An unbiased, cell-based screen for Pax protein inhibitors identifies triazol derivatives that suppress target gene expression

SUMMARY

The Pax family of developmental control genes are frequently deregulated in human disease. In the kidney, Pax2 is expressed in developing nephrons but not in adult proximal and distal tubules, whereas polycystic kidney epithelia or renal cell carcinoma continue to express high levels. Pax2 reduction in mice or cell culture can slow proliferation of cystic epithelial cells or renal cancer cells. Thus, inhibition of Pax activity may be a viable, cell type specific therapy. We designed an unbiased, cell-based, high throughput screen that identified triazolo pyrimidine derivatives that attenuate Pax transactivation ability. We show that BG-1 inhibits Pax2 positive cancer cell growth and target gene expression, but has little effect on Pax2 negative cells. Chromatin immunoprecipitation suggests these inhibitors prevent Pax protein interactions with the histone H3K4 methylation complex at Pax target genes in renal cells. Thus, these compounds may provide structural scaffolds for kidney specific inhibitors with therapeutic potential.

eTOC Blurb

Bradford et al report the identification of small molecule inhibitors for the Pax2 family of developmental regulatory proteins. Such inhibitors can suppress proliferation of Pax2 expressing cancer cells and may provide scaffolds for future anti-cancer compounds.

Graphical Abstract

INTRODUCTION

The Pax (paired-box) family of DNA binding proteins encompass a family of developmental regulatory proteins that are also associated with a variety of disease states. Pax2, along with Pax5 and Pax8, constitute subclass II of the Pax family and are expressed during early stages of central nervous system, urogenital, B-lymphoid, and thyroid development (Adams et al., 1992; Bouchard et al., 2002; Mansouri et al., 1998; Plachov et al., 1990; Stoykova and Gruss, 1994; Torres et al., 1995; Urbanek et al., 1997; Urbanek et al., 1994). Pax2 is highly expressed in the intermediate mesoderm and is required for conversion of metanephric mesenchymal cells to the epithelia of the developing kidney nephrons (Rothenpieler and Dressler, 1993; Torres et al., 1995). Mice with homozygous Pax2 null alleles exhibit bilateral renal agenesis and die before birth. Whereas, Pax2 haploinsufficiency in both mice and humans produces hypoplastic kidneys and is associated with pediatric and adult onset chronic renal disease (Bower et al., 2012; Sanyanusin et al., 1995; Schimmenti et al., 1997). Pax2 expression remains strong in proliferating epithelia derived from the intermediate mesoderm but is down-regulated as nephron epithelial cells mature and exit mitosis (Dressler and Douglass, 1992; Laszczyk et al., 2020). However, Pax2 expression remains in the collecting ducts and renal medulla, which are derived from the branching ureteric bud. Pax8 is also expressed in developing nephron epithelia, but unlike Pax2, continues to be expressed in adult proximal and distal tubules. In adults, Pax2 and Pax8 cooperate to regulate urine concentration in the renal medulla through the activation of urea transporters and aquaporins (Laszczyk et al., 2020).

Despite its essential function in kidney development, deregulated Pax2 expression underlies several renal deformities and diseases. In transgenic mice, persistent overexpression of Pax2 leads to renal cyst formation similar to nephrotic syndromes (Dressler et al., 1993). Ectopic Pax2 expression is observed in polycystic kidney disease (PKD) (1994; Ostrom et al., 2000; Qin et al., 2012; Winyard et al., 1996), in renal cell carcinoma (Daniel et al., 2001; Gnarra and Dressler, 1995; Gokden et al., 2008; Igarashi et al., 2001), Wilms’ tumor (Dressler and Douglass, 1992), and in certain cancers of the female reproductive tract (Rabban et al., 2010; Zhai et al., 2010) and the nervous system (Burger et al., 2012). In both renal cancer cells and in polycystic kidney epithelia, reduced gene dosage or antisense oligonucleotide inhibition of Pax2 reduces proliferation and slows cyst growth (Gnarra and Dressler, 1995; Ostrom et al., 2000; Stayner et al., 2006). A reduction of Pax2 expression by siRNAs also sensitizes renal cancer cells to cisplatin induced apoptosis, suggesting possible new avenues of intervention (Hueber et al., 2006).

Studies with the Pax2/5/8 family of proteins indicates both activating and repressive functions at potential targets genes. While these three Pax proteins have an identical DNA binding paired domain, the carboxy terminal domain of Pax2/5/8 have unique features that are thought to regulate function. The carboxy terminal domain of Pax2 is necessary for transcriptional activation and is phosphorylated at Serine and Threonine residues in response to c-Jun-N-terminal Kinase (JNK) signaling (Cai et al., 2002; Lechner and Dressler, 1996). Pax2/5/8 can recruit the KMT2D histone H3K4 methyltransferase complex to chromatin through interactions with the adapter protein PTIP, resulting in H3K4me2/3 at Pax binding sites (Patel et al., 2007). The Pax2/5/8 proteins also contain an octapeptide sequence that binds the co-repressor Grg4/Tle4 (Eberhard et al., 2000). Expression of Grg4 recruits a protein phosphatase (PPM1b) that can dephosphorylate the carboxy terminal domain of Pax2 to displace PTIP and recruit a Polycomb repressor complex (Abraham et al., 2015; Patel et al., 2012). Thus, Pax proteins can act as epigenetic regulators to imprint active or repressive histone marks on chromatin depending on interactions with specific co-factors (Patel et al., 2014).

Given that ectopic or deregulated expression of Pax proteins can promote survival and proliferation of specific cell types in cancer and renal disease, inhibition of Pax protein activity may be a viable route for developing more cell-type specific therapies (Grimley and Dressler, 2018). Using homology modeling and computational screening, the small molecule EG1 was identified that targets the DNA binding domain of Pax2 with approximately 50% maximal inhibition at 10 μM (Grimley et al., 2017). In this report, we describe an unbiased, cell based, high-throughput screen for small molecules that inhibit Pax2 transcriptional activation. More than 60,000 chemical compounds were screened to ultimately yield three structurally similar triazolo pyrimidine compound derivatives that each exert inhibitory actions on Pax2 transactivation ability. These compounds can significantly alter the proliferation and viability of human Pax2-positive renal carcinoma cells in vitro. Putative Pax2 targets, such as Cadherin-6, UTA-1, and AVPR2, are reduced in expression when cells are cultured with inhibitors. Furthermore, chromatin immunoprecipitation experiments indicate that these inhibitors do not affect DNA binding of Pax proteins to chromatin, rather they inhibit the ability of Pax proteins to recruit PTIP and promote H3K4 methylation at Pax binding sites.

