Discovery of an In Vivo Chemical Probe for BCL6 Inhibition by Optimization of Tricyclic Quinolinones

B-cell lymphoma 6 (BCL6) is a transcriptional repressor and oncogenic driver of diffuse large B-cell lymphoma (DLBCL). Here, we report the optimization of our previously reported tricyclic quinolinone series for the inhibition of BCL6. We sought to improve the cellular potency and in vivo exposure of the non-degrading isomer, CCT373567, of our recently published degrader, CCT373566. The major limitation of our inhibitors was their high topological polar surface areas (TPSA), leading to increased efflux ratios. Reducing the molecular weight allowed us to remove polarity and decrease TPSA without considerably reducing solubility. Careful optimization of these properties, as guided by pharmacokinetic studies, led to the discovery of CCT374705, a potent inhibitor of BCL6 with a good in vivo profile. Modest in vivo efficacy was achieved in a lymphoma xenograft mouse model after oral dosing.


■ INTRODUCTION
The B-cell lymphoma 6 (BCL6) protein is a transcriptional repressor required for the expansion of germinal center Bcells. 1,2 Expression is highly regulated in normal B-cells allowing for rapid proliferation during somatic hypermutation. Dysregulation of and dependence on BCL6 is regularly observed in B-cell lymphomas. 3,4 BCL6 acts as a transcriptional repressor to a broad range of genes via the recruitment of corepressors (NCOR, SMRT, or BCOR) to its dimeric BTB domain, which enables binding to key sites on DNA. 5,6 The protein−protein interaction (PPI) between BCL6 and its corepressors has been targeted as a potential therapy for BCL6-driven lymphomas. Promising small molecule inhibitors of the BTB domain have been reported by us and others with limited suitable in vivo candidates for cancer studies. 7−15 We recently reported a new tricyclic quinolinone scaffold that demonstrated improved binding affinity to BCL6 (CCT372064: BCL6 HTRF IC 50 = 4.8 nM) compared to our earlier benzimidazolone and quinolinone series (Figure 1). 13 We then optimized this scaffold for the degradation of BCL6 through modification of the 2-pyrimidine substituent of this core. Degradation required piperidine in this position decorated with either 3-methyl or 4,4-difluoro. Introducing polarity onto piperidine to decrease the log D was crucial to lower in vivo clearance and also increase the potency of inhibition and degradation. Ultimately, this optimization culminated in the discovery of CCT373566, an extremely potent (DC 50 = 0.7 nM) in vivo degrader that demonstrated strong antiproliferative activity in cells but showed only modest in vivo efficacy. 16 In this study, we report the discovery of a potent in vivo BCL6 inhibitor. Further optimization of our tricyclic quinolinone series was needed in search of an oral tool compound to validate the role of BCL6 inhibition in tumor models in mice. Initially, we aimed to improve potency through additional exploration of the 2-pyrimidine substituent. However, while potent BCL6 inhibitors (cellular IC 50 < 20 nM) were discovered, in vivo studies were limited by low free compound concentrations. Achieving an optimal pharmacokinetic profile required careful balancing of physicochemical properties that led us to decrease the molecular weight (MW) of our series. The optimization resulted in the discovery of compound CCT374705, an inhibitor with potent antiproliferative effects in vitro and sustained exposure above the predicted required concentrations.

■ RESULTS AND DISCUSSION CHEMISTRY
Final compounds were generally obtained by a single-step nucleophilic aromatic substitution (S N Ar) reaction from dichloropyrimidine intermediate 2 (Scheme 1) or, for pyridine substituents, by palladium-catalyzed cross-coupling reactions (Scheme 2). While compound 8 could also be prepared via the sequential S N Ar route, yields for the final step were low. As such, a new route was developed (Scheme 3); this required the synthesis of the substituted pyrimidine 8c, with the final compound prepared under acidic conditions. All compounds were obtained from a common aniline intermediate 1. The synthesis of 1 has been published; 13 however, previously, the synthetic route required a chiral SFC separation. Commercially sourced starting material (3S)-3-amino-3-cyclopropyl-2,2-difluoro-propan-1-ol hydrochloride, 19, was supplied at ∼85% ee. An in-house route to this compound as a single enantiomer was established (Scheme 4).
