1. NFIB expression is correlated with high proliferation, low apoptosis and metastasis in SCLC
Previous experiments have confirmed that NFIB is an oncogene in SCLC and accelerates SCLC initiation and progression. Fristly, to further clarify the specific evidence of NFIB participation in SCLC, we determined the amplifications of NFIB in human SCLC tissues and cells. The positive rate of NFIB staining was 80.7% (96/119) in SCLC tissues and 90.8% (59/65) in brain metastatic small cell carcinoma tissues, while NFIB was almost not expressed in carcinoid and LCNEC cells (Fig. 1A, B and Figure S1A). The positive rate of NFIB protein in brain metastatic small cell carcinoma tissues was significantly higher than that in primary SCLC tissues. High levels of NFIB was closely related to regional lymph node metastasis, distant metastasis and TNM stage (Table 1) and was also beneficial to intrapulmonary metastasis and vascular invasion in SCLC patients (Fig. 1B). In addition, NFIB was expressed in SCLC and adjacent normal fresh tissues, and the expression level of NFIB was different among different SCLC patients. Among 24 patients, the expression level of NFIB mRNA in 21 cancer tissues was higher than that in adjacent normal tissues (Fig. 1C). Moreover, NFIB expression was somewhat variability in different SCLC cell lines (Figure S1B and C).
Table 1
Correlations between NFIB expression and clinical/pathological features in SCLC.
Clinicopathologic features | Total, n | % | NFIB low | NFIB high | P |
n | % | n | % | value |
Gender | Male | 74 | 62.2 | 15 | 20.2 | 59 | 79.8 | 0.738 |
| Female | 45 | 37.8 | 8 | 17.8 | 37 | 82.2 | |
Age | ≤ 60 | 67 | 56.3 | 9 | 13.4 | 58 | 86.6 | 0.065 |
| >60 | 52 | 43.7 | 14 | 26.9 | 38 | 73.1 | |
Tobacco status | Non-smoker | 13 | 10.9 | 2 | 15.4 | 11 | 84.6 | 0.642 |
| Former smoker | 23 | 19.3 | 6 | 26.1 | 17 | 73.9 | |
| Active smoker | 83 | 69.8 | 15 | 18.1 | 68 | 81.9 | |
Histology type | Pure SCLC | 106 | 89.1 | 18 | 17 | 88 | 83 | 0.128 |
| Mixed SCLC | 13 | 10.9 | 5 | 38.5 | 8 | 61.5 | |
Tumor size(cm) | ≤ 3 | 75 | 63.1 | 17 | 22.7 | 58 | 77.3 | 0.336 |
| >3 | 44 | 36.9 | 6 | 13.6 | 38 | 86.4 | |
Syn | Negative | 17 | 14.3 | 15 | 88.2 | 2 | 11.8 | < 0.001 |
| Positive | 102 | 85.7 | 8 | 7.8 | 94 | 92.2 | |
CgA | Negative | 35 | 29.4 | 19 | 54.3 | 16 | 45.7 | < 0.001 |
| Positive | 84 | 70.6 | 4 | 4.8 | 80 | 95.2 | |
Ki-67 | Negative | 21 | 17.6 | 13 | 61.9 | 8 | 38.1 | < 0.001 |
| Positive | 98 | 82.4 | 10 | 10.2 | 88 | 89.8 | |
Lymph node metastasis | Negative | 61 | 51.3 | 17 | 27.9 | 44 | 72.1 | 0.020 |
| Positive | 58 | 48.7 | 6 | 10.3 | 52 | 89.7 | |
Distant metastasis | Negative | 105 | 88.2 | 23 | 21.9 | 82 | 78.1 | 0.041 |
| Positive | 14 | 11.8 | 0 | 0 | 14 | 100 | |
TNM staging | Ⅰ | 52 | 43.7 | 15 | 28.8 | 37 | 71.2 | 0.006 |
| Ⅱ | 33 | 27.7 | 6 | 18.2 | 27 | 81.8 | |
| Ⅲ | 20 | 16.8 | 2 | 10 | 18 | 90 | |
| Ⅳ | 14 | 11.8 | 0 | 0 | 14 | 100 | |
Data were expressed as n (%). P < 0.05 was considered statistically significant. |
SCLC means small cell lung cancer. |
To determine the biological function of NFIB, we used shRNA to interfer NFIB in several human SCLC cell lines. While shRNA-mediated NFIB knockdown caused a dramatic increase in apoptosis and a corresponding decrease in proliferation and vice versa (Fig. 1D-G). Notably, knockdown of NFIB mainly increased early apoptosis of tumor cells. Collectively, these data from both human patients and cells support an oncogenic role for NFIB in SCLC.
