Introduction

Chronic lymphocytic leukemia (CLL) is an age-associated B-cell malignancy characterized by the accumulation of hyperdiploid B-1 cells in the bone marrow, spleen and blood.1 It is the most common lymphoid malignancy in the western hemisphere. Since its first documented diagnosis more than 150 years ago, the etiology of CLL is largely unknown and it remains incurable with current therapy.2 FCR therapy—combination of Fludarabine, Cyclophosphamide and Rituximab—is the new gold standard for CLL therapy.3 Although the response rate to therapy is higher as compared with other cancers, almost all patients relapse because of persistence of minimal residual disease.4, 5 Hence, novel treatment strategies need to be developed. Modulating microRNA expression is one such promising but underexplored therapeutic area.6 We have previously shown that miR-15a/16-1 upregulation is one such promising therapeutic strategy.7

The 13q14 (region that encodes miR-15a/16-1 in humans) deletion is the most common chromosomal abnormality in CLL, occurring in 50–60% of patients.8 It is believed to encode critical tumor suppressor genes, as it is frequently deleted or silenced in various other malignancies like prostate cancer, mantle cell lymphoma and multiple myeloma.9, 10, 11 Detailed cytogenetic analysis has revealed the presence of a 130-kb minimal deleted region centromeric to the marker D13S272 that contains several candidate tumor suppressor genes like Dleu1, Dleu2, Dleu5 and Dleu7.12, 13 However, currently only Dleu2 (host gene of miR-15a/16-1) and Dleu7 have been demonstrated to have tumor-suppressive functions in CLL.14, 15

MicroRNAs are often located in intronic regions within host genes, which can be both coding and non-coding host genes.16 Mir-15a/16-1 is encoded within an intronic region of the non-coding Dleu2 gene in both human and mouse, and is transcribed off the Dleu2 promoter. A point mutation (in several CLL patients and NZB mice—de novo mouse model of CLL) and a point deletion (in NZB mice) in the 3′-flanking region of mir-16-1 was discovered and was associated with 50% reduction in the expression of mature miR-15a/16-1 in patients as well as in NZB mice and LNC cell line (NZB-derived mouse B-CLL line).17, 18, 19 Correcting the reduced miR-15a/16-1 level gives rise to growth inhibitory effect.20 To develop strategies to modulate miR-15a/16-1 levels, it is imperative to understand the molecular mechanisms that control its expression.

BSAP (B-cell-specific activator protein), encoded by the Pax5 gene, acts as a transcription factor and contains a DNA-binding domain, and recently it has been shown that BSAP negatively regulates Dleu2, the host gene of mir-15a/16-1 in mouse lymphoma cells.21 BSAP is expressed at the pro-B-cell stage and is maintained until the plasma cell stage is reached.22 BSAP can function either as an oncogene or a tumor suppressor depending on the cell type.23 BSAP can result in increased or decreased gene expression and this can be regulated by additional proteins that BSAP is capable of interacting with via its protein-binding domain. BSAP overexpression generally confers proliferative phenotype in lymphoid malignancies, especially B-ALL.24, 25 In light of this background, we explored the BSAP-Dleu2 regulation in mouse and human CLL cells. It has become increasingly clear that combination therapies are much more effective at fighting cancer (reviewed in Humphrey et al.26). Hence, we also report herein the combined effect of BSAP knockdown and histone deacetylase (HDAC) inhibition (HDAC activity is increased in CLL) on miR-15a/16-1 levels and malignant cell death.

Results

BSAP levels are increased and inversely correlate with miR-15a/16 levels in B-1 malignant cells from CLL patient PBMCs

PBMCs from untreated CLL or age-matched normal controls were stained for surface expression of CD19 and CD5 and intracellular levels of BSAP. Cells were gated on CD19+ (B-cell gate) and the mean fluorescence intensity (MFI) of BSAP determined (Figure 1a). The BSAP levels in a non-B-cell source, patient T cells (CD3+, CD19−), is shown for comparison. The CLL B cells demonstrated increased expression of BSAP when compared with non-CLL sources. However, all sources of B cells had increased expression of BSAP relative to their T-cell population. In addition, the CLL B cells were sorted into two different B-1 populations, CD19+CD5+ BSAPhi and CD19+CD5+BSAPlo (Figure 1b). RNA was obtained from the sorted populations and analyzed by polymerase chain reaction (PCR) for the expression of miR-15a. The B-1 cells with high BSAP had reduced levels of miR-15a relative to the expression in B-1 cells with low expression of BSAP (Figure 1b). Similarly, as miR-15a/16-1 targets Bcl-2, the expression of Bcl-2 was low in the BSAP low-expressing CLL cells, suggesting that the resultant high levels of miR-15a/16-1 permit these cells to readily undergo apoptosis. Indeed, the BSAP low cells are the minor population of B-1 cells in the CLL patient. Hence, we hypothesized that by knocking down BSAP via small interfering (si)BSAP, miR15a/16-1 levels will be increased and their target Bcl-2 decreased, which would lead to the induction of apoptosis.

