At High Levels, Constitutively Activated STAT3 Induces Apoptosis of Chronic Lymphocytic Leukemia Cells

Chronic lymphocytic leukemia (CLL) is characterized by a gradual accumulation of monoclonal, dysfunctional mature-appearing lymphocytes (1). Traditionally, CLL cells were thought to have a prolonged lifespan because of a deregulated apoptosis pathway (2–4). Studies performed during the past decade revealed that CLL cells proliferate and undergo spontaneous apoptosis. Ki-67, typically expressed in proliferating cells, was found in an appreciable number of CLL bone marrow cells, and TUNEL also revealed apoptosis in CLL bone marrow cells (5). Deuterated water studies showed that 0.1–1.75% of the CLL clone is regenerated daily (6). However, the increment in peripheral blood (PB) CLL cell counts is typically lower than is expected from the cells’ proliferation rate. Therefore, it was assumed that CLL cell proliferation is accompanied by spontaneous apoptosis.

STAT3 is a latent cytoplasmic transcription factor that upon phosphorylation and dimerization shuttles to the nucleus, binds to DNA (7), and activates STAT3-regulated genes. In various solid tumors and hematologic malignancies, including CLL, STAT3 is constitutively activated and provides neoplastic cells with proliferation signals and a survival advantage (8–17).

Because in CLL cells, STAT3 is constitutively phosphorylated on serine 727 residues and activates antiapoptotic genes (10, 12–14), we hypothesized that high STAT3 levels inversely correlate with the apoptosis rate in CLL cells. Surprisingly, we found that when STAT3 levels were sufficiently high, STAT3 no longer protected CLL cells from apoptosis. Instead, STAT3 induced the expression of proapoptotic genes, activated the caspase-3 gene promoter, and induced apoptosis in CLL cells.

PB samples were obtained from 64 patients with CLL who were treated at The University of Texas MD Anderson Cancer Center Leukemia Clinic between April 2008 and May 2011, after approval was obtained from the Institutional Review Board, and written informed consent was obtained from the patients. PB counts of 320 consecutive patients with CLL were evaluated to determine the means ± SD of lymphocyte counts in the 10% of patients with the highest counts (n = 32; mean 140,000 ± 49,738 × 10⁹/l) and the 10% of patients with the lowest counts (n = 32; mean 12,800 ± 4,564 cells × 10⁹/l). Of the 64 patients, 54 (84%) had not received any prior treatment for CLL. The remaining 10 previously treated patients were evenly distributed between the subgroups of patients with a high or low lymphocyte count. The clinical characteristics of all the patients are depicted in Supplemental Table I.

PB cells were fractionated using Ficoll Hypaque 1077 (Sigma-Aldrich, St. Louis, MO). The low-density cellular fraction was used immediately or frozen for additional studies.

Western immunoblotting was performed as previously described (18). Briefly, CLL cell extract was prepared. The protein concentration was determined using a Micro BCA protein assay reagent kit (Thermo Scientific, Pierce, Rockford, IL). Cell lysates were denatured and, following electrophoresis, transferred to a nitrocellulose membranes. The membranes were incubated with monoclonal mouse anti-human STAT3 (BD Biosciences, Palo Alto, CA), polyclonal rabbit anti-human phosphoserine (serine 727) STAT3 (Cell Signaling Technology, Beverly, MA), caspase-3 (Cell Signaling Technology), cleaved caspase-3 (BD Biosciences), monoclonal STAT1 (BD Biosciences), phosphoserine STAT1 (serine 727) (BD Biosciences), or monoclonal mouse anti-human β-actin (Sigma-Aldrich) HRP-conjugated secondary Abs (GE Healthcare, Amersham, Buckinghamshire, U.K.). Proteins were visualized via an ECL detection system (GE Healthcare).

Densitometry analysis was performed using an Epson Expression 1680 scanner (Epson America, Long Beach, CA). Densitometry values were normalized by dividing the numerical value of each sample signal by the numerical value of the signal from the corresponding actin protein levels used as loading control.

The rate of cellular apoptosis was analyzed using double staining with a Cy5-conjugated Annexin V kit and propidium iodide (PI; BD Biosciences) according to the manufacturer’s instructions. After incubation for 10 min in the dark at room temperature, the samples were analyzed on an FACSCalibur flow cytometer (BD Biosciences). Cell viability was calculated as the percentage of Annexin V–positive cells.

