Growth Factor-Specific Signaling Pathway Stimulation and Gene Expression Mediated by ErbB Receptors

The mechanisms by which receptor tyrosine kinases (RTKs) utilize intracellular signaling pathways to direct gene expression and cellular response remain unclear. A current question is whether different RTKs within a single cell target similar or different sets of genes. In this study we have used the ErbB receptor network to explore the relationship between RTK activation and gene expression. We profiled growth factor-stimulated signaling pathway usage and broad gene expression patterns in two human mammary tumor cell lines expressing different complements of ErbB receptors. While the growth factors EGF and NRG1 similarly stimulated Erk1/2 in MDA-MB-361 cells, EGF acting through an EGF receptor/ErbB2 heterodimer preferentially stimulated PKC, and NRG1 β acting through an ErbB2/ErbB3 heterodimer preferentially stimulated Akt. The two growth factors regulated partially overlapping yet distinct sets of genes in these cells. In MDA-MB-453 cells, NRG1 β acting through an ErbB2/ErbB3 heterodimer stimulated prolonged signaling of all pathways examined relative to NRG2 β acting through the same heterodimeric receptor species. Surprisingly, NRG1 β and NRG2 β also regulated partially overlapping but distinct sets of genes in these cells. These results demonstrate that the activation of different RTKs, or activation of the same RTKs with different ligands, can lead to distinct profiles of gene regulation within a single cell type. Our observations also suggest that the identity and kinetics of signaling pathway usage by RTKs may play a role in the selection of regulated genes. lysed 1 ml 1X sample µ l lysates phosphorylated anti-pAkt(S473), rabbit anti-pp70(T389, T421, S424), rabbit anti-pp90(S381), anti-pPKC(pan), rabbit anti-phospho-c-Myc(T58, S62), rabbit anti-phospho-c-Jun(S63, S73), and rabbit anti-pCREB(S133). Filters were stripped and re-probed with antibodies to actin (Sigma).


Introduction
Polypeptide growth factor hormones act on individual cells within tissues or on pluripotent stem cells to induce responses that contribute to development, tissue maintenance and repair, or disease state. Depending on cell type and the identity of the growth factor presented to the cell a variety of responses are possible, including proliferation, differentiation, apoptosis, survival, migration and fate specification. A fundamental question in growth factor signaling concerns the mechanisms by which specificity is generated: how do different growth factors elicit different responses within a single cell type, and how do different cell types respond differently to a single growth factor? Cellular responses to growth factors are mediated by cell surface receptor tyrosine kinases (RTKs) that possess an intrinsic protein tyrosine kinase activity. Growth factor binding stimulates receptor dimerization and autophosphorylation on several tyrosine residues, and phosphorylated tyrosines provide docking sites for intracellular signaling proteins that contain src homology-2 (SH2) or protein tyrosine binding (PTB) domains (1-2). The sequence-specific recruitment of signaling proteins couples activated RTKs to intracellular signaling cascades that propagate signals to the nucleus to elicit initial changes in gene expression. Growth factorstimulated expression of immediate early genes (IEGs), or genes whose induction does not require protein synthesis, then lays the foundation for the ultimate cellular response (3)(4)(5).
Over the past decade several signaling cascades connecting activated RTKs to the nucleus have been characterized (6). For example, activation of the Erk serine/threonine kinases follows the Through the phosphorylation of multiple tyrosine residues each RTK has the capacity to stimulate several different signaling cascades. By independently targeting different subsets of genes or by acting in a combinatorial manner to regulate gene expression, the different signaling pathways have the potential to mediate a variety of cellular responses. However, many different RTKs utilize identical signaling pathways in mediating diverse responses to growth factors.
Hence, a major question has become how signaling cascades couple growth factor receptors with specific gene expression patterns to mediate a diverse array of cellular responses.
