**5. Results**

#### **5.1. Characterization of PRLR isoform specific monoclonal antibodies**

We have recently prepared and characterized PRLR isoform specific polyclonal antibodies [5]. However, a limitless source of polyclonal antibody is restricted because it is generated by repeated immunization of a rabbit. Each antibody preparation has the potential to vary in its specificity as well. Monoclonal antibodies are more homogeneous and can produce a near inexhaustible supply of reagent. We used the same peptide sequence as for the preparation of isoform specific polyclonal antibodies and, again, demonstrate specificity for the individual isoforms. CHO cells were transiently transfected with cDNA containing the individual PRLR isoforms; western blot analysis from cell lysates indicate that the antibodies detect the corresponding protein without cross-reactivity (Figure 1A).

By immunocytochemical analysis using transiently transfected CHO cells, we were able to positively stain for isoform specific PRLR. As shown in Figure 1B, no cross-reactivity was identified using the antibodies on cells transfected with non-corresponding PRLR cDNA. By both western and immunochemical analysis we were able to demonstrate the specificity of the PRLR isoform specific monoclonal antibodies.

#### **5.2. Immunofluorescent staining of human breast tissue**

Once we determined the specificity of the isoform specific monoclonal antibodies on transfected cells, we tested them on histosections obtained from reduction mammoplasty specimens. In order to examine the localization patterns of the PRLR, we stained the tissues simultaneously with LF and either SF1a or SF1b antibodies. Similar, striking, distribution of isoform expression was observed, regardless of whether paraffin-embedded or frozen sections were used. As seen in Figure 2, LF expression appeared primarily basal, whereas SF localized more luminally.

**5. Results** 

solution for 30 min at 37o C in a humidity chamber to prevent evaporation. PRLR primary antibody-oligonucleotide conjugates were diluted in PLA Probe Diluent to 5 g/ml, and applied to slides overnight at 4o C in a cooled humidity chamber. On the second day, slides were washed 3 times in 1X Duolink Wash Buffer A to remove excess antibodyoligonucleotide conjugates. Bound antibody-oligonucleotide conjugates were ligated together for 30 min at 37o C according to the manufacturer's instructions. Subsequently, ligated templates were amplified for 2 hr at 37o C according to the manufacturer's protocol, washed twice for 10 min in Duolink 1X Wash Buffer B, and once for 1 min in 0.1X Wash Buffer B prior to mounting with Duolink Fluorescence mounting media (which also contains

DAPI for nuclear counterstaining) and observation under a fluorescent microscope.

**5.1. Characterization of PRLR isoform specific monoclonal antibodies** 

antibodies detect the corresponding protein without cross-reactivity (Figure 1A).

according to the instructions provided by OLINK.

the PRLR isoform specific monoclonal antibodies.

localized more luminally.

**5.2. Immunofluorescent staining of human breast tissue** 

Image acquisition was performed using a Zeiss Axioscope M1 under 40X magnification and an AxioCam HR. Image analysis was performed using the Duolink ImageTool software

We have recently prepared and characterized PRLR isoform specific polyclonal antibodies [5]. However, a limitless source of polyclonal antibody is restricted because it is generated by repeated immunization of a rabbit. Each antibody preparation has the potential to vary in its specificity as well. Monoclonal antibodies are more homogeneous and can produce a near inexhaustible supply of reagent. We used the same peptide sequence as for the preparation of isoform specific polyclonal antibodies and, again, demonstrate specificity for the individual isoforms. CHO cells were transiently transfected with cDNA containing the individual PRLR isoforms; western blot analysis from cell lysates indicate that the

By immunocytochemical analysis using transiently transfected CHO cells, we were able to positively stain for isoform specific PRLR. As shown in Figure 1B, no cross-reactivity was identified using the antibodies on cells transfected with non-corresponding PRLR cDNA. By both western and immunochemical analysis we were able to demonstrate the specificity of

Once we determined the specificity of the isoform specific monoclonal antibodies on transfected cells, we tested them on histosections obtained from reduction mammoplasty specimens. In order to examine the localization patterns of the PRLR, we stained the tissues simultaneously with LF and either SF1a or SF1b antibodies. Similar, striking, distribution of isoform expression was observed, regardless of whether paraffin-embedded or frozen sections were used. As seen in Figure 2, LF expression appeared primarily basal, whereas SF **Figure 1.** Characterization of PRLR isoform antibodies. **A**. Western blot analysis indicating specificity of monoclonal antibodies. CHO cells were transiently transfected with isoform specific PRLR cDNA as described. Cell lysates were prepared and proteins separated by PAGE. Each isoform specific lysate was probed with each isoform specific antibody. Western blot analysis was performed twice with separate lysates. Data shown are a representative autoradiogram. Molecular weights are marked as indicated: LF = 93kDa, SF1a = 55kDa, SF1b = 43kDa. **B**. Fluorescent immunocytochemical analysis. CHO cells were transiently transfected with isoform specific PRLR cDNA as described. Specific staining was observed. The negative control lacks primary antibody. Data shown are representative of triplicate experiments. Magnification 20X.

