Context dependent effects of amplified MAPK signaling during lung adenocarcinoma initiation and progression

SUMMARY

Expression of oncogenic Kras^G12D initiates lung adenomas in a MAPK signal-dependent manner from only a subset of cell types in the adult mouse lung. Amplification of MAPK signaling is associated with progression to malignant adenocarcinomas, but whether this is a cause or consequence of disease progression is not known. To better understand the effects of MAPK signaling downstream of Kras^G12D expression, we capitalized on the ability of Braf inhibition to selectively amplify MAPK pathway signaling in Kras^G12D-expressing epithelial cells. MAPK signal amplification indeed promoted the rapid progression of established adenomas to malignant adenocarcinomas. However, we surprisingly observed a greater number of overall tumor-initiating events after MAPK signal amplification, due to induced proliferation of cell types that are normally refractory to Kras^G12D-induced transformation. Thus, MAPK signaling in the lung is thresholded not only during malignant progression, but also at the moment of tumor initiation.

Graphical abstract

INTRODUCTION

Lung adenocarcinoma is the most common subtype of lung cancer in the U.S. and accounts for approximately 40% of the 224,000 new cases of lung cancer that were predicted in 2016. KRAS is the most commonly mutated proto-oncogene in lung adenocarcinoma, and its activation likely occurs during the earliest stages of tumorigenesis (2014; Guerra et al., 2003; Jackson et al., 2001; Johnson et al., 2001). The molecular etiology of lung adenocarcinoma is diverse, and while personalized therapies for this disease have expanded greatly over the last decade, no targeted therapies currently exist for patients with KRAS mutations.

Somatic expression of oncogenic Kras from its endogenous locus initiates tumors in limited cell- and tissue-types (Guerra et al., 2003; Johnson et al., 2001; Mainardi et al., 2014; Sutherland et al., 2014). In the lung epithelium of the adult mouse, Kras^G12D -expression induces neoplastic transformation and the initiation of adenomatous lesions that can progress to resemble human lung adenocarcinoma (Guerra et al., 2003; Jackson et al., 2001). The cellular origin of these lesions has been extensively studied and depending on experimental contexts and the methodology used to activate oncogenic Kras expression, multiple cell types have been shown to give rise to adenocarcinoma in the lung (Kim et al., 2005; Rowbotham and Kim, 2014; Sutherland et al., 2014; Xu et al., 2012). However, alveolar type II (AT2) cells that reside in the alveolar spaces and express Sftpc (SPC) are the predominate cell of origin for lung adenocarcinoma (Sutherland et al., 2014; Xu et al., 2012). Less well understood are the involvement of rare cell types that are CC10/SPC dual positive. Cells positive for both these marks can be found commonly at the broncho-alveolar duct junction (BADJ) or more rarely in the alveolar spaces and appear to also be susceptible to transformation by oncogenic Kras (Kim et al., 2005; Sutherland et al., 2014; Xu et al., 2012). In contrast, although Scgb1a1 (CC10) -positive club cells that line the proximal bronchiolar airways can initiate tumor formation after activation of oncogenic Kras, compared to their SPC-positive counterparts, these cells are relatively recalcitrant to the oncogenic effects of Kras^G12D expression (Sutherland et al., 2014; Xu et al., 2014; Xu et al., 2012). These observations imply a heterogeneity of responses to Kras activation amongst different cell populations in the lung. However, it remains unclear whether the cells that do not initiate tumorigenesis upon Kras^G12D expression have activated failsafe tumor suppression programs to inhibit tumor formation, or if they are simply less sensitive to oncogenic Kras expression.

Several lines of evidence suggest that oncogenic Kras signaling through the Raf/Mek/Erk pathway (hereafter referred to as MAPK pathway), is necessary for both lung adenocarcinoma initiation and progression. Genetic deletion or pharmacological inhibition of the MAPK pathway limits tumor growth in Kras-driven models of lung adenocarcinoma (Blasco et al., 2011; Engelman et al., 2008; Trejo et al., 2012), and a Phase II study of the Mek1/2 inhibitor selumetinib observed a prolongation of progression-free survival in KRAS-mutant NSCLC (Janne et al., 2013). Although activation of endogenous oncogenic Kras expression and subsequent MAPK signaling in some lung epithelial cells is sufficient to initiate low grade lesions in the mouse, further amplification of the MAPK signaling pathway is associated with malignant progression (Feldser et al., 2010; Junttila et al., 2010). However, oncogenic Kras signals through multiple effector pathways that can impact tumorigenesis (Cully and Downward, 2008; Drosten et al., 2010; Drosten et al., 2014; Malumbres and Barbacid, 2003), a subset of which are required for efficient Kras-driven lung adenocarcinoma formation or tumor maintenance (Engelman et al., 2008; Kissil et al., 2007). Thus, it is unclear whether MAPK signaling per se can provide adequate cellular cues to initiate neoplastic transformation and whether the amplified MAPK signaling observed in advanced stage adenocarcinoma is the driver of tumor progression or simply the result of cellular state changes associated with malignancy.