First, cyclopropanecarboxaldehyde was condensed with Ellman′s (S)-sulfinamide in the presence of magnesium sulfate and catalytic PPTS to yield imine 19a. 17 An organozinc reagent  Table 8, Starting from Common Intermediate 1 a was prepared from ethyl bromodifluoroacetate and immediately quenched with 19a via a Reformatsky reaction to yield 19b. Initially, this reaction was conducted on a moderate scale (<2 g) using preactivated zinc dust. 18 To avoid handling excessive quantities of activated zinc dust, an in situ method of activation was used on larger scale (>15 g) reactions. In an adapted procedure previously developed for multikilogram scale-up, zinc activation was achieved using DIBAL-H in the presence of a small quantity (5%) of imine. 19 Slow addition of the remaining imine produced a manageable exotherm during the reaction to 19b. The Reformatsky reaction showed diastereoselectivity (∼80% ee) similar to the commercially available amino alcohol as seen by 1 H NMR. Column chromatography could not separate the diastereoisomers at this stage. However, after the reduction of the ester by sodium borohydride to alcohol 19c, the diastereoisomers were separated by normal phase column chromatography. The tert-butylsulfinyl group was removed in acid to yield the >99% ee amino alcohol.
Improving Potency. To probe the effect of BCL6 inhibition in an in vivo xenograft model in mice, we wanted to ensure that we could achieve sustained coverage of our inhibitor. It has been shown previously that to induce an antiproliferative effect in BCL6-high cells, concentrations of degraders or inhibitors of BCL6 must be sustained for several days. 10,13,20 Based on this in vitro antiproliferative data, we hypothesized that it would be necessary to employ a dosing regimen that maintained a free concentration in vivo above the free IC 90 , as calculated from our cellular NanoBRET assay, for the duration of our in vivo studies. This approach would ensure that we are achieving sufficient and continual target engagement and, therefore, truly testing the therapeutic hypothesis of BCL6 inhibition within a xenograft model.
Our previous studies on chiral piperidine-substituted tricyclic quinolinone analogues led to the discovery of extremely potent degraders of BCL6. 16 Additionally, during these investigations, we observed striking SAR between piperidine enantiomers. While each pair of piperidine stereoisomers showed binding affinity similar to the BTB domain of BCL6, as measured by our biochemical TR-FRET assay (Table 1), only one isomer was shown to induce degradation of the protein. Optimization led to the discovery of cis-3,5-substituted degrader CCT373566, which demonstrated subnanomolar cellular degradation of BCL6 and displayed an acceptable pharmacokinetic profile and ADME properties (Table 1). In vivo studies of CCT373566 showed that free plasma concentrations remained above the calculated free DC 50 levels for over 24 h when dosed at 50 mg/ kg. The tricyclic quinolinone substituted with the opposite cispiperidine enantiomer CCT373567 demonstrated similar biochemical affinity but did not induce degradation of BCL6 and thus appeared as a good candidate BCL6 inhibitor. Unfortunately, while the cellular activity of degrader CCT373566 was excellent (DC 50 = 0.7 nM), CCT373567 showed a large decrease in activity in our NanoBRET cellular inhibition assay (IC 50 = 25.9 nM). If, as expected, CCT373567 had in vivo pharmacokinetics similar to CCT373566, we would require a dose significantly higher than 50 mg/kg to guarantee sustained inhibition of BCL6.
To improve upon CCT373567, we must either improve the cellular potency and/or increase the free exposure of the compound through optimization of PK properties ( Figure 2).
Our previous studies on degrader tricyclic quinolinone analogues demonstrated that biochemical potency gains could be achieved via modification of the 2-amino substituent of the pyrimidine ring. 16,20 Additionally, we showed that modification of this part of the molecule considerably impacts the overall physicochemical and in vitro properties.
In our alternate benzimidazolone series, morpholine and piperazine groups were among the most potent inhibitors. 20 We hypothesized that we could increase the affinity of CCT373567 through the replacement of the 6-membered piperidine with these other heterocycles.
We tested a morpholine substituent on our tricyclic quinolinone core (3, Table 2). This compound showed biochemical affinity similar to BCL6 (3.2 nM) as CCT373567, and a modest improvement in mouse microsomal clearance (57 vs 90) was also observed. The morpholine analogue 3 also showed an improvement in permeability and efflux compared to CCT373567.
