Structural basis for ion selectivity in potassium-selective channelrhodopsins.

KCR channelrhodopsins (K+-selective light-gated ion channels) have received attention as potential inhibitory optogenetic tools but more broadly pose a fundamental mystery regarding how their K+ selectivity is achieved. Here, we present 2.5-2.7 Å cryo-electron microscopy structures of HcKCR1 and HcKCR2 and of a structure-guided mutant with enhanced K+ selectivity. Structural, electrophysiological, computational, spectroscopic, and biochemical analyses reveal a distinctive mechanism for K+ selectivity; rather than forming the symmetrical filter of canonical K+ channels achieving both selectivity and dehydration, instead, three extracellular-vestibule residues within each monomer form a flexible asymmetric selectivity gate, while a distinct dehydration pathway extends intracellularly. Structural comparisons reveal a retinal-binding pocket that induces retinal rotation (accounting for HcKCR1/HcKCR2 spectral differences), and design of corresponding KCR variants with increased K+ selectivity (KALI-1/KALI-2) provides key advantages for optogenetic inhibition in vitro and in vivo. Thus, discovery of a mechanism for ion-channel K+ selectivity also provides a framework for next-generation optogenetics.

Photoisomerization pathway of the microbial rhodopsin chromophore in solution.

Photoisomerization is a key photochemical reaction in microbial and animal rhodopsins. It is well established that such photoisomerization is highly selective; all-trans to 13-cis, and 11-cis to all-trans forms in microbial and animal rhodopsins, respectively. Nevertheless, unusual photoisomerization pathways have been discovered recently in microbial rhodopsins. In an enzymerhodopsin NeoR, the all-trans chromophore is isomerized into the 7-cis form exclusively, which is stable at room temperature. Although, the 7-cis form is produced by illumination of retinal, formation of the 7-cis form was never reported for a protonated Schiff base of all-trans retinal in solution. Present HPLC analysis of retinal oximes prepared by hydroxylamine reaction revealed that all-trans and 7-cis forms cannot be separated from the syn peaks under the standard HPLC conditions, while it is possible by the analysis of the anti-peaks. Consequently, we found formation of the 7-cis form by the photoreaction of all-trans chromophore in solution, regardless of the protonation state of the Schiff base. Upon light absorption of all-trans protonated retinal Schiff base in solution, excited-state relaxation accompanies double-bond isomerization, producing 7-cis, 9-cis, 11-cis, or 13-cis form. In contrast, specific chromophore-protein interaction enforces selective isomerization into the 13-cis form in many microbial rhodopsins, but into 7-cis in NeoR.

A Novel Color Switch of Microbial Rhodopsin.

Function of animal and microbial rhodopsins starts by light absorption of the retinal chromophore. The absorption maximum wavelength (λmax) of rhodopsins is determined by the energy gap between the electronically ground (S0) and first excited (S1) state of the retinal chromophore, and the color tuning mechanism is one of the central topics in rhodopsin research. "Color switches", color-determining residues, are red- and blue-shifting amino acids at the same position in two rhodopsins, whose exchange causes spectral blue- and red-shifts, respectively, in each rhodopsin. As mutation easily destroys elaborate chromophore-protein interactions, the known color switches in microbial rhodopsins are limited; the L/Q switch in C-helix (TM3), the A/TS switch in G-helix (TM7), and the G/P switch in F-helix (TM6). Here, we report a novel color switch of microbial rhodopsins, which is located in D-helix (TM4). In this color switch, the red- and blue-shifting amino acids are Asn (N) and Leu (L)/Ile (I), respectively. As Asn and Leu/Ile are polar and nonpolar amino acids, respectively, and the position is located near the ß-ionone ring, the N/LI switch matches the general rule of color tuning by polarity. The N/LI switch is also useful for optogenetics, as many ion-transporting rhodopsins contain blue-shifting amino acids, such as L and I, at that position.

Functional analysis of LAT3 in prostate cancer: Its downstream target and relationship with androgen receptor.

