This Article Abstract Full Text (PDF) All Versions of this Article: 26/3/400    most recent Final Manuscript Author Manuscript Purchase Article View Shopping Cart Alert me when this article is cited Alert me if a correction is posted Citation Map Services Email this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Request Copyright Permission Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Goffin, V. Articles by Kelly, P. A. Search for Related Content PubMed PubMed Citation Articles by Goffin, V. Articles by Kelly, P. A.

Development and Potential Clinical Uses of Human Prolactin Receptor Antagonists

Institut National de la Santé et de la Recherche Médicale, Unit 584 (V.G., S.B., P.T., P.A.K.), Paris Cedex 15, France; Université René Descartes-Paris V (V.G., S.B., P.T., P.A.K.), 75730, Paris Cedex 15, France; and Service d’Endocrinologie et Médecine de la Reproduction (P.T.), Hôpital Necker-Enfants Malades, 75743 Paris Cedex 15, France

Correspondence: Address all correspondence and requests for reprints to: Dr. Vincent Goffin, Institut National de la Santé et de la Recherche Médicale Unit 584, Faculté de Médecine Necker, 156, rue de Vaugirard, 75730 Paris Cedex 15, France. E-mail: goffin{at}necker.fr

	Abstract

Top Abstract I. Introduction II. Pathophysiology of PRL III. Autocrine PRL and... IV. Potential Indications for... V. Development of PRL... VI. Discussion VII. Perspectives References

 
There is a large body of literature showing that prolactin (PRL) exerts growth-promoting activities in breast cancer, and possibly in prostate cancer and prostate hyperplasia. In addition, increasing evidence argues for the involvement of locally produced (autocrine) PRL, perhaps even more than pituitary-secreted (endocrine) PRL, in tumor growth. Because dopamine analogs are unable to inhibit PRL production in extrapituitary sites, alternative strategies need investigation. To that end, several PRL receptor antagonists have been developed by introducing various mutations into its natural ligands. For all but one of these analogs, the mechanism of action involves a competition with endogenous PRL for receptor binding. Such compounds are thus candidates to counteract the undesired actions of PRL, not only in tumors, but also in dopamine-resistant prolactinomas. In this review, we describe the different versions of antagonists that have been developed, with emphasis on the controversies regarding their characterization, and the limits for their potential development as a drug. The most recently developed antagonist, 1–9-G129R-hPRL, is the only one that is totally devoid of residual agonistic activity, meaning it acts as pure antagonist. We discuss to what extent this new molecule could be considered as a lead compound for inhibiting the actions of human PRL in the above-mentioned diseases. We also speculate on the multiple questions that could be addressed with respect to the therapeutic use of PRL receptor antagonists in patients.

I. Introduction

II. Pathophysiology of PRL

A. Pathologies linked to PRL levels

B. PRL, hyperplasia, and cancer

C. Summary

III. Autocrine PRL and Tumors

A. Breast

B. Human prostate

C. Regulation of autocrine PRL expression

D. Mouse models

E. Summary

IV. Potential Indications for PRLR Antagonists

A. Breast and prostate tumors

B. Prolactinomas

V. Development of PRL Receptor Antagonists

A. Mechanism of PRL receptor activation

B. Rational development of human PRLR antagonists

VI. Discussion

A. The best candidate antagonist

B. Which place for PRLR antagonists in human therapy?

C. Limited disadvantages of PRLR antagonists

D. Impact on public health

VII. Perspectives

	I. Introduction

Top Abstract I. Introduction II. Pathophysiology of PRL III. Autocrine PRL and... IV. Potential Indications for... V. Development of PRL... VI. Discussion VII. Perspectives References

 
PROLACTIN (PRL) WAS discovered in 1928 as a pituitary factor able to stimulate mammary gland development and lactation in rabbits, as well as the production of crop milk in pigeons. A few years later, the name prolactin was given, based on its ability to stimulate milk production. The protein structure of ovine PRL was elucidated in the late 1960s, and the cDNA and gene structures were elucidated in the early 1980s. In humans, the PRL gene is located on chromosome 6 and encodes a mature protein of 199 amino acids (23 kDa), including six cysteine residues involved in three intramolecular disulfide bonds. PRL shares high structural and functional similarity with two other polypeptide hormones, GH and placental lactogen (PL). It is thought that the genes encoding these proteins evolved from a common ancestral gene by duplication. More recently, newly identified proteins such as proliferin, proliferin-related-protein, somatolactin, or several PRL-like proteins have been added to the PRL/GH/PL family based on sequence similarities. These proteins are assumed to share a common three-dimensional (3D) structure, named a four-helix bundle, and characterized by four antiparallel -helices (for reviews, see Refs. 1, 2, 3, 4 and original references therein).

More than 300 separate biological activities have been attributed to PRL; these can be subdivided into the following categories: functions linked to reproduction; endocrinology and metabolism; control of water and electrolyte balance; growth and development; brain and behavior; and finally, immunoregulation and protection (5). These biological functions are mediated by a specific membrane receptor that was cloned by our group in 1988 (6); it is referred to as the PRL receptor (PRLR), or the lactogen receptor. As is true for their respective ligands, the receptors of PRL and GH (GHR) are closely related (7) and were two pioneering members of the now well-known cytokine receptor superfamily (8). These receptors are non-tyrosine kinase, single-pass transmembrane chains. One of their major common features is the 3D structure of their extracellular, ligand-binding domain, which folds into two ß-sheets each containing seven ß-strands (9, 10, 11, 12). Although the PRLR gene is unique in each species, various isoforms resulting from alternative splicing have been described, first in rodents, then in ruminants, and more recently in humans (5, 13, 14). With few exceptions, isoforms within a given species exhibit identical extracellular (210 amino acids) and transmembrane (24 amino acids) domains, and only differ by the length and/or the composition of their intracellular domain. In contrast to tyrosine kinase receptors, cytokine receptors transduce the signal via associated kinases, which are recruited by the receptor (when not prebound to the latter) and activated upon ligand binding. Several tyrosine or serine/threonine kinases are involved in PRLR signaling. Although description of these complex signaling pathways falls beyond the scope of this review, we should mention that PRLR signaling mainly (but not exclusively) involves Janus kinase (Jak), MAPK, and Src kinases, which each trigger specific cascades (Fig. 1). In many instances, the interaction of the PRLR with these kinases, or with other proteins involved in positive and negative regulation of signaling (adapters, phosphatases, suppressor of cytokine signaling, etc.), has been mapped to identified regions of the cytoplasmic domain. For example, the major PRLR-associated kinase, Jak2, requires the integrity of the membrane-proximal, proline-rich domain called box-1, whereas signal transducer and activator of transcription (Stat) 5 preferentially binds to the C-terminal phosphotyrosine of the receptor. These two examples illustrate that PRLR isoforms, which differ essentially in their cytoplasmic domains, elicit distinct biological properties depending on their ability to activate some signaling cascades but not others. In general, short receptor isoforms have been attributed dominant-negative activities with respect to the long isoform, because they bind PRL with high affinity but fail to activate some major signaling pathways, such as Jak2/Stat5 (5). On the other hand, certain short isoforms have been shown to rescue phenotypes observed in PRLR heterozygote mice (15), indicating that our understanding of short receptor functions is still incomplete. The reader is referred to previous reviews specifically dedicated to PRLR signaling for additional details (5, 13, 16).

