Integrator is part of the Notch transcriptional supercomplex
To investigate the mechanism of Notch transcriptional regulation, we sought to identify key downstream components that link the NTC to activation of transcription. Therefore, we purified Notch1/CSL-dependent complexes from nuclear extracts prepared from a Notch-driven T-cell lymphoma cell line (4084). Nuclear lysates were fractionated by size exclusion chromatography and the fraction with Notch1-containing complexes was purified by CSL-DNA affinity FPLC. An isogenic Myc-driven T-cell lymphoma cell line (6780) served as a negative control for purification since it lacks activated Notch and Maml. Eluate fractions containing Notch were subjected to LC-MS/MS to determine the identities of co-eluting proteins (Fig. 1a). This analysis revealed that we had co-purified 13 out of 15 known components of the Integrator complex with Notch, as they were not identified in the control cell line using the same purification scheme. These data indicate that INT is a component of the Notch transcriptional supercomplex and plays a role in Notch-mediated transcription.
In order to validate the results obtained from the purification and MS analysis in a system relevant to human cancer, we performed ChIP assays in Notch-dependent EAC cell lines OE19 and OE33. These cell lines were classified as Notch-dependent because they exhibit an inhibition of the cell growth as well as decreased viability and transcription of Notch target genes when treated with DAPT, which is a gamma-secretase inhibitor that inhibits Notch activity [21, 22]. Using Integrator subunits 1 and 11 (INTS1—the largest core subunit and INTS11—a catalytic subunit) as indicators of the INT complex in the ChIP analysis, we investigated the occupancy of INT on the HES1, HES4, and CCND1 promoters (Fig. 1b and Additional file 3: Fig. S2). We observed both INTS1 and INTS11 ChIP signals on the promoters in both cell lines together with ChIP signals of CSL, Notch1, Maml1, and NACK. This result indicates that INT co-localizes with both the NTC and NACK on Notch target genes. Therefore, we validated the purification and MS analysis in the Notch-dependent tumor cells.
Both INT and NACK are critical for transcriptional activation mediated by Notch
Previously, we have reported that the protein NACK plays an essential role in Notch-mediated transcriptional regulation. NACK is recruited to the NTC following the acetylation of Maml1 on two residues Lys188 and Lys189 by p300. Both p300 activity and wild-type Maml are required for the function of NACK [7]. Moreover, we have discovered that NACK is required for the recruitment of RNAPII to Notch-activated promoters. However, the mechanism of RNAPII recruitment by NACK is unclear. Since we observed INTS1 and INTS11 co-localization with NACK on the Notch target promoters, we explored the relationship between NACK and INT in Notch-mediated transcription. We depleted NACK or INTS11 in OE19 and OE33 EACs via RNA interference and analyzed transcription of several key Notch1 target genes (CCND1, NOTCH3, and HES1). The depletion of either INTS11 or NACK resulted in the attenuation of transcription of Notch target genes. When INTS11 was knocked down in both OE19 and OE33 cells via RNA interference, transcription of canonical Notch target genes was significantly reduced (Fig. 1c). Similarly, when NACK was knocked down by siRNA, we observed a reduction of transcription of a similar set of genes (Fig. 1d). These results indicate that NACK and INT are both critical for transcriptional activation mediated by Notch.
NACK is required for Notch1, Integrator, and RNAPII on the Notch target promoter
In order to investigate the relationship between NACK and other co-factors on a Notch-dependent promoter, we depleted NACK via RNA interference in OE33 and analyzed the occupancy of Notch1, the active form of RNAPII phosphorylated at Serine 5 (RNAPII-S5P), INTS1, and INTS11 on the HES1 promoter via a ChIP assay. When NACK was depleted, we observed a profound decrease of RNAPII-S5P ChIP signal on the HES1 promoter (Fig. 2a), in line with previously published data [7]. Similarly, the occupancy of Notch1, INTS1, and INTS11 on the HES1 was also decreased when NACK was depleted, which indicates that NACK is important for the stabilization of the NTC and localization of INT to the promoter (Fig. 2a).