One model suggests that RTKs send general signals through a limited number of pathways, and that these signals are interpreted in the target cell by context-specific transcription factors. A series of recent studies with invertebrate model systems support this view. Analysis of the regulatory regions of marker genes for specific late-stage developmental events in Drosophila Sweeney et al.,9 mismatch at the thirteenth residue. Expression of each gene sequence is reflected in the fluorescence of the identical series relative to the mismatch series, and is reported by GeneChip software as an "average difference" (AD) value (for a further discussion see ref. 11). AD values reported for genes on chips corresponding to growth factor-treated samples were normalized to control sample chips by scaling to equivalent total average differences as previously described (41). A threshold value of 30 was assigned to genes on all chips with a reported AD value of less than 30. Genes were then sorted for growth factor response according to these normalized, thresholded AD values. Genes that exhibited a difference in AD values between growth factor treatment and control of less than 100 in at least one of duplicate experiments were discarded as either not expressed or not reproducibly regulated by growth factor. To be considered regulated by growth factor, the AD value for a gene must have been 2.5 fold higher or lower with growth factor treatment relative to control in both of duplicate experiments. To be considered preferentially regulated by a given growth factor, the fold stimulation or suppression by one growth factor must have been 1.5 fold greater or lesser than by the other in both of duplicate experiments. express very low levels of ErbB4, which with prolonged exposure was observed to be stimulated exclusively by NRG1β (not shown).
The initial impact of differential receptor activation is on the first step in signal transduction, the recruitment of SH2 and PTB domain-containing proteins to activated receptors. We examined the association of signaling proteins with activated receptors first by blotting receptor immunoprecipitates with antibodies to signaling proteins (Fig. 1A, lower two panels). As expected, EGF preferentially stimulated the recruitment of the adaptor protein Grb2 to EGF receptor while NRG1β preferentially stimulated the recruitment of this protein to ErbB3. NRG1β also preferentially stimulated the association of p85, the 85 kDa subunit of phosphoinositide 3kinase (PI3K), with ErbB3, as previously demonstrated (40).
We also examined the growth factor-stimulated recruitment of tyrosine-phosphorylated proteins to complexes with the adaptor proteins Grb2 and Shc, the protein tyrosine phosphatase SHP2, p85 and the negative regulatory protein cbl (Fig. 1B). In general each of the signaling proteins associated with the EGF receptor upon EGF treatment, and with 185 kDa ErbB receptors upon NRG1β treatment. The exception was cbl, which as suggested previously (45) responded preferentially to EGF stimulation. However, other ligand-dependent differences in recruitment were also observed. A notable example was p85, which associated similarly with tyrosine phosphorylated proteins at 47 kDa, 55 kDa and 185 kDa after EGF and NRG1β treatment, but preferentially associated with other tyrosine-phosphorylated proteins in the 90-130 kDa range in response to EGF. These observations confirm that a major outcome of differential ErbB receptor activation by different EGF-like growth factors is the differential recruitment of intracellular by guest on July 10, 2020 http://www.jbc.org/ Downloaded from Sweeney et al.,12 signaling proteins into complexes with activated receptors and other tyrosine phosphorylated proteins.
As expected, differential ErbB receptor stimulation by EGF and NRG1β in MDA-MB-361 cells resulted in the differential stimulation of intracellular serine/threonine kinase cascades, as determined by the phosphorylation of these enzymes in response to growth factor treatment (Fig.   2). The extent and kinetics of activation of Erk1 and Erk2 were similar for the two growth factors, perhaps reflecting the similar levels of Grb2 recruitment to activated receptors. EGF more potently stimulated the phosphorylation of two PKC isoforms, consistent with its preferential recruitment of phospholipase C gamma (46). In contrast, NRG1β was reproducibly stronger in stimulating the phosphorylation of Akt at serine 473, consistent with the stronger recruitment of p85 to ErbB3 in response to this factor. Interestingly, EGF preferentially stimulated the phosphorylation of p70S6 kinase. Although both Akt and p70S6 kinase phosphorylation have both been demonstrated to depend on PI3K, the regulation of p70S6 kinase is complex and not yet fully understood (47).
The ultimate outcome of RTK stimulation by growth factors is the phosphorylation of nuclear factors to elicit changes in transcriptional regulation. To determine whether differential signaling through ErbB receptors might influence transcription factor regulation, we examined the phosphorylation state of three transcription factors, Myc, Jun and CREB, in response to EGF and NRG1β in MDA-MB-361 cells. Although the phosphorylation of these factors was stimulated by both growth factors, EGF reproducibly elicited a stronger response than did NRG1β (Fig. 3A).