View larger version (44K): [in this window] [in a new window]   FIG. 1. The main signaling cascades triggered by the PRLR in target cells. Ligand-induced activation of the PRLR triggers several signaling cascades. The main pathway involves the tyrosine kinase Jak2, which in turn activates three members of the Stat family, Stat1, Stat3, and mainly Stat5. The MAPK pathway is another important cascade activated by the PRLR. It involves the Shc/Grb2/Sos/Ras/Raf intermediaries upstream to MAPK kinase and Erk-1/-2 kinases. Connections between the Jak-Stat and MAPK pathways have been suggested. Activation of other signaling pathways has been reported; among these, Src kinases are known to play an important role in proliferation, e.g., in breast cancer cells. Clearly, this figure represents a simplified view of PRLR signaling.

	II. Pathophysiology of PRL

Top Abstract I. Introduction II. Pathophysiology of PRL III. Autocrine PRL and... IV. Potential Indications for... V. Development of PRL... VI. Discussion VII. Perspectives References

Hyperprolactinemia is one of the most frequent endocrine diseases, especially in women. Excess PRL secretion leads to galactorrhea (spontaneous lactation), amenorrhea (lack of menstrual cycle), and sterility, which again emphasizes the close relationships between PRL, gonadotropic functions, and mammary gland physiology. There are several causes leading to hyperprolactinemia, such as specific physiological status (e.g., pregnancy, lactation), pharmacological treatment [dopamine antagonist (neuroleptic) drug therapy], various pathologies altering the neuroendocrine control mechanisms regulating PRL secretion, and, of course, pituitary adenomas of lactotrope (PRL-secreting) cells, termed prolactinomas (19). In the majority of the cases, prolactinomas are efficiently treated by synthetic analogs of dopamine, which is the physiological negative regulator of PRL synthesis and secretion in the pituitary (19, 20, 21) (see Section IV.B).

Although hypoprolactinemia is a very rare pathology, such a human model could be helpful for understanding the vast number of biological actions that have been attributed to PRL (5). The pituitary transcription factor 1 (Pit-1) is a major player in the regulation of PRL gene transcription in the pituitary (22). Although mutations of Pit-1 have been reported to lead to hypoprolactinemia in humans as well as in the mouse (23), defects in Pit-1 result in combined hormonal deficiencies, because lactotrope, thyrotrope, and somatotrope axes are all impaired. Therefore, patients harboring Pit-1 mutations are not very informative with respect to PRL failure per se. Isolated hypoprolactinemia was recently described in patients unable to lactate, a phenotype correlated with low, or even undetectable, PRL levels (24). Unfortunately, no information was reported regarding the state of mammary gland development. A genetic origin of hypoprolactinemia was suspected in these patients, because very low PRL levels were also detected in one of the mothers (24); this observation, however, remains to be confirmed. In summary, as indicated above, a definitive human model of PRL deficiency is lacking to more clearly understand the functions of this hormone.

The involvement of PRL in auto-immune diseases, such as systemic lupus erythematosus, has also been suggested, based on the observation that the disease is accentuated postpartum, when PRL levels are elevated to maintain lactation (25). Nevertheless, PRL levels are not necessarily abnormally elevated in patients suffering from lupus, and the possible correlation between circulating PRL levels and the severity of the disease remains controversial (26). Finally, although treatment using bromocriptine, a dopamine agonist, sometimes leads to improvement, additional studies are required to confirm this observation (27).

B. PRL, hyperplasia, and cancer
1. Experimental evidence.
For more than three decades, PRL has been suspected of being involved in tumor proliferation. As the main target tissue of PRL, the breast has been used as the prototype model to investigate the tumor growth-promoting potency of PRL. The proliferative activity of PRL has been clearly demonstrated in vitro, using various mammary tumor epithelial cell lines derived from either mice or humans (reviewed in Refs. 13 and 28, 29, 30, 31). Interestingly, PRL was also reported to affect the proliferation of various human prostate cell lines (32, 33), which emphasizes that PRL may exert similar growth-promoting effects on both organs. Various animal models have further strengthened the protumor potency of PRL in vivo. For example, the growth-promoting action of exogenous PRL was reported in animal models with spontaneous or carcinogen-induced mammary tumors (34, 35), which corroborates the growth-inhibiting action of dopamine agonists in similar models (36). More recently, transgenic mice overexpressing systemic PRL were generated, and their main phenotypes were prostate hyperplasia, which appears very early in life (3–4 months of age), and mammary neoplasia, which appears in older animals (1 yr) (37, 38). In these models, however, whether the effects of PRL are direct or involve alteration of other systemic hormones (e.g., steroids) remains to be demonstrated. In mice deficient for PRL or for its receptor (knockout models), the appearance and/or the development of virus-induced mammary (39) or prostate (40) tumors was delayed, or even inhibited, arguing for the permissive role of PRL in tumor growth. The growth-stimulating action of PRL was also observed when human mammary (41) or prostate (42) tumor cell lines were xenografted into immunodeficient mice. Thus, PRL appears to be a protumor factor in a wide variety of models (13, 43, 44). Multiple mechanisms could be involved, because PRL acts on cell division (30, 33, 45, 46), cell survival (antiapoptotic effect) (43, 47, 48, 49, 50), cell motility (51), and possibly angiogenesis regulation (52) (reviewed in Refs.13 and 53). It is important to note that the protumor actions of PRL do not require steroid hormones, which are major factors in the hormone dependence of mammary (estrogens) and prostrate (androgens) tumor growth (30, 43, 54, 55). This, however, does not preclude these two classes of hormones from acting in concert to promote tumor growth more efficiently. For example, estradiol and testosterone positively regulate the expression of the PRLR in rat prostate (56), whereas receptors for estradiol, progesterone, and PRL are mutually regulated in human mammary tumor cell lines (57).

2. Epidemiological evidence.
Epidemiological studies investigating PRL as a possible risk factor for the development of breast or prostate cancers are somewhat limited, and for most, conclusions are ambiguous. It is important to note that these studies were aimed at evaluating this risk with respect to high or low PRL levels within the normal range, meaning they involved normal subjects with respect to prolactinemia. Unfortunately, we are not aware of any large-scale study investigating whether the risk of developing cancers is increased in hyperprolactinemic patients, or in those patients whose PRL levels could not be normalized after treatment with dopamine agonist.

In postmenopausal women, these studies at best indicated positive but nonsignificant association of PRL levels with breast cancer risk (58, 59, 60, 61), or they failed to detect any association (62, 63). Epidemiological studies involving premenopausal patients were similarly sparse (58, 59, 61, 62, 64, 65). Actually, only two recent studies from Hankinson’s group demonstrated a clear correlation between PRL levels and breast cancer risk in postmenopausal women. The first study correlated PRL levels in the higher quartile of the normal range with an increased risk (by a factor of 2) of developing breast cancer compared with the lower quartile (13, 66). Although it is always delicate to interpret a punctual variation of hormone levels, we should mention that this correlation was independent of estradiol levels. The second study from the same group showed that this risk mainly affected estrogen receptor-positive tumors (relative risk, 1.78; 95% confidence interval, 1.28–2.5), with a risk increased to 1.94 for estrogen receptor-positive/ progesterone receptor-negative tumors, although in this case it was not strictly significant (95% confidence interval, 0.99–3.78) (67). It is presumably because these two studies involved a much larger number of patients (the Nurse Health Study cohort involves more than 30,000 women, with 306 and 851 breast cancer patients in the 1999 and 2004 studies, respectively) that it achieved a significant correlation when previous studies performed on much smaller numbers of patients only achieved positive, but not significant, correlation.