After thawing in hot water, cells were washed twice with RPMI 1640 (Life Technologies), and TRIzol (Invitrogen, Carlsbard, CA) was added. The RNA was isolated using an RNeasy purification kit (Qiagen, Valencia, CA). RNA quality and concentration were analyzed with a NanoDrop spectrophotometer (ND-1000; NanoDrop Technologies, Wilmington, DE).

We used 500 ng total RNA in one-step quantitative RT-PCR (qRT-PCR; Applied Biosystems, Foster City, CA) analysis with an ABI Prism 7700 sequence detection system (Applied Biosystems) using a TaqMan gene expression assay for MAPK 8, KRAS, phospholipase Cγ2, p-ERK, calpain9, and caspase-3 according to the manufacturer’s instructions. Samples were run in triplicate, and relative quantification was performed by using the comparative threshold cycle method (13).

Hybridization was performed with Human Gene 2.0 ST arrays (Affymetrix, Santa Clara, CA) according to the manufacturer’s instructions. Briefly, 100 ng total RNA from each sample was reverse transcribed to cDNA, followed by overnight in vitro transcription to generate cRNA, which was also reverse transcribed. The quality of cDNA and fragmented cDNA was assessed using an Agilent Bioanalyzer system. Microarrays were hybridized, washed, and stained, and images were scanned using the Affymetrix GeneChip Command Console and analyzed with the Affymetrix Expression Console. The t test with Welch correction (variance not assumed equal) was used to compare gene expression levels, and the analysis was done using Partek Genomics Suite 6.6 (Partek, St. Louis, MO). Network analysis was performed on 1001 genes for which levels were upregulated (p < 0.01 in samples with at least a 1.5-fold increase in lymphocyte counts) using the Ingenuity pathway analysis (Ingenuity Systems, www.ingenuity.com). The data are provided at http://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE68721.

Levels of cleaved poly (ADP-ribose) polymerase (PARP) were determined using an Invitrogen Cleaved PARP ELISA kit (Invitrogen) according to the manufacturer’s instructions. Briefly, microtiter wells were coated with mouse anti-rabbit uncleaved (116 kD) or cleaved (85 kD) PARP Abs. Cleaved PARP levels were assessed by comparing the color intensity of the tested samples with that of the standard cleaved PARP protein levels provided in the kit using a microplate autoreader (Model #EL309; Bio-Tek Instruments, Winooski, VT).

The human STAT3 sequence was obtained from GenBank NM_139276, and its full-length coding sequence was generated via PCR from the human STAT3 cDNA clone (OriGene, SC124165). The primers used for generation of rSTAT3 were: STAT3-Full, 5′-GCGGCGGCGGCCGCCCCGCCGCCACCATGGCCCAATGGAATCAGCTACAG-3′ and STAT3, 3′-GCGGCGCTCGAGTAACATGGGGGAGGTAGCGCAC-5′. The sense primer contained a NotI site, a Kozac codon, and a gene-specific sequence (starting from ATG), and the antisense primers contained an XhoI site and a coding sequence ending at 2313 bp with a mutant stop codon. The PCR products were digested with NotI and XhoI and subcloned into the mammalian expression vector pBudCE4.1 (Invitrogen). GFP was cloned into another cloning site with HindIII/XbaI. MM-1 cells were transfected via electroporation with a Gene Pulser Xcell electroporation system (Bio-Rad, Hercules, CA) set at 120 mV with one pulse. Transfection efficiency was detected using fluorescence microscopy and flow cytometry. After the transfection, cells were incubated for 24 or 48 h with 50 ng/ml recombinant human IL-6 (Life Technologies, Frederick, MD) and harvested for qRT-PCR analysis.

Full-length human STAT3 was amplified from the human STAT3 cDNA clone (SC124165; OriGene, Rockville, MD), cloned using SalI/EcoRV into pENTR11(Invitrogen), and sequence-verified. Both GFP and STAT3 were transferred into pInducer20 (plasmid 44012; Addgene, Cambridge, MA), and after overnight infection, protein expression was induced using 1 μg/ml doxycycline (Sigma-Aldrich). The lentiviral vector was produced by transfecting the plasmid into HEK293T cells using the Lipofectamine 2000 transfection reagent (Invitrogen). pINDUCER20 STAT3 and the GFP expression vectors were transfected with the packaging vectors pMDLg/pRRE, pRSV-Rev, and pMD2.G (Addgene). HEK293T cells were expanded in culture, and their supernatant was used to infect the CLL cells as previously described (10).