Sweeney et al., 13
Differential Gene Expression in MDA-MB-361 Cells. Since EGF and NRG1β stimulated overlapping yet distinct patterns of intracellular signaling events in MDA-MB-361 cells, we examined whether the two growth factors also stimulated differences in gene expression. To test this we used Affymetrix HuGeneFL Array chips to simultaneously assess changes in transcript levels of ~5600 genes after one hour of growth factor treatment. We observed that while many transcripts were similarly elevated by both EGF and NRG1β, many were preferentially elevated by one or the other growth factor. The fold stimulations observed for a subset of genes is plotted in Fig. 3B.
Our analysis revealed 132 genes whose transcript abundance reproducibly changed in response to one or both growth factors. 92 mRNAs were enhanced at least 2.5 fold with growth factor treatment, while in two separate experiments 40 mRNAs were suppressed by at least 2.5 fold.
Interestingly, despite its weaker potency in stimulating several signaling cascades NRG1β was the stronger factor in regulating mRNA levels. 44 mRNAs were preferentially elevated by NRG1β, where the fold induction by this growth factor was at least 1.5 fold greater than that of EGF, compared with 18 mRNAs that were preferentially elevated by EGF. Likewise, 20 mRNAs were preferentially suppressed by NRG1β by at least a 1.5 fold margin compared to 10 mRNAs that were preferentially suppressed by EGF. Table 1 shows an abbreviated list of genes regulated by growth factors in MDA-MB-361 cells.
(A full list of growth factor-responsive genes may be found with the supplementary material).
Particularly noteworthy is the distribution of known immediate early genes (IEGs), indicated in  (38).
While NRG1β very efficiently induced morphological changes in these cells consistent with their differentiation (48) NRG2β had negligible activity in this assay (38,49). Paradoxically, the difference in the biological activities could not be explained by differences in receptor occupation or gross receptor tyrosine phosphorylation; NRG2 stimulated the tyrosine phosphorylation of the ErbB2 and ErbB3 receptors to the same extent as NRG1. However in examining individual signaling protein recruitment to activated receptors, differences became apparent. NRG1β was more potent than NRG2β in stimulating the recruitment of Grb2, Shc, SHP2 and p85 to ErbB2 and in stimulating the recruitment of Grb2 to ErbB3. The results of those previous studies are summarized in Table 2.
To further confirm differential signaling by NRG1β and NRG2β, we examined the association of a broad range of tyrosine-phosphorylated proteins with individual signaling proteins (Fig. 4). Because of the attenuated response of signaling pathways to NRG2β compared to NRG1β one might expect NRG2β to regulate fewer genes or to cause smaller responses in the same set of genes. Indeed, a subset of mRNAs, many of which were previously identified as immediate early genes, were preferentially elevated in response to NRG1β (Fig. 6B). However, a significant number of mRNAs were preferentially elevated in response to NRG2β compared to NRG1β, and  (Table 4).
Moreover, the response of a given gene to a single growth factor can be quite disparate depending on cellular context. Since in each case cells were treated with 30 nM NRG1β for one hour, and in each case signaling is mediated predominantly by an ErbB2/ErbB3 heterodimer, the response of genes to this growth factor may be directly compared. Ten of the nineteen overlapping mRNAs were elevated in response to NRG1β in both cell lines. These mRNAs were also responsive to the other growth factors, and nine of these ten have been previously identified as IEGs. In contrast, several mRNAs showed an opposite response to NRG1β in the two cell

Discussion
In this study we have used the ErbB receptor network to address the question of whether activation of different RTKs leads to identical or distinct profiles of gene expression. We first provide a surface-to-nucleus survey of signaling events triggered in response to two different EGF-like growth factors over the course of an hour after presentation to cultured human mammary tumor cells. Then using DNA microarray technology we provide a broad comparison of the gene transcripts that are induced or suppressed in the cells after an hour of growth factor treatment. It should be noted that while our expression studies focus on genes that respond with immediate early kinetics, we have not limited our analyses to cycloheximide-insensitive growth factor-induced genes. These relaxed criteria more fully reflect the breadth of the response to different growth factors, and also allow for an analysis of genes that are suppressed in response to growth factor treatment. It should also be noted that this is not intended to be a comprehensive analysis of genes that are regulated by EGF-like growth factors in mammary tumor cells, but rather a survey of genes that uncover trends in cellular responses to RTK activation.