Activating mutations of the BRAF proto-oncogene are a frequent driver event in melanoma, but also occur less commonly in a variety of other cancer types including lung adenocarcinoma (Davies et al., 2002). Vemurafenib is a potent and selective Braf inhibitor approved for the treatment of BRAF^V600E mutant melanoma (Bollag et al., 2010; Bollag et al., 2012). However, vemurafenib and other selective Braf inhibitors also amplify MAPK signaling downstream of oncogenic Kras or activated receptor tyrosine kinases (Hatzivassiliou et al., 2010; Heidorn et al., 2010; Poulikakos et al., 2010). This paradoxical effect promotes cell proliferation in experimental tumor models driven by oncogenic Ras and the onset of secondary RAS-driven malignancies in patients (Boussemart et al., 2016; Callahan et al., 2012; Su et al., 2012). Braf inhibitors thus represent an important but underutilized tool for studying the role of oncogenic MAPK signaling during tumor initiation and progression.

To better understand the impact of amplified MAPK signaling in lung adenocarcinoma, we utilized the vemurafenib-related compound PLX4720 to modulate the intensity of oncogenic MAPK signaling downstream of Kras^G12D in the mouse. We utilized well-characterized models of lung adenocarcinoma that rely on Cre-dependent activation of a Kras^LSL-G12D allele alone or in combination with p53 inactivation (Jackson et al., 2005; Jackson et al., 2001). We demonstrate that amplified MAPK signaling is sufficient to drive the histological progression of adenomas to malignant carcinomas. Additionally, we unexpectedly discovered that MAPK signal amplification promoted neoplastic transformation of lung cell types that did not respond initially to Kras^G12D activation, but initiated tumorigenesis rapidly upon MAPK signal amplification. We explored the cell of origin of these lesions using adenoviral vectors that drive cell-type restricted Cre expression and performed lineage tracing studies with fluorescent reporter alleles. We show that club cells residing in the proximal bronchiolar airway, which are relatively refractory to the transforming effects of endogenous oncogenic Kras expression, initiate tumorigenesis rapidly upon the specific amplification of MAPK signaling and contribute to tumor formation.

RESULTS

Amplification of MAPK signaling downstream of oncogenic Kras^G12D activation unexpectedly promotes tumor initiation

To control the onset of MAPK signal amplification downstream of oncogenic Kras, we first sought to determine whether the Braf-specific inhibitor PLX4720 could alter MAPK signaling within lung tumors initiated in Kras^LSL-G12D/+;p53^flox/flox mice. We initiated lung tumors in cohorts of Kras^LSL-G12D/+;p53^flox/flox mice via inhalation of adenoviral vectors expressing Cre from a ubiquitous promoter (Ad:CMV-Cre) (DuPage et al., 2009). One week post infection, mice were treated with either control or PLX4720-containing diets (Figure 1a). Kras^LSL-G12D/+;p53^flox/flox mice maintained on PLX4720 became moribund rapidly and succumbed to excessive lung tumor burdens significantly earlier than control cohorts (Figure 1b). As expected at this terminal endpoint, multiple advanced tumors demonstrating large, pleomorphic nuclei with a high degrees of nuclear atypia and aberrant mitotic figures were present in the lungs of PLX4720 treated mice. However unexpectedly, the lungs were overrun with alveolar and airway/bronchiolar hyperplasia (AAH) that likely contributed to the reduced survival (Figure 1c). Using a previously defined tumor grading scheme, we observed that the histological spectrum of tumors was significantly more advanced after PLX4720 treatment compared to control (Figure 1d) (Jackson et al., 2005; Nikitin et al., 2004). Additionally, the number of AAH lesions increased after PLX4720 treatments, but these differences were obscured by the coalescing pattern of their growth and did not achieve statistical significance (Figure 1c,d). Tumor cells in PLX4720 treated mice displayed a higher rate of proliferation and higher levels of phosphorylation specifically for Mek1/2 (pMek) and Erk1/2 (pErk), but not alternative key pathways important in Kras-mediated lung tumor initiation or maintenance (Figure 1e, Figures S1 and S2) (Castellano et al., 2013; Green et al., 2015; Molina-Arcas et al., 2013).

Figure 1.

MAPK signal amplification promotes early and widespread lung tumor growth in Kras^LSL-G12D/+;p53^flox/flox mice. (a) Kras^LSL-G12D/+;p53^flox/flox mice were infected with Ad:CMVCre by forced inhalation to initiate lung tumors. One week post Cre, treatment with control or PLX4720-containing diet was initiated and maintained for the indicated durations. (b) Kaplan-Meier survival analysis of Ad:CMVCre infected mice that were maintained on control or PLX4720 containing diets, p<0.0001, log-rank (Mantel–Cox) test. (c) H&E images of lung tumors and hyperplasias from mice at the survival endpoint. (d) Tumor grades in treated with control or PLX4720 for survival analysis, (contingency χ² test for trend, p<0.0001, Student’s t-test of mean number of AAH per mouse, p=0.08).(e) Immunohistochemistry for pMek and pErk five weeks post Ad:CMVCre. (f-g) Kras^LSL-G12D/+;p53^flox/flox mice were infected with 5×10⁷ or 5×10⁶ pfu of Ad:CMVCre as indicated. (f) Percent lung tumor burden after five weeks after Ad:CMVCre infection, p<0.0001. (g) Tumor number in lungs of mice five weeks after Ad:CMVCre infection, p=0.0001. (h) Whole lung lobes (upper panel) and lung sections (middle and lower panels) analyzed by fluorescence microscopy from Kras^LSL-G12D/+; p53^flox/flox; Rosa26^Motley/+ mice five weeks after Ad:CMV-Cre infection. Scale bars are 25 microns. Error bars represent SEM from average. Also see Figures S1 and S2.

