|Home | About | Journals | Submit | Contact Us | Français|
Parathyroid hormone-related protein (PTHrP) is the causative factor of the paraneoplastic syndrome humoral hypercalcemia of malignancy (HHM) and it also contributes to osteolytic metastases, both of which are common complications of squamous carcinomas of the lung. Inhibition of autocrine EGFR signaling has been shown to reduce plasma calcium and PTHrP concentrations in two lung squamous cell carcinoma xenograft models of HHM. The purpose of this study was to investigate the mechanism by which EGFR is activated and stimulates PTHrP gene expression in lung squamous carcinoma cell lines (SCC). Amphiregulin (AREG) was the only EGFR-ligand that could be consistently detected in conditioned media from the SCC lines and reduction of its expression either by siRNA or by precipitating antibodies reduced PTHrP mRNA expression as effectively as EGFR targeted inhibition. Using siRNA knockdown or inhibitors to upstream regulators of AREG shedding including TACE, Src/Lck and Gi/o, also reduced PTHrP mRNA expression. We determined that blockade of autocrine AREG-EGFR signaling does not affect PTHrP mRNA stability. Of the three PTHrP promoters (P1, P2, P3), P1 mRNA could be reduced by nearly 100% with an EGFR inhibitor, and both EGF and AREG stimulated P1 mRNA by ~5-fold. Finally, ectopic expression of EGFR in a receptor-low but AREG expressing cell line increased PTHrP mRNA levels in vitro, and induced the capability to cause HHM and rapid osteolytic growth in vivo. Taken together, we provide evidence that AREG stimulation of EGFR results in high levels of PTHrP gene expression, contributing to cancer-associated bone pathology.
Lung cancer is the leading cancer-related cause of death in the United States, with non-small cell lung cancers (NSCLC) accounting for 90% of all lung derived tumors (1). Among NSCLCs, squamous cell carcinomas (SCCs) are most frequently associated with cancer-mediated diseases of bone such as the paraneoplastic syndrome humoral hypercalcemia of malignancy (HHM) and bone metastasis (2, 3). The causative factor of HHM is parathyroid hormone-related protein (PTHrP), which was first purified from a human lung SCC (3). Detectable or increased plasma concentrations of PTHrP have been found in over 80% of hypercalcemic patients with solid tumors, although the precise molecular mechanisms that activate PTHrP gene expression in these tumors are unknown (4). SCCs of the lung and other sites produce very high levels of PTHrP in part due to the transcriptional activity of the P1 promoter, which is not activated in most other cancer cell types that express PTHrP (5, 6). Lung SCC cells trigger lytic bone metastases which is speculated to be driven in part by production of PTHrP by the tumor cells (2). Today, lung cancer associated HHM has become less common in the United States due to earlier detection and treatment of primary tumors. However, NSCLC bone metastasis and associated hypercalcemia remain a common clinical problem, and better treatment options for this form of disease are needed.
There is emerging evidence that autocrine epidermal growth factor receptor (EGFR) signaling regulates PTHrP gene expression in a variety of cell types (7-9). In a series of experiments using primary normal human keratinocytes, Cho et al. reported that the addition of precipitating antibodies to amphiregulin (AREG) and EGFR tyrosine kinase inhibitor PD153035 (PD) reduced PTHrP mRNA levels up to 80% (7). Treatment of EGFR-positive breast cancer cell lines with either PD or AREG-neutralizing antibodies reduced PTHrP mRNA levels by 60 to 70% (8). Finally, inhibition of autocrine EGFR signaling has been shown to reduce plasma PTHrP and total calcium concentrations in two lung SCC xenograft models of hypercalcemia (9).
The epidermal growth factor or ErbB family of receptor tyrosine kinases consists of four unique receptors: EGFR, ErbB2 (HER2/neu), ErbB3, and ErbB4 (10). The EGFR is overexpressed in the majority of lung SCCs, while ErbB2 and ErbB3 protein levels are elevated less frequently (9, 11, 12). Lung SCCs also produce EGFR ligands, mainly AREG and transforming growth factor-α (TGFα) (9, 13). EGFR ligands are synthesized as membrane-bound precursors that require proteolytic cleavage for release of soluble ligand. Once cleaved, EGFR ligands are capable of binding and activating the EGFR in an autocrine/paracrine manner. The tumor-necrosis factor α-converting enzyme (TACE/ADAM17) has been implicated as the matrix metalloprotease that cleaves EGFR ligands, including AREG, in several human cancers (14-16). TACE has been proposed to be the major sheddase that cleaves AREG in non-small cell lung cancer (15), but human airway trypsin-like protease, ADAM 10 and other MMPs have also been reported to release the ligand (17-19). Numerous cell types that lack TACE, but produce other ligands have been shown to be defective in EGFR signaling (20, 21). Moreover, it has been reported that TACE plays a critical role in the cellular cross-talk (transactivation) between G protein-coupled receptors (GPCRs) and the EGFR signaling pathway (15, 22). In SCCs of the head and neck, GPCR activation of TACE and subsequent shedding of AREG appears to be mediated by Src family members (15, 22).
The purpose of this study was to investigate the mechanism of autocrine EGFR stimulation of PTHrP gene expression in HHM-inducing lung SCCs. In this manuscript we report that PTHrP gene expression in three SCC lines is controlled by TACE, AREG and EGFR signaling. AREG-dependent PTHrP gene expression can be blocked by pertussis toxin, but this signaling does not appear to be cSrc dependent. In these lines, AREG-EGFR signaling dramatically impacts transcripts from the P1 promoter,but also influences transcript production from the other promoters. Furthermore, the reconstitution of the TACE-AREG-EGFR signaling system in a lung SCC line is sufficient to produce HHM and rapid osteolytic growth in animal models.
