![[Feedback]](/icons/banner/feedbackACT.gif){kind=icon}
![[For Subscribers]](/icons/banner/subscriberACT.gif){kind=icon}
![[Archive]](/icons/banner/archiveACT.gif){kind=icon}
![[Search]](/icons/banner/searchACT.gif){kind=icon}
![[Contents]](/icons/banner/tocACT.gif){kind=icon}
|
|
|
|
|
||
kinases1 Department of Biochemistry, University of Lausanne, Epalinges CH-1066, Switzerland; 2 Cutaneous Biology Research Center, Massachusetts General Hospital, Charlestown, Massachusetts 02129, USA; 3 Laboratory of Cancer Pharmacogenomics, Fondo "Edo Tempia," Biella 13900, Italy; 4 Department of Surgery, Centre Hospitalier Universitaire Vaudois, Lausanne CH-1011, Switzerland; 5 Department of Dermatology, Centre Hospitalier Universitaire Vaudois, Lausanne CH-1011, Switzerland; 6 Department of Dermatology, Massachusetts General Hospital, Boston, Massachusetts 02114, USA; 7 Division of Cancer Biology and Tissue Engineering, Tufts University Dental School, Boston, Massachusetts 02111, USA
 |
Abstract |
|---|
 Top Abstract ResultsDiscussionMaterials and methodsAcknowledgmentsReferences |
|---|
kinases, two key effectors of small Rho GTPases previously implicated in neoplastic progression. Thus, the Notch1 gene is a p53 target with a role in human tumor suppression through negative regulation of Rho effectors.
[Keywords: Notch; p53; ROCK/MRCK; stem cells; squamous cell carcinoma; in vivo siRNA delivery]
Received August 22, 2006; revised version accepted January 18, 2007.
). The biological function of this pathway is critically dependent on context-specific interactions with other signaling pathways, with possibly opposite roles in growth/differentiation of different cell types, at different developmental stages and/or under normal versus pathological conditions.
The Notch gene family encodes evolutionarily conserved type 1 transmembrane receptors that are activated by ligand binding and proteolytic cleavage, with release of the Notch intracellular domain (Artavanis-Tsakonas et al. 1999). The best characterized "canonical" pathway of Notch activation involves translocation of this molecule to the nucleus, where it associates with the DNA-binding protein CSL (CBF-1 or RBP-J
in mammalian cells), converting it from a repressor into an activator of transcription (Mumm and Kopan 2000; Lai 2002). While the cytoplasmic domain of Notch can directly bind to CBF-1, binding of Notch to a second ancillary protein, Mastermind-like 1 (MAML1) or related family members, is required for elevated levels of CSL-dependent transcriptional activation (through recruitment of further transcription coactivators such as p300) (Petcherski and Kimble 2000; Wu et al. 2000; Oswald et al. 2001). A physical demonstration of a high-molecular-weight Notch–CBF-1–Mastermind complex has been obtained (Fryer et al. 2002) as well as an elucidation of its structural organization (Nam et al. 2006; Wilson and Kovall 2006). Importantly, a 51-amino-acid peptide (MAM51) corresponding to the highly conserved N-terminal region of the MAML1 protein has been shown to function as a specific suppressor of Notch-CSL transcription, by competing for MAML1 binding to the complex and thereby preventing downstream transcriptional activation (Weng et al. 2003, 2004; Maillard et al. 2004; Balint et al. 2005).
The best characterized targets of Notch/CBF-1/MAML-mediated activation are members of the HES and HERP families of basic helix–loop–helix (bHLH) transcriptional repressors (Iso et al. 2003). However, several other direct targets of Notch/CBF-1 transcription have been identified, which can be induced by Notch activation in a cell-type-specific manner. In particular, in mouse keratinocytes, the Notch pathway exerts growth-inhibitory (Rangarajan et al. 2001) and tumor-suppressing functions (Nicolas et al. 2003; Proweller et al. 2006) that can be attributed to CBF-1 (RBP-J)-dependent up-regulation of p21WAF1/Cip1 with its consequent suppression of specific Wnt family members (Devgan et al. 2005). However, in human keratinocytes, Notch1 activation leads to little or no increase of p21WAF1/Cip1 expression, and causes more long-term suppression of growth and induction of differentiation, which can be attributed only in part to a "noncanonical" mechanism involving suppression of p63 expression (Nguyen et al. 2006). Little is known of the program of gene expression under specific Notch/CBF-1/MAML control in human keratinocytes.
