PRL + G129R for 24 hours. G129R alone or PRL + G129R significantly decreased expression of factors associated with the PI3K/Akt signaling pathways, but increased abundance of both total and phosphorylated FOXO3a and EIF-4EBP1, a transcriptional repressor. Inhibition of PI3K/Akt is known to upregulate and activate multiple receptor tyrosine kinases including the FOXO-dependent pathway (ref. 40; Fig. 4A). The functions of FOXO3a in the cell cycle, metabolism, and cell death are closely related to its nuclear distribution (41). We utilized multi-photon confocal microscopy to examine the effects of G129R alone or PRL/G129R on subcellular localization of FOXO3a and PRLR in Ishikawa and Hec1A cells. We observed a predominantly cytoplasmic distribution of FOXO3a (as shown by red arrows, Fig. 4B) and membrane-associated PRLR expression when the cells were treated with mannitol or PRL. However, treatment with G129R alone or in combination with PRL resulted in substantial high levels of nuclear FOXO3a in these two uterine cancer cell lines (as shown by white arrows, Fig. 4B).

To validate this finding, we collected subcellular fractions of Ishikawa and Hec1A cells under hormone-depleted conditions and treatment with mannitol (10 μg/mL) as control, PRL (0.1 μg/mL), G129R (10 μg/mL), or combined PRL + G129R. In comparison with PRLR_LF (at ∼100 kDa), PRLR_SF was expressed predominantly as approximately 40 kDa in total cell lysates or cytoplasmic fractions of Ishikawa (Fig. 4C, top) and Hec1A (Fig. 4D, top) cells; levels of PRLR_SF were not changed when cells were treated with mannitol, PRL, G129R, or PRL + G129R, implying that the activity of PRL/PRLR_SF rather than the receptor itself was affected by G129R (Fig. 4C and D, top). Phosphorylated PI3K and Akt were among the factors downregulated by G129R alone or in combination with PRL, while FOXO3a and EIF-4EBP1 were upregulated by G129R.

Furthermore, in comparison with mannitol-treated cells, we observed a slightly reduced level of FOXO3a in the nuclear fraction (NER) of Ishikawa (0.9-fold) and Hec1A (0.8-fold) cells under PRL treatment, but a much higher increase of FOXO3a in Ishikawa and Hec1A NERs under G129R or PRL/G129R treatment (Fig. 4E and F). Phosphorylated (pSer574) FOXO3a levels from NERs were also higher in Ishikawa cells (5.2-fold) and Hec1A cells (1.5-fold) when treated with G129R. Furthermore, in comparison with mannitol-treated NERs, we observed a similar level of EIF-4EBP1 in the Ishikawa NER (1.0-fold) and a 2.8-fold increased EIF-4EBP1 in Hec1A NER under PRL treatment. Meanwhile, the EIF-4EBP1 levels were 2.2-fold higher in Ishikawa NER and 2.6-fold higher in Hec1A NER under G129R treatment, and 2.9-fold higher in Ishikawa NER and 5.2-fold higher in Hec1A NER under PRL/G129R treatment (Fig. 4E and F). Phosphorylated (pSer65) EIF-4EBP1 levels from NERs also reached peak-high in Ishikawa and Hec1A NERs when treated with G129R or PRL/G129R (Fig. 4E and F). LaminB1 was applied as a loading control for NER in both cell lines, but due to faster proliferation rate and lower senescence in Hec1A cells than Ishikawa cells, the levels of LaminB1 in whole-cell lysate from Ishikawa cells were lower than in Hec1A cells (42). The relative ratio of density for each comparison was quantified by ImageJ software and is represented in Fig. 4E and F.

Collectively, these results showed that blockade of the PRL/PRLR axis with G129R decreased cell viability, inhibited proliferation, and increased the G₀-phase population. Treatment with G129R also led to elevated expression of nuclear FOXO3a and EIF-4EBP1.