Figure 1
figure 1

BSAP levels are elevated in CLL B cells and inversely correlated with miR-15a levels. (a) Representative histogram overlay of intracellular BSAP levels in PBMCs from normal controls (n=5) or CLL (n=11). Control CD19+ (B cells) (in blue) are decreased relative to CLL B cells (in red). For comparison, CLL patient T cells (CD3+, dashed histogram) and the isotype (gray shaded histogram) are shown. The average delta MFI±s.d. (change in MFI relative to isotype) is shown in the graph to the right. (b) CLL B cells were gated on CD19+ and sorted into CD5+BSAPhi and CD5+ BSAPlow populations. The mean expression levels of miR-15a and Bcl2 in the sorted populations, measured using 100 cell PCR, are shown in the graph. *P<0.05, n=11, two-tailed paired Student’s t-test.

Full size image

Negative regulation of miR-15a/16-1 by BSAP can be exploited for CLL therapy

CLL is a very heterogeneous disease, but downregulation of miR-15a/16-1 remains the single most commonly occurring pathological event in >70% of CLL patients.27 Hence, we hypothesized that by attenuating the BSAP-mediated negative regulation of the promoter of the host gene for miR-15a/16-1, we can increase miR-15a/16-1 levels in ex vivo-treated patient PBMCs. Total PBMCs isolated from patient blood were used for the BSAP knockdown studies instead of purified B-1 cells, as >90% of the PBMCs were malignant CD19+CD5+ B-1 cells (Figure 2a). BSAP expression was knocked down by about 25% using siRNA-BSAP as compared with the Neg. CTRL siRNA (Figure 2d). The reduced expression of BSAP translated into a significant increase in the expression of miR-15a/16-1 (Figure 2b). Different patients exhibited different kinetics of BSAP knockdown and hence the peak reduction in BSAP is shown. BSAP positively regulates CD19 expression and post siRNA-BSAP treatment, CD19 expression was reduced, further validating the success of the BSAP knockdown (Figure 2e). As the CLL patient PBMCs are not actively cycling, increase in the percentage of apoptotic cells was used as a readout for biologically significant increase in miR-15a/16-1 expression, instead of cell cycle analysis. The percentage of Annexin V+ cells (apoptotic cells) was significantly higher in the siRNA-BSAP-treated group as compared with the Neg. CTRL-treated group (Figure 1f), indicating that reduction in BSAP induced by siRNA-BSAP increased both the level of miR-15a/16-1 and the amount of apoptosis.

Figure 2
figure 2

BSAP knockdown in ex vivo CLL patient PBMCs increases miR-15a/16-1 and apoptosis. (a) Representative histogram overlay of BSAP expression in CD19+CD5+ cells from PBMCs of CLL patients. Neg.CTRL- (gray filled) and siRNA BSAP- (black open) treated cells 24 h post-transfection. (b) Mean peak relative quantitation (RQ) PCR±s.d. of miR-15a and Dleu2 post siRNA BSAP or negative control treatment. (c) Regression analysis of levels of Dleu2 mRNA levels versus mR15a. diamond=neg siRNA, circle+siBSAP, cross=sorted CD5+ B cells (d) Average reduction in BSAP expression following treatment with Neg.CTRL or siRNA BSAP transfection. %Reduction in BSAP=(MFI treated−MFI mock) × 100/MFI mock. BSAP, 24 h post-transfection. (e) Mean delta CD19 MFI as compared with mock-treated cells in Neg.CTRL (gray) and siRNA BSAP (black). (f) Mean % change in apoptotic cells in Neg.CTRL (gray) and siRNA BSAP (black) % change=(%AnnexinV+ treated−%AnnexinV+ mock). *P<0.05, n=7.

