An Orally Available Small Molecule BCL6 Inhibitor Effectively Suppresses Diffuse Large B Cell Lymphoma Cells Growth in Vitro and in Vivo

Background: The transcription factor B cell lymphoma 6 (BCL6) is an oncogenic driver of diffuse large B cell lymphoma (DLBCL) and mediates lymphomagenesis through transcriptional repression of its target genes by recruiting corepressors to its N-terminal broad-complex/tramtrack/bric-a-brac (BTB) domain. Blocking the protein-protein interactions of BCL6 and its corepressors has been proposed as an effective approach for the treatment of DLBCL. However, BCL6 inhibitors with excellent drug-like properties are rare. Hence, the development of BCL6 inhibitors is worth pursuing. Methods: We screened our internal chemical library by luciferase reporter assay and Homogenous Time Resolved Fluorescence (HTRF) assay and a small molecule compound named WK500B was identied. The binding anity between WK500B and BCL6 was evaluated by surface plasmon resonance (SPR) assay and the binding mode of WK500B and BCL6 was predicted by molecular docking. The function evaluation and anti-cancer activity of WK500B in vitro and in vivo was detected by immunouorescence assay, Real-Time Quantitative PCR, cell proliferation assay, cell cycle assay, cell apoptosis assay, enzyme-linked immunosorbent assay (ELISA), germinal centre (GC) formation mouse model and mouse xenograft model. Results: WK500B engaged BCL6 inside cells, blocked BCL6 repression complexes, reactivated BCL6 target genes, killed DLBCL cells and caused apoptosis as well as cell cycle arrest. In animal models, WK500B inhibited germinal centre formation and DLBCL tumor growth without toxic and side effects. Moreover, WK500B showed favourable pharmacokinetics and presented superior druggability compared to other BCL6 inhibitors. Conclusions: WK500B druggability