These data suggest that MAPK amplification is able to more rapidly drive tumors to a higher histological state, while increasing the amount and number of early lesions normally present in the model. We therefore investigated the effects of PLX4720 at earlier time points post Kras^G12D activation. Five weeks post infection, we detected a dramatic increase in the lung tumor burden of PLX4720 treated mice. PLX4720 treated animals harbored approximately 8-fold higher tumor burden than control mice infected with an equivalent dose of Ad:CMV-Cre and similar tumor burden to controls when infected with 10-fold less Ad:CMV-Cre (Figure 1f). Again, MAPK amplification not only increased the proliferation rate of the resulting lesions, but also increased the overall number of hyperplasias and small adenomas that initiated. The number of tumors observed in PLX4720 treated mice was approximately 8-fold greater than controls and similar to the number of tumors that developed in mice that were infected with 10-fold less Ad:CMV-Cre (Figure 1g). To determine whether the increased number of tumors observed after MAPK signal amplification resulted from a greater frequency of tumor initiating events rather than the biased detection of larger, more rapidly proliferating PLX4720-treated lesions, we traced the lineages of individual tumor clones using Kras^LSL-G12D/+;p53^flox/flox;Rosa26^Motley/+ mice (Caswell et al., 2014). The Motley allele recombines stochastically to express one of three distinct fluorescent proteins in nascent Kras^G12D/+;p53^∆/∆ tumors. Fluorescent imaging of lung tissues revealed that tumor clones were both larger and more frequent in PLX4720 treated mice, and we observed many tumor clones growing in close proximity (Figure 1h). These data suggest that amplification of MAPK signaling downstream of Kras^G12D expression in lung epithelium not only promotes cellular proliferation, but unexpectedly increases the frequency of tumor initiation.

We next sought to determine whether the increased tumor initiation and tumor burden observed after wildtype BRAF inhibition in emerging Kras^G12D/+;p53^∆/∆ hyperplasia was dependent on MAPK signaling. One week after infection with Ad:CMV-Cre, mice were treated with control, PLX4720 alone, or PLX4720 plus the Mek1/2 inhibitor PD0325901 (Figure 2a). Consistent with our initial results, relatively few lesions were detected in control treated Kras^LSL-G12D/+;p53^flox/flox mice five weeks after tumor initiation, while PLX4720 treatment significantly increased tumor number and tumor burden. However, Mek inhibition completely abrogated the enhanced tumor initiation and increased tumor burden elicited by PLX4720 treatment (Figure 2b-d). While PLX4720/PD0325901 treatment led to the loss of pErk staining in the small hyperplasias that could be detected, as expected pMek staining remained elevated compared to control tumors (Figure 2e). MAPK signal amplification was also apparent in lung tissue lysates after PLX4720 treatment and was significantly diminished by Mek inhibition (Figure 2f). This observation is consistent with PLX4720 activating pathway signaling upstream of Mek1/2, while PD0325901 treatment abrogates signaling downstream of Mek1/2. These results suggest that the tumor promoting effects of Braf inhibition are Mek dependent.

Figure 2.

Enhancement of tumor growth from pharmacological MAPK signal amplification is Mek1/2 dependent. (a) Kras^LSL-G12D/+;p53^flox/flox mice were infected with 5×10⁷ pfu of Ad:CMVCre to initiate tumor formation. One week post infection, control, PLX4720 (Braf inhibitor), or PLX4720/PD0325901 (Braf inhibitor/Mek1/2 inhibitor) treatments were maintained for 4 weeks. (b) H&E images of whole lung lobes (upper panel) and high magnification images (lower panel) from areas showing hyperplasias. Percent lung tumor burden (c) and tumor number (d) analysis five weeks after Ad:CMVCre infection. (e) Immunohistochemistry for pMek and pErk in mice five weeks post-Ad:CMVCre. (f) pErk, total Erk (tErk), and Hsp90 expression in whole lung lysates of normal, control, PLX4720, or PLX4720/PD0325901 treated mice t=1-5 weeks. Scale bars are 25 microns. Error bars represent SEM from average. Also see Figure S3.

To further confirm the on-target effect of PLX4720, and because PLX4720 has not previously been shown to limit the oncogenic effects of Braf^V600E expression in autochthonous lung adenocarcinomas, we modeled the effects of PLX4720 in Braf^CA/+;p53^flox/flox mice (Figure S3a) (Dankort et al., 2007; Shai et al., 2015). As expected, activation of Braf^V600E expression and p53 deletion lead to rapid onset of large adenomas and adenocarcinomas six weeks post infection (Figure S3b). However, PLX4720 treatment beginning one week post infection nearly eliminated detectable tumor growth when measured at either two or six weeks after initiation (Figure S3b,c). Consistent with other tumor model systems, these results demonstrate that in the same tumor type, PLX4720 limits the oncogenicity of Braf^V600E, whereas in the context of wildtype Braf in Kras^LSL-G12D/+;p53^flox/flox mice, PLX4720 promotes oncogenic MAPK signaling downstream of Kras^G12D.