To investigate the roles of FOXO3a and EIF-4EBP1 in the inhibitory effects of G129R in uterine cancer, we employed two sets of siRNAs to knockdown the expression of FOXO3a and EIF-4EBP1 in Ishikawa and Hec1A cells. SiRNA1 against EIF-4EBP1 and siRNA1 against FOXO3 yielded more than 90% knockdown compared with scrambled RNA (sc-siRNA) and siRNA2s (Supplementary Fig. S6A and S6B). We next determined the effects of G129R on the proliferation of Ishikawa and Hec1A cells pretreated with sc-siRNA, siRNA-FOXO3, or siRNA-4EBP1. Different from sc-siRNA–pretreated Ishikawa cells, in which G129R significantly reduced the EdU⁺ population in comparison with mannitol-treated cells (**, P < 0.01) or PRL-treated cells (***, P < 0.001), there were no significant changes between G129R-treated cells and mannitol- or PRL-treated cells when the cells were pretreated with siRNA-FOXO3 or siRNA-4EBP1 (Fig. 5A). Similar results were observed in Hec1A cells (Fig. 5B). It is also noteworthy that siRNA knockdown in both Ishikawa and Hec1A cells reduced cell proliferation in general, which possibly is attributable to the impaired S-phase entry and G₀-phase accumulation (43). Knockdown of 4EBP1 also showed similar effects on reducing cell proliferation in comparison with sc-siRNA in both uterine cancer cell lines independent of PTEN status, which is consistent with a previous report in breast cancer cells (44).

We further investigated the effects of G129R on cell-cycle progression as reflected by G₂–M-phase and sub-G₀-phase populations. In Ishikawa cells (Fig. 5C), G129R alone or in combination with PRL significantly reduced the G₂–M-phase population (indicated by open bars in the figure) to a lower level than mannitol treatment in both the nontransfected and sc-siRNA–pretreated groups, but did not have such an effect in either the FOXO3-knockdown (siRNA-FOXO3) or EIF-4EBP1–knockdown (siRNA-4EBP1) groups (P < 0.05). The sub-G₀-phase populations (gray pattern bars) induced by G129R alone or PRL + G129R were approximately 4.5-fold higher than those treated with mannitol or PRL in Ishikawa nontransfected (control) or sc-siRNA–transfected cells; however, the sub-G₀-phase populations induced by G129R alone or PRL + G129R did not differ from those induced by mannitol or PRL in Ishikawa siRNA-FOXO3- or siRNA-4EBP1–knockdown cells. In Hec1A cells (Fig. 5D), in nontransfected (control) or sc-siRNA–transfected cells, PRL treatment increased the G₂–M-phase population (open bars) by about 2-fold in comparison with mannitol-treated cells, but G129R alone or in combination with PRL significantly reduced the G₂–M-phase population to lower levels than PRL treatment (P < 0.05). This effect of G129R was reversed by siRNA-FOXO3 or siRNA-4EBP1; there was no significant difference in the G₂–M-phase population after PRL, G129R, or PRL + G129R treatment. The sub-G₀–G₁-phase populations (gray pattern bars) in Hec1A nontransfected (control) or sc-siRNA–transfected cells treated with G129R or PRL + G129R were approximately 3-fold higher than in cells treated with PRL; however, the sub-G₀-phase populations induced by G129R alone or PRL/G129R did not differ from those induced by PRL in Hec1A siRNA-FOXO3- or siRNA-4EBP1–knockdown cells.

To determine the effects of G129R on this EIF-4EBP1- and FOXO3a-mediated translational suppression in uterine cancer cells, we treated Ishikawa and Hec1A cells (sc-siRNA, siRNA-FOXO3, or siRNA-4EBP1) with mannitol (10 μg/mL) as control, PRL (0.1 μg/mL), G129R (10 μg/mL), or combined PRL + G129R for 24 hours. Using an AHA incorporation assay, which measures active protein synthesis by detecting the incorporation of methionine into proteins, we compared the effects of PRL stimulation and inhibition on protein synthesis by determining the ratios of nascent protein in treated cells and controls. In Ishikawa sc-siRNA cells, G129R alone or in combination with PRL significantly reduced protein synthesis (by ∼40%) in comparison with PRL alone (Fig. 6A). While knockdown of FOXO3a in Ishikawa cells increased global protein synthesis by more than 50%, treatment of these FOXO3a-knockdown Ishikawa cells with G129R alone or PRL + G129R did not reduce protein synthesis to a greater extent than PRL or mannitol (Fig. 6A).