Background
Diffuse large B cell lymphoma (DLBCL) which arises from the germinal centre (GC) and exhibits highly aggressive and rapid progression is the most common form of non-Hodgkin lymphoma (NHL) 1,2,3 .
Approximately 60%-70% of DLBCL patients can be cured by using a rst-line therapy regimen that consists of the combination of a CD20 monoclonal antibody with cyclophosphamide, hydroxyrubicin, oncovin and prednisone (R-CHOP) 4 . Despite major advances in the treatment of DLBCL, up to 50% of the patients will relapse or become refractory to this treatment and die of their disease 4,5 . The development of chimeric antigen receptor (CAR)-T cell therapy may be a boon for relapsed/refractory patients, however, not all patients respond to CAR-T therapy, and only approximately 50% of those progress within 12 months 6 . Novel therapies that improve the outcome for DLBCL remain imperative.
The BTB/POZ transcriptional repressor BCL6, is a master regulator essential for the GC reaction, which is necessary for humoral immunity and has emerged as a critical therapeutic target for DLBCL 7,8,9 . BCL6 contains six C 2 H 2 zinc nger domain at its C-terminus, which interacts with DNA sequences 10 . The Nterminal BTB-POZ domain mediates dimerization and recruits the corepressors silencing mediator of retinoid and thyroid receptor (SMRT), nuclear receptor corepressor (N-CoR) and BCL6 corepressor (B-CoR) to a speci c groove motif 11,12,13 , which is required for the repression of target genes involved in DNA damage responses and proliferation checkpoints 14,15,16,17 . The middle portion, RD2, interacts with the corepressor MTA3 and represses genes involved in plasma cell differentiation 18 . Dysregulation of BCL6 expression associated with promoter translocations or point mutations has been shown to play a central role in DLBCLs 7,10,19,20 . Constitutive expression of BCL6 in mice increases the risk of the development of DLBCL, similar to that observed in human disease 21,22 . Delivery of shRNA to DLBCL cells, which results in the loss of BCL6 biological activity, kills DLBCL cells, indicating that BCL6 is an attractive target for DLBCL treatment 17 . A peptidomimetic BCL6 inhibitor RI-BPI inhibits the recruitment of BCL6 BTB corepressors and causes de-repression BCL6 target genes by binding to the BCL6 BTB domain 23 . More importantly, it can kill DLBCL cell lines and DLBCL patient cells 23 , demonstrating that inhibition of the BCL6 BTB -corepressor interaction and reactivation of BCL6 target genes are a useful therapeutic strategy for DLBCL treatment.
At present, several BCL6 small molecule inhibitors have been reported, which demonstrates the feasibility of developing BCL6 inhibitors that directly bind to the BCL6 BTB domain 24,25,26,27,28,29,30 . However, no such inhibitor has been enrolled in a clinical-trial. For example, the compounds 79 − 6 and FX1 directly bind to BCL6 BTB , reactivate BCL6 target genes and inhibit the proliferation of DLBCL cells in vitro and in vivo 24,25 . However, the weak activities of 79 − 6 and FX1 and the fact that they are based on the skeleton of rhodamine may be detrimental and potential limitations for further drug development 31,32 . Other reported small molecule inhibitors can bind to BCL6 BTB and disrupt the interaction between BCL6 and corepressors in a micromolar range or even a nanomolar range in vitro, but their effects on the biological functions of BCL6 in vivo have not been further studied, which may be due to their poor pharmacokinetic/pharmacodynamic properties or other reasons 26,27,28,29,30 . Therefore, there is still a long way to go in terms of screening novel BCL6 inhibitors with high druggability for DLBCL therapy.
We showed that WK500B, a novel synthetic small molecule compound, directly bound to BCL6 BTB , signi cantly inhibited the BCL6 BTB /SMRT interaction, reactivated BCL6 target genes in a concentrationdependent manner, killed DLBCL cell lines and caused cell cycle arrest and apoptosis. More importantly, WK500B showed favourable pharmacokinetics, resulting in the disruption of the germinal centre formation and the a nity maturation of immunoglobulins in mice at a low dose, and it strongly suppressed the growth of DLBCL in vivo. Furthermore, no cases of toxic effects were noted. Taken together with the fact that WK500B is more potent both in vitro and in vivo, is orally available and shows superior druggability compared with other reported BCL6 inhibitors, it seems that WK500B could be a promising therapeutic drug candidate for DLBCL.

Luciferase reporter assay
The reporter construct (GAL4) 5 -TK-LUC and the GAL4-DBD-BCL6 BTB expression plasmid were kindly provided by Dr. Ari Melnick (Weill Cornell Medical College, Department of Haematology/Oncology, New York, NY, USA.) A TK-Renilla luciferase plasmid was included as an internal control (Promega, WA, USA).
293T cells were cotransfected with (GAL4) 5 -TK-LUC and GAL4-DBD-BCL6 BTB or GAL4-DBD plus a TK-Renilla reporter construct using Lipofectamine 2000 (Thermo Fisher Scienti c). After transfection for 6 h, the cells were treated with the compounds for 24 h. A dual luciferase assay was performed according to the manufacturer's guidelines (Promega, WA, USA).

BCL6 Protein and SMRT Peptide
The DNA segment coding for the BCL6 BTB domain (5-129) with mutations in C8Q, C67R and C84N was cloned into a vector containing a C-terminal GST tag. The construct was expressed in E. coli BL21 (DE3), and the culture was incubated at 18 ℃ with 0.5 mM IPTG for 18 h. BCL6 BTB protein was puri ed by glutathione agarose resin and dialyzed with PBS. The peptide SMRT (1414-1430) with a 6His tag was synthesized by Abace Biology Company.

Homogenous Time Resolved Fluorescence (HTRF) Assay
The HTRF assay in 384-well plates (Greiner Bio-one, 784045) was performed according to the manufacturer's instructions (Cisbio). Brie y, a total volume of 8 µL of BCL6-GST protein and 6His-SMRT peptide were added to each well, and then 2 µL of diluent buffer with the compounds to be tested was added. After 1 h, a total volume of 10 µL of anti-6His-XL665 (Cisbio) and anti-GST-Tb (Cisbio) was added to reach a 20 µL nal volume. After overnight incubation, the plate was read on a microplate reader (BioTek Cytation5) at 665 nm and 620 nm.

