2024 Volume 47 Issue 9 Pages 1504-1510
Signal transducer and activator of transcription 3 (STAT3) is a pleiotropic factor involved in multiple vital biological processes and a key mediator of gene transcription in response to cytokines, growth factors and aberrant activation of oncogenic signaling. STAT3 has two splicing isoforms, STAT3α and STAT3β, derived from alternative splicing of exon 23 within pre-mRNA. STAT3β differs from STAT3α by replacement of 55 amino-acid residues in the C-terminal transactivation domain with 7 specific amino acids. Thus, a shorter STAT3β was originally regarded as a dominant negative isoform of STAT3α. Recently accumulating evidence from independent studies have shown STAT3 splicing isoforms confer distinct and overlapping functions in many fundamental cellular regulatory steps such as cell differentiation, inflammatory responses, and cancer progression. However, relatively little is known about the mechanisms of STAT3 pre-mRNA splicing, and it remains undiscovered which chemical compounds or bioactive substances can induce the STAT3β expression. In this study, we generated a potent reporter for detection of alternative splicing of STAT3 pre-mRNA optimized for the screening of function-known chemical library, and successfully identified entinostat, a histone deacetylase inhibitor, as a novel inducer of STAT3β through modulating mRNA splicing. Our findings demonstrate that alternative splicing of STAT3 can be regulated by a compound, providing an important clue for understanding the regulation mechanisms of the expression balance of STAT3 isoforms in a chemical biology approach. Entinostat is likely to be a promising seed compound for elucidating how the higher ratio of STAT3β expression impacts on biological responses associated with Janus kinase (JAK)/STAT3 signaling pathway.
Signal transducer and activator of transcription 3 (STAT3) is a member of STAT protein family and a multifunctional transcription factor that mediates signal transduction from the cell membrane to the nucleus in various cellular responses.1) Classically, activation of STAT3 depends on phosphorylation at the tyrosine residue 705 (Y705) by Janus kinases which are activated in response to ligands, such as interleukin (IL)-6 family cytokines, epidermal growth factor and interferons.2–4) Activated STAT3 forms homo- or heterodimers and translocates to the nucleus, and then dimeric STAT3 acts as a transcriptional activator for target genes according to respective ligands.5)
STAT3 exists in two splicing isoforms, the full-length STAT3α and the truncated STAT3β, which are generated using an alternative 3′ acceptor site in exon 23 of STAT3 pre-mRNA. STAT3α (770 amino acids) conserves 55-residue transactivation domain (TAD) at the C-terminal, whereas STAT3β (722 amino acids) has seven unique amino acids whose function remains uncertain.6,7) Consequently, due to the presence or absence of TAD, STAT3β was initially described as an impaired transcriptional factor or a dominant negative factor against signal transduction and gene expression regulated by STAT3α.7) In addition, the physiological significance of STAT3β has not been emphasized in in vivo studies, because of the lower expression levels of STAT3β compared with STAT3α.
