Introduction

Parathyroid hormone-related protein or PTHrP was originally identified as the tumor product responsible for the clinical syndrome of humoral hypercalcemia of malignancy (HHM), a common metabolic complication of many types of cancer . It is now appreciated that PTHrP is the product of a wide variety of normal tissues and that its role in HHM is the result of an intriguing evolutionary relationship with parathyroid hormone . Furthermore, recent data have demonstrated that PTHrP plays important roles in the development of several organs, including the mammary gland . In this review, we will discuss the current knowledge of the roles that PTHrP plays during mammary development, during lactation and in breast cancer. There are now several in-depth reviews of PTHrP biology available, and the reader is referred to these for a detailed discussion of PTHrP and its gene . However, before proceeding, we will briefly touch upon several themes of PTHrP biology that will serve as necessary background for a full understanding of PTHrP's actions in the mammary gland.

Humoral Hypercalcemia of Malignancy - The Isolation and Characterization of PTHrP

Humoral hypercalcemia of malignancy refers to a syndrome of diffuse osteoclastic bone resorption associated with a tumor distant from the skeleton. Patients suffering from this syndrome have a biochemical profile of hypercalcemia and hypophosphatemia similar to that seen with primary hyperparathyroidism. In fact, historically it was hypothesized that the tumors associated with this syndrome might be making parathyroid hormone (PTH) in an ectopic fashion. However, studies performed in the early 1980's suggested that HHM was caused by a circulating substance that was clearly distinct from PTH, although it had similar actions on PTH-target organs such as bone and kidney . These observations laid the groundwork for the purification of PTHrP from human tumors or cell lines associated with HHM , and this was followed rapidly by the isolation of PTHrP cDNA's in 1987 and the elucidation of the genomic structure of the human PTHrP gene in 1989 .

In the decade that has elapsed since the initial purification of PTHrP, we have learned much about this peptide. Human PTHrP is encoded by a complex gene containing 8 exons spanning 15 kilobases of DNA . In addition, the PTHrP genes of several other species have now been characterized, and it appears that the majority of the PTHrP coding region has been highly conserved throughout evolution. For instance, human and chicken PTHrP are 98% conserved through their first 111 amino acids. In all species, the first 13 amino acids of PTHrP are highly homologous to the same amino acids of PTH (8 are identical) and, although the peptide regions from amino-acids 14-34 differ in their primary amino acid sequence, PTHrP and PTH may share similar secondary structure in these regions. As will be discussed in greater detail below, it is now known that this homology reflects the fact that the PTHrP and PTH genes share a common ancestry and are members of a small gene family. This information was particularly enlightening in regard to the pathogenesis of HHM, since these regions of shared primary and secondary structure correspond to the PTH-like regions that are important for receptor binding and activation. Indeed, many studies have now documented that amino-terminal fragments of PTHrP and PTH itself share the use of the same receptor, termed the PTH/PTHrP receptor (see below).

Another aspect of PTHrP biology that is now firmly established is the fact that the primary transcript of PTHrP is a polyprotein, similar to pro-opiomelanocortin (POMC) . The amino acid sequence of PTHrP contains multiple clusters of basic amino acids arranged in pairs, triplets and quadruplets that serve as post-translational processing sites. As a consequence, there are multiple forms of PTHrP, including several species containing the "PTH-like" amino-terminus, as well as mid-region and C-terminal fragments, all of which appear to have distinct biological actions subserved by distinct receptors. The study of the post-translational processing of PTHrP is ongoing, but it is clear that, similar to the case of POMC, different tissues expressing the PTHrP gene may generate different peptides with distinct biological profiles.

Amino-terminal fragments of PTHrP make use of a G protein-coupled, 7-transmembrane-spanning receptor, termed the PTH/PTHrP receptor. This receptor was cloned in 1991, and it is the "classical" PTH receptor involved in systemic calcium homeostasis . Prior to its isolation, several groups had shown that amino-terminal fragments of PTH and PTHrP were equipotent in stimulating this receptor . However, once its expression pattern was examined, it became clear that this protein also serves as a PTHrP receptor, and it was subsequently named the PTH/PTHrP receptor. Like PTHrP, the PTH/PTHrP receptor is widely expressed and is found in many tissues that are not involved in the regulation of calcium homeostasis. Furthermore, in these sites it is often expressed in cells adjacent to those cells expressing PTHrP . Although it is now clear that amino-terminal PTHrP acts through the PTH/PTHrP receptor, the receptors that recognize other fragments of PTHrP remain uncharacterized. There appear to be receptors for the mid-region fragment of PTHrP, and there may be other receptors that recognize the amino-terminal portions of PTHrP as well . The characterization and isolation of these other PTHrP receptors remains an area of active research.