Another major pathway that has been linked to control of keratinocyte cell fate determination, stem cell potential, and/or differentiation is that triggered by small GTPases of the Rho/Rac and CDC42 family (Xu et al. 2003; Benitah et al. 2005; Dotto and Cotsarelis 2005; Wu et al. 2006). Twenty small GTPases have been identified, which can be grouped in five distinct families (Jaffe and Hall 2005). The "on" or "off" state of these molecules is controlled by a large number of GTP exchange factors (GEFs) as well as GTPase-activating proteins (GAPs), with a positive and negative regulatory function, respectively, in response to several cues. The GTPases in "on" configuration activate, by direct binding, several effectors, in particular, protein kinases, with a cascade of downstream signaling events impinging on a variety of cellular processes (Jaffe and Hall 2005). Increased Rho signaling has also been positively connected with development and/or progression of tumors of various origins (Sahai and Marshall 2002). Importantly, no genetic mutations have been uncovered that are responsible for activation of this pathway in tumors, unlike those found with ras. Instead, increased expression of the Rho A or C GTPases and/or of their ROCK1/2 effectors has been described in several tumors, through as yet uncharacterized mechanisms (Sahai and Marshall 2002; Gomez del Pulgar et al. 2005). The MRCK kinase, an effector of CDC42 with functional overlap with ROCK1/2, has also been suggested to have a role in tumor progression (Benitah et al. 2004) and/or invasion (Wilkinson et al. 2005).
An inverse relationship exists between induction of differentiation and tumor development. Consistent with its role in maintaining lymphocyte stem cell populations, the role of Notch signaling in promoting human T-cell leukemia is by now well demonstrated (Ellisen et al. 1991; Weng et al. 2004). A proto-oncogenic function of this pathway in human breast and ovarian carcinogenesis (Weijzen et al. 2002; Pece et al. 2004; Hopfer et al. 2005; Ayyanan et al. 2006) and in melanoma progression (Balint et al. 2005) is also emerging. In contrast, the possibility that Notch signaling plays an alternative tumor-suppressing function in human cancers of other types, where it can promote differentiation, has only been suggested on the basis of activated Notch1 overexpression studies (Linnoila et al. 2000; Shou et al. 2001; Sriuranpong et al. 2001; Talora et al. 2002; Qi et al. 2003). We report here that the Notch1 gene is a p53 target in human keratinocytes with a role in tumor suppression of this cell type through negative regulation of the ROCK1/2 and MRCK kinases.
|
Results |
|---|
|
TopAbstractResultsDiscussionMaterials and methodsAcknowledgmentsReferences |
|---|
Notch signaling promotes commitment of human keratinocytes to differentiation and restricts their growth potential (Nguyen et al. 2006). Relative to normal human primary keratinocytes, immunoblot and RT–PCR analysis showed that Notch1 protein and mRNA were reduced in a panel of skin and oral squamous cell carcinoma (SCC) cell lines (Fig. 1A,B). Notch2 expression was also decreased, although less consistently (Fig. 1C). A similar reduction of Notch1 expression was also found in a panel of skin SCCs relative to normal epidermis controls, with a parallel reduction in levels of the HES1 gene, a common target of the "canonical" Notch pathway (Fig. 1D,E). Reduced Notch1 expression was also confirmed by laser capture microdissection (LCM) and real-time RT–PCR analysis of additional skin SCCs versus surrounding normal epidermis (Fig. 1D, right panel).
|
). As shown in Figure 1F, many genes that are up- or down-regulated by Notch activation in keratinocytes exhibited an opposite pattern of expression in the tumors, including, as expected, HES1; other Notch targets like the OAS1 and IF144 interferon-responsive genes that we recently identified (Nguyen et al. 2006); and, as considered in detail further below, the ROCK2 and MRCK (CDC42BPA) kinase genes.