Since NACK is required for the presence of INTS11 and RNAPII-S5P on the promoter, we reasoned that an inability of NACK to bind the NTC should result in the absence of INTS1, INTS11, and RNAPII-S5P on the Notch target promoter, which should lead to the transcriptional block. To test this hypothesis, we assayed the reconstitution of the Notch transcriptional supercomplex in HEK293T cells by transfecting pcDNA3 vectors encoding various combinations of the NTC proteins. HEK293T cells express INTS1, INTS11, and CSL, but lack endogenous Notch, Maml, and NACK, thus allowing us to program the cells with various protein components of the NTC as wild-type (WT) or mutant versions and assess the activity. When cells were transfected with Notch1 and Maml1, there was a minimal induction of HES5 transcription as compared to the cells transfected with an empty pcDNA vector (Fig. 2b, lanes 1 and 2). However, when NACK was added to the transfection, we observed a dramatic dose-dependent increase in HES5 transcription (Fig. 2b, lanes 3 and 4). This result established the HEK293T recapitulation assay as a bone fide model for Notch-mediated transcription. We next sought to examine the effect on Notch-mediated transcription when we utilized previously described mutants of Maml1 and NACK that attenuate Notch-mediated transcription. Maml1(2S) harbors the K188R/K189R mutations that are no longer sites of acetylation. This mutant version results in the inability of p300 to acetylate Maml1 and thus prevents binding of NACK to the NTC [7]. The NACK(K) mutant possesses a K1002A mutation in a critical active site residue that renders NACK unable to bind the NTC [7]. To establish a baseline for the occupancy of Notch1 and INTS11 on the HES5 promoter, HEK293T cells were transfected with Notch1, Maml1 and NACK, and the occupancy of Notch1, INTS1, and INTS11 was determined by ChIP analysis. The ChIP assay revealed that the Notch1, INTS1, and INTS11 occupancy on the HES5 promoter substantially increased over the control condition (pcDNA3-transfected) (Fig. 2c). When WT Maml1 was substituted with the K188R/K189R (Maml1(2S)) mutant, there was a complete loss of Notch1, INTS1, and INTS11 binding. The same result was obtained when WT NACK was replaced with the K1002A mutant (NACK(K)). Both mutants prevent NACK from binding the NTC, which results in the absence of Notch1 and INT on the promoter. In turn, this leads to the loss of RNAPII-S5P on the HES5 promoter as we have shown previously [7]. Together these data indicate that NACK is essential for the stabilization of the NTC, subsequent recruitment of INT and RNAPII to the Notch target promoter, and thus activation of transcription.
Integrator is required for RNAPII-S5P, SPT5, and cyclin T1 on the Notch target promoter
Similarly, we examined the effect of INTS11 depletion on transcriptional co-factors on the HES1 and HES4 promoters in OE33. When INTS11 was depleted, there was no decrease in the occupancy of Notch1, Maml1, or NACK on both promoters indicating that INT is not required for the stabilization of the NTC or recruitment of NACK to the Notch-dependent promoters (Fig. 2d, f). In contrast, we observed a decreased occupancy of the active phosphorylated form of RNAPII (RNAPII-S5P), but not unmodified RNAPII, on both HES1 and HES4 promoters (Fig. 2e, g). This result indicates that INT is not required for the recruitment of RNAPII, but for its activation on the Notch target promoters. In addition, we evaluated the occupancy of SPT5 and Cyclin T1 (CCNT1) on the promoter of HES1 when INTS11 was knocked down. These co-factors are known to interact with RNAPII and the INT complex and are involved in the initiation of transcription. CCNT1 is a subunit of the positive transcription elongation factor b (P-TEFb) needed for the elongation phase in transcription. SPT5 is part of DRB sensitivity-inducing factor (DSIF) that can act either as a negative or positive elongation factor depending on its phosphorylation status by P-TEFb [23,24,25,26]. With INTS11 depletion, we observed a decrease in the occupancy of both SPT5 and CCNT1 on the promoter indicating INT requirement for these co-factors (Additional file 4: Fig. S3).
