JAG1 overexpression contributes to Notch1 signaling and the migration of HTLV-1-transformed ATL cells

The HTLV-I-transformed (IL-2 independent) cell lines (MT2, MT4, C8166, and C91PL), ATL-like cell lines (ATLT, ATL25, ED-40515(−), and TL-Om1), and ALL cell lines (Jurkat and Molt4) were grown in RPMI 1640 with 10% fetal bovine serum. The HTLV-I-immortalized (IL-2 dependent) cell lines (LAF and 1185) and the ATL-like cell lines (ATL43T, ATL55T, KOB, KK1, SO4, and LM-Y1) were grown in media with 20% serum and 50 U/mL IL-2. The ATL patient samples used in this study were previously described in another publication [13, 16]. Samples were obtained after informed consent after internal institutional review board approval, respecting the regulations for the protection of human subjects. The present study control samples consist of isolated peripheral blood mononuclear cells (PBMCs) from healthy HTLV-1-negative donors that have been previously reported [16]. Pharmacological inhibitors used to treat cells include STAT3 inhibitor, S3I-201 (Calbiochem), and NFAT inhibitor (Cayman Chemical Company).

RNA was extracted using TRIzol (Invitrogen) followed by DNase I treatment and reverse transcription with an RNA-to-cDNA synthesis kit (Applied Biosystems). The StepOnePlus Real-time PCR System (Applied Biosystems) was used in the study to quantify the expression of the genes of interest. The following primers were used in this study with iTaq Universal SYBR green (Bio-Rad): GAPDH (S-GAAGGTGAAGGTCGGAGTC and AS-GAAGATGGTGATGGGATTTC), STAT3 (S-GATTGACCAGCAGTATAGCCGCTTC and AS-CTGCAGTCTGTAGAAGGCGTG), pre-miR-124a (S-AGGCCTCTCTCTCCGTGTTC and AS-CAGCCCCATTCTTGGCATTC), JAG1 (S-ATCGTGCTGCCTTTCAGTTT and AS-GATCATGCCCGAGTGAGAA), JAG2 (S-GTCGTCATCCCCTTCCAGT and AS-CTCCTCATTCGGGGTGGTAT), DLL4 (S-AGGCCTGTTTTGTGACCAAG and AS-GTGCAGGTGTAGCTTCGCT), IL-8 (S-CTGATTTCTGCAGCTCTGTGTG and AS-CAGACAGAGCTCTCTTCCATCAG), RelA (S-CTCTGCTTCCAGGTGACAGT and AS-TCCTCTTTCTGCACCTTGTC), Hes1 (S-CTGGAAATGACAGTGAAGCACCT and AS-ATTGATCTGGGTCATGCAGTTG), Hey1 (S-CCGAGATCCTGCAGATGACC and AS-CCCGAAATCCCAAACTCCGA), and VEGF (S-TCTACCTCCACCATGCCAAGT and AS- GATGATTCTGCCCTCCTCCTT). NFATc1 was detected using iTaq Universal Probes (Bio-Rad) with the following primers: NFATc1 (S-CCATCCTCTCCAACACCAAA, AS-GTCTCTCCTTTCCGAAGTTCAA, and probe-ACTGTGCCGGAATCCTGAAACTCA).

Cells were collected, washed twice with PBS, and stained with the antibodies according to the manufacturer’s instructions. The samples were then washed twice with PBS before analysis with a LSR II flow cytometer. Fixation with PFA 4% was included for JAG1 antibody staining before incubation with the antibody. The following antibodies were used: FITC anti-human JAG1 (Sino Biological Inc.), FITC Mouse IgG2a isotype control (BD PharMingen), and PE anti-Human DLL4 (Biolegend).