MAPK signal amplification promotes Kras^G12D-driven lung adenoma progression

Amplified MAPK signaling has been previously observed to be a common characteristic of many high grade Kras^G12D-driven lung tumors (Feldser et al., 2010). However, because copy number gains of oncogenic Kras alleles are a frequent event in both human and mouse lung adenocarcinomas, it is unclear whether it is the specific amplification of the MAPK pathway signaling, additional genetic alterations selected during tumorigenesis, or one of the many other oncogenic pathways downstream of activated Kras that drives malignant progression (2014; Cully and Downward, 2008; Malumbres and Barbacid, 2003). To control the onset of MAPK signal amplification downstream of oncogenic Kras in established tumors, we allowed adenomas to form in Kras^LSL-G12D/+; p53^flox/flox mice before treating with PLX4720. At seven weeks post tumor initiation, a time point when most tumors have not naturally amplified MAPK signaling nor progressed to be adenocarcinoma, we initiated two weeks of PLX4720 treatment, then assessed MAPK signaling status within lung tumors by immunohistochemical staining and western blotting (Figure 3a-c)(Robles-Oteiza et al., 2015). While tumors from control mice rarely had high levels of pMek or pErk staining, we observed widespread and strong staining of pMek and pErk in tumors from PLX4720 treated mice (Figure 3b). Consistent with MAPK signal amplification promoting cell proliferation, PLX4720 treatment also increased the percentage of Nkx2-1 positive lung tumor cells that were labeled with BrdU, and significantly increased tumor burden (Figure 3d-f)

Figure 3.

MAPK amplification drives cell proliferation and tumor progression in established p53-deficient tumors. (a) Kras^G12D/+;p53^flox/flox mice are infected Ad:CMVCre to initiate tumor formation. Starting 4 or 7 weeks post-Cre mice were treated with either control or PLX4720 until 9 weeks. (b) H&E, pMek, pErk, and Arf staining of lung tumors from mice on control or PLX4720 t=7-9 weeks. (c) Western blot analysis of lysates from individually isolated tumors for pErk, tErk, and Hsp90, mice treated t=4-9 weeks. (d) BrdU and Nkx2-1 lung tumor staining from control or PLX4720 treated mice. Mice were injected with BrdU at days 6 and 13 after starting control or PLX4720 treatments. (e) Quantification of BrdU and Nkx2-1 dual positive cells from mice on control or PLX4720 treatments from t=7-9 weeks, p<0.0001. (f) Percent lung tumor burden, p=0.026 and (g) tumor grades in mice treated from t=7-9 weeks with control or PLX4720, (contingency χ² test for trend, p=0.0002). Scale bars are 100 microns. Error bars represent SEM from average. Also see Figure S4.

To determine whether amplified MAPK signaling induced by acute two week PLX4720 treatment altered progression of lung adenomas, we assessed the expression status of the Arf tumor suppressor. Our previous study demonstrated that increased MAPK signaling was coincident with the acquired expression of the Arf tumor suppressor and could mark tumors that had progressed to higher grade (Feldser et al., 2010). We found consistently elevated Arf expression in tumors from mice treated with PLX4720 that demonstrated intense MAPK signaling, and no evidence of Arf expression in tumors in control mice or in normal tissue of either cohort (Figure 3b, Figure S4a). Furthermore, histopathological staging of these tumors, based on previously described guidelines (Jackson et al., 2005; Nikitin et al., 2004), revealed a skewed distribution in PLX4720 treated mice which predominantly displayed grade 3 and 4 tumors, in contrast to control mice, whose tumors were normally distributed with a majority of grade 2 or 3 tumors (Figure 3g). Indicative of their early carcinoma status, all tumors, regardless of control or PLX4720 treatment, were Nkx2-1 positive and Hmga2 negative (Figure S4a) (Winslow et al., 2011). Interestingly, in addition to MAPK amplification driving established tumor progression of these established tumors, we also observed an increase in lung hyperplasia (Figure S4b,c). This increase in alveolar hyperplasia observed upon delayed amplification of MAPK signaling is consistent with the expansion of dormant cells that were recalcitrant to transformation despite expressing Kras^G12D. Taken together, these results demonstrate the ability of PLX4720 to cooperate with oncogenic Kras^G12D to drive MAPK signal amplification, enhance rates of cellular proliferation, promote lung adenocarcinoma progression, and awaken dormant oncogene expressing cells to initiate tumorigenesis.