RESULTS

A trio of Pax2 inhibitors identified by cell-based HTS

To discover small molecule chemical modulators of Pax2 with diverse mechanisms of action we designed an unbiased cell-based high-throughput screening assay. Human embryonic kidney (HEK293) cells containing an integrated luciferase construct driven by a Pax Response Sequence (PRS4-Luc) were transiently transfected with a Pax2 expression construct (Fig. 1A) and thereafter used to screen a primary set of 69,125 diverse chemical compound library in the Center for Chemical Genomics (CCG) (Fig. 1B, C). Each of the 69,125 chemical compounds from the primary set of selected structures were initially tested once at a concentration of 8 μM. Of the 69,125 compounds initially tested, 2,688 were identified as having activity in our HTS assay. Whereby, 1,960 were inhibitory and 728 were excitatory chemical modulators of Pax2 activity. Each plate contained transfected negative control for inhibition, CMV-Pax2b with DMSO alone, and a positive control for inhibition Pax2^C63Y, a mutant that cannot bind PRS4 and activate luciferase (Cai et al., 2003; Grimley et al., 2017).

Figure 1. HTS for small-molecule modulators of Pax2 transcription activation.

HEK293 cells carrying the integrated PRS4-Luc reporter construct (A) were transiently transfected with either a wildtype or mutant Pax2 expression vector. B) The transfected cells were plated in 384-well plates and treated with either vehicle (DMSO) or compound for 18 hours before luminescence was read. HTS showing negative control wells (blue; CMV-Pax2b), positive control wells (red; CMV Pax2b^C63Y), and sample wells (green; CMV-Pax2b with compound). The graph shows percent activity, which is determined on a per plate basis and is based on the averaged raw luminescence values for the positive and negative control wells. Since the positive control used in this screen is the DNA-binding deficient Pax2 mutant, samples with positive activity values (top of graph) are inhibitors while those with negative activity values (bottom of graph) are activators of Pax2 mediated transcription activation. C) Summary of HTS strategy, re-testing, confirmation, and remining results.

All potential hits were re-screened in triplicate in the PRS4-luc cells. A counter screen utilized BRE-Luc cells (Bradford et al., 2019), which expressed low-level, constitutive luciferase but did not respond to Pax2 expression, as a means to detect general inhibition of transcription, translation, or luciferase enzymatic activity. These triage measures yielded 233 inhibitors and 110 activators that were deemed suitable candidates for dose-response analysis. The compounds were first dose-response tested using stock solutions in the CCG compound collection. Thereafter, confirmatory dose-response testing followed using fresh powder purchased from commercial sources. In total, 48 inhibitor and 2 activators of Pax2 were identified from the initial HTS screen, counter screen, and dose response testing. Many of the top 50 hits displayed satisfactory dose-response activity and propitious potencies with single digit IC₅₀’s. Of the 50 hits, 37 were singletons, while 11 clustered into 3 subgroups with similar scaffolds. However, the largest cluster contained metabolically unstable structures. Therefore, we initiated additional screens in the CCG compound collection to expand the initial 50 hits and identify metabolically stable scaffolds that would be amenable for SAR analysis and secondary screening assays.

We utilized cheminformatics tools to remine the entire CCG library to identify the untested scaffolds in our primary HTS screen. A 75% similarity search of the entire CCG library using the initial 50 hits as queries returned 369 untested compounds. Dose-response analysis was carried out on the 369 compounds by using 8-point titration curves generated with a 1.67 dilution factor, again in PRS4-Luc cells transiently transfected with a Pax2 expression construct. The titration curves were devised in the same manner as the primary HTS titration curves, with a concentration range from 1.38 μM to 50 μM.

The dose-response analysis yielded 122 active compounds. A cluster analysis of the chemical structures was performed on 122 compounds using the DataWarrior software (Sander et al., 2015) resulting in 34 clusters with 70% similarity. A final set of 15 compounds were selected for purchase of fresh powder based on the most active compound in terms of pAC₅₀ from the dose-response testing in each cluster and consideration of the overall structural diversity. Some clusters were eliminated from further analysis due to the reactive filter flags, lack of activity range within clusters, and the undesirable chemical structures. We also considered the overall structural diversity when selecting the final set of compounds for purchase of the fresh powder. The pAC₅₀ values for the 15 selected compounds ranged from 4.44 to 6.06 μM. Using 384 well plates, we performed in duplicate 8-point titration assays of the 15 commercially acquired compounds using both PRS4-Luc cells and BRE-Luc cells. Testing from the fresh powder confirmed the dose-response activities of the 15 compounds generated from CCG stock solutions. Further, counter screening analysis demonstrated the selective activity of BG-1, BG-2, and BG-3 in PRS4-Luc cells over the BRE-Luc cells (supplemental Fig. S1 A–C). However, BG-1 and BG-2 did indicate signs of cytotoxicity and/or decreased selectivity above the 17.93 μM and most evidently at the 50 μM concentrations tested, whereas compound BG-3 did not indicate signs of cytotoxicity.

This trio of compounds each contain a triazolo pyrimidine core and collectively have a range of activities in the PRS4-Luc cells. Using 12-point titration curves generated with a dilution factor of 2.0, we found the compounds displayed sigmoidal dose-response curves with inflection points at 1.5 μM (BG-1), 3.2 μM (BG-2), and 5.5 μM (BG-3) (Fig. 2A). Fold reductions of 0.58, 0.78, and 0.82 at 1.56 μM were also calculated (Fig. 2A, B). Chemical structures of this trio of compounds also indicate resistance to metabolic degradation which is a favorable property for in vivo validation studies (Fig. 2C). Moreover, a distinct range of activities across the compound trio are induced by nominal changes across their chemical formulas (Fig. 2C). Since BG1 had the lowest IC₅₀, we also tested its ability to inhibit Pax5 and Pax8 activation of the PRS4-Luc reporter (Fig. 2D). Given the homology between Pax2, 5, and 8 and their ability to activate the PRS4-Luc reporter, it was not entirely surprising that all three Pax proteins were similarly inhibited by BG1.

Figure 2. Dose-dependent inhibition of Pax2 dependent activation by a trio of triazolo pyrimidine derivatives.