Surface plasmon resonance (SPR) Assay
A Biacore T200 instrument (GE Healthcare) equipped with a CM5 sensor chip (GE Healthcare) was used to monitor binding interactions via SPR. The BCL6 BTB protein was immobilized on CM5 using aminecoupling at a ow rate of 10 µL/min in 10 mM sodium acetate buffer (pH 4.0). Then, 0.1 mg/mL BCL6 BTB protein was injected for 420 s, and the surface was blocked by 1 M ethanolamine. The kinetics and a nity studies were performed in 25 mM HEPES, pH 7.5, 150 mM NaCl, 0.05% (v/v) Tween-20 and 5% DMSO at 25°C. Various concentrations of compounds were injected into the ow system. Data analysis was performed using Biacore T200 Plus Evaluation Software (GE healthcare).
All animals were obtained from the Animal Centre of East China Normal University. All procedures involving animals were approved by the Animal Investigation Committee of the Institute of Biomedical Sciences, East China Normal University.
2.6. Immuno uorescence assay SUDHL4 cells were exposed to WK500B for 24 h, xed with 4% paraformaldehyde, permeabilized with 0.3% Triton X-100 and blocked with 1% BSA. The cells were incubated with the antibodies against BCL6 (CST, 14895) and SMRT (Invitrogen, MA1-842) overnight. Cells were then stained with secondary antibodies and DAPI was used to visualize the nuclei. Images were acquired on a confocal microscope (Leica).

Cell Proliferation Assay
Cell viability was tested by an MTS assay (Promega, WA, USA). Cells were seeded in 96-well plates and then treated with the indicated concentrations of compounds for 72 h. Then, 20 µL of MTS was added and incubated at 37°C for 1-2 h. The absorbance at 490 nm was measured by a SpectraMax 190 plate reader, and the IC 50 value was calculated using GraphPad Prism. In the drug combination assay, the CI value was calculated by Calcusyn software.
2.9. Cell Cycle Analysis SUDHL4 cells were treated with WK500B for 24 h. Then, the cells were harvested, washed with PBS and xed with cold 70% ethanol at 4°C for 12 h. Subsequently, the cells were washed twice with PBS and incubated with 50 µg/mL propidium iodide and 10 µg/mL RNase for 30 min in the dark at room temperature. Cells were stored at 4°C until analysis by ow cytometry (FACS Calibur, BD Biosciences).

Cell apoptosis assay
SUDHL4 cells were treated with WK500B for 24 h. Cells were collected, washed with PBS resuspended in binding buffer andincubated with Annexin V-FITC and propidium iodide in the dark for 20 min. Samples were analysed immediately by using ow cytometry (FACS Calibur, BD Biosciences).

Immunization of mice
To induce the GC reaction, 8-week-old male C57BL/6 mice were immunized with 100 µg of 4-hydroxy-3nitrophenylacetyl (NP)-(Santa Cruz) coupled chicken gamma globulin (CGG) (Cedarlane). Two days later mice were randomly assigned to the indicated groups and treated daily with drugs or vehicle (0.5% methyl cellulose) by intragastric gavage for 12 days. Mice were sacri ced at the peak of the GC response. Spleens were collected for ow cytometry and histology analysis, and serum was processed for ELISA assays.

Immuno uorescence histology
Mouse spleens were frozen in OCT compound (Sakura), and 6 µm-thick cryostat sections were stained with biotin-conjugated peanut agglutinin (Sigma) and IgD (BD Pharmingen) at 4°C overnight followed by incubation with donkey anti-rat IgG(H + L)-Alexa Fluor 488 (Invitrogen) and streptavidin-Cy3 (Biolegend) for 2 h at room temperature. After washing in PBST, the slides were mounted in Prolong Gold anti-fade reagent (Molecular Probes). Images were acquired by a microscope, and ImageJ software (NIH) was used to quantify the GC areas.
2.14. ELISA Immunoglobulin levels in serum were measured by ELISA. The titres of isotype-speci c antibodies were determined for the binding of NP-speci c antibodies to NP 5 -BSA and NP 23 -BSA on coated plates according to the manufacturer's protocol (Southern Biotech). OD values at 450 nm were measured by a plate reader (BioTek).