Initially we wanted to compare ErbB content and PTHrP mRNA levels in the low EGFR expressing lung SCC HTB-182/NCI H520 line (23), to other lines (RWGT2 and HARA) that induce hypercalcemia in animals (9, 24, 25). Strong immunoreactive bands for EGFR were present in the RWGT2 and HARA samples, but not in the HTB-182 extracts (Fig. 1A). Minimal ErbB2, ErbB3, and ErbB4 immunoreactivity was also detected in the SCC lines. The MCF-7 breast cancer line was used as both a negative control for EGFR expression and a positive control for ErbB2, ErbB3, and ErbB4 expression. Thus, high levels of EGFR are expressed in the hypercalcemia-inducing lung SCC lines, but other ErbB family members are minimally expressed, even in the HTB-182 cell line.
We then evaluated the expression of PTHrP mRNA levels in the lung SCC lines, RWGT2, HARA, and HTB-182 cells by quantitative real-time PCR (QRT-PCR) (Fig. 1B). RWGT2 cells express very high levels of PTHrP mRNA transcripts. HARA cells also produce substantial PTHrP mRNA, ranging between 25 to 50% that of RWGT2 cells (Fig. 1B). In comparison, HTB-182 cells expressed less than one-tenth as much PTHrP mRNA as RWGT2 cells; however, even this is 10 to 100-fold more than most human breast cancer cell lines (8).
We wanted to determine if the low EGFR expressing HTB-182 SCC line could produce hypercalcemia in a mouse xenograft model. Athymic nude mice injected with 1×106 HARA or RWGT2 became hypercalcemic (defined as total calcium levels ≥ 12 mg/dl) within 60 days, as previously reported (9) (Fig. 1C). The HTB-182 cell line was unable to induce hypercalcemia even when tumors were grown to a diameter of >2 cm, which took longer than 60 days (Fig. 1C). We also characterized ErbB and PTHrP expression in the tumors. As expected, HARA and RWGT2 tumor lysates had robust EGFR expression (Fig. 1D), as well as low levels of ErbB2 (not shown), but ErbB3 and ErbB4 could not be detected in these unprecipitated lysates. In contrast, the HTB-182 tumor lysates lacked detectable levels of ErbB proteins using this methodology. Also, the relative PTHrP/GAPDH mRNA ratios among the lines remained similar to those observed in vitro (Fig. 1D). Thus, growth in vivo did not alter relative EGFR or PTHrP mRNA expression for the three SCC lines, and only those lines that expressed high levels of the receptor had the capacity to induce hypercalcemia in vivo.
We then wanted to determine the EGF-like ligand expression of the RWGT2, HARA and HTB-182 cell lines. We used sandwich ELISAs (R&D Systems, Minneapolis, MN) to determine the relative concentrations of AREG, Betacellulin (BTC), epidermal growth factor (EGF), heparin-binding epidermal growth factor (HB-EGF) and TGFα in the medium for each cell line (kits for epiregulin and epigen are not available). Consistent with previous mRNA expression measures (9) the concentration of AREG secreted into the medium by our panel of SCC lines was approximately one to two orders of magnitude greater than HB-EGF, the only other ligand that was consistently detectable in the RWGT2 line (Fig. 2 A). The shed protein levels for BTC, EGF and TGFα in the media did not consistently exceed the minimum limit of detection for these ELISAs (not shown). Furthermore, media harvested from the RWGT2 cell line had AREG protein levels that were consistently 10 times greater than the HARA and HTB-182 cell lines. These findings suggest that AREG would likely drive autocrine EGFR signaling in the two hypercalcemia-inducing SCC lines and the high levels of this ligand may contribute to greater PTHrP levels in the RWGT2 line.
To confirm the specificity of the impact of various inhibitors on PTHrP mRNA expression, we used siRNA to knockdown EGFR and AREG in the HARA line. We confirmed knockdown of the EGFR protein by western blotting with an EGFR antibody, which indicated EGFR immunoreactivity was decreased by ~50% (Fig. 2B). QRT-PCR analysis revealed that PTHrP mRNA levels were reduced by ~50% following transfection with an EGFR siRNA construct, compared to a scrambled control (Fig. 2B). To determine the role of AREG signaling in PTHrP gene expression, we used siRNA to specifically knockdown AREG levels in the HARA cells. siRNA AREG knockdown resulted in a ~50% decrease in AREG concentrations in the conditioned media and resulted in a ~30% decrease in PTHrP mRNA expression (Fig. 2D). We then treated the cells with an AREG neutralizing antibody (10 μg/ml) and blunted PTHrP mRNA expression levels by ~75% and ~50% in the RWGT2 and HARA cell lines, respectively (Fig. 2E). These results suggest that AREG is the primary ligand activating EGFR-induced PTHrP gene expression; however, other mechanisms contribute to control of the PTHrP gene in these lines.