However, several studies have demonstrated that the distinct functions of STAT3β contribute to multiple cellular processes, such as cell differentiation, embryo development, inflammatory response and cancer malignancy.8) Furthermore, since a large number of STAT3β-specific target genes have been identified, it has been thought that STAT3β is not the transcriptionally inactive isoform but a modulator of gene expression following cytokine stimulation, in its own right.9) The physiological importance of STAT3 has been shown by the early embryonic lethality of Stat3 knock-out (KO) mice, and STAT3β can rescue the embryonic lethality of STAT3α deficiency and itself induces the expression of STAT3 target genes.10,11) The expression and phosphorylation level of STAT3α and STAT3β determines granulocyte differentiation with their unique kinetic patterns.12–14) STAT3β can also functionally compensate the KO of STAT3 during astrocyte differentiation which is dependent on STAT3 activation.15) In inflammatory responses, STAT3β functions as a negative regulator of the expression of lipopolysaccharide (LPS)-responsive genes in liver, and STAT3β deficient macrophages produce more inflammatory cytokines, tumor necrosis factor α and IL-6, in response to LPS and interferon than wild-type ones.11,16) In the vast majority of studies on cancer biology, STAT3α has been established as a tumor promoter, because constitutive activation of STAT3α is associated with malignant transformation, abnormal cell proliferation, apoptotic resistance, angiogenesis and metastasis in a wide variety of cancers.17,18) Recently, STAT3β has been considered to exert the potent antitumorigenic effects for acute myeloid leukemia (AML), melanoma, esophageal squamous cell carcinoma (ESCC), breast, lung, and colon cancers.19–25) The redirection of endogenous alternative splicing from STAT3α to STAT3β using morpholino oligomers induces tumor regression in a xenograft model.26) Clinically, the higher expression of STAT3β significantly correlates with a favorable prognosis and prolongs overall survival in AML and ESCC patients.19,22,27) Accumulating evidence has indicated that STAT3β does not simply function as a truncated form of STAT3α with dominant negative effects but widely exerts its distinctive properties on various biological mechanisms with or without associated STAT3α activation.28,29)
Collectively, enhancing the expression of STAT3β by modulating endogenous splicing would regulate the biological processes and disease progression underlying the distinct functions of STAT3 splicing isoforms, therefore, the regulation of STAT3 alternative splicing has been considered to be a potential therapeutic target. However, chemical compounds which induce a preferential splicing of STAT3β are still unidentified. Here we describe the construction of a potent STAT3-splicing reporter to detect a preferential splicing of STAT3β and the identification of an inducer of STAT3β through screening of chemical libraries. Our results show that entinostat, a histone deacetylase (HDAC) inhibitor, induces STAT3β via switching STAT3 alternative splicing. As mentioned above, STAT3β has been expected as a high-potency anti-inflammatory or -tumorigenic molecule, thus the discovery of a chemical inducer of STAT3β might provide a novel seed of medicines and therapeutic targets (e.g., RNA binding proteins) associated with STAT3 alternative splicing.
The cDNA of firefly luciferase was obtained from pGL4.10[luc2] vector (Promega, Madison, WI, U.S.A.) by restriction enzyme digestion (XhoI/XbaI) and subcloned into myc-pcDNA3 vector.30) And additionally, to avoid the translation of luciferase according to unintentional open reading frame (ORF), we deleted the start codon in luc2 cDNA by performing site-directed mutagenesis with PfuUltra II Fusion HS DNA Polymerase (Agilent Technologies, Santa Clara, CA, U.S.A.) and primers (5′- CTGTTGGTAAAGCCACCGAAGATGCCAAAAACA-3′ and 5′-TGTTTTTGGCATCTTCGGTGGCTTTACCAACAG-3′) (pcDNA3-myc-Luc2). A human STAT3 genomic DNA fragment (minigene) spanning exon 22 (25 nt) to exon 24 (43 nt) was amplified by KOD FX Neo (TOYOBO, Osaka, Japan) using HeLa genomic DNA and STAT3-specific primers with overhanging restriction enzyme sites for BamHI and XhoI, respectively (5′-AAAGGATCCAGACCAAGTTTATCTGTGCGACAC-3′ and 5′-AAACTCGAGGTAGCGCACTCCGAGGTCAAC-3′) and subcloned into the site of BamHI/XhoI of pcDNA3-myc-Luc2 vector to produce pcDNA3-myc-STAT3 minigene-Luc2 (STAT3-splicing reporter). Additionally, amino acids encoded by wild-type minigene spanning exon 22 (25 nt) to exon 24 (43 nt) is not consistent with that of endogenous STAT3 in Myc-Luciferase fusion protein, and then one premature termination codon (PTC) appears in the β-type spliced transcript derived from STAT3-splicing reporter without a base substitution. Therefore, to abolish the PTC of the transcript through the process of β-type splicing, the above-described forward primer for cloning minigene was designed for a single base substitution of T to C at the seventh base from the 3′-end of exon 22 (an underlined base). Mutant STAT3 minigene with deletion of 3 bases ACC (fifth to seventh bases at 5′-end of exon 23) was constructed with QuikChange Site-Directed Mutagenesis Kit (Agilent Technologies) and primers (5′-CCCATTTCCTACAGAACGTGCAGCAATACCATTGACCTG-3′ and 5′- CAGGTCAATGGTATTGCTGCACGTTCTGTAGGAAATGGG-3′) using STAT3 minigene as a template, and cloned into pcDNA3-myc-Luc2 vector, similarly (STAT3 (ΔACC)-splicing reporter). All constructs were confirmed by sequencing.