With respect to prostate pathologies, there are even fewer reports than for breast cancer. The recent Northern Sweden Health and Disease Cohort study involving nearly 30,000 men, including 144 subjects diagnosed with prostate cancer (68), concluded that there is no correlation between PRL levels and the risk to develop prostate cancer. To the best of our knowledge, there is no epidemiological study investigating PRL as a possible risk factor for developing prostate hyperplasia. The only available report is a recent prospective, case-control study involving only 20 men with prolactinoma (69). Although experimental models have indicated that hyperprolactinemia correlates with prostate hypertrophy (see Section II.B.1), this report surprisingly suggested the opposite. Although this is presumably due to the fact that decreased levels of male steroids in these patients predominate over the putative growth-promoting effect of PRL, this is puzzling regarding the intrinsic relationship between PRL levels and prostate diseases.

3. Clinical evidence.
At the therapeutic level, only a few trials have been reported, and the results obtained from the treatment of breast or prostate cancer patients with dopamine agonists are disappointing. Although bromocriptine was shown to normalize PRL levels in metastatic breast cancer and prostate carcinoma patients, it was not found to provide significant benefit to breast cancer (70, 71) or prostate cancer (72) patients. One report indicated that lowering PRL levels using bromocriptine improved the antitumor actions of chemotherapeutic agents such as Taxotere (73), which is an interesting issue that requires confirmation.

C. Summary
The implication of PRL in human diseases is mainly based on two criteria: 1) the correlation between PRL levels and the pathology of interest; and 2) the efficiency of dopaminergic analogs to treat the pathology. Based on these criteria, hyperprolactinemia is the only pathology unanimously recognized by clinicians as related to PRL. Despite the numerous experimental data arguing for a tumor growth-promoting activity of PRL on the breast and prostate tissues, the transposition to the human context is highly debated, because the two criteria to assess the PRL dependence of a given disease are at best only partly met (breast cancer), or clearly not fulfilled (prostate diseases). As discussed in Section III, these criteria should probably be revisited in view of data accumulated within the last decade.

	III. Autocrine PRL and Tumors

Top Abstract I. Introduction II. Pathophysiology of PRL III. Autocrine PRL and... IV. Potential Indications for... V. Development of PRL... VI. Discussion VII. Perspectives References

 
Several laboratories involved in PRL research, including ours, have tried to reconcile the opposing conclusions drawn from experimental and epidemiological/therapeutic studies with respect to the protumor potency of PRL (31, 74, 75). One tempting hypothesis came from evidence that PRL acts not only as an endocrine hormone, but also as an autocrine-paracrine growth factor (Fig. 2). The involvement of autocrine PRL in breast cancer has been recently discussed by Clevenger et al. in their outstanding review (13); therefore, only key reports, including those on prostate cancer, and articles published since the Clevenger review article, are mentioned below.

View larger version (37K): [in this window] [in a new window]   FIG. 2. Endocrine and autocrine/paracrine ligands of the PRLR. In humans, there are three different ligands of the PRLR, i.e., PRL, GH, and during gestation, PL. Secretion of PRL and GH by the pituitary can be inhibited using inhibitors of synthesis (dopamine and somatostatin analogs). In contrast, there is currently no known negative regulator of extrapituitary production of these hormones, acting locally via an autocrine/paracrine mechanism. We propose that PRLR antagonists could be used to counteract the protumor action of PRLR ligands by preventing receptor activation.

B. Human prostate
As observed in the breast, expression of PRL and of its receptor is detected in both normal tissue and prostate tumors in humans (86), thereby arguing for the physiological relevance of the autocrine-paracrine loop. The level of PRLR expression is elevated in dysplasia, suggesting that PRL might participate to the preneoplastic stages of tumor development (87). Although no report indicates that the PRLR is overexpressed in human prostate cancer, which contrasts with what is observed in breast cancer, a recent study from Nevalainen’s group (88) showed that the expression of PRL was correlated to the grade of the tumor (Gleason score), meaning the higher the grade, the higher the amount of PRL detected by immunostaining. One of the major signaling molecules activated by the PRLR is the transcription factor Stat5 (5); interestingly, Nevalainen and colleagues also reported that activated Stat5: 1) acts as a survival factor in prostate cancers (89); and 2) is also more abundantly detected in high Gleason score cancers (88). Although it is premature to suggest that autocrine PRL acts as a survival factor in human prostate cancer by maintaining high levels of activated Stat5, this hypothesis seems reasonable.

C. Regulation of autocrine PRL expression
One major difference between mouse models and human tissues is that PRL is permanently expressed in human breast and prostate (i.e., independently of the physiological status), whereas it is either restricted to gestation/lactation (mammary tissue), or even undetectable (prostate) in mouse tissues (13). In humans, but not in animals, the PRL gene is regulated at the transcriptional level by two distinct promoters (for reviews, see Refs. 20 and 90). The proximal promoter, also referred to as "pituitary" promoter, covers approximately 5 kb upstream of the transcription site, in which the 250 bp just before the transcription initiation site (in exon 1b) are necessary and sufficient for transcription. The second promoter, referred to as "extrapituitary" or "lymphoid" promoter, includes approximately 3 kb upstream of exon 1a (itself located 5.8 kb upstream of the initiation site) and was initially described as directing PRL expression in lymphoid and decidual cells (90). Depending on promoter usage, the PRL mRNAs differ in length by 134 bp but encode a strictly identical mature protein. Conventional wisdom has linked the proximal PRL promoter to pituitary gene expression (involving Pit-1 as major activating transcription factor and dopamine as major negative regulator), and the decidual/lymphocyte promoter to extrapituitary gene expression (independent of Pit-1 and dopamine). This is presumably a simplified view of the reality, because analysis of PRL cDNA obtained from various human breast cancer cell lines or biopsies showed that both types of mRNA (differing in their 5' untranslated region) are present, reflecting a duality of promoter usage (91). Still more surprising, a recent study has shown that in the SK-BR-3 human mammary tumor cell line, the pituitary promoter, and not the decidual/lymphocyte promoter, is active, despite the absence of Pit-1 in mammary cells (92). In view of these two studies, the alternative promoter usage in pituitary vs. nonpituitary tissues is a concept that obviously needs to be revisited.

D. Mouse models
A recent study involving transplantation of wild-type (WT) or PRL-deficient mouse mammary epithelium into immunodeficient mice showed reduced proliferation at the end of pregnancy in the latter, in agreement with the fact that mammary expression of PRL is restricted to gestation/lactation (93). To mimic the permanent expression of local PRL described in human tissues, tissue-specific transgenic mice were recently generated. The first two publications in which these models were investigated indicate that local expression of PRL in prostate or mammary tissues leads to benign prostate hyperplasia (94) and mammary neoplasia (95), respectively. Although these new animal models need to be further analyzed, it is clear that their major phenotypes are very similar to those reported earlier in transgenic mice expressing PRL systemically (37, 38). Not only do these observations argue for the intrinsic protumor potency of autocrine PRL in vivo, but they also suggest that this activity is really PRL-specific, because circulating levels of androgen and PRL were unchanged or minimally elevated, respectively, in the prostate-specific PRL transgenic model. The impact of autocrine PRL on mammary tumorigenesis may even be predominant over that of circulating PRL, because the various strains of transgenic mice overexpressing systemic PRL develop mammary tumors with similar characteristics, irrespective of their circulating levels of PRL (38). We have recently developed a transgenic model in which the hPRL transgene is controlled by the promoter of milk protein WAP (whey acidic protein). This recombinant system provides not only a mammary-specific, but also a time-specific overexpression of the transgene, because this promoter is active from around midgestation until the end of lactation. Interestingly, sporadic PRL overexpression leads to various functional and histological abnormalities. Histological analysis of young and midage animals (<1 yr) identified the appearance of benign tumors, with dystrophies aggravating with successive pregnancies. We are awaiting the analyses of older animals to ascertain whether they develop malignant tumors, as reported for transgenic animals in which expression of the PRL transgene is permanent (38, 95), or whether this difference relates to the intrinsic characteristics of the experimental models used by the investigators, e.g., the genetic background. Obviously, detailed analysis of this animal model will help elucidate the actions of autocrine PRL in the mammary gland.