By examining the crystal structure of CCT373566, 16 we hypothesized that bridged bicyclic derivatives of morpholines and piperazines could fill the space between the two solventexposed surfaces of the binding pocket more effectively and may therefore be more potent toward BCL6 ( Figure 3). Additionally, we hoped that increasing the 3-dimensional shape of the molecules could compensate for the loss of solubility that might result from increasing lipophilicity. We explored several bridged morpholine and piperazine analogues ( Table 2).
The bridged compounds (4−7) were found to be extremely potent in our TR-FRET assay, IC 50 ≤ 3 nM. However, the transition from biochemical to cellular inhibition varied between morpholine and piperazine compounds. Both isomers of the bridged N-acyl piperazine (6 and 7) suffered a large loss in affinity in cells (26.9 and 41.0 nM) and were disappointingly less potent than CCT373567. The two bridged morpholine compounds, 4 and 5, represented our most potent compounds in cells so far, 10.7 and 12.5 nM, respectively. Moreover, the addition of the bridge led to better permeability (6.7/4.6 vs 3.3) and efflux ratio (3.0/3.5 vs 6.0) compared to 3.
To fill more of the wedge-shaped pocket (Figure 3), we prepared a larger bicycle containing morpholine with an endohydroxy-piperidine, i.e., endo-3-oxa-9-azabicyclo[3.3.1]nonan-7-ol (8). It was hypothesized that we could maintain good permeability and solubility through the presence of the polar OH group with an intramolecular hydrogen bond. The combination of the larger bridge and H-bond of 8 resulted in a cellular potency of 4.5 nM, with increased solubility and lower mouse microsomal clearance compared to CCT373567.
We decided to conduct further profiling in vivo on our two most potent cellular compounds to date, 4 and 8. 8 was our most potent compound in cells and demonstrated good solubility and lower microsomal clearance. Although slightly less potent than 8, compound 4 exhibited a significantly lower efflux ratio, but the main concern was poorer solubility.
In Vivo Profiling of 4 and 8. We carried out low-dose pharmacokinetic studies, administering at 1 mg/kg i.v. (n = 3) and 5 mg/kg p.o. (n = 3) in female Balb/C mice. All mice appeared normal post dosing and 24 h post dose. Each compound had a clear solution when formulated, however, 4 came out of the solution as a cloudy suspension just before PO dosing. Consequently, 4 was poorly bioavailable (21%) compared to 8 (66%) ( Table 3), resulting in a much lower maximum concentration (C max ). Additionally, 4 had a higher free IC 90 than 8 and, as a result, the concentration, once corrected for protein binding, never reached this required value when dosed at 5 mg/kg ( Figure 4). While both compounds have low total in vivo clearance, the lower clearance of 4 compared to 8 (4.9 mL/min/kg vs 7.3 mL/min/kg) leads to a longer half-life (1.12 h vs 0.68 h) (Table 3). However, once corrected for protein binding, the unbound clearance of 4 was higher than 8, consistent with the results from our in vitro microsomal assay. Assuming linear PK, the free concentration of 8 was predicted to remain above the free IC 90 (as calculated from the NanoBRET assay) for 6 h when orally dosed at 50 mg/kg ( Figure S1).
Despite the maximum free concentration (unbound C max ) of 8 reaching well above the free IC 90 , the short half-life (<1 h) results in lower concentrations of the compound at later time points. Conversely, the unbound C max of 4 is low and limited by poor solubility and bioavailability. The poorer free cellular potency of 4 also results in the requirement for higher concentrations of compound needed to exceed the free IC 90 . Neither compound fulfilled our desired pharmacokinetic profile. We hypothesized that we needed to target a compound with a profile similar to 8 but with decreased efflux. Alternatively, a compound with properties similar to 4 but with increased potency and solubility could also fulfill our desired target profile. Both compounds also show high mouse microsomal clearance, which could also be improved, although total in vivo clearance is low.
Reducing Molecular Weight to Decrease Efflux. The PK results of 4 and 8 demonstrated that the properties of our tricyclic inhibitors needed to be carefully balanced to improve coverage above the free IC 90 .