L-type amino acid transporter 3 (LAT3, SLC43A1) is abundantly expressed in prostate cancer (PC) and is thought to play an essential role in PC progression through the cellular uptake of essential amino acids. Here, we analyzed the expression, function, and downstream target of LAT3 in PC. LAT3 was highly expressed in PC cells expressing androgen receptor (AR), and its expression was increased by dihydrotestosterone treatment and decreased by bicalutamide treatment. In chromatin immunoprecipitation sequencing of AR, binding of AR to the SLC43A1 region was increased by dihydrotestosterone stimulation. Knockdown of LAT3 inhibited cell proliferation, migration, and invasion, and the phosphorylation of p70S6K and 4EBP-1. Separase (ESPL1) was identified as a downstream target of LAT3 by RNA sequencing analysis. In addition, immunostaining of prostatectomy specimens was performed. In the multivariate analysis, high expression of LAT3 was an independent prognostic factor for recurrence-free survival (hazard ratio: 3.24; P = .0018). High LAT3 expression was correlated with the pathological T stage and a high International Society of Urological Pathology grade. In summary, our results suggest that LAT3 plays an important role in the progression of PC.

Development of an In Vitro Chloroplast Splicing System: Sequences Required for Correct pre-mRNA Splicing.

Chloroplast genomes in land plants include approximately 20 intron-containing genes. Most of the introns are similar to the group II introns found in fungi, algae and some bacteria, but no self-splicing has been reported. To analyze splicing reactions in chloroplasts, we developed a tobacco (Nicotiana tabacum) chloroplast-based in vitro system. We optimized the splicing reaction using atpF precursor messenger RNA (pre-mRNA). Our system requires a high ATP concentration, whereas ATP is not necessary for self-splicing group II introns. Self-splicing group II introns possess two exon-binding sites (EBS1 and 2) complementary to two intron-binding sites (IBS1 and 2) in the 3' end of 5' exons, which are involved in 5' splice-site selection. Using our in vitro system and atpF pre-mRNA, we analyzed short sequences corresponding to the above EBSs and IBSs. Mutation analyses revealed that EBS1-IBS1 pairing is essential, while EBS2-IBS2 pairing is important but not crucial for splicing. The first 3' exon nucleotide determines the 3' splice sites of self-splicing introns. However, mutations to this nucleotide in atpF pre-mRNA did not affect splicing. This result suggests that the mechanism underlying chloroplast pre-mRNA splicing differs partly from that mediating the self-splicing of group II introns.

Molecular Properties and Optogenetic Applications of Enzymerhodopsins.

The cyclic nucleotides cAMP and cGMP are ubiquitous secondary messengers that regulate multiple biological functions including gene expression, differentiation, proliferation, and cell survival. In sensory neurons, cyclic nucleotides are responsible for signal modulation, amplification, and encoding. For spatial and temporal manipulation of cyclic nucleotide dynamics, optogenetics have a great advantage over pharmacological approaches. Enzymerhodopsins are a unique family of microbial rhodopsins. These molecules are made up of a membrane-embedded rhodopsin domain, which binds an all trans-retinal to form a chromophore, and a cytoplasmic water-soluble catalytic domain. To date, three kinds of molecules have been identified from lower eukaryotes such as fungi, algae, and flagellates. Among these, histidine kinase rhodopsin (HKR) is a light-inhibited guanylyl cyclase. Rhodopsin GC (Rh-GC) functions as a light-activated guanylyl cyclase, while rhodopsin PDE (Rh-PDE) functions as a light-activated phosphodiesterase that degrades cAMP and cGMP. These enzymerhodopsins have great potential in optogenetic applications for manipulating the intracellular cyclic nucleotide dynamics of living cells. Here we introduce the molecular function and applicability of these molecules.

Epigenetic modifications in prostate cancer.