Integrator is over-expressed in Notch-dependent esophageal adenocarcinoma and it is required for EAC cell growth and tumorigenesis
We demonstrated that INTS11 depletion resulted in a decrease of Notch target transcription in Notch-dependent EAC cell lines OE19 and OE33 (Fig. 1c). Since Notch has a critical role in driving the neoplastic phenotype in EAC, we sought to determine if INT played a role in Notch-driven tumorigenesis of EAC. We determined that EAC cell lines OE19, OE33, FLO1, SKGT2, and EAC42 overexpressed all the INT subunits examined (INTS1, 3, 6, 9, 10, 11, 13) as compared to the non-tumorigenic esophageal cell line Het1A, indicating the importance of INT in these cancer cells (Fig. 3a). Furthermore, when we depleted INTS11 using a doxycycline-inducible shRNA vector in OE33, cells failed to form colonies in the induced state. Colonies were readily formed when the shRNA was not induced, nor was colony formation affected with an irrelevant control vector either induced or not (Fig. 3b). Similarly, when INTS11 was knocked down by siRNA in two other EAC cell lines OE19 and FLO1, the cells failed to form colonies compared to the cells transfected with control siRNA. Moreover, the non-tumorigenic Het1A cells efficiently formed colonies when transfected with either siRNA to INTS11 or siControl (Fig. 3c).
To evaluate whether INT is required for Notch-driven tumor growth in vivo, we inoculated nude mice with OE33 cells harboring a doxycycline-inducible shRNA against INTS11 and allowed tumor formation. When tumors reached > 200mm3, shRNA was induced by a doxycycline treatment. Over the course of the treatment, we observed that shGFP OE33 tumors (control) continued to grow and reached an average size of > 550 mm3 while shINTS11 OE33 tumors (2 independently derived clones) failed to grow and significantly decreased in size. By the end of the treatment, almost all mice in shINTS11 groups had no measurable tumors (Fig. 3d). Similarly, the knockdown of Notch or NACK in EAC causes attenuation of Notch signaling and cell growth arrest both in vitro and in vivo, as we have reported previously [4, 7, 18]. Together these findings indicate that the NACK-INT mode is critical for maintaining active Notch-mediated transcription to drive cell proliferation and tumorigenesis in Notch-dependent EAC.
INTS11 depletion results in apoptosis via G2/M cell cycle arrest
We demonstrated that depletion of INTS11 decreased the transcription of Notch target genes in EAC leading to tumor growth failure. As the downregulation of Notch has been associated with cell growth inhibition and stimulation of apoptosis in various types of cancer [27,28,29], we sought to further determine whether the INTS11 depletion results in apoptosis in EAC via flow cytometry. Under conditions of INTS11 KD in OE19 cells, there was a significant time-dependent decrease in a cell viability (Fig. 4a, b) with a concomitant increase in a number of apoptotic cells compared to the control (Fig. 4c, d). Furthermore, we observed increased levels of the proapoptotic proteins BAK1 and cleaved PARP and a decrease in the anti-apoptotic protein BCLXL by western blot analysis (Fig. 4e, lanes 2 and 4). The changes in protein levels of BAK1 and BCLXL (proteins of early apoptosis) began to emerge at 72 h following INTS11 KD, whereas, an increase in cleaved PARP (late apoptosis) was observed only after 96 h, indicating the proper temporal relationship between early and late apoptosis in these cells. Furthermore, we observed a block in cell cycle progression at G2/M at 96 h, with a concomitant decrease in cells in G1, indicating that cells were failing to progress through the cell cycle and arresting at G2/M resulting in apoptosis (Fig. 4f–h). When we evaluated the levels of BAK1 and BCLXL as well as the cell cycle upon knockdown of Notch1, we obtained results similar to those with INTS11 knockdown. We observed an increased expression of BAK1, decreased expression of BCLXL, and G2/M cell cycle arrest (Additional file 5: Fig. S4). Together, these data indicate that the inhibition of Notch-dependent transcriptional regulation mediated by INTS11 KD directly contributes to the cell cycle arrest and activation of apoptosis in EAC.