Prolactin Receptor Isoforms in Human Breast Cancer 203

**Figure 3.** Fluorescent immunohistochemical analysis from breast cancer tissues. Representative photos taken from breast tumor specimens. LF is indicated by green fluorescence; SF is indicated by red

With the exception of the MDA-MB-231 cells, similar patterns of expression were observed in all breast cell lines tested. In the ER positive MCF7 cells, dimers of all six potential combinations were observed in the absence of stimulation, and were maintained throughout the stimulation with PRL (Figure 4A). However, in the ER negative MDA-MB-231 cells, a much more complex pattern appeared (Figure 4B). We observed changes in the levels of several dimers over the course of PRL stimulation, suggesting that the different dimer pairings may be utilized by the cell to mitigate responses to prolonged PRL stimulation.

While performing these experiments, we noticed a significant amount of PRLR dimers at either the perinuclear or nuclear region of the cell. Several receptors in the mammary gland have been shown to undergo nuclear translocation including EGFR, FGFR, and GHR [13]. However, there is little information known about the nuclear localization of PRLR. Several reports in the literature have offered contradicting evidence for PRLR nuclear translocation

In a preliminary study of PRLR isoform nuclear translocation, MDA-MB-231 cells were

fluorescence. LF = long form; SF = short form. Magnification 40X.

[19-21].

stimulated with PRL

**Figure 2.** Fluorescent immunohistochemical analysis from normal breast tissues. Representative photos taken from breast reduction mammoplasty specimens. Images in left panels are taken from paraffin embedded tissues; middle panels are imaged from frozen sections; right panels are taken from paraffin embedded normal lactating mammary tissues. Negative control indicates no primary antibody used. LF is indicated by green fluorescence; SF is indicated by red fluorescence. LF = long form; SF = short form. Magnification 40X.

However, the expression patterns localized quite differently in cancer tissue. The PRLR isoforms appeared to colocalize, but were only present in specific areas (Figure 3). The pattern seemed more random, concentrated in "hot spots," particularly for SF1a.

#### **5.3.** *In situ* **proximity ligation assay monitors PRLR dimerization**

The Duolink reagents allow for specific protein-protein interactions to be examined *in situ*. Using antibodies to PRLR-LF, SF1a and SF1b we generated PRLR isoform-specific PLA probes in order to monitor every possible dimer combination using the OLINK ProbeMaker kit. This kit allows for the direct conjugation of the oligonucleotide (PLUS or MINUS end) to the primary antibody, which is essential for examining homodimers. Using these probes, we examined the six possible dimer pairings (LF/LF, SF1a/SF1a, SF1b/SF1b, LF/SF1a, LF/SF1b, SF1a/SF1b) in six established breast cancer cell lines: T47D, ZR75-1, MDA-MB-231, MDA-MB-468, MCF7, and SK-BR-3 in the absence and presence of PRL stimulation. In addition to the specific antibodies, we included non-specific isotyped control antibodies as negative controls (i.e. LF/NSR, where NSR is a non-specific rabbit IgG in place of a specific rabbit antibody against PRLR). After the proximity ligation reaction has taken place, a single fluorescent red dot is observed. This dot represents a single molecular interaction between the two proteins being interrogated.

Magnification 40X.

the two proteins being interrogated.

**Figure 2.** Fluorescent immunohistochemical analysis from normal breast tissues. Representative photos taken from breast reduction mammoplasty specimens. Images in left panels are taken from paraffin embedded tissues; middle panels are imaged from frozen sections; right panels are taken from paraffin embedded normal lactating mammary tissues. Negative control indicates no primary antibody used. LF is indicated by green fluorescence; SF is indicated by red fluorescence. LF = long form; SF = short form.