Prostate cancer is a major cause of cancer-related deaths among men worldwide. In addition to genomic alterations, epigenetic alterations accumulated in prostate cancer have been elucidated. While aberrant deoxyribonucleic acid hypermethylation in promoter CpG islands inactivates crucial genes associated with deoxyribonucleic acid repair, cell cycle, apoptosis or cell adhesion, aberrant deoxyribonucleic acid hypomethylation can lead to oncogene activation. Acetylation of histone is also deregulated in prostate cancer, which could cause aberrant super-enhancer formation and activation of genes associated with cancer development. Deregulations of histone methylation, such as an increase of trimethylation at position 27 of histone H3 by enhancer of zeste homolog2 overexpression, or other modifications, such as phosphorylation and ubiquitination, are also involved in prostate cancer development, and inhibitors targeting these epigenomic aberrations might be novel therapeutic strategies. In this review, we provide an overview of epigenetic alterations in the development and progression of prostate cancer, focusing on deoxyribonucleic acid methylation and histone modifications.

Adrenalectomy in Japanese patients with subclinical Cushing syndrome: 1-mg dexamethasone suppression test to predict the surgical benefit.

OBJECTIVES: To investigate whether the result of the 1-mg dexamethasone suppression test can predict the improvement of comorbidities after adrenalectomy in patients with subclinical Cushing syndrome. METHODS: This retrospective study included 117 subclinical Cushing syndrome patients who underwent adrenalectomy. The numbers of prescribed drugs for metabolic comorbidities and the clinical variables at diagnosis were compared with those at the follow up. Patients were classified into subgroups according to the result of the 1-mg dexamethasone suppression test. RESULTS: Significant improvements in blood pressure, serum cholesterol and body mass index were observed. Furthermore, a significant improvement in glycated hemoglobin was observed in patients with diabetes mellitus. These improvements led to a discontinuation or reduction of prescribed drugs after surgery. In addition, the greatest reduction of prescribed drugs was observed in patients whose serum cortisol levels were between 1.8 and 3.0 µg/dL after the 1-mg dexamethasone suppression test. CONCLUSIONS: The result of the 1-mg dexamethasone suppression test can be a useful factor predicting the improvement of comorbidities after adrenalectomy. Current data might give us a new insight into the decision-making for the treatment of subclinical Cushing syndrome.

Contralateral adrenal width predicts the duration of prolonged post-surgical steroid replacement for subclinical Cushing syndrome.

OBJECTIVES: To identify pre-treatment factors affecting the duration of post-surgical steroid replacement in patients undergoing adrenalectomy for subclinical Cushing syndrome. METHODS: The present retrospective analysis included 64 patients who underwent unilateral laparoscopic adrenalectomy for subclinical Cushing syndrome. Adrenal tumor and contralateral adrenal sizes together with various clinical factors were studied in association with the duration of post-surgical steroid replacement. Adrenal tumor and contralateral adrenal size were measured at the level of the maximum transverse plane of the adrenal glands using computed tomography scan or magnetic resonance imaging. Cox's proportional hazards model was used for the statistical analysis. RESULTS: All 64 patients were treated with post-surgical steroid replacement after adrenalectomy. The median duration of the steroid treatment was 6 months. When assessing the duration of post-surgical steroid replacement, contralateral adrenal volume <0.745 cm3 , contralateral adrenal width <6.15 mm and serum cortisol after a 1-mg dexamethasone suppression test >2.65 µg/dL were significant predictors of prolonged post-surgical steroid treatment on univariate analysis. On multivariate analysis, contralateral adrenal width <6.15 mm was the only independent predictive factor for the prolonged post-surgical steroid replacement. CONCLUSIONS: Contralateral adrenal width seems to represent a significant predictive factor for the duration of post-surgical steroid replacement in subclinical Cushing syndrome patients. Pre-surgical assessment of image findings might help clinicians determine the total duration of steroid therapy after adrenalectomy.

Cooperation between the chloroplast psbA 5'-untranslated region and coding region is important for translational initiation: the chloroplast translation machinery cannot read a human viral gene coding region.