Xenograft tumor growth
The SUDHL4 xenograft tumour models were developed by injecting 1 × 10 7 cells into male SCID mice (6-8 weeks of age). The mice were grouped randomly when the volume of the tumour nodules reached 100 mm 3 and were treated by gavage injection with the indicated compounds or vehicle for 18 days, and the bodyweight and tumour dimensions were measured. The tumour volume was calculated using the following equation: tumour volume = length × width × width × 0.52. At the completion of the study, the mice were euthanized, and tumours and major organs were collected.

H&E staining and IHC
Tumours and tissue specimens were xed and embedded in para n. Sections were cut from the para n blocks. For IHC staining, samples were carried out using the VECTASTAIN ABC kit (Vector). Anti-Ki67 (1:250; Catalogue #ab15580, Abcam) was used as the primary antibody. For H&E staining, samples were stained with haematoxylin and eosin according to standard protocols.

Statistical analysis
The results are expressed as the mean ± SD. The signi cance of the difference between two groups was analysed by Student's t test. For animal experiments, data were analysed by two-way ANOVA. All experiments were performed at least three times, except for the animal experiments. All statistical analyses were carried out using GraphPad Prism 5.0 software. The signi cant differences in the means were determined at the level of *P < 0.05, ** P < 0.01 and ***P < 0.001.

Identi cation of WK500B as a novel BCL6 BTB inhibitor.
To identify novel BCL6 BTB inhibitors with improved activities both in vitro and in vivo, a preliminary screen was performed of a library of approximately 500 compounds using a luciferase reporter assay, which allowed us to assess BCL6-mediated transcriptional repression activity, and 2 hits were identi ed at a concentration of 10 µM among the investigated ones. Intriguingly, these effective compounds shared the same structural skeleton of diaminopyrimidine ( Supplementary Fig. 1a). Therefore, more analogs were further synthesized and modi ed based on the structural of diaminopyrimidine. Subsequently, those derivatives were then subjected to secondary screen to evaluate BCL6-mediated transcriptional repression activity and the disruption of BCL6 BTB /SMRT complex formation by a Homogenous Time-Resolved Fluorescence (HTRF) assay ( Supplementary Fig. 1a). FX1, which was previously reported as a BCL6 inhibitor 25 , was included as the positive control. Among the investigated derivatives, 7 compounds exhibited strong repression to BCL6 in luciferase reporter assay and inhibited BCL6 and SMRT interaction in HTRF assay (Fig. 1a). WK500B was identi ed as the most active analogues among the investigated compounds (Fig. 1b). WK500B almost completely disrupted the transcriptional repression function of the BCL6 BTB domain at 10 µM, which abrogated the repression of the (GAL4) 5 -TK-LUC reporter by the GAL4-DBD-BCL6 BTB fusion protein, whereas FX1 had little effect on the BCL6 BTB domain at the same concentration (Fig. 1c). Consistent with its greater inhibitory activity against BCL6 BTB versus that of FX1, WK500B blocked the interaction of the corepressor SMRT with BCL6 BTB with an IC 50 of 1.37 ± 0.12 µM, and the IC 50 of FX1 was 29.78 ± 2.45 µM (Fig. 1d). To study whether WK500B could bind to the BCL6 BTB , a surface plasmon resonance (SPR) assay was performed 33 . Surprisingly, WK500B directly bound to the BCL6 BTB domain with a dissociation constant (K D ) value of 1.61 µM (Fig. 1e). In addition, we also performed a molecular docking assay to predict the binding mode of the WK500B and BCL6 BTB domain.
As shown in Fig. 1f, the carbonyl oxygen of the acyl residues and the N atom of the pyridinyl on the left part of WK500B were bound to the relevant amino acid residue R24. In addition, two hydrogen bonds were formed between the N-H of the piperazine ring of WK500B and the N atom of the imidazolyl group of His14. In short, these results suggest that WK500B could bind to the BCL6 BTB domain.
The above results indicated that WK500B bound to the BCL6 BTB domain with high a nity and disrupted BCL6 BTB/SMRT complex formation in the extracellular space. To assess the impact of WK500B on the BCL6 BTB/SMRT interaction inside cells, an immuno uorescence assay was performed in SUDHL4 cells. As we predicted, the colocalization of BCL6 with SMRT was disrupted by WK500B (Fig. 2a), and endogenous BCL6 expression was not affected by WK500B (Fig. 2a, Supplementary Fig. S2a). Targeting the BCL6 BTB domain could avoid the adverse effects resulting from the complete abrogation of BCL6 functions 34,35 , and WK500B had no effect on in ammatory responses ( Supplementary Fig. S2b). These ndings established that WK500B bound to the BCL6 BTB domain and blocked recruitment of SMRT by BCL6.
BCL6 is a transcriptional repressor that plays a key role in GC formation and the pathogenesis of DLBCLs by repressing the expression of downstream target genes through binding with corepressors at gene promoters 20,36 . To determine the effect of WK500B on BCL6 transcriptional repression activity, we examined the expression of BCL6 target genes in DLBCL cell lines (SUDHL4 and Farage) that were exposed to WK500B. As shown in Fig. 2B, WK500B signi cantly increased the reactivation of the expression of known BCL6 target genes (p53, CDKN1A, CXCR4 and CD69) 14,15,17,25 compared with the vehicle, while the positive control compound FX1 showed little effect at the same concentration. To further explore the selectivity of WK500B, Toledo cell line, which lacks BCL6 expression was then exposed to WK500B at 10 µM for 24 h. In contrast to SUDHL4 and Farage, WK500B had no effect on any of these genes in the BCL6-independent Toledo cell line (Fig. 2c). Together, these data show that WK500B disrupts BCL6 corepressor recruitment and speci cally reactivates BCL6 target genes in BCL6-dependent DLBCLs.