Of the compounds tested in Table 2, 8 is the most potent in cells and has the lowest microsomal clearance and comparatively good solubility. The major disadvantage compared to the other compounds tested is low permeability and high efflux. Thus, to improve upon 8, we should aim to maintain potent cellular IC 90 and good solubility while optimizing for a decreased efflux. The compounds in Table 2 demonstrate that while log D has some influence on the efflux, the best indicator of poor permeability and high efflux in this series is an increased topological polar surface area (TPSA). Therefore, our next aim was to reduce the TPSA as much as possible to reduce the efflux while also ensuring that compounds remain soluble enough for dosing at higher concentrations. We hypothesized that lowering the molecular weight (MW) of our amine-based pyrimidine substituents would be a good strategy to allow us to decrease the TPSA without decreasing solubility. Reducing the MW of a compound reduces the need for the inclusion of polar solvating groups to compensate for additional lipophilic groups (e.g., methylene groups of morpholine and piperazine). Azetidine represents the lowest MW of possible cyclic amine substituents, and we investigated a number of examples (Table 4). The azetidines remained very potent in our biochemical TR-FRET assay (≤10 nM), however, there was a small general decrease in cellular potency compared to the six-membered cyclic amines of Table 2. Pleasingly, as hypothesized, the caco-2 data was found to correlate better with TPSA than log D, suggesting that our strategy of decreasing TPSA does lead to improvement in permeability and efflux. An example of this is the comparison of 3-methoxy-substituted azetidine, 10, with 3cyano azetidine, 12. Both were measured to have a similar log D, 2.7 and 2.6, although there was a larger difference in TPSA, 94 and 108, respectively. Whereas 10 showed good permeability (15 × 10 −6 cm·s −1 ) and a low efflux ratio (1.4), 12 was poorly permeable (2.8 × 10 -6 cm·s −1 ) and had a substantially higher efflux ratio (22).
Unfortunately, the microsomal clearance in the mouse of the azetidine series was generally poorer than the 6-membered rings, except for the 3,3-difluoroazetidine, 11, which had the lowest microsomal clearance measured to date. The difluoro analogue was also the most potent azetidine in our cellular NanoBRET assay (31 nM), with a low efflux ratio. We therefore chose 11 as the best representative of this subseries to test in vivo.
In Vivo Profiling of 11. A pharmacokinetic study, dosing at 1 mg/kg i.v. (n = 3) and 5 mg/kg p.o. (n = 3), was carried out in female Balb/C mice. All mice appeared normal post dosing and 24 h post dose. The half-life of 11 was increased compared to 8, despite a lower volume of distribution, due to a drastic improvement in total in vivo clearance ( Table 5). As observed previously for 4 and 8, there was a good correlation between our total microsomal clearance measured in vitro and our unbound clearance in vivo. However, while the low solubility of 11 did not result in precipitation in the dosing solution, the bioavailability was moderate. This, alongside an increase in plasma protein binding, leads to a reduced free concentration, C max , compared to the more potent inhibitor 8. Despite this, the decreased total and free clearance led to the highest free drug concentrations at longer time periods to date ( Figure S2). Importantly, the free concentration of 11 was expected to remain above the free NanoBRET IC 90 for ∼8 h when dosed at 50 mg/kg, assuming linear PK, representing an improvement compared to all previous in vivo experiments (Figure 5).
Removing Substituents to Increase Solubility. Despite the improvement in coverage gained by 11, we aimed to find a compound that could give us an even longer time above IC 90 in vivo. The initial free concentration of the difluoroazetidinecontaining compound was limited by bioavailability, and we thus sought to improve the solubility further without drastically modifying the TPSA or log D. We hypothesized that we could achieve this by exploring compounds with even lower MW. Interestingly, our previously reported inhibitor CCT372064 (14) (Figure 1) contained no substituent in the 2-position of pyridine, has a considerably lower MW than 11, and yet was found to be reasonably potent in both our biochemical and cellular assays (Table 6). 13 Despite this, no in vivo experiments were conducted due to poor permeability and high efflux. As the potency of CCT372064 was achieved with 6-chloro and 5nitrile substitutions of pyridine, we investigated other substituted pyridines (Table 6).  16 The yellow area highlights space in the pocket to be filled with bridged compounds. The backbones of the two BCL6 monomers are colored as blue and pink ribbons. Selected water molecules are shown as red spheres and Hbonds as yellow dashed lines.  We sought to lower the TPSA of CCT372064 by replacing 5nitrile with a 5-Cl substituent, 15. Pleasingly, this change did reduce the TPSA (68 vs 92), and a large improvement in permeability (9.1) and efflux (2.9) was observed for 15.
