Introduction
The mammalian peptide hormone relaxin and its human analogue H2 relaxin or relaxin-2 are well known for their matrix-modifying capacity. They induce extensive tissue remodeling via upregulation of matrix metalloproteinases (MMP) and angiogenesis [
1,
2]. Since this is also a key feature of malignant invasion, they have been implicated in cancer progression. However, their role in this context is still unclear.
The most unambiguous data exist for prostate cancer. In xenografts of relaxin-2-overexpressing prostate cancer cell lines, tumour growth and neoangiogenesis are significantly enhanced [
3]. The opposite can be observed after transfection of an antagonistic relaxin-2 analogue lacking the binding domain for the relaxin receptor RXFP1 [
4]. Increased production of relaxin-2 as well as relaxin by tumour and stromal cells has been demonstrated in advanced prostate cancers in humans and mice [
5,
6]. Relaxin enhances proliferation, adhesion and invasion in vitro as well as tumour growth in the TRAMP mouse model in vivo, while deficiency of RXFP1 antagonizes this effect [
5,
7]. A tumour-promoting function of relaxin-2 has been demonstrated also in thyroid cancer where it fosters invasion in vitro via increased invadopodia formation and MMP upregulation as well as xenograft growth in nude mice [
8,
9]. Relaxin-2 expression was also enhanced in advanced human endometrium cancers correlating with unfavourable clinical outcome [
10].
In breast cancer, the situation is much less clear. Elevated expression of relaxin-2 has been demonstrated in neoplastic mammary tissues as compared to their benign counterparts [
11]. Porcine relaxin as well as relaxin-2 was found to influence tumour cell proliferation in a biphasic way regarding time course and concentration. While low amounts and short-term application resulted in enhanced in vitro growth of MCF-7 and MDA MB-231 cells, high concentrations and long-term exposure yielded the opposite effect and diminished the growth of the respective xenografts in nude mice [
12‐
14]. We showed that 5-day exposure of MCF-7 and SK-BR3 cells to 100 ng/ml of porcine relaxin as well as recombinant relaxin-2 was followed by MMP upregulation and enhanced invasiveness. This could be inhibited by the antagonistic relaxin-2 analogue B-R13/17 K [
15,
16]. Consistently, relaxin-2 serum levels were significantly elevated in patients with metastatic breast cancer correlating with short survival [
17].
To further complicate matters, relaxins do not only act on the tumour cells but also on the benign cells of the surrounding stromal compartment. Components of the tumour microenvironment are essentially involved in malignant progression, in particular, the tumour-associated macrophages (TAM) which are characterized by a so-called M2 phenotype with tumour-promoting function [
18]. We have recently shown that not only the TAM, infiltrating from the peripheral blood, but also resident macrophages at the site of metastasis are critical for the colonization of distant organs [
19,
20]. Interestingly, Figueiredo et al. [
21] have shown that relaxin inhibits expression of the typical M1-cytokine IL 1β in rat macrophages, thus indicating a potential role for relaxins in the tumour-associated phenotype shift to M2.
Given the impact of microenvironment effects on tumour growth we searched for a model where the influence of relaxins on breast cancer progression could be studied without confounding factors, such as xenografting, artificial cancer induction/injection or immunodeficiency. We therefore chose the Tg(MMTV-erbB2) mouse model, where breast cancers arise spontaneously due to transgenic erbB2-overexpression [
22]. Since erbB2-overexpressing tumours are clinically aggressive and often metastasize into the brain [
23], we additionally used the organotypic brain slice coculture, an ex vivo model recently established by our group [
19], to study the effect of porcine relaxin as well as the human brain isoform relaxin-3 on the colonization of the central nervous system.
Discussion
While there is increasing evidence of a tumour-promoting role of relaxins in prostate and several other cancers, the data for breast cancer are still contradictory. Here we show that porcine relaxin significantly enhances growth of breast cancers which had developed spontaneously in an erbB2-overexpressing mouse model. This was associated with a significantly higher proliferation rate in tumours from relaxin-treated animals as well as with upregulation of RXFP1 expression. Additionally, relaxin-treated animals had significantly higher serum levels of E2 and P4, which was accompanied by increased expression of the respective receptors in the tumours. Induction of RXFP1 either directly by its ligand or indirectly via E2 which, in turn, can be upregulated by relaxin [
30]), has already been demonstrated in the marmoset monkey [
30] and in the pig [
31]. In the mouse cervix, activation of ERα-signaling was necessary to enable the proliferative effect of relaxin [
32]. This suggests that in our model relaxin stimulates tumour growth by one or both of the following options: either directly via upregulation of its own receptor or indirectly via induction of sex hormones which then enhance proliferation through their corresponding receptors.
As an indication that relaxin did not only act on tumour but also on stromal cells, we found significantly elevated amounts of infiltrating TAM in tumours from relaxin-treated animals. Although the TAM were localized predominantly in the stromal compartment, they also infiltrated into the tumour tissue itself. This demonstrates that interaction between tumour and stroma in this spontaneous model could occur freely without interference of local obstacles, such as formation of a surrounding fibrous capsule, or of species barriers as often the case in artificially induced cancers in both immunocompetent and deficient mice [
5,
12,
13]. Since high amounts of TAM are well-known to confer an unfavourable clinical outcome [
33], relaxin-induced TAM infiltration may have additionally contributed to enhanced tumour growth.