WK500B signi cantly inhibits DLBCL growth and induces cell cycle arrest and apoptosis in vitro.
Most previous studies have shown that BCL6 target genes are essential for DLBCL survival and reactivating these target genes can cause DLBCL cell lethality 9,25,35 . Given that WK500B signi cantly reactivated BCL6 target genes (Fig. 2b), ve DLBCL cell lines (SUDHL4, SUDHL6, OCI-LY7, Farage and DOHH2) were exposed to WK500B for 72 h, and the viabilities of these cells were detected through an MTS assay. As expected, WK500B effectively inhibited DLBCL proliferation at low concentrations, and the IC 50 values of different DLBCL cell lines were equal to 1 µM (Fig. 3a, b). In addition, four normal cell lines, including the human normal liver cell line LO2, human skin broblast cell line HAF, human normal colon epithelium cell line NCM460 and human prostatic cell line PNT1A, were also exposed to WK500B to further explore its selectivity toward normal cells. Excitingly, WK500B had little killing effect on the normal cells even at 10 µM (IC 50 > 10 µM), indicating that WK500B selectively inhibited DLBCL proliferation with low toxicity toward normal cells (Fig. 3a, b). Moreover, WK500B had an even more signi cant effect on inhibiting DLBCL proliferation, and its IC 50 values were 30-fold lower than those of FX1 (Fig. 3b).
Subsequently, SUDHL4 cells with lentivirus expressing BCL6 shRNA were treated with WK500B. The results indicated that WK500B could effectively inhibit control vector cells in the investigated range of 1.25 − 10.0 µM in a dose-dependent manner, while it showed little effect on BCL6 knockdown cells in the same concentration range, supporting the notion that WK500B inhibits DLBCL cell growth by targeting BCL6 (Fig. 3c).
BCL6 mediates the survival and cell cycle progression of B-cell lymphoma cells 37 . We determined whether WK500B affected DLBCL cell cycle progression and cell apoptosis by using ow cytometry. The results (Fig. 3d, e) indicated that WK500B induced signi cant cell cycle arrest at the S phase and caused the dose-dependent induction of apoptosis at doses of 2.5, 5.0 and 10 µM. The positive control FX1 did not induce apoptosis or cell cycle arrest at the same concentration (data not shown). Taken together, these results demonstrated that WK500B induced the death of DLBCLs in vitro.