EIF-4EBP1 inhibits initiation of translocation by binding to eukaryotic initiation factor-4E (eIF4E) and preventing recruitment of the translational machinery (45). As expected, knockdown of EIF-4EBP1 in Ishikawa cells increased the nascent protein synthesis (>2-fold) compared with sc-siRNA Ishikawa cells when treated with control (mannitol; P < 0.01; Fig. 6A). In Ishikawa EIF-4EBP1–knockdown cells, G129R alone or in combination with PRL did not change protein synthesis to a greater extent than PRL or mannitol (Fig. 6A). Similar effects were noted in Hec1A cells (Fig. 6B; *, P < 0.05 compared with PRL-treated sc-siRNA cells with G129R or G129R/PRL-treated sc-siRNA cells; **, P < 0.01 compared with mannitol-treated sc-siRNA cells with siRNA-4EBP1 cells). Hec1A cells, in which FOXO3a or EIF-4EBP1 was knocked down, synthesized significantly more protein than Hec1A sc-siRNA cells (*, P < 0.05; **, P < 0.01; ns, not significant) and the inhibitory effects of G129R on protein synthesis in sc-siRNA cells were abrogated by depletion of functional FOXO3a or EIF-4EBP1 (Fig. 6B; ns, not significant compared with PRL-treated siRNA-FOXO3a cells with G129R- or G129R/PRL-treated siRNA-FOXO3a cells). Taken together, our preclinical results suggest that G129R has a novel inhibitory mechanism as an antagonist to the PRL/PRLR_SF complex in uterine cancer cells (depicted in Fig. 6C).

Recent studies have shown that PRL is closely associated with malignancy, particularly in gynecologic cancer (3), and is a discriminative biomarker for early detection of endometrial cancer (4). In this study, we discovered that a short PRLR isoform, PRLR_SF, is the predominantly expressed isoform in human uterine cancer. Furthermore, our results reveal a novel mechanism of action of G129R in blocking the PRL/PRLR_SF axis in uterine cancer cells. When G129R competes with PRL and binds to PRLR_SF, it blocks cell proliferation and progression of the cell cycle and reduces PI3K/Akt activity. In addition, G129R treatment accelerates the nuclear translocation of FOXO3a, which releases the activated form of EIF-4EBP1 to initiate translational suppression and reduce nascent protein synthesis.

The expression of diverse PRLR isoforms in a tissue-specific manner, resulting in different signaling activities, has been reported previously (46). We observed a high level of pSer65-4EBP1 in lysates of both Ishikawa and Hec1A cells, and because pSer65-4EBP1 has been reported to prevent recruitment of the translational machinery (45), we speculate that PRLR_LF transcripts could be blocked from translation, while PRLR_SF may still undergo translation given that it has the shortest transcript of the three transmembrane isoforms of the PRLR gene (23, 24). The unique PRLR isoform expression in uterine cancer led us to target the PRL/PRLR axis with a PRL antagonist, G129R. Our RPPA results show that FOXO3a and translation suppressor EIF-4EBP1 played critical roles in mediating the inhibitory effects of G129R in proliferation and cell-cycle progression of uterine cancer cells, which is quite different from the previously reported effects of PRLR_LF in mediating autophagic cell death (37). Our studies also suggest FOXO3a and EIF-4EBP1 as potential prognostic markers for the clinical application of G129R in uterine cancer. In future studies, the dynamics of cellular responses to G129R binding to PRLR_SF should be characterized in detail, including the ligand-initiated endocytic signaling pathways upon binding (47) in uterine cancer cells.

