Metastasis competence can be acquired early in tumorigenesis, although its underlying molecular intricacies remain unclear. A recent study provides key information about the function of L1CAM in conferring metastasis-initiation potential and chemoresistance in colorectal cancer by hijacking epithelial regenerative mechanisms.
Metastasis is considered a leading cause of cancer-related deaths, but the mechanisms through which some cancer cells become competent to disseminate from the primary site and regrow in a distant organ are not fully understood1. Recent findings are starting to provide important clues about the identity of metastatic-initiating cells (MICs) and the features that distinguish them from their non-metastatic counterparts1,2. In this issue of Nature Cancer, Ganesh et al.3 add to this body of literature by describing an MIC population that is defined by expression of the neuronal cell adhesion molecule L1CAM and is able to drive carcinoma propagation, metastasis and chemoresistance.
In normal tissue epithelia, specific populations of adult stem cells reside within specialized niches and can self-renew, generate differentiated progeny and respond to injury, thereby contributing to the long-term maintenance of tissue homeostasis4. Interestingly, many common cancer types are maintained by a population of cancer stem cells (CSCs)5 that function similarly to their normal counterparts in healthy tissue.
CSCs are known to display molecular and functional heterogeneity, an attribute that appears to be established early during tumorigenesis, given that genetically distinct CSC subclones are already present in primary tumors5,6,7,8. Some of these clones fade or become dominant during metastasis and in response to chemotherapy5,6,7,8. In this sense, CSCs not only ensure the long-term maintenance of tumors but also contribute to metastasis, long-term tumor quiescence, radio- and chemoresistance, and tumor relapse5,6,7,8.
However, many matters regarding the origins of metastasis, and more specifically of MICs, remain under debate. For instance, are the cells capable of initiating primary tumors in the same manner as those that initiate metastasis? Which attributes genuinely define CSC populations with the capacity to initiate primary tumors versus metastases? Understanding these aspects of tumor progression is of fundamental importance for the design of new effective treatments to tackle metastasis.
In previous studies, the same group showed that L1CAM is a determining factor for the spontaneous spreading of brain-metastatic cells along the abluminal side of the endothelium in a manner dependent on the transcription factor YAP (yes-associated protein 1)3,9. This function of L1CAM has also been confirmed in models of lung and bone metastasis3,9. Following up on these findings, Ganesh et al. analyzed normal human colon and colorectal cancer (CRC) samples and observed that, intriguingly, L1CAM expression was absent from the normal human colonic epithelium but was expressed in cancer cells, and was further enriched in both matched metastasis and post-therapy relapse tumors. When assayed for organoid-initiation in vitro and in vivo, purified L1CAMhi cells displayed greater tumor-initiating capacity than their L1CAMlow counterparts. This tumorigenic ability of L1CAMhi cells was inhibited after L1CAM silencing in organoids orthotopically transplanted into mice.
In the homeostatic intestine, LGR5 (leucine-rich-repeat-containing G-protein-coupled receptor 5), a target of the Wnt signaling pathway, is a marker of stem cells10. LGR5+ crypt base columnar cells (CBCs) are multipotent and capable of self-renewal during homeostasis and regeneration10. A second stem-cell population exists alongside CBCs, the so called +4 stem cells, because of their location with respect to the base of the crypt. These cells are normally quiescent and do not contribute to daily homeostasis, but after injury they become active and replenish the pool of CBCs, which regenerate the tissue11. Of note, LGR5+ cells have been identified as the origin of intestinal tumors through oncogenic driver mutations12.
To gain insight into the mechanisms underlying the biology of L1CAMhi tumor cells, Ganesh et al. studied whether LGR5 and L1CAM expression overlapped. By using single-cell messenger RNA–sequencing technology, they showed that L1CAM+ cells do not significantly co-express LGR5 yet robustly display a regenerative stem cell transcriptomic signature that emerges in mouse intestinal epithelium after damage13,14. These data prompted the investigators to test whether L1CAM might be upregulated during the process of epithelial regeneration. To this end, they used the dextran sodium sulfate–damage mouse model, which leads to crypt loss followed by a period of tissue recovery. On the basis of results with different mouse models in which L1cam was deleted in specific intestinal epithelial cells, the authors found that L1CAM is induced during tissue regeneration and is essential for colon regeneration and mucosal recovery.
The authors studied whether L1CAM might play a role during adenoma formation and/or adenocarcinoma progression. Although development, regeneration and malignant transformation share common molecular pathways, regeneration is rigorously controlled, whereas tumorigenesis entails superseding physiological constraints on growth4,5. In a series of in vivo experiments, the investigators studied mice that develop benign adenoma and/or malignant intestinal adenocarcinoma. Additionally, they included right- and left-sided CRC strategies to discard specific-side-related molecular features of the disease. Intriguingly, L1CAM was not necessary for adenoma initiation but was required for adenocarcinoma local growth and, more importantly, metastatic dissemination of tumor cells to the liver and lungs. These processes were independent of both tumor sidedness and cancer cell genotype. Additionally, the L1CAM cancer cells were resistant to irinotecan treatment, and L1CAM inhibition considerably improved the response of the tumors to chemotherapy.