WK500B exhibits favourable pharmacokinetics and abrogates germinal centre formation in vivo.
To determine whether WK500B could serve as the prototype of BCL6 inhibitor for further clinical development, a human liver microsome study was primarily carried out to evaluate in vitro metabolism of WK500B. Intriguingly, WK500B had good metabolic stability in vitro (Supplementary Table 1). Encouraged by this result, pharmacokinetic analyses were subsequently performed after oral administration (p.o., 10 mg/kg) and intravenous injection (i.p., 5.0 mg/kg) of WK500B in mice to further gain insights into the preclinical pharmacokinetics. As shown in Supplementary Table 2, the plasma half-life (T 1/2 ) values were 7.93 ± 0.81 h for p.o. administration and 6.41 ± 0.87 h for i.p. injection, respectively, and relatively low plasma clearance (CL) was observed, which indicated that WK500B has improved metabolic stability in vivo. The mean values of the apparent volume of distribution (V d ) were very high for both p.o. and i.p. administration, which suggested that WK500B was widely distributed in tissues. In addition, the area under the curve (AUC 0−∞ ) was high for both p.o. and i.p. administration, indicating that WK500B was absorbed easily. Moreover, WK500B has excellent oral bioavailability (F = 82.49%). These results demonstrated that WK500B is a bioavailable candidate for oral administration and has improved druglikeness.
BCL6 is important for the formation of GCs for humoural immunity, and mice engineered with a mutation of BCL6 BTB have normal B cell development but fail to form GCs and show disruption of the a nity maturation of immunoglobulins 34 . To examine the impact of WK500B on GCs, C57/BL6 mice were immunized with NP 18 CGG and treated by gavage with daily doses of WK500B and FX1 (positive control) at 50 mg/kg two days later. After 12 days of dosing, the mice were sacri ced. As expected, the WK500B treatment group showed signi cantly abolished GC formation (GL7 + FAS + B220 + ) compared with the vehicle group (Fig. 4a). Moreover, the WK500B treatment group exhibited better inhibitory activity, with a frequency of B cells of 0.4%, than the FX1 group, in which the frequency of GC B cells was 0.9% (Fig. 4a). In addition, spleens were also subjected to immuno uorescent staining to assess the number of GCs. Ig D staining revealed normal B cell follicular structures, and staining with peanut agglutinin, a GC B cellspeci c marker, showed a profound loss of GCs (Fig. 4b), similar to the results of Flow cytometric. Notably, WK500B had a more signi cant GC inhibitory effect than FX1. Follicular helper T (Tfh) cells contribute to the development of GC B cells, and BCL6 also acts a pivotal part in Tfh cells 34,38 . In mice with BCL6 BTB mutations, the frequency of Tfh cells was depleted compared with that in BCL6 +/+ mice 34 . Similar to that observed in mice with conditional deletion of BCL6 in GC B cells, the proportion of Tfh cells (CXCR5 + PD1 + CD4 + ) was much lower in the WK500B treatment group than that in the untreated group (Fig. 4c).
Impaired immunoglobulin a nity maturation is found in mice with BCL6 BTB mutations 34 , and we next examined whether WK500B can recapitulate this phenotype by enzyme-linked immunosorbent assay (ELISA). As expected, WK500B treatment group brought about antibody-de cient mice whose titres of high-a nity immunoglobulin G1 speci c to NP5-BSA as well as the total IgG1 speci c to NP23-BSA were dramatically lower than the blank control group (Fig. 4d). Collectively, these results demonstrated that WK500B could effectively inhibit BCL6 biological functions in vivo.
3.5. WK500B effectively suppresses the growth of lymphoma without toxic effects in vivo.
To further evaluate whether WK500B could protect against DLBCL proliferation in vivo, SCID mouse xenograft models were established by subcutaneous injection of SUDHL4 DLBCL cells. When visible tumours reached approximately 100 mm 3 in size, mice were injected with 12.5 mg/kg/day and 25 mg/kg/day WK500B or 25 mg/kg/day FX1 (positive control) or vehicle by oral administration. The tumour volume and mouse body weight were detected every three days. As expected, WK500B signi cantly inhibited tumour growth and had a more potent tumour-inhibitory effect than FX1 (Fig. 5a, b and c). Importantly, the expression of BCL6 target genes in tumours was more strongly reactivated in the WK500B-treated groups than in the control group, which was consistent with the in vitro observations (Fig. 2b), providing pharmacodynamic evidence of WK500B engaging with BCL6 in tumour tissue (Fig. 5e). In addition, the tumour tissues were analysed by immunohistochemistry (IHC). Compared with control group, the proliferation marker Ki67 was dramatically decreased in the WK500B-treated groups, which con rmed the anticancer effects of WK500B in vivo (Fig. 5f). Furthermore, we tested whether WK500B bring about toxic effects in mice. No obvious toxicity was observed according to the measurement of the mouse body weight (Fig. 5d), and the mice had no abnormal behaviour or side effects during the treatment period. Moreover, no obvious organ damage was observed based on H&E staining ( Supplementary Fig. 3). To further evaluate the toxicity of WK500B in mice, C57BL/6 mice received WK500B orally at 50 mg/kg/day for 14 days, and the biochemical parameters and complete blood counts from C57BL/6 mice were examined. Intriguingly, WK500B had negligible effects on mice treated with compound (Supplementary Tables 3 and 4). Hence, these data demonstrated that WK500B is an effective anti-lymphoma agent in vivo and is nontoxic to animals.
Since lymphomas are typically highly heterogeneous, BCL6 inhibitors in combination with other targeted agents are required. BCL6 has been found to cooperate with EZH2 and PRMT5 to mediate GC formation 39,40 . Then, to evaluate whether WK500B has synergistic effects with EZH2 inhibitor or PRMT5 inhibitor, we exposed SUDHL4 cells to a combination of the PRMT5 inhibitor GSK591 or the EZH2 inhibitor GSK343 with WK500B, and we observed increased de-repression of BCL6 target genes compared to that induced by either compound alone (Fig. 6a). Next, the combinatorial activity was determined by administering increasing doses of WK500B with GSK591 or GSK343 and measuring cell viability. As expected, the combination of GSK591 with WK500B more signi cantly suppressed DLBCL cell growth than either GSK591 or WK500B alone, resulting in a synergistic combination effect (Fig. 6b). Consistent with the ndings for GSK591, synergistic killing activity in SUDHL4 cells was also observed when combining GSK343 with WK500B (Fig. 6b). Collectively, the results indicate that WK500B synergizes with the EZH2 and PRMT5 inhibitors.