Unfortunately, despite the improved permeability of the dichloro compound, there was a 5-fold decrease in the cellular potency compared to CCT372064. Additionally, the solubility was poor (6 μM).
We decided to investigate the 6-position with a larger (methyl − 16) and smaller (fluoro − CCT374705 (17)) substitution. The addition of a 6-methyl group to our inhibitor 16 did not have the desired outcome, the cellular potency decreased (129 nM), and, in addition, the microsomal clearance was significantly increased (112). The smaller fluoro substitution of CCT374705 conferred similar biochemical and cellular potencies (6 and 22 nM) to CCT372064. The 5-chloro-6-fluoro substituted compound showed good permeability (17) and a low efflux ratio (2.3). We then decided to swap the two substituents to see if we could find any improvement. The 6chloro-5-fluoro substituted pyridine, 18, was less potent and had poorer permeability and efflux compared to CCT374705.
Therefore, the best combination of 5,6-substitution was found to be 5-chloro-6-fluoropyridine CCT374705, demonstrating good potency in our biochemical (6 nM) and cellular (22 nM) assays. In addition to the good potency of CCT374705, the 5chloro-6-fluoro substitution displayed the lowest microsomal clearance of all our inhibitors to date (12).   The crystal structure of CCT374705 bound to BCL6 was solved and showed an identical binding conformation as CCT372064 ( Figure 6). The key interactions, as seen in all previously reported tricyclic quinolone structures, were maintained (pyridine sandwiched between Tyr58 and Asn21, with a π-π interaction with Tyr58, H-bond interactions with Met51, Ala52, and Glu115, and the 7-membered ring filling a subpocket defined by residues His14, Asp17, Val18, and Cys53 of BCL6) ( Figure S3). The two pyridines of CCT374705 and CCT372064 are well aligned, with the 5,6-substitutions (Cl, F and CN, Cl) occupying the same space in both structures, pointing toward Leu25.
Due to the combination of good cellular potency and promising in vitro PK properties, we chose to conduct a pharmacokinetic study on CCT374705 to see if these favorable properties translated in vivo. Safety profiling of CCT374705 was carried out, and all targets (78) showed a K d above 1 μM, although some targets (11/78) did show activity at concentrations below 10 μM (see Supporting Information). A kinase panel (468) confirmed selective activity and minimal off-target interactions.
In Vivo Profiling of CCT374705. A pharmacokinetic study, dosing at 1 mg/kg i.v. (n = 3) and 5 mg/kg p.o. (n = 3), was carried out in female Balb/C mice. All mice appeared normal Table 6. Structure−Activity Relationships of 5-and 6-Substituted Pyridines c a Data represent the geometric mean of at least three replicates. See Tables S1 and S2 for full statistics. b Measured log D determined using the Chrom log D method. c † indicates n = 1 and * indicates n = 2. post dosing and 24 h post dose. The oral bioavailability of CCT374705 was moderate (48%), only showing a modest increase on that of 12. However, this improvement, combined with low total clearance and the lowest plasma protein binding (ppb) measured thus far, resulted in the highest free concentrations in vivo ( Figure S1). The unbound clearance was improved compared to all previously tested compounds leading to the best half-life and longest exposure above the free IC 90 observed ( Table 7). The free concentration was found to remain above the calculated free NanoBRET IC 90 for over 8 h when orally dosed at 5 mg/kg (Figure 7).