How and why L1CAM becomes expressed on the pre-cancerous intestinal epithelium, as it does during tissue regeneration, is not entirely clear. Nevertheless, seeking to explore the underlying mechanisms, the authors discovered that binding of the transcriptional repressor NSRF (also known as REST) to an enhancer element of the L1CAM locus significantly decreased after disintegration of the colon epithelium. The decrease in REST occupancy at the L1CAM enhancer was a key determinant of the derepression of L1CAM expression and occurred concomitantly with the displacement of E-cadherin from the cell membrane. Importantly, disruption of epithelial integrity was essential for the initiation of metastatic spreading, thus indicating that this process operates during metastasis.
The work of Ganesh et al. (Fig. 1) makes an important step in the molecular characterization of CRC MICs. The results also raise many interesting new questions. For instance, is there any functional relationship between L1CAMhi and L1CAMlow cells? Do L1CAMhi cells depend on L1CAMlow cells to develop metastasis? Although the authors describe a certain degree of plasticity between L1CAMhi and L1CAMlow states, it will be of interest to determine why all L1CAMlow cells are not equally competent to give rise to L1CAMhi cells, as well as the in vivo importance of this molecular heterogeneity with respect to metastasis. Interestingly, phenotypic plasticity has been reported in cancer and metastatic stem cells, which can transit freely between active and inactive states2,7,8. In addition, strong evidence indicates that metastatic spreading is assisted by a supportive microenvironment and that MICs present very specific metabolic dependencies2,5. Given that L1CAM expression appears to be spatially restricted within the tumor architecture, investigating whether the signals that activate the transition between the L1CAMhi and L1CAMlo states are also dependent and niche-derived and/or metabolic cues will be interesting.
Another interesting finding pertains to the chemotherapeutic vulnerability of CRC cells after L1CAM is depleted. L1CAMhi cells are characterized as slowly proliferative chemoresistant cancer cells that promote tumor relapse after chemotherapy treatment. Other investigations have shown that quiescent cancer cells are an important source of chemoresistance6,7,8. Moreover, several of these studies strongly suggest that the population of MICs corresponds to these chemoresistant populations6,7,8. For instance, CD36+ MICs were initially identified on the basis of their unique ability to metastasize when orthotopically transplanted into mouse models of oral squamous cell carcinoma, breast cancer, ovarian cancer and melanoma2,15. Importantly, CD36+ cells exhibit a quiescent cell-cycle profile and have greater resistance to chemotherapy than CD36– cells in certain types of leukemia, pancreatic and renal cancer, among others7. These studies highlight two very important issues: first, a strong molecular relationship appears to link metastasis and chemoresistance, although further work is necessary to provide full understanding; second, caution is necessary when choosing treatment, because some therapies might lead to the activation of quiescent and pro-metastatic cell populations, thus ultimately promoting the formation of much more aggressive tumors after therapy is discontinued.
Finally, the discovery of REST as a transcriptional regulator that can determine whether the L1CAM-related metastatic program is active raises many new questions. The release of REST from chromatin has been associated with large changes in gene expression. It will be interesting to explore other factors acting in the transcriptional program regulated by REST, and whether this pathway is also required for establishing the molecular program that defines metastatic cells in different tumor types.
The work of Ganesh et al. provides an exciting basis for investigating these questions and elucidating the cellular and molecular mechanisms underlying metastasis initiation.
Lambert, A. W., Pattabiraman, D. R. & Weinberg, R. A. Cell 168, 670–691 (2017).
Pascual, G. et al. Nature 541, 41–45 (2017).
Ganesh, K. et al. Nat. Cancer https://doi.org/10.1038/s43018-019-0006-x (2020).
Blanpain, C. & Fuchs, E. Science 344, 1242281 (2014).
Batlle, E. & Clevers, H. Nat. Med. 23, 1124–1134 (2017).
Chen, J. et al. Nature 488, 522–526 (2012).
Ye, H. et al. Cell Stem Cell 19, 23–37 (2016).
Sánchez-Danés, A. et al. Nature 562, 434–438 (2018).
Valiente, M. et al. Cell 156, 1002–1016 (2014).
Er, E. E. et al. Nat. Cell Biol. 20, 966–978 (2018).
Barker, N. et al. Nature 449, 1003–1007 (2007).
Montgomery, R. K. et al. Proc. Natl Acad. Sci. USA 108, 179–184 (2011).
Barker, N. et al. Nature 457, 608–611 (2009).
Ayyaz, A. et al. Nature 569, 121–125 (2019).
Ladanyi, A. et al. Oncogene 37, 2285–2301 (2018).
The authors declare no competing interests.
About this article
Cite this article
Pascual, G., Benitah, S.A. L1CAM links regeneration to metastasis. Nat Cancer 1, 22–24 (2020). https://doi.org/10.1038/s43018-019-0014-x