Mechanistic Study of TTF-1 Modulation of Cellular Sensitivity to Cisplatin

The lung lineage master regulator gene, Thyroid Transcription Factor-1 (TTF-1, also known as NKX2-1), is used as a marker by pathologists to identify lung adenocarcinomas since TTF-1 is expressed in 60 ~ 70% of lung ADs. Much research has been conducted to investigate roles of TTF-1 in lung cancer biology. But, how it modulates cellular chemosensitivity remains poorly characterized. Our study shows that TTF-1 sensitizes the KRAS-mutated A549 and NCI-H460 lung cancer cells to cisplatin, a common chemotherapy used to treat lung cancer. This chemosensitization activity does not appear to be mediated by a TTF-1-imposed alteration on nucleotide excision repair. Mechanistically, TTF-1 induced a reduction in p-AKT (S473), which in turn activated glycogen synthase kinase 3 (GSK3) and reduced β-catenin. Intriguingly, in the EGFR-mutated NCI-H1975 and HCC827 cells, TTF-1 desensitized these cells to cisplatin; concomitantly, TTF-1 conferred an increase in p-AKT. Finally, the conditioned media of TTF-1-transefected cells sensitized TTF-1− cells to cisplatin, implicating that the TTF-1-driven chemosensitization activity may be dually pronged in both intracellular and extracellular compartments. In short, this study highlights the enigmatic activities of TTF-1 in lung cancer, and calls for future research to optimally manage chemotherapy of patients with TTF-1+ lung ADs.

relationship between TTF-1 and cellular chemosensitivity. Although TTF-1 is a transcription factor that is not directly druggable by small molecules, studies such as the present one may shed new light on exploiting TTF-1 to improve lung cancer management. Case in point is our recent work which has identified TTF-1 as a putative indicator of lung cancer vulnerability to statins 24 . Maeda et al. 6 reporting that the rat Ttf-1 transgene sensitizes the TTF-1 − human lung AD cell line A549 to cisplatin, we investigated how the human TTF-1 would impact cisplatin sensitivity of A549 cells. The expression of the TTF-1 transgene in A549-TTF-1 cells was documented by immunoblotting and RT-QPCR in our published studies 24,25 . TTF-1-transfectant cells and a control -the empty vector (EV) transfectant cells -were treated with cisplatin for 48 hr (Fig. 1A). IC50 calculations implicated a significant increase (>80%) in cisplatin sensitivity of A549-TTF-1 cells (6.14 μM ± 0.78 μM SEM) in comparison with the EV control (11.28 μM ± 1.47 μM SEM, Fig. 1B). Next, we conducted dose-response analyses by subjecting cells to 72-hr cisplatin treatments. In these experiments, we included an additional control -cells transfected with the homeodomain deletion (HDD) mutant TTF-1 which lacks DNA-binding activities 9 . Again, the TTF-1 transfectant cells were more sensitive to cisplatin than the EV and HDD control cells (Fig. S1A,B). It is important to note that the TTF-1 transgene did not significantly alter the proliferation of A549 cells (Fig. S1C). However, this TTF-1-dependent, heightened cisplatin sensitivity did not extend to another platinum chemotherapy -carboplatin (Fig. S2A,B); nor could we detect a TTF-1-dependent differential sensitivity toward gemcitabine (Fig. S2C,D), a chemotherapy sometimes used in combination with cisplatin for advanced nonsmall cell lung cancer (NSCLC) 26 . Therefore, we focused subsequent studies on cisplatin which is a standard chemotherapy for NSCLC 27 . It is interesting to note that the HDD mutant of TTF-1 did not appear to be localized in the nucleus as the wildtype protein was (Fig. 1G), based on the robust appearance of overlapping immunofluorescence of green (TTF-1) and blue (DAPI for DNA) only preferentially in the A549-TTF-1 cells (represented by the teal color). To substantiate this finding, we probed the expression of full-length TTF-1 and HDD mutant in the cytoplasmic and nuclear extracts of the A549 transfectant cells. The immunoblots demonstrated that while the full-length TTF-1

Diverse impacts of TTF-1 on cellular cisplatin vulnerability. Since A549 cells harbor a mutant