Full size image

BSAP regulates miR-15a/16-1 expression at the level of host gene transcription

Many microRNAs are encoded within the introns of other bigger coding or non-coding genes, with or without their own promoters. In the latter case, the expression of the encoded miR would depend on the promoter status of the host gene. To confirm that BSAP regulates Dleu2 transcription (host gene for mir-15a/16-1), the Dleu2 mRNA level was measured in the CLL patient PBMCs following siRNA-BSAP. A significant increase was observed in the expression of Dleu2 following BSAP knockdown (Figure 2b). The increase in Dleu2 was found to be positively correlated to the level of miR-15a/16-1 (Figure 2c). Hence, we conclude that BSAP inhibits mir-15a/16-1 expression by inhibiting the transcription of its host gene, Dleu2. This was further verified by fluorescence in situ hybridization (FISH) analysis at a single-cell level to quantitate simultaneously the amount of Pax5 mRNA (which encodes the BSAP protein) and the amount of Dleu2 mRNA. In non-CLL B cells, the amount of Pax5 (BSAP) was decreased relative to the CLL B cells (Figure 3). In addition, the amount of Pax5 expression was inversely related to the amount of Dleu2 on a single-cell level (Figure 3), with CLL B cells having low levels of miR-15a/16 expression and high levels of BSAP and non-CLL B cells expressing lower levels of BSAP but higher levels of miR15a.

Figure 3
figure 3

In situ RNA FISH analysis of Dleu2 and Pax5 mRNA levels. Cells were grown on poly L-lysine-coated coverslips and hybridized with single-molecule FISH probes. Dleu2 probes were labeled with TMR and Pax5 probes were labeled with Alexa 594. The z-stacks obtained from each fluorescence channel were merged and coded to represent Pax5 (in red) and Dleu2 (in green). Each spot represents single mRNA molecule.65 The right panel shows the merger of both channels and the chart represents the mean values from multiple analysis±s.e.m.

Full size image

BSAP interacts with Dleu2 promoter

It is currently unknown whether BSAP regulates Dleu2 promoter directly or indirectly. Promoter analysis using CONSITE failed to show any canonical BSAP binding sites in the human Dleu2 promoter. However, we cannot rule out the possibility that BSAP might be present in a complex with some of its binding partners at the Dleu2 promoter. Binding sites for some of the proteins that BSAP has been known to interact with were found in the human Dleu2 promoter (c-Myb, AML1 and E2F). We performed a chromatin immunoprecipitation (ChIP) assay on Daudi cells to assess BSAP-Dleu2 promoter interactions. Daudi is a B lineage cell line (Burkitt’s lymphoma) and has BSAP expression.28 Similar to the positive control of CD19 promoter, the Dleu2 promoter was enriched in the BSAP pulldown. No enrichment of the negative control Kras promoter was observed (Figures 4a and b). The ChIP assay was performed using four CLL PBMC samples in four independent pulldown experiments (Figure 4c). The Dleu2 promoter region was also enriched in the CLL samples and was similar to a known promoter, which encodes a BSAP binding site, the CD19 promoter. In contrast, the Kras promoter, which does not encode a BSAP binding site, was not pulled down in CLL samples (Figure 4c). These data suggest that a BSAP binding site is present at the Dleu2 promoter in human B cells. However, a canonical BSAP binding site was found at +676 position in the mouse Dleu2 gene. A ChIP assay was performed on murine B cells, the LNC cells, to assess BSAP-Dleu2 promoter interactions. As we knew the position of the putative BSAP binding site in mouse Dleu2, we were able to design a ChIP assay having an internal control in addition to the glyceraldehyde 3-phosphate dehydrogenase control as shown in Figure 4d. Indeed, the Dleu2 promoter fragments were enriched in the BSAP pulldown and the amount of enrichment was inversely proportional to the distance from the BSAP binding site (Figure 4e). The upstream sequences (U) were not as enriched by the BSAP pulldown as were the downstream sequences (D), which contained the BSAP binding site. The presence of the Dleu2 U region in the BSAP pulldown may be because of DNA fragments, which contain both the D and U region. The human data indicating BSAP interaction with the Dleu2 promoter were validated in the mouse system, which has a canonical BSAP binding site in its promoter.