Cell Culture, Transfection and Stable Cell Line GenerationK562 cells were cultured in RPMI 1640 (Nacalai Tesque, Kyoto, Japan), supplemented with 10% fetal bovine serum (Sigma-Aldrich, St. Louis, MO, U.S.A.), 100 units/mL penicillin and 100 µg/mL streptomycin (Nacalai Tesque). K562 cells were transfected with pcDNA3-myc-Luc2, STAT3-splicing reporter or STAT3 (ΔACC)-splicing reporter by electroporation using NEPA Type II (Nepa Gene, Chiba, Japan). At 24 h after electroporation, these cells were cultured with 0.75 mg/mL G418 (Nacalai Tesque) for over 2 weeks. All cells were maintained at 37 °C in 5% CO2.
Chemical Compounds and AntibodiesA total of approximately 2160 compounds for screening were provided by the Medical Research Support Center, Graduate School of Medicine, Kyoto University. Entinostat/MS-275 and romidepsin/FK228 were purchased from Cayman Chemical (Ann Arbor, MI, U.S.A.). Vorinostat/suberoylanilide hydroxamic acid (SAHA), Trichostatin A (TSA) and RGFP966 were purchased from InvivoGen (San Diego, CA, U.S.A.), FUJIFILM Wako Pure Chemical Corporation (Osaka, Japan), and Sigma-Aldrich, respectively. Anti-Stat3 (D3Z2G) rabbit monoclonal antibody (#12640), anti-β-actin (13E5) rabbit monoclonal antibody (#4970) and anti-rabbit immunoglobulin G, horseradish peroxidase linked antibody (#7074) were obtained from Cell Signaling Technology (Danvers, MA, U.S.A.).
Cell-Based Screening of Function-Known Chemical Library10000 K562 cells expressing each reporter in 50 µL medium (phenol red free) per well were seeded on white 96-well plates. For each plate, STAT3-splicing reporter expressing K562 cells were in all columns except the last one, and the first column was assigned as controls treated with dimethyl sulfoxide (DMSO). pcDNA3-myc-Luc2 or STAT3 (ΔACC)-splicing reporter expressing K562 cells were in the upper or lower half of the last column, respectively. The former was used for background correction and the latter was for positive controls. These controls in the last column were also treated with DMSO. Compounds diluted in the same medium were administered to the rest of wells, and the plates were incubated for 24 h. Subsequently, the mixture of Passive Lysis Buffer (Promega) and luciferase assay substrate solution (E151A and E152A, Promega) was added to wells, and luminescence signals were measured using Fluoroskan Ascent FT (Thermo Fisher Scientific, Waltham, MA, U.S.A.).
RNA Extraction, Semiquantitative RT-PCR and Quantitative PCR AnalysisTotal RNA was extracted using TRIzol reagent (Invitrogen, Waltham, MA, U.S.A.) and purified using RQ1 RNase-Free DNase (Promega). Reverse transcription was performed with ReverTra Ace qPCR RT Master Mix (TOYOBO). Semiquantitative RT-PCR was carried out with SapphireAmp Fast PCR Master Mix (TaKaRa Bio, Shiga, Japan) and PCR products were applied to polyacrylamide gel electrophoresis. Real-time RT-PCR was performed with PowerUP SYBR Green Master Mix (Applied Biosystems, Waltham, MA, U.S.A.). Fluorescent detection and data analyses were done using CFX Connect Real-Time PCR System (BIO-RAD, Hercules, CA, U.S.A.). Primer information is provided in Supplementary Materials.