Discussion
B cells transiting the GC reaction manifest many hallmarks of cancer cells, and the deregulation of BCL6 expression leads to B cell malignant transformation. BCL6 has been documented a convincing therapeutic target in DLBCL and other GC-derived lymphomas 7 . Inhibitors that disrupt interactions between BCL6 BTB and its corepressors show anticancer activity and lack toxicity 23,37 . Although a series of BCL6 inhibitors have been reported (Fig. 7a), the IC 50 values of BCL6 inhibitors are high and molecules with a higher a nity for BCL6 have not been studied in vivo (Fig. 7b), which is perhaps due to their poor pharmacokinetic properties or other reasons 26,27,29 . Hence, the development of clinical-grade BCL6 inhibitors with improved drug-likeness is an urgent mission.
Here, we conducted a screen by using a luciferase reporter assay and an HTRF assay to identify a novel potent BCL6 inhibitor, WK500B, which directly bound to BCL6 BTB , signi cantly reactivated the expression of BCL6 target genes, impacted the growth of DLBCLs in vitro and induced apoptosis and cell cycle arrest. BCL6 BTB mutant mice had an impaired GC response after immunization with a T cell-dependent antigen 34 . WK500B disturbed the biological functions of BCL6 and phenocopied the BCL6 BTB phenotype in vivo, which signi cantly inhibited GC formation, reduced the proportion of Tfh cells and impaired immunoglobulin a nity maturation. Moreover, WK500B suppressed the growth of DLBCL cells free of eliciting detectable off-target effects against normal tissues. No inhibitors that directly target BCL6 have been approved by the FDA, which may be due to their poor pharmacokinetic/pharmacodynamic properties 26,27,28,29,30 . WK500B displayed favourable pharmacokinetics with a long plasma half-life (T 1/2 ) and low plasma clearance (CL) as well as a high apparent volume of distribution (V d ) and area under the curve (AUC 0−∞ ). Extraordinarily, WK500B also has excellent oral bioavailability, which is not possessed by other BCL6 inhibitors. Taken together, WK500B is orally available and shows superior druggability compared with other reported BCL6 inhibitors (Fig. 7a, b).
In addition, other aspects of the BCL6 mechanism have been harnessed to design combined targeted therapy regimens 41 . Combinations of WK500B with an EZH2 inhibitor and a PRMT5 inhibitor showed enhanced anti-lymphoma activity and synergistic effects on the de-repression of BCL6 target genes.
Future clinical trials will be important to investigate the e cacy of targeting BCL6 in combination with other therapies. Recent reports have implicated BCL6 as an important therapeutic target in solid tumours, including glioblastoma 42 , breast cancer 43 , ovarian cancer 44 and non-small cell lung cancer (NSCLC) 45 .
Moreover, an increasing number of studies have shown that targeting the BCL6 BTB domain can overcome drug tolerance in cancer cells, 46,47,48,49,50 . We conjectured that WK500B may kills other cancer cells and overcome drug tolerance, which need more experiments to verify our hypothesis. This is of great signi cance to expand the scope of BCL6 inhibitors.