KRAS G12S allele, we next explored how a TTF-1 transgene would impact the cisplatin sensitivity of another KRAS mutant-containing lung cancer cell line (NCI-H460, KRAS Q61H ). Although NCI-H460 (referred to as H460) is more sensitive to cisplatin (IC50 ~ 4.5 μM) than A549, the TTF-1 transgene decreased the cisplatin IC50 of H460 to 3.1 μM (a 31% decrease from 4.5 μM, Fig. 1C,D). In addition to KRAS, EGFR is another important lung oncogene whose mutation is mutually exclusive from the mutations of KRAS in lung adenocarcinomas 28 . To investigate how a TTF-1 transgene would modulate cisplatin sensitivity under the EGFR mutant background, stable retrovirally transduced cells of NCI-H1975 (EGFR L858R+ , referred to as H1975 hereafter) carrying either the wt-TTF-1 transgene or the empty vector were analyzed for cisplatin sensitivity (Fig. 1E,F). Surprisingly, the TTF-1 transgene desensitized H1975 cells to cisplatin, contrary to the observations based in A549 and H460 cells. We transfected TTF-1 into another EGFR mutant lung cancer cell line (HCC827, EGFR exon 19 deletion) which is intrinsically more resistant to cisplatin than H1975 cells. Again, the TTF-1 transgene desensitized the EGFR-mutated HCC827 cells to cisplatin (Fig. S4A,B). In a human lung cancer cell line that does not harbor either KRAS or EGFR mutations, i.e. NCI-H1437 (TP53 R267P ), intriguingly, TTF-1 did not significantly alter its cisplatin sensitivity (Fig. S4C,D). An implication of these observations is that the cisplatin sensitivity modulation by TTF-1 may be genotype-dependent, warranting a larger study to examine this relationship. Nevertheless, the molecular mechanism of how TTF-1 confers cisplatin sensitivity in A549 cells remains to be determined and is the focal point of subsequent studies.

TTF-1 does not influence cellular ultraviolet (UV) light sensitivity. DNA damages induced by cispla-
tin mediate cisplatin-induced cytotoxicity and are repaired by NER. A possible mechanism of TTF-1-dependent enhancement of cisplatin sensitivity is that TTF-1 reduces cellular NER activity. However, our observation of cellular carboplatin sensitivity not influenced by TTF-1 argues against this hypothesis. To substantiate this line of research, we assessed the UV light sensitivity of A549-based transfectant cells, in view of the fact that NER is the main mechanism to repair UV-induced DNA photoproducts in human cells 29 . The clonogenic survival assay results, as shown in Fig. 2A, demonstrated that there was no statistically significant difference in the UV light sensitivity between A549-TTF-1 cells and the two control cells. Using Immunoslot Blot assays 30 to examine the repair kinetics of two major UV-induced DNA lesions (6-4 photoproducts and cyclopyrimidine dimers (CPDs)), we did not detect a significant difference in the repair of 6-4 photoproducts (Fig. 2B). However, the CPD repair appeared to trend slightly higher in the A549-TTF-1 cells. Finally, we quantified the damage-containing DNA oligonucleotides excised by NER in vivo 31 . The results again indicated the NER capacity was actually higher in A549-TTF-1 cells (Fig. 2C). These findings suggest that the TTF1-dependent enhancement of cisplatin sensitivity is unlikely mediated by a suppression of NER. Other TTF-1-dependent signaling events may be at play to heighten cellular vulnerability to cisplatin in the A549 cells.
TTF-1 modulates β-catenin. Multiple considerations motivated us to examine β-catenin as a mechanistic lead mediating TTF-1-dependent enhancement of cisplatin sensitivity: (i) β-Catenin and TTF-1 functionally interact in the context of lung and thyroid development 32 , (ii) β-Catenin directly regulates TTF-1 transcription 33 , and (iii) β-Catenin confers cellular cisplatin resistance 23,34 . We hypothesized that TTF-1 may feed negatively back to suppress β-catenin expression, which is expected to result in an enhancement of cisplatin sensitivity. To investigate how β-catenin expression may respond to a TTF-1 transgene, we used immunoblotting to analyze three pools of β-catenin expression (total, nuclear, and cytoplasmic) in the A549-based transfectant cells. Interestingly, TTF-1 appeared to significantly reduce the nuclear presence of β-catenin (Figs 3A-C and S7), which was also supported by immunofluorescence analyzed by confocal microscopy, as β-catenin appeared to be more restricted to plasma membrane (Fig. 3D).