Next, we evaluated TACE protein levels in the SCC lines. The MDA-MB-231 breast cancer cell line was used as a positive control for TACE expression. We detected a robust immunoreactive TACE band in each of the lung SCC lines that was most intense in the RWGT2 and HTB-182 cell lines (Fig. 3A). Treatment of the hypercalcemia-inducing RWGT2 and HARA cell lines with 10 μM of a broad spectrum MMP inhibitor, GM6001, that also targets TACE resulted in an approximate 50% reduction of AREG shed into the media by RWGT2 cells (Fig. 3D). A similar trend was observed in the HARA cell line, although this result was not statistically significant. We also measured PTHrP mRNA expression by QRT-PCR analysis of the treated the RWGT2 and HARA cell line with GM6001. This inhibitor reduced PTHrP mRNA expression by ~30% in RWGT2 and 50% in HARA (Fig. 3E). We treated HARA cells with siRNA to TACE and found a ~45% reduction in TACE protein expression (Fig. 3B). We also determined that the expression of shed AREG was decreased by ~60% and the PTHrP mRNA reduced ~30% (Fig. 3C) after siRNA transfection. Taken together, the inhibition of TACE reduces AREG levels in the media and PTHrP mRNA levels in the cell lines (Figs. 3C-E).
We wanted to determine if previously identified inhibitors of upstream regulators of TACE could also influence PTHrP gene expression. It has previously been shown that stimulation of G-protein coupled receptors (GPCR) can activate the TACE and this could be blocked with pertussis toxin (15). On that basis, we treated the cells with the Gi/o protein inhibitor pertussis toxin (100ng/mL). We observed a complete loss of AREG shedding in the RWGT2 (black bars) and ~60% decrease in the HARA cells (Fig. 4A, white bars). Consistent with a blockade in EGFR transactivation, pertussis toxin was able to decrease the expression of PTHrP mRNA by 50% in RWGT2 ; however this repression was more pronounced (~80%) in HARA cells (Fig. 4B)
Previous reports implicated c-Src as a major intermediate between GPCR signaling and TACE-mediated AREG shedding in SCC lines derived from head and neck cancer (26). We used an inhibitor of Src and Lck kinases (Src-1,10 μM, 24 hours) to treat the RWGT2 and HARA lines. This inhibitor failed to reduce the release of AREG protein into the media or decrease PTHrP mRNA levels in the RWGT2 cell line (Fig. 4A&B black bars), but moderately decreased (~20%) the AREG levels in HARA line. In contrast the Src-1 inhibitor greatly reduced (~90%) PTHrP mRNA expression in the HARA cell line. (Fig. 4B white bars). Thus, a Src /Lck inhibitor could not inhibit the majority of AREG shedding in these lung SCC lines, but yet efficiently inhibited PTHrP gene expression in the HARA line.
Next, we examined how endogenous EGFR signaling regulated PTHrP gene expression. In order to investigate mRNA stability, RWGT2 and HARA cells were pretreated with 5 μg/ml of the transcription inhibitor actinomycin D, in the presence or absence of 1 μM of the EGFR tyrosine kinase inhibitor PD153035 (PD). At various time points after actinomycin D treatment, no significant difference in PTHrP mRNA levels was observed as compared to the control groups, suggesting that blockade of EGFR signaling did not alter PTHrP mRNA stability (Fig. 5A&B). We used QRT-PCR analysis to investigate PTHrP promoter-specific transcripts in the RWGT2 and HARA cell lines based on primers developed by Richard et al. (27). Of the three PTHrP promoters (P1, P2, P3), P1 mRNA was reduced below detection levels with the EGFR tyrosine kinase inhibitor PD153035 (PD), but the P1/P2 and P3 specific primers detected less substantial decreases in transcript levels (Fig. 5D&E). Treatment with exogenous EGF (17 nM) and AREG (93 nM) stimulated P1 mRNA by ~5-fold in the RWGT2 (black bars) and HARA (white bars) cell lines, but also stimulated the P1/P2 and P3 transcripts (Fig. 5B). Thus, in these lung SCC lines, endogenous AREG appears to regulate PTHrP gene expression though transcripts derived from all three promoters (Fig. 5).
The observation that the HTB-182 SCC line does not express high levels of the EGFR made this cell line an attractive model for ectopic receptor expression studies. Therefore, we stably transduced the HTB-182 cell line with a recombinant retrovirus bearing an EGFR expression vector (HTB-182/EGFR cells) or an empty vector (HTB-182/LXSN), and isolated a G418 resistant population of cells. Western blotting confirmed the presence of EGFR in the HTB-182/EGFR cells (Fig. 6A). We determined that AREG was present in the media at 50 to 70 pg/ml in the HTB-182/EGFR or HTB-182/LXSN, whereas BTC, EGF, HB-EGF, were below the limit of detection for these commercial ELISAs; however, we did consistently detect TGFα in the HTB-182/LXSN line (Fig. 6B). Next, we measured PTHrP mRNA expression of the HTB-182/EGFR cells and observed an increase in PTHrP mRNA relative to vector control cells (Fig. 6C). Moreover, the addition of exogenous AREG (93 nM) or EGF (17 nM) to HTB-182/EGFR cells stimulated PTHrP mRNA expression an additional 4- to 5-fold over the control cells, respectively (Fig. 6C). We also treated the HTB-182/EGFR or HTB-182/LXSN cell lines with the EGFR kinase inhibitor PD153035 (PD, 1 μM), AREG antibody (αAREG, 10 μg/mL), pertussis toxin (PT, 100 ng/mL), Src kinase inhibitor 1 (Src, 10 μM), and TACE inhibitor (GM, 10 μM) and measured PTHrP mRNA expression (Fig. 6D). All of the inhibitors that repressed PTHrP mRNA levels in the HARA cell line also significantly reduced transcript levels in the HTB-182/EGFR cell line, but failed to have an impact on the HTB-182/LXSN cells. Exogenous AREG stimulation of the HTB-182/EGFR cell line increased P1and P3-derived PTHrP transcripts 6-fold; however, exogenous EGF was less effective in increasing P1 and P3 transcripts than in the RWGT2 or HARA lines (Fig. 6E-G). Thus, ectopic expression of high levels of the EGFR increased PTHrP gene expression in the HTB-182 line and sensitized this expression to inhibitors that regulate AREG shedding and EGFR signaling.