Caspase-3 promoter fragments were transfected into MM-1 cells via electroporation as described above. Each construct included a luciferase reporter gene 770 bp upstream to the transcription starting site of the caspase-3 gene that did not include γ-IFN activation sequence (GAS)–like elements or 854 bp upstream to the caspase-3 gene transcription starting site that included a single GAS-like element. The luciferase activity of unstimulated or IL-6–stimulated MM-1 cells was assessed 24 h after transfection using a Dual-Luciferase Reporter Assay System (Promega) and a Sirius luminometer V3.1 (Berthold Detection Systems, Pforzheim, Germany). The luciferase activity of each of the human caspase-3 promoter constructs was determined by calculating the constructs’ luciferase activity relative to the activity of the Renilla luciferase produced by the PGL4-17 control vector.

A chromatin immunoprecipitation (ChIP) assay was performed using a SimpleChIP Enzymatic Chromatin IP Kit (Cell Signaling Technology, Boston, MA) according to the manufacturer’s instructions. Briefly, cells were cross-linked with 1% formaldehyde for 10 min at room temperature and then harvested and incubated on ice for 10 min in lysis buffer. Nuclei were pelleted and digested with micrococcal nuclease. Following sonication and centrifugation, sheared chromatin was incubated with anti-STAT3 or rabbit serum (negative control) overnight at 4°C. Then, protein G beads were added, and the chromatin was incubated for 2 h in rotation. Ab-bound protein–DNA complexes were eluted and subjected to PCR analysis. The primer sets used to amplify the caspase-3 promoter putative STAT3 binding sites were as follows: set 1, 5′-914TCCCAACAGCCGGCTTAA-3′ and 3′-849AAGAAGCCTGGTTTGGC-5′; set 2, 5′-1172CACGACCCAAAATAGGAA-3′ and 3′-1047GCTTGTTGGCAAATGCCT-5′; set 3, 5′-1299GTCCCTGAATCTGACTTC-3′ and 3′-1177CATTTCAGACCCTGAAGC-5′; and set 4, 5′-1706GTAGCTGGGATTACAGGT-3′ and 3′-1625CAAGATGGTGAAACCCTG-5′.

Nondenatured cellular nuclear extracts were prepared using a NE-PER extraction kit (Thermo Scientific Pierce, Rockford, IL). Nuclear protein extracts were incubated with biotin-labeled caspase-3, c-myc, and STAT3 promoters’ DNA probes in binding buffer for 30 min on ice. All probes were synthesized by Sigma-Genosys (The Woodlands, TX). Following incubation, the samples were separated on a 5% polyacrylamide gel, transferred onto a nylon membrane, and fixed on the membrane via UV cross-linking. The biotin-labeled probe was detected with strepavidin-HRP (Gel-Shift Kit; Panomics, Fremont, CA). The control consisted of 7-fold excess unlabeled cold probe. To test the effect of STAT3, anti-STAT3 Abs (BD Biosciences) or mouse IgG1 (BD Biosciences) were added with the nuclear extracts (10, 19).

Because in CLL STAT3 is constitutively phosphorylated on serine 727 residues (10), and because STAT3 promotes the proliferation and survival of cancer cells, including CLL cells (20, 21), we assumed that CLL cells from patients with a high tumor burden would harbor high levels of STAT3. To test this hypothesis, we obtained PB cells from patients with a low (n = 32) and high (n = 32) lymphocyte count. Compared with patients with low lymphocyte counts (mean: 12,800 ± 4,654 × 10⁹/l), patients with high counts (mean: 145,000 ± 49,738 × 10⁹/l) had advanced disease, high levels of β₂-microglobulin levels, and unmutated IgH gene status (Supplemental Table I). Western blot analysis revealed that CLL cells from patients with a high lymphocyte count harbored higher levels of total STAT3 compared with patients with low counts (p = 0.03), whereas overall the levels of serine p-STAT3 were similar in patients with high and low counts as assessed by densitometry (Fig. 1A). High lymphocyte counts correlated with high levels of STAT3 (r_s = 0.5; p < 0.0001) (Fig. 1B). Because STAT1 is constitutively phosphorylated on serine 727 residues (22) and induces apoptosis in CLL cells (23), we assessed the levels of STAT1 and pSTAT1 in patients with high (n = 4) and low (n = 4) lymphocyte counts. Contrary to STAT3, levels of STAT1 and serine p-STAT1 were similar in patients with high and low lymphocyte counts (Fig. 1C).