However, the expression patterns localized quite differently in cancer tissue. The PRLR isoforms appeared to colocalize, but were only present in specific areas (Figure 3). The

The Duolink reagents allow for specific protein-protein interactions to be examined *in situ*. Using antibodies to PRLR-LF, SF1a and SF1b we generated PRLR isoform-specific PLA probes in order to monitor every possible dimer combination using the OLINK ProbeMaker kit. This kit allows for the direct conjugation of the oligonucleotide (PLUS or MINUS end) to the primary antibody, which is essential for examining homodimers. Using these probes, we examined the six possible dimer pairings (LF/LF, SF1a/SF1a, SF1b/SF1b, LF/SF1a, LF/SF1b, SF1a/SF1b) in six established breast cancer cell lines: T47D, ZR75-1, MDA-MB-231, MDA-MB-468, MCF7, and SK-BR-3 in the absence and presence of PRL stimulation. In addition to the specific antibodies, we included non-specific isotyped control antibodies as negative controls (i.e. LF/NSR, where NSR is a non-specific rabbit IgG in place of a specific rabbit antibody against PRLR). After the proximity ligation reaction has taken place, a single fluorescent red dot is observed. This dot represents a single molecular interaction between

pattern seemed more random, concentrated in "hot spots," particularly for SF1a.

**5.3.** *In situ* **proximity ligation assay monitors PRLR dimerization** 

**Figure 3.** Fluorescent immunohistochemical analysis from breast cancer tissues. Representative photos taken from breast tumor specimens. LF is indicated by green fluorescence; SF is indicated by red fluorescence. LF = long form; SF = short form. Magnification 40X.

With the exception of the MDA-MB-231 cells, similar patterns of expression were observed in all breast cell lines tested. In the ER positive MCF7 cells, dimers of all six potential combinations were observed in the absence of stimulation, and were maintained throughout the stimulation with PRL (Figure 4A). However, in the ER negative MDA-MB-231 cells, a much more complex pattern appeared (Figure 4B). We observed changes in the levels of several dimers over the course of PRL stimulation, suggesting that the different dimer pairings may be utilized by the cell to mitigate responses to prolonged PRL stimulation.

While performing these experiments, we noticed a significant amount of PRLR dimers at either the perinuclear or nuclear region of the cell. Several receptors in the mammary gland have been shown to undergo nuclear translocation including EGFR, FGFR, and GHR [13]. However, there is little information known about the nuclear localization of PRLR. Several reports in the literature have offered contradicting evidence for PRLR nuclear translocation [19-21].

In a preliminary study of PRLR isoform nuclear translocation, MDA-MB-231 cells were stimulated with PRL

Prolactin Receptor Isoforms in Human Breast Cancer 205

6 hr prior to Duolink analysis to examine steady-state nuclear localization of PRLR dimer formation using confocal microscopy (Figure 5). Of the various combinations of PRLR isoform dimers that were examined, nearly all exhibited some nuclear localization. The exception was for SF1a/SF1a homodimers, which formed perinuclear rings but showed little to no nuclear localization (Figure 5B). Homodimers of the two short forms or a heterodimer of the two short forms (Figure 5B, SF1a/SF1a; Figure 5C SF1b/SF1b; Figure 5D, SF1a/SF1b) exhibited significantly higher levels of nuclear localization while LF homodimers (Figure 5A) and the two LF-short form heterodimers (Figure 5E, LF/SF1a; Figure 5F, LF/SF1b) showed modest levels of nuclear localization. These results confirm that nuclear localization exists for the human prolactin receptor isoforms and may suggest a new role for them as potential transcriptional regulators. Pathways indicating the intranuclear transport and function of PRL and its receptor targeting intracellular actions, such as transcription, have been demonstrated [13]. Increases in transcription may effect cell proliferation and/or differentiation. However, the function of perinuclear localization has

**Figure 5.** PRLR isoform dimers exhibit different localization patterns. MDA-MB-231 cells were stimulated for 6 hr with 500 ng/ml recombinant human PRL, fixed and examined by confocal

microscopy for PRLR dimer localization using Duolink. Dimer combinations: A. LF/LF; B. SF1a/SF1a; C. SF1b/SF1b; D. SF1a/SF1b; E. LF/SF1a; F. LF/SF1b. Images were acquired using a 20X objective with a 2X

yet to be explored.

zoom.

**Figure 4.** Duolink analysis of breast cancer cell lines for PRL-dependent PRLR dimerization. Representative images are shown for each dimer pairing in both MCF7 (panel A) and MDA-MB-231 (panel B) cells at 0, 5, 10 and 60 min post-stimulation with PRL. Magnification 40X. Nuclei stained (shown in blue) with DAPI, Duolink signal shown in red. Each red dot represents a single dimerization event.