Given the very frequent alterations of the p53 pathway in skin SCCs (Backvall et al. 2004), an attractive possibility was that the down-modulation of Notch signaling in these tumors is linked, at least in part, to compromised p53 function. RT–PCR analysis of an established p53 target gene with a tumor-suppressing function, Btg2 (Boiko et al. 2006), showed that its expression in SCCs was generally decreased (Fig. 1G). Expression of another "classical" p53 target, p21WAF1/Cip1, was decreased in some SCCs but not in others (Fig. 1G), consistent with the complex regulation of this gene (Gartel and Tyner 1999) and its similarly reported heterogeneous expression in tumors (e.g., see Erber et al. 1997). For a more global analysis of endogenous p53 targets in keratinocytes, we knocked down p53 expression by small interfering RNA (siRNA) transfection of these cells. Genes up- or down-regulated as a consequence of p53 suppression in human primary keratinocytes were compared with the expression profiles of skin SCCs versus normal epidermis. This analysis showed a concordance of genes that are similarly deregulated in the two cases (Fig. 1H). Among these was p53 itself, whose levels were significantly reduced in the tumors that we analyzed, as well as the Notch1 and HES1 genes, and ROCK2.
To more directly assess to what extent p53 function is linked to control of Notch1 gene expression, we compared the effects of knocked-down versus increased p53 expression. Primary human keratinocytes were either transfected with p53-specific siRNAs, as before, or treated with nutlin, an HDM2 inhibitor that causes increased levels of endogenous p53 (Vassilev et al. 2004). Notch1 expression (at RNA and protein levels) was induced as a result of increased p53, while it was down-modulated in cells with p53 knockdown (Fig. 2A). Infection with a recombinant adenovirus expressing wild-type p53 caused also an up-regulation of Notch1 expression not only in human primary keratinocytes, but, to an even greater extent, in SCC13 and SCC028 SCC cells, consistent with the reported loss of endogenous p53 activity in these cells (Fig. 2B; Boyle et al. 1993; Zhao et al. 2006).
|
; van Es et al. 2005; Radtke et al. 2006). HCT116 is a well-characterized human colon carcinoma cell line with wild-type p53 (Bunz et al. 1998). In contrast to keratinocytes, knockdown of p53 in these cells did not down-modulate Notch1 expression (Fig. 2C), while conversely, increased p53 expression, by either adenoviral infection or nutlin treatment, led to no up-regulation (Fig. 2B,C).
Chromatin immunoprecipitation (ChIP) assays were used to assess whether the Notch1 gene is a direct target of p53 and whether there are differences in binding of the p53 protein to the Notch1 promoter in keratinocytes versus colon carcinoma cells. Nucleotide sequence analysis indicated the presence of two putative p53-binding sites in the Notch1 promoter, at position –3.55 kb as previously reported (Wei et al. 2006), and at –0.8 kb from the translational start site (Fig. 2D). ChIP assays of human primary keratinocytes showed binding of endogenous p53 protein to both sites, with no binding to an unrelated internal sequence (Fig. 2D). No positive binding signal was obtained by similar ChIP assays with a cancer cell line (PC3) that lacks p53 (Zhao et al. 2006; data not shown). Quantitative ChIP assays by real-time PCR confirmed the results with human keratinocytes and showed even higher levels of p53 binding in HCT116 cells (Fig. 2E), indicating that the differential regulation of the Notch1 gene by p53 in the two cell types occurs downstream from promoter recognition and binding.
Inhibition of Notch signaling in human primary keratinocytes promotes aggressive SCC formation
To evaluate the functional consequences of decreased Notch signaling, as observed in the clinically occurring SCCs, we infected primary human keratinocytes with a retroviral vector expressing a 51-amino-acid peptide (MAM51) corresponding to the N-terminal region (amino acids 13–74) of the MAML1 protein. This peptide competes for MAML1 binding to the Notch/CBF-1 complex, thereby preventing downstream transcription (Weng et al. 2003). As expected, the MAM51-expressing keratinocytes showed a significant reduction in "canonical" Notch signaling, as assessed by promoter activity assays with a Notch/CBF-1 reporter and real-time RT–PCR for HERP1, an established endogenous Notch1 target gene (Fig. 3A,B). Besides CBF-1-dependent transcription, Notch activation can also induce the NF-B pathway through a separate mechanism unlikely to involve MAML1 function (Nguyen et al. 2006; Shin et al. 2006). In fact, MAM51 expression had no effect on NF-B signaling in keratinocytes under either basal or differentiating conditions, or after expression of oncogenic ras (Fig. 3C,D).