To determine if the reconstitution of AREG-EGFR signaling in the HTB-182 cell line could induce hypercalcemia, athymic nude mice were subcutaneously injected with 1×106 HTB-182, HTB-182/LXSN, and HTB-182/EGFR cells, and serum calcium levels were monitored. We determined that the HTB-182/EGFR tumors maintained receptor expression in vivo and contained higher levels of PTHrP than the HTB-182/LXSN control (Fig. 7A&B). As shown in Figure 7C, HTB-182/EGFR tumor-bearing mice became hypercalcemic, whereas parental HTB-182 and the HTB-LXSN mice maintained serum calcium levels similar to non-tumor bearing controls (Fig. 7C). Thus, HTB-182/EGFR cells expressed higher levels of PTHrP mRNA than parental or vector bearing cells, and acquired the ability to produce hypercalcemia in nude mice.
Since PTHrP has been established to drive osteolytic growth of cancers within the bone (28, 29), we compared the growth of HTB-182/EGFR and HTB-LXSN cells injected the metaphysis of the tibia of athymic nude mice. One to two-weeks after injection of 2×104 cells into the tibia, we observed small X-ray lucent regions in both the HTB-182/EGFR and HTB-182/LXSN injected bones (Fig. 8A). However, 3-weeks after injection the X-ray detectable lesions in the HTB-182/EGFRinjected tibias took on the appearance of holes in the bone, and a significant difference in lesion area as compared to HTB-182/LXSN was evident at 4-weeks (Fig. 8B). At this time, the tumor bearing and non-injected tibias were removed, fixed, embedded, and sectioned. Bone sections were stained using hematoxylin and eosin, tartrate resistant acid phosphatase (TRAP) histochemistry and EGFR immunohistochemistry. As shown in Figure 8F, in regions where the HTB-182/EGFR tumor cells occupied the morrow cavity cortical bone was eroded and tumor and other cells were often present outside of the bone. In contrast, HTB-182/LXSN tumors tended to fill the marrow cavity with little impact on the cortical bone (Fig. 8E). Histomorphometry indicated that total area occupied by tumor cells tended to be larger in the HTB-182/LXSN as compared to the HTB-182/EGFR-bearing legs, but this difference was not significant (Fig. 8D). EGFR antibodies stained cells in the mouse bone marrow, and also intensely labeled the cell periphery in the HTB-182/EGFR tumor cells, whereas this labeling was not present in the HTB-182/LXSN tumor cells, confirming continued ectopic expression of the receptor in the bone microenvironment (Fig. 8H). A 2.8-fold increase in osteoclasts/bone surface area was observed in the TRAP stained HTB-182/EGFR-injected legs as compared to the uninjected legs from mice bearing HTB-182/LXSN cells (Fig 8C). This increase in osteoclasts was observed in the growth plate and periosteum, as well as in the diaphysis both within and around the tumor (Fig. 8G). In contrast, HTB-182/LXSN-bearing legs had no increase in these bone resorbing cells as compared to non-tumor cell-injected legs (Fig. 8C). The number of osteoclasts in the non-injected and injected legs of the HTB-182/LXSN-bearing mice and their distribution (the growth plate and perosteum-perichondrium junction) was typical of mice at 7 to 8-weeks of age. Taken together, these finding suggest reconstitution of AREG-EGFR signaling leads to aggressive osteolytic growth by the HTB-182 lung SCC line.
GPCR transactivation-induced TACE mediated cleavage of ligands appears to be a common means of activating the EGFR in both physiological and pathological circumstances (14, 30). In this manuscript, we present new data that suggests a pertussis toxin-sensitive transactivation of the EGFR is a major regulator of PTHrP gene expression in human lung SCC lines that induce osteolytic lesions and humoral hypercalcemia in mice. Such a transactivation mechanism has been reported to mediate calcium receptor activation of PTHrP in HEK 293 cells, the PC-3 human prostate cancer cell line and the hypercalcemia-inducing rat Leydig cancer cell line H-500 (31-33). However, in these cells the ligand involved appeared to be HB-EGF (31), whereas this study indicates AREG to be the transactivating agonist (Figs. (Figs.2A2A and and6C).6C). These SCC lines expressed high levels of TACE, and transfection with a siRNA to TACE, as well as GM6001, decreased the shedding of AREG and the expression of PTHrP transcripts in HARA and RWGT2 cell lines. The magnitude of reduction of PTHrP mRNA transcripts observed with TACE inhibition suggests that this MMP is responsible for at least a portion of the EGFR transactivation observed in these lines.