Figure 4
figure 4

Dleu2 promoter occupancy by BSAP in human and mouse. (a) Representative data of promoter occupancy in human B-cell line (Daudi) following immunoprecipitation with IgG (gray) or BSAP (black) antibody. (b) Delta % Input=(% Input BSAP IP−% Input IgG) indicates promoter enrichment over pull down with nonspecific IgG antibody in Daudi cells. (c) Cumulative data of promoter enrichment ChIP experiments in CLL PBMCs, n=4. (d) Promoter analysis in mouse, schematic of the Dleu2 promoter in mouse, the transcription start site (Dleu2 TSS), the flanking upstream (U) and downstream (D) regions are indicated and the total region is approximately 1 kb. The BSAP binding site is shown as a star. Chart enrichment of CD19, glyceraldehyde 3-phosphate dehydrogenase, Dleu2 D and Dleu2 U fragments in LNC, the NZB mouse CLL cell line (mean values of three independent experiments). *P<0.05, Student’s t-test. Error bars indicate±s.e.m.

Full size image

BSAP knockdown leads to cell cycle arrest and apoptosis in NZB

We and others have previously shown that miR-15a/16-1 expression is reduced by almost 50% in CLL (in both patients and NZB mice).20 Using microarray-based expression analysis, negative correlation was shown between BSAP and Dleu2 (host gene of miR-15a/16-1) levels in Myc5 cells.21 Myc5 is a B-lymphoma cell line derived by transducing p53-null bone marrow cells with c-Myc retrovirus. These cells can oscillate between B-cell and macrophage lineages based on the culture conditions and BSAP levels in vitro.29 On the contrary, LNC is a natural malignant B-1 cell line derived from the spontaneously occurring murine model of human CLL and has constitutively very high levels of BSAP expression and low levels of miR-15a/16-1 expression (Supplementary Figure S1). In light of the above disparities in the two systems, we wanted to confirm whether BSAP could negatively regulate miR-15a/16-1 levels in our system.

The NZB-derived CLL cell line LNC was transfected with either 3 μg non-targeting negative control siRNA (Neg.CTRL) or with 3 μg siRNA BSAP (siRNA-BSAP), using Amaxa Nucleofection (Lonza, Walkersville, MD, USA), to transiently knockdown BSAP. The expression of BSAP was reduced as early as 12 h post-transfection and the suppression began to diminish after 48 h. Flow cytometric analysis via histogram overlays of BSAP levels on a per cell basis showed a decrease in BSAP in siRNA-treated cells compared with Neg.CTRL-treated cells as early as 12 h post-transfection (Figure 5a). The average BSAP MFI at different time-points post-transfection is shown in Figure 5b and was found to be lowest at 24 h post siRNA-BSAP treatment when compared with the Neg.CTRL at that time-point (P<0.05). BSAP protein was reduced by 53% at 24 h as compared with the Neg.CTRL. The miR-15a levels were measured in cells with successful knockdown of BSAP to determine the potential causal relationship between BSAP and miR-15a/16-1 expression in our system. The miR-15a level was measured using TaqMan MicroRNA Assays (Applied Biosystems, Carlsbad, CA, USA) and an approximately twofold increase in miR-15a levels was observed in the siRNA-BSAP-treated cells at 24 h (P<0.05) (Figure 5c). Interestingly, the increase in miR-15a expression corresponded with the peak reduction in BSAP, suggesting a strong influence of BSAP on the expression of the microRNA. LNC cells mimic aggressive CLL and are very rapidly dividing. We and others have shown that an increase in miR-15a/16-1 leads to cell cycle arrest and reduced proliferation.30, 31, 32 Initially, we have shown that reducing BSAP levels results in an increase in miR-15a/16-1 expression. To test the biological significance of this, the increased miR-15a/16-1 should result in decreased downstream targets like cyclin D, leading to cell cycle arrest. As compared with the Neg.CTRL-treated cells, which have baseline level of BSAP and miR-15a/16-1, the siRNA-BSAP-treated cells exhibited a significant increase in the percentage of cells in the G1 phase and a reduction in the percentage of cells in the S phase (Figures 5d and e, P<0.05). Moreover, we also observed a decrease in the expression of cyclin D1 in the siRNA-BSAP-treated cells as compared with the Neg.CTRL-treated cells (Figure 5f), indicating that the cell cycle arrest is a direct consequence of increased miR-15a/16-1, as it has been shown to be an important regulator of cyclin D1.