Previous studies showed that ∼1% of the entire CLL clone divides daily (24). Because the actual increase in the PB lymphocyte count is lower than the expected lymphocyte count estimated using the CLL cell proliferation rate, we hypothesized that CLL cell proliferation is associated with concomitant spontaneous apoptosis. Using Annexin V and PI staining, we assessed the spontaneous apoptosis rate of fresh PB low-density cells obtained from patients with CLL and found that CLL cell apoptosis rates correlated with WBC counts (r_p = 0.88; p = <0.0001) (Fig. 2A). For example, the spontaneous apoptosis rate of CLL PB low-density cells (>98% lymphocytes) from a patient with a WBC count of 16 × 10⁹/l was 23%, whereas the spontaneous apoptosis rate of CLL PB cells from a patient with a WBC count of 101 × 10⁹/l was 65% (Fig. 2B). Because similar experiments (data not shown) revealed a similar trend, we sought to determine whether high lymphocyte counts correlated with a high apoptosis rate.

To test our hypothesis, we first analyzed the gene signature profile of CLL cells from 9 patients with high (n = 4) and low (n = 5) lymphocyte counts. The same trend in gene expression was validated by qRT-PCR, and pathway analysis revealed that the apoptosis signaling pathway was enriched with genes for which levels were high in CLL cells from patients with high lymphocyte counts (p = 0.02). Specifically, caspase-3, MAPK 8, KRAS, phospholipase Cγ2, protein kinase C, calpain 9, and caspase-3 were incorporated in this pathway (Supplemental Fig. 1).

Taken together, these results suggest that the apoptosis signaling pathway is activated in CLL cells from patients with high lymphocyte counts and that those cells undergo spontaneous apoptosis at a high rate. To further delineate these findings, we obtained cells from patients with CLL with high (n = 16) and low (n = 16) lymphocyte counts and, using ELISA, assessed the levels of cleaved PARP. PARP, known to protect cellular integrity, is cleaved and inactivated by cleaved caspase-3. We found that the median level of cleaved PARP was significantly higher in cells from patients with high lymphocyte counts (median 0.03; range 0.02–0.07) than in cells from patients with low lymphocyte counts (median 0.06; range 0.03 - 0.46), (p < 0.007) (Fig. 3).

Although STAT3 is a transcription factor that provides CLL cells with a survival advantage, and its levels are high in lymphocytes from patients with CLL with high lymphocyte counts (Fig. 1), CLL cells from patients with high lymphocyte counts undergo spontaneous apoptosis at a higher rate than cells obtained from patients with low lymphocyte counts who harbor lower STAT3 levels (Figs. 2, 3). To determine whether high levels of STAT3 induce apoptosis, we transfected MM-1 cells with a vector containing the rSTAT3 DNA construct and used Annexin/PI staining to assess the cellular apoptosis rate. As shown in Fig. 4A, 85% of cells transfected with STAT3 underwent apoptosis, as opposed to 59% of cells that were transfected with the empty vector (44% difference).

Because the protein levels of caspase-3, an essential proapoptotic protein activated downstream of both the intrinsic and extrinsic apoptotic pathways, were significantly increased in CLL cells from patients with a high lymphocyte count, and overexpression of STAT3 in MM-1 cells induced apoptosis (Fig. 4A), we sought to determine whether there is a cause-and-effect association between STAT3 and caspase-3. Therefore, we transfected MM-1 cells with rSTAT3 DNA and using qRT-PCR assessed STAT3 and caspase-3 RNA levels 24 h after the transfection. Transfection of MM-1 cells with STAT3 induced a 7.5-fold increase in STAT3 expression levels and a 2.5-fold increase in caspase-3 expression levels (Fig. 4B). Similarly, infection of CLL cells with a lentivirus harboring the full-length human STAT3 induced an increase in STAT3 and caspase-3 expression, an increase in STAT3 and cleaved caspase-3 protein levels (Fig. 4C), and an increase in apoptosis rate of CLL cells (Fig. 4D).

