Clinical association of progesterone receptor isoform A with breast cancer metastasis consistent with its unique mechanistic role in preclinical models

Breast oncogenesis may occur over decades. The majority (> 78%) of initial breast cancer diagnoses are in women over 50 years of age [1] and median age, 61 years [2]. The majority of breast cancer (BCa) cases are of the luminal subtype (L-BCa), which is estrogen receptor (ER)-positive [3] and continues to express ER into advanced stages [4]. Primary L-BCa is exquisitely sensitive to anti-estrogens. Unfortunately, L-BCa has frequently metastasized prior to diagnosis and the metastatic tumors may also appear long after cessation of therapy [5, 6]. At least 20% of BCa patients have ER+ tumors showing distal metastasis [7]. Metastatic L-BCa is typically incurable. Our previous studies have mechanistically linked invasiveness and metastasis of L-BCa to a novel hormonal mechanism [8, 9].

Growth of ER+ breast tumors is supported by estrogen but the hormone suppresses tumor invasiveness, regardless of hormone sensitivity of growth and suppresses tumor progression [10‐16]. In contrast, progestins support invasiveness and metastasis in L-BCa cells [17, 18]. A single gene encodes two isoforms of the progesterone receptor (PR), PR-A and PR-B, via alternative promoter usage. PR-B only differs from PR-A in having an amino-terminal segment of 164 residues containing the activation function, AF3 [19]. Despite overlap in the gene regulatory patterns of PR-B and PR-A, the two isoforms also show a clear distinction in terms of the strength of this regulation and also in the target genes that they activate or repress [19‐21]. The two PR isoforms can heterodimerize and regulate a small set of unique genes [20, 22]. PR-A and PR-B show similar levels of expression in normal breast but breast oncogenesis is associated with loss of this coordinate expression, mostly resulting in a high PR-A:PR-B ratio, in early as well as progressed lesions [23]. The higher ratio of PR-A to PR-B is generally due to PR-A overexpression and has been reported to be negatively associated with disease free survival (DFS), although the mechanistic underpinnings were not established [24].

The pre-menopause plasma estrogen level is 1.4 nM − 1.6 nM (follicular phase), and 3.6 nM - 4.2 nM (luteal phase) [25]. Progesterone levels in the plasma range from < 4 nM in the follicular phase, up to > 50 nM in the luteal phase [26]. After menopause, the circulating hormone levels show a marked decline (median value for estrogen of 0.14 nM and median value for progesterone of 0.13 nM), but hormone levels of up to ~ 1 nM may be retained in breast tissue [27, 28]. Previous in vitro studies in the literature had suggested progesterone induces invasiveness of BCa cells principally via PR-B [29, 30] although clinical observations noted more frequent elevation of PR-A in ductal carcinoma in situ (DCIS) and invasive breast lesions [23] and associated PR-A rather than PR-B with lower DFS [24]. Notably, those in vitro studies were conducted using relatively high progesterone concentrations (corresponding to luteal stage, pregnancy or pharmacological plasma levels) and excluded estrogen signaling [29‐33].

We have reported studies covering the full ranges of pre-menopausal and post-menopausal levels of estrogen and progesterone, conducted with both estrogen and progesterone signaling present, using T47D, ZR-75-1 and BT474 cells as well as isogenic recombinant T47D cells expressing a single PR isoform [8]. Previous studies of the effects of high dose progesterone on invasion and metastasis utilized models in which estrogen signaling was absent. Therefore we took into account the possibility that cross-talk with estrogen signaling may influence regulation of invasiveness by progestins at physiological hormone levels. At concentrations < 0.01 nM, estrogen strongly suppressed invasiveness of ER+ BCa cells. At relatively low concentrations (< 1 nM), progestins abrogated inhibition of invasiveness caused by estrogen. It took relatively higher concentrations of progestins, (in the range of 5 nM - 50 nM) to induce invasiveness in the absence of estrogen, in a progestin dose-dependent manner. It was the PR-A isoform that mediated rescue of invasiveness from estrogen regulation by progesterone and this effect was uninfluenced by PR-B [8]. In contrast, estrogen-independent induction of invasiveness by progestin was mediated by PR-B and this occurred at either pregnancy-associated or pharmacological levels of progestin [8]. When PR-A was 2–3 fold overexpressed in PR-A+/PR-B+ cells, a lower concentration of progestin (< 0.2 nM) completely rescued invasiveness. These hormonal effects on invasiveness were independent of HER2 status. Thus progesterone supports BCa cell invasiveness by counteracting estrogen, acting exclusively through PR-A, at hormone concentrations corresponding to the full physiological ranges of estrogen and progesterone levels, both prior to and after menopause [8].