p53 restricts MAPK-mediated carcinomatous progression

To determine if amplified oncogenic MAPK signaling downstream of Kras^G12D can drive tumor progression in the presence of normal p53 function, we induced lung tumors in Kras^LSL-G12D/+;p53^+/+ mice. One week post infection with Ad:CMV-Cre, we initiated treatment with control or PLX4720 and analyzed animals five or eight weeks post infection (Figure 4a). Ad:CMV-Cre infected Kras^LSL-G12D/+;p53^+/+ mice that were treated with PLX4720 had a dramatic expansion of tumor cells throughout the lung compared to control mice (Figure 4b). The lungs of PLX4720 treated mice were predominated by tumor with an average of 74% of the lung area covered by tumor cells compared to <10% in control mice (Figure 4c). Consistent with this massive expansion of tumor, the tumor cells were significantly more proliferative after PLX4720 treatment (Figure 4d and Figure S5a). Surprisingly, PLX4720 treatment led only to a modest, but consistent increase in pErk levels within the tumors, which contrasted with the robust induction of pErk observed in p53 deficient tumors treated with PLX4720 (Figures 1e and 4e). To better investigate the levels of MAPK amplification induced by PLX4720 in the presence or absence of p53, we analyzed lung lysates from Ad:CMV-Cre treated Kras^LSL-G12D/+;p53^+/+ and Kras^LSL-G12D/+;p53^flox/flox mice following control or PLX4720 treatments. Consistent with our immunohistochemical staining, PLX4720 treatment increased MAPK signaling in lungs from both cohorts of mice, but the levels of MAPK signaling is restricted by the presence of p53 (Figure 4f,g). The diminished activation of the MAPK pathway in the presence of wildtype p53 suggests that a relatively small degree of MAPK signal amplification is sufficient to drive cell proliferation, but that high levels of MAPK signal amplification might surpass a signal threshold that activates a p53 response that would induce tumor suppressive programs.

Figure 4.

MAPK signal amplification leads to massive expansion of low grade tumors despite wildtype p53. (a) Kras^LSL-G12D/+;p53^+/+ mice infected with Ad:CMV-Cre were treated one week later with control or PLX4720. (b) H&E images of whole lung lobes 8 weeks after Ad:CMV-Cre infection. (c) Percent lung tumor burden analysis in mice 8 weeks after Ad:CMV-Cre infection. (d) Quantification of phosphorylated-Histone 3 (p-H3) positive cells. (e)) H&E and immunostaining for pMek and pErk of lung sections 5 weeks after Ad:CMV-Cre infection. (f) pErk, tErk, and Hsp90 expression from whole lung lysates of Kras^LSL-G12D/+ and Kras^LSL-G12D/+;p53^flox/flox mice treated with control or PLX4720 from t=1-5 weeks. (g) Quantification of MAPK signaling in lysates from whole lung lobes of Kras^LSL-G12D/+;p53^+/+ and Kras^LSL-G12D/+;p53^flox/flox mice expressed as a ratio of normalized pErk to tErk for each sample. (h) Tumor grades in Kras^LSL-G12D/+;p53^+/+ mice 8 weeks after Ad:CMV-Cre infection. Cumulative tumor number score is indicated. Average number of lesions per mouse was significantly higher in the PLX4720 group (p=0.02), but tumor grade distribution was not different (contingency χ² test for trend p=0.6968). Scale bars are 100 microns. Error bars represent SEM from average. Also see Figure S5.

To determine whether p53 restricts the progression of Kras^G12D-driven lung adenomas with amplified MAPK signaling, we assessed the histopathology of tumors in control and PLX4720 treated Kras^LSL-G12D/+;p53^+/+ mice. Lesions in PLX4720 treated mice were predominantly diffuse alveolar/bronchiolar hyperplasia. The coalescing pattern of hyperplasias and neoplastic tumors in PLX4720 treated mice made it difficult to distinguish individual lesions. As a result, we conservatively estimated the number of tumors per mouse that displayed characteristics of each tumor grade. Although the overall number of tumors detected was significantly greater in PLX4720 treated mice, the distribution of tumor grades did not differ from controls (Figure 4h). We also observed a marked increase in the incidence of alveolar hyperplasias centered on the airways and bronchiolar hyperplasias after PLX4720 treatment (Figure S5b). These data support a model in which even a relatively small degree of MAPK signal amplification significantly increases the frequency of Kras^G12D-driven tumor initiation and increases the rate of proliferation in lesions, but cannot drive malignant progression in the presence of a p53-regulated threshold.

Amplification of MAPK signaling expands lung adenocarcinoma cell of origin

SPC-expressing AT2 cells were previously identified as the predominant cell of origin for Kras^G12D-induced lung adenocarcinoma. In contrast, CC10-expressing club cells in proximal bronchiolar airways, appear to be relatively insensitive to Kras^G12D-induced transformation (Kim et al., 2005; Sutherland et al., 2014; Xu et al., 2014; Xu et al., 2012). However, because amplification of MAPK signaling in combination with Kras^G12D expression lead to increased numbers of tumors surrounding proximal bronchiolar airways, where AT2 cells typically do not reside, we sought to determine the cell type of origin of these additional PLX4720-induced tumors. We therefore utilized adenoviral vectors expressing Cre recombinase under a CC10- or SPC- specific promoter to initiate tumors in a cell-type restricted manner (Sutherland et al., 2011; Sutherland et al., 2014). Addition of the Rosa26^LSL-YFP allele further permitted the identification and lineage tracing of infected cells. We infected Kras^LSL-G12D/+;p53^flox/flox;Rosa26^{LSL-YFP/LSL-YFP} mice with either Ad:SPC-Cre or Ad:CC10-Cre and began control or PLX4720 treatments one week post infection (Figure 5a). Five weeks post infection, we analyzed lung tumor growth in each group of mice. Consistent with previous reports, in control mice tumors initiated by Ad:SPC-Cre were numerous and localized predominantly to the alveolar spaces, whereas tumors initiated with Ad:CC10-Cre were less frequent and localized mainly in the distal airways near the bronchioalveolar duct junction (Figure 5b, Supplementary Figure 6a)(Sutherland et al., 2014; Xu et al., 2014). PLX4720 treatment profoundly increased tumor burden in both Ad:SPC-Cre and Ad:CC10-Cre cohorts five weeks post infection and increased pErk levels specifically in YFP-positive cells in the alveolar space or lining the airways, respectively. (Figure 5b and Figure S6b). Interestingly, early Kras^G12D-expressing bronchiolar hyperplasias induced by PLX4720 treatment harbored a mixture of proliferating YFP-positive and YFP-negative cells (Figure S6c). This observation suggests that MAPK amplification in Kras^G12D-expressing CC10-positive cells has non-cell autonomous proliferative effects on adjacent epithelial cells in the bronchiolar airway. However, it is unclear whether these YFP-negative epithelial cells might contribute to tumor initiation and adenoma establishment, because all progressing lesions eventually became uniformly YFP positive.