Cells were grown on a coverslip slide coated with poly-Lysine or cytospined onto a coverslip. Cells were then fixed with 4% PFA. Immunohistochemistry (IHC) was performed using an EXPOSE HRP/DAB Detection IHC Kit (Abcam), with JAG1 (R&D Systems) and DLL4 (Abcam) antibodies, counterstaining with Mayer’s hematoxylin (Lillie’s modification) and Bluing reagent (ScyTek). Images were taken with a Nikon Eclipse 80i microscope (Nikon Instruments, Inc., Melville, NY) with a × 60 objective lens and a Nikon DSFI1 camera.

The HTLV-1 Tax gene was cloned into the pTRIPZ, lentiviral inducible vector, engineered to become Tet-On. The pTRIPZ vector contains puromycin resistance, which was used to select a stable cell line. The stable cell lines were incubated in the absence or presence of doxycycline to induce the expression of the viral protein Tax. Tax and HBZ genes were also cloned in the pSIH1-green fluorescent protein (GFP) lentiviral vector. A shRNA against JAG1 was cloned into the pSIH1-GFP vector. The pTRIPZ and miR-124a/pTRIPZ stable lines and the miR-124a/pCDNA construct are previously described [16]. Luciferase assays were performed using the Dual-Luciferase Reporter System (Promega). The wild-type 3′UTR of NFATc1 was cloned into a modified pGL3-Promoter luciferase vector (Promega) with the primers S-GGTCTAGATTGCCACATTGGAGCACTCAGTTCAGC and AS-CCGAATTCCGGCTTTATTGGATCTATTTCCTAACTAC. Mutant NFATC1 3′UTR sites were generated using the site-directed mutagenesis kit (Stratagene).

Cell lysates were separated on SDS-PAGE gels followed by electroblotting to polyvinylidene difluoride membranes and probed with Actin (sc-1615), NFATC1 (7A6), and a Tax mouse monoclonal antibody from the NIH AIDS Reagent Program, HTLV-I Tax Hybridoma (168B17), and with appropriate secondary antibodies from Santa Cruz Biotechnology.

Cell viability and proliferation were measured by Cell Proliferation Kit II (XTT) (Roche) according to the manufacturer’s instructions. One hundred microliters of cells were seeded in a 96-well plate, and 50 μL of XTT labeling mixture was added to each well and incubated for 4–6 h. Spectrophotometry was used to measure the absorbance at 450 and 620 nm. The results were plotted as mean, and the standard deviation is shown from at least two independent experiments.

Cells were plated in a 12-well plate, and when the cells reached confluence, they were treated with 3 μg/ml of JAG1-neutralizing antibody. After 3 days, the media was removed and replaced with fresh media with 3 μg/ml of JAG1-neutralizing antibody. After 3 days, the p1000 tip was used to scratch the plate. The plate was then washed twice with PBS, and fresh media with neutralizing antibody (3 μg/ml) was added. After 48 h, the cells were fixed with cold methanol (MeOH) and then stained with crystal violet dye (0.5% MeOH) for 20 min at room temperature. Images were taken with an Olympus 1x71 Inverted Microscope with a × 40 objective lens.

Experiments in figures were performed multiple times in duplicate. Representative results were shown in the final figures. P values were calculated by using paired and two-tailed Student’s t test. In the figures, asterisk indicates p value < .05, two asterisks indicate p value < .01, and three asterisks indicate p value < .001. Correlation analysis was performed by using Pearson’s correlation. The Pearson’s correlation coefficient, coefficient of determination, and p values are reported in the figures.

