Cellular quiescence: How TGFβ protects cancer cells from chemotherapy

ABSTRACT

Using a functional proliferation reporter we identified quiescent tumor propagating cancer cells (TPCs) in intact squamous cell carcinomas, and found that TGFβ signaling controls their reversible entry into a growth arrested state, which protects TPCs from chemotherapy. TPCs with compromised TGFβ/Smad signaling can't enter quiescence and subsequently die from chemotherapy.

Autocommentary

Cellular quiescence has long been speculated to account for chemotherapy resistance and consequently cancer cell survival, allowing for seeding of recurrent tumors following remission.¹ Still, the unbiased, functional isolation of quiescent tumor propagating cancer cells (TPCs) based on their slow proliferation rate in intact solid tumors has been technically challenging. The molecular mechanisms dictating quiescence, therapy resistance, and the seeding of new tumors after treatment remains to be defined in most cancers. To investigate quiescence molecularly and functionally, and to determine its relevance as a drug resistance mechanism, we developed an experimental approach that enabled us to identify quiescent TPCs in cutaneous squamous cell carcinomas (SCCs), uncover the signaling pathways that govern their quiescence, and functionally demonstrate their differential response to chemotherapy directly within tumors.⁵

SCCs are heterogeneous tumors which are sustained by basal TPCs located along the tumor stroma interface³ where they express high levels of α6β4 and β1 integrins⁴ and Sox2,⁵ before they down-regulate these markers and differentiate into suprabasal SCC cells without proliferative potential. To directly identify proliferative heterogeneity within the self-renewing basal layer of intact SCCs, we isolated TPCs from tumors.⁴ infected them with our virally encoded proliferation reporter, and transplanted them orthotopically onto mice to establish secondary tumors.² Our proliferation reporter expresses red fluorescent protein (RFP) to constitutively label the tumor parenchymal lineage, along with a Doxycycline repressible Histone H2B green fluorescent protein (H2BGFP). H2BGFP is a long-lived protein that incorporates into nucleosomes, and its transcription can be turned off with Doxycycline, which allows actively proliferating cells to only incorporate unlabeled histones as they replicate their genome. Proliferating cells will thereby lose half their H2BGFP intensity with each cell division, while quiescent cells retain this label.⁶ Pulse-chase time course experiments revealed label-retaining and non-label retaining basal SCC cells, which were both able to establish secondary tumors that are similar to one another and to their parent in serial transplantation experiments (Fig 1).

Figure 1.

Chemotherapy compromises proliferative but not quiescent tumor propagating squamous cell carcinoma cells. Tumor propagating cancer cells (TPCs) interconvert (bi-directional arrows) between quiescent (green) and proliferative (purple) states before they differentiate into squamous cell carcinoma (SCC) cells without proliferative potential (pink). The ability of TPCs to reversibly growth arrest depends on TGFβ signaling. While both, proliferative and quiescent TPCs are able to establish tumors that are identical to one another and their parent in serial transplantation studies, quiescent TPCs survive chemotherapy and sustain tumorigenesis after treatment. TGFβ inhibition prior to chemotherapy however allows all TPCs to actively cycle, which significantly increases their response to chemotherapy.

The interconversion of quiescent and proliferative TPCs upon transplantation suggested that the contrasting proliferative behaviors of TPCs could be dynamic, rather than hierarchical traits within these tumors. Consistent with this idea we found that label retaining and non-label retaining cells could be further separated into transferrin receptor⁷ (CD71) high and low populations, which are actively cycling, or growth arrested at the experimental endpoint, respectively. These behaviors underscore that basal TPCs can switch dynamically between quiescent and proliferative states in SCCs (Fig 1).