|
|
-secretase cleavage at the cell membrane (Artavanis-Tsakonas et al. 1999). Chemical inhibitors of -secretase activity can suppress endogenous Notch activity and have been proposed as possible therapeutic agents for cancers involving Notch activation (Miele et al. 2006). Our findings suggested that these same agents could have an opposite tumor-promoting function in keratinocyte tumor development. To assess this possibility, primary human keratinocytes infected with an oncogenic ras-transducing retrovirus were tested by skin-reconstitution grafting assays, under conditions that allow formation, under normal conditions, of well-stratified epidermis (Dotto et al. 1989). Grafted mice were treated three times per week with either the -secretase inhibitor DAPT or DMSO vehicle alone. In control-treated mice, H-rasV12-expressing keratinocytes produced a hyperplastic epithelium with only limited sites of invasion (Fig. 4D, top panels). In contrast, in mice treated with the -secretase inhibitor, there was aggressive tumor formation comparable with that caused by H-rasV12- and MAM51-coexpressing keratinocytes. Even in these cases, widespread areas of spindle cell transformation, with concomitant positive staining for vimentin and keratin expression, were observed (Fig. 4D, bottom panels).
Inhibition of Notch signaling suppresses keratinocyte commitment to differentiation and expands populations with stem cell potential
MAM51-expressing keratinocytes in culture exhibited no increased proliferation that could explain their in vivo tumorigenic behavior (Supplementary Fig. 1). Rather than enhanced growth of the total cell population, the tumor-promoting effects of decreased Notch signaling may be connected with decreased commitment to differentiation and expansion of keratinocyte subpopulations with stem cell potential, the likely targets of carcinogenesis (Perez-Losada and Balmain 2003). This possibility was assessed by several complementary assays.
Loss of keratinocyte adhesion to the substrate triggers important features of terminal differentiation, including irreversible withdrawal from the cell cycle (Watt et al. 1988). Gene expression analysis showed a substantial induction of Notch1 gene expression, as well as HES1 and HERP1 (Fig. 5A), in keratinocytes after various times of suspension culture. This was associated with an induction of endogenous p53 target genes like p21WAF1/Cip1, Gadd45, and Btg2 (Fig. 5B). Both pathways were found to be functionally involved. In fact, after several hours of suspension culture, the fraction of keratinocytes that was able to reattach to the dish and resume growth was substantially increased by Notch suppression, via MAM51 expression (Fig. 5C) or DAPT treatment (data not shown). A protection against the irreversible commitment to differentiation was also observed with keratinocytes with p53 knockdown (Fig. 5D).
|
6
4) (Rangarajan et al. 2001) as well as Wnt ligand family members (Devgan et al. 2005), which are positively connected with keratinocyte proliferation. Upon induction of differentiation, keratin 1 and involucrin expression were up-regulated in the MAM51-expressing keratinocytes to a lesser extent than in controls (Fig. 5E). Conversely, integrin 6 levels were increased in MAM51-expressing keratinocytes rather than suppressed as in the controls (Fig. 5F); integrin 4 and Wnt7a levels were higher in MAM51-expressing keratinocytes already under growing conditions, and were less reduced with differentiation (Fig. 5F). In three-dimensional organotypic cultures, which recapitulate the normal vertical differentiation program occurring in vivo (Margulis et al. 2005), control keratinocytes produced a well-differentiated and stratified epidermis, while MAM51-expressing cells generated a lesser stratified epidermis with substantially reduced keratin 1 expression and up-regulation and aberrant suprabasal distribution of integrin 6 (Fig. 5G).
Clonogenic behavior of primary human keratinocytes provides a widely used assay for populations with high self-renewal potential (Rochat et al. 1994). In parallel with the above findings, after prolonged passaging, the fraction of keratinocytes with clonal growth capability was significantly higher in MAM51-expressing cultures than controls (Fig. 5H). This was accompanied by an expanded subpopulation with elevated integrin 6 (integrin 6bright) and decreased CD71 expression (CD71dim), which are associated with putative stem cells (Fig. 5I; Li and Kaur 2005).