We used RT-PCR to test the expression of Notch receptor ligands JAG1, JAG2, DLL1, and DLL4 in HTLV-I-infected immortalized (IL-2-dependent) and transformed (IL-2-independent) cell lines compared to the HTLV-I-uninfected T cell line, Jurkat, and PBMCs isolated from healthy donors. Generally, Notch receptor ligands JAG2 and DLL1 were downregulated when compared to normal PBMCs (Fig. 1c, d). Overexpression of the Notch receptor ligand JAG1 was detected in five of six HTLV-I cell lines tested when compared to HTLV-I-negative cells. Only HTLV-I-immortalized 1185 cells did not significantly overexpress JAG1 (Fig. 1a). To confirm that the JAG1 ligand was overexpressed on the cell surface of HTLV-I-infected cells, we used JAG1 antibody staining followed by FACS analysis. Our analysis confirmed high cell surface expression of JAG1 (Fig. 1b), suggesting that it may play a role in the constitutive activation of Notch signaling in HTLV-I-infected cells. Finally, expression of the Notch receptor ligand DLL4 was variable in HTLV-I-infected cell lines compared to HTLV-I-negative cells, but was overexpressed on the cell surface of MT4 and C8166 transformed cells (Fig. 1e, f). We next investigated the expression of Notch receptor ligands JAG1 and DLL4 in a series of ATL patient-derived cell lines. These cell lines are of ATL origin and display varying levels of the HTLV-I oncoprotein, Tax (Fig. 2a). Overexpression of JAG1 was detected in seven out of ten ATL cell lines tested (Fig. 2b), and cell surface expression was confirmed by FACS and IHC analysis (Fig. 2c, d). In contrast, DLL4 was found to be overexpressed in only two ATL cell lines, ATLT and ATL25 (Fig. 2e). These results were validated by using FACS and IHC, using Jurkat cells as a negative control (Fig. 2f, g).

We next investigated the molecular mechanism associated with increased JAG1 expression in HTLV-I-transformed cell lines. The viral Tax gene encodes for a 40-kDa nuclear phosphoprotein that has pleiotropic effects on HTLV-I-infected cells [17]. Tax has been shown to positively regulate the transcription of many cellular genes. Furthermore, JAG1 expression was elevated in all ATL lines that had detectable Tax expression (Fig. 2a). JAG1 expression was lower in ED-40515(−), ATL43T, and Tl-Om1, where Tax expression from cDNA was undetectable by PCR analysis. To test whether Tax can activate JAG1 expression, we transiently expressed Tax in Molt4, an HTLV-I-negative T cell line, where Tax is not expressed. Successful delivery and expression of Tax by lentiviral transduction was confirmed by RT-PCR (Fig. 3a). Expression of JAG1 following transduction of Tax was investigated by FACS and RT-PCR. Results from these studies confirmed a strong induction of JAG1 mRNA and cell surface expression in Tax-transduced cells (Fig. 3b, c). In contrast, parallel analyses on the same transduced cells indicated that Tax had no significant effects on DLL4 expression (Fig. 3b, c). These results were also independently confirmed using an inducible Jurkat Tax Tet-On cell line (data not shown). Since HTLV-I Tax is a potent inducer of the NF-κB pathway, we investigated if Tax-mediated JAG1 overexpression occurred in an NF-κB-dependent manner. To this end, we constructed and transduced Tax-expressing MT4 cells with an IκBα (NF-κB inhibitor alpha) dominant negative mutant lentivirus able to suppress Tax-mediated NF-κB canonical activation [18]. As expected, transduction with IκBα-DN resulted in significant suppression of interleukin-8 (IL-8) mRNA expression (Fig. 3g), a well-known NF-κB target gene [19]. Consistent with a role for Tax-mediated NF-κB activation in controlling JAG1 expression, JAG1 mRNA expression was also downregulated following transduction of the IκBα dominant negative mutant (Fig. 3g). Notably, the Notch1 target gene, hairy and enhancer of split 1 (Hes1), was also significantly suppressed (Fig. 3g), suggesting that Tax-mediated JAG1 expression partly contributes to Notch signaling. In order to confirm that the IκBα-DN plasmid used could indeed block Tax transactivation of NF-κB, we transfected cells with NF-κB luciferase (Fig. 3h). Tax expression increased NF-κB luciferase, but was blocked in the presence of increasing IκBα-DN plasmid. The HTLV-I HBZ gene is reportedly expressed at the mRNA level in most ATL cells, although its expression seems quite variable among samples and laboratories [20]. We tested if HBZ might be involved in the activation of JAG1 or DLL4 expression in ATL cells. Transduction of Molt4 cells with an HBZ lentiviral vector was performed, and both JAG1 and DLL4 expression were analyzed by RT-PCR and immunostaining. Our data suggests that HBZ does not play any significant role in the expression of JAG1 and/or DLL4 in ATL cells (Fig. 3d–f).