Figure 5
figure 5

BSAP knockdown leads to increased miR-15a/16-1 and cell cycle arrest in vitro: 3 μg siRNA BSAP or non-targeting siRNA (Neg.CTRL) was transiently transfected into the murine NZB CLL cell line LNC and analyzed at the indicated time-points. (a) Representative histogram overlay of cells transfected with either siRNA BSAP (black dotted line) or with Neg.CTRL (gray filled) and stained with BSAP-PE antibody. (b) The BSAP MFI±s.e.m. at indicated time-points post-transfection with siRNA BSAP (black bar) or with Neg.CTRL (gray bar), n=3, *P<0.05. (c) Average miR-15a levels 24 h post-transfection, n=6, *P<0.05. (d) Percentage of cells in the the G1 phase and the S phase of the cell cycle at 36 h post-transfection, n=7, *P<0.01. (e) Percent change in cells in the G1 phase and the S phase at 36 h post-transfection. Percent change=((%G1 or %S in siRNA BSAP group−%G1 or %S in Neg.CTRL group)/(%G1 or %S in Neg.CTRL group)) × 100, n=7, *P<0.01, large effect size. (f) Decrease in the cyclin D1 MFI in the siRNA BSAP group in relation to the Neg.CTRL group at 24 h post-transfection, n=3. Statistics used: Two-tailed paired Student’s t-test.

Full size image

Combination treatment with HDAC inhibitor

Next, we wanted to assess whether the antiproliferative effect of BSAP knockdown can be enhanced by using an HDAC inhibitor. LNC cells were treated with siRNA-BSAP and the HDAC inhibitor suberoylanilide hydroxamic acid (SAHA) (clinically known as vorinostat). Dose–response curves were calculated for SAHA (data not shown) and a suboptimal dose of 0.8 μM was selected for the combination treatment. An additive effect was observed on the increase in the %G1 (Figure 6a) and decrease in %S (Figure 6b) as compared with the untreated cells. However, a synergistic effect was observed in the percentage of cells undergoing apoptosis when treated with the siRNA BSAP and SAHA together (Figure 6c). In addition, we observed a significant increase in miR-15a/16-1 expression following treatment (Figure 6d). To confirm that the observed cell cycle and apoptosis effects are mediated by increased miR-15a/16-1, LNC cells were treated with SAHA+siRNA BSAP+antagomiR to miR-15a and miR-16-1. The antagomiR lead to a stable reduction in miR-15a/16-1 up to 48 h post-transfection (Figure 6e). This gave rise to a partial rescue from increased malignant cell death (Figure 6f). In conclusion, these findings indicate that the combination of HDAC inhibition and BSAP knockdown leads to a significantly higher increase in malignant cell death in part by increasing the level of miR-15a/16-1.

Figure 6
figure 6

HDAC inhibitor enhances the apoptotic effect of BSAP knockdown: LNC cells were treated with 0.8 μM SAHA alone or in combination with 2.5 μM siRNA-BSAP. The delta values were calculated in relation to untreated cells. (a) Relative quantitation (RQ) miR-15a 36 h post treatment with siRNA BSAP, SAHA and SAHA+siRNA BSAP (SA+siRNA). (b) Percentage of AnnexinV+ cells after indicated treatments. *P<0.05, combination versus single-agent treatment, horizontal line (−) indicates predicted value if effect of siRNA BSAP and SAHA were additive, #P<0.05, observed increase in apoptosis significantly different from the predicted additive value for the combination treatment. (c) Change in percentage of cells in the G1 phase after indicated treatments. (d) Change in percentage of cells in the S phase after indicated treatments. (e) Bar graph showing average % Annexin V+ cells in SAHA+siRNA BSAP (SA+siRNA) and SAHA+siRNA BSAP+antgomiR to miR-15a/16-1 (SA+siRNA+antagomiR) groups at 24 h. (f) Line graph showing the level of miR-15a in SA+siRNA (solid line) and SA+siRNA+antagomiR (dashed line) at 24, 36 and 48 h post treatment. *Significant difference in the miR-15a levels between the two groups at that time-point, paired Student’s t-test, n>3, error bars indicate±s.e.m.