2. NFIB binds to sequences in BIK and BAK genes and represses pro-apoptotic factors
To corroborate our findings of NFIB association with apoptosis, we analyzed proteins involved in this process by western blotting. Repression of NFIB increased levels of the pro-apoptotic factors, BIK and BAK, and reduced the ratio of BCL-2/ BAX (Fig. 2A). Meanwhile, NFIB overexpression unsurprisingly led to different outcomes of related genes, although to a lesser extent than knockdown (Fig. 2B). One interesting finding is that NFIB binds to the promoters of the pro-apoptotic factors confirmed by ChIP-PCR in H2227, which lacked NFIB amplification, by means of comparison with overexpression NFIB (Fig. 2C). The rescue experiment in H209 also supports this result (Fig. 2D-F). Taken together, NFIB inhibition in SCLC cells was sufficient to upregulate the expression of the pro-apoptotic factors, BIK and BAK, and promote apoptosis in these cells. Our data suggest that NFIB expression is integral to human SCLC cell line viability through its suppressing role of apoptosis.
3. Notch1 is required for the effect of NFIB on chemosensitivity
We next examined whether NFIB can influence the acquisition of chemosensitivity in SCLC cells. A major factor in the dismal survival rates for SCLC is rapid relapse following standard-of chemotherapy (often cisplatin or carboplatin with etoposide)20. Therefore, we chose cisplatin and etoposide for the subsequent chemotherapy resistance trial. Wonderingly, NFIB knockdown tumor cell lines survived better than control cell lines in response to cisplatin and etoposide (Fig. 3A and B). To further explore the specific mechanisms by which NFIB affects chemoresistance in SCLC cells, we applied GDS4794, GSE62021 and GSE60052 profiles to screen for genes associated with chemoresistance in SCLC (Fig. 3C). We were surprised to find that NFIB is highly correlated with Notch genes which independently contributed to chemoresistance and intratumoral heterogeneity21,22. One of the putative NFI target gene, Hes1, was of particular interest because of its role as a critical transcriptional target of Notch signal pathway10. Hes1 has previously been shown to be important for chemotherapy resistance acquisition of SCLC cells and to be a negative prognostic factors for OS and PFS in SCLC patients11. Meanwhile, Notch signal pathway related genes, Notch1, Notch2, Notch4 and Hes1, was increased abundance following cisplatin treatment in mouse models of SCLC23. This model would suggest that Notch signaling activity is low in NFIBhigh SCLC cell lines and predict that decreased NFIB signaling would result in increased expression of Notch1 and Hes1. To examine this possibility, we first explored the transcriptional regulation of Hes1 and Notch1 by NFIB. Adenoviral transfer of shNFIB induced an increasing in Notch1, Notch2 and Hes1 expression levels (Fig. 3D and E). Active NFIB caused a progressive reduction of Notch1, Notch2 and Hes1, reciprocal to the increase of NFIB (Fig. 3E and I). These data suggest that NFIB inhibits Notch1, Notch2 and Hes1 gene expression in SCLC cells.
To extend upon the finding that NFIB represses Notch1 expression, we wanted to further determine if the activity of Notch signal pathway was regulated by NFIB. Firstly, we treated H446-shNFIB with the γ-secretase inhibitor dibenzazepine (DBZ)24. Application of DBZ led to a significantly decreased expression of Notch1, Notch2 and Hes1 (Fig. 3F). Moreover, inhibition of Notch activity was found to restore chemosensitivity in H446 shNFIB (Fig. 3G and H). Meanwhile, stimulation of H2227-NFIB with active Notch1 intra-cellular domain(N1ICD) rescovered expression levels of Notch1, Notch2 and Hes1 (Fig. 3I). Although overexpression of NFIB failed to change the chemosensitivity of SCLC cells, overexpression of Notch1 in H2227-NFIB could significantly reduce the drug sensitivity of cells to cisplain and etoposide (Fig. 3J and K). The rescue experiment in H209 also proved this opinion (Figure S2A-F). Collectively, these results imply that NFIB impacts chemosensitivity by targeting Notch1.