PTH and PTHrP: A Gene Family

As noted above, the amino-terminal portion of PTHrP and PTH are highly homologous. This knowledge generated a great deal of excitement when PTHrP was first isolated because it provided a rationale for the similarities between hyperparathyroidism and HHM. With the cloning of the PTHrP gene it became clear that the homologous amino-terminal regions of PTH and PTHrP were the result of a true evolutionary relationship between the PTH and PTHrP genes. The human PTHrP gene is located on the short arm of chromosome 12, while the human PTH gene is located on the short arm of chromosome 11 ; these two chromosomes arose as the result of an ancient tetraploidization event . In addition, there is a similar organization to the PTH gene and the portions of the PTHrP gene encoding the prepro and coding regions of PTHrP. These facts suggest that PTH and PTHrP arose from a common ancestral gene as a result of the above-mentioned tetraploidization event, some 200 to 300 million years ago.

This time frame is consistent with the apparent emergence of PTH as a distinct gene in amphibians . (Although there have been several preliminary reports of PTH-like sequences in fishes, no distinct gene has been cloned.) The PTHrP gene has been characterized as far back as chickens and appears to be highly conserved over evolutionary time, but there is currently no information as to the presence of a distinct PTHrP gene in either amphibians or fishes. Whatever the exact point of divergence from their ancestral gene, it is clear that these two genes have evolved separately. The PTH gene has a simple organization; it is expressed solely by the parathyroid chief cells, and PTH functions as a classical peptide hormone to regulate systemic calcium homeostasis. In contrast, the PTHrP gene has a complex organization; it is expressed in a wide variety of cells, and PTHrP appears to function to influence cellular growth and differentiation in an autocrine and/or paracrine fashion.

What has emerged from this information is a picture of PTH and PTHrP as two ancestrally-related ligands that share a common receptor. The systemic calcium metabolism-regulating functions of PTH are mediated by PTH/PTHrP receptors in the classical target tissues of bone and kidney, and the functions of PTHrP are mediated by PTH/PTHrP receptors on a variety of different cell types located in tissues not participating in systemic calcium metabolism . Obviously, if this is the case, there must be mechanisms to ensure that the correct ligand talks to the correct receptor population, and HHM represents the consequences of a breakdown of the barriers that normally separate the realms of PTH and PTHrP. Tumors both overproduce PTHrP and disrupt the mechanisms that normally exclude it from the circulation, allowing PTHrP to alter calcium homeostasis by gaining access to the PTH/PTHrP receptors in bone and kidney that are normally reserved for its cousin, PTH. As we will discuss below, the only other time in which PTHrP gains access to the circulation appears to be during lactation, where it may play a specialized role in calcium homeostasis.

PTHrP as a Developmental Regulatory Molecule.

Soon after its discovery, it became clear that PTHrP was expressed during embryogenesis in a wide variety of tissues derived from all three germ layers . Recent studies have found a similar widespread distribution of PTH/PTHrP receptor expression during fetal life . In fact, PTHrP and the PTH/PTHrP receptor have been reported to be one of the earliest peptide hormone/receptor pairs to be detected during mouse development, being expressed from the late morula stage onward . The exact role that PTHrP plays during early development is still unclear, but recent studies have suggested that PTHrP has important functions during organogenesis. At this point, there are two well documented areas in which PTHrP has been shown to regulate developmental events. The first is fetal bone development. Here, PTHrP appears to regulate the rate of chondrocyte differentiation and apoptosis within the growth plate of the developing long bones , and appears to act downstream of Indian hedgehog and BMP's . The second is the regulation of epithelial-mesenchymal interactions during the formation of epithelial organs. In this regard, PTHrP has been implicated in the regulation of the development of hair follicles , teeth (William Philbrick, unpublished observations), the epidermis, the lungs, and the mammary gland . To date, we have the most detailed information about PTHrP's effects on mammary development, and thus we will discuss this in detail.