the regulation of β-catenin by TTF-1 may involve GSK3. Constitutively active GSK3 is a key component of the β-catenin destruction multiprotein complex, which targets Wnt effector β-catenin for phosphorylation, ubiquitination, and destruction 35,36 . Regulation of the two GSK3 isoforms (α and β) is mediated by inhibitory serine phosphorylation (S21-GSK3α and S9-GSK3β) and by activating tyrosine phosphorylation (Y279-GSK3α and Y216-GSK3β) 37 . By immunoblotting, a decrease in the inhibitory p-S21 of GSK3α was detected in the TTF-1 transfectant cells (Figs 4A,B and S8), implicating a putative activation of GSK3α. Concurrently, the total GSK3β and the activating p-Y216 of GSK3β appeared to increase in the TTF-1 transfectant cells (Fig. 4A,B), suggesting a simultaneous stimulation of both GSK3 isoforms by TTF-1. Presumably, GSK3 activation underlies the TTF-1-triggered loss of nuclear β-catenin through the β-catenin destruction complex. To substantiate this finding, we used a well-validated GSK3 inhibitor (LiCl) 38,39 to probe the system. The rationale was that LiCl would dampen the TTF-1-dependent activation of GSK3 which in turn should reverse the enhanced cisplatin sensitivity conferred by TTF-1. As shown in Fig. 4C, LiCl desensitized A549-TTF-1 cells to cisplatin as predicted. On the other hand, LiCl appeared to exert a limited effect on the cisplatin sensitivity of the control A549-EV cells whose GSK3 activity is lower than that in A549-TTF-1 cells. These data demonstrate the mechanistic importance of the GSK3/β-catenin axis in TTF-1-driven cisplatin sensitization.
We were particularly interested in AKT because a direct transcriptional target of TTF-1, i.e. receptor tyrosine kinase-like orphan receptor 1 (ROR1), acts functionally upstream of AKT 13 . Indeed, as reported by Yamaguchi et al. 13 , ROR1 was upregulated in A549-TTF-1 cells (Fig. S5). We therefore examined the AKT activation status, represented by p-S473, in the A549 based transfectant cells. Figure 5(A,B) shows that p-AKT was significantly www.nature.com/scientificreports www.nature.com/scientificreports/ depressed in A549-TTF-1 cells; concomitantly, a known downstream target of AKT (p-PRAS40, Proline-rich Akt substrate 40 kDa) 40 was depressed as well in A549-TTF-1 (Figs 5A and S9). Importantly, in the H460 cell system, in which the TTF-1 transgene also conferred cisplatin sensitization, the TTF-1 transgene similarly inhibited p-AKT, p-PRAS40, and β-catenin (Figs 5C and S10). However, in the two EGFR-mutated cells (H1975 and HCC827) where the TTF-1 transgene conferred cisplatin desensitization, the TTF-1 transgene appeared to increase p-AKT (Figs 5D and S11). These observations are in line with the key finding presented in Fig. 4 that GSK3 activation (a known consequence of AKT deactivation) downregulates β-catenin and leads to cisplatin sensitization. To further corroborate these findings, we treated control and TTF-1 transfectant cells with a pan-AKT inhibitor (MK-2206) 41 . The anticipated outcome was that MK-2206 would phenocopy TTF-1 in sensitizing A549-EV cells to cisplatin. As shown in Figs 5E and S12, MK-2206 obliterated p-AKT signals as expected. Moreover, MK-2206 further suppressed the inhibitory S9 phosphorylation of GSK3β, conferring a decline in the nuclear β-catenin levels of the two control A549 transfectant cells (EV and HDD, Fig. 5E). In terms of cisplatin sensitivity, it was as expected that MK-2206 treatments increased the cisplatin sensitivity of the A549-EV control cells (Fig. 5F,G). Overall, these data highlight the role of the AKT/GSK3/β-catenin in TTF-1-dependent cisplatin sensitization.
The secretome of A549-TTF-1 cells also harbors the cisplatin sensitization activity. Our data so far establish mechanistically that the intracellular signaling axis (AKT/GKS3/β-catenin) mediates the cisplatin sensitization activity of TTF-1. In view of our published work that TTF-1 reprograms the secretome of lung cancer cells 25 , we hypothesize that the extracellular secretome, i.e. the conditioned media (CM), of A549-TTF-1 cells may also harbor the activity sensitizing naive A549 cells to cisplatin. To test this hypothesis, we used the CM harvested from A549-TTF-1 or the negative control cells (A549-EV) and transplanted them either homotypically or heterotypically onto recipient cells in the presence of increasing concentrations of cisplatin. Interestingly, we found that A549-EV cells incubated with the CM of A549-TTF-1 cells (denoted A549-TTF-1-CM) were more sensitive to cisplatin than the same cells incubated with A549-EV-CM (Fig. 6A,B). Conversely, similar media transplant experiments using the already sensitized A549-TTF-1 as the recipient cells didn't reveal a further increase in cellular cisplatin sensitivity regardless of the type of CM imposed (Fig. 6A,B). To identify the CM www.nature.com/scientificreports www.nature.com/scientificreports/ compartment containing the cisplatin sensitization activity, we used differential ultracentrifugation to segregate CM into two compartments -(i) exosomes and (ii) exosome-depleted CM (EDCM). An exosomal marker (CD9) 42 and nanoparticle tracking analysis were employed to characterize the exosome and EDCM fractions (Fig. S6). The cisplatin-sensitization activity appears to segregate with EDCM because A549-TTF-1-EDCM preferentially enhanced the cisplatin sensitivity of the control A549-EV cells but not that of A549-TTF-1 cells (Fig. 6C-F); whereas the exosomes, regardless of its source cells, did not alter the cisplatin sensitivity of either A549-EV or A549-TTF-1 cells (Fig. 6G,H).