Investigations of the upstream control of TACE, failed to pin down specific molecular targets in these lung SCC lines. Src was found to exclusively mediate activation of TACE in head and neck SCC line PCI-37A (22), but another family member (Yes) has been implicated in this role in other lines (26). The c-Src/Lck inhibitor modestly decreased the release of AREG in the HARA, and had no impact in the RWGT2 line, raising the possibility that other signaling events may be involved in TACE activation in these lung SCC lines. In contrast, the SRC-1 inhibitor efficiently repressed PTHrP mRNA levels ~90% in both HARA and the HTB-182/EGFR cell lines. These findings coupled with the enhanced PTHrP mRNA repression with pertusis toxin suggest additional roles for G-proteins and Src/Lck kinases beyond EGFR transactivation in control of PTHrP gene expression in the HARA and HTB-182/EGFR cell lines that are not apparent in the RWGT2 line (Figs. (Figs.44 and and6D6D).
Very little is known about the G-protein mediators that transduce GPCR signals to TACE. Pertussis toxin has been reported to block EGFR transactivation in other cancer cell lines (15), and we found the Gi/o inhibitor efficiently reduced the cleavage of AREG and the PTHrP gene expression (Figs. (Figs.44 and and6D)6D) in all lines tested. We did not investigate the role of specific GPCRs found in NSCLC such as CXCR family or frizzled receptors (34-36) in the control of PTHrP gene expression. A variety of autocrine or paracrine/endocrine GPCR agonists could redundantly stimulate EGFR transactivation in lung SCC. The well established role of the calcium-sensing receptor in EGFR transactivation (31, 32) and reported high levels of expression of this receptor in NSCLC (34) raises the intriguing possibility that humoral hypercalcemia induced by some lung SCCs could be driven by a vicious cycle of PTHrP production being stimulated by increased calcium levels of the microenvironment or the circulation.
The production of high levels of PTHrP by hypercalcemia-inducing SCCs has been associated with activated PTHrP gene transcription (6, 37). Human PTHrP expression is regulated by three promoters: P1, P2 (GC rich) and P3 (TATA) (38). Most human cancers utilize the P3 promoter to upregulate PTHrP (27, 39); however, tumors of squamous cell origin uniquely express high levels of transcripts from the P1 promoter (27, 39). We had previously found that AREG-EGFR signaling regulated PTHrP gene expression via the MAP kinase pathway and the P3 promoter in primary human keratinocytes grown in serum free media, as well as breast cancer cell lines (7, 8). As expected, the lung SCC lines used in this study expressed P1-derived transcripts (Fig. 5), and it appears endogenous and exogenous EGFR signaling greatly impacts transcript levels derived from this promoter, as well as to a lesser extent the other promoters. Blockade of endogenous AREG-EGFR signaling appeared not to impact overall PTHrP mRNA stability, as measured by our coding region primers that detect sequences found in all PTHrP transcripts (Fig. 5A). A previous study using a human keratinocyte cell line grown in serum containing media (similar to those used in this study) reported that exogenous EGF treatment impacted PTHrP increased transcription from all 3 promoters and enhanced PTHrP mRNA stability (39). We speculate that the inconsistencies between the literature and our findings stem from comparing EGFR stimulation triggered by exogenous EGF to a blockade of endogenous AREG-EGFR signaling. Emerging evidence from our lab and several others indicates AREG stimulates a limited EGFR tyrosine phosphorylation, as compared to ligands such as TGFα or EGF, and this is likely activate a subset of downstream signaling pathways (Gilmore and Foley unpublished data) (8, 40-43). Taken together, we speculate that the generation of HHM by SCCs requires high levels of PTHrP gene expression driven by the P1 promoter, which can be efficiently activated by the AREG-EGFR signaling.
Although lagging behind breast and prostate cancer, there is an expanding understanding of molecular signaling pathways that direct the growth of lung cancer cells in the bone microenvironment. As a whole, NSCLCs are reported to produce aggressive osteolytic lesions in human patients (1). The production of bone morphogenetic proteins (BMPs), receptor for nuclear factor kappa ligand (RANKL) and possibly PTHrP appears to underlie the growth of mixed osteolytic and osteoblastic lesions by the lung adenocarcinoma cell line, A549 (44). A large cell lung cancer subline of line NCI-H460 produces aggressive osteolytic lesions after intracardiac injection due to TGFβ responsiveness and upregulation of global MMP activity (45). One of the few studies to analyze factors associated with lung cancer progression in matched sets of tumor and bone samples found that the expression of the EGFR observed in primary NSCLC tumors was maintained in subsequent bone metastases (46), suggesting that the signaling by the receptor could be active in the bone microenvironment. We have found that ectopic expression receptor induced rapidly growing X-ray detectable lesions and increased numbers of osteoclasts implying that AREG-EGFR signaling contributed to osteolytic growth in the bone (Fig. 8). We speculate that EGFR activation of PTHrP gene expression observed in the HTB-182 line lead to the osteolytic growth. In addition, EGFR signaling has been reported to increase the expression of other cytokines/growth factors such as IL-1, Il-8, M-CSF and MMP-2 associated with osteolytic growth in both cancer and ostelblast like cells (40, 47-49). In addition, PTH receptor signaling on cells of the osteoblasts lineage activates AREG gene expression which could in turn further activate PTHrP gene expression in lung cancer cells within the bone microenvironment (50). As is emerging from the breast cancer-bone metastasis field (51), it is likely there are redundant growth factor pathways by which lung cancer cells manipulate the bone microenvironment to induce osteolysis. Thus, broadly targeting AREG-EGFR signaling may provide a means to control the expression of many factors that facilitate the growth of lung NSCLC metastases within the bone.