The Role of PTHrP in Embryonic Mammary Gland Development

Recent studies from our laboratory have documented that PTHrP is necessary for mammary development . In the absence of a functional PTHrP gene, mammary development fails during embryonic life, and no mammary epithelial ducts are formed (Fig. 1). As described below, this appears to result from a failure of epithelial/mesenchymal interactions around the time of the initiation of ductal branching morphogenesis.

Before discussing the mammary phenotype of PTHrP-knockout mice in detail, it is useful to review the embryonic development of the murine mammary gland . This is essentially a two-step process. The first step involves the formation of five pairs of mammary buds. These structures form as the result of an invagination of epidermal cells along the mammary streak, a thickening of ventral epidermis stretching between anterior and posterior limb buds bilaterally. The mammary streak is first visible on E10 and the mammary buds are fully formed by E12. On E13, the fetal testes begin to produce androgen, and in male embryos this leads to the destruction of the mammary rudiments. In female mice, the mammary buds remain quiescent until E16 when they undergo a transition to the second step of embryonic development, the formation of the initial ductal tree. This process involves the elongation of the mammary bud, its penetration into the mammary fat pad precursor and the initiation of ductal branching morphogenesis. By birth, this process gives rise to a primary mammary epithelial tree consisting of 15-20 branching ducts contained within the mammary fat pad. This initial pattern persists until puberty, at which time the mature virgin gland is formed through a second round of branching morphogenesis, regulated by circulating hormones (see below).

In the absence of PTHrP, mammary development fails at the transition between the two phases of development outlined above . In PTHrP-knockout mice, mammary buds form normally. However, between E16 and E18, they fail to undergo the initial round of ductal branching morphogenesis (Fig. 2). Instead, in PTHrP-knockout embryos, the mammary rudiment remains bud-like in appearance, does not penetrate into the fat pad precursor, and becomes surrounded by a dense condensation of stroma. Subsequently, the mammary epithelial cells degenerate and, by birth, disappear altogether. At this point, it is not clear how the epithelial cells die. Our studies thus far do not support either necrosis or classical apoptosis as a mechanism (Fig. 3). We do know that this phenotype is the result of the loss of amino-terminal PTHrP, for PTH/PTHrP receptor knockout mice have an identical phenotype. Finally, reintroduction of PTHrP into the epithelial cells of the PTHrP knockout mammary buds at E15-E16 by breeding a keratin 14-PTHrP transgene onto the homozygous PTHrP-null background results in the restoration of near-normal embryonic mammary development (Fig. 4). Hence, it is clear from these experiments that amino-terminal PTHrP must interact with the PTH/PTHrP receptor in order for mammary epithelial development to proceed past the mammary bud stage.

In order to begin to understand the mechanisms by which PTHrP might act during embryonic mammary development, we next determined the cellular localization of PTHrP and PTH/PTHrP receptor expression by in situ hybridization . PTHrP mRNA is expressed at high levels in mammary epithelial cells of the embryonic mammary bud from E12 on, as well as in the growing ducts during the initial phases of ductal branching morphogenesis at E18 (Fig. 5A-C). PTHrP expression is especially intense in the cells located peripherally, adjacent to the basement membrane. It is also expressed in developing hair follicles, but there appears to be minimal expression within the embryonic epidermis in general. In contrast to the epithelial pattern of PTHrP expression, PTH/PTHrP receptor expression is restricted to mesenchymal cells (Fig. 5D-F). At E12, PTH/PTHrP receptor mRNA is expressed throughout the ventral dermis, including the dense mammary mesenchyme. At E18, as the mammary ducts grow and invade the developing fatty stroma, PTH/PTHrP receptor mRNA expression continues to be expressed in stromal cells immediately surrounding the growing mammary duct. Therefore, within the embryonic mammary gland, it appears that PTHrP and the PTH/PTHrP receptor represent a epithelial/mesenchymal signaling unit in which PTHrP is produced by mammary epithelial cells and interacts with its receptor on mammary mesenchymal cells. In the context of the phenotypes of the PTHrP and PTH/PTHrP knockout mice, this would suggest that PTHrP initiates a sequence of epithelial/mesenchymal interactions that lead to the formation of the primary epithelial duct system. Specifically, PTHrP appears to be an epithelial signal that modulates stromal cell function in a fashion that is critical for the transition of the mammary bud into an active phase of ductal branching morphogenesis.