6 hr prior to Duolink analysis to examine steady-state nuclear localization of PRLR dimer formation using confocal microscopy (Figure 5). Of the various combinations of PRLR isoform dimers that were examined, nearly all exhibited some nuclear localization. The exception was for SF1a/SF1a homodimers, which formed perinuclear rings but showed little to no nuclear localization (Figure 5B). Homodimers of the two short forms or a heterodimer of the two short forms (Figure 5B, SF1a/SF1a; Figure 5C SF1b/SF1b; Figure 5D, SF1a/SF1b) exhibited significantly higher levels of nuclear localization while LF homodimers (Figure 5A) and the two LF-short form heterodimers (Figure 5E, LF/SF1a; Figure 5F, LF/SF1b) showed modest levels of nuclear localization. These results confirm that nuclear localization exists for the human prolactin receptor isoforms and may suggest a new role for them as potential transcriptional regulators. Pathways indicating the intranuclear transport and function of PRL and its receptor targeting intracellular actions, such as transcription, have been demonstrated [13]. Increases in transcription may effect cell proliferation and/or differentiation. However, the function of perinuclear localization has yet to be explored.

204 Prolactin

event.

**Figure 4.** Duolink analysis of breast cancer cell lines for PRL-dependent PRLR dimerization. Representative images are shown for each dimer pairing in both MCF7 (panel A) and MDA-MB-231 (panel B) cells at 0, 5, 10 and 60 min post-stimulation with PRL. Magnification 40X. Nuclei stained (shown in blue) with DAPI, Duolink signal shown in red. Each red dot represents a single dimerization

**Figure 5.** PRLR isoform dimers exhibit different localization patterns. MDA-MB-231 cells were stimulated for 6 hr with 500 ng/ml recombinant human PRL, fixed and examined by confocal microscopy for PRLR dimer localization using Duolink. Dimer combinations: A. LF/LF; B. SF1a/SF1a; C. SF1b/SF1b; D. SF1a/SF1b; E. LF/SF1a; F. LF/SF1b. Images were acquired using a 20X objective with a 2X zoom.

#### **6. Discussion and further research**

Prolactin signaling through the LF of the PRLR has been well studied [13]. Two short forms of the PRLR (SF1a and SF1b) share common extracellular and transmembrane domains with the LF, yet contain different cytoplasmic sequences due to alternative splicing of exon 11 [1,2]. The role of these short forms in PRLR function remains elusive, in part due to a lack of specific reagents. Both SF1a and SF1b contain the Box1 binding site for Jak2 and MAPK proteins, yet lack the Box 2 site (and downstream phosphorylation sites) required for STAT binding [2, 14]. By lacking part of intracellular signaling domain, the short forms of the PRLR act as dominant negatives of the LF in transfected cells and its effects most likely occur through heterodimerization with LF [5]. While both short forms can function as a dominant negative, SF1b is the more potent negative regulator of LF function.

Prolactin Receptor Isoforms in Human Breast Cancer 207

dimer may be formed to limit the inhibitory activity of each monomer towards the LF by a

In addition to confirming the existence of all possible dimer pairings, we were able to show that the different PRLR dimers peaked at different times during the PRL time course. These data suggest that upon stimulation with PRL, regulatory events occur within the cell to favor a particular signaling outcome. For example, an increase in LF/SF1b dimers is predicted to sequester otherwise functional LF monomers into inactive dimers with SF1b. These dimers (and similarly LF/SF1a dimers) would be capable of binding PRL, but would most likely be unable to transmit the signal from ligand binding to the intracellular milieu. Thus, the dominant negative nature of the PRLR short forms could occur by isolating PRLR LF monomers and also result in diminished effects of PRL on cells. Further studies would be required to better understand the dimerization time course for PRLR in these and other

The confirmation of the nuclear localization of PRLR dimers is an important finding. Conflicting reports were found in the literature [13, 19, 20], however our study confirms that PRLR dimers can be found in the nucleus of cultured breast cancer cells. Interestingly, the amount of nuclear localized PRLR was different for the various dimer pairings tested in MDA-MB-231 cells. In general, the SF dimer combinations (SF1a/SF1a, SF1a/SF1b, and SF1b/SF1b) exhibited higher levels of nuclear localization than dimers containing at least one monomer of LF. In addition, only SF1a homodimers localized to a perinuclear region, and appeared as rings around the nucleus. Together these data suggest a potential transcriptional role for PRLR. Recent work performed in the Walker laboratory [21] demonstrated that, in prostate cancer, the expression of SF1b was upregulated after treatment with the PRL inhibitor S179D. This increase in repression also led to an upregulation of p21 and the vitamin D receptor (VDR), both known to affect differentiation and apoptosis. *In vitro* studies also confirmed that long-term overexpression to SF1b decreased the growth and migration of prostate cancer cells, in addition to enhancing cellmatrix interactions and cell-cell aggregation [22]. Abnormalities in the vitamin D endocrine system have been linked to many disorders, including cancer [23]. Strong epidemiological associations were made between vitamin D deficiency and breast and prostate cancers. The VDR system may arrest the tumor cell cycle at G1/G0 through several mechanisms such as by induction of p21 gene transcription [24] or by inducing the synthesis and/or stabilization of p27 [25]. Recent work in tumor-derived endothelial cells has implicated VDR as an antiproliferative factor inducing cell cycle arrest in G1/G0 and tumor angiogenesis [26]. Loss of VDR in this system or in mammary epithelial cells may affect differentiation and apoptosis or modulate intracellular signaling routes. These changes in the tumor microenvironment could potentially result in aberrant angiogenic signaling pathways, possibly even enhancing angiogenesis for a more efficient delivery of chemotherapeutic drugs. PRL and its signaling pathways could be exploited to search for more effective therapies in prostate and breast cancer [27]. Potential synergy with various chemotherapeutics may be another aspect of the