RWGT2 cells were obtained from Dr. Theresa Guise (University of Texas Health Center, San Antonio, TX) and maintained in minimum essential medium alpha medium (MEMα), supplemented with 10% fetal bovine serum, 1% L-glutamine, and 1% penn/strep. The HARA cell line was obtained from Dr. Haruo Iguchi (Shikoku Cancer Center, Matsuyama, Japan) and maintained with RPMI-1640 medium, and supplemented as described above. The HTB-182 (NCI H520) cell line was purchased from ATCC (Manassas,VA) and maintained as described for the HARA cell line. In vitro experiments were performed when the RWGT2, HARA, and HTB-182 cell lines were ~90% confluent. The RWGT2 cell line is derived from a human lung SCC bone metastasis, the HARA cell line from a human primary lung SCC, and the HTB-182 cell line from a primary human lung SCC (52-55).
Stable-expressing HTB-182/EGFR and HTB-182/LXSN cell lines were made using a retroviral transduction approach (56). The recombinant retroviral constructs, pLXSN-EGFR and pLXSN, were described previously (57, 58).
For the in vitro ligand-stimulating experiments, cells were treated with either 17 nM EGF (Sigma) or 93 nM human recombinant AREG (R&D Systems). These are saturating concentrations determined from stimulating cells with series of concentrations ranging from 0.017 nM - 170 nM for EGF and .09 nM - 930 nM for AREG. The lysates were resolved by SDS-PAGE and then immunoblotted for general phosphotyrosine status (not shown) (4G10, Upstate). For the in vitro EGFR tyrosine kinase inhibitor studies, PD153035 (Tocris) was dissolved in dimethylsulfoxide (DMSO) and used at a final concentration of 1 μM. PD153035 is an anilinoquinazoline that inhibits the tyrosine kinase activity of the Epidermal Growth Factor Receptor, ErbB2/HER2/Neu, and ErbB4/HER4 (59). This molecule functions by inhibiting ATP binding to the tyrosine kinase domain (59). It exhibits approximately 50-fold selectivity for the EGFR over ErbB2 and approximately 20-fold selectivity for the EGFR over ErbB4 (60). PD153035 was added to cells for 3 or 6 hours, while control cells were treated with media containing 0.01% DMSO (vehicle). For the ligand studies, control goat IgG and goat anti-AREG antibodies (R&D Systems) were resuspended in sterile phosphate-buffered saline (PBS). Cells were treated for 24 hours with 10 μg/ml of either goat IgG or the anti-AREG antibody. Cells were treated 16-24 hours with GM6001 (Calbiochem) which is a broad spectrum metalloproteinase inhibitor that targets the active site zinc atom was used at a final concentration of 10 μM which is the recommended dosage for cultured cell work (61) (62). Src kinase inhibitor 1 (c-Src/Lck inhibitor; Calbiochem), was dissolved in DMSO and applied to cells overnight at a concentration of 10 μM. Src kinase inhibitor 1 is a quinazolinamine that targets ATP binding pocket and inhibits tyrosine kinase activity. It exhibits 20 to 50-fold greater selectivity for Src/LcK than serine threonine or other tyrosine kinases (61). Pertussis toxin was dissolved in sterile water and applied to cells for 16-24 hours at a concentration of 100 ng/mL (Sigma). All inhibitors were used at doses that did not produce alterations in cell morphology when placed on cells for 24 to 48 hours. For Western blot analysis, the following primary antibodies were used: anti-EGFR rabbit polyclonal antibody (1:1,000), anti-ErbB2 (1:1,000), anti-ErbB3 (1:1,000) and anti-ErbB4 (1:200) (Santa Cruz Biotechnology); anti-phosphotyrosine mouse monoclonal antibody (1:1000) (Upstate Biotechnology); anti-TACE/ADAM 17 rabbit monoclonal antibody (1:1,000) (QED Bioscience Inc.); anti-β-tubulin mouse monoclonal antibody (1:20,000) (Developmental Studies Hybridoma Bank). Secondary antibodies include: goat anti-mouse and goat anti-rabbit antibody (1:5,000) conjugated to horseradish peroxidase (KPL laboratories).
The HARA line was selected due to its ability to be efficiently transfected (up to ~70% of cells) using the siRNA transfection reagents and a GFP reporter construct, whereas only 10 to 20% of RWGT2 cells were transfected using this construct (data not shown). The siRNAs were all initially ordered from Santa Cruz as part of a starter kit that contained at least 3 individual siRNAs or pools of siRNAs. These were then tested individually and the lot chosen for further studies which has the most substantial decrease in the protein of interest. For each individual transfection 2 × 105 cells/well were plated into a 6-well dish. The EGFR siRNA, TACE siRNA, AREG siRNA, or scrambled control siRNA (sc-29301, sc-36604, sc-39412, sc-37007; Santa Cruz Biotechnology) duplexes were used at concentrations ranging from 0.5 to 0.75 μg in 100 μl of siRNA Transfection Medium (sc-36868; Santa Cruz). The media-duplex solution was further combined with 6 μl of siRNA transfection reagent (sc-29528; Santa Cruz) in 100 μl of siRNA transfection medium, incubated for 30 minutes at room temperature, combined with 800 μl of transfection medium, and placed on freshly washed cells (~50% confluent). Cells were placed in a incubator for 7 hours at 37°C, followed by the addition of 1 ml of fresh medium supplemented with 20% FBS. Cells were incubated an additional 24 hours, followed by another media change with normal medium, and harvested 48 hours after the initial addition of the siRNA duplexes.