ROCK1/2 and MRCK are negative Notch target genes that are required for tumor development
Little is known of the mechanisms involved in control of human keratinocyte stem cell potential by the "canonical" Notch/CBF-1 pathway. Another major pathway that has been recently linked to control of keratinocyte self-renewal and/or differentiation is that triggered by small GTPases of the Rho/Rac and CDC42 family (Benitah et al. 2005; Grossi et al. 2005; Wu et al. 2006). Microarray analysis suggested that genes for the ROCK2 and MRCK kinases, two major effectors of the RhoA and CDC42 GTPases, respectively (Ishizaki et al. 1996; Leung et al. 1998), are down-regulated by Notch activation in primary human keratinocytes with an opposite up-regulation in tumors (Fig. 1F). Real-time RT–PCR confirmed that expression of these genes is suppressed by activated Notch1 expression in human primary keratinocytes, and is similarly down-modulated by the "canonical" Notch effector HES1 (Fig. 6A). Several fully conserved HES-binding sites are present in the promoter region of the ROCK2 and MRCK genes (Supplementary Fig. 2). Antibodies suitable for immunoprecipitation of the endogenous HES1 protein for analysis are not available. However, HES1 binds to a specific transcriptional corepressor, Tle-1 (Stifani et al. 1992), which can be readily immunoprecipitated by existing antibodies. ChIP assays with these antibodies showed that the Tle-1 protein is indeed associated with a predicted high-affinity HES1-binding region of the ROCK2 promoter, while no such association was detectable at a low-affinity, more downstream region (Fig. 6B).
|
, ROCK2, and the closely related ROCK1 gene was substantially up-regulated in the MAM51-expressing keratinocytes, while that of 
Np63, which is controlled by Notch through a CBF1-independent mechanism (Nguyen et al. 2006), was not significantly modulated (Fig. 6C). Increased expression of ROCK1, ROCK2, and MRCK was also observed in primary human keratinocytes after DAPT treatment (Fig. 6D). In parallel with these findings, expression of one or more of these genes was up-regulated in the skin SCCs that we examined (Fig. 6E) as well as in the SCC-derived cell lines (data not shown). Up-regulation of ROCK2 was also found by immunohistochemical analysis of additional clinically occurring SCCs, with similar results with experimental tumors formed by the MAM51 and ras keratinocytes (Fig. 6F, top and bottom panels, respectively).
Our finding that p53 is an upstream positive regulator of Notch raised the interesting possibility that p53 may also control ROCK1/2 and MRCK expression. Human keratinocytes with knockdown p53 expression exhibited a substantial increase of these genes, while, conversely, increased p53 expression caused a down-modulation of ROCK1/2 and MRCK similar to that caused by activated Notch1 (Fig. 7, left and right panels, respectively). A block of endogenous Notch activity by DAPT did not counteract the suppressing effects of p53 on ROCK1 and 2 expression (Fig. 7D,E), while it partially prevented those on MRCK (Fig. 7F), indicating that p53 controls these genes in a manner that is parallel and, in the case of MRCK, partially dependent on the Notch pathway.
|
up-regulation, we "knocked down" expression of these genes by siRNAs (Fig. 8A). Consistent with their functional redundancy (Wilkinson et al. 2005), only limited effects were observed by individual knockdown of these genes (data not shown). In contrast, the combined knockdown of these kinases reduced the elevated levels of integrins 6 and 4 and Wnt7a expression of MAM51 keratinocytes (Fig. 8B), while enhancing differentiation of these cells (as assessed by keratin 1 expression) (Fig. 8C). In parallel with their emerging role in cell cycle control and transcription, the best characterized function of ROCK1/2 and MRCK kinases is in cell motility and invasion (Wilkinson et al. 2005). Consistent with our other findings, time-lapse microscopy showed a substantially increased mobility of MAM51-expressing versus control keratinocytes, such mobility being suppressed by ROCK1/2 and MRCK knockdown (Fig. 8D). In parallel, MAM51 expressing keratinocytes exhibited also a much higher capability to migrate across a membrane filter barrier than control keratinocytes (Fig. 8E).
|
). Using this approach, we treated mice, which were either injected subcutaneously or grafted with the H-rasV12/MAM51 keratinocytes, with combined siRNAs for the ROCK1/2 and MRCK kinases or scrambled siRNA control. As shown in Figure 9A, treatment with the specific siRNAs caused a significant reduction of tumor formation in several animals, with small and well-differentiated lesions or even, in the case of grafted cells, a hyperplastic epithelium without overt signs of malignancy. The results of this set of experiments were further confirmed with a substantially larger number of animals (eight mice per group) for statistical validation of the results. Treatment with the ROCK1/2 and MRCK siRNAs suppressed tumor growth over time (Fig. 9B), with a significant reduction of final tumor size (Fig. 9C).
|
|
Discussion |
|---|
|
TopAbstractResultsDiscussionMaterials and methodsAcknowledgmentsReferences |
|---|
). However, the importance of epigenetic mechanisms for down-modulation of negative growth regulatory genes and tumorigenesis is being increasingly recognized (Ting et al. 2006). We have shown here that expression of the Notch1 gene is significantly down-modulated in skin SCCs, in parallel with suppressed Notch activity. These findings are of likely functional significance as suppression of Notch signaling, by either genetic or pharmacological manipulations, is sufficient to elicit, together with oncogenic ras, oncogenic conversion of primary human keratinocytes.