A) Pax2 transfected PRS4-Luc cells cultured in increasing concentrations of BG inhibitors were used to generate a 12-point dose response curves in duplicate (DF = 2.0). The maximum and minimum concentrations tested were 50 μM and 0.02 μM, respectively (grey circles). Both positive and negative controls consisted of Pax2 transfected PRS4-Luc cells treated with DMSO (black squares) and PRS4-Luc cells lacking Pax2 and without compound or vehicle treatment (black diamonds). The accompanying bar plot reports the fold reduction (FR) of luminescence in BG treated relative to vehicle treated Pax2 transfected PRS4-Luc cells. B) Tabulation of IC₅₀’s and FR at 1.56 μM for the BG inhibitor trio. The PubChem SID numbers are listed below the designated compound name. C) Chemical structures of the BG trio of Pax2 inhibitors. DF: dilution factor, FR: fold reduction, IC₅₀: 50%-maximal inhibitory concentration, and vehicle: DMSO/D6. Three-parameter curve fitting was used to fit logistic/sigmoidal curves to the dose-response data (Hill Slope = −1) using GraphPad Prism 8. Error bars represent one standard deviation from the calculated mean data points.

To directly assess metabolic degradation, we examined compound stability when exposed to mouse liver microsomes fortified with NADPH. The percent of compound remaining was measured at 5, 10, 15, 30, 45, and 60 minutes (Fig. 3A), generating a half-life estimation of 7.89, 46.24, and 25.39 minutes for BG-1, BG-2, and BG-3 respectively when plotted (Fig. 3B). The high clearance drug verapamil was used as a control. Combined, the structure and half-life suggest the compounds may have moderate stability in vivo. To examine in vivo clearance, a pharmacokinetic analysis was performed. Mice received a single dose of each compound orally (15 mg/kg) or intravenously (5 mg/kg), and blood collected at the time points indicated (Figure 3C–E). After isolation of plasma, concentration of each compound was determined and graphed over time (Fig. 3C–E). While BG-1 had the shortest half-life based on the microsomal stability assay (Fig. 3A, B), when given intravenously it had the lowest rate of clearance compared to BG-2 and BG-3 (Fig. 3 C–E). This suggests that out of the three compounds, BG-1 may be the most viable candidate for in vivo use.

Figure 3. Metabolic stability and pharmacokinetics.

BG-1, BG-2, and BG-3 were tested for stability in vitro with NADPH-fortified pooled mouse liver microsomes. (A) Samples were withdrawn and activity halted at 5, 10, 15, 30, 45, and 60 minutes. The percent of compound remaining at each time-point is recorded and (B) graphed as a function of time, including the control high clearance drug verapamil (blue). BG-1 red, BG-2 green, BG-3 purple. (C-E) In vivo plasma clearance was examined after dosing mice once intravenously (IV; 5 mg/kg; bule) or orally (PO; 15 mg/kg; red). Blood was collected at 0.5, 2, 4, and 7 hours, plasma isolated and percent of each compound calculated and graphed as a function of time for (C) BG-1, (D) BG-2, and (E) BG-3. Note that while BG-2 was the most stable in vitro, BG-1 was the most stable compound in vivo.

Pax2 inhibitors negatively regulate RCC proliferation and survival

Class II Pax family members are integral to malignant cell growth and survival in proliferative renal diseases. Inhibition of Pax2 with antisense oligonucleotides has been shown to decelerate the growth of human renal tumor cells (Gnarra and Dressler, 1995; Hueber et al., 2008; Hueber et al., 2006). Therefore, to examine the effects of the HTS identified Pax2 inhibitors on proliferating human renal carcinoma cells (RCC), we subjected Pax2 expressing RCC for 24 hours to increasing concentrations of BG-1 and assessed for levels of phosphorylated histone H3 (pHH3), a marker of mitotic cells (Fig. 4). A significant reduction in pHH3 levels was observed at the IC₅₀ of BG-1, 1.5 μM, as well as at 15 μM of BG-1 24 hours after treatment (Fig. 4A). Conversely, in ovarian carcinoma (SKOV-3) cells lacking Pax2 expression (Grimley et al., 2017) there was no significant change in levels of pHH3 between vehicle (DMSO) and BG-1 treated cells at each concentration tested (Fig. 4B). Similarly, in human embryonic kidney (HEK293) cells lacking Pax2 expression, increasing concentrations of BG-1 did not significantly change levels of pHH3 (Fig. 4C). These results indicate that Pax2 expressing RCC have decreased mitotic activity in the presence of BG-1. To assess for cell survival after treatment with BG-1, we conducted luciferase ATP assays (Fig. 5). RCC and SKOV-3 were treated overnight with increasing BG-1 concentrations, similar to our proliferation assays. Both experimental cell lines were then lysed to measure luminescence, which is proportional to the quantity of ATP, a marker of metabolically active cells. At each concentration tested, BG-1 significantly and dose-dependently lowered the percentage of viable RCCs (Fig. 5A). A significant drop in metabolically active cells was observed in SKOV-3, however, only at tenfold the IC₅₀ of BG-1, 15 μM (Fig. 5B). That BG-1 has a significant effect on both the proliferation and viability of Pax2 expressing RCC at its IC₅₀ and not Pax2 negative SKOV-3 cells, suggests that BG-1 may be interacting with a specific target that regulates Pax2 activity in RCC.

Figure 4. BG-1 inhibits renal carcinoma cell proliferation.

Immunoblots of whole cell protein lysates from Pax2-positive-RCC cells (A), Pax2-negative-SKOV-3 cells (B), and HEK293 cells (C) treated for 24 hours with increasing concentrations of BG-1. Membranes were probed for phospho-HH3, total-HH3, and β-Actin. Quantification of phospho-HH3 abundance by densitometry accompanies each immunoblot. For each lane, the signal of phospho-HH3 was normalized to β-Actin to control for loading variability. Final data are expressed as the mean (n= 3 biological replicates) ratio of phospho-HH3 to β-Actin with error bars representing one standard deviation from the mean. ***p = 0.0001 to 0.001, ****p < 0.0001, ns (not significant): p ≥ 0.05 relative to vehicle. Significance was determined by ANOVA, Dunnett’s post-hoc multiple comparisons test.

Figure 5. BG-1 decreases survival and Cdh6 expression in renal carcinoma cells.