Figure 5.

MAPK signal amplification promotes tumor initiation in AT2 and club cells. (a) Kras^G12D/+;p53^flox/flox;Rosa26^LSL-YFP/+ mice were infected with Ad:SPC-Cre or Ad:CC10-Cre to initiate tumor formation. One-week post infection, control or PLX4720 treatments began and lung tissue was analyzed four weeks later. (b) Immunohistochemistry of serial lung sections for YFP and pErk. Scale bars are 50 microns. (c) Immunofluorescence for YFP (yellow). Scale bars are 100 microns. (d) Contingency tables of YFP-positive singlet cells from (c). Significance indicated, fisher’s exact test. Also see Figure S6.

To better assess whether MAPK amplification downstream of Kras^G12D expression promotes the expansion of cells that are otherwise indifferent to oncogenic Kras expression, we identified YFP-positive cells by immunofluorescence analysis, and determined the number of cells in each lesion. We observed only small YFP-positive lesions that typically consisted of less than five cells in control mice five weeks after tumor initiation. We also frequently observed single YFP positive cells (singlets) in control mice suggesting that these cells failed to proliferate after Kras^G12D induction (Figure S6d). As expected, these singlets and small lesions were predominantly localized to the alveolar space in Ad:SPC-Cre infected mice and to the airways in those infected with Ad:CC10-Cre (Figure 5c). MAPK signal amplification after PLX4720 treatment had only a small, albeit statistically significant, effect on the expansion of small lesions in mice infected with Ad:SPC-Cre. In contrast, Ad:CC10-Cre infected mice displayed both an increase in the number of large lesions (5 or more cells) after treatment with PLX4720, as well as a reduction in the number of singlet lesions (Figure 5d and Figure S6d). These data suggest that Kras^G12D-expressing CC10-positive club cells within the proximal airways are quiescent, but are a competent source of tumor initiating cells upon MAPK signal amplification.

CC10-positive lesions originating in the proximal bronchiolar airways transdifferentiate toward an AT2 cell fate after MAPK signal amplification

Expression of Kras^G12D in cells at the terminal airway initiates tumor growth at relatively low frequency in cells dual positive for CC10 and SPC (Kim et al., 2005; Sutherland et al., 2014; Xu et al., 2012). As these lesions expand, they transdifferentiate toward an AT2 cell fate losing CC10 (Kim et al., 2005; Sutherland et al., 2014; Xu et al., 2014; Xu et al., 2012). Because our data suggests that MAPK amplification promotes increased transformation of CC10-positive club cells in more proximal airways, we determined whether these lesions similarly transdifferentiate toward an AT2 cell fate after MAPK amplification, or whether they maintain expression of CC10 and Sox2 which typify club cells. To address this question, we analyzed lungs from Kras^G12D/+;p53^flox/flox;Rosa26^{LSL-YFP/LSL-YFP} mice infected with Ad:SPC-Cre or Ad:CC10-Cre followed by control or PLX4720 treatment. Five weeks after Ad:Cre delivery, YFP-positive lesions were evaluated for their expression of SPC and CC10. Both hyperplasias and small adenomas arising in Ad:SPC-Cre infected mice were uniformly SPC-positive and CC10-negative and MAPK amplification via PLX4720 treatment did not alter this expression pattern (Figure 6a,b). Interestingly, despite their CC10-expressing origin, the YFP-positive adenomas that formed by this time point in Ad:CC10-Cre infected mice were also consistently SPC-positive, CC10-negative, and Sox2-negative. These adenomas were typically located near the BADJ at the distal terminus of the airway epithelium and protruded into the alveolar space (Figure 6a-c). However, the hyperplastic lesions within the more proximal airway epithelium that are specifically observed after MAPK amplification were positive for both SPC and CC10 (Figure 6a,b). These hyperplastic lesions were also positive for club cell marker Sox2 (Figure 6c). However, more advanced adenomas arising off the proximal bronchiolar airways in PLX4720-treated mice displayed a transitional staining pattern in which YFP-positive tumor cells within the airway epithelium retain Sox2 expression, but cells extending into the adjacent alveolar space became Sox2-negative (Figure 6c,d). Overall, the proportion of YFP-positive, Sox2-negative lesions arising in Ad:CC10-Cre infected Kras^G12D/+;p53^flox/flox;Rosa26 ^{LSL-YFP/LSL-YFP} mice is significantly enriched after MAPK signal amplification (Figure 6d). These data indicate that proximal airway club cells, although intrinsically indifferent to the effects of oncogenic Kras signaling, are induced to proliferate, transdifferentiate toward a AT2 cell state, and form tumors upon amplification of MAPK signaling downstream of oncogenic Kras^G12D expression.