importance of dimerization and localization of PRL and its receptors in breast cancer.

yet-to-be discovered mechanism.

breast cancer cell lines.

To facilitate the study of the PRLR short forms, we generated isoform-specific polyclonal antibodies and demonstrated a breast cancer specific pattern of expression. While all tumor biopsies expressed both LF and SF1b equally, the presence of SF1a was detected mainly in ductal breast carcinoma biopsies and not in those derived from lobular carcinomas [5]. These data suggest diagnostic potential for the PRLR short forms.

The mere expression profiles of these proteins may not serve as sufficient biomarkers, as PRLR dimers are the functional unit for PRL-mediated events. Therefore, an analysis of the ratios of the various potential dimers may be needed to generate a more complete understanding of PRLR signaling. Techniques such as BRET and FRET have been used to examine PRLR dimer formation [14-17]. However, these techniques collectively constrain the user to cells that express tagged constructs and prevent analysis of tissue samples. Coimmunoprecipitation has been used with moderate success for PRLR dimer analysis and avoids the need for tagged constructs, but lacks quantitation, requires specific antibodies, and is devoid of any cellular localization information. To overcome these technical challenges, we employed an *in situ* proximity ligation assay developed by OLINK. This technique allows for the sensitive detection of protein-protein interactions in unmodified cells and tissue. Using the newly generated rabbit and mouse monoclonal PRLR isoformspecific antibodies, we examined PRLR dimer formation in breast cancer cell lines in the absence and presence of PRL stimulation. Due to the flexibility of the method, we were able to monitor all possible homo- and hetero-dimers concurrently.

We were able to confirm that PRLR homo- and hetero-dimers form in the absence of PRL, similar to reports by others [14, 17]. Similar reports have shown that members of the EGFR family can form dimers in the absence of ligand [18]. In total, we observed homodimers of LF/LF, SF1a/SF1a, and SF1b/SF1b and heterodimers of LF/SF1a, LF/SF1b, and SF1a/SF1b in two different, but well-established breast cancer cell lines, MCF7 and MDA-MB-231. While SF homodimerization has been shown, heterodimerization of the short forms has been examined but not proven [16]. However, these assays utilized PRLR receptors C-terminally tagged with GFP or Rluc to perform BRET analysis, which may somehow block the detection of the SF heterodimer. The functional consequence of the SF1a/SF1b heterodimer is not known. Given that both short forms lack the majority of the intracellular domain, this dimer may be formed to limit the inhibitory activity of each monomer towards the LF by a yet-to-be discovered mechanism.

206 Prolactin

**6. Discussion and further research** 

Prolactin signaling through the LF of the PRLR has been well studied [13]. Two short forms of the PRLR (SF1a and SF1b) share common extracellular and transmembrane domains with the LF, yet contain different cytoplasmic sequences due to alternative splicing of exon 11 [1,2]. The role of these short forms in PRLR function remains elusive, in part due to a lack of specific reagents. Both SF1a and SF1b contain the Box1 binding site for Jak2 and MAPK proteins, yet lack the Box 2 site (and downstream phosphorylation sites) required for STAT binding [2, 14]. By lacking part of intracellular signaling domain, the short forms of the PRLR act as dominant negatives of the LF in transfected cells and its effects most likely occur through heterodimerization with LF [5]. While both short forms can function as a

To facilitate the study of the PRLR short forms, we generated isoform-specific polyclonal antibodies and demonstrated a breast cancer specific pattern of expression. While all tumor biopsies expressed both LF and SF1b equally, the presence of SF1a was detected mainly in ductal breast carcinoma biopsies and not in those derived from lobular carcinomas [5].