4. NFIB directly represses Notch1 expression in SCLC
To further characterize the transcriptional regulation of Notch1 by NFIB, we used IHC and IF to examine NFIB interacted with Notch1 in SCLC tissues and cells. It is worth noting that the expression of NFIB was markedly reduced in patients after chemotherapy, accompanied by a corresponding increase in expression levels of Notch1 and Hes1 (Fig. 4A) and was negatively correlated with Notch1 and Hes1 in SCLC human tissues (Fig. 4B). Significantly, NFIB, Notch1 and Hes1 showed co-localisation in SCLC cells (Fig. 4C). Furthermore, we predicted that NFIB directly binds to the promoter of Notch1 by JASPAR. To confirm this prediction, we carried out ChIP analysis in H2227-NFIB cells using primers corresponding to the region of Notch1 promoter. Bands corresponding to the Notch1 promoter between − 368 to − 388bp and − 1038 to − 1058bp were clearly detected and enriched following immunoprecipitation with an NFIB antibody compared to rabbit IgG and the − 1038 to − 1058bp resulted in significant enrichment in binding, respectively(Fig. 4D).
Next, to test the transcriptional regulatory activity of sequences that bound the Notch1 promoter, we used a luciferase reporter system. The activity of an Notch1 promoter-luciferase reporter, or a mutant version in which NFIB consensus sites have been mutated (Mut), was determined in NFIB overexpression lines (H2227-NFIB). The immediate Notch1 promoter showed a significantly reduced level of luciferase activity with cotransfection of NFIB-pGL3 (Fig. 4E). Together, these findings reveal that NFIB occupies the endogenous Notch1 promoter and can repress Notch1 promoter-driven gene expression.
5. NFIB knockdown accelerates the conversion of cell phenotype by targeting Notch activation
Previous studies have shown that Notch1 and Hes1 were essential for maintaining the balance between NE and non-NE cells phenotype9,25 and promoting phenotypic switching from NE to non-NE phenotype11. We hypothesized that the influence of NFIB on SCLC cells chemosensitivity may be related to cell phenotype transition which was regulated by Notch activation. As hypothesized, expression levels of NE markers, UCHL1, SYP, CALCA, CHGA and CHGB, were decreased in NFIB knockdown cells, instead the epithelial marker EPCAM was accordingly increased (Fig. 5A, B and Figure S2C). Cell lines of SCLC cells transformed from typical NE SCLC floating cluster to adherent growth, further suggestive of a change in differentiation (Fig. 5C). While this expression pattern was partially reversed after application of DBZ, expression levels of NE markers and EPCAM returned similar to levels before NFIB knockdown (Fig. 5D-F). Similarly, this expression pattern is mitigated by stimulated with N1ICD (Fig. 5G-I and Figure S2D). Meanwhile, NFIB expression was positively correlated with expression levels of SYP, CHGA and KI-67 in SCLC tissues (Figure S2G, Table 1). Collectively, these data further indicate that NFIB can inhibit the activation of Notch signal pathway and relate to the transition from NE to nonNE cell phenotype, therefore affect chemosensitivity of SCLC cells.
6. Inhibition of Notch activity reverses the effect of NFIB knockdown on chemosensitivity in vitro and vivo
Additional chemotherapy resistance assays was applied to confirm the clinical significance of Notch signal pathway in the effect of NFIB on chemotherapy resistance through a combination of the Notch antagonist tarextumab and cisplatin in NFIB knockdown SCLC cells. Moreover, tarextumab markedly decreased protein levels of Notch1 and Notch2 confirmed by western blotting (Figure S3A). Importantly, tarextumab treatment was found to impair chemoresistance in these cells (Fig. 6A and Figure S3B, C). Our experimental results are highly indicated that Notch1 is necessary for the affect of cisplatin resistance increasing induced by NFIB knockdown. Notably, the impact of NFIB overexpression on chemoresistance was altered upon Notch1 activation in H2227 cells (Fig. 6B and Figure S3D, E).
To further support the effect of NFIB on chemosensitivity and explore potential clinical applications, we established models in vivo. The nude mice injected with H446 cells after NFIB knockdown showed smaller volume of subcutaneous transplanted tumor and significantly lower sensitivity to cisplatin treatment than controls (NC). Moreover, combining tarextumab with cisplatin and etoposide strikingly decreased the tumor volumes of H446 shNFIB cells, whereas the administration of cisplatin and etoposide or tarextumab alone could not retard tumor growth (Fig. 6C-E). Similar to in vitro results, immunohistochemical analysis results indicated a decreased levels of NFIB, Notch1 and Hes1 in the combination-treated group compared with other groups (Fig. 6F). At the same time, multi-point puncture of subcutaneous tumors showed that the expression of NE genes was missing associated with increased expression of epithelial markers at large part of tissues after NFIB knockdown (Fig. 6G and H). These findings suggested that knockdown NFIB may relieve the inhibition effect on Notch1 and initiate the transformation of cell phenotype to reduce chemosensitivity, but this process will not be reversed by overexpression of NFIB.