Full size image

Discussion

In this study, we found that in CLL patients, B cells constitutively express higher levels of BSAP than do control sources of B cells. However, within a CLL malignant clone, a minority of the B-1 cells had relatively low levels of BSAP. Analysis of the two types of B-1 cells demonstrated that increased expression of the microRNA miR-15a/16-1 was found in the patient B-1 subpopulation with the higher expression of BSAP (minor population). To further dissect the role of BSAP in the regulation of the levels of miR-15a/16-1, addition of siRNA pools with antisense regions to the Pax5 gene (which encodes BSAP) resulted in an increase in miR-15a/16-1 levels. Taken together, these studies indicate that BSAP is a negative regulator of miR-15a/16-1. The BSAP-mediated regulation of miR-15a/16-1 levels is occurring at the level of the host gene transcription as seen by the significant positive correlation between Dleu2 and miR-15a/16-1 levels that is dependent on the BSAP levels and independent of downstream mutations in that loci. This was also observed at the level of single-cell RNA FISH analysis. We have also shown that BSAP interacts with the Dleu2 promoter (direct interaction in mice and indirect interaction in humans). This finding is very interesting given that the human Dleu2 promoter lacks a canonical BSAP binding site. We speculate that BSAP interacts with another protein that binds to the Dleu2 promoter directly and this complex is responsible for the repression of the Dleu2 promoter. This report concentrated on Dleu2 promoter repression and other studies have also indicated that the Dleu2 promoter regulates miR15a/16 levels.33 However, a potential promoter region located near the mir15a/16-1 loci within Dleu2, which has myb binding sites, has been reported and myb interacts with BSAP.34 Further studies need to be performed to identify binding partners of BSAP and additional promoter regulation of miR15a/16.

Earlier studies focused mainly on the role of BSAP in the B-cell lineage development program.35 This study highlights the role of BSAP in cancer cell survival via its regulation of an important pro-apoptotic microRNA cluster, mir-15a/16-1. BSAP appears to act as an oncogene in CLL and has an important role in downregulating mir-15a/16-1, a known tumor suppressor in CLL. Other groups have also demonstrated the role of BSAP in regulating apoptosis.36 Previously, we demonstrated that inhibition of BSAP leads to decreased proliferation and apoptosis in malignant B-1 cells from the NZB murine model of CLL.37, 38, 39 In this study, we have shown that knocking down BSAP leads to malignant cell death not only in our murine CLL cell line but also in ex vivo patient PBMCs via upregulation of miR15a/16 levels. We propose that targeting BSAP using siRNA or small-molecule inhibitors could serve as a novel therapeutic strategy for CLL. The role of BSAP in tumorigenesis is highly cell type specific, in that it can act both as an oncogene (non-Hodgkin’s lymphoma, neuroblastoma) or as tumor suppressor (multiple myeloma).40, 41, 42, 43 Targeting BSAP has already been shown to have therapeutic effects in other types of cancers like small-cell lung cancer.44 BSAP has been shown to be a positive regulator of CD19 and indirectly to control c-myc levels via regulation of CD19.45 In these studies, BSAP binding to the promoter region of CD19 in the ChIP analysis was verified and the additional binding to the promoter of Dleu2 (and miR15a/16-1) determined. The elevated levels of BSAP found in CLL malignant B-cell clones may have multiple effects in addition to negative regulation of miR15a/16 via repression of Dleu2 including elevation of c-myc via upregulation of CD19.

Because the bicistronic miR-15a/16-1 is a tumor suppressor microRNA,46 a wide variety of malignancies involve the repression of this microRNA cluster. There are different mechanisms by which this repression occurs. Several other studies have also found that the Dleu2 host gene promoter is repressed. Recently, c-Myc has been shown to repress the Dleu2 promoter via recruitment of HDAC3.47 HDAC activity is increased in a variety of cancers including CLL.48, 49, 50 HDACs have emerged as attractive targets in the treatment of both solid and hematological malignancies.51, 52, 53 Several groups have reported the efficacy of HDAC inhibitors in inducing CLL cell death in vitro and in clinical trials.54, 55 In addition to inducing direct cell death, HDAC inhibition also increases the immunogenicity of CLL cells, thereby facilitating antitumor immune response.56, 57 HDAC has also been shown to repress the Dleu2 promoter in CLL malignant B-1.58 Our results support these previous findings. We have shown that similar to BSAP-mediated Dleu2 repression, HDAC-mediated repression is also reversible and it opens up avenues for therapeutic intervention. Based on the data presented in this paper, we propose that BSAP and HDACs co-operate to bring about repression of the Dleu2 promoter in CLL. This study is critical because for the first time it simultaneously addresses the interaction between three important players (13q14 locus, BSAP and HDAC) involved in CLL pathogenesis. Out of the many genetic alterations found in CLL, the frequency of only 13q14 deletions is significantly elevated in MBL (monoclonal B-cell lymphocytosis), the precursor stage of CLL.59 In addition, epigenetic silencing of 13q14 has been found in CLL.60 This finding indicates that 13q14 silencing via its deletion or epigenetic modification is one of the first events leading to CD19+CD5+ B-cell expansion and understanding the regulation of this locus will be instrumental in delineating CLL etiology. A recent paper by Laurie et al.61 that analyzed clonal mosaicism in 50 000 subjects further underscores the importance of the 13q14 region in CLL.61 In this study, the authors discovered that 13q deletions were over-represented in normal subjects with age and its presence increased the chances of a future CLL diagnosis.