A) Pax2-postive-RCC and B) Pax2-negative-SKOV-3 cells were subjected to increasing concentrations of BG-1 for 18 hours. Luminescence, which is proportional to ATP levels was used to measure metabolically active cells. Note that BG-1 significantly reduces Pax2-positive-cell viability in a dose-dependent manner. *p = 0.01 to 0.05, **p = 0.001 to 0.01, ****p < 0.0001, ns (not significant): p ≥ 0.05 relative to vehicle. Significance was determined by ANOVA, Dunnett’s post-hoc multiple comparisons test. C) Immunoblot of whole cell lysates from Pax2-positive RCC cells subjected to increasing concentration of BG-1 for 24 hours. Membranes were probed for Cdh6, Pax2, and β-Actin. The immunoblot is accompanied by the densitometry quantification of Cdh6 abundance (D). For each lane, the signal of Cdh6 was normalized to β-Actin to control for loading variability. Final data are expressed as the mean (n= 3 biological replicates) ratio of Cdh6 to β-Actin with error bars representing one standard deviation from the mean. *p = 0.01 to 0.05, ***p = 0.0001 to 0.001, ns (not significant): p ≥ 0.05 relative to vehicle. Significance was determined by ANOVA, Dunnett’s post-hoc multiple comparisons test.

Cadherin6 (Cdh6) features a classical cadherin structure with five extracellular cadherin (EC) domains, a single transmembrane domain, and a highly conserved catenin interacting cytoplasmic domain (Nollet et al., 2000; Shapiro and Weis, 2009). Like all cadherins, Cdh6 proteins concentrate at adherends junctions and require calcium molecules to catalyze cell-cell adhesion events. Cdh6 expression in early nephrogenesis is coincident with Pax2 and persists in fetal mammalian kidneys (Cho et al., 1998) but is restricted in adult kidneys (Bialucha et al., 2017). However, elevated levels of Cdh6 transcripts and proteins have been detected in renal and ovarian tumor tissue sections (Bialucha et al., 2017). Deregulated Cdh6 expression is associated with the progression of renal carcinomas as well as unfavorable patient outcomes (Paul et al., 1997; Shimazui et al., 1998). In addition, Cdh6 is an emerging target gene of Pax8. Pax8 silencing in fallopian tube secretory epithelial (FT194) cells results in reduced levels of Cdh6 gene expression (de Cristofaro et al., 2016). Therefore, we investigated the expression levels of Cdh6 in human Pax2-positive-RCC after BG-1 dosing (Fig. 5C, D). By immunoblotting, we observed a significant reduction of Cdh6 protein levels in human RCC at 1.5 μM with no effect on Pax2 levels. However at 15 μM BG-1, Pax2 levels were also reduced, coincident with Cdh6. At the highest concentration, the data suggest that Pax2 protein may be destabilized in these cells, perhaps through a disruption of protein-protein or protein-chromatin interactions. These data suggest a window of specificity for Pax targets and that inhibition of Pax protein activities may be beneficial for renal cancer therapy.

BG-1 and BG-2 inhibit H3K4 methylation at Pax target genes.

The renal medulla functions to concentrate urine and retain water. Pax2 and Pax8 are both expressed in the adult renal collecting duct epithelium and are essential for urine concentration through the regulation of Aquaporins, Urea transporters (UTA1, UTA3), and Arginine Vasopressin receptor 2 (AVPR2) (Abraham et al., 2021; Laszczyk et al., 2020). Such proteins are regulated in part by increased osmolality, which in the renal medulla can range up to 1200 mOsm/kg. Mouse Inner Medullary Collecting Duct (mIMCD) cells were used to examine the effects of Pax inhibition on the expression of known targets and to monitor assembly of the histone H3K4me complex. When cultured under increased osmolality conditions simulating 500 mOsm/kg, IMCD cells show a marked increase in Pax2 and Pax8 protein expression. Despite their homology, Pax2 and Pax8 have some unique targets in IMCD cells, with Pax2 activating Avpr2 (Abraham et al., 2021) at cis-regulatory elements and Pax8 activating the Urea transporters UTA-1 and UTA-3 encoded by the Slc14a2 gene (Laszczyk et al., 2020). Thus, we utilized chromatin immunoprecipitation (ChIP) to examine Pax2 binding at known AVPR2 cis-regulatory elements in response to high salt (Fig. 6). In high salt, Pax2 protein levels increase, regardless of inhibitors, whereas PTIP protein levels are unaffected (Fig. 6A). Avpr2 expression increases approximately 4 fold but not in the presence of 5 μM BG1 or BG2 (Fig. 6B). Chip experiments show no effect of BG-1 or BG-2 on increased Pax2 binding to the Avpr2 promoter in high salt (Fig. 6C). However, BG-1 and BG-2 inhibited the recruitment of PTIP and prevented histone H3K4me3 and RNA PolII recruitment (Fig. 6 E–F). As described previously (Abraham et al., 2021), Pax8 did not bind the same AVPR2 cis-regulatory element as Pax2 (Fig. 6D), indicating some degree of specificity despite an identical DNA binding paired domain.

Figure 6. Pax2 inhibitors disrupt PTIP recruitment and H3K4 methylation at the AVPR2 promoter.

A) Westerns blots for the indicated proteins after culture in 300 mM NaCl (−) or 500 mM NaCl (+) for 24 hours, with either DMSO (cnt) or 5 μM BG-1 or BG-2 as indicated Note the high-salt induction of Pax2 protein in IMCD cells, regardless of inhibitors. B) RT-PCR measures AVPR2 expression levels in control or BG treated cells with and without high salt (*p<0.01, by one-way ANOVA and Tukey’s test). C-G) Chromatin immunoprecipitation from treated IMCD cells as indicated, using primers for Pax2 binding site at the AVPR2 promoter. Note binding of Pax2 is unaffected by inhibitors (C), whereas Pax8 does not bind well (D). BG-1 and BG-2 prevent recruitment of PTIP (E), H3K4 methylation (F), and PolII recruitment (G). All ChIP data (C-G) was from 3 experiments with error bars representing 1 SD, *p<0.01 by unpaired student t-test.