E. Summary
The above-mentioned studies demonstrate the tumor growth-promoting potency of autocrine PRL in various models and suggest that in human breast and prostate cancers, this action might, at least in part, involve overexpression of PRL itself and/or of its receptor. Although the relative contribution of local vs. systemic PRL to the growth of tumors cannot be accurately evaluated yet, the autocrine/paracrine loop can be proposed as a possible explanation regarding the paradox that PRL is a tumor growth promoter in many experimental models, whereas clinical/epidemiological evidence is still lacking. First, autocrine PRL is secreted into its local environment and is thus assumed not to contribute (or only very modestly) to circulating PRL levels. This parameter has been totally ignored in epidemiological studies that only take into account serum PRL levels. Second, because dopamine does not regulate PRL synthesis in extrapituitary tissues (20), dopamine analogs used in clinical trials involving breast cancer patients presumably failed to affect local PRL expression.

The hypothesis that local PRL participates in the proliferation of breast and prostate tumors has become a major axis of research in many well-established laboratories in the field (13, 35, 95, 96). Although a number of highly respected articles have been published in the last decade, this autocrine mechanism of action still needs to be further studied, to be better understood, and most importantly, to be definitely accepted as a relevant factor involved in the promotion of tumor growth in humans. To that end, the positive role of autocrine PRL in human tumor growth could be demonstrated by treating patients with a compound capable of blocking its effects.

	IV. Potential Indications for PRLR Antagonists

Top Abstract I. Introduction II. Pathophysiology of PRL III. Autocrine PRL and... IV. Potential Indications for... V. Development of PRL... VI. Discussion VII. Perspectives References

 
A. Breast and prostate tumors
The first strategy to block the effects of autocrine PRL in tumors should be to inhibit its synthesis. In addition to our limited understanding of the promoters and transcription factors involved in PRL transcription in extrapituitary sites, extracellular factors regulating extrapituitary PRL expression also need to be identified. Various hormones, growth factors, peptides, or neurotransmitters were evaluated for their ability to regulate PRL synthesis, especially in the uterus and in mammary tissue (92, 97) (for a review, see Ref. 35). However, none of them seems to exert per se a major role on extrapituitary PRL production similar to that of dopamine in the pituitary.

An alternative approach would be to develop compounds able to inhibit the actions mediated by autocrine PRL, rather than its production. PRLR antagonists can be defined as hPRL analogs that bind but do not activate the PRLR; hence, they prevent endogenous PRL from exerting its effects by a competitive mechanism. These compounds appear to be an attractive alternative strategy to circumvent the absence of any known negative regulator of PRL expression in extrapituitary sites (Fig. 2). This would permit attaining a dual goal: 1) to demonstrate the involvement of local PRL in the proliferation in human breast or prostate tumors; and, if conclusive, 2) to constitute a new and unique class of molecules, acting at the level of receptor activation. The development of such compounds is detailed in Section V.

B. Prolactinomas
The treatment of prolactinomas involves medical treatment, surgery, or radiotherapy (19, 21). Irradiation tends to be less frequently used because it affords the lowest benefit/risk ratio. For microadenomas (<10 mm), there is no established clinical consensus whether medical treatment or surgery should be the first-line therapy. The success rates of transphenoidal surgery are highly dependent on the experience and skill of the surgeon; this remains the favored option in many centers, especially those with established success rates. For macroadenomas (>10 mm), medical treatment is the first-line therapy because the efficiency of surgery is much more limited. Bromocriptine was the pioneering dopamine analog, after which more potent drugs with improved activity (increased half-life, fewer side effects) were subsequently developed (pergolide, quinagolide, cabergoline) (19). Dopamine agonists normalize circulating PRL levels by decreasing its production, which among other effects, restores fertility in many women presenting with hyperprolactinemia. These drugs also lead to rapid tumor shrinkage, which is an important parameter for the well-being of patients, independent of fertility, because various symptoms associated with pituitary adenomas, such as visual field disturbance or headache, are improved (19).

Ten to 20% of patients presenting with prolactinomas are dopamine-resistant. Although the definition of dopamine resistance can vary among studies, it involves a mix of patients who, at best, partially respond to dopamine agonists without complete PRL normalization, and at worst, do not respond at all (98). Although there are probably several causes for the lack of dopamine agonist responsiveness, it most likely involves a decrease of dopamine D2 receptor expression at the membrane of lactotrope cells, which is amplified by a decrease of the G protein that couples this receptor to downstream adenyl cyclase (98, 99). To date, the strategies to circumvent dopamine agonist resistance include switching to another dopamine agonist, increasing the dose beyond conventional doses to see whether some response is observed, surgery, or radiotherapy. PRLR antagonists could also represent an alternative medical treatment, because they should be able to block the actions of PRL, if not its production.

	V. Development of PRL Receptor Antagonists

Top Abstract I. Introduction II. Pathophysiology of PRL III. Autocrine PRL and... IV. Potential Indications for... V. Development of PRL... VI. Discussion VII. Perspectives References

 
A. Mechanism of PRL receptor activation
Receptor activation through homo-, hetero-, or oligomerization is a common rule within the hematopoietic cytokine receptor superfamily (8). As one of the pioneering members of this receptor superfamily, the GHR constituted a paradigm for elucidating the molecular rules of receptor activation during the late 1980s. Mutational, biochemical, and structural investigations of hGH have clearly demonstrated that this hormone possesses two functionally important regions, each of which interacts with one receptor chain (9, 100, 101). The initial model of GHR activation suggested that receptor homodimerization occurs in a two-step process (100). First, one molecule of GH binds to one molecule of GHR through its binding site 1, of high surface and affinity, leading to the formation of an intermediate and inactive complex of 1:1 stoichiometry (one ligand bound to one receptor). Second, the hormone involved in this intermediary complex recruits a second (identical) GHR molecule through its binding site 2, of lower surface and affinity, leading to the formation of an active trimeric complex in which the receptor is homodimerized (hGH-hGHR₂). Similar activation mechanisms were next reported for erythropoeitin, leptin, and thrombopoietin receptors (8).

The PRLR was rapidly assumed to be activated through a similar mechanism. Indeed, most of the observations reported for the formation of the hGH-hGHR₂ complex were also made for the interaction of hGH with the lactogen receptor, including the existence of two binding sites of hGH overlapping those involved in the interaction with the hGHR (10, 102). With respect to PRL, the specific ligand of the PRLR, we performed extensive mutational studies of hPRL aimed first at determining the existence and second at characterizing two binding sites homologous to those described in hGH. These studies confirmed that hPRL possesses two regions that are functionally involved in the activation of the receptor, strongly suggesting that the PRLR is also activated by PRL through homodimerization (Fig. 3A). These studies were reviewed in a previous issue of Endocrine Reviews (1) and are therefore not detailed in this article.