Immunoblotting for the EGF family of receptors was performed as described in Gilmore et al., 2007. Briefly, confluent cells were washed with cold PBS, and lysed in isotonic EBC lysis buffer (50 mM Tris pH 7.4, 120 mM NaCl, 0.5% NP40). Cell lysates were harvested into microcentrifuge tubes, incubated on ice for 20 minutes, centrifuged for 10 minutes at 15,000 rpm, and supernatants were transferred to fresh tubes. Protein concentrations were determined using a Bradford protein assay, and concanavalin A-sepharose beads were used to precipitate glycosylated proteins (including ErbB receptors) from cell lysates. Both the precipitated or nonprecipitated lysates (used for TACE) were resolved by SDS-PAGE on a 7.5% polyacrylamide gel, and transferred to PVDF membranes (BioRad). Membranes were blocked in 5% milk:TBS-T (Tris 200 mM, pH 7.4, 150 mM NaCl, and 0.05% Tween-20)(63) and probed with the relevant primary antibody overnight at 4°C. The immunoblots were then washed 3 × 5 minutes in TBS-T buffer, probed with the corresponding secondary antibody conjugated to horseradish peroxidase for 1 hour and washed 3× for 5 minutes in TBS-T. Primary antibody binding was detected using a goat anti-mouse or goat anti-rabbit antibody conjugated to horseradish peroxidase and enhanced chemiluminescence (Santa Cruz). The immunoblots were then stripped and reprobed with an anti-β-tubulin mouse monoclonal antibody (Santa Cruz) and detected as described above.
One milligram sections were removed from frozen tumors. The sections were pulverized using a mortar and pestle in EBC lysis buffer supplemented with protease inhibitor cocktail and protein phosphatase 1 and 2 cocktail (Sigma). The lysates were then quantified, resolved by SDS-PAGE and immunoblotted as described above.
HARA, RWGT2, HTB-182, HTB-182/EGFR and HTB-182/LXSN cell lines were grown to confluence in 12-well dishes followed by serum starvation for 24 hours. Conditioned media was collected and PMSF was added to a final concentration of 1 mM. Samples were centrifuged for 10 minutes at 4°C, and supernatants were transferred to new microcentrifuge tubes. AREG, BTC, EGF, TGFα and HB-EGF protein levels were determined from the conditioned media using the DuoSet ELISA kit per manufacturer's instructions (R&D Systems).
Total RNA was isolated and purified from cells using the mini RNA isolation II kit (Zymo Research Corp.) according to the manufacturer's instructions. Reverse transcription (RT) of total RNA was performed in a final volume of 50 μl. One microgram of total RNA was treated with RQ1 DNase (Promega Corp.) and 1 U RNasin RNase inhibitor (Promega Corp.), in 1 × PCR buffer, 5 mM MgCl2, and 1 mM dNTP (New England BioLabs) at 37°C for 30 minutes prior to first strand cDNA synthesis. After DNase treatment, the total RNA was subjected to reverse transcription using 1 U Moloney murine leukemia virus (MuLV) reverse transcriptase and 2 μM random hexamers (Applied Biosystem) at 42°C for 30 minutes. The reverse transcriptase was then inactivated by heating to 75°C for 5 minutes, followed by a 5 minute cooling period at 4°C.
Quantitative real-time reverse transcriptase PCR (QRT-PCR) was performed using the DyNAmo HS SYBR Green qPCR master mix (New England BioLabs) according to the manufacturer's instructions and employing 2 μl of cDNA. PCR reactions were performed in a DNA Engine Opticon System (MJ Research Inc.). The PCR cycling consisted of an initial incubation at 95°C for 15 minutes to activate HotStartTaq DNA polymerase (from??), followed by 45 cycles of denaturation at 94°C for 15 seconds, annealing for 30 seconds, and extension at 72°C for 30 seconds. Specificity of the PCR products was determined using a melting curve analysis from 55°C to 95°C. Samples lacking a template and non-RT-treated samples were included as negative controls. Detected mRNA transcript levels of PTHrP P1, P1/P2, and P3 promoters were normalized to GAPDH, and data is expressed as the ratio of the message of interest to GAPDH. The results represented in the figures were derived from at least 3 separate experiments. The qPCR primer sequences used were previously described (7, 27); forward PTHrP-all transcripts primer (5′- GTCTCAGCCGCCGCCTCAA -3′) and reverse PTHrP-all transcripts primer (5′- GGAAGAATCGTCGCCGTAAA -3′), forward GAPDH (5′- CATGGAGAAGGCTGGGGCTC -3′) and reverse GAPDH (5′- GATGGCATGGACTGTGGTCA -3′), forward P1 primer (5′-CAGCCAGAAGAGCAGAGAGAA) and reverse P1 primer (5′-GCGAGTTGAAAACCGAGCG), forward P1/P2 (5′-GAAGCAACCAGCCCCCAGA) and reverse P1/P2 (5′-TGAGACCCTCCACCGAGC), forward P3 primer (5′-GGAGAAAGCACAGTTGGAGTAGC) and reverse P3 (5′-TCTTTTGAGGCGGCGGCTGA).
