Background
The family of human epidermal growth factor (EGF) tyrosine kinase receptors (HER) includes HER1 (also known as the EGF receptor), HER2, HER3 and HER4. These receptors play important roles in diverse cellular processes, including but not limited to, cell growth, proliferation and migration [
1]. Once activated, HER receptors initiate the recruitment of intermediate signaling proteins, which subsequently activate downstream signal cascades that trigger the cellular responses [
2]. HER2 receptors lack a ligand-binding domain and HER3 receptors lack intrinsic tyrosine kinase activity [
3]. Even so, HER2 and HER3 form dimers with other ligand-bound HER receptors, and thereby participate in signal transduction. Upon ligand binding, HER1 and HER4 are quickly phosphorylated and activated. Receptor activation can result in the release of their cognate ligands, which then act as a positive feedback loop through autocrine/paracrine signaling.
Aberrant HER receptor signaling, either due to overexpression or mutation of one or more HER receptors or due to abnormal production of their ligands, contributes to the development and progression of a broad spectra of human cancers, including breast, colon, lung, ovarian, and head and neck cancers [
4‐
7]. Since portions of these proteins are all released to the extracellular environment, HER receptors and their ligands are not only potential therapeutic targets for the treatment of these cancers, but also potential cancer biomarkers [
8‐
11].
A number of HER ligands have been identified as cancer biomarkers, including EGF, amphiregulin (AREG), heparin-binding EGF-like growth factor (HB-EGF), and transforming growth factor-α (TGF-α) [
12‐
14]. These ligands are tightly associated with HER receptor expression in a variety of cancer types. For example, studies have demonstrated a number of HER ligands are expressed and correlated with expression of HER receptors in breast cancer patients, and high expression of certain HER ligands are related to the biological aggressiveness of the tumors [
15]. All of these ligands are initially synthesized as membrane-anchored proteins [
3]. Soluble ligands are released through a process called "shedding", which involves proteolytic cleavage on the extracellular side of the transmembrane domain. Shedding is the last step in the secretion of the biologically active ectodomain of the ligands. Similar to HER ligands, HER receptors undergo shedding during both physiological and pathological conditions. In general, this process is thought to represent one of several feedback mechanisms that prevent prolonged receptor activation. Metalloproteases, including the disintegrin and metalloproteases (ADAMs), are recognized as the major mediators of receptor and ligand ectodomain shedding [
3,
16,
17].
Serum concentrations of secreted HER ligands and HER receptors have been investigated rigorously as potential prognostic factors and therapeutic indicators for many cancer types. However, numerous studies suggest that no single protein biomarker assay may have sufficient sensitivity and specificity to be used clinically, especially for early detection. In particular, the tumor microenvironment appears to be a highly regulated system. Its secretome consists of substantial numbers of proteins that are processed through regulated secretory pathways. There is considerable evidence that secretion of these proteins is altered due to a variety of physiological or pathological conditions. Therefore, in addition to HER receptors and ligands, many other groups of circulating proteins have been examined as potential prognostic factors in diagnosis of human cancers.
One such group of proteins is the cytokines [
18,
19]. In the pathogenesis of carcinogenesis, pro-inflammatory cytokines such as interleukin-1α (IL-1α), tumor necrosis factor (TNFα), and regulated upon activation, normal T cell expressed and secreted (RANTES), can regulate host responses to infection, inflammation and various immune responses. Pro-inflammatory cytokines can also induce expression of adhesion molecules and metalloproteases, both of which are involved in the process of tumor invasion [
20]. Besides cytokines, fibrogenic and angiogenic factors, including basic fibroblast growth factor (bFGF) [
21], platelet-derived growth factor (PDGF) [
22], vascular endothelial growth factor (VEGF) [
23], hepatocyte growth factor (HGF) [
24], insulin-like growth factor (IGF) [
25], are capable of stimulating mitogen signaling pathways and are involved in a wide variety of cellular processes. Although these proteins are not known to directly interact with the HER receptors, it has been demonstrated that most of them can manipulate HER-regulated signal pathways, commonly by transactivation of these receptors [
26‐
28]. Furthermore, increased expression of these proteins in breast cancer is associated with overexpression of HER family members [
29,
30]. Hence, an evaluation of the relationship between HER expression and cytokines may add valuable information for cancer prognostics.