ImmunoblotCells were harvested with radio-immunoprecipitation assay (RIPA) buffer (50 mM Tris–HCl (pH 7.5), 150 mM NaCl, 0.5% sodium deoxycholate, 0.1% sodium dodecyl sulfate (SDS), 1 mM ethylenediaminetetraacetic acid (EDTA), 1% NP40 substitute) containing a protease inhibitor cocktail (Nacalai Tesque), and then were sonicated. Lysates were processed with SDS sample buffer and applied to SDS-polyacrylamide gel electrophoresis (PAGE). After electrophoresis separation, the proteins were transferred onto a polyvinylidene difluoride (PVDF) membrane. Immunoblotting was performed with the antibodies described above. Chemiluminescence was detected with Chemi-Lumi One Super kit (Nacalai Tesque), and luminescence images were analyzed by Amersham Imager 680 (GE Healthcare Life Sciences, Marlborough, MA, U.S.A.).
Statistical AnalysisAll statistical tests were performed in GraphPad Prism 10. p < 0.05 was considered statistically significant.
To identify small molecules that efficiently induce STAT3β through modulating alternative splicing of STAT3 pre-mRNA, we constructed a firefly luciferase-based splicing reporter that reflects the use of the splice site at the first 50 nucleotides (nt) of exon 23, namely β-type splicing (STAT3-splicing reporter). In this reporter, Myc-derived peptide coding sequence and the human STAT3 gene fragment (minigene) spanning exon 22 (25 nt from its 3′-end) to exon 24 (43 nt from its 5′-end) were fused upstream of the cDNA of luciferase (details are described in Materials and methods). With STAT3-splicing reporter system, β-type splicing leads to expression of Myc-Luciferase fusion protein without any stop codons. In contrast, α-type splicing abrogates the expression of Myc-Luciferase fusion protein because 50 nt-frameshift brings out-of-frame and PTC to reporter’s transcript (Fig. 1A). Incidentally, the minigene subcloned into splicing reporters is not consistent with the ORF of endogenous STAT3. Due to this reporter construction, several PTCs appears in the α-type spliced transcript including the first 50 nt of exon 23 and the frameshift leads to the out-of-frame of luciferase. We also generated the mutant splicing reporter with deletion of ACC, 5–7th bases at 5′-end of exon 23 in human STAT3 minigene, for using positive control of β-type splicing (STAT3 (ΔACC)-splicing reporter), with reference to Zammarchi et al.’s research paper which shows that ΔACC deletion mutant markedly induces a switch from STAT3α to β.26)
(A) Schematic diagram of STAT3-splicing reporter. (B) RT-PCR analysis of total RNA extracted from K562 parental and reporter stable cell lines expressing the pcDNA3-myc-Luc2, STAT3-splicing reporter or STAT3 (ΔACC)-splicing reporter. (C) Relative luciferase activity of K562 cells expressing those reporters, versus STAT3-splicing reporter. The error bars mean ± S.D. p-Value was evaluated by Student’s t-test.
To evaluate whether STAT3-splicing reporter recapitulates the expression level of endogenous STAT3 mRNA, and STAT3 (ΔACC)-splicing reporter produces more β-type spliced mRNA in K562 cells, we established K562 cells stably expressing each reporter and determined splicing patterns of transcripts derived from reporters by performing RT-PCR (Fig. 1B). The results showed that the PCR products derived from STAT3-splicing reporter mimicked the mRNA expression levels of endogenous STAT3α and STAT3β (lane 3), and that ΔACC deletion mutant in splicing reporter increased the ratio of the transcript through β-type splicing (lane 4). In addition, the expression levels of endogenous STAT3α and STAT3β were not affected by the introduction of splicing reporters (lanes 1–4). Then to examine whether there is a correlation with the expression level of β-type spliced transcript and its translated product, Myc-Luciferase fusion protein, we performed luciferase assay using K562 cells stably expressing those reporters. Luciferase activity theoretically could represent the expression levels of Myc-Luciferase fusion protein derived from the transcript via β-type splicing. As expected, an increase in transcript via β-type splicing elevated luciferase activity (Fig. 1C).