In summary, we have shown that in CLL, promoter regulation by a transcriptional repression via BSAP and HDAC co-coordinately repress Dleu2, the host gene of miR-15a/16-1, and targeting this loop leads to malignant cell death via increase in mature miR-15a/16-1 expression (Supplementary Figure S2).

Materials and methods

Patient samples and cell lines

A measure of 4–5 ml of blood was collected in ethylenediaminetetraacetic acid (EDTA) tubes from untreated CLL patients after obtaining informed consent in accordance with UMDNJ (now Rutgers) and East Orange Veterans Administration New Jersey Health Care System human subjects Institutional Review Board. PBMCs were isolated from patient blood using Ficoll-Hypaque solution according to the manufacturer’s instructions (Stem Cell Technologies Inc., Vancouver, BC, Canada). Human Burkitt’s lymphoma cell line Daudi (ATCC No. CCL-213, Manassas, VA, USA) was used. In addition, the NZB- (murine model of human CLL) derived malignant B-1 cell line LNC was also used.62 Both cell lines were maintained in RPMI 1640 supplemented with 10% fetal bovine serum, 1% sodium pyruvate, 1% penicillin–streptomycin at 37 °C and 5% CO2.

siRNA nucleofection

A total of 2.5 × 106 PBMCs were nucleofected with 3 μg of ON-TARGETplus SMARTpool—siRNA targeting the Pax5 gene that encodes the BSAP protein (hereafter referred to as siRNA-BSAP) or ON-TARGETplus Non-targeting Pool siRNA (Neg.CTRL) as a control (Dharmacon, Lafayette, CO, USA) or nothing (mock) using Human B Cell Nucleofector Kit program with the AMAXA Instrument (Lonza, Switzerland). Similarly, 2.5 × 106 LNC cells were nucleofected with 3 μg of ON-TARGETplus SMARTpool—Mouse siRNA BSAP or ON-TARGETplus Non-targeting Pool siRNA as a control (Dharmacon) or nothing (mock) using Cell Line Nucleofector Kit T, program G-016 (AMAXA Instrument). % Reduction in BSAP protein was calculated as ((siRNA BSAP MFI−Neg.CTRL MFI)/Neg.CTRL MFI) × 100. The cells were harvested at different time points (24, 36, 48, 72 and 96 h) for further analysis. Wherever indicated, 1 μM each of antagomiRs to miR-15a and miR-16-1 (Dharmacon) were also nucleofected.

Surface and intracellular flow cytometry

Approximately 0.5 × 106 cells were stained with indicated surface antibodies for 25 min at 4 °C. Cells were then fixed and permeablized using BD Cytofix/Cytoperm kit (BD Biosciences, Franklin Lakes, NJ, USA) according to the manufacturer’s instructions and stained with the indicated intracellular antibodies for 30 min. Antibodies for surface markers were anti-human CD19-FITC (BD Biosciences), anti-human CD5-PE.Cy7 (BD Biosciences), and for intracellular markers were anti-human/mouse Pax5-PE (eBioscience, San Diego, CA, USA) and anti-mouse cyclin D1-AF647 (Santa Cruz Biotechnology Inc., Santa Cruz, CA, USA). The stained cells were acquired on BD LSR II (BD Biosciences) and analyzed using the FlowJo Software (TreeStar Inc., Ashland, OR, USA).

Sorting of B-1 cells and 100 cell PCR

CLL patient PBMCs were stained with CD19-FITC, CD5-PE.Cy7 and BSAP-PE antibodies. The CD19+ population in the lymphoid gate was further sorted into CD5+BSAPhi (B-1 BSAP hi) and CD5+BSAPlow (B-1 BSAP low) using the BD FACS Aria II cell sorter. Owing to the low number of cells obtained after sorting, we performed 100 cell PCR to measure the miR-15a/16-1 and Bcl-2 levels in the sorted subpopulations. Briefly, for the microRNA measurement, 100 cells in 4.84 μl 1 × phosphate-buffered saline were heat disrupted at 95 °C for 10 min to release RNA and immediately kept on ice. This extract was then used to prepare cDNA using the TaqMan microRNA RT Kit (Applied Biosystems), followed by real-time PCR. Bcl-2 levels were measured using the Power SYBR Green Cells-to-Ct Kit (Ambion Inc., Austin, TX, USA).