In regard to potential biomarkers like growth factors and cytokines, there has been a great deal of research on how these proteins alter epithelial cell function and downstream cell-signaling pathways. In contrast, there is very little known about how changes in epithelial cell processes regulate the secretion of these proteins. A better understanding of these cellular mechanisms is needed if we are to gain useful mechanistic insight into tumor biology based on circulating biomarker data. Recent evidence suggests that circulating levels of breast cancer biomarkers vary with stage [
31] and with HER2 receptor status [
19]. In this study, we use human mammary epithelial cell lines (HMEC) expressing different levels of HER receptors to examine effects on the secretion patterns of these potential biomarkers. HER1-3 are the most commonly studied HER receptors. Therefore, we focused our study on a parental HME cell line with endogenous HER1 expression, as well as derived cell lines that were transfected to overexpress either HER2 or HER3, while still maintaining the basal HER1 expression. Since these HMEC lines were originally derived from normal breast cells, they are not carcinoma cell lines, and are similar to normal mammary epithelial cells in that they require HER1 activation for proper proliferation and migration responses [
32]. In this study, we treat cells with a single concentration of EGF to activate HER1, and then examine the effects on protein secretion in all three HMEC lines. Overall, this study provides novel insight into the underlying molecular processes that regulate biomarker secretion and illustrates how HER2 and HER3 co-expression can affect the secretion of a variety of bioactive proteins that are important in breast cancer development.
Methods
Materials
EGF (human recombinant) was purchased from Peprotech (Rocky Hill, NJ). Protease inhibitor cocktail III, as well as all signaling pathway inhibitors, including PI3-kinase inhibitors LY 294002 and wortmannin, MEK inhibitors PD 98059 and U0126, were purchased from Calbiochem (La Jolla, CA). All capture and detection antibodies, including the commercial source and catalog numbers, that were used here for the sandwich ELISA protein microarrays have been previously described [
33]. All other reagents were from Sigma Chemical Company (St. Louis, MO), unless otherwise indicated.
Generation of HER Cell Lines
Human mammary epithelial cell line (HMEC) 184A1L5 was graciously provided by Martha Stampfer (Lawrence Berkeley Laboratory, Berkeley, CA) and maintained at 37 °C in 5% CO
2/air in DFCI-1 medium supplemented with 12.5 ng/ml EGF as described [
34,
35].
Both HER2 and HER3 expressing cell lines were derived using a retrovirus-based strategy, as described previously [
36,
37]. Briefly, transfected cells expressing HER2 or HER3 were screened in DFCI-1 medium with the addition of 250 μg/ml G418 (InVitrogen, Carlsbad, CA) or 2 μg/ml puromycin (Sigma, St. Louis, MO), respectively. The abundances of HER receptors on these transfected cells were characterized using flow cytometry, with Alexa-488 conjugated mAb 7C2 against HER2 and phycoerythrin conjugated anti-HER3 antibody (R&D Systems Inc., Minneapolis, MN). Individual clones of retrovirus transfected HMEC were isolated using cloning rings (Fisher Scientific Inc., Pittsburgh, PA). Expression levels of HER receptors were determined by flow cytometry and western blotting.
Three cell lines were examined in this study. They include the parental cell line 184A1L5 expressing endogenous HER1, which is abbreviated as parental "HER1" cell line in this study. The transfected cell line that overexpresses HER2, as well as the basal HER1 receptor, is abbreviated as "HER2" cell line here; the transfected cell line that overexpresses HER3, as well as the basal HER1 receptor expression, is abbreviated as the "HER3" cell line. Levels of the HER receptors in these cell lines were quantified as described before [
36,
37]. Results from this analysis showed that there are approximately 2x10
5 HER1, 3x10
4 HER2 and 2x10
3 HER3 receptors on each parental HER1 cell, 1.5x10
5 HER1, 6x10
5 HER2 and 2x10
3 HER3 on each HER2 cell, and 2x10
5 HER1, 3 10
4 HER2 and 2.8 10
4 HER3 on each HER3 cell. It has been reported that the well characterized breast cancer cell line SK-BR3 express about 2x10
5 HER1, 6x10
5 HER2 and 1x10
4 HER3 receptors [
38,
39]. Based on these reports and another evaluation of HER receptor concentrations in breast cancer cells [
40], our HER2 or HER3 cell lines express HER2 or HER3, respectively, at levels that are comparable to those found in breast cancer cells that overexpress these receptors.