We therefore wondered whether quiescent and proliferative TPCs respond differently to the chemotherapeutic agent 5-fluorouracil (5-FU), which is often used to treat these carcinomas.⁸ Indeed, chemotherapy damaged DNA preferentially in proliferative TPCs, which are induced to die as a response to treatment. However, quiescent TPCs were only slightly harmed by chemotherapy and they survived to initiate new tumors even more effectively after 5-FU exposure.⁵ It is intriguing to speculate that chemotherapy may further purify and select a stem-like population, but genetic changes caused by imprecise DNA repair, or epigenetic changes in response to cellular stress and inflammation, could also make surviving, quiescent TPCs more tumorigenic. Irrespective of the mechanisms, successful treatment seems to depend on a complete response to the initial chemotherapy, which is much more potent in actively proliferating cells, and which may benefit from combinatorial approaches that stimulate quiescent TPCs to re-enter the cell cycle prior to treatment.

Consistent with this idea, our differential gene expression analyses identified TGFβ signaling as an essential regulator of TPC quiescence in SCCs. Functional studies revealed that TPCs proliferate continuously in Tgfbr2-deficient SCCs, although they could still permanently cell cycle arrest and differentiate into SCC cells without proliferative potential. Nevertheless, their compromised ability to reversibly withdraw from the cell cycle increased their response to chemotherapy and limited their ability to seed new tumors after treatment. Moreover, Smad2/3 ChIP-seq on TGFβ1 stimulated cells, combined with chromatin accessible regions defined by ATAC-seq on TPCs isolated from SCCs, identified direct transcriptional targets of TGFβ/Smad signaling. Although Smad2/3 bound elements were more accessible in quiescent TPCs, they were also accessible in proliferative TPCs, suggesting that TPCs are poised to respond quickly to growth arrest signals. These analyses revealed that TGFβ/Smad signaling directly regulates cell cycle genes, independent of p21, which was not needed to efficiently cell cycle arrest. Taken together, these data suggest that targeting quiescence signaling mechanisms in combination with conventional chemotherapy could be a viable therapeutic option. Still, further work will be required to determine whether TGFβ mediated quiescence is a general mechanism of therapy resistance in different cancer models.

Head and Neck (HN)-SCCs share many histological and mutational similarities with our tumor model,⁹ so we decided to interrogate HNSCC patient data¹⁰ to determine whether our murine quiescent TPC signature was relevant to the classification of HPV-negative SCCs that show complete response, or progression with chemotherapy. These comparisons revealed a significant overlap only between quiescent TPCs and progressive HNSCC patient specimens. Additionally, TGFβ1 signaling was the most significantly activated pathway in progressive HNSCCs, suggesting that TGFβ1 mediated cellular quiescence could have significant therapeutic implications for therapy resistance and recurrence in patients. Further work on a larger patient cohort will determine if our quiescent TPC signature has prognostic value and it will identify biomarkers that may predict therapeutic responses prior to treatment and help personalize the treatment of HNSCC patients.

Our experimental approaches allowed us to overcome technical limitations that precluded the direct identification and isolation of quiescent TPCs so far, and we have now started to grasp their molecular and functional characteristics in intact SCCs. Although our study proves the principle that slow cycling or quiescent cells can be directly identified in solid tumors, additional studies in other cancer models and primary patient material will be required to better understand how proliferative heterogeneity is established and maintained, and how it enables therapy resistance and recurrent tumor growths in patients. Further identification and testing of quiescent populations in other cancer models could determine if growth arrest is ultimately responsible for chemotherapy resistance, or whether additional mechanisms inherent to a cell's quiescent state, such as metabolic and enzymatic activities, are needed to protect the cells from treatment. Our approach will allow us to interrogate directly how tumor cells respond to treatment and how TPCs that don't respond to the initial treatment can drive disease progression despite intended intervention.

Funding Statement

This research was supported by grants from the National Institutes of Health R01CA181111 and the American Cancer Society RSG-16-033-01-DDC to M.S. and a T32 CA009161 (Levy PI) fellowship to J.B.

Disclosure of potential conflicts of interest

There are no potential conflicts of interest to disclose.