Figure 6.

MAPK signal amplification drives transdifferentiation of club cells toward AT2 cell fate. Experimental scheme is identical to Figure 5a. (a,b) Immunostaining of lung sections from Kras^LSL-G12D/+;p53^flox/flox;Rosa26^LSL-YFP/+ mice; DAPI (white), YFP (yellow), CC10, (cyan), and SPC (red). Pink box indicates SPC-positive adenoma adjacent to CC10-positive club cells in the airway. Green box highlights hyper-proliferation in the airway positive for SPC and CC10. (c) Immunostaining for YFP (yellow), Sox2 (cyan), and DAPI (white) in the airways and adenomas. Hyperplasia in the airway demonstrates YFP-positive cells that have lost Sox2 expression (white arrows). Pink arrow highlights YFP-positive neoplasia retaining Sox2 expression along airway transitioning to Sox2-negative adenomas in the alveolar space. (d) Quantification of YFP+ lesions based on Sox2 expression status, Fisher’s exact test p<0.0001. Scale bars are 100 microns. (e) Model: MAPK signal amplification alters tumor initiation and progression in lung epithelial cell types. See text for details.

DISCUSSSION

Our results provide insight into the cell type- and context-specific crosstalk between the Kras, MAPK, and p53 pathways during lung adenocarcinoma initiation and progression. Our data demonstrate that a major impediment to transformation in the lung is the relative proclivity of the initiating cell to respond to oncogenic MAPK signaling downstream of Kras^G12D activation. By modulating the intensity of only a single arm of oncogenic Kras signaling, we establish that MAPK signaling is distinctly thresholded in different lung epithelial cell types. By governing cellular responses to oncogenic cues and regulating the probability of tumor initiation after oncogenic Kras activation, these thresholds act as a major determinant of transformation, growth, and lung tumor progression. Subsequent to neoplastic initiation, further progression of early stage adenomas is limited by the relatively low levels of MAPK signaling induced by endogenous Kras^G12D expression, but can be overcome by additional amplification of MAPK signaling. However, this enhanced tumor progression and MAPK signaling is opposed by the presence of wildtype p53 expression. Our data, therefore, suggest that the oncogenic effects of MAPK signaling are thresholded in at least two contexts: first in a cell type-specific manner during tumor initiation, and second during the p53 regulated adenoma-to-carcinoma transition.

Cellular resistance to Kras-mediated transformation has been observed in many cell types, including club cells, but the molecular basis of this phenomena has not previously been fully identified (Guerra et al., 2003; Xu et al., 2012). Our data suggest that, given the appropriate molecular cues, CC10-positive club cells in proximal bronchiolar airways and, to a lesser extent, CC10 and SPC dual positive cells at the broncho-alveolar duct junction can be a major source of tumor initiating cells after Kras^G12D expression. Tumors arising from club cells in the proximal airway are initially CC10 and Sox2 positive, but ultimately lose these markers upon expansion to the adenomatous state where they begin to express the AT2 cell marker SPC (Figure 5). Work by Xu et al. (Xu et al., 2014) and Sutherland et al. (Sutherland et al., 2014) posited that cells more distally positioned in the adult airway have a greater ability to initiate tumorigenesis in response to oncogenic Kras, and that Kras activity can drive the transdifferentiation of these distally located CC10-positive club cells toward an AT2 cell fate. Our data extend these models, and further suggest that lung adenocarcinoma initiation is governed by MAPK signal thresholds in CC10-positive cells in the proximal bronchiolar airway that are distinct from the oncogenic thresholds of CC10-positive cells at the bronchiolar terminus, both of which can be overcome through MAPK signal amplification.

Extensive study of the cell of origin in Kras-driven lung adenocarcinoma suggests that SPC-expressing AT2 cells are the predominant initiating cell type in these tumors under common experimental conditions (Reviewed in Rowbotham and Kim, 2014). However, hyperplasias have also been observed after Kras^LSL-G12D activation in rare CC10 and SPC dual positive cells (Kim et al., 2005; Xu et al., 2012). These dual positive cells either reside within the airway epithelium at the broncho-alveolar duct junction or in the alveolar spaces proximal to airways. Although we readily detected these dual positive cells at the broncho-alveolar duct junction in our study, we failed to identify any AT2 cells expressing CC10. While experimental models utilizing CC10-CreER transgenes have successfully identified CC10-positive AT2 cells as the cell of origin in Kras induced lung adenocarcinoma, others that rely on adenoviral delivery of CMV-Cre or CC10-Cre to lung epithelial cells have failed to do so (Jackson et al., 2001; Kim et al., 2005; Sutherland et al., 2014). As a result, we were likely limited in our ability to detect this CC10-positive AT2 cell population. However, it would be of great interest to investigate whether these CC10/SPC dual positive AT2 cells differentially respond to MAPK signal amplification using the appropriate model system.