Cell Culture and Treatment
Only cells that were less than fifteen passages from the original frozen stock were used in this study. Each cell line was seeded at 3.0x105 cells per well of 6-well plates (BD Falcon, Franklin Lakes, NJ) in DFCI-1 culture medium, and were allowed to grow to confluence prior to treatment. Before EGF treatment, cultured cells were fasted for 14-18 h in serum-free, DFCI-1 medium that lacked all supplements except 0.1% bovine serum albumin. Cells were then washed twice with buffered saline, pH 7.2-7.4 (PBS), followed by the addition of 1 ml fresh serum-free medium with 12 ng/ml EGF added. Activated cells were incubated at 37 °C for up to 24 h and samples were collected at fixed time points after EGF addition. Immediately prior to harvesting, cells were chilled by placing the culture dishes on ice.
Sample Collection and Processing
Conditioned medium (CM) from the cultured cells was transferred to microcentrifuge tubes, and centrifuged at 2000 g for 5 min at 4 °C in order to remove any particulates or cell debris. An aliquot of each supernatant was then transferred to another tube that contained a 1/10th volume of 1% casein and green fluorescent protein in PBS (Bio-Rad, Hercules, CA), such that the final concentration of the green fluorescent protein was 100 pg/ml. Green fluorescent protein was used as the antigen in an internal calibrant assay based on a sandwich ELISA. The fluorescent signal from the capture antibody in this assay was used for data normalization using a custom bioinformatics program, ProMAT Calibrator, as described below. The cells on these culture plates were then washed twice with cold PBS, and harvested by adding 200µl of lysis buffer (50 mM pH 7.4 HEPES, 150 mM NaCl, 0.5% NP40, 1 mM PMSF, 1 mM Na3VO4 and 1% protease inhibitor cocktail III). Cell lysates were collected and centrifuged at 18,500 g for 10 min. The protein concentration of cell lysates was measured using the Bicinchoninic Acid protein quantitation kit (Sigma, St. Louis, MO) and averaged (± standard deviation) 1.6 ± 0.3 mg/ml. Since 200µl of lysis buffer was added, we estimate that about 0.32 mg of protein was collected from each plate.
In certain experiments, cultured cells were pre-incubated with a single inhibitor for 1 h prior to EGF addition. These inhibitors, and their concentrations, were 20 µM LY 294002, 0.2 µM Wortmannin, 25 µM PD98059, 10 µM U0126. Stock solutions of each reagent were individually prepared in DMSO and the final concentration of DMSO in the culture medium was 0.1% in all treatments, including controls. At least two independent experiments were performed, with five biological replicates (i.e., five cell culture dishes that were individually processed) in each experiment, for all results described here. Samples were stored at -80°C until analysis.
ELISA Microarray Analysis
Concentrations of individual proteins in CM and cell lysates were quantitatively measured using sandwich enzyme-linked immunosorbent assay (ELISA), as previously described in detail [
33]. Briefly, ELISA microarray chips were custom manufactured using aminosilanated, 25x75 mm glass slides (Erie Scientific, Portsmouth, NH) stamped with a hydrophobic barrier that was used to create 16 wells on each slide. The ELISA reagents used in these analyses have been previously evaluated and shown to have no cross-reactivity and to be able to quantitatively detect purified antigens that were spiked into human serum [
41]. The capture antibodies were suspended in PBS at concentrations ranging from 0.5 to 1.0 mg/ml. These antibodies were printed using a GeSiM noncontact NanoPlotter NP2 printer (Quantum Analytics, Foster City, CA). Sixteen identical chips were printed on each slide, such that each chip was isolated by a hydrophobic barrier. Each capture antibody and control reagent was printed in quadruplicate (once in each quadrant) on each chip. Successful printing was confirmed using the RedReflect capability on the ScanArray ExpressHT (PerkinElmer, Santa Clara, CA) laser scanner. The printed slides were blocked with 1% casein in PBS at 4 °C and, after washing, stored desiccated and under vacuum at -20°C until use.