View larger version (42K): [in this window] [in a new window]   FIG. 3. Assay sensitivity determines the apparent biological properties of hPRL analogs. A, Receptor activation. This panel is a theoretical representation of the formation of functional receptor dimers with respect to the concentration (x-axis) and the nature (hPRL, blue line; G129R-hPRL, red line) of the ligand. Similarly to the model initially proposed for the GHR (100 ), the PRLR is assumed to be activated in two steps. First, one molecule of PRL binds to one molecule of receptor through its binding site 1, leading to an inactive 1:1 complex. Then, this intermediary complex binds to a second molecule of receptor through the hormone binding site 2, leading to the formation of an active trimeric 1:2 complex. At high concentrations, these hormones are believed to saturate receptors through their binding site 1, leading to preferential formation of 1:1 inactive complexes. This model of sequential receptor dimerization explains why bell-shaped curves are frequently observed in PRL dose-response bioassays (self-antagonism follows agonism). Due to alteration of binding site 2 in G129R-hPRL, the bell-shaped curve of receptor dimer formation by this analog has a much lower amplitude. B, Bioactivity. This panel schematically represents the dose-response curves obtained for hPRL (solid blue line) and G129R-hPRL (solid red line) in four bioassays routinely used in our laboratory [proliferation of Nb2, Ba/F-hPRLR, or T-47D cells, and LHRE-luciferase assay in human embryonic kidney (293-HEK) fibroblasts]. Dotted lines schematically represent the dose-dependent formation of functional receptor dimers by hPRL (blue) and G129R analog (red), as represented on panel A. The threshold of receptor dimers sufficient to produce maximal response in a given assay is represented by the horizontal black hatched line. For both ligands, bioactivity curves (solid) superimpose those of dimer formation (dotted) until this threshold is achieved. Assay sensitivity is defined with respect to the concentration of hPRL required to produce maximal biological response. In the most sensitive assay (Nb2 cell proliferation, left), very low levels of stimulation are needed to achieve maximal proliferation (reflected by the low threshold of receptor dimers required for maximal response), possibly because the division process faces intracellular limiting factors. Therefore, although G129R-hPRL is able to induce only low amounts of 1:2 complexes (see A), the level of functional receptor dimers is sufficient to promote (sub-)maximal cell proliferation. Note, however, that the agonistic part of the curve is shifted toward high concentrations for all site 2 analogs compared with WT hormone. In contrast, in the less sensitive assay (transcriptional LHRE-luciferase, right), the biological response seems directly proportional to the level of receptor activation (reflected by the high threshold of receptor dimers required for maximal response), presumably because it is not influenced by limiting factors. Therefore, the low level of 1:2 complexes formed by G129R-hPRL is by far insufficient to promote any significant biological response. Bottom panels represent competitive antagonistic biossays, in which a fixed concentration of hPRL (producing a given biological response represented by the horizontal blue line) is competed for by increasing concentrations of G129R-hPRL, leading (or not) to a decrease of bioactivity (antagonism). Depending on the assay used, the interpretation of G129R-hPRL properties can thus vary from weak agonist, partial agonist, to full antagonist. What happens in vivo is believed to resemble "sensitive assays" more than "low sensitive assays", which is supported by the agonistic activities of G129R-hPRL in transgenic mice.

B. Rational development of human PRLR antagonists
1. Mutation of the helix 3 glycine as a paradigm to generate antagonists.
The paradigm mutation for shifting PRL/GH hormones from agonists to antagonists was discovered by Kopchick et al. (107) in the late 1980s. With the aim of developing a GH molecule containing perfectly amphiphilic -helices, these authors substituted an arginine for the natural glycine in the third helix of bovine GH, generating so-called G119R-bGH. Although potentialization of GH properties was expected, this single mutation was shown to inhibit growth when G119R-bGH was expressed in transgenic mice. At that time, the molecular mechanism underlying this functional antagonism was unknown, and it is only when the 3D structure of the hGH-hGHbp₂ complex was solved that the effect of the Gly to Arg substitution could be understood (9). The helix 3 of GH is involved in the second binding site of the hormone to its receptor. The small side chain of the glycine residue (numbered 120 in human and 119 in bovine GH) maintains a cleft within the helix, into which one tryptophane residue (Trp 104) of the GHR can dock upon binding; when a residue larger than an alanine is substituted for the glycine, docking of the tryptophane is impaired, and consequently, the interaction of such modified GH with the second receptor chain is prevented. However, because the first binding site of the analog is not altered, the hormone-receptor interaction can still occur normally via this site. The mechanism of antagonism of glycine mutants is thus based on a competition with WT GH for binding to the receptor (Ref. 107 and references therein). Based on the conventional model of sequential GHR dimerization, it was initially believed that hGH-G120R is able to form only an inactive 1:1 complex with the receptor, in agreement with the stoichiometry of the hGH-hGHbp complex that could be crystallized (108). However, the recent evidence that the membrane receptor is predimerized rather suggests that the glycine mutants interact with a receptor dimer but are unable to induce the conformational change required to achieve an active receptor. Whatever occurs at the molecular level, functional receptor dimerization is impaired, whereas binding persists.

2. First-generation PRLR antagonist, G129R-hPRL.
In the early 1990s, we developed the first-generation PRLR antagonist based on the assumption that GH and PRL shared common mechanisms of receptor activation. The identification of two binding sites in hPRL, which were assumed to induce PRLR homodimerization, led us postulate that the prototype mutation performed in hGH to generate a GHR antagonist—replacement of the helix 3 glycine—would have the same effect in PRL. The so-called G129R-hPRL analog was generated (Fig. 4) and characterized using the classical Nb2 cell proliferation assay, which was at that time the unique cell-based bioassay available. Unexpectedly, in this assay not only did G129R-hPRL fail to antagonize PRL-activated cell proliferation, but it was also shown to exert intrinsic mitogenic activity, although the dose-response curve was displaced toward the right, i.e., higher concentrations (109) (Fig. 3B). In other words, G129R-hPRL appeared to be a weak agonist and not an antagonist. At first sight, these results suggested that our initial hypothesis about the consequences of the mutation was incorrect. This was unlikely, however, because the glycine substitution rapidly appeared to be a general concept to generate competitive antagonists not only in GHs, but also in PLs, irrespective of their origin (55, 110). Although G129R-hPRL displayed no antagonism, the glycine substitution obviously disturbed the binding process, because dose-response curves obtained with hPRL and G129R-hPRL were not superimposable. Interestingly, whereas we failed to detect self-antagonism of WT hPRL at high concentration in the Nb2 assay (reflected by a bell-shaped curve; Fig. 3B), this phenomenon was clearly marked for G129R-hPRL analog (109). Self-antagonism is a phenomenon that was initially described by Wells and colleagues (111, 112) as a logical consequence of the model of GHR activation by sequential dimerization (Fig. 3A): when the ligand is present at high concentration, the formation of 1:1 complexes (involving ligand site 1) is favored and the bioactivity decreases. The higher the difference of affinity between sites 1 and 2 in favor of site 1 (which is the case in hGH), the lower the concentration at which self-antagonism occurs. Hence, in the absence of a detectable bell-shaped curve for hPRL in the Nb2 assay, even when very high concentrations (>100 µg/ml) were tested, we hypothesized that the affinity of both hPRL binding sites should be roughly identical, thereby maintaining the preferential formation of 1:2 complexes up to very high concentrations (no detectable self-antagonism), whereas the affinity of site 2 should be altered in G129R-hPRL analog, thereby rendering self-antagonism detectable at workable concentrations. A recent report confirmed these assumptions, because the affinities of both hPRL binding sites for hPRLBP were found to be identical when measured by surface plasmon resonance (106). With respect to G129R-hPRL, the affinity of site 2 was shown to be decreased by one log unit (106), which correlates with our findings that the global affinity of this analog is 10-fold lower compared with that of WT hPRL (113). In summary, based on these initial studies, G129R-hPRL was suspected to be a weak agonist (and not an antagonist) because the affinity of site 2 was not sufficiently altered to completely prevent the interaction of the ligand with the second PRLR chain and thereby, to generate an antagonist. The ability of the glycine mutation to generate an antagonist in GH but not in PRL was therefore interpreted as somehow linked to the distinct features of these hormones.