The mere expression profiles of these proteins may not serve as sufficient biomarkers, as PRLR dimers are the functional unit for PRL-mediated events. Therefore, an analysis of the ratios of the various potential dimers may be needed to generate a more complete understanding of PRLR signaling. Techniques such as BRET and FRET have been used to examine PRLR dimer formation [14-17]. However, these techniques collectively constrain the user to cells that express tagged constructs and prevent analysis of tissue samples. Coimmunoprecipitation has been used with moderate success for PRLR dimer analysis and avoids the need for tagged constructs, but lacks quantitation, requires specific antibodies, and is devoid of any cellular localization information. To overcome these technical challenges, we employed an *in situ* proximity ligation assay developed by OLINK. This technique allows for the sensitive detection of protein-protein interactions in unmodified cells and tissue. Using the newly generated rabbit and mouse monoclonal PRLR isoformspecific antibodies, we examined PRLR dimer formation in breast cancer cell lines in the absence and presence of PRL stimulation. Due to the flexibility of the method, we were able

We were able to confirm that PRLR homo- and hetero-dimers form in the absence of PRL, similar to reports by others [14, 17]. Similar reports have shown that members of the EGFR family can form dimers in the absence of ligand [18]. In total, we observed homodimers of LF/LF, SF1a/SF1a, and SF1b/SF1b and heterodimers of LF/SF1a, LF/SF1b, and SF1a/SF1b in two different, but well-established breast cancer cell lines, MCF7 and MDA-MB-231. While SF homodimerization has been shown, heterodimerization of the short forms has been examined but not proven [16]. However, these assays utilized PRLR receptors C-terminally tagged with GFP or Rluc to perform BRET analysis, which may somehow block the detection of the SF heterodimer. The functional consequence of the SF1a/SF1b heterodimer is not known. Given that both short forms lack the majority of the intracellular domain, this

dominant negative, SF1b is the more potent negative regulator of LF function.

These data suggest diagnostic potential for the PRLR short forms.

to monitor all possible homo- and hetero-dimers concurrently.

In addition to confirming the existence of all possible dimer pairings, we were able to show that the different PRLR dimers peaked at different times during the PRL time course. These data suggest that upon stimulation with PRL, regulatory events occur within the cell to favor a particular signaling outcome. For example, an increase in LF/SF1b dimers is predicted to sequester otherwise functional LF monomers into inactive dimers with SF1b. These dimers (and similarly LF/SF1a dimers) would be capable of binding PRL, but would most likely be unable to transmit the signal from ligand binding to the intracellular milieu. Thus, the dominant negative nature of the PRLR short forms could occur by isolating PRLR LF monomers and also result in diminished effects of PRL on cells. Further studies would be required to better understand the dimerization time course for PRLR in these and other breast cancer cell lines.

The confirmation of the nuclear localization of PRLR dimers is an important finding. Conflicting reports were found in the literature [13, 19, 20], however our study confirms that PRLR dimers can be found in the nucleus of cultured breast cancer cells. Interestingly, the amount of nuclear localized PRLR was different for the various dimer pairings tested in MDA-MB-231 cells. In general, the SF dimer combinations (SF1a/SF1a, SF1a/SF1b, and SF1b/SF1b) exhibited higher levels of nuclear localization than dimers containing at least one monomer of LF. In addition, only SF1a homodimers localized to a perinuclear region, and appeared as rings around the nucleus. Together these data suggest a potential transcriptional role for PRLR. Recent work performed in the Walker laboratory [21] demonstrated that, in prostate cancer, the expression of SF1b was upregulated after treatment with the PRL inhibitor S179D. This increase in repression also led to an upregulation of p21 and the vitamin D receptor (VDR), both known to affect differentiation and apoptosis. *In vitro* studies also confirmed that long-term overexpression to SF1b decreased the growth and migration of prostate cancer cells, in addition to enhancing cellmatrix interactions and cell-cell aggregation [22]. Abnormalities in the vitamin D endocrine system have been linked to many disorders, including cancer [23]. Strong epidemiological associations were made between vitamin D deficiency and breast and prostate cancers. The VDR system may arrest the tumor cell cycle at G1/G0 through several mechanisms such as by induction of p21 gene transcription [24] or by inducing the synthesis and/or stabilization of p27 [25]. Recent work in tumor-derived endothelial cells has implicated VDR as an antiproliferative factor inducing cell cycle arrest in G1/G0 and tumor angiogenesis [26]. Loss of VDR in this system or in mammary epithelial cells may affect differentiation and apoptosis or modulate intracellular signaling routes. These changes in the tumor microenvironment could potentially result in aberrant angiogenic signaling pathways, possibly even enhancing angiogenesis for a more efficient delivery of chemotherapeutic drugs. PRL and its signaling pathways could be exploited to search for more effective therapies in prostate and breast cancer [27]. Potential synergy with various chemotherapeutics may be another aspect of the importance of dimerization and localization of PRL and its receptors in breast cancer.