Discussion
How TTF-1 may impact cellular responses to cisplatin, a standard chemotherapy for lung cancer, was not well characterized. In the 2012 study by Maeda et al. 6 , it was shown that the rat Ttf-1 transgene sensitized the human lung mucinous AD cell line (A549) to cisplatin. However, the molecular mechanism of this observation was not thoroughly investigated. In this study, we first reproduced the observation of Maeda et al. using the human TTF-1 transgene. Next, we surmised that the impact of TTF-1 on cellular cisplatin sensitivity may be connected to NER in view of the report that TTF-1 interacts with Damaged DNA binding protein 1 (DDB1) 5 . DDB1 is part of a ubiquitin E3 ligase complex (DDB1-CUL4ADDB2) that initiates NER by recognizing damaged chromatin with concomitant ubiquitination of core histones at the lesion 43,44 . Thus, the known interaction between TTF-1 and DDB1 may compromise the NER-directed repair of cisplatin-induced DNA damages, which would presumably lead to a heightened sensitivity to cisplatin. However, by multiple DNA repair assays we did not obtain data supporting an involvement of NER in TTF-1-dependent regulation of cisplatin sensitivity. Therefore, we turned to β-catenin which is known to modulate cellular sensitivity to cisplatin 23,34 . The consideration that motivated us to investigate β-catenin was partially based on the documented positive regulation of E-cadherin by TTF-1 in A549 cells 45 . β-Catenin is involved in two apparently unrelated functions: (i) being part of a large protein complexes involved in E-cadherin-mediated cell adhesion, and (ii) providing the transcriptional activation domain within a nuclear complex with T cell factor family of transcription factors 46 . Thus, TTF-1-dependent regulation www.nature.com/scientificreports www.nature.com/scientificreports/ of E-cadherin would likely perturb β-catenin as well. Consistent with our hypothesis, TTF-1 reduced the levels of β-catenin in the nucleus. Since GSK3α/β are well-known regulators of β-catenin via the β-catenin destruction complex, we next analyzed various regulatory phosphorylations of GSK3α/β in response to TTF-1 (Fig. 4A,B). The specific alterations we detected included: (i) a reduction of inhibitory S21 phosphorylation of GSK3α, (ii) an increase of total GSK3β, and (iii) an increase of activating Y216 phosphorylation of GSK3β. Because only a small portion of β-catenin is associated with the GSK3-driven β-catenin destruction complex, it is known that the changes in nuclear β-catenin levels (rather than cytoplasmic or total nuclear β-catenin levels) are more reflective of β-catenin destruction complex activity 37 . Consequently, it is less surprising that only the nuclear β-catenin levels were significantly reduced by the TTF1-dependent upregulation of GSK3 activities in this study. Nevertheless, a corollary of our observations is that there appears to be a general activation of GSK3 by TTF-1. Importantly, the roles of GSK3 and β-catenin are further supported by the GSK3 inhibitor (LiCl) which dampened the cisplatin sensitization activity of TTF-1 (Fig. 4C). We then focused on AKT as it is one of the kinases that control GSK3 by phosphorylating GSK3 at the inhibitory serine residues, given the knowledge that ROR1, a downstream target of TTF-1, regulates AKT 13 . In both A549 and H460 cell lines harboring a mutant KRAS allele, TTF-1 suppressed AKT activation which presumably unleashed GSK3 activation. Consistently, the AKT inhibitor MK-2206 sensitized the control A549-EV cells to cisplatin, mimicking TTF-1. In the two KRAS mutant backgrounds (A549 and H460), TTF-1 manifested the cisplatin sensitization phenotype. However, in H1975 and HCC827 cells which harbor EGFR mutations, TTF-1 elicited cisplatin desensitization. This is reminiscent of the known dimorphic phenotypes of Ttf-1 on tumorigenesis: being tumor-suppressive in the Kras mutant background 6,10,12 but oncogenic in the Egfr mutant context 6 . Perhaps, how TTF-1 reprograms cellular chemosensitivity is also genotype dependent (Fig. 7). Future studies will be needed to solidify this concept.