In order to generate standard curves for each of the ELISA analyses, a single mixture containing all the antigens was prepared in 0.1% casein in PBS and containing 100 pg/ml green fluorescent protein. This stock solution of the standard mixture was aliquoted and stored at -80°C, and an aliquot was thawed on a daily basis for each ELISA microarray analyses. For this analysis, the standard stock solution was serially diluted 3-fold to create at least 7 dilutions of the standard mixture. Each dilution, and an antigen-free blank, was analyzed on three seperate chips.
Several incubation steps are included in processing the ELISA microarray chips [
41]. Washes were performed between each incubation step by submerging the slides in PBS containing 0.05% Tween-20 (PBS-T). Either 20 μl of the standard mixture or an individual, diluted sample was then pipetted onto each of three replicate chips. Samples were then incubated overnight in a closed chamber with saturated humidity and gentle mixing. Slides were then incubated with a cocktail of all detection antibodies in 0.1% casein/PBS buffer for 2 h. The level of biotinylation was then increased using the biotinyltyramide amplification system (Perkin-Elmer). Finally, slides were submerged in 1 μg/ml of Cy3- or Alexa 647-conjugated streptavidin in PBS-T, and incubated for 1 h in dark. Slides were then quickly rinsed with deionized water and dried. These slides were imaged with a ScanArray Express HT laser scanner. ScanArray Express software was used to quantify the spot fluorescence intensity from the scanned images. Spot fluorescent data were then processed and analyzed using Protein Microarray Analysis Tool (ProMAT) and ProMAT Calibrator [
42‐
44], both of which are open-source, freeware programs (available at
http://www.pnl.gov/statistics/ProMAT/) that we developed specifically for processing ELISA microarray data. Standard curves were fit to a four-parameter logistic model and used to estimate the individual protein concentrations in each sample.
RT-PCR Analysis
Quantitative assessment of mRNA expression was performed by real-time RT-PCR. Total RNA was extracted using RNeasy kit (Qiagen, Valencia, CA). The HMECs used in the quantitative RT-PCR analyses were cultured and harvested in parallel with the cells used in the comparable protein experiments in the CM and cell lysates, although the RNA studies were conducted using cells raised in seperate dishes. Complementary DNA were synthesized from total RNA via reverse transcription using ImProm II reagents (Promega, Madison, WI) and oligo-dT priming. The following primers were employed in the quantitative real-time RT-PCR:
cyclophilin A, forward GAGCTGTTGCAGACAAAGTTC and reverse CCCTGGCACATGAATCCTGG;
AREG, forward CGGAGAATGCAAATATATAGAGCAC and reverse CGTTCACCGAAATATTCTTGC;
TGF-α, forward TTGCTGCCACTCAGAAACAG and reverse ATCTGCCACAGTCCACCTG;
RANTES, forward CCTCATTGCTACTGCCCTCT and reverse GGTGTGGTGTCCCGAGGAATA;
IL-1α, forward GCATACGGGTCCTGGCATCTTGTCC and reverse ATGGTGATCTTCTTGCGGCTCTTGC.
PCR reactions were carried out in a Roche Lightcycler II using 20 ng cDNA and FastStart DNA Master
PLUS SYBR Green I reagent according to the manufacturer's instructions (Roche Applied Science, Indianapolis, IN). The PCR cycle parameters were set as denaturation at 95 °C for 10 s, annealing at 55 °C for 5 s and elongation at 72 °C for 10 s for 45 cycles. Melting curve analyses were performed from 60 °C to 95 °C in 0.5 °C increments. RNA from individual cell culture plates was prepared for each treatment group in five replicates. Quantitative RT-PCR data for AREG, TGF-α, IL-1α, and RANTES were normalized based on cyclophilin A transcriptional level [
45].
Quantification of Erk/Akt activation
Phosphorylated Erk1 and Akt levels of cell lysates were quantified by conventional ELISA techniques in a 96-well microplate. R&D System's DuoSet IC ELISA kits were used in these measurements, as described before [
37]. The manufacture's protocols, including the lysis buffer, were followed in all the ELISA measurements. Before each ELISA assay, protein concentrations of the cell lysates were measured using the Bicinchoninic Acid protein quantitation kit (Sigma, St. Louis, MO).