Screening of Chemical Compounds for Identification of β-Type Splicing InducersTo identify chemical compounds promoting β-type splicing, we performed luciferase assay-based screening of function-known chemical library (approximately 2160 compounds) using the established K562 cells stably expressing STAT3-splicing reporter, STAT3 (ΔACC)-splicing reporter (as a positive control) or pcDNA3-myc-Luc2, reporter backbone without insertion of minigene (for background correction) (Fig. 2A). In the first screening whose Z’-factor is 0.41, we selected 95 compounds inducing luciferase activity than mean of positive controls minus 1.75-fold standard deviation (S.D.) at concentration of 10 µM (Fig. 2B). Next, to narrow candidates, we additionally tested dose-dependency of those compounds for luciferase activity (one example is shown in supplementary Fig. 1), and then inspected the reporter-produced transcripts undergone β-type splicing by real-time (data not shown). As a result of our sequential evaluation based on reporter system, entinostat (MS-275), a selective class I HDAC inhibitor, was extracted as a candidate of β-type inducer.
(A) Workflow for screening. Approximately 2160 compounds were screened with K562 cells expressing STAT3-splicing reporter. Experimental details are described in Materials and Methods. (B) Scatter plots representative the activity of firefly luciferase. Black and gray plots showed the luciferase activity of STAT3-splicing reporter-cells with the stimulation of test compounds. Blue and red plots indicated the luciferase activity of STAT3- and STAT3 (ΔACC)-splicing reporter cells under the treatment of DMSO in all tested 96 well-plates, respectively. Purple plot represented that of entinostat (indicated by a red arrow).
To validate the effect of entinostat on the induction of endogenous STAT3β, we analyzed the expression levels of STAT3 splicing isoforms, performing semiquantitative RT-PCR, real-time quantitative (q)PCR and immunoblotting. Consequently, entinostat significantly upregulated endogenous mRNA expression and protein levels of STAT3β in K562 cells at a concentration of 2.5 µM but not of STAT3α (Figs. 3A–D). Whereas 5 µM entinostat increased the mRNA expression of STAT3α (approximately 1.5-fold) not only STATβ compared with control, thus entinostat whose a concentration more than 5 µM is likely to affect the transcriptional activation of STAT3 gene in addition to enhancing β-type splicing in STAT3β induction. An impact of entinostat on alternative splicing of STAT3 would precede the transcriptional activation of STAT3 gene in STAT3β induction at least with the stimulation of 2.5 µM entinostat.
The expression of mRNA and proteins of STAT3 splicing isoforms in K562 cells treated with DMSO (as control) or entinostat (2.5, 5 µM) for 24 h were assessed by semiquantitative RT-PCR (A), real time RT-qPCR (B) and immunoblot analysis (C), respectively. Results of RT-qPCR were normalized to the expression of Cyclophilin A and quantified using the ΔΔCT method. (D) Densitometry analysis of STAT3α and STAT3β from immunoblots (C) was carried out using ImageJ software. The histogram shows the mean levels of STAT3β/α ratio relative to control. The error bars mean ± S.D. p-Values were evaluated by one-way ANOVA followed by Dunnett’s multiple comparisons test. (p-values, N.S.: not significant, versus control.)