We also examined Pax8 binding to the Slc14a2 promoter, which drives expression of UTA1 and UTA3 (Fig. 7). In high salt, increased Pax8 protein are coincident with increased UTA1 and UTA3 mRNA levels and increased Pax8 binding to the promoter (Fig. 7A–D). However, 5 μM of either BG-1 or BG-2 significantly inhibited increased UTA1/3 mRNA expression in high salt with no effect on Pax8 protein levels or binding to the Slc14a2 promoter (Fig. 7A–D). As with the AVPR2 promoter, 5 μM BG-1 or BG-2 inhibit the recruitment of PTIP and the H3K4methyltransferase complex to prevent increased H3K4me3 and PolII binding (Fig. 7E–G). These data suggest a mechanism for BG-1/2 inhibition that prevents PTIP-Pax protein interactions to inhibit recruitment of MLL3/4 (KMT2C/D) histone methyltransferases. This could be due to binding of BG-1/2 to the carboxy-terminal domain of Pax proteins known to interact with PTIP (Lechner and Dressler, 1996; Lechner et al., 2000), or binding to the BRCT domains of PTIP that are thought to recognize P-Serine residues (Manke et al., 2003; Yu et al., 2003), or even inhibition of a signaling pathway that drives phosphorylation of the Pax carboxy-terminal domains. However, we could not determine any discernible effects of BG1 on the phosphorylation of Pax proteins in our cell systems (Fig. S2).

Figure 7. Pax2 inhibitors disrupt Pax8 dependent PTIP recruitment and H3K4 methylation at the Slc14a2 promoter.

A) Western blot shows 24 hour high-salt induction of Pax8 protein in IMCD cells, regardless of 5 μM BG-1 or BG-2 inhibitors. B) RT-PCR measures UTA1 expression levels in control or BG treated cells with and without high salt (*p<0.01). C) RT-PCR measures UTA3 expression levels in control or BG treated cells with and without high salt (*p<0.01). D-G) Chromatin immunoprecipitation from treated IMCD cells as indicated, using primers for Pax8 binding site at the Slc14a2 promoter. Note binding of Pax8 is unaffected by inhibitors (D), whereas BG-1 and BG-2 prevent recruitment of PTIP (E), H3K4 methylation (F), and PolII recruitment (G). For RT-PCR (B, C), *p < 0.01, by one-way ANOVA and Tukey’s test. The ChIP data (D-G) was from 3 experiments with error bars representing 1 SD, *p<0.01 by unpaired student t-test.

In a previous report, we identified another small molecule (EG1) through a virtual screen for DNA binding antagonists (Grimley et al., 2017). A fixed 10 μM amount of EG1 is able to reduce the IC₅₀ of BG1 when used in combination in our Prs4-Luc reporter cells (supplemental Fig. S3). Given that BG1 and EG1 may be working through different mechanisms, by inhibiting both protein-protein interactions and DNA-protein interactions an added benefit may be achieved.

DISCUSSION

In this report, we identified small molecules that modulate the transcriptional activity of the Pax2 family of nuclear DNA binding proteins and developmental regulators. Using unbiased HTS screening strategies, we identified a trio of triazolo pyrimidine derivatives that inhibit Pax mediated transcription activation of a reporter gene at single digit micromolar concentrations. The most active compound, BG-1, significantly inhibited proliferation of Pax2 positive cancer cells while repressing a putative target gene, Cdh6. Cancer cell lines that were Pax negative did not show significant inhibition of proliferation or viability by BG-1 at similar concentrations. Thus, there appears to be a concentration window of specific efficacy for Pax mediated gene regulation that has little impact on general transcription or translation.

Three selective Pax2 inhibitors, BG-1, BG-2, and BG-3 were identified from a set of 15 compounds by a series of screens and counter screens. Preliminary structure activity relationship (SAR) analysis shows that all three compounds are structurally very similar containing a triazolo pyrimidine core. BG-1 and BG-2 differ only in the substitutions of the phenyl ring. It is notable that the substitution at the phenyl ring is allowed at all of the three positions, ortho, meta, and para. The dual substitutions at the ortho and meta position make compounds slightly more active than the single substitution at the para position given the similar sizes of substituents, methyl and chlorine. The main difference between them (BG-1 and BG-2) and BG-3 is the linkage between the triazolo pyrimidine and the phenyl rings: piperazine for BG-1 and BG-2 and amine for BG-3, but the activities for both linkages appear comparable.

We found that BG-1 and BG-2 are from the same cluster, which is the biggest cluster with 23 compounds out of the 122 dose-response actives. The BG-3 cluster is also relatively large with 7 compounds. Besides these two clusters, there is another cluster that contain triazolo pyrimidine scaffold albeit the activities of the compounds in that cluster, pAC₅₀ were too lower, ranging 3.55 to 4.18. There were a total of 33 compounds with the triazolo pyrimidine core in the 122 dose-response actives. Now that we identified 3 compounds from the two clusters, it will be worthwhile to test other compounds in the same clusters to extend the SAR for Pax2 inhibition.

To begin to characterize the mechanisms of action of BG-1 and BG-1, we examined their ability to repress additional known target genes in mouse IMCD cells. These cells express both Pax2 and Pax8, whose levels increase rapidly in response to high osmolality. Despite their identical paired domains, the Pax2 and Pax8 proteins have some unique targets in the IMCD cells, where Pax2 binds to an activates the Avpr2 promoter (Abraham et al., 2021) and Pax8 binds to and activates the Slc14a2 promoter (Laszczyk et al., 2020). We used this feature to examine the expression of Avpr2 and UTA1/3 and the specific interactions of Pax proteins with the adaptor protein PTIP on chromatin. Both BG1 and BG2 did not affect Pax2, Pax8, or PTIP protein levels at the 5 μM concentrations used, but significantly inhibited increased Avpr2 and UTA1/3 expression in response to high salt. Although Pax protein binding to cis-regulatory elements was not affected by BG-1/2, recruitment of PTIP was attenuated by both small molecules. PTIP is an essential component of the Mll3/4 (KMT2C/D) histone H3K4 methyltransferase complex, as demonstrated in a variety of contexts, including B-cells (Daniel et al., 2010; Schwab et al., 2011), developing embryos (Kim et al., 2009; Patel et al., 2007), and adult kidneys (Soofi et al., 2020). In the absence of PTIP recruitment, no increase in H3K4me3 or RNA Pol II recruitment is observed upon increased Pax2 or Pax8 expression in mIMCD cells at either the AVPR2 or Slc14a2 locus. Thus, BG-1/2 prevent potential Pax-PTIP interactions that are necessary for recruitment of MLL3/4 and required co-factors.