RNA isolation and quantitation

Total RNA was extracted using Trizol reagent (Invitrogen, Carlsbad, CA, USA) according to the manufacturer’s instructions. MicroRNA-specific cDNA was prepared using the TaqMan MicroRNA Reverse Transcription Kit (Applied Biosystems) according to the manufacturer’s instructions. The following pre-made TaqMan Assays (Applied Biosystems) were used for real-time quantitation—mmu-miR-15a (Assay ID 000389) and U6 (Assay ID 001973). Dleu2 transcripts were quantified from random hexamer-primed cDNA using TaqMan hsa-Dleu2 Assay (Applied Biosystems) (Assay ID Hs00863925 m1). MicroRNA and Dleu2 levels were normalized to U6 (mouse and human) and 18s rRNA (human), respectively.

Cell cycle and apoptosis assay

For cell cycle analysis, 0.5 × 106 were washed with 1 × phosphate-buffered saline and resuspended in 300 μl hypotonic propidium iodide solution, acquired on a BD Calibur IV cytometer (BD Biosciences, Franklin Lakes, NJ, USA) and analyzed using the ModFit LT Software (Verity Software House, Topsham, ME, USA). For apoptosis quantitation, cells were stained with Annexin V-PE (BD Bioscience) according to the manufacturer’s protocol, acquired on BD Calibur IV.

Chromatin immunoprecipitation

In total, 5 × 106 cells were crosslinked with 0.37% formaldehyde and sonicated to obtain 1-kb or 400-bp fragments using the S220 sonicator (Covaris Inc., Woburn, MA, USA). The ChIP-ready chromatin was pulled down with either goat anti-human Pax5 (Santa Cruz Biotechnology) or normal goat immunoglobulin G (IgG) (Santa Cruz Biotechnology). EpiTect One Day ChIP Kit (Qiagen) was used. The pulled down DNA was then used for real-time SYBR Green PCR of Dleu2 promoter using primers described in Bougel et al.63 CD19 promoter was used as a positive control and Kras promoter was used as the negative control. Similar protocol was followed for the mouse cell line. Dleu2 was primed at two separate sites (with and without the BSAP binding site as an internal control). CD19 was used as a positive control and glyceraldehyde 3-phosphate dehydrogenase was used as a negative control. Refer Supplementary materials (Supplementary Table S1) for mouse primer sequences.

HDAC inhibitor treatment

SAHA (clinically known as vorinostat57, 64) (Selleckchem, Houston, TX, USA) was dissolved in 100% ethanol to obtain a 10 mM stock. The final working concentration used was 0.8 μM.

RNA FISH

Dleu2 and Pax5 RNAs were imaged using single-molecule FISH probes as described previously.65 Briefly, a set of 35 probes was designed to hybridize to each target RNA and was synthesized with a 3′ amino modification from Biosearch Technologies (Novato, CA, USA). The individual probes for a given target were pooled in equimolar amounts and then coupled with succinimidyl ester of either TMR (for Dleu2) or Alexa 594 (for Pax5). The coupled fraction was purified using high-performance liquid chromatography and the concentration was determined using nanodrop. The coverslips were washed with 1 × phosphate-buffered saline, fixed in 4% formaldehyde, permeabilized with 70% ethanol and hybridized with the Dleu2 TMR and Pax5 Alexa 594 probes. Hybridization was carried out overnight at 37 °C. The coverslips were washed (with 10% formamide in 2 × SSC) to remove unbound probes and imaged using Zeiss wide field fluorescence microscope (Carl Zeiss, Thornwood, NY, USA). For each image, z-stacks were obtained and merged to get the final image. The image acquisition was carried out by Openlab software (Perkin-Elmer, Waltham, MA, USA) and numbers of mRNAs were counted using custom written alogrithms in MATLAB (MathWorks, Natick, MA, USA).66

Statistics

Data were analyzed with a paired Student’s t-test, unless otherwise specified. P<0.05 was considered significant.