These data are consistent with our earlier findings that both porcine relaxin and human relaxin-2 enhance breast cancer cell invasion [
15,
16]. In vivo, we could show that elevated relaxin-2 serum levels in breast cancer patients correlate with metastatic disease [
17] and high RXFP1 mRNA levels are an independent marker of metastasis and shortened survival in dogs [
34]. In contrast to these results, other authors described reduced growth of relaxin-2-overexpressing MDA-MB 231 cells [
13] as well as a differentiating effect of relaxin on MCF-7 cells in nude mice [
12]. Both groups used xenograft models in immunodeficient animals with all the potential problems regarding the microenvironment interaction mentioned above. Additionally, the MDA-MB 231 transfectants produced up to 50-fold enhanced local relaxin-2 concentrations [
13]. This is clearly higher than in our model where systemic relaxin levels were around 10
−10 M and only about fivefold higher in relaxin-treated animals than in controls. The amount of relaxin is an important variable, since Sacchi et al. [
14] have shown that MCF-7 cells respond to higher versus lower relaxin concentrations with opposite functional outcome, levels of <8 × 10
−10 M inducing tumour cell proliferation.
The biphasic concentration effect may also explain the divergent data of another group [
35,
36] which generated high intratumoural concentrations of relaxin either by direct overexpression of relaxin in the injected breast cancer cells or by delivery via infiltrating monocytic cells with artificial relaxin overexpression. Although there was excessive matrix remodeling around the tumours which usually is a hallmark of cancer progression, the tumours grew slower than the controls. This surprising effect was attributed to facilitated accessibility for lymphocytes and macrophages with antitumour activity. Supposedly, local relaxin levels were rather high in this model, RLN1 mRNA expression being described as more than fivefold increased in the monocyte approach and 50–180 fold in the transfected tumour cells. Since high concentrations of relaxin have been described to induce a cytotoxic TH1-shift in CD4 + T cells [
37], this would in fact support the concept of a T-cell induced antitumour response in this model. However, in vivo tumour-infiltrating T-cells and macrophages usually display a tumour-promoting TH2 and M2 phenotype. Figueiredo et al. [
21] have shown that concentrations of relaxin-2 as low as 0.02 ng/ml shift the macrophage phenotype from M1 to M2. Consistently, in our model with relatively low relaxin levels, enhanced infiltration by TAM upon relaxin treatment was not associated with inhibition of tumour growth but with the opposite.
Until now the influence of relaxins on spontaneous metastasis formation is unclear. Since we could not detect any metastases during the life span of our mice, we addressed this question in a brain slice coculture model [
19]. There we had shown before that tumour cells need the active help of microglia, the resident macrophages of the brain, for successful invasion. First, microglia is attracted and accumulates at the site of the potential tumour cell entry. Activated microglia then serve as vehicles and guidance structures for the malignant cells into the brain. Porcine relaxin increased microglia accumulation and invasion of each of the tumour cell lines investigated in this model, independent of the species of origin as well as of erbB2 and hormone receptor expression. The effect was dose-dependent and became most significant in higher concentrations. These concentrations are higher than those used in our previous Boyden chamber experiments [
15], however, since relaxin is distributed via diffusion within the brain slice, they are necessary to ensure sufficient penetration. To further characterize the receptors which mediate the pro-invasive effect, we applied the brain-specific relaxin-3 which binds with high affinity to its primary receptor RXFP3 but can also interact with lower affinity with RXFP1. Both receptors were present in the brain, but only RXFP1 was expressed on the tumour cells. Synthetic human relaxin-3 induced a similar pro-invasive effect as porcine relaxin. In comparison, the specific RXFP3-agonist R3/I5 did not significantly enhance invasion. Thus, it can be assumed that most of the pro-invasive effect is mediated by RXFP1 both on tumour cells and their stromal recipients, the microglia.
These results are consistent with a recent clinical observation in a patient with hormone receptor-negative breast cancer [
38]. There, we found an unusually fulminant progression from a single small osteolytic lesion to widely disseminated metastases in multiple organs during hormonal treatment for oocyte harvest before the planned chemotherapy. This was associated with induction of high relaxin-2 serum levels, strongly suggesting a tumour-promoting effect of relaxin-2 either directly on the tumour cells or via the microenvironment.
Taken together, in a mouse model of spontaneous erbB2-positive breast cancer, where interactions between tumour cells and the stromal compartment are not hampered by species barriers and immunodeficiency, relaxins clearly promote tumour progression. This is obviously due to effects of relaxins on the tumour cells but also on the microenvironment, in particular, the macrophages. We also demonstrate that relaxins can foster metastatic colonization of the brain, again, via direct action on the tumour cells but also on the recipient tissue. These findings strongly argue in favour of a tumour-promoting role for relaxins in breast cancer. Given the context-dependency of relaxin action, further experiments with special attention to an intact microenvironment and to relaxin dosage are warranted.