Little is known of the upstream mechanisms controlling Notch1 expression. The p53 tumor suppressor gene is mutated or lost in a large fraction of human tumors, including skin SCCs (http://p53.free.fr) (Olivier et al. 2002), and, even when not mutated, p53 function can be compromised by secondary genetic and/or epigenetic events (Backvall et al. 2004). In human cell types where it promotes tumor formation, Notch signaling was reported to act upstream of p53, suppressing its function (Beverly et al. 2005; Mungamuri et al. 2006). In contrast, we have shown here that the Notch1 gene is a downstream positive target of p53 in both normal and cancer-derived keratinocytes. A similar regulation of Notch1 expression by p53 occurs in prostate cancer cells, another cell type in which increased Notch signaling causes growth inhibition with associated tumor-suppressing effects (K. Lefort, P. Ostano, G. Chiorino, and G.P. Dotto, in prep.). In contrast, no such regulation was observed in HCT116 colon carcinoma cells, in parallel with the proposed tumor-promoting function of Notch signaling in this context (Radtke et al. 2006). This differential regulation occurs downstream from endogenous p53 binding to the Notch1 promoter and is likely due to an interplay with other cell-type-specific determinants of gene expression. In other ongoing work, we have found that activity of a reporter with 2 kb of the Notch1 promoter is also under positive p53 control. However, together with the p53-binding region, we have identified another part of the Notch1 promoter, overlapping with the transcriptional start site, that is differentially active in primary keratinocytes versus cancer cells (C. Lambertini and G.P. Dotto, in prep.). It is also worth noting a putative p53-binding site that is highly conserved between mouse and human Notch1 promoters in their more upstream regions (at position –5339 from the ATG initiation site of the human gene). Thus, as with many other genes, transcriptional control of the Notch1 gene is likely to result from p53 acting in concert with other key regulatory proteins at both proximal and distant sites.
Loss-of-function experiments were required to establish a possible tumor-suppressing function of Notch in human cells. Expression of the MAM51 peptide from the Mastermind-like 1 protein was previously used to suppress Notch signaling and inhibit tumorigenic behavior of acute T-cell leukemia cells as well as melanomas (Weng et al. 2003; Maillard et al. 2004; Balint et al. 2005). In sharp contrast, we have found that this same approach increased dramatically oncogenic transformation of primary human keratinocytes by a H-rasv12 oncogene, with formation of SCCs with widespread areas of epithelial mesenchymal transformation and clear cell cancer formation resembling those of aggressive tumors. Similar tumor-promoting effects were caused by pharmacological suppression of endogenous Notch activity by a -secretase inhibitor, which, in other cellular contexts, has been proposed as a possible anti-neoplastic agent (Miele et al. 2006).
Previous studies have reported oncogenic conversion of human primary keratinocytes by concomitant expression of oncogenic H-rasv12 and suppression of NF-B, through a mechanism that promotes growth through TNF/JNK activity (Dajee et al. 2003; Zhang et al. 2004). Since Notch activation leads to induction of NF-B (Nickoloff et al. 2002; Nguyen et al. 2006; Shin et al. 2006), an attractive possibility was that the tumor-suppressing function of Notch in keratinocytes is mediated by NF-B. However, while suppressing "canonical" CBF-1-dependent transcription, MAM51 expression did not affect NF-B activity and had no direct growth-promoting effects on the total cell population. Telomerase activity in MAM51-expressing versus control keratinocytes was also not increased (data not shown). This is consistent with our other findings that Notch suppression, rather than causing an overall enhancement of growth, reduces the commitment of keratinocytes to differentiation with a relative increase, over time, of only a minor population of cells with high self-renewing clonogenic potential. As pointed out in a recent review (Merlo et al. 2006), acquisition of an "immortalized" cell phenotype is not necessarily a characteristic of the vast majority of cells in primary tumors. What is emerging instead is a consensus picture of "cancer stem cells," representing a minor, possibly slow-cycling population that gives rise to all other heterogeneous tumor cell populations. Besides deregulation of target stem cell populations, further genetic instability and mutational events are known to occur as tumors develop (Merlo et al. 2006). In the experimental situation that we have established, tumor formation is detectable as early as 2–3 wk after injection of cells; it will be interesting to determine whether specific cell populations and additional genetic events are selected over time, in particular in the serial tumor transplants that we have initiated.