Following birth, the murine mammary gland undergoes little development until the onset of puberty at 3-4 weeks of age . During sexual maturation, under hormonal influence, the distal ends of the mammary ducts form specialized structures called terminal end buds, which serve as the sites of active cellular proliferation and differentiation during ductal morphogenesis. By 8-10 weeks of age, the epithelial duct system has grown to the borders of the mammary fat pad and the terminal end buds disappear, leaving the typical branched duct system found in the adult virgin gland. After the onset of pregnancy there is another round of ductular proliferation that leads to the production of terminal ducts, and this is then followed by the formation of lobulo-alveolar structures. It appears that, in addition to embryonic development, PTHrP may also play a role in the epithelial/mesenchymal interactions governing these later rounds of ductal morphogenesis.

As discussed in the section on lactation, it was known that PTHrP is expressed in the pregnant and lactating mammary gland . More recently, our laboratory has shown that PTHrP and the PTH/PTHrP receptor are also expressed in the preadolescent and adolescent mammary gland, as well as during early to mid pregnancy . As noted in embryonic development, by in situ hybridization, PTHrP appears to be expressed in epithelial cells and the PTH/PTHrP receptor appears to be predominantly expressed in stromal cells surrounding the PTHrP-expressing epithelial cells. However, overall, we found expression levels to be less intense in the post natal gland and, in addition, expression of both PTHrP and the PTH/PTHrP receptor appeared to be restricted to terminal end buds. No PTHrP mRNA expression was detected in mature mammary ducts during either puberty or pregnancy, although we could detect some expression of PTH/PTHrP receptor mRNA in periductal stromal cells, especially during early pregnancy. These studies demonstrate that during the post-natal phases of active mammary ductular morphogenesis, PTHrP is expressed in epithelial cells, and the PTH/PTHrP receptor is expressed in mammary stromal cells. In addition, it appears that the expression of PTHrP and the PTH/PTHrP receptor are most intense in regions of the mammary gland that are actively proliferating and undergoing ductal morphogenesis, the terminal end-buds.

Consistent with these observations, studies in transgenic mice have shown that overexpression of PTHrP in mammary myoepithelial cells, driven by the keratin 14 (K14) promoter, results in abnormal mammary ductal development during puberty and early pregnancy (Fig. 6) . During puberty, PTHrP-overexpression results in severe impairments in both the overall rate of ductal proliferation as well as in the pattern of side branching. This results in a delay in ductal growth into the mammary fat pad and in a much simpler overall duct structure in transgenic mice. In addition, PTHrP-overexpression results in an impairment of terminal duct development during early pregnancy. These effects are mediated by amino-terminal PTHrP acting through the PTH/PTHrP receptor, as overexpression of PTH produced a similar phenotype and local administration of PTHrP(1-36) within the fat pads of normal mice also impairs ductular proliferation . These studies suggest that PTHrP continues to participate in epithelial-mesenchymal interactions in the post-natal mammary gland and may contribute to the regulation of branching ductal morphogenesis during puberty and early pregnancy.

The experiments outlined above suggest that PTHrP, produced by mammary epithelial cells, acts to modulate stromal cell function at several different phases of ductal morphogenesis. We would propose that PTHrP interacts with the PTH/PTHrP receptor on mammary stromal cells and that this interaction is critical to the ability of the stroma to support proper ductal branching morphogenesis. In addition to the phenotypes of PTHrP-knockout and overexpressing mice detailed above, several experiments support this hypothesis. First, we have found that primary cultures of mouse mammary stromal cells express PTH/PTHrP receptor mRNA, but no PTHrP mRNA. Conversely, mammary epithelial cells from freshly isolated organoids, express PTHrP mRNA but not PTH/PTHrP receptor mRNA. These results lend further support to the epithelial/mesenchymal pattern of expression of PTHrP and PTH/PTHrP receptor in the mammary gland. In addition, mammary stromal cells in culture bind amino-terminal PTHrP and respond with an increase in intracellular cAMP . Together, these observations confirm that mammary stromal cells contain functional PTH/PTHrP receptors. Finally heterotypic tissue recombination experiments have documented that PTHrP signaling is necessary for mesenchymal cells to support epithelial morphogenesis . Mammary epithelium from PTH/PTHrP receptor knockout embryos was able to form ducts when combined with normal mesenchyme and transplanted beneath the kidney capsule of athymic mice. However, mesenchymal cells from PTH/PTHrP receptor knockout embryos were not able to support the outgrowth of either receptor knockout epithelium or normal epithelium. These results support the idea that stromal cells must interact with PTHrP in order to support normal epithelial morphogenesis.