As discussed herein, our laboratory has recently published the utility of PRLR isoformspecific antibodies in characterizing breast cancer samples as either lobular or ductal in nature [5]. The generation of monoclonal antibodies to the various PRLR isoforms provides a more renewable and reliable supply of these antibodies, and were crucial to the dimerization studies. To our knowledge, this is the first time that the Duolink *in situ* proximity ligation assay enabled detection of all possible PRLR dimer combinations. The technique is robust, quantitative and relatively easy to use.

Prolactin Receptor Isoforms in Human Breast Cancer 209

mammary glands from ovariectomized female mice lacking both E and P, neither P nor PRL stimulate DNA synthesis in epithelial or adjacent stromal cells. However, simultaneous injection of P and PRL results in a significant synergistic effect on DNA synthesis in both cell types [31]. Similarly, disruption of expression and distribution of PR in the mammary gland results in a parallel disruption of expression and distribution of the PRLR suggesting interregulation of these receptors [32]. Expression of the PRLR in human cells is regulated through a complex system of promoters. The PRLR gene in humans has six exon 1s, the most generic of which (PRLR promoter region III) is functional in human breast cancer cells. T47D human breast cancer cells treated with P overexpress the PRLR and activate PRLR promoter III. This promoter lacks a classical P response element. Thus P acts through the cooperative interaction of the PR with the transcription factors C/EBP and an adjacent Sp1A

site to confer P responsiveness leading to increased expression of the PRLR [33].

are now available.

**7. Conclusion** 

**Author details** 

**Acknowledgement** 

**8. References** 

effective targets for prevention and treatment.

These examples of the complex interactions of E, P, and PRL underscore the importance of understanding all three hormones' actions and their receptor interactions, distribution, and functions in the normal human breast and in breast cancer. Effective tools for these studies

Just as the identification of the ER and PR are now routinely performed on all clinical breast cancer samples, screening for PRLR status and isoform profile may also one day become a routine procedure in the characterization of breast cancers. Isoform specific antibodies to the PRLR can be powerful tools in detection and characterization of breast and other cancers and provide valuable insights into the important PRLR signaling pathways that could be

Erika Ginsburg, Christopher D. Heger, Paul Goldsmith and Barbara K. Vonderhaar *Center for Cancer Research, National Cancer Institute, National Institutes of Health, USA* 

This research was supported by the Intramural Research Program of the NIH, National Cancer Institute. We thank Sarah J. Tarplin for her assistance with the immunostaining and to Dr. Jodie M. Fleming for helpful discussions. The authors have no competing interests.

[1] Trott JF, Hovey RC, Koduri S, Vonderhaar BK (2003) Alternative splicing to exon 11 of human prolactin receptor gene results in multiple isoforms including a secreted

prolactin-binding protein. *J Mol Endocrinol* 30(1):31-47.

With the advent of the tissue microarray (TMA) [28, 29], researchers are able to examine large populations of tissues from patients on a single glass slide. There are many commercially available TMAs as well, including arrays for breast cancer patients. These arrays have typically been evaluated for ER/PR/HER2 status and, in many cases, have both normal and diseased tissue for each patient. Given that OLINK provides a Brightfield Duolink kit, several studies could be envisioned. First, since the Duolink assay can also be used to obtain quantitative total expression of a target, the application of this technique to TMAs would provide a significantly better means of analyzing the expression levels of targets (rather than high, medium, and low expression). These data could also be combined with expression profiles of the PRLR dimer complexes and ultimately compared with other markers to identify more specific breast cancer signatures. Further correlations could be made with the levels of the hormone PRL itself, which could help further diagnose patients.

The "hormonal responsiveness" of breast cancer has been known for a long time. Unfortunately, this phrase has only been used to indicate responsiveness to estrogen (E) and progesterone (P). However, several other hormones including PRL, growth hormone, and thyroid hormone have been shown to play roles in normal breast development and function and their various roles in breast cancer are being investigated. The complex interplay of the three major hormones involved in normal breast development, E, P, and PRL, is well documented. In model systems that include human breast cancer cells in culture as well as rodent models, E, P, and PRL have all been shown to stimulate growth. In fact, when compared directly with E acting on MCF-7 human breast cancer cells, PRL is a more potent mitogen than E [8].

Additionally, these model systems have demonstrated that each of these hormones acts as more than just a mitogen. Complex systems of inter-regulation exist between these steroids, PRL and their receptors. PRL stimulates growth through both an endocrine as well as an autocrine/paracrine mechanism. The expression of autocrine PRL in human breast cancer cells is regulated by E through action on the PRL promoter [30]. Expression of the PRL gene is regulated by two promoters, the proximal promoter utilized primarily in the pituitary and the distal promoter utilized in breast cancer cells. E directly induces hPRL gene expression in T47D human breast cancer cells through the action of a functional non-canonical E response element (ERE) and an AP1 site on the distal promoter. Both the ERE and AP1 sites are required for full induction of promoter activity through the alpha form of the ER.