Conclusions
In summary, WK500B not only disrupted BCL6 biological functions in vitro and in vivo but also presented superior druggability, and combination therapy with WK500B and other targeted chemotherapeutic agents yielded dramatic antitumour effects. Therefore, WK500B could potentially be used as a novel effective and orally available anticancer agent for treatment of DLBCL.

Declarations
Ethics approval and consent to participate All animal experimental protocols were approved by the Animal Investigation Committee of the Institute of Biomedical Sciences, East China Normal University.

Consent for publication
All the authors agree to the content of the paper and their being listed as a co-author of the paper.

Availability of data and materials
All data generated or analyzed during this study are included in this article and its supplementary les.
(e) The effect on apoptosis of SUDHL4 cells that were treated with WK500B for 24 h.  WK500B suppresses BCL6-driven DLBCL growth in vivo. SUDHL4 cells were injected subcutaneously into SCID mice. After 18 days of treatment, the mice were sacri ced, and the tumours were photographed (a) and weighed (b). Scale bars, 1 cm. Tumour volumes (c) and body weights (d) of mice treated with different compounds were measured every 3 days. (e) mRNA expression of the BCL6 target genes from tumours was measured by RT-qPCR assays. (f) Images of IHC staining of Ki67 are shown. Scale bars, 50 μm. (*, P < 0.05; **, P < 0.01; ***, P < 0.001 versus control).  Comparison of reported small molecular BCL6 inhibitors and WK500B. (a) Chemical structures of reported small molecular BCL6 inhibitors and WK500B. (b) Comparison of reported small molecular BCL6 inhibitors and WK500B. Y, indicates that the title requirement is met. N, indicates that the datum was not reported. Otherwise, the violation is listed in detail. cLogP and tPSA were calculated using Chemdraw.