Chloroplast mRNA translation is regulated by the 5'-untranslated region (5'-UTR). Chloroplast 5'-UTRs also support translation of the coding regions of heterologous genes. Using an in vitro translation system from tobacco chloroplasts, we detected no translation from a human immunodeficiency virus tat coding region fused directly to the tobacco chloroplast psbA 5'-UTR. This lack of apparent translation could have been due to rapid degradation of mRNA templates or synthesized protein products. Replacing the psbA 5'-UTR with the E. coli phage T7 gene 10 5'-UTR, a highly active 5'-UTR, and substituting synonymous codons led to some translation of the tat coding region. The Tat protein thus synthesized was stable during translation reactions. No significant degradation of the added tat mRNAs was observed after translation reactions. These results excluded the above two possibilities and confirmed that the tat coding region prevented its own translation. The tat coding region was then fused to the psbA 5'-UTR with a cognate 5'-coding segment. Significant translation was detected from the tat coding region when fused after 10 or more codons. That is, translation could be initiated from the tat coding region once translation had started, indicating that the tat coding region inhibits translational initiation but not elongation. Hence, cooperation/compatibility between the 5'-UTR and its coding region is important for translational initiation.

Arabidopsis Pol II-Dependent in Vitro Transcription System Reveals Role of Chromatin for Light-Inducible rbcS Gene Transcription.

In vitro transcription is an essential tool to study the molecular mechanisms of transcription. For over a decade, we have developed an in vitro transcription system from tobacco (Nicotiana tabacum)-cultured cells (BY-2), and this system supported the basic activities of the three RNA polymerases (Pol I, Pol II, and Pol III). However, it was not suitable to study photosynthetic genes, because BY-2 cells have lost their photosynthetic activity. Therefore, Arabidopsis (Arabidopsis thaliana) in vitro transcription systems were developed from green and etiolated suspension cells. Sufficient in vitro Pol II activity was detected after the minor modification of the nuclear soluble extracts preparation method; removal of vacuoles from protoplasts and L-ascorbic acid supplementation in the extraction buffer were particularly effective. Surprisingly, all four Arabidopsis Rubisco small subunit (rbcS-1A, rbcS-1B, rbcS-2B, and rbcS-3B) gene members were in vitro transcribed from the naked DNA templates without any light-dependent manner. However, clear light-inducible transcriptions were observed using chromatin template of rbcS-1A gene, which was prepared with a human nucleosome assembly protein 1 (hNAP1) and HeLa histones. This suggested that a key determinant of light-dependency through the rbcS gene transcription was a higher order of DNA structure (i.e. chromatin).

Treatment of locally advanced prostate cancer (Stage T3).

OBJECTIVE: Formerly, locally advanced prostate cancer exhibited poorly prognosis. In the late 1990s, new surgical and radiation technologies were introduced in combination with androgen deprivation. To evaluate respective strategies, outcomes were examined. PATIENTS AND METHODS: Between 2001 and 2010, 224 patients with T3N0M0 (10.9% of all prostate cancer cases) were treated with prostatectomy, external beam radiation therapy with/without androgen deprivation or hormone alone. Complete records were obtained by the end of 2015. RESULTS: Operation group first started without adjuvant treatment and prostate-specific antigen (PSA) relapse occurred in 39% of cases. Radiation therapy group was alternatively divided into two subgroups, that received either monotherapy or combination with androgen deprivation, and PSA relapse rates were 65 and 16%, respectively. High rates of PSA relapse in both the operation and radiation therapy groups were observed in patients without adjuvant therapy, but after relapse androgen deprivation proceeded favorable outcomes. In the radiation subgroups, PSA relapse rates were different, but both subsequent survival rates were the same. This may be due to the effect of androgen deprivation after relapse, indicating effect of delayed therapy. PSA relapse rate in the hormone therapy group was 25% and after relapse, patients applied to treatment with other hormonal and anticancer drugs. Overall survival rates were 91, 88 and 67% in the operation, radiation therapy and hormone therapy groups, respectively. CONCLUSION: Aggressive treatment with short-term androgen deprivation for locally advanced prostate cancer could be beneficial and not harmful when suitable candidates are selected. Delayed androgen deprivation was effective for no adjuvant patients after PSA relapse.

Additional pathway to translate the downstream ndhK cistron in partially overlapping ndhC-ndhK mRNAs in chloroplasts.