Because overexpression of STAT3 resulted in increased caspase-3 RNA levels, we sought to determine whether STAT3 directly activates the caspase-3 gene. We conducted a sequence analysis of the caspase-3 gene promoter and identified within the promoter six GAS-like elements thought to be STAT3 binding sites (Fig. 5A, top panel). Then, to determine whether STAT3 activates the caspase-3 promoter, we transfected MM-1 cells with the luciferase reporter gene driven by truncated fragments of the caspase-3 promoter. Because in unstimulated MM-1 cells STAT3 is not activated (25), we incubated the cells with IL-6 that induces phosphorylation of STAT3 on tyrosine 705 residues (10, 26, 27). IL-6–induced luciferase activity was detected in the 854-bp fragment that harbors putative STAT3 binding sites but not in shorter fragments lacking those sites (Fig. 5A, bottom panel). Then, using ChIP, we found that DNA fragments detected by primers 1, 2, and 4, but not primer 3, each corresponding to a single GAS-like element (Fig. 5A), coimmunoprecipitated with STAT3 (Fig. 5B), suggesting that STAT3 binds to sites 1, 2, and 4 (but not site 3) of the caspase-3 promoter. To confirm that STAT3 binds to the caspase-3 gene promoter, we performed EMSA using a biotinylated DNA probe of binding site 1 of the caspase-3 promoter. MM-1 nuclear protein extract bound to and formed a complex with the caspase-3 DNA promoter, and the binding was completely reversed by the addition of excess unlabeled probe or attenuated by anti-STAT3 Abs, whereas no binding was detected when a biotinylated mutated site 1 DNA probe was used (Fig. 5C).

To validate that STAT3 binds to the caspase-3 gene promoter in CLL cells as well, we incubated nuclear extracts of PB CLL cells from two patients with high lymphocyte counts with the biotinylated DNA probe of binding site 1 as described in the previous section. CLL cell nuclear extracts bound to and formed complexes with the caspase-3 DNA promoter fragment, and the binding was completely reversed by the addition of excess unlabeled probe (Fig. 6A). Unlike the isotype control IgG, anti-serine p-STAT3 Abs attenuated the binding, and no binding was detected when a biotinylated mutated site 1 DNA probe was used instead of the site 1 probe. In addition, using ChIP, we found that CLL cell chromatin fragments that were immunoprecipitated with anti-STAT3 Abs coimmunoprecipitated with DNA of caspase-3 and the STAT3-regulated genes STAT3, p21, c-Myc, retinoic acid–related orphan receptor 1, and vascular endothelial growth factor-C (Fig. 6B). Taken together, these data suggest that STAT3 binds to the caspase-3 promoter.

Like in other cancers, in CLL, STAT3 provides the neoplastic cells with a survival advantage (10). However, overexpression of STAT3 induced a significant increase in caspase-3 expression and induced apoptotic cell death. Furthermore, CLL cells from patients with a high, but not low, lymphocyte count expressed high levels of STAT3 and underwent spontaneous apoptosis at a high rate. Therefore, using Western immunoblotting, we assessed the levels of caspase-3 in CLL cells from patients with high and low lymphocyte counts and found that caspase-3 and cleaved caspase-3 levels were significantly higher in cells from patients with high counts (Fig. 6C). We then plotted the levels of caspase-3 as a function of the levels of total STAT3 and found that caspase-3 remained constant across a wide range of STAT3 levels and increased only when a threshold of STAT3 protein level was reached (Fig. 6D), suggesting that the binding affinity of p-STAT3 to the caspase-3 promoter is relatively weak and that p-STAT3 activates the transcription of caspase-3 only when its levels are sufficiently high.

To test this hypothesis, we first used ChIP. Serial dilutions of CLL cell DNA of chromatin fragments coimmunoprecipitated with anti-STAT3 Abs were prepared, and STAT3 target genes were detected via PCR. As opposed to p21, c-Myc, or STAT3, caspase-3 was detected only in the least diluted fraction, suggesting that STAT3 binds to the caspase-3 promoter with a low affinity (Fig. 6E). We also analyzed the DNA that coimmunoprecipitated with STAT3 by using qRT-PCR. High c-Myc, p21, and STAT3 but low caspase-3 levels were detected. As in MM-1 cells, DNA of the Caspase-3 promoter was detected using primer sets 1, 2, and 4 but not with primer set 3 (Fig. 6F). These results confirmed that STAT3 binds to the caspase-3 promoter with low affinity.