Abnormal proliferation of renal epithelial cells is observed in cystic and dysplastic renal disease, polycystic kidney disease, and renal cell carcinoma, all of which continue to express high levels of Pax2 protein. Existing therapies for managing proliferative renal diseases are limited with generally poor patient outcomes. Though several drugs are currently FDA approved for the treatment of renal carcinoma, various resistance mechanisms to inhibitor drug therapy have impeded the ability to acquire effective long-term treatments with minimal side effects (Singh, 2021). Improved renal disease interventions may require broader approaches that encompass the optimization of chemical scaffolds shown to modulate more tissue and cell type specific pathways, such as the Pax2/8 proteins and their targets. The complete loss of Pax2 and Pax8 in adult mice results in severe renal abnormalities including the inability to concentrate urine due to loss of UTA1/3 and various Aquaporins (Laszczyk et al., 2020). Unlike a genetic deletion, it seems unlikely that any small molecule inhibitor would result in a complete loss of Pax function at the concentrations used. However, all Pax genes are haploinsufficient in mouse embryonic development and in adult human syndromes, suggesting that even a 50% reduction of gene dosage can have significant effects on proliferation and differentiation. Indeed, this is what has been observed in PKD as Pax2^+/− heterozygotes show reduced cyst growth and disease progression in two different PKD mouse models (Ostrom et al., 2000; Stayner et al., 2006). Thus, it may not be necessary to achieve 100% inhibition in order to significantly impact disease progression by Pax inhibitors.

In summary, the trio of BG lead compounds identified provide a scaffold for the development of additional specific Pax inhibitors that can be optimized and tested in animal models of polycystic kidney disease or in renal or reproductive cancers that express Pax2 or Pax8. The mechanism of action of BG-1/2 suggest inhibition of epigenetic modifications that are necessary for Pax target gene expression. Such inhibition may also impact non-Pax2/8 dependent mechanisms that involve PTIP or similar protein-protein interactions. Whether this inhibition is due to direct interactions of BG-1/2 with active binding sites or due to effects on Pax phosphorylation that may impact the ability to bind PTIP remains to be investigated and could lead to additional insights and applications.

SIGNIFICANCE

Developmental control genes that specify cell fates and regulate the rapid growth and differentiation during mammalian embryogenesis are often deregulated and can promote disease states in adults. In the kidney, the Pax 2 and Pax8 proteins are highly expressed in renal cancers and polycystic kidney disease. Genetic loss of function suggests that Pax proteins are necessary for continue proliferation and/or survival of cystic or cancerous renal cells. Thus, suppression of Pax protein activity may present new drugable, cell-type specific pathways for therapies. In this report, we describe a robust, cell-based high throughput screen for antagonists of Pax2 mediated transcriptional activation. After screening more than 69,000 small molecules and derivatives of initial hits, we define a family of structures that can inhibit Pax2 and Pax8 mediated transactivation in our reporter system. Analyses with cultured renal epithelial cells demonstrate specific repression of Pax2 and Pax8 targets with our lead molecules by inhibiting the recruitment of the adapter protein PTIP and the assembly of histone methylation complexes at target promoters. Furthermore, the lead compounds can slow proliferation of Pax2 positive renal cancer cells with little effect on Pax2 negative cell lines. The significance of our findings are three-fold. First, we demonstrate a proof of principle that nuclear developmental control proteins can be targeted by small molecules. Second, that such proteins may reflect cell type specific pathways for inhibiting cancer cell growth. Third, that small molecules can prevent the assembly of epigenetic regulatory complexes which may ultimately have broader applications. The structures identified in our report can serve as the basis for further development of small molecules with greater specificity and tolerance.

STAR Methods

RESOURCE AVAILABILITY

Lead contact

Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, Gregory R. Dressler (dressler@med.umich.edu).

Materials availability

The lead compounds used in this study were purchase from a commercial supplier.

Data and code availability

All data reported in this paper will be shared by the lead contact upon request.

This paper does not report original code.

Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.

EXPERIMENTAL MODEL AND SUBJECT DETAILS

HEK293s (ATCC CRL-1573) were engineered to contain an integrated PRS4-Luc (clone #33) reporter or BRE-Luc reporter construct and clonally isolated after selection. Cells were maintained in complete medium (Dulbecco’s modified Eagle’s medium (DMEM) containing 4.5 g/L D-glucose and 584 mg/L L-glutamine supplemented with 10% FBS and 1% Pen/Step). Both renal (RCC) and ovarian (SKOV-3) carcinoma cell lines were maintained in Roswell Park Memorial Institute (RPMI) 1640 medium (containing 2 g/L D-glucose, 300 mg/L L- glutamine, and 1mg/L glutathione supplemented with 10% FBS and 1% Pen/Step). The mouse IMCD cells were cultured in Dulbecco’s modified Eagle medium/nutrient mixture F-12 supplemented with 10% fetal bovine serum and 100 U/mL penicillin and 100 mg/mL streptomycin (Media from Life Technologies, Carlsbad, CA). All cell lines employed in this study were cultured in a 37°C environment containing 5% CO₂.

METHOD DETAILS

Transfection Procedure

As described previously (Grimley et al., 2017), PRS4-Luc (clone #33) reporter cells were seeded into Corning 100 mm tissue culture-treated culture dishes at 3.0 × 10⁶ and grown overnight. The following morning, PRS4-Luc cells were approximately 50% confluent and media exchange was performed with 5 ml of fresh complete media lacking 1% Pen/Strep. PRS4-Luc cells were then transiently transfected with or without mouse cDNA expression vectors (CMV-Pax2b, CMV-Pax2^C63Y), or CMV-Luc. The transfection plasmid DNA-lipid complex consisted of: 4 μg of plasmid DNA combined with 910 μL of Opti-MEM I (Gibco) for a total of 914 μL, as well as, 14 μL of Lipofectamine 2000 reagent (Life Technologies) and 900 μL of Opti-MEM I (Gibco) for a total of 914 μL. The transfection plasmid DNA-lipid complex reagent was mixed and allowed to incubate for 20 minutes at room temperature. The plasmid DNA-lipid complex was then added dropwise to the overnight cultured PRS4-Luc cells described above. The PRS4-Luc cells containing the transfection reaction were then allowed to incubate for 6 hours at 37°C and 5% CO₂. Transfection reaction plates were gently shaken at 30 mins and 60 mins post-transfection.