The molecular mechanisms by which PTHrP exerts its effects on mammary stromal cells remain unclear. Studies have shown that the stroma secretes growth factors that regulate epithelial morphogenesis . In addition, the stroma contributes to the extracellular matrix, whose composition can have profound influences on epithelial cell behavior . The mammary stroma is also the source of matrix mettalloproteinases that are involved in remodeling the ECM during ductal branching morphogenesis . All of these molecules are potential downstream stromal effectors of PTHrP’s actions on epithelial development, and experiments are currently underway to examine PTHrP’s effects on their expression in cultured mammary stromal cells. However, irrespective of the exact stromal response to PTHrP, it is worth noting that these experiments underscore the truly reciprocal nature of the epithelial/mesenchymal interactions that guide epithelial morphogenesis. Although stromal cells influence the form and function of the epithelium, the epithelial cells clearly participate in the regulation of their own fate, for without epithelial signals, such as PTHrP, the stromal cells are incompetent to direct epithelial morphogenesis.

Role of PTHrP During Lactation

Lactating humans supply between 300 - 400 mg of calcium per day to their offspring in breast milk, a requirement that puts them into negative calcium balance . Many studies have now demonstrated that this excess calcium is supplied by the maternal skeleton, and it is well known that lactating humans transiently lose between 5 - 8% of their skeletal mass during the first 6 months of lactation . This phenomenon is even more impressive in rodents; it has been reported that lactating rats may loose up to 35% of their trabecular bone mass in the first 21 days after delivery . Furthermore, in both humans and rodents, this bone loss has been shown to be associated with a state of increased bone turnover . There has been great interest among bone biologists in understanding the mechanisms underlying this mobilization of calcium from the maternal skeleton. Studies examining the classical calciotrophic hormones have documented that neither PTH nor 1,25 (OH)₂D seem to be necessary or sufficient to account for the mineral fluxes that take place during this period . Thus, there appears to be a "missing link" in the control of maternal mineral metabolism and the mobilization of calcium for milk production. When it was discovered that PTHrP was produced by the lactating breast (see below), this molecule became a natural candidate as this "missing link" and, as we will review in the following paragraphs, there is growing evidence to support such an idea.

Thiede and Rodan were the first to report that the lactating mammary gland produced PTHrP. They demonstrated a rapid and transient increase in PTHrP mRNA and protein content in the mammary glands of lactating rats in response to suckling . Early in lactation, the induction of PTHrP expression appeared to be dependent on the suckling-induced rise in serum prolactin, because it could be reproduced by the injection of prolactin into unsuckled puerperal animals and also could be blocked by the administration of bromocriptine, an inhibitor of prolactin secretion by the pituitary . It would appear that most of the PTHrP produced during lactation is secreted into milk, where it has been found in high concentrations . PTHrP has now been found in the milk of a number of mammalian species, and levels have generally been approximately 1,000-fold higher than those seen in the circulation of patients with HHM and 10,000-fold higher than those found in normal controls . In addition to its regulation by prolactin, PTHrP production may also be regulated by local factors. Thompson and colleagues have demonstrated that unilateral milking of a goat will cause an increase in PTHrP concentrations in the milk from the side which was emptied, but not from the contralateral side . Furthermore, it has been shown that PTHrP concentrations in milk rise with the duration of lactation in the rat, and that PTHrP mRNA expression becomes prolactin-independent after day 16 of lactation in the rat . Finally, two recent studies have suggested that there are different isoforms of PTHrP in milk, and that the concentrations of non-amino-terminal fragments of PTHrP correlated with the calcium content of the milk . This is especially interesting since mid-region PTHrP has been implicated in the regulation of calcium transport across the placenta during pregnancy ; perhaps PTHrP plays a similar role in facilitating calcium transport into milk. Clearly, there remains much to be learned about the regulation of PTHrP production and secretion into milk during lactation.