The drastic increase in free concentration at later time points was a significant improvement on all previous in vivo results. The antiproliferative activity of CCT374705 was tested in 14-day assays in a range of BCL6-dependent (HT, Karpas 422, SU-DHL-4, and OCI-Ly1) and independent (OCI-Ly3) cell lines ( Table 8). The most potent effect was seen in both OCI-Ly1 and Karpas 422 cell lines, where sub-100 nM GI 50 was observed. As seen with our previously reported inhibitors, despite being high BCL6-expressing, the cell lines HT and SU-DHL-4 were less sensitive to inhibition of BCL6. 13,16 We performed a PK linearity study in SCID mice with CCT374705, dosed at 5 (n = 3), 20 (n = 3), and 50 (n = 3) mg/ kg p.o. in a solution formulation previously developed for degrader CCT373566 (Figure 8). 16 A linear increase in exposure was observed with increasing dose, and we found that free concentrations of CCT374705 above the free IC 90 were achieved at 20 and 50 mg/kg dosing for over 15 and 24 h, respectively. The in vitro results established that CCT374705 showed the strongest antiproliferative effect in Karpas 422 cells so an in vivo study was undertaken to determine if the promising in vitro efficacy could be replicated in a Karpas 422 xenograft model in mice. The compound was dosed orally (50 mg/kg, bid, n = 10) for 35 days and was well tolerated with no body weight losses observed. Post treatment analysis showed that free concentrations of CCT374705 remained well above calculated free IC 90 for over 12 h post dose ( Figure 9A). Unlike in the case of degraders (depletion of BCL6), there is no obvious biomarker for inhibitors of BCL6. We used quantitative RT-PCR to measure gene expression changes in a panel of known BCL6 target genes in different DLBCL lines (data not shown) and found that ARID3A was consistently derepressed. We were able to detect a significant increase in ARID3A mRNA expression in our CCT374705-treated Karpas 422 tumor xenografts, indicat-  ing that we were achieving modulation of BCL6 activity ( Figure  9B). Despite sustained exposure of CCT374705 and clear in vivo target engagement, as demonstrated by ARID3A mRNA levels, only a modest slowing of tumor growth was observed compared to the vehicle control group ( Figure 9C). After 35 days, the tumor growth inhibition ratio (T/C) of the CCT374705treated group was 0.55.

■ CONCLUSIONS
The aim of this work was to find a potent compound to investigate the effect of BCL6 inhibition in an in vivo mouse model. Our primary aim was to optimize the cellular potency of our previously published inhibitor CCT373567, the nondegrading isomer of our potent BCL6 degrader CCT373566. While we were able to improve the cellular potency of our inhibitors through the addition of polarity on our 2-pyrimidine substituents, we struggled to balance the increase in polarity with pharmacokinetic performance, particularly efflux (8). We decided to focus on the reduction of TPSA, which was found to be the property best correlated to our in vitro efflux ratios. Due to the MW of our compounds, the removal of polarity negatively affected our physicochemical properties and solubility, particularly preventing adequate dosing (4). We decided to reduce the MW of our inhibitors, switching from 6-membered piperidines and morpholines to 4-membered azetidines. Pleasingly, where TPSA was decreased, the azetidine-substituted compounds did show a decrease in efflux. Despite a decrease in potency, 11 showed improved exposure compared to 8 at later time points due to lower in vivo clearance. However, total and free concentrations were limited by low bioavailability, and we decided to further decrease the MW of our inhibitors to ensure higher solubility. We focused on pyridine-containing compounds substituted in both the 5-and 6-positions, analogous to our previously reported tricyclic quinolinone inhibitor CCT372064. The Cl-and F-substituted pyridine-containing compound CCT374705 had the best balance of properties, including increased solubility and microsomal clearance similar to CCT374284. The free concentrations in vivo were the best seen for all our inhibitors, with the free concentration of CCT374705 remaining above the free IC 90 for over 24 h when dosed at 50 mg/kg. Notably, it demonstrated significantly improved unbound in vivo clearance compared to all previously tested compounds. Ultimately, we found that optimization of the pharmacokinetic profile of our inhibitors through decreases in molecular weight led to better in vivo inhibitors, regardless of any reduction in potency.
We chose to progress CCT374705 to an efficacy study using a Karpas 422 xenograft model, the cell line in which CCT374705 had the most antiproliferative effect. The goal of this work was to study the effect of BCL6 inhibition in vivo. To assess this, we wanted to ensure complete coverage of our inhibitor (>IC 90 ) for the entirety of the study. To achieve this high level of sustained exposure, the study was conducted with a twice-daily 50 mg/kg oral dosing regimen. We were able to confirm target engagement with BCL6 by measuring an increase in ARID3A mRNA expression. PK/PD analysis showed that free concentrations of CCT374705 remained well above the free IC 90 for over 12 h post dose. However, similarly to our previously reported efficacy results with our BCL6 degrader CCT373566, only moderate in vivo efficacy was observed.