Luciferase Reporter Assays for HTS

Adherent PRS4-Luc reporter cells, transiently transfected with CMV-Pax2b, or control CMV-Pax2b^C63Y and CMV-Luc were dissociated from Corning 100 mm tissue culture-treated dishes and counted using the Countess II. PRS4-Luc cells were then diluted to 300 cells/μL and seeded manually or using a Multidrop Combi dispenser at either 30 × 10³ cells/well in 96-well plates or 6 × 10³ cells/well in 384-well plates, which were prepared in advance with screening molecules and DMSO. BRE-Luc cells were prepared in the same manner. The test plates were then incubated overnight in a 37°C environment containing 5% CO₂. The next day, a BioTek ELx405 microplate washer was used to lower the volume of medium and an equal volume of Steady-Glo (Promega) was added to the remaining culture medium in each well using a Multidrop Combi dispenser. Following Steady-Glo cell lysis, luminescence of either 8-point titration curves (384-wells) or 12-point titration curves (96-wells) was measured on a PHEARstar (BMG Labtech) luminometer.

Sources of drug-like molecules for HTS

The 69,125 compounds tested in the primary HTS campaign were sampled from the CCG library housed at the Life Science Institute at the University of Michigan. The CCG library is curated with diverse structures that have drug/lead-like physicochemical properties which meet Lipinski’s rules for bioavailable/orally active drugs, which possess molecular weights < 500, hydrogen-bond donors < 5, hydrogen-bond acceptors < 10, and compound hydrophilicities (cLogP) < 5 (Lipinski et al., 1997). The following CCG library collections were utilized: ChemDiv 100K (1.39% hit ratio (HR)), MS Spectrum 2000 (3.3% HR), Pilot Prestwick (2.5% HR), Pilot LOPAC (2.81% HR), Navigator Pathways (12.06% HR), Pilot NCC (1.31% HR), and Navigator Kinase (13.82% HR). Retested plates had a HR of 10%. For dose-response studies, several of the following CCG compound collections tested positive for active compounds: CB_3K (5 actives),

ChemDiv 20K (11 actives), MS Spectrum 2000 (12 actives), Navigator Kinase (2 actives), ChemDiv 100K (190 actives), Navigator Pathways (12 actives), MS2000_NCC_Focused (52 actives), Pilot Prestwick (6 actives), and Pilot LOPAC (13 actives). For confirmation dose-response studies, fresh powder samples of the dose-response actives were acquired from commercial sources and included the top 50 “hits” in this study. To aid identification of a Pax2 inhibitor series for secondary screens and SAR analysis, the CCG chemical library was re-mined based on the structures of top 50 “hits”. Over 300 scaffolds were sampled for dose-response studies from several CCG chemical collections including: ChemDiv 20K collection (65 actives), ChemDiv 100K collection (53 active compounds), NCI collection (1 actives), MB 24K collection (2 actives) and the CMLD collection (1 actives). For confirmation dose-response studies, fresh powders were sourced from Aldrich Market Select (AMS). The purity of compounds ranged from 90–92%, and they were synthesized by AMS suppliers including: Princeton BioMolecular Research (1 compound), Vitas-M Laboratory, Ltd. (7 compounds), InterBioScreen Ltd. (1 compound), and ChemDiv, Inc (6 compounds). The trio of triazolo pyrimidine derivatives (BG-1, BG-2, and BG-3) highlighted in this study were initially sampled from the ChemDiv 100K collection. Instant JChem 17.20.0 (ChemAxon, Boston, MA) was used for structure database management, search, and prediction

Luciferase ATP Assays

As previously carried out (Grimley et al., 2017), adherent RPMI 1640 (Gibco) cultured RCC or SKOV-3 cells were dissociated from tissue culture vessels and counted using the Countessä II. Diluted cells were then seeded at 500 cells/well in 96-well plates and allowed to adhere overnight during incubation. The next day, increasing concentrations of either screening compounds or DMSO were added to the test plates and incubated overnight. The following day the ELx405 (BioTek) microplate washer was used to lower the volume of culture medium and an equal volume of CellTiter-Glo (Promega) was added directly to the remaining culture medium in each well using a Multidrop Combi dispenser. To further induced cell lysis, test plates were allowed to shake on the Multidrop Combi dispenser plate carrier for 2 minutes following CellTiter-Glo^Ò addition. The PHEARstar (BMG Labtech) luminometer was used to measure luminescence, which is proportional to ATP levels (a measure of metabolically active cells).

Pharmacokinetics

The microsomal stability assay was performed by the Pharmacokinetics Core at the University of Michigan College of Pharmacy. 1μM of each compound was incubated with 200μg pooled mouse microsomes in phosphate buffer (3.3mM MgCl2) with 4mg NADPH. Samples were collected at 5,10, 15, 30, 45, and 60 minutes, the reaction quenched with cold acetonitrile, and after precipitation of protein, supernatant was used for LC/MS/MS analysis. The percentage of compound remaining was plotted over time.

For the in vivo pharmacokinetic assay, each compound at 3mg/mL in PBS containing 20% DMSO and 50% PEG-400 was administered intravenously (5mg/kg) or orally (15mg/kg). At 0.5, 2, 4, and 7 hours, blood samples were collected. After centrifugation, plasma was isolated and subjected to LC-MS/MS. A semi-logarithmic plot of the compound plasma concentration over time was plotted.

Immunoblotting

Cells were counted by using the Countessä II (Life Technologies) and seeded into 12- well plates at 5.0 × 10⁴ cells/well. Following overnight incubation and cell attachment, increasing concentrations of either screening compounds or DMSO were added to the test plates and incubated for 24 hours. The treatment medium was aspirated the following day and cells were washed once in 1X PBS. The cells were then subsequently harvested in 2 X sodium dodecyl sulfate gel-loading buffer, transferred to 1.5 mL collection tubes, and boiled for 5 minutes at 95°C.

8% or 15% sodium dodecyl sulfate-polyacrylamide gels were then utilized to resolved equal amounts of total protein from cell lysates. Electrophoretic transfer of separated proteins followed at 4°C to Immobilon-FL polyvinylidene difluoride membranes (Millipore). Following transfer, membranes were allowed to dry with subsequent reactivation in methanol. Membranes were then rinsed in 1X TBS and blocked for 1hr in 5% bovine serum albumin or 5% non-fat dry milk dissolved in 1X TBS-T (1X TBS + 0.1% Tween 20). After masking non-specific binding sites, the membranes were then incubated at 4oC overnight with primary antibodies, diluted in 1% non-fat dry milk prepared in 1X TBS-T, consisting of one of the following: anti-Rb-Cadherin-6 (Cho et al., 1998), anti-Rb-Pax2 (Dressler and Douglass, 1992), anti-Rb-b-Actin (Sigma-Aldrich; A2066), anti-Rb-phospho-Histone H3 (ser10) (Millipore; 06–570), anti-Rb-Histone H3 (Abcam; ab1791). The next day, membranes were washed in 1% non-fat dry milk prepared in 1X TBS-T and incubated for 30 mins or 1 hour at room temperature in anti-Rb-HRP diluted in 1% non-fat dry milk (1X TBS-T) or 5% non-fat dry milk (1X TBS). Membranes were washed in 1X TBS-T or 1X TBS and rinsed twice with 1X TBS followed by incubation with ECL western blotting substrate (Pierce; 32106). Finally, membranes were incubated with HyBlot ESÔ autoradiography film (Denville Scientific), developed on a film processor, and quantified using ImageJ (Version 2.0.0- rc-69/1.52p).