We previously showed that in mouse keratinocytes, Notch signaling exerts a growth-inhibitory and tumor-suppressing function that can be attributed to up-regulation of p21WAF/Cip1 expression (Rangarajan et al. 2001; Devgan et al. 2005). However, in human keratinocytes, Notch1 activation leads to no increase of p21WAF/Cip1 expression and leads to down-regulation of p63 through a "noncanonical" CBF-1-independent pathway (Nguyen et al. 2006), which, as we have shown here, is unaffected by MAM51 expression. Nothing is known on the mechanisms whereby "canonical" Notch/CBF-1/MAML signaling restricts the growth potential of human keratinocytes and suppresses carcinogenesis. We have found that one such mechanism involves regulation of the ROCK1/2 and MRCK kinases, two key effectors of RhoA and CDC42 GTPases, respectively (Ishizaki et al. 1996; Leung et al. 1998). Expression of both kinases is down-modulated in keratinocytes by activated Notch1, while it is up-regulated by MAM51 expression as well as DAPT treatment, and it is increased in keratinocyte-derived SCC cell lines and tumors. The promoter regions of the ROCK2 and MRCK genes contain several fully conserved HES-binding sites, and their expression is suppressed by HES1, a classical Notch–CBF-1 target with transcription repressor function (Artavanis-Tsakonas et al. 1999), to the same extent as activated Notch1. Importantly, expression of these genes is also under negative control of p53, through a parallel and, in the case of MRCK, partially Notch-dependent mechanism. Biologically, ROCK1/2 and MRCK have been recently implicated in tumor progression (Benitah et al. 2004) and invasion (Wilkinson et al. 2005). By a gene knockdown approach, we have found that up-regulation of these genes is required for the effects of Notch suppression in keratinocytes, including deregulated integrin and differentiation marker expression and increased cell motility. More importantly, our findings that the tumorigenicity of MAM51- and ras-expressing keratinocytes is impaired by the knockdown of the ROCK1/2 and MRCK genes in vivo implicate these kinases as critical targets in control of human keratinocyte tumor formation.
|
Materials and methods |
|---|
|
TopAbstractResultsDiscussionMaterials and methodsAcknowledgmentsReferences |
|---|
Culturing of primary human keratinocytes (Nguyen et al. 2006), suspension cultures (Watt et al. 1988), organotypic (Margulis et al. 2005) cultures, and clonogenicity assays (Rochat et al. 1994) were as previously reported. SCC011, SCC012, SCC022, and SCC028 cells were provided by Dr. J. Rocco (Massachusetts General Hospital, Boston, MA), and SCC13 cells were provided by Dr. J. Rheinwald (Brigham and Women’s Hospital, Boston, MA), while other cells were from American Type Culture Collection (ATCC). LZRS-H-rasV12 (Dajee et al. 2003) and MSCV-MAM51 (Andriani et al. 2003) retroviruses were provided by Drs. P. Khavari (Stanford University, Stanford, CA) and J. Aster (Brigham and Women’s Hospital, Boston, MA), respectively. The Ad-p53 virus was from Dr. S. Lee (Massachusetts General Hospital, Charleston, MA), while the Ad-Notch1 virus was previously described (Rangarajan et al. 2001). Conditions for retrovirus and adenovirus production and infection were as previously reported (Rangarajan et al. 2001; Nguyen et al. 2006). The pRS-p53 vector was from Dr. Reuven Agami (The Netherlands Cancer Institute, Amsterdam, The Netherlands).
Cell motility and migration assays
Cell motility was measured by time-lapse microscopy, with cells under relatively sparse conditions in 60-mm wells, and analyzed on an inverted Axiovert 100 microscope in a temperature- and CO2-controlled chamber, taking pictures every 3 min over an 8-h period.