Entinostat is one of the selective inhibitors of HDAC1, HDAC2 and HDAC3 categorized as class I HDACs.31) To examine whether STAT3β expression is dependent on HDAC inhibitory effects, we analyzed the expression levels of STAT3α and STAT3β mRNA by RT-qPCR analysis in K562 cells stimulated with SAHA, TSA, RGFP966 and romidepsin, representatives of HDAC inhibitors whose IC50 values determined from cell-free assay are lower than that of entinostat.31–33) As shown in Fig. 3, entinostat at a concentration of 2.5 µM significantly induced the mRNA expression of STAT3β but not STAT3α, therefore the concentration of those HDAC inhibitors was determined to be 2.5 µM, except for romidepsin (0.25 µM). The results showed that both STAT3α and STAT3β mRNA were induced by treatment with SAHA, TSA and romidepsin except for RGFP966 with a similar proportion (Fig. 4A). In addition, the protein levels of STAT3α and STAT3β in K562 cells treated with the same as above were assayed by immunoblotting. SAHA, TSA and romidepsin increased both STAT3α and STAT3β protein levels but not entinostat, which showed a correlation with the results of mRNA analysis (Fig. 4B). These results indicate that the increase of STAT3β under stimulation of SAHA, TSA and romidepsin predominantly depend on the transcription of STAT3 gene without changing the balance of STAT3 splicing isoforms.
(A) STAT3α and STAT3β mRNA in K562 cells treated with the indicated HDAC inhibitors (2.5 µM, except for romidepsin whose concentration was 0.25 µM) for 24 h were measured by RT-qPCR analysis. Cyclophilin A was used for the normalization of target genes, and the results were quantified using the ΔΔCT method. The error bars mean ± S.D. p-Values were evaluated by one-way ANOVA followed by Dunnett’s multiple comparisons test. (p-values, N.S.: not significant, versus control.) (B) The protein levels of STAT3α and STAT3β in K562 cells treated with the indicated HDAC inhibitors (2.5 µM, except for romidepsin whose concentration was 0.25 µM) for 24 h were assayed by immunoblotting.
The development of STAT3 inhibitors has been well-studied, because STAT3 is considered to play a pivotal role in cancer malignancy. But STAT3 also functions as a molecule essential to inflammatory and immune responses, it is often thought that the administration of STAT3 inhibitor could lead side effects on the immune system. In addition, the STAT3α-to-β splicing switch inhibits tumorigenicity more than total STAT3 knockdown.26) With respect to STAT3 splicing switch, the introduction of morpholino oligomers or A-to-I RNA editing have been reported as methods for modulating of STAT3 alternative splicing,26,34,35) whereas the application of antisense oligonucleotides or RNA editing could bring risks of immunogenicity and off-target effects, especially former often led translation inhibition and nonsense mediated mRNA decay. Thereby the increasing STAT3β/α ratio by splicing redirection with chemical compounds is considered effective regarding the avoidance of the above concerns.
As chemical splicing modulators, spliceostatin A, E7107 and H3B-8800, the representative SF3B inhibitors, are known to target the spliceosome containing SF3B, a core component of U2 snRNP, resulting in impairment of the assembly of the spliceosome through a common pharmacophore consisting of a conjugated diene.36) The chemical structure of entinostat does not contain a conjugated diene, suggesting that the impact of entinostat on STAT3 splicing could not depend on the inhibitory effect on SF3B.
In this study, we established a potent reporter of STAT3 splicing and identified entinostat as the first compound to induce STAT3β via splicing switch. Entinostat might directly or indirectly regulate the STAT3 pre-mRNA recognized by the spliceosome, and it would be interesting to find out entinostat-targeted RNA binding proteins and to unveil the detail mechanisms associated with ones as a future experiment.
This work was supported by JSPS KAKENHI (JP26870308 to S.M.), JST the establishment of university fellowships towards the creation of science technology innovation (JPMJFS2146), JST SPRING (JPMJSP2101) and partially Research Support Project for Life Science and Drug Discovery (Basis for Supporting Innovative Drug Discovery and Life Science Research (BINDS)) from AMED under Grant Number JP23ama121034. We thank Saya Oshizuki, Chizuru Tsuzuki and Asuka Hata for technical supports.
S. M. conceived the project and designed the research. M. K. and S. M. performed experiments. S. M., N. K., and K. S. contributed to design experiments. M. K., S. M., and K. S. performed data analysis. S. M. wrote the original draft, and N. K. and K. S. reviewed and edited the draft.
The authors declare no conflict of interest.
This article contains supplementary materials.