Data Analysis
Concentrations of secreted proteins are presented in pg per ml of conditioned medium (CM). The total volume of medium/dish was a constant within and across studies. Concentrations of individual proteins in cell lysates are presented in ng per mg lysate protein. We used total (i.e., accumulated) secreted protein in the temporal studies for comparisons across the cells lines. An alternative approach would have been to use non-stimulated cells as the control, and present the difference between stimulated and non-stimulated cells. This approach was not used because the cells were first fasted overnight in serum-free medium. To continue to deprive the cells of growth factors would have led to cell death during the remaining 24 hr study, and therefore was not a reasonable option. Rather, the approach used allowed us to compare cell lines in parallel and directly demonstrate the involvement of HER2 or HER3 in protein secretion following HER1 activation.
Statistical differences between cell lines following EGF stimulation and/or signaling pathway inhibition were initially determined by analysis of variance, and then delineated using Fisher's protected least significant difference test using StatView 5.0.1 software (SAS Institute). A significance level of 0.05 was used in all cases.
Discussion
In this study, we examined protein secretion patterns in HMEC upon HER1 receptor activation. This group of proteins not only includes HER receptors and ligands, but a variety of MMPs, cytokines and growth factors that regulate cellular behavior in both normal and pathological conditions. All of these proteins have been associated with the development of a variety of epithelial cancers, including breast cancer; and many of them have been investigated as potential cancer biomarkers. Our goal in this study was to better understand the underlying mechanism that links HER receptor activation and biomarker secretion. In particular, by examining three HMEC lines that express different levels of HER receptors, we attempted to determine the influence of these receptors on the biomarker secretion and their regulatory mechanisms.
Our results suggest that increased HER2 and HER3 expression potentiates ligand induced autocrine and paracrine signaling resulting from EGF activation of HER1. Elevated levels of HER2 or HER3 expression increased HER1 ligand secretion, suggesting an increase in the autocrine positive feedback loop that is associated with HER1 activation (Figure
1). The differences in EGF consumption rate among the three HMEC lines clearly imply that the presence of HER2 or HER3 accelerates HER1 ligand consumption. In contrast to EGF (which is not synthesized by the HMECs), the changes in concentration of AREG, HB-EGF and TGF-α reflect the net balance of two essential processes in this autocrine system: ligand cleavage from the synthesized precursors on cell membrane and ligand binding/consumption through receptor capture and internalization. Because there is a large molar-excess of EGF in the culture medium compared to the secreted HER1 ligands in the early time points of our experiments, it seems likely that the high concentrations of EGF effectively outcompetes the relatively low concentrations of secreted ligands for binding sites on HER1. Thus, the steady accumulation of the secreted HER1 ligands in the CM during those early time points is likely to primarily reflect the rate of ligand shedding. At later time points, when the EGF levels are mostly depleted, the loss of the secreted ligands from the CM likely reflects consumption associated with HER1 binding. All three secreted ligands can contribute to autocrine signaling through HER receptors [
32], and all four HER1 ligands assayed here possess similar EC
50 values for activating HER1 receptor. Even so, ligand binding has been reported to be strengthened when HER1 dimerizes with either HER2 or HER3 [
55]. Since, the first step in ligand consumption is binding to the receptor, followed by endocytosis of the receptor-ligand complex and protein degradation, it seems likely that enhancement of HER1 ligand-binding affinities due to increased HER1 dimerization with HER2 or HER3 accounts for the increased EGF consumption observed in the HER2 and HER3 cell lines in the early time points, and increased consumption of other HER ligands at the later time points. It should be noted that, once endocytosed, some HER ligands are not degraded, but they are recycled back to the cell membrane along with HER receptors and are then released from the cell. Compared to HER1, recycling of HER2/HER3 is more prominent [
56]. This recycling property potentially may contribute to the increased accumulation of TGF-α in HER2 and HER3 cell lines.