Real-Time Reverse Transcription PCR

RNA was prepared from cells using the TRIzol RNA Isolation system (Life Technologies). 1 μg of total RNA was reverse-transcribed into cDNA using random hexamers and the SuperScript III First Strand kit (Life Technologies) as described by the manufacturer. cDNA was diluted 20-fold and quantitative real-time PCR performed with the iTaq SYBR Green master mix (Bio-Rad) in a 7900HT Real-Time PCR system (Applied Biosystems, Foster City, CA). Each reaction was performed in triplicate; b-actin was used as an endogenous control to normalize values. Primers pairs for PCR for b-actin, Pax2, and Avpr2 are listed in Table S1. Annealing temperature was 60°C. The Avpr2 primers flank exons 2 and 3, and generate a 164 base pair fragment that was verified by DNA sequencing. Primers for UTA1 and UTA3 were as described previously (Laszczyk et al., 2020). Results were analyzed with the Applied Biosystems Prism 7900 system software. mRNA expression was calculated using the ddCT method (Schmittgen and Livak, 2008) as performed previously (Abraham et al., 2015). The normalized expression of Pax2 and Avpr2 in cells grown in 300 mOsm/kg media was set to 1.0 and expression in the other cells is relative to each respective control.

Chromatin Immunoprecipitation (ChIP) and Real-Time PCR

To examine known Pax targets in epithelial cells, mIMCD3 cells in triplicate were pretreated in DMSO or 5 mM BG1 or BG2 for 4 hours, then treated with control media or media supplemented with NaCl to raise osmolality to 500 mOsm/kg for 24 hours with or without BG1/2. Cells were crosslinked for 10 minutes by the addition of formaldehyde to a final concentration of 1%, followed by quenching in 0.125 M glycine for 5 minutes. Chromatin was prepared and ChIP was performed as described previously (Abraham et al., 2015). The precipitated DNA was reconstituted in sterile water and real-time PCR quantitation of precipitated genomic DNA relative to inputs was performed in triplicate using IQ SYBR GREEN with ROX mastermix (Bio-Rad) in a Prism 7900 Sequence Detection system (Applied Biosystems). Primer sets within the Avpr2 an Slc14a2 promoter regions were as described (Abraham et al., 2021; Laszczyk et al., 2020). All ChIP data are presented as percent input after subtraction of nonspecific binding to rabbit IgG. In all experiments, nonspecific binding was less than 0.005% of input.

QUANTIFICATION AND STATISTICAL ANALYSIS

Statistical Analysis

Analysis of variance (one-way) with post-hoc Dunnett’s multiple comparisons test and multiple t-tests (one unpaired t-test per row) were used to assess significance and calculate p-values in GraphPad Prism 8 (Version 8.4.2). For each test, alpha was set to 0.05% as the threshold for statistical significance. Statistical details for each experiment are found in the figure legends.

KEY RESOURCES TABLE.

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
Rabbit Cadherin-6	Cho et al., 1998	N/A
Rabbit Pax2	Dressler et al., 1992	N/A
Rabbit PTIP	Lechner et al., (2000)	N/A
Rabbit Pax8	Proteintech	Cat#10336-1-AP
Rabbit Beta-Actin	Sigma-Aldrich	Cat#A2066
Rabbit phospho-Histone H3 (Ser10)	Millipore	Cat#06-570
Rabbit Histone H3	Abcam	Cat#ab1791
Rabbit phospho-MAPK Family Sampler kit	Cell Signaling Technologies	Cat#9910T
Rabbit MAPK Family Sampler kit	Cell Signaling Technologies	Cat#9926
Rabbit Histone H3K4me3	Abcam	Cat#ab8580
RNA Polymerase II CTD	Abcam	Cat#ab5408
Bacterial and virus strains
Biological samples
Chemicals, peptides, and recombinant proteins
BG-1 (CCG-191845)	ChemDiv/Molport	PubChem ID 124738812
BG-2 (CCG-130202)	ChemDiv/Molport	124682279
BG-3 (CCG-191087)	ChemDiv/Molport	124738112
Critical commercial assays
Steady-Glo	Promega	Cat#E2520
CellTiter-Glo	Promega	Cat#G7572
iTaq SYBR Green master mix	Bio-Rad	Cat#1725121
Deposited data
Experimental models: Cell lines
PRS4-Luc (HEK293)	Patel, et al. 2012	N/A
BRE-Luc (HEK293)	Bradford, et al. 2019	N/A
Human Renal Carcinoma	Gnarra and Dressler 1995	N/A
Human Ovarian Carcinoma SKOV3	ATCC	HTB-77
Mouse IMCD3	ATCC	CRL-2123
Experimental models: Organisms/strains
Oligonucleotides
See Supplemental Table S1
Recombinant DNA
CMV-Pax2b mouse	Grimley et al., 2018	N/A
CMV-Pax2C63Y mouse	Grimley et al., 2018	N/A
CMV-Luc	Grimley et al., 2018	N/A
Software and algorithms
Instant JChem 17.20.0	ChemAxon, Boston, MA	https://chemaxon.com/products/instant-jchem
Prism 8.4.2	GraphPad Software, San Diego, CA	https://www.graphpad.com/scientific-software/prism/
MScreen	(Jacob et al., 2012)	https://mscreen.lsi.umich.edu/
ImageJ 2.0.0	(Schindelin et al., 2012)	https://imagej.net/software/fiji/
Other

Highlights.

A cell-based reporter is designed to facilitate screening for Pax2 inhibitors.

The inhibitors could suppress cancer cell proliferation and known Pax target genes.

The inhibitors block assembly of histone methylation complexes at Pax2 binding sites.