If the PTHrP produced within the lactating breast is to serve an endocrine function to mobilize skeletal calcium stores for milk production, in addition to being secreted into the milk, it must also be secreted into the systemic circulation. Here the data is somewhat conflicting, but the weight of evidence now appears to suggest that PTHrP is secreted into the general circulation from the lactating breast. There is evidence for a suckling-associated increase in phosphaturia in the rat and cow, as well as suckling-associated increases in urinary cAMP excretion in the rat . These findings would support the possibility that significant concentrations of PTHrP make it into the systemic circulation after suckling, since PTHrP is known to have both of these effects in the kidney. In addition, in thyroparathyroidectomized cows (unable to secrete PTH), this suckling-associated phosphaturia was completely abolished by treatment with a PTH/PTHrP receptor antagonist . There have now been a handful of studies that have attempted to measure PTHrP in the circulation of lactating animals and humans . It appears that PTHrP is present in the venous drainage of the mammary gland in lactating goats and cows . Most, but not all, studies in humans have found elevated levels of PTHrP in the plasma of lactating women . In one large study, levels of PTHrP during lactation were reported to correlate positively with serum prolactin levels and bone turnover markers and negatively with estradiol levels and with bone density measurements .

These data suggest the following hypothesis. PTHrP production by the lactating mammary gland is stimulated by suckling in response to both systemic factors (prolactin) and local factors within the mammary gland itself. The majority of PTHrP so produced is secreted into milk, but some is released into the systemic circulation in order to stimulate bone resorption, mobilizing calcium for further milk production and perhaps facilitating its transport into milk. This is an attractive hypothesis, but two points are worth noting. First, it is possible that the PTHrP affecting bone metabolism does not originate from the mammary gland itself. Circulating levels of PTHrP been found to be elevated in patients with hyperprolactinemia secondary to pituitary adenomas, and it was suggested the pituitary was the source of the PTHrP in these patients . This could also be true during lactation. In addition, it has been shown that osteoblasts possess prolactin receptors, and it is known that these cells also secrete PTHrP . Hence, it is distinctly possible that PTHrP is secreted into the bone microenvironment by osteoblasts, in response to prolactin, and that changes in mineral ion homeostasis do not rely on PTHrP secreted from the breast. Second, passive immunization of lactating mice with anti-PTHrP antibodies does not influence the calcium content of milk, nor does it affect calcium homeostasis in the lactating mother . Therefore, although PTHrP may play some role in mineral metabolism during lactating, its exact contribution remains a matter of debate. The resolution of this issue is of some importance, for an endocrine function for PTHrP during lactation would represent a physiological "PTH-like" effect of PTHrP and would perhaps explain some of the evolutionary pressure that has maintained the homology between the amino-termini of these molecules and their sharing of the same receptor.

In addition to the possible role(s) of PTHrP in calcium metabolism, there have been two other suggested functions for PTHrP during lactation. First, based on PTHrP's actions as a vasodilator, it has been suggested that PTHrP may regulate blood flow to the mammary gland. In support of this concept, it has been shown that the injection of PTHrP into the mammary arteries of dried ewes increased mammary blood flow and antagonized the vasoconstrictive effects of endothelin . Furthermore, a preliminary report has shown that the nutrient arteries of the inguinal mammary glands of rats express PTHrP mRNA and that PTHrP expression was increased five-fold in response to suckling and/or prolactin . Second, it has been suggested that the PTHrP found in breast milk affects calcium metabolism and/or GI development within the neonate, but there is little experimental evidence to support any function for milk-derived PTHrP in the neonate .

PTHrP's Role in Breast Cancer

It has long been appreciated that breast cancer cells have a propensity to metastasize to bone . This "osteotrophism" results in an incidence of skeletal metastases as high as 70% in series of women with advanced disease . A significant proportion of these patients also develop hypercalcemia, both due to local bone destruction or osteolysis as well as in association with classical HHM . It has been known that PTHrP was the cause of humoral hypercalcemia associated with breast cancer for some time, but exciting new work also suggests that this molecule may be involved in the local ostolysis associated with these tumors and, in fact, may play an important role in the osteotrophism itself.