To interrogate the role of the ER-PR-A crosstalk in metastasis, we investigated selective crosstalk mechanisms of PR-A with ER via microRNAs (miRNAs), which have an extensive role in estrogen signaling [9]. Among miRNAs regulated by estrogen or progesterone, miR-92a-3p and miR-26b-5p mediated the PR-A–ER crosstalk. Progesterone plus PR-A suppressed induction of miR-92a-3p and repression miR-26b-5p by estrogen. This resulted in regulation of genes involved in invasiveness and metastasis and also completely rescued invasiveness in vitro. In murine xenograft models, when miR-92a-3p was constitutively expressed or when miR-26b-5p was inhibited, metastasis was profoundly suppressed, similar to knockdown of PR [9]. Notably, these two miRNAs did not regulate each other and, based on quantitative changes in mRNAs of genes involved in invasion and/or metastasis using a commercially available pathway cDNA array, we found that the two micro RNAs mediate independent hormone-regulated pathways controlling invasion and metastasis that are convergent. Therefore, the strong suppression of invasiveness by even very low concentrations of estrogen (~ 0.01 nM) may be the net effect of sub-optimal regulation of these two miRNAs by estrogen. In primary tumors, PR-A expression showed negative correlation with miR-92a-3p and positive correlation with miR-26b-5p, but the strength of this correlation was moderate, as expected because of variable stromal content in resected tumor specimens, the likely influence of variability in the range of plasma progesterone levels in postmenopausal women and differences in expression of a compliment of coregulators and downstream effectors. The mechanistic studies suggest that hormonal crosstalk of PR-A with ER enables metastasis of luminal BCa, notably by opposing regulation of critical microRNAs by estrogen. This occurs even at hormone levels corresponding to plasma hormone status in the follicular stage and post-menopause.

Public RNAseq databases on the transcriptome profiles of breast tumors, such as The Cancer Genome Atlas (TCGA), do not distinguish PR-A, because it has multiple transcripts arising from alternative promoter usage that only differ from the PR-B transcript with respect to the positions of their 5′ ends. Therefore, in this study, we undertook a prospective analysis to clinically validate the role of PR-A in metastasis. We further investigated the possibility of isoform-selective modulation of PR-A.

As described above, previous studies have rigorously elucidated a mechanistic model of breast cancer metastasis that should predict the manner in which expression of ER, PR-A and PR-B are associated with L-BCa metastasis (Fig. 1). In this model, PR-A is induced by estrogen and thus ER limits its own ability to suppress invasion and metastasis. Additionally, PR-A may also be further up-regulated in L-BCa via mechanisms that are independent of estrogen. PR-B on the other hand requires relatively high physiological levels of progesterone (premenopausal luteal stage and pregnancy associated levels) to induce metastasis. Therefore, the model predicts a strong clinical association of PR-A with metastasis of primary L-BCa but, at best, a weak association of PR-B with metastasis. The model also predicts that the association of PR-A with metastasis should be uninfluenced by the relative expression of PR-B as PR-B influences metastasis by a separate mechanism and only at high progesterone levels. At the same time the model predicts a likely negative association of ER with metastasis. The mechanistic studies also predict that expression of HER2 will not affect the association of PR-A with metastasis.