View larger version (43K): [in this window] [in a new window]   FIG. 4. Structural representation of hPRL analogs referred to as PRLR antagonists. One of the 20 sets of NMR coordinates obtained by Keeler et al. (138 ) was used to illustrate the 3D structure of hPRL (top left), as well as to locate the mutations performed in the three analogs of interest: S179D-hPRL (top right), G129R-hPRL (bottom left), and 1–9-G129R-hPRL (bottom right). Human PRL folds in an antiparallel four -helix bundle. N-to-C orientations of the four helices (h1 to h4, color-coded) and of the three connecting loops (L1 to L3) are indicated by thin arrows. Global locations of binding sites 1 and 2 are deduced from mutagenesis studies: site 1 mainly involves helix 4, loop 1, and, to a lesser extent, helix 1; whereas site 2 is formed by surfaces of helices 1 and 3, and possibly involves the N terminus. Glycine 129 is anticipated to create a cleft in site 2, which permits the docking of large residues of the PRLR; therefore, its replacement by an Arginine in G129R-hPRL (surrounded by dotted blue circle and indicated by a large blue arrow) presumably disturbs the interaction of hPRL with the second PRLR molecule. Removal of the nine first N-terminal residues in 1–9-G129R-hPRL (surrounded by dotted gray circle and indicated by a large gray arrow) likely enhances this effect, possibly by preventing conformational changes of the second receptor molecule within the hormone-receptor complex, or even by abolishing any interaction of hPRL with this second receptor. In contrast, S179D substitution (surrounded by dotted red circle and indicated by a large red arrow) is not anticipated to interfere with receptor docking, because this serine faces the hydrophobic core of the protein and should thus not be involved in any binding site. The mechanism of action of S179D-hPRL remains obscure.

The apparently conflicting data obtained for G129R-hPRL in the bioassays described above led us to the conclusion that parameters other than species specificity direct the agonistic vs. antagonistic properties of this analog. Extensive comparison of bioassay features [species origin of cells and receptors, number of receptor expressed, biological response measured (proliferation/transcriptional activity/signaling), etc.] provided evidence that assay sensitivity is the key parameter responsible for the readout of biological properties. Sensitive bioassays are defined as those requiring a very low level of stimulation (low PRL concentration) to exhibit a biological response (e.g., Nb2 cell proliferation), whereas low sensitive bioassays are those requiring a high level of stimulation (high PRL concentration) to exhibit a biological response (luciferase reporter assay). As indicated above, G129R-hPRL maintains the ability to induce low levels of functional receptor dimers (Fig. 3A), because site 2 exhibits a significant and measurable affinity for the hPRLBP (106). In sensitive bioassays, the level of activated receptors is sufficient to elicit a detectable response; therefore G129R-hPRL acts as a weak agonist (dose-response curve displaced to the high concentrations) and/or a partial agonist (never reaches the maximal activity displayed by WT hPRL), but not as an antagonist. In contrast, in low sensitive assays, the level of activated receptors is not significant to elicit a biological response; therefore the analog acts as an antagonist and not an agonist (119) (Fig. 3B). It should be noted that assay sensitivity is determined not only by the intrinsic responsiveness of the cell or animal models used but also by the type of experiment that is performed in a given system. For example, whereas maximal proliferation of Nb2 cells is achieved at 1 ng/ml of PRL, maximal phosphorylation of signaling proteins in these cells requires up to 50–100 ng/ml PRL, and sometimes more. Hence, G129R-hPRL acts as an agonist in proliferation assay (109), whereas it will be considered as an antagonist in signaling studies (120), despite the fact that the same cells are used in both types of experiments. Determining whether an analog is agonist or antagonist is thus sometimes very difficult.

In summary, our findings strongly suggest that G129R-hPRL is intrinsically a weak, partial agonist, because it is still able to achieve some level of PRLR activation when added in sufficient amount. In low sensitive bioassays, the agonistic activity is not displayed because the level of receptor activation is below the threshold leading to detectable responses; therefore the antagonistic effect predominates (119). This interpretation is not unanimously accepted, because some controversy remains about the intrinsic actions of G129R-hPRL. Chen et al. (47) claimed that this antagonist per se promotes apoptosis of breast cancer cells. Indeed, the antiapoptotic effect of hPRL involves up-regulation of the survival factors bcl-2 and TGF and down-regulation of the apoptotic factor TGFß1, whereas G129R-hPRL was reported to have opposing effects on these targets, to activate capsase-3, and to up-regulate bax expression (48, 117, 121). Because these changes were observed without concomitant addition of hPRL, it suggests that G129R-hPRL could activate specific signaling cascades resulting in biological effects opposed to those of hPRL. This remains to be demonstrated, although it is unlikely because our data clearly show that when G129R-hPRL activates the PRLR, it exerts PRL-like rather than anti-PRL actions (109, 119). Alternatively, the proapoptotic effect of G129R-hPRL could reflect a competitive antagonism toward autocrine hPRL, produced by breast cancer cells (76). Although the reliability of this hypothesis could not be assessed in studies from Chen’s group, it is again unlikely, because autocrine hormones are particularly difficult to antagonize (122, 123), especially with an analog otherwise demonstrated to be a partial agonist. The controversy also applies to in vivo models. Transgenic mice expressing G129R-hPRL or hPRL under the control of the ubiquitous metallothionein promoter were recently generated by Chen’s group (121). The action of G129R-hPRL was particularly unclear because the expression of bax and cytochrome c proteins in the mammary gland appeared to be regulated by both ligands in an opposing manner in animals 6 months of age (low bax/cytochrome c in hPRL transgenics, high in G129R-hPRL transgenics), but in a similar manner at 9 months of age (high bax and cytochrome c in both transgenics). These varying phenotypes might reflect that effects are borderline, maybe due to the low level of expression of the transgene (5–10 ng/ml, i.e., even lower than endogenous PRL levels), which certainly fails to compensate for the lower affinity of G129R-hPRL. In contrast, transgenic mice expressing higher levels of G129R-hPRL (100–1000 ng/ml) (J. J. Kopchick, B. Kelder, P. A. Kelly, and V. Goffin, unpublished results) clearly highlight the agonistic activity of this analog, because these animals display many symptoms of hyperprolactinemia, such as prostate hypertrophy, abnormal morphology of the mammary gland, or reproductive disorders. This is in direct agreement with our in vitro observations. In summary, even if these conflicting data cast a shadow on our understanding of G129R-hPRL, it is clear that this first-generation PRLR antagonist is able to exhibit PRL-like actions in some experimental situations, including in vivo, which obviously precludes this PRL analog from being considered as a good candidate for drug development.