All experiments that used animals were approved by the Indiana University IACUC and were performed in compliance with stipulations of that body. Age matched Foxn1nu nude mice (Harlan) were injected subcutaneously with 1×106 RWGT2, HARA, HTB-182, or HTB-182/EGFR cells. Subcutaneous tumors were observable approximately 10 days after injection. Mice were monitored and weighed every week, until tumor size was greater than 1 cm in diameter. From this point, mice were monitored every other day for increases in tumor size, weight loss and general comfort level. In addition, serum calcium concentrations were evaluated every 10 days. Tumor-bearing mice were euthanized when hypercalcemia (≥12 mg/dL) developed or tumor diameter was ≥ 2 cm. Terminal blood samples obtained via cardiac puncture were centrifuged for serum separation and collection and stored at −80°C. Tumor tissue was harvested, weighed, snap-frozen in liquid nitrogen or fixed in 10% neutral buffered formalin, and stored at either −80°C for protein analysis, or embedded in paraffin blocks for future histological evaluation.
Total serum calcium concentrations were monitored throughout the course of the experiment using a commercial QuantiChrom Calcium Assay Kit (BioAssay Systems). Tail venipuncture was used to collect blood via capillary action into a non-heparin coated capillary tube (Fisher Scientific). Serum samples from each experimental subject were placed in tubes and stored at −80°C for calcium analysis. Samples and commercial standards were aliquoted into 96-well plates per manufacturer's instructions (BioAssay Systems), and the intensity of the colorimetric reaction was measured at 612 nm in a standard 96-well plate reader. Final calcium concentrations were calculated against a standard curve and reported as mg/dL.
Either 2×104 HTB-182/LXSN, HTB-182/EGFR or HARA cells, used as a positive control, were injected into the left tibia of age matched Foxn1nu nude mice (Harlan). Weekly radiography was used to monitor the progression of tibial tumors (20 kV for 45 seconds; Faxitron). Lesions were first detected 14-days post injection. The endpoint of these studies was 28 days as the X-ray detectable lesion in the HTB-182/EGFR animals approached the limit designated by the animal care and use protocol. Both the HTB-182/LXSN and HTB-182/EGFR groups of mice were sacrificed at 28 days. The lytic areas were determined using image J software version (NIH).
Bones were removed from the mice fixed in 10% buffered formalin for 48 hours and then placed in 70% EtOH. Bones were embedded (undecalcificed) in methylmethacrylate. The polymerized blocks were sectioned at 6 μm in the frontal plane, mounted on charged microscope slides. Hemotoxylin and eosin staining was done on 3 sections of each bone. The sections were submerged in a 1:1 mixture of xylene and chloroform to remove the plastic. They were then cleared in 3 × 5 minutes of xylene and rehydrated with alcohols. The sections were then stained hemotoxylin and eosin.
Tartrate resistant alkaline phosphatase staining was done on 6 sections of each bone to detect active osteoclasts. Sections were incubated in methyl methacrylate for 3 × 60 minutes with agitation. They were cleared in 3 × 20 minutes of acetone and rehydrated with alcohols. The sections were first stained for Von Kossa, which stains calcified bone black. The sections were then pretreated for 1 hour with 0.2 M Tris (pH 9.0) and then incubated with TRAP stain following the manufacture's protocol (Sigma). The sections were the photographed immediately.
Histological sections were submerged in a 1:1 mixture of xylene and chloroform, cleared in xylene, and rehydrated with alcohols. The sections were the decalcified with 1% acetic acid and then epitope retrieval was done with pepsin solution (Zymed labs). The sections were blocked with normal horse serum for 30 minutes and primary antibody (mouse anti-human EGFR; Zymed Labs) was placed on the sections for 48 hours at 4°C. The sections were then incubated in peroxidase blocking solution (30% H2O2 in methanol) for 10 minutes. Secondary antibody was then added for 30 minutes at room temperature (biotinylated horse anti-mouse IgG, Vector Labs). Primary antibody was detected using HRP-Streptavidin (Vector Labs) and DAB peroxidase (Vector Labs). The sections were then counterstained with methyl-green.
Micrographs of H&E stained tumor bearing bone sections (10X, 6 sections from distinct regions of the bone were used to generate measures) and area of tumor bearing regions was measured with Image J software. The average tumor area per section is reported in 8D. Osteoclasts were counted on serial 10X micrographs that captured the entire section (TRAP and Von Kossa stained) of the tumor bearing bones and the control non-injected bones from mice bearing the HTB-182/LXSN tumors in the left leg. Also the entire bone surface area including the preriosteum and endosteum was measured from these sections (64). The mean of the average number of osteoclasts per section divided by bone surface length of the section are reported in 8C, and the distribution of osteoclasts within the bone was carefully noted.
Results of in vitro experiments are expressed as the mean ± SD of triplicate or quadruplicate measures of independent replicates for a single experiments. Results of in vivo experiments are expressed as the mean ± SEM of 3 to 6-replicates of samples taken from 4 or 5 individual animals. All statistical comparisons were based on two–tailed analysis of the Student's t-test. A probability value of P<0.05 was considered to be significant.
We thank Drs. Theresa A. Guise and Gregory R. Mundy for the use of the RWGT2 cell line, and Dr. Hauro Iguchi for the use of the HARA cell line. Ruth Sanders is thanked for her careful reading and editing of the manuscript. Alan Rosenbaum and Erin Chrismore are thanked for help with osteoclast counts and histomorphometery. This work was supported by National Institutes of Health grant; DK067875, Indiana University Biomedical Research Grant to J.F., the Walther Cancer Center Postdoctoral Grant to J. L. G., and Susan G. Komen Postdoctoral Award to J. L. G., the National Center for Research Resources K01RR021879 to G.L., National Institutes of Health grant AR53237 to A.R., and IU School of Medicine and Medical Sciences dissertation support for R.G.
Conflict of Interest
The authors state no conflict of interest.