PRLR expression in tumor stromal cells could biologically relate to tumorigenesis. Both endocrine and autocrine/paracrine activities of PRL are involved in organ development and tumor growth/progression. Interestingly, PRLR_SF was reportedly the isoform predominantly expressed in endothelial cells derived from microvascular and macrovascular origins of both endocrine and nonendocrine organs (37). Blocking the activity of the PRL/PRLR axis in endothelial cells with a receptor-specific antagonist such as a G129R derivate markedly reduced PRL-induced angiogenic signaling of endothelial cells (48). However, the mechanism whereby PRL antagonism inhibits the angiogenic properties of endothelial cells has remained unclear. Paik and colleagues reported that FOXO3a^−/− mouse models of gynecologic malignancies had higher rates of pituitary adenoma and vascular abnormalities than FOXO3a^WT mice (49). In contrast, EIF-4EBP1 has been reported in several studies to show a strong tumor-suppressive effect in compensating for hypoxia-increased protein synthesis in tumor and tumor-associated endothelial cells (46, 50), and knockdown of EIF-4EBP1 in cancer-associated fibroblasts abrogated tumor chemoresistance (51). Therefore, our functional studies showing G129R-induced increases in nuclear FOXO3a expression and in EIF-4EBP1 activities provide new hints about the stromal effects of PRL antagonism. In summary, this study reported a unique pattern of PRLR_SF expression in uterine cancer and prognostic potential for FOXO3a and EIF-4EBP1 in initiating cell death following treatment with G129R in uterine cancer.

A.K. Sood reports personal fees from Merck and Kiyatec, stock ownership from BioPath (shareholder), and grants from Esperance and MTrap outside the submitted work. No potential conflicts of interest were disclosed by the other authors.

Y. Wen: Conceptualization, resources, data curation, software, formal analysis, supervision, funding acquisition, validation, investigation, visualization, methodology, writing-original draft, project administration, writing-review and editing. Y. Wang: Data curation, software, formal analysis. A. Chelariu-Raicu: Data curation, investigation, methodology. E. Stur: Investigation, methodology. Y. Liu: Investigation, methodology. S. Corvigno: Formal analysis, investigation, methodology. F. Bartsch: Investigation, writing-original draft, writing-review and editing. L. Redfern: Investigation, methodology. B. Zand: Methodology. Y. Kang: Investigation, methodology. J. Liu: Resources. K.A. Baggerly: Resources, data curation, software. A.K. Sood: Conceptualization, resources, supervision, funding acquisition, writing-review and editing.

Portions of this work were supported by the Department of Defense Ovarian Cancer Research Program W81XWH2010335 (Y. Wen), NIH Uterine SPORE P50CA098258 (to Y. Wen and A.K. Sood), National Comprehensive Cancer Network (to Y. Wen), Marsha Rivkin Center for Ovarian Cancer Research (to Y. Wen), the NIH P50CA217685 (to A.K. Sood), R35CA209904 (to A.K. Sood), P30CA016672 (used for the Bioinformatics Shared Resource and the Functional Proteomics RPPA Core Facility), the American Cancer Society Research Professor Award (to A.K. Sood), the Frank McGraw Memorial Chair in Cancer Research (to A.K. Sood), and the Blanton-Davis Ovarian Cancer Research Program (to Y. Wen and A.K. Sood). A. Chelariu-Raicu was supported by funds from Deutsche Forschungsgemeinschaft (CH 1733/1-2). E. Stur is supported by Ovarian Cancer Research Alliance (OCRA number FP00006137). In addition, we thank Dr. Walter N. Hittelman at MD Anderson Cancer Center for help with multiphoton confocal imaging, Donald Payne and Michael Redman at Oncolix Inc. for supplying G129R, and Kathryn Hale and Sunita Patterson for their editorial work.

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.