The goal of our group was thus to develop pure PRLR antagonists, i.e., PRL analogs devoid of the partial agonistic activity detected in G129R-hPRL. We explored various strategies, all of which involved the introduction of additional mutations into this initial antagonist. Obviously, one strategy was to completely abolish the interaction involving binding site 2, to prevent the formation of functional receptor dimers. Attempts to block site 2 more efficiently than in G129R-hPRL were unsuccessful. For example, combination of two substitutions individually hindering site 2 (G129R and A22W) (109, 113) yielded a misfolded protein that could never be reliably characterized (V. Goffin, unpublished observation). In one of our former structure-function studies of hPRL, we suggested that the greater the difference in the affinity of each binding site (site 1 higher), the lower the agonistic activity of the hormone (124). Therefore, we investigated various strategies aimed at increasing site 1 affinity to favor the formation of inactive 1:1 complexes. First, we inserted into G129R-hPRL sequence other amino acid substitutions previously shown to increase site 1 affinity (e.g., Q71A and Q74A mutations) (125). The corresponding double mutants failed to exhibit significantly increased antagonistic properties (S. Kinet, S. Pastoret, V. Goffin, and J. A. Martial, unpublished data), which probably correlates with the relatively low advantage provided by the glutamine replacements alone (2- to 3-fold increase of affinity) (125). Second, based on the observation that the affinity of hGH site 1 for the hPRLRbp is dramatically increased by the coordination of one zinc ion (126), we substituted a glutamate for the natural aspartate 183 in hPRL, to reconstitute in the latter a zinc binding site identical to that found in the hGH-hPRLR complex (124). Although hPRL intrinsically binds zinc, which does not affect biological properties (127), the D183E substitution conferred zinc sensitivity to the hPRL-hPRLR interaction, which now interfered with biological properties. However, the effect was opposite to that expected, because the mutation per se was detrimental to site 1 affinity (124). Hence, the double mutant D183E/G129R-hPRL was shown to exhibit increased agonistic activity compared with G129R-hPRL, because the reduction of site 1 affinity due to the D183E substitution also decreased the difference of affinity with respect to site 2. Again, these results were opposite to our initial expectation. The last strategy that we investigated involved the combination of the G129R mutation with another substitution that had just been reported to generate a strong antagonist, namely the S179D substitution, described in Section V.B.3 (128).

3. Molecular mimic of phosphorylated PRL, S179D-hPRL.
PRL exists in several molecular isoforms resulting from various posttranslational modifications, including proteolytic cleavage, glycosylation, and phosphorylation (129). In rat PRL, serine 177 was identified as the major phosphorylation site (130). Based on the observation that phosphorylated rat PRL antagonizes PRL actions in proliferation assays, Walker’s group (128) generated a molecular mimic of phosphorylated PRL by substituting an aspartate for the topologically equivalent residue in the human sequence, serine 179 (Fig. 4). Aspartate is indeed accepted as a satisfactory molecular mimic of phosphorylated serine. In their first report, these authors showed that S179D-hPRL strongly antagonized the mitogenic activity of hPRL in Nb2 cells, thereby suggesting that the observations made for phosphorylated rat PRL could be extrapolated to the human hormone (128). However, in contrast to classical PRL/GH antagonists harboring a mutation of the helix 3 glycine, S179D-hPRL appeared to antagonize hPRL effects through a noncompetitive inhibition of receptor activation. This encouraged Walker and colleagues to investigate the molecular bases of this unusual, although apparently highly potent mechanism of antagonism. These authors showed that S179D-hPRL strongly activated Stat5, while minimally activating Jak2 (131). This led the authors to hypothesize that the antagonistic activity of S179D-hPRL could result from the activation of signaling molecules/pathways other than those known to be involved in hPRL signaling.

The results reported in these two initial reports were controversial, because, in our hands, S179D-hPRL failed to antagonize hPRL and instead, it was clearly shown to act as an agonist with respect to the proliferation of Nb2 cells and of human breast cancer cell lines, as well as to the activation of Jak2/Stat5 and MAPK signaling pathways in these cells (120). S179D-hPRL was even shown to be a superagonist in some situations, such as on the activation of a Stat5 reporter gene (LHRE-luciferase). The possible reasons for such conflicting results, despite using similar experimental models, were extensively debated in our corresponding publication (120), but none of them could really account for the discrepancies. More recent reports from Walker’s group (132, 133, 134, 135, 136, 137) partly reconciled these opposing findings. Indeed, in normal and tumor mammary cells, S179D-hPRL was shown to activate common pathways with WT hPRL, but with some qualitative, quantitative, or kinetic differences. For example, although both ligands activate Jak2/Stat5, the ratio of tyrosine vs. serine phosphorylation of Stat5 was lower in cells stimulated with S179D-hPRL than with hPRL (132). Some observations remain confusing, however, because WT hPRL was more effective than S179D-hPRL to activate the MAPK pathway in normal mouse mammary cells HC11 (132), but less effective in human breast cancer cells MCF-7 (133). In terms of biological activities, both PRL-like and anti-PRL actions were attributed to S179D-hPRL. For example, this analog has been shown to reduce tumor incidence of prostate cancer cells injected into nude mice (42), to affect maternal behavior in nulliparous female rats (134) and alter the development of pup tissues (135), and to inhibit growth of rat mammary gland (136) and of human breast cancer cells (133), indicating biological effects opposed to those normally exerted by WT PRL. In contrast, S179D-hPRL was reported to promote lobuloalveolar differentiation and casein expression in rat mammary gland (136) and to be even more potent than WT hPRL on bone tissue (137), which indicates that this analog also displays PRL-like properties in some circumstances.

