INTRODUCTION
One of the most intriguing questions in mammary gland biology is why prolactin (PRL) which is clearly involved in normal breast development in rodents and humans, and clearly plays a role in rodent mammary cancer [1, 2] has not been accepted as a player in human breast cancer. For any hormone, such as estrogen, to be accepted as playing a role in breast cancer, three criteria are applied. First, there must be specific receptors for the hormone on the cancer cells. Second, the hormone must induce a biological response. Third, the course of the disease must be altered when the action of the hormone is specifically inhibited at the target (as is the case with antiestrogens and estrogen effects), or the source of the hormone is removed, ie. through ovariectomy. When these same criteria are applied to PRL, a role for this hormone in breast cancer can be argued.
CRITERION 1: PROLACTIN RECEPTORS
The presence of specific receptors for PRL has been demonstrated in both normal and malignant mammary glands. PRL receptors belong to the cytokine hematopoietic family of receptors [3]. The members of this superfamily are single membrane-spanning receptors organized into three domains comprising an extracellular ligand binding domain, a hydrophobic transmembrane domain and an intracellular domain containing a proline rich motif. The three different forms of the PRL receptor differ in their cytoplasmic domain (Fig. 1). The long form of the PRL receptor is able to induce differentiation as measured by induction of milk protein gene expression [4]. The short form of the receptor acts as a negative regulator of PRL-induced differentiation [5]. We have shown [6] that the short form of the mouse PRL receptor, as the long form [7], is able to induce mitosis, but not differentiation
PRL receptors have been demonstrated in over 70% of breast biopsy samples by specific binding assays [8-11]. Mammary cells, both normal and malignant, contain long and short forms of the receptor. The ratio of long and short forms is unknown. In human breast cancer, the correlation of disease parameters with hormone binding to ER and PR is clear. However, it is less clear for the PRL receptor because of its multiple size and charged forms. Interaction of PRL with its receptor induces dimerization of the membrane associated receptor [12]; a variety of dimeric combinations are possible. The physiological significance of homo- vs. heterodimerization is an open question.
Our preliminary data [13] show that more than 90% of breast cancer surgical samples were positive for PRL receptor mRNA, but the amount varies considerably. No mRNA for the intermediate form of the receptor was detected in our samples in contrast to the report by Clevenger [14]. No correlation was found between the presence of mRNA for PRL receptors and ER status. The cancerous tissue contained significantly more message for the PRL receptor than did adjacent, non-involved tissue from the same patient [13]. Using in situ hybridization, PRL receptor mRNA was found in normal breast, inflamatory lesions (mastitis), benign proliferative breast disease (fibroadenoma, papilloma, adenosis, epitheliosis), intraductal carcinoma or lobular carcinoma in situ, and invasive ductal, lobular or medullary carcinoma [15]. No correlation with the level of PRL receptor mRNA was found according to the histological type of lesion. These findings, as well as the presence of PRL receptors in breast stromal cells, were confirmed by immunohistochemistry.
CRITERION 2: BIOLOGICAL RESPONSE
In the mammary gland, PRL is both a differentiating agent and a mitogen. It is the mitogenic action which is important in breast cancer. In the normal gland, PRL's most fundamental actions are at the level of growth and maintenance of the morphology of the mammary gland. Although the ovarian steroids, estrogen and progesterone, are involved in the ductal growth and branching; lobuloalveolar development and extensive growth of the alveolar cells require PRL [16, 17]. The effects of PRL on cell growth are synergistic with the effects of progesterone [18]. The steroid hormone appears to act, in part, by increasing the level of PRL receptors.
In order to firmly establish the mitogenic action of PRL in mammary cells, its effects on cells in culture needed to be examined. Primary breast biopsy samples grown in nude mice respond to lactogenic hormones with increased growth [19]. Malarkey et al. [20] found that physiological levels of human PRL and human growth hormone (hGH) increased the population doubling of primary breast tumor cultures. Both T47D and MCF-7 human breast cancer cells respond to PRL stimulation when grown as solid tumors in nude mice [2].
Direct effects of PRL on growth of human breast cancer cell lines in culture can be demonstrated only under proper conditions. PRL-stimulated growth of MCF-7 cells was greater in the presence of 1% charcoal stripped serum (CSS) compared to 10% CSS [21]. The maximal effect was observed at 100-250 ng/ml human PRL. Growth was also stimulated by hGH, human placental lactogen and ovine PRL but required higher concentrations to achieve the same effect (Fig. 2). Thus MCF-7 cells were more sensitive to the mitogenic effect of human PRL than to other lactogens. Bovine PRL had no effect on the growth of these human breast cancer cells. Charcoal stripping removes bovine lactogens from serum [21].
To date, 80% of human breast cancer cell lines examined respond to PRL as a mitogen. The ER positive MCF-7, ZR-75.1 and T47D cells [22] all respond to PRL in the presence of CSS. Lemus-Wilson et al. [23] also showed that PRL acts as a mitogen in MCF-7 and ZR-75.1 when grown in serum-free media. The ER negative cell line, T47Dco also showed an increase in growth in the presence of PRL when grown in the presence of CSS [24] or under serum-free conditions [25].
Effects of PRL on the growth of breast cancer cells are modulated by the presence of other growth factors and hormones. Melatonin, the primary hormone from the pineal gland, completely blocked human PRL induced growth of MCF-7 and ZR-75.1 cells [23]. Bovine PRL when added simultaneously with human PRL, blocked the effect of human PRL on the growth of MCF-7 cells [21]. As little as 50-100 ng/ml of bovine PRL was able to block the human PRL induced growth of these cells (Fig. 3). In contrast, Lochnan et al. [26] reported that bovine PRL was equally potent with human PRL in the stimulation of the ß-casein promoter in CHO cells transfected with the human PRL receptor gene. Bovine PRL also is an effective mitogen in normal mouse mammary cell lines. The ability of bovine PRL to act as an antagonist of human PRL may be unique to the mitogenic action of PRL in human breast cancer cells.
CRITERION 3: RESPONSE TO HORMONE ABLATION
Historically, the third criterion has been the most difficult to satisfy in terms of PRL's action in human breast cancer. From 60-85% of human breast cancer biopsies contain immunologically detectable PRL [27, 28] while specific PRL receptors have been demonstrated in over 70% of the biopsy samples [8-11]. However, there is no clear correlation between circulating PRL levels [29-31] or PRL receptor content [32] and the etiology or prognosis of the disease (Reviewed by Vonderhaar [22]).
When patients were treated with PRL inhibiting ergot drugs that significantly diminish circulating pituitary PRL, no change in disease state was observed [33, 34]. Operating on the assumption that the lack of effect may have been due to the presence of hGH, which is also a lactogen, Manni et al. [35] administered a combination therapy of bromocriptine and somatostatin to a group of women with advanced breast cancer. Circulating levels of PRL, detected by a single assay, were abolished nearly completely in 8 of 9 patients, whereas hGH levels were suppressed in 7 of 9 patients during treatment. Although overall antitumor effects could not be assessed reliably because the patients entering the study had been pretreated heavily with chemotherapeutic agents, only one patient experienced disease stabilization. In a similar study, Anderson et al. [36] treated patients long-term with bromocriptine and the long-acting, superpotent somatostatin analogue, octretide, and found that there was no evidence of disease progression for periods up to 6 months in 4 of 6 patients with advanced breast cancer who had failed first and second line endocrine therapies. While immunoreactive PRL, GH and IGF-I in 24 hr profiles of serum were greatly reduced by these treatments, diurnal peaks of bioactive lactogenic hormone as well as GH levels were still apparent although much reduced.
Thus the lack of a definitive correlation between PRL levels and suppression of pituitary synthesis with prognostic and/or diagnostic parameters in human breast cancer has resulted in this hormone's potential role in the disease being overlooked. Recently, however, our laboratory and others have found an explanation which may resolve this problem, ie. breast cancer cells make their own PRL and it may act as an autocrine/paracrine factor in human breast cancer.
Although human PRL was first characterized as a 22-25 kDa pituitary hormone, in recent years synthesis of PRL and PRL-like molecules by a variety of tissues other than the pituitary has been reported [37]. Endogenous production of PRL by breast tumors has long been suspected. Following surgical removal of the pituitary in humans, the circulating levels of all pituitary hormones become undetectable, except for PRL. Circulating levels of PRL remained at 30-80% of the pre-surgical levels for as long as 10 months in breast cancer patients who received a total hypophysectomy [38]. Patients given bromocriptine plus somatostatin persistently maintained low levels of circulating bioactive PRL [36]. While these data could result from other PRL-like molecules circulating in the blood, it was also possible that PRL itself was produced by peripheral tissues. Both normal tissue and tumors appear to generate this hormone. PRL is also produced by the brain, placenta, uterus, dermal fibroblasts and the immune system [39-42].
Steinmetz et al. [43] first showed by in situ hybridization, that PRL gene transcripts are present in secretory mammary epithelial cells in pregnant rats. PRL mRNA has also been demonstrated by both Northern analysis and PCR in mammary glands from lactating rats [44], goats and sheep [45] suggesting local synthesis of PRL mRNA by this gland. More recently, the presence of PRL was demonstrated by PCR in some primary human breast carcinomas [13, 14] and in human breast cancer cells in culture in our laboratory [46].
We determined that PRL is synthesized by human breast cancer cells and acts in an autocrine manner to stimulate cell proliferation with a panel of anti-PRL monoclonal antibodies and by use of the antisense technique. Using a panel of monoclonal antibodies against human PRL, proliferation of both T47Dco and MCF-7 human breast cancer cell lines was inhibited by 20-90% [46]. In addition, when T47Dco cells were treated with antisense RNA directed against the gene encoding for pituitary PRL, significant growth inhibition (>50%) was obtained [47]. RT-PCR, followed by Southern analysis using human PRL cDNA as the probe, confirmed the presence of PRL mRNA in T47Dco and MCF-7 cells [46]. In all, over 80% of all breast cancer cell lines tested contained PRL mRNA. In addition, we found that 100% of primary breast cancer biopsies also contained mRNA for PRL [13]. When human breast cancer cells are grown as solid tumors in nude mice, the resulting tumors contain PRL gene transcripts (Fig. 4) [47].
That the protein is actually synthesized and secreted by the cells in culture was confirmed by metabolically labelling T47Dco cells with 35S-cysteine. Conditioned media and cell extracts both contain a 22kDa protein precipitated by an anti-human PRL monoclonal antibody [46]. Conditioned media prepared from T47Dco cells stimulated the PRL responsive Nb2 rat lymphoma cells to grow in a concentration dependent manner. These cells respond to picogram quantities of lactogens. The level of biological activity in the conditioned media is equivalent to 0.7 ug/ml (14.5 pg PRL/cell) of pituitary PRL as measured by the Nb2 assay and is approximately 30% of the amount normally produced by the rat pituitary cell line, GH3 [48, 49]. By using a specific RIA for human pituitary PRL, 0.35ug/ml of PRL protein was detected. The activity in the conditioned media, like that of the human pituitary PRL, was abolished when the media were pretreated with an antiPRL antibody [46].
Our preliminary data in the human breast cancer cell lines found no correlation of ER status with PRL receptors or with the ability to synthesize PRL [13]. In addition, we found that 73% of primary breast cancer surgical samples also contained mRNA for PRL. In the majority of cases, the amount of mRNA for PRL and its receptors is significantly elevated in cancerous vs. the adjacent, non-involved tissue from the same patient. There was no correlation between the presence of PRL mRNA and ER status [13].
The nature of the post-translational alteration of PRL synthesized by the mammary gland and breast cancer cells is unknown. Our observations that a panel of monoclonal antibodies directed against pituitary PRL vary in their ability to recognize the PRL produced by breast cancer cells [46] suggests that there are marked differences in post-translational modifications between the pituitary and the mammary gland. PRL achieves its many diverse activities through different post-translational modifications such as glycosylation and phosphorylation [50, 51].
The hormones and growth factors which regulate expression of PRL by human breast cancer cells in vivo and in vitro are also currently unknown. Recent studies [52] have shown that PRL synthesis is regulated in T47D cells by the distal promoter used by decidua and lymphocytes rather than the proximal promoter used by the pituitary.
CONCLUSION
The demonstration of synthesis of PRL by human breast cancer cells suggests that manipulation of pituitary PRL is not a valid approach to therapy for this disease and that new approaches based on the concept of an autocrine/paracrine PRL may be necessary. These observations suggest that the use of specific drugs [24, 25] or analogues of the hormone which antagonize PRL action at the target tissue [53-57] presents a viable alternate approach to therapy for diseases that are PRL sensitive.