Diagnostic evaluation of PR status in primary breast cancer does not typically distinguish PR isoforms. However, two major clinical studies have suggested an association between PR-A and progression of L-BCa. In one study, coordinate expression of PR-A and PR-B observed in normal breast epithelial cells and in benign breast tumors was reportedly frequently disrupted in both DCIS and in invasive lesions, where a PR-A: PR-B ratio > 1 was more frequent than a ratio < 1 [23]. As the extent of dysregulated expression varied from cell to cell within the lesions, the authors suggested that loss of coordinate expression of the PR isoforms is an early event in the development of malignancy with potential implications in tumor progression [23]. Another study conducted using axillary node-positive breast tumors noted that an elevated PR-A: PR-B ratio was more frequently the result of overexpression of PR-A and that patients harboring primary tumors with higher PR-A: PR-B ratios were ~ 2.8 – fold more likely to relapse [24]. Our more recent in vitro mechanistic and pre-clinical model studies have rigorously established that PR-A causes invasiveness and metastasis in response to progesterone levels in its pre- and post-menopausal ranges by counteracting the effect of estrogen [8, 9]. Moreover, elevated expression of PR-A hyper-sensitizes the receptor to progesterone levels at the low end of its post-menopausal range [8, 9]. We have previously established and reported that in primary ER+ breast cancer tumors, the mRNA levels of PR-A and PR-B correlate with their respective protein levels [9]. The present study directly links this mechanistic role of PR-A with metastasis of L-BCa in a patient population while offering a basis for the previous clinical observations noted above.

The distinctive relationships of the expression levels of PR-A, PR-B and ER in primary L-BCa lesions with node involvement is entirely consistent with predictions based on the previously established mechanistic model (Fig. 1) in which PR-A and PR-B have independent roles in promoting invasiveness and metastasis, with PR-A alone being responsive to the entire range of pre- and post-menopausal progesterone levels, functioning by counteracting the influence of estrogen/ER. As predicted from the model, PR-A was strongly associated with node involvement both in a univariable analysis and after adjusting for age, pathologic type, tumor grade and PR-B. In contrast, PR-B did not show an association with node involvement. ER actually showed a negative correlation with node involvement in the univariable analysis but in the multivariable analysis, this association was weaker, suggesting a more complex role for ER in the pathophysiology of L-BCa. The patients included in this study had predominantly HER2-negative tumors and there was no apparent influence of HER2 on the association of PR-A with node involvement, consistent with the previous mechanistic studies.

Together with previous studies, this study supports the view that progesterone, acting largely through PR-A, contributes to metastasis of L-BCa. However, even in post-menopausal women, it would not be feasible to completely inhibit the actions of progesterone to suppress invasion and metastasis of L-BCa. This is because in these women progesterone is needed to check proliferative effects of estrogen on the uterus. In mice in which the PR-A isoform was selectively knocked out, estrogen and even progesterone (acting via PR-B) caused hyperplasia of the uterine epithelium [40]. However, this rodent model data on the physiological role of PR-A in the uterus do not necessarily appear to be able to predict the likely side effects of pharmacologically selectively modulating PR-A in humans to suppress metastasis of L-BCa, when the following observations are considered together. First, we noted in this study that the general transcriptional activities of PR-A and PR-B are differentially sensitive to mifepristone, with PR-A being inhibited in a much lower dose range of mifepristone than PR-B. Second, as previously reported by us, a low dose (1 nM) of mifepristone completely inhibited the ability of PR-A to rescue invasiveness of L-BCa cells from suppression by estrogen [8]. In contrast, under the same conditions, mifepristone had no effect on the estrogen-independent induction of invasiveness by PR-B [8]. Finally, in (premenopausal) women treated long-term daily with low dose mifepristone as a potential contraceptive, endometrial hyperplasia was not a significant concern (41). It is therefore possible that in humans, PR-A is selectively targeted by low dose mifepristone, with minimal side effects on the uterus. From a mechanistic standpoint, hormone-dependent transcriptional activities of PR-A and PR-B are quite distinct from each other both in their patterns of regulation of mRNAs [19‐21] and micro RNAs [9]. There is a clear structural basis for this functional difference between the PR isoforms, i.e., the amino-terminal 164 residues of PR-B that is lacking in PR-A. Therefore it should be possible in principle to develop selective modulators of PR-A that could potentially offer an effective means of intervention in progression of L-BCa in certain patient groups such as post-menopausal women at high risk.

Taken together, the current clinical study, our previous detailed mechanistic studies and other related studies in the literature strongly support the view that, throughout a woman’s pre- and post-menopausal years, PR-A is a major driver in promoting invasiveness and metastasis of L-BCa by suppressing specific aspects of estrogen/ER action. The studies also suggest that PR-A isoform specific modulation is achievable as a possible intervention for progression of L-BCa in clinically appropriate situations.