Cell migration was measured using a Matrigel Invasion Chamber kit (BioCoat and FluoroBlok; Becton Dickinson Bioscience) according to the manufacturer’s instructions. Cells (1 x 105/200 µL in DMEM–0.1% FCS) were added to the upper compartment and incubated for 24 h at 37°C. The lower compartment contained DMEM–5% FCS as chemoattractant. After removal of nonmigratory cells from the upper surface of the filter, cells that had passed to the lower surface were fixed and stained with the Diff Quick Stain Kit. Quantification was achieved by post-labeling of migrated cells with a fluorescent dye, followed by fluorescence intensity determination.
Subcutaneous tumorigenicity and keratinocyte grafting assays
For in vivo tumorigenicity assays, control and MAM51-expressing keratinocytes were infected with the LZRS-H-rasV12 retrovirus and, 24 h after infection, were brought into suspension, admixed with Matrigel (BD Biosciences), and injected (106 cells per injection) subcutaneously in NOD/SCID or nude mice. Tumors were dissected 4–6 wk later. Grafting assays were performed as previously described (Dotto et al. 1989). Briefly, keratinocytes were brought into suspension and injected (5 x 106 cells per injection) into a silicone transplantation chamber implanted onto the back of nude mice. For the in vivo DAPT treatment, this compound was injected subcutaneously at the site of grafting or, in some cases, directly into the transplantation chambers.
Skin SCC samples
Human SCC samples were obtained with Massachusetts General Hospital institutional approval as discarded material during Mohs micrographic surgery. For controls, normal skin was shaved very superficially with a #15 scalpel blade with minimal dermal contamination. Alternatively, samples of SCC and normal epidermis were frozen in OCT and used for LCM using an AutoPix Automated Laser Capture Microdissection System, followed by one round of linear RNA amplification as we described previously (Mammucari et al. 2005).
Quantitative real-time RT–PCR and microarray analysis
Conditions for RNA preparation, real-time RT–PCR, and microarray analysis were as previously described (Mammucari et al. 2005). The list of gene-specific primers is provided in Supplementary Table 3.
Immunodetection techniques
Conditions for immunohistochemistry (Rangarajan et al. 2001), immunoblotting (Lefort et al. 2001), and FACS analysis (Li and Kaur 2005) were as described. The list of antibodies used for these studies is provided in the Supplemental Material.
siRNA transfection and in vivo delivery
Cells were transfected as described (Nguyen et al. 2006) with 200 nM siRNAs for validated human ROCK1, ROCK2, and MRCK/ (Dharmacon) in parallel with nonspecific control duplex VIII (Dharmacon), or with validated stealth RNA interference (RNAi) Duopak for p53 (Invitrogen) and corresponding Stealth RNAi control. For in vivo delivery, 200 pmol of each siRNA (ROCK1, ROCK2, MRCK/) or 600 pmol of the nonspecific control duplex VIII were complexed with Jet-PEI (Polyplus) as described (Grzelinski et al. 2006) and injected i.p. into mice every 3 d for 4 wk.
ChIP assays
These assays were carried out as previously described (Nguyen et al. 2006) with the modifications indicated in the Supplemental Material. The primers of various regions of the human Notch1 promoter were 5'-AGCAGGGTACCCGCCACAGGA -3' and 5'-GACTCGGTGCAGGGACCATGG-3' (p53-A), 5'-ACAATCCGTCCGCGAGGCACT-3' and 5'-AGGATCGATC GCCGCCAGACA-3' (p53-B), and 5'-ATTTGGCAAGAGGT GTG-3' and 5'-AGGCTGGGAGGGACCTAGGAC-3' (intervening sequence C). The primers of various regions of the human ROCK2 promoter were 5'-CATGGACAAGATGCCAG GACG-3' and 5'-CAGGGTGGTGACTTCACGTAC-3' (HES-A), 5'-ATTGTGAGGACCCCGAGACAC-3' and 5'-ACTCCCCAG TTTGGATGTGCC-3' (HES-B), and 5'-GCTGAGAATGTAA ACGGGTAGCG-3' and 5'-ATTTTGCCTTGTTGCCCAGGC TG-3' (intervening sequence C).
|
Acknowledgments |
|---|
|
TopAbstractResultsDiscussionMaterials and methodsAcknowledgmentsReferences |
|---|
|
Footnotes |
|---|
