Hopping between differentiation states in lung adenocarcinoma

Abstract

The work by Cheung et al., published in this issue of Cancer Cell, demonstrates another example of how lineage-specific transcriptional regulators of differentiation, GATA6 and HOPX, can control the fate of lung adenocarcinoma progression.

Lineage-specific transcriptional regulators govern differentiation status during normal lung development as well as in lung adenocarcinoma. Aberrant activation or inactivation of transcriptional regulators important for lineage commitment have been frequently observed in hematopoietic malignancies such as PAX5 deletion and TAL1 translocation in acute lymphocytic leukemia, and RUNX1 and RARA translocations in acute myelogenous leukemia. There has been emerging evidence that tissue-specific differentiation programs also become dysregulated during cancer evolution in solid tumors. ETS family members are frequently translocated in prostate cancer and a lineage restricted genomic amplification of developmental transcription factors occurs frequently in solid tumors, as exemplified by MITF in melanoma, NKX2-1 in lung adenocarcinoma and SOX2 in lung and esophageal squamous cell carcinomas.

In the orderly development of tissues, temporally and spatially controlled mitogenic signaling directs cell lineages to proliferate and/or migrate. However, over the course of cancer evolution, cells accumulate genomic and/or epigenomic alterations to adapt their lineage identity to sustained oncogenic signal activation. These adapted differentiation states often predict the fate of cancer progression. For example, poorly differentiated lung adenocarcinomas, the most divergent differentiation state from the normal respiratory unit, generally have the worst prognosis, while for subtypes of lung adenocarcinoma that more closely resemble normal progenitors, the prognosis varies among subtypes classified based on architectural patterns that resemble normal differentiation status (Yoshizawa et al., 2011) (Figure 1).

Figure 1.

Metastatic potential varies among subtypes of lung cancer defined by transcriptional regulation of cell lineages. Transcriptional regulators active in different cell lineages are shaded in green (WNT/TCF), yellow (SOX2), orange (FOXA1), red (NKX2-1) and blue (GATA6/HOPX). Arrows indicate differentiation. Curved arrows indicate malignant transformation.

Studies using animal models of lung adenocarcinoma have shown that tumors with dysregulated developmental transcriptional regulators indeed present with differential biological properties. In the case of NKX2-1, the most well-characterized example of a transcriptional regulator of lung adenocarcinoma differentiation, stochastic loss of mouse Nkx2-1 promotes dedifferentiation and metastasis in a model of Kras-driven lung adenocarcinoma (Winslow et al., 2011), while discrete Nkx2-1 deletion induces endodermal trans-differentiation and increases tumor initiation and progression (Maeda et al., 2012; Snyder et al., 2013). However, the consequences of an altered differentiation program are context-dependent as shown by Maeda et al. where haplo-insufficiency of Nkx2-1 conversely decreased the incidence and progression of Egfr-driven lung adenocarcinomas (Maeda et al., 2012). These results are consistent with the heterogeneity of human lung adenocarcinomas where tumors with low NKX2-1 expression have generally worse prognoses while oncogenic amplification of NKX2-1 is also found in a subset of the disease.

Another developmental pathway, WNT/TCF signaling, has also been shown to mediate lung adenocarcinoma progression and metastasis. The TCF4 transcriptional signature is associated with human lung adenocarcinoma recurrence, and maintenance of TCF4 activity through its downstream effectors, including LEF1 and HOXB9, in metastatic derivatives of lung adenocarcinoma cell lines is required for metastatic potential in a xenograft model (Nguyen et al., 2009). In highly metastatic Lkb1-deficient Kras-driven lung tumors, progression to metastasis is associated with signatures of β-catenin activity, epithelial-mesenchymal transition (EMT) and focal adhesion (Carretero et al., 2010). In addition, in a transgenic mouse model, constitutively active β-catenin cooperates with oncogenic Kras to enhance tumor development and de-differentiation to an immature distal epithelial lineage through increased expression of Id2 and Sox9 and a decreased expression of Hopx (Pacheco-Pinedo et al., 2011). Thus, there seems to be interplay between WNT/TCF signaling and transcriptional regulators of differentiation in the lung.