Having optimized the in vivo pharmacokinetic profile of CCT374705, this compound is a suitable probe to examine the role and function of BCL6 in diseases within mice models.

■ EXPERIMENTAL SECTION
All in vivo experiments were carried out according to the UK guidelines for animal experimentation. Cell lines were supplied by Public Health England, UK (ECACC). Cell lines were authenticated by STR profiling and were routinely screened for Mycoplasma using an in-house PCRbased assay.
General Synthetic Information. All anhydrous solvents and reagents were obtained from commercial suppliers and used without further purification. Evaporation of solvent was carried out using a rotary evaporator under reduced pressure at a bath temperature of up to 60°C. Flash column chromatography was carried out using a Biotage purification system using SNAP KP-Sil or Sfar cartridges or on reversephase mode using SNAP Ultra C18 cartridges. Semipreparative separations were carried out using a 1200 Series Preparative HPLC over a 15-min gradient elution. Microwave-assisted reactions were carried out using a Biotage Initiator microwave system. The final compounds were purified to ≥95% purity. NMR data was collected on a Bruker Avance 500 spectrometer equipped with a 5 mm BBO/QNP probe or on a Bruker Avance Neo 600 spectrometer equipped with a 5 mm TCI Cryo-Probe. NMR data is presented in the form of chemical shift δ (multiplicity, coupling constants, integration) for major Figure 9. In vivo efficacy of CCT374705. (A) The PK/PD study with CCT374705 at 50 mg/kg po. BCL6 levels in the tumor were quantified using capillary electrophoresis and normalized to a GAPDH loading control and are shown as black (vehicle-treated) or white (compoundtreated) bars. Free compound levels at 4 and 12 h are shown (purple dots); the dashed, black lines represent the calculated free, cellular (from NanoBRET assay) IC 50 and IC 90 values (nM). (B) Relative ARID3A mRNA levels after treatment with CCT374705 (purple) at 50 mg/kg PO as measured by TaqMan PCR assay. Sampling took place at 12 h post dose. (C) Tumor growth during the efficacy study with CCT374705 at 50 mg/kg PO BID for 35 days. Tumor xenografts were prepared by subcutaneous injection of 1 × 10 7 Karpas 422B cells in female SCID mice, with dosing of compound commencing 21 days after injection, to mice with xenografts of ∼0.5 cm 3 , as described in more detail in the Supporting Information. Sampling took place at 4 and 12 h post dose. All experiments were carried out according to the U.K. guidelines for animal experimentation. diagnostic protons, given in parts per million (ppm) relative to tetramethylsilane (TMS), referenced to the internal deuterated solvent. HRMS was assessed using an Agilent 1200 series HPLC and a diode array detector coupled to a 6120 time-of-flight mass spectrometer with a dual multimode APCI/ESI source or on a Waters Acquity UHPLC and a diode array detector coupled to a Waters G2 QToF mass spectrometer fitted with a multimode ESI/APCI source.
Molecular formula strings with associated biochemical assay data and calculated properties (CSV) Selectivity data for CCT374705 (XLSX) Protein production, purification, and crystallography (BCL6 constructs used, methods for expression, purification, and crystallography, including data collection, processing, and refinement); biological assay conditions (methods for TR-FRET, MSD, and cell proliferation assays); physicochemical and in vitro DMPK assay conditions (methods for NMR and HPLC solubility, log D, microsomal clearance, caco-2, and protein binding). In vivo PK and PD experimental methods, including preparation of tumor xenografts. Further analytical data for compound 21 (CCT374705), including LCMS, chiral HPLC, and qNMR; summary statistics and individual replicate values of TR-FRET and NanoBRET assays (Tables S1 and S2); crystallographic data collection and refinement statistics (Table S3); composition of solution formulation for CCT374705 (Table S4); qNMR purity analysis for CCT374705 (Table S5); free mean mouse (BALB/c) blood concentrations (nM) of 9 after PO dosing at 5 mg/kg