Since Tax expression is retained in only 30% of ATL cells and HBZ did not play a significant role in activation of JAG1, we next investigated alternative pathways that could explain the high levels of JAG1 in ATL cells [21]. MicroRNAs (miRNAs) are transcriptional and post-transcriptional regulators of gene expression that are involved in many pathological conditions, including cancer [22]. A previous study identified JAG1 as a direct target of miRNA-124a [23]. Interestingly, our laboratory has shown that miR-124a is significantly downregulated in transformed cell lines and in primary ATL patient samples [16]. In addition to its direct effects on JAG1, miR-124a can also target STAT3 and NFATc1, two transcription factors known to regulate JAG1 gene expression (Fig. 4a). We have previously shown that STAT3 is a direct target of miR-124a [16]. In order to demonstrate a role for miR-124a inhibition of NFATc1, we first confirmed that NFATc1 is a direct target of miR-124a by 3′UTR luciferase assays. We cloned the full-length NFATc1 3′UTR into a pGL3-luciferase vector and showed that miR-124a overexpression led to a statistically significant loss in NFATc1 3′UTR activity (Fig. 4b). The NFATc1 3′UTR contains three miR-124a binding sites. Mutation of individual miR-124a binding sites within the NFATc1 3′UTR led to a loss of miR-124a inhibition to varying degrees (Fig. 4c). However, loss of all three miR-124a binding sites led to complete loss of miR-124a inhibition (Fig. 4c). In addition, mutated miR-124a lost the ability to inhibit NFATc1 3′UTR luciferase activity compared to wild-type miR-124a (Fig. 4b). To confirm that miR-124a can target NFATc1 protein levels, we generated miR-124a/pTRIPZ inducible 293T cells. Induction of miR-124a expression led to a loss of NFATc1 protein levels (Fig. 4d), whereas no loss was found in control, pTRIPZ, inducible cells. Since miRNA regulation of a target gene is context-dependent, we next investigated the role of miR-124a in the overexpression of JAG1 in the context of ATL cells. To this end, we used three TET-ON inducible cell lines expressing miR-124a under the control of an inducible promoter. The addition of doxycycline to the culture media efficiently induced expression of miR-124a (Fig. 4e), which was associated with a significant decrease in JAG1 levels of expression, in two out of three ATL lines (Fig. 4f). Expression of miR-124a in ATLT still led to a decrease in JAG1 expression; however, the level was not significant, possibly due to varying degrees of miR-124a expression after doxycycline induction. Finally, we found that miR-124a could not only suppress the expression of the STAT3 gene in ATL cell lines, but could also suppress NFATc1 mRNA (Fig. 4f). Consistent with these observations, miR-124a was able to inhibit both STAT3- and NFAT-dependent signaling as demonstrated by reduced luciferase activity from promoter reporter assays (Fig. 4g).