All HER ligands measured in this study can initiate phosphorylation of HER receptors and lead to downstream signal transduction, including the activation of MEK/Erk and PI3K/Akt pathways [
1]. Activation of Erk and Akt can provide critical cell mitogenic and survival signals required for tumor progression. Components of these two pathways are frequently abnormal in a variety of cancer tumors and may represent biologically relevant targets for anti-cancer therapy [
57‐
59]. In addition, MEK pathway activation can induce transcriptional activation of HER ligands like TGF-α [
59]. Our observation that inhibitors of MAPK/Erk largely blocked HER ligand secretion emphasizes the importance of this pathway in HER autocrine signaling. We also identified a regulatory role of PI3K/Akt in this process (Figure
5). We do not know of any prior reports that identify this pathway in HER1 autocrine signaling. Our study finds that both of these signaling pathways can regulate HER activation by stimulating ligand synthesis and, subsequently, shedding.
In addition to HER ligands, we detected the shedding of HER1 and HER2 in HMEC. Changes in the circulating levels of HER receptors have been reported in a variety of different cancers. Elevated circulating HER2 levels have been identified in a group of patients with HER2-positive breast cancer with significantly worse outcome [
60‐
62]. In contrast, patients diagnosed with ovarian, head and neck, non-small cell lung cancer display lower serum HER1 concentrations, and changes of serum HER1 concentration have been correlated with the efficiency of cancer treatment [
11]. In this study, we did not observe prominent HER1 or HER2 shedding in HMEC, even in cells that overexpress HER2 (Figure
6). Our results are consistent with a recent report that investigated HER1 shedding in malignant cells [
63]. Their results showed that among a few malignant cell lines, HER1 shedding only occurred in cell lines that express more than 7x10
5 HER1 per cell and were treated for 8 h or more with phorbol 12-myristate 13-acetate.
MMPs are widely expressed by many types of cells. High levels of MMPs are associated with tumor invasion and metastasis [
54]. Previously, we reported that TNF-α stimulation could induce production of MMP9 in our parental HMEC, the HER1 cell line, and that this process was dependent on autocrine signaling that activated HER1 [
64]. In this study, we examined the secreted levels of MMP1, 2 and 9 after EGF treatment. The results show that MMP2 levels are similar among all the three HMEC lines and that EGF treatment does not increase MMP2 levels. Thus, it appears that MMP2 secretion is not regulated by HER receptor activity. In contrast, MMP1 expression and secretion were induced in response to HER signaling. Notably, increased levels of the HER2 receptor enhance basal MMP1 secretion (Figure
8A). In parental cells and HER2 cells, induction of MMP1 expression can be triggered only through ligand-induced HER1 activation. Interestingly, induction of MMP9 secretion by EGF signaling was only observed in parental cells, implying the presence of HER2 or HER3 may repress MMP9 secretion in response to HER1 activation by an undefined mechanism.
Our conclusion that MMP secretion is mediated by the Erk pathway is in agreement with previous reports [
65‐
68]. We also observe that the Akt pathway has a role in MMP secretion (Figure
8). As inhibition of MMP secretion through blockade of either pathway is only effective when secretion is stimulated by HER receptor activation, these results indicate that Erk and Akt pathways mediate MMP secretion through HER signaling in HMEC. Overall, these results suggest that secretion of these three MMP forms are regulated by distinct molecular processes and, thus, that their use as biomarkers may provide differential insight into cell-signaling processes occurring in breast cancer cells.
Cytokines play an important role in the pathogenesis of cancers. Secretion of most cytokines undergo trafficking through the Golgi [
69], and are not dependent on proteolytic shedding. In this study, we found that activation of HER signaling could trigger the release of cytokines in the HMEC systems, apparently through increased synthesis of the cytokines (Figure
10).
Continuous accumulation of both PDGF and VEGF was observed in all the HMEC lines. Even so, our results suggest that the regulatory mechanisms that control the secretion of these two proteins are distinct. In HMEC, secretion of PDGF seems to be a constitutive process that is independent of Erk and PI3K/Akt pathways. In contrast, synthesis and secretion of VEGF is a direct and spontaneous response to HER signaling through EGF stimulation, and both MAPK/Erk and PI3K/Akt pathways appear to have a role in VEGF secretion. Our results are consistent with previous studies that demonstrate that HER1 regulates VEGF expression in a variety of different cancer cell lines [
70].
Competing interests
The authors declare that they have no competing interests.
Authors' contributions
All authors contributed to the study design, data analysis and approved the final manuscript.