We have utilized the MDA-MB-361 cells to activate distinct ErbB receptor heterodimers and downstream signaling pathways. We confirm that stimulation with different EGF-like growth factors in these cells results in different patterns of receptor activation, which in turn leads to the differential recruitment of SH2 and PTB domain-containing proteins to receptors and other tyrosine phosphorylated proteins. In these cells EGF stimulates the tyrosine phosphorylation of the EGF receptor and ErbB2, while NRG1β stimulates the tyrosine phosphorylation of ErbB2 and ErbB3. The similar level of recruitment of Grb2 to EGF receptor in response to EGF and to by guest on July 10, 2020 http://www.jbc.org/ Downloaded from Sweeney et al.,18 ErbB3 in response to NRG1β likely underlies the similar activation of the Erk1 and Erk2 kinases by the two growth factors. On the other hand, the preferential recruitment of p85 to ErbB3 in response to NRG1β translates into an increased activation of the Akt kinase by this factor. The preferential stimulation of PKC by EGF may be related to previous observations that PLC-γ is more strongly stimulated by EGF receptor activation than by ErbB3 activation (46,50,51) Although both growth factors stimulated the phosphorylation of the Myc, Jun and CREB transcription factors, EGF appeared to be reproducibly stronger in doing so. An analysis of the phosphorylation of other transcription factors such as the STATs and forkhead family members could uncover NRG1β-specific nuclear factors.
Most of the known IEGs were similarly stimulated by both growth factors in these cells, suggesting that these genes may be downstream of a common pathway. A strong candidate is Erk1/2-mediated transcriptional regulation through the serum response promoter element. Other genes that are similarly stimulated by both factors could be IEGs that have not previously been reported, or genes whose expression is regulated in response to IEG induction. The IEGs that preferentially respond to one of the growth factors may contain additional promoter elements sensitive to pathways specific to signaling by that factor. Although more kinase cascades were preferentially stimulated with EGF, NRG1β was the stronger growth factor in regulating gene expression. This might be explained by the preferential stimulation of Akt by NRG1β. Aktregulated genes are of acute interest because of the known role for PI3K/Akt signaling in mediating cell survival (52, 53). More work is needed to determine which pathways preferentially affect specific mRNA levels. These issues are currently being addressed using pharmacological and mutagenesis methods to suppress specific pathways.

Sweeney et al., 19
Our findings contrast those of a previous study that concluded that in immortalized mouse fibroblasts, different RTKs and diverse RTK-stimulated signaling pathways mediate the expression of primarily overlapping sets of immediate early genes (11). Whether the differences in conclusions arise from the complexity inherent in the ErbB signaling network, or whether fibroblasts are pre-programmed to respond to a variety of stimuli with the expression of a defined set of genes remains to be resolved.
We have utilized the MDA-MB-453 cells to activate the same ErbB receptor heterodimer with different growth factors, and the results point to another mechanism by which early gene expression may be regulated. In these cells, we found no major differences in the identity of the signaling cascades activated by the growth factors NRG1β and NRG2β. Instead we observed that each of the pathways was induced with abbreviated kinetics with NRG2β relative to NRG1β, possibly resulting from the suppressed ability of NRG2β to stimulate the recruitment of signaling proteins to ErbB2 (38). This in turn resulted in marked differences in gene expression, which may underlie the dramatic difference in biological potencies between the two factors (38,49). This interpretation is consistent with results from other systems where signaling strength or duration is thought to impact biological response. For example, overexpression of some RTKs in PC12 pheochromacytoma cells is sufficient to confer ligand-dependent differentiation (54, 55).
Interestingly, despite its dominance in stimulating signaling pathways and differentiation, In summary, using pathway profiling and global gene expression analysis methods we demonstrate that differences in signal transduction pathway usage by the ErbB RTKs precede differences in gene regulation by growth factors (Fig. 7). These results indicate that RTK signaling is not generic and that specific pathways or combinations of pathways may target specific genes. Hence, EGF-like growth factors have the potential to serve as more than a "go"     Lysates from NRG1β-or NRG2β-treated cells were blotted with antibodies specific for the indicated phosphorylated kinases, and the normalized signals plotted.