In this study, we did not detect differential sensitivity at higher cisplatin concentrations for the NCI-H460 transfectant cells. This perplexing observation may be explained by the concept of different drug resistance mechanisms being deployed at high vs low concentrations of chemotherapies like cisplatin (active vs passive resistance per Stewart et al. 47 ). The two control transfectant cells (EV and HDD) of NCI-H460 can be considered as having a shoulder on their dose-response curve at low cisplatin concentrations (in comparison with the dose-response curve of NCI-H460-TTF-1 cells). This shoulder may represent the "active resistance" caused likely by a saturable resistance mechanism that could be overloaded at higher concentrations of cisplatin. At higher cisplatin concentrations, this saturated resistance mechanism becomes inactive. Thus, the control cells are responding in the same way as the TTF-1 transfectant cells.
Our observations of the intracellular signaling axis (AKT/GSK3/β-catenin) putatively mediating the cisplatin sensitization activity of TTF-1 does not preclude an involvement of TTF-1-directed extracellular events also contributing to the cisplatin sensitization phenotype. TTF-1 reprograms the secretome as evidenced by the work of Taguchi et al. who identified a protein signature of Ttf-1 in the plasma of mouse models of lung ADs 48 . Moreover, our recent study reported an alteration of secretome and exosomal cargos by TTF-1 25 . By using the CM www.nature.com/scientificreports www.nature.com/scientificreports/ of A549-TTF-1 cells to educate the control A549-EV cells, we showed that the CM of A549-TTF-1 sensitized the control A549-EV cells toward cisplatin. These observations implicate the presence of a TTF-1-driven extracellular component that contributes to cisplatin sensitization and raise intriguing mechanistic questions such as whether the CM of TTF-1 + cells decreases nuclear β-catenin levels in recipient cells and the potential roles of secreted Wnt inhibitory molecules (e.g., Dickkopf-related proteins) in the TTF-1-dependent phenotype of cisplatin sensitization. More research is needed to resolve these mechanistic questions and identify the culprit extracellular factors. In summary, this study provides the mechanistic insight on how TTF-1 increases cellular sensitivity to cisplatin, a chemotherapy commonly administered to lung AD patients. Given the fact that TTF-1 immunoreactivity is routinely analyzed for lung cancer patients, hopefully the results described herein inspire others to comprehensively investigate how TTF-1 modulates chemotherapeutic responses and translate the findings to improve lung cancer management.
Conditioned Media and exosomes. Cells were plated and allowed to grow for 48 hours in RPMI-1640 with exosome depleted 1% FBS. The conditioned media was collected and centrifuged at 500 × g for 10 min to remove detached and dead cells. The supernatant was then collected and spun at 2,000 × g for 20 min to remove debris. The supernatant was again collected and spun at 20,000 × g for 30 min to remove larger microvesicles. The conditioned media was further centrifuged at 100,000 × g for 90 min. The supernatant represents the soluble fraction of the secretome (i.e., EDCM). The pellet was washed with an alkaline buffer (150 mM Na 2 CO 3 , pH 11) to break up protein aggregates and centrifuged again for 100,000 × g for 90 min. The exosome pellet was resuspended in PBS and kept at −80 °C. Quantification of the exosome concentration was performed using NTA (NanoSight NS300). For CM-based cisplatin sensitivity experiments, CM was collected as described above and was homotypically or heterotypically transplanted onto either A549-EV or A549-TTF-1 cells (replacing what would otherwise be fresh media) and allowed to incubate with cisplatin for 48 hrs. Cell viability was analyzed with cell titer blue as discussed above. The methods for conducting UV light sensitivity and NER assays are presented in the supplementary section. statistical analysis. Quantitative experimental data were determined using biological replicates. Two-tailed t-tests and one-way ANOVAs were used to analyze data for statistical significance as described in the figure legends. *p ≤ 0.05; **p ≤ 0.01; ***p ≤ 0.001. P value > 0.05 is considered not significant.

Data Availability
All data generated or analyzed during this study are included in this published article (and its Supplementary Information files).