To further delineate these findings, we performed EMSA. Using serial dilutions of CLL cell nuclear extracts, we found that the binding of those extracts to the biotinylated caspase-3 promoter DNA probe was low compared with the binding of c-Myc or STAT3 promoter probes across all tested dilutions (Fig. 6G). Furthermore, the binding of nuclear extracts of CLL cells from patients with low lymphocyte counts (n = 4) was significantly lower than that of patients with high counts (n = 4) (Fig. 6H), suggesting that nuclear extracts from patients with CLL with high lymphocyte counts bind to the caspase-3 promoter with a higher affinity.

In this study, we show that when present at high levels, p-STAT3 activates caspase-3 and induces apoptosis. For decades, the relatively slow accumulation of CLL cells in most patients was attributed to a prolonged lifespan of the leukemia cells (2–4, 28). However, evidence from recent years suggests that CLL cells proliferate and die, cell birth and death rates are similar, and the actual CLL cell count represents a net effect of clonal turnover (24).

Like in several cancers, in CLL, STAT3 is constitutively activated (8, 11, 17, 19, 29, 30). Activated STAT3 functions as an oncogene that induces proliferation and protects the neoplastic cells from apoptosis (10, 14, 31, 32). STAT3 activates the STAT3 gene, and as a result, CLL cells harbor high levels of STAT3 (10). As anticipated, STAT3 levels correlated with the patients’ lymphocyte count. However, in patients with high counts, STAT3 induced apoptosis rather than providing the cells with a survival advantage. This functional switch occurred only in cells with significantly high STAT3 levels. Likewise, overexpression of STAT3 induced apoptosis in MM-1 cells and in CLL cells from patients with relatively low lymphocyte counts.

In various neoplasms, STAT1 and STAT3 undergo similar posttranslation modifications (33). Consistent with our data, in CLL, both STAT1 and STAT3 are constitutively phosphorylated on serine 727 residues (10, 22) but not on tyrosine residues (34–36). However, whereas activated STAT3 under most conditions protects the cells from apoptosis (10, 12, 13, 22), activated STAT1 induces apoptosis (37). Remarkably, unlike STAT3, STAT1 or serine p-STAT1 levels did not correlated with CLL patients’ lymphocyte counts, suggesting that the proapoptotic effect of STAT3 is independent of STAT1.

Theoretically, at high levels, STAT3 might promote apoptosis either directly or indirectly by no longer activating antiapoptotic genes. In a previous study, we found that in acute myeloid leukemia, GM-CSF, known to induce myeloid cell proliferation, also upregulated and activated proapoptotic caspases by activating the JAK–STAT signaling pathway (38). The results of that study suggested that STAT activation induces apoptosis by directly activating the apoptotic cell death machinery. In the current study, we found that the transcript levels of the STAT3-regulated antiapoptotic genes Bcl-2 and MCL-1 were similar in CLL cells from patients with high and low lymphocyte counts, suggesting that these antiapoptotic genes are not downregulated in cells from patients with high lymphocyte counts. However, in cells from patients with high lymphocyte counts, caspase-3 levels were significantly high, suggesting that when present at high levels, STAT3 activates caspase-3.

In mammalian cells, STAT3 is ubiquitously expressed as a latent isoform and is activated upon phosphorylation on tyrosine 705 residues (39). In circulating CLL cells, STAT3 is constitutively phosphorylated on serine but not tyrosine residues (10, 22), and both serine– and tyrosine–p-STAT3 activate a similar repertoire of genes (12, 13, 19, 36). By using a luciferase assay of IL-6–stimulated MM-1 cells, in which IL-6 induces tyrosine–p-STAT3 (25), we found that p-STAT3 activates the caspase-3 promoter. These data were confirmed by ChIP and an EMSA with Abs that detect phosphorylated and unphosphorylated STAT3 isoforms. In CLL cells, ChIP confirmed that STAT3 binds with low affinity to the caspase-3 promoter, and an EMSA revealed that specific anti-serine p-STAT3 Abs significantly attenuated the binding of pSTAT3 to the caspase-3 promoter, suggesting that serine–p-STAT3 rather than its unphosphorylated form activates the caspase-3 promoter in CLL cells.