We and others have previously demonstrated that high levels of MAPK signaling correlate with malignant progression in Kras^G12D-expressing, p53-deficient tumors. Additionally, restoration of p53 gene function in established tumors selectively invoked anti-cancer responses in high grade tumor cells, suggesting that high MAPK levels are incompatible with p53 expression (Feldser et al., 2010; Junttila et al., 2010). However, whether increased levels of MAPK signaling directly contributed to tumor progression or were simply a by-product of the transition to a malignant state was unclear. Our data now show that lung adenomas rapidly advance to low-grade carcinomas after MAPK signal amplification, but only in the absence of p53 expression (Figures 3 and 4). Thus, our data demonstrate that MAPK signal amplification is indeed the driving force behind lung adenocarcinoma progression and confirm that amplified MAPK signaling and tumor progression is constrained by p53. Intriguingly, inflammatory stimuli such as viral infection, tobacco smoke, or chemically-induced injury can also promote tumor initiation and progression after Kras^G12D expression in the lung (Mainardi et al., 2014; Moghaddam et al., 2009; Takahashi et al., 2010) or pancreas (Daniluk et al., 2012; Gidekel Friedlander et al., 2009). Future studies are required to determine whether these effects are mediated by inflammation-induced increases in MAPK signaling, and whether patients suffering from chronic inflammatory conditions may benefit from treatment with MAPK pathway inhibitors. In addition, our work also suggests the possibility that Braf inhibition could potentiate the effects of p53-activating therapies and may represent a unique strategy to increase their therapeutic efficacy.

EXPERIMENTAL PROCEDURES

Animal studies and treatments

Animal studies were performed under strict compliance with Institutional Animal Care and Use Committee at University of Pennsylvania. Kras^LSL-G12D (Jackson et al., 2001), p53^flox/flox (Jackson et al., 2005), p53^XTR/XTR (Robles-Oteiza et al., 2015), Braf^CA/+ (Dankort et al., 2007), Rosa26^FlpO-ER (Lao et al., 2012), Rosa26^{LSL-Luciferase} (Safran et al., 2003), Rosa26^LSL-YFP (Srinivas et al., 2001), and Rosa26^LSL-Motley (Caswell et al., 2014) mice have previously been described. Mice were administered Ad:CMV-Cre (Ad5CMVCre) as noted at 5×10⁶ or 5×10⁷ pfu/mouse, Ad:SPC-Cre (Ad5SPCCre) at 2×10⁸ pfu/mouse, or Ad:CC10-Cre (Ad5CC10Cre) obtained from University of Iowa Viral Vector Core by inhalation as previously described to initiate tumor formation (DuPage et al., 2009; Sutherland et al., 2014). Rodent Lab Diet (AIN-76A) was formulated with PLX4720 (Plexxikon) at 417 mg/kg (Tsai et al., 2008) and PD0325901 (Plexxikon) 7 mg/kg (Trejo et al., 2012). Mice were maintained on the indicated diets for the described time periods. BrdU incorporation was performed by injecting mice at 30 mg/kg (Sigma B5002) i.p. one week and 12 hours prior to necropsy. Mice in survival studies were monitored for lethargy, labored breathing, and weight-loss, at which time animals were euthanized. Whole lung lobe live fluorescent imaging on Rosa26^Motley mice was performed using MZ16FA microscope and DFC300FX camera (Leica).

Histology

Lung tissues were prepared as previously described (Robles-Oteiza et al., 2015) and stained as indicated in Supplemental Experimental Procedures. Vector detection reagents were used as directed. Photomicrographs were captured on a Leica DMI6000B inverted light and fluorescent microscope.

Immunodetection of lung and tumor lysates

Snap frozen lung lobes and tumors were homogenized in NP40 buffer, resolved on NuPage BT gels, and transferred to PVDF. Blocking and antibody detection were performed in Odyssey Blocking Buffer (TBS). Blots were probed as indicated (see Supplemental Experimental Procedures). LiCor Odyssey and software was used for detection and quantification. pErk and tErk were normalized based on control values between blots and the ratio of normalized pErk to tErk was determined for each sample.

Histological quantification

ImageJ software was used for tumor quantification. Data points represent individual mice for tumor number burden graphs and one tumor for BrdU and p-H3. PennVet Comparative Pathology Core determined individual tumor grades by using established tumor-grading schemes in experiments, where possible (Jackson et al., 2005; Nikitin et al., 2004). In experimental cases where individual tumors were impossible to determine, a cumulative tumor number score was assigned per mouse. In this grading scheme 1 indicates 5 or less tumors, 2 equals 6 to 10 tumors, and 3 represents more than 10 tumors of that grade.

Statistical analysis

Significance was analyzed using Prism Software. For tumor number and burden, BrdU+, and p-H3, unpaired student’s t-test were performed. For variance in tumor grade distribution, YFP, Sox2/YFP contingency analyses, chi-squared test for trend was used. Log-rank (Mantel–Cox) tests were performed to determine significance in survival studies.