Similarly, the interplay of P and PRL is well established in various models. PR and PRLR are co-expressed temporally and spatially in the developing mouse mammary gland. In mammary glands from ovariectomized female mice lacking both E and P, neither P nor PRL stimulate DNA synthesis in epithelial or adjacent stromal cells. However, simultaneous injection of P and PRL results in a significant synergistic effect on DNA synthesis in both cell types [31]. Similarly, disruption of expression and distribution of PR in the mammary gland results in a parallel disruption of expression and distribution of the PRLR suggesting interregulation of these receptors [32]. Expression of the PRLR in human cells is regulated through a complex system of promoters. The PRLR gene in humans has six exon 1s, the most generic of which (PRLR promoter region III) is functional in human breast cancer cells. T47D human breast cancer cells treated with P overexpress the PRLR and activate PRLR promoter III. This promoter lacks a classical P response element. Thus P acts through the cooperative interaction of the PR with the transcription factors C/EBP and an adjacent Sp1A site to confer P responsiveness leading to increased expression of the PRLR [33].

These examples of the complex interactions of E, P, and PRL underscore the importance of understanding all three hormones' actions and their receptor interactions, distribution, and functions in the normal human breast and in breast cancer. Effective tools for these studies are now available.

#### **7. Conclusion**

208 Prolactin

mitogen than E [8].

As discussed herein, our laboratory has recently published the utility of PRLR isoformspecific antibodies in characterizing breast cancer samples as either lobular or ductal in nature [5]. The generation of monoclonal antibodies to the various PRLR isoforms provides a more renewable and reliable supply of these antibodies, and were crucial to the dimerization studies. To our knowledge, this is the first time that the Duolink *in situ* proximity ligation assay enabled detection of all possible PRLR dimer combinations. The

With the advent of the tissue microarray (TMA) [28, 29], researchers are able to examine large populations of tissues from patients on a single glass slide. There are many commercially available TMAs as well, including arrays for breast cancer patients. These arrays have typically been evaluated for ER/PR/HER2 status and, in many cases, have both normal and diseased tissue for each patient. Given that OLINK provides a Brightfield Duolink kit, several studies could be envisioned. First, since the Duolink assay can also be used to obtain quantitative total expression of a target, the application of this technique to TMAs would provide a significantly better means of analyzing the expression levels of targets (rather than high, medium, and low expression). These data could also be combined with expression profiles of the PRLR dimer complexes and ultimately compared with other markers to identify more specific breast cancer signatures. Further correlations could be made with the levels of the hormone PRL itself, which could help further diagnose patients. The "hormonal responsiveness" of breast cancer has been known for a long time. Unfortunately, this phrase has only been used to indicate responsiveness to estrogen (E) and progesterone (P). However, several other hormones including PRL, growth hormone, and thyroid hormone have been shown to play roles in normal breast development and function and their various roles in breast cancer are being investigated. The complex interplay of the three major hormones involved in normal breast development, E, P, and PRL, is well documented. In model systems that include human breast cancer cells in culture as well as rodent models, E, P, and PRL have all been shown to stimulate growth. In fact, when compared directly with E acting on MCF-7 human breast cancer cells, PRL is a more potent

Additionally, these model systems have demonstrated that each of these hormones acts as more than just a mitogen. Complex systems of inter-regulation exist between these steroids, PRL and their receptors. PRL stimulates growth through both an endocrine as well as an autocrine/paracrine mechanism. The expression of autocrine PRL in human breast cancer cells is regulated by E through action on the PRL promoter [30]. Expression of the PRL gene is regulated by two promoters, the proximal promoter utilized primarily in the pituitary and the distal promoter utilized in breast cancer cells. E directly induces hPRL gene expression in T47D human breast cancer cells through the action of a functional non-canonical E response element (ERE) and an AP1 site on the distal promoter. Both the ERE and AP1 sites

are required for full induction of promoter activity through the alpha form of the ER.

Similarly, the interplay of P and PRL is well established in various models. PR and PRLR are co-expressed temporally and spatially in the developing mouse mammary gland. In

technique is robust, quantitative and relatively easy to use.

Just as the identification of the ER and PR are now routinely performed on all clinical breast cancer samples, screening for PRLR status and isoform profile may also one day become a routine procedure in the characterization of breast cancers. Isoform specific antibodies to the PRLR can be powerful tools in detection and characterization of breast and other cancers and provide valuable insights into the important PRLR signaling pathways that could be effective targets for prevention and treatment.