We next investigated the expression of JAG1 mRNA in freshly isolated uncultured PBMCs from acute ATL patients. Our analysis shows that DLL4 was not consistently overexpressed in patient samples (data not shown). In contrast, JAG1 was significantly overexpressed in 12/17 ATL samples (70.6%) when compared to a healthy donor PBMC (Fig. 5a). Since Tax expression is variable to undetectable in ATL patients and NF-κB can induce JAG1 expression, we then tested whether the elevated JAG1 levels in ATL patients were due to an activated NF-κB pathway. Expression of the p65 subunit of NF-κB, RelA, is one of several markers used to test for NF-κB activity. The analysis of RT-PCR expression data for RelA and JAG1 showed no correlation between RelA expression and JAG1 in ATL patients (Fig. 5b). This was in contrast to IL-8, a known downstream target of NF-κB, which strongly correlated with RelA expression in ATL patient samples (r = 0.5997). This indicates that NF-κB regulation of JAG1 in HTLV lines occurs through Tax expression and suggests that a Tax-independent mechanism elevates JAG1 RNA in ATL patients. Our previous work demonstrated that over half of ATL patients have high gene expression of STAT3 [16]. We found that NFATc1 gene expression was also high in a majority of ATL patients (data not shown). We then tested the expression of STAT3 and NFATc1 by RT-PCR and found both genes were positively correlated with JAG1 expression in ATL patients (STAT3, r = 0.433, and NFATc1, r = 0.479) (Fig. 5b). This suggests that STAT3 and NFATc1, which are at least partially regulated by miR-124a, could transcriptionally upregulate JAG1 expression in ATL patients. To further test this hypothesis, we treated an ATL line (ED-40515(−)) with specific pharmacological inhibitors to STAT3 (S3I-201) and NFAT (NFAT inhibitor, inNFAT). RT-PCR confirmed that inhibition of either the STAT3 or the NFAT pathway leads to loss of JAG1 expression in ATL lines; however, the result was only statistically significant for S3I-201 (Fig. 5c, d). The inNFAT prevents dephosphorylation of calcineurin-mediated dephosphorylation of NFAT, thereby preventing calcineurin binding. To test if the inhibitor was working, we performed RT-PCR on NFATc1 levels, since NFATc1 can auto-amplify its own transcription [24]. We found only a 40% loss in NFATc1 transcription following the addition of inNFAT to ED-40515(−) (Fig. 5d). It is possible that higher concentrations of drug are needed in order to fully downregulate the NFAT pathway in ATL cells.

In order to validate the significance of JAG1 overexpression in the activation of Notch1 signaling in HTLV-I-transformed cells, we interrupted JAG1 signaling in ATL cells. Efficient downregulation of JAG1 was obtained by lentiviral delivery of JAG1 shRNA into the ATL line, ATL55T (Fig. 6a). Suppression of JAG1 expression was associated with downregulation of the Notch target genes VEGF (vascular endothelial growth factor), Hes1, and Hey1 (hairy-related transcription factor 1) (Fig. 6a). These results confirm that in ATL cells, JAG1 signaling is involved in the activation of the Notch1 pathway. To confirm these results, we then exposed HTLV-I-transformed cells to a JAG1-neutralizing antibody, which can selectively block the binding between the ligand and its receptor. In this assay, we used Jurkat and ATL55T, which display low and high levels of JAG1 expression, respectively. Exposure of Jurkat cells to a JAG1-neutralizing antibody had no significant effect on the expression of the Notch target genes, Hes1 and Hey1 (Fig. 6b). In contrast, blockade of JAG1 by neutralizing antibody resulted in strong inhibition of Hes1 and Hey1 expression in ATL55T (Fig. 6b). Together, these results confirmed that JAG1 overexpression is partly involved in the activation of Notch signaling in ATL cells and may represent a novel therapeutic target. To test the biological significance of JAG1 in ATL cells, we measured cellular proliferation following inhibition of JAG1 signaling by a neutralizing antibody. Our data showed that short-term blockade of JAG1 for 48 h was not associated with inhibition of cellular growth for ATL55T cells (Fig. 6c). In agreement with these findings, propidium iodide staining followed by FACS analysis demonstrated no significant cell death in ATL55T or ATL43T cells following transient inhibition of JAG1 (Fig. 6d). It is possible that JAG1 was not sufficiently inhibited or that alternative pathways are used. It is also possible that inhibition for longer periods are required to significantly halt ATL cell proliferation and survival. Additional studies may be warranted to answer these questions. Since JAG1 overexpression has been implicated in metastasis, we then investigated the role of JAG1 in ATL tumor cell migration by wound assays in vitro [25]. Results presented in Fig. 6e demonstrate a reduction in wound healing 48 h after inhibition of JAG1, suggesting that in ATL cells, high JAG1 expression may contribute to tumor cell migration in vitro.