The study by Cheung et al. in this issue describes an “alveolar” gene signature compiled from published transcriptomic data of several experimental alveolar differentiation phenotypes, that further anti-correlates with an embryonic stem cell signature, and can cluster lung adenocarcinomas into two subgroups: “alveolar-like” and “distal airway stem cell (DASC)-like” (Cheung et al., 2013). The “alveolar-like” subgroup represents histologically more differentiated lung adenocarcinomas with decreased incidence of metastasis. The alveolar-like subgroup was characterized by increased expression levels of NKX2-1, GATA6 and their known interactor HOPX, consistent with the findings in non-metastatic primary murine Kras-driven lung adenocarcinomas (Winslow et al., 2011).

Cheung et al. further demonstrate that GATA6 and HOPX cooperatively suppress in vitro invasion, and metastatic potential in xenograft models using human lung adenocarcinoma cell lines. GATA6, another key transcriptional regulator in lung morphogenesis, is essential for the maturation of alveolar type I epithelial cells. Furthermore, HOPX has been shown to regulate alveolar cell maturation in vivo by suppressing surfactant protein production downstream of NKX2-1 and GATA6 activity in the developing airway (Yin et al., 2006). Therefore, it is plausible that the cooperative function of GATA6 and HOPX is to direct a subpopulation of alveolar type II cells towards a more mature state which will then terminally differentiate into alveolar type I cells. Furthermore, it suggests that the repression of this pro-differentiation function confers a more invasive and metastatic lung adenocarcinoma phenotype (Figure 1).

Interestingly, Cheung et al. found these effects were not mediated through β-catenin activity or EMT, but rather by suppression of squamoid differentiation, the gene signature of which includes expression of KRT6A and KRT6B. While GATA6 has been shown to suppress expansion of bronchiolar epithelial cells through inhibition of the β-catenin pathway in developing and regenerating lungs, Gata6 is not required for the earliest stages of lung branching, but is expressed throughout the developing foregut endoderm (Zhang et al., 2008). In the esophagus, where stratified squamous epithelium is native, GATA6 serves as an amplified lineage oncogene in metaplastic adenocarcinomas (Lin et al., 2012). Therefore, the role of GATA6 may generally be to suppress differentiation to the squamous epithelial lineage.

Another interesting finding of the Cheung et al. study is that, unlike Nkx2-1 in mouse models of lung cancer, the genotypes of the major driver oncogenes, KRAS and EGFR, in the two lung adenocarcinoma cell lines primarily used in this study, did not have differential effects on metastatic phenotype. Consistent with this result, the “alveolar” subgroup is not associated with particular genotypes of KRAS, EGFR, STK11 and TP53. However, the high level of genomic complexity in human cancer cell lines makes it difficult to interpret the influence of a single genotype, and it would be of interest to test whether these observations are genotypespecific within genetically defined autochthonous in vivo models.

The work by Cheung et al. provides mechanistic insight into how lineage factors play roles in maintaining cellular identity within tumors, in some cases deleterious for cancer fitness, invasion and/or metastasis while advantageous in other cases. Cancer cells are typically characterized by reduced developmental plasticity, or fixation at a particular developmental point within a lineage, that are ordinarily tightly regulated in tissue development in part through the accumulation of genomic alterations. While studies of genomic alterations that lead to activation or inactivation of signaling molecules in lung adenocarcinoma have been fruitful, the molecular regulators of lung adenocarcinoma differentiation, and indeed of differentiation in normal lung alveolar epithelium, remain poorly understood. Therefore, further investigation of dysregulated transcriptional networks in solid tumors may provide clues to the stages of normal lung epithelial development as well as identify additional important tumor vulnerabilities.