The chloroplast NAD(P)H dehydrogenase (NDH) C (ndhC) and ndhK genes partially overlap and are cotranscribed in many plants. We previously reported that the tobacco ndhC/K genes are translationally coupled but produce NdhC and NdhK, subunits of the NDH complex, in similar amounts. Generally, translation of the downstream cistron in overlapping mRNAs is very low. Hence, these findings suggested that the ndhK cistron is translated not only from the ndhC 5'UTR but also by an additional pathway. Using an in vitro translation system from tobacco chloroplasts, we report here that free ribosomes enter, with formylmethionyl-tRNA(fMet), at an internal AUG start codon that is located in frame in the middle of the upstream ndhC cistron, translate the 3' half of the ndhC cistron, reach the ndhK start codon, and that, at that point, some ribosomes resume ndhK translation. We detected a peptide corresponding to a 57-amino-acid product encoded by the sequence from the internal AUG to the ndhC stop codon. We propose a model in which the internal initiation site AUG is not designed for synthesizing a functional isoform but for delivering additional ribosomes to the ndhK cistron to produce NdhK in the amount required for the assembly of the NDH complex. This pathway is a unique type of translation to produce protein in the needed amount with the cost of peptide synthesis.

Processing of the 5'-UTR and existence of protein factors that regulate translation of tobacco chloroplast psbN mRNA.

The chloroplast psbB operon includes five genes encoding photosystem II and cytochrome b 6 /f complex components. The psbN gene is located on the opposite strand. PsbN is localized in the thylakoid and is present even in the dark, although its level increases upon illumination and then decreases. However, the translation mechanism of the psbN mRNA remains unclear. Using an in vitro translation system from tobacco chloroplasts and a green fluorescent protein as a reporter protein, we show that translation occurs from a tobacco primary psbN 5'-UTR of 47 nucleotides (nt). Unlike many other chloroplast 5'-UTRs, the psbN 5'-UTR has two processing sites, at -39 and -24 upstream from the initiation site. Processing at -39 enhanced the translation rate fivefold. In contrast, processing at -24 did not affect the translation rate. These observations suggest that the two distinct processing events regulate, at least in part, the level of PsbN during development. The psbN 5'-UTR has no Shine-Dalgarno (SD)-like sequence. In vitro translation assays with excess amounts of the psbN 5'-UTR or with deleted psbN 5'-UTR sequences demonstrated that protein factors are required for translation and that their binding site is an 18 nt sequence in the 5'-UTR. Mobility shift assays using 10 other chloroplast 5'-UTRs suggested that common or similar proteins are involved in translation of a set of mRNAs lacking SD-like sequences.

Translation of partially overlapping psbD-psbC mRNAs in chloroplasts: the role of 5'-processing and translational coupling.

The chloroplast psbD and psbC genes encode the D2 and CP43 proteins of the photosystem II complex, and they are generally cotranscribed. We report studies on the basic translation process of tobacco psbD-psbC mRNAs using an in vitro translation system from tobacco chloroplasts. The primary transcript has an unusually long 5'-UTR (905 nt). We show that it is translatable. Processing of the 5'-UTR greatly enhances the translation efficiency of the psbD cistron. A striking feature is that psbD and psbC cistrons overlap by 14 nt. Removal of the psbD 5'-UTR plus the start codon and introduction of a premature termination codon in the psbD cistron considerably reduce the translation efficiency of the downstream psbC cistron. These results indicate that translation of the psbC cistron depends largely on that of the upstream psbD cistron and thus shows translational coupling; however, a portion is independently translated. These observations, together with the presence of monocistronic psbC mRNAs, suggest that the psbD and psbC cistrons are translated via multiple processes to produce necessary amounts of D2 and CP43 proteins.

[Clinical outcomes after combined therapy with dutasteride in patients with unsuccessful trial without catheter after treatment with an alpha1-adrenergic receptor blocker monotherapy for acute urinary retention caused by prostatic hyperplasia].

OBJECTIVE: The outcome of trial of voiding without catheter in patients treated combination therapy with dutasteride and alpha1-adrenergic receptor blocker for acute urinary retention caused by benign prostatic hyperplasia was not reported. We evaluated the clinical efficacy of combination therapy with dutasteride in patients with unsuccessful trial without catheter after treatment with an alpha1-adrenergic receptor blocker monotherapy for acute urinary retention caused by benign