Caspase-3 levels increased only when STAT3 levels reached a threshold. High levels of STAT3 were required to induce this effect because the binding affinity of p-STAT3 to the caspase-3 gene promoter is relatively low. Hence, the functional switch of STAT3 depends on its DNA-binding affinity. However, DNA-binding affinity cannot be predicted by a consensus sequence per se. STAT3–DNA binding may vary depending on background DNA sequences in the vicinity to the GAS element, the chromatin status, and concomitant binding of STAT or DNA structures to other proteins, adaptor molecules, and other factors (40). For example, we identified several GAS-like elements in the caspase-3 promoter and used primers to detect them. However, one primer (primer 3), constructed to detect a STAT3-specific (TTCN3GAA) binding element, did not detect STAT3–caspase-3–DNA binding in MM-1 or CLL cells. Although analyzing STAT3 binding sites in the genome of different species, we did not detect conserved sequence blocks within the caspase-3 promoter. However, we identified several putative STAT3 binding sites in the caspase-3 promoter of all examined organisms.

STAT3 binds with high affinity to the antiapoptotic Bcl-2 and MCL-1 gene promoters and with low affinity to the caspase-3 gene promoter, suggesting that in CLL cells from patients with high lymphocyte counts, the activation of the caspase-3 gene overcomes in part the protective effect of antiapoptotic genes, and apoptotic cell death ensues.

Caspase-3 is an inactive zymogen that is usually activated by the intrinsic and the extrinsic apoptotic pathways (41). By activating the caspase-3 gene, STAT3 increases the level of caspase-3 and predisposes it to proteolytic cleavage by upstream caspases such as caspase-8 or caspase-9. In CLL cells, STAT3 levels correlated with the levels of cleaved but not uncleaved caspase-3 possibly because as its levels increased, caspase-3 underwent rapid cleavage. Unlike other caspases that possess autocatalytic activity, an intrinsic safety catch prevents the autocatalysis of caspase-3. Upon removal of the safety catch by acidification or other means, caspase-3 undergoes autocatalytic cleavage (42). Whether autocatalysis or upstream caspases activate caspase-3 in CLL cells remains to be determined.

Whereas in normal cells extracellular stimuli activate STAT3, in CLL, STAT3 is constitutively phosphorylated and activated. p-STAT3 forms dimers, shuttles to the nucleus using the karypherin nuclear transport system, binds to DNA, and, following nuclear phosphatase-induced dephosphorylation, STAT3 translocates to the cytosol using nuclear export chromosome region maintenance 1 (10). In the nucleus, p-STAT3 binds primarily to high-affinity DNA-binding sites, and when present at excess, p-STAT3 dimers bind to low-affinity DNA-binding sites. In the cytosol, a cytoplasmic serine kinase phosphorylates the USTAT3 on serine 727 residues (Z. Estrov, unpublished observations); in addition, USTAT3 binds to the NF-κB in competition with IκB (19). The levels of serine p-STAT3 detected via Western immunoblotting were similar in CLL cells of patients with high or low lymphocyte counts, most likely because this assay is not suitable to detect miniscule, albeit biologically significant, changes in p-STAT3 levels. Given that p-STAT3 activates the STAT3 gene and induces the production of STAT3 protein (10), cellular levels of total STAT3 protein represent a better parameter of the activity of STAT3. Therefore, it is very likely that in CLL cells from patients with high lymphocyte counts, both USTAT3 and p-STAT3 levels are high.

USTAT3 activates NF-κB (19), and, like the NFAT, NF-κB is constitutively activated in unstimulated CLL cells. Together, these transcription factors orchestrate a transcription program that further protects CLL cells from apoptosis (19, 43–45). Like STAT3, NF-κB levels correlate with disease burden (46), and although generally viewed as an antiapoptotic transcription factor, NF-κB also harbors proapoptotic properties (47).

Excessive activation of STAT3 induces apoptosis of CLL cells. Whether this versatile effect represents a physiological negative-feedback mechanism used in normal cells to counteract extreme proliferation or unwarranted prolonged cellular survival remains to be determined.

We thank Markeda Wade of the Department of Scientific Publications at The University of Texas MD Anderson Cancer Center for editing the manuscript.

The authors have no financial conflicts of interest.