PTHrP has now been reported to be produced by both cultured breast cancer cell lines as well as by primary human breast cancers . Several histological surveys have demonstrated that about 50% - 60% percent of primary breast tumors express PTHrP . In addition, a series of observations has demonstrated that by both immunohistochemistry and in situ hybridization, PTHrP is more commonly expressed in breast cancer metastatic to bone as compared to other sites of metastasis or within the primary tumor itself . These observations suggest that either PTHrP expression within a given population of tumor cells might be involved in establishing the bony metastasis, or that once within the bone microenvironment, tumor cells upregulate PTHrP expression. There is now evidence for both of these possibilities. First, several clinical studies have shown that PTHrP expression within a primary breast tumor is predictive of bony metastases. Second, in an elegant series of animal experiments, Guise and colleagues have demonstrated that levels of PTHrP expression within breast cancer cells influenced in the likelihood of metastasizing to bone. When introduced into the left ventricle of nude mice, MDA-MB-23 cells engineered to overexpress PTHrP produced 3 times more bone metastases than native MDA-MB-23 cells, which express low levels of PTHrP. In addition, despite the low levels of PTHrP produced by native MDA-MB-23 cells, treatment of the mice with PTHrP antisera was able to reduce the number and size of bone metastases produced by these cells by up to five-fold (Fig. 7) . Therefore, there is now good evidence to suggest that PTHrP production facilitates the establishment of breast cancer cells within the skeleton.

If PTHrP is involved in the osteotrophism of breast cancer cells, it most likely does so through its ability to activate osteoclastic bone resorption. In order for a metastatic cell to take up residence in bone, it must be able to recruit the resident osteoclasts to resorb mineralized tissue . This osteolysis releases a series of growth factors from the bone matrix that, in turn, nourish the tumor cells, giving them a growth advantage and causing more local bone resorption . It is thought that this viscous cycle underlies the pernicious behavior of breast cancer metastatic to bone and, as a proof of concept, several large clinical trials have now demonstrated that inhibition of osteoclastic bone resorption in patients with breast cancer, using a class of drugs known as bisphosphonates, inhibits both the occurrence and growth of bony metastases . Again, there is evidence from the Guise group that upregulation of PTHrP expression in breast cancer cells within the bone environment may play a role in the establishment of the viscous cycle described above. TGF-b is plentiful in the bone matrix and is released into the microenvironment upon osteoclastic bone resorption . Furthermore, TGF-b has been shown to upregulate PTHrP expression in breast cancer cell lines . By introducing a mutant, dominant negative TGF-b type II receptor into MDA-MB-23 cells, Guise and colleagues were able to abolish the upregulation of PTHrP by TGF-b and to reduce the occurrence and growth of bone metastases in their animal model .

In summary, the concept that emerges from these data is that PTHrP, by virtue of its ability to activate osteoclastic bone resorption, helps a breast cancer cell obtain a foothold in bone. Once established in the skeleton, the tumor cells, responding to factors like TGF-b that are released from the bone matrix, upregulate their production of PTHrP, which contributes to further bone resorption and the continued growth of the tumor cells. If this is true, PTHrP and its receptor(s) may be useful therapeutic targets for therapy aimed at preventing bone metastases.

Summary

PTHrP is involved in the physiology and pathophysiology of the mammary gland from start to finish. First, it is a necessary participant in the regulation of mammary mesenchymal cell function during embryogenesis, and it appears to continue to participate in epithelial-mesenchymal interactions during later stages of ductal morphogenesis. Second, it appears to play a role in mobilizing calcium from the maternal skeleton during lactation. Finally, it appears to be an important player in the osteotrophism that makes metastatic breast cancer such a devastating disease. It is likely that further research into the roles that this remarkable molecule plays in the mammary gland will continue to shed light on both PTHrP physiology as well as mammary gland biology itself. In addition to the areas highlighted in this review, there remain unexplored, yet important questions to address. For instance, does PTHrP play a role in mammary epithelial cell differentiation or transformation? Does it regulate tumor growth by influencing tumor-stromal interactions? Is it involved in the earlier stages of metastasis, or only in the bone-seeking behavior of tumor cells? Does it regulate calcium transport across the mammary epithelium and into milk? We look forward to the answers to these and other unanticipated questions that we are certain will arise in the future.