It is clear that the mechanism of action of S179D-hPRL is nonconventional, sometimes varying, and presumably very complex. Because S179D-hPRL was reported to regulate PRLR expression, part of its specific properties may result from changes in the ratio of the various isoforms, which are known to differ in their ability to activate signaling pathways (5). However, a clear picture is still lacking, because S179D-hPRL was shown to induce expression of long PRLR isoforms in human breast cancer cells MCF-7 (133) and otherwise to up-regulate expression of the short receptor in normal mouse mammary cells HC11 (132). Such regulation is anticipated to result in opposing actions of S179D-hPRL in target cells. With respect to the interaction with the PRLR, S179D-hPRL clearly works differently than helix 3 glycine mutants. When the latter are assumed not to activate (or almost not to activate) the receptor and to act through a competitive mechanism, which requires that they are used in molar excess (see Section V.B.2), S179D-hPRL obviously activates the receptor, which results in biological responses that could be opposed to those triggered by WT hPRL. One intriguing observation concerns the concentrations at which S179D-hPRL was reported to exert anti-PRL actions. Although we showed that its affinity for the hPRLR is 20-fold lower compared with hPRL (123), which correlates with dose-response curves displaced to the right in several bioassays, it is surprising that in vitro, concentrations as low as 0.1 nM (2 ng/ml) significantly inhibited the growth of human prostate tumor cell lines potentially induced by autocrine PRL (42). Similarly, animal treatment involving implants of osmotic minipumps (42, 136) achieved circulating levels of S179D-hPRL around 2–3 nM (50 ng/ml), which is not that different from endogenous PRL levels (95) and certainly insufficient to compensate for its lower affinity (given the difference of binding affinity between hPRL and S179D-hPRL, a 1:1 molar ratio should correspond to a 20:1 activity ratio). If these observations could be confirmed, they would suggest that the hormone-receptor interaction should transmit the "S179D-hPRL signal" very efficiently to achieve anti-PRL actions, despite its low concentration in these experiments. Understanding the specific features of the molecular interaction between S179D-hPRL and the hPRLR might solve part of the mystery. Serine 179 is in helix 4 (138), which contains several binding site 1 determinants (1), and points toward the inside of the four helix bundle. Accordingly, this residue is not anticipated to be involved in a direct interaction with the receptor. However, its mutation into Asp is expected to result in local disturbance of binding site 1 (as is putative phosphorylation of hPRL in the physiological context), which possibly correlates with the difficulty to correctly refold S179D-hPRL in vitro (120). How such potential structural changes lead to a ligand exhibiting properties so different from the natural ligand remains to be elucidated. Whether the biological properties of S179D-hPRL are mediated by the only known PRLR, or whether S179D-hPRL is also able to bind and activate another (class of) receptor(s) remains unknown. To the best of our knowledge, however, there are no experimental data to support such a hypothesis.

In summary, the current hypothesis regarding this particular analog is that it would stimulate various signaling pathways triggered by the PRLR, but for some unknown reason, this molecular interaction results in differential activation of downstream targets, modulating the resulting biological effects. The putative antitumor potency of S179D-hPRL relies on its proposed ability to favor (or maintain) tissue differentiation at the expense of proliferation. Obviously, additional studies are required to understand this mutant, which will surely necessitate careful interpretation.

4. Second-generation PRLR antagonist, 1–9-G129R-hPRL.
In view of the intrinsic agonistic activity of S179D-hPRL, our initial attempt to improve G129R-hPRL properties by combining G129R and S179D substitutions was rapidly abandoned. During the course of routine structure-function studies of hPRL, we generated a series of hPRL analogs mutated at the N terminus, because this region is the most divergent within the PRL/GH/PL family (2). With respect to the first helix, the N terminus contains five amino acids in primate GHs and PLs (which evolved from the same ancestral gene) (3), 14 residues in hPRL (including a highly conserved disulfide bond between cysteines 4 and 11) (129), and 17 residues in ruminant PLs (which evolved from the PRL lineage). Based on the fact that all of these hormones are lactogenic, the N terminus was conventionally considered to be functionally not important for binding to the PRLR. However, the crystal structure obtained for the trimeric complex between ovine PL (oPL) and the rat PRLBP demonstrated that the N terminus of oPL plays a critical role in the second binding site (12). Obviously, such interaction does not occur in the hGH-hPRLR complex, because hGH lacks homologous N-terminal residues. This evidence validated our former observation that binding mechanisms of lactogens are hormone-specific and involve binding determinants that are, for some, located at positions that are topologically nonequivalent (1). This prompted us to investigate the functional involvement of the N terminus in hPRL biological properties. We engineered several N-terminal-deleted hPRLs, involving removal of the first nine residues (mutant 1–9-hPRL), to mimic the N terminus of hGH up to the first 14 residues (1–14-hPRL), to delete the entire N-terminal loop. In brief, we found that deletion of the first nine amino acids slightly increased receptor binding affinity and biological activity, an effect presumably mediated by moderate site 1 enhancement, whereas deletion of residues 1–14 slightly decreased binding affinity and biological activity, presumably by affecting site 2 functionality (139).

Although the effects of N-terminal deletions were relatively modest, these mutations were introduced into G129R-hPRL (Fig. 4), because site 1 enhancement and site 2 alteration were expected to be alternative ways to improve the antagonistic properties of this first-generation antagonist (see Section V.B.2). Unexpectedly, despite the fact that 9 and 14 residue deletions have opposite effects on biological properties of hPRL, double mutants (1–9-G129R-hPRL and 1–14-G129R-hPRL) displayed almost superimposable dose-response curves in all bioassays (123). Similarly to G129R-hPRL, they exhibit 10-fold reduced affinity for the human receptor compared with WT hPRL, which confirms the detrimental effect of glycine substitution (106). With respect to in vitro cell bioassays, N-terminal deletions did not improve antagonistic properties, which was again opposite to our expectation. However, and this is the key point, they are markedly different from G129R-hPRL in agonistic assays. Although the latter displays agonistic activities in sensitive bioassays (Ba/F-LP or Nb2 cell proliferation assays) (Fig. 3B), the new analogs failed to stimulate even minimal proliferation. Thus, the absence of residual agonism confers to N-terminal-deleted G129R analogs the great advantage of acting as pure PRLR antagonists. The molecular reasons underlying the abolition of the residual agonistic activity after removal of the N terminus are still unclear. We can speculate that, as observed in oPL, this region is involved in an interaction within binding site 2 and that the abolition of these contacts due to the N-terminal deletions contributes to preventing any interaction with the second receptor; this is very speculative, however.

We have shown that the agonistic properties of G129R-hPRL were also displayed in transgenic animals expressing this analog. Therefore, it was important to assess the pure antagonistic properties of the new compounds in vivo as well. Treatment of mice with very high doses of these new antagonists failed to reveal any detectable agonism. In addition, when coinjected with hPRL into WT mice, a 50-fold molar excess of antagonist totally abolished all PRL-mediated signals investigated (activation of MAPK in mammary tissue and of Stat3 and Stat5 in the liver). The high molar ratio of antagonist vs. PRL was as expected, based on its 10-fold reduced affinity, and it perfectly correlated with in vitro observations. Under the same conditions, G129R-hPRL was much less efficient (123), probably due to its intrinsic agonistic properties. Finally, the ability of second-generation antagonists to counteract the effects mediated by autocrine PRL was tested using probasin-PRL transgenic mice, which overexpress PRL only in the prostate (94). We showed that the morphological phenotype of prostate hyperplasia was accompanied by constitutive activation of PRLR-mediated signaling pathways, such as MAPKs and Stat5. Short-term treatment (30–60 min) with 1–9-G129R-hPRL markedly reduced, or even abolished the activation of these signaling molecules, whereas under identical conditions, G129R-hPRL and S179D-hPRL failed to efficiently oppose effects of locally produced PRL (123). It is noteworthy that efficient inhibition of autocrine PRL in probasin-PRL transgenics required an elevated circulating concentration of 1–9-G129R-hPRL (1–2 µM), which suggests either that expression of the transgene leads to a very high local concentration of PRL or that the high concentration of the 1–9-G129R-hPRL measured in serum is not that actually found within prostate tissue. While awaiting the generation of transgenic mice expressing 1–9-G129R-hPRL, the long-term efficiency of this antagonist was evaluated by implanting osmotic minipumps into probasin-PRL transgenic mice. Despite the relatively low concentration of hormone released by this approach (2–3 nM in serum), the analysis of gene expression profiles clearly discriminated between the action of first- and second-generation antagonists in the prostate: G129R-hPRL resulted in slight but uniform up-regulation of gene expression, whereas 1–9-G129R-hPRL only down-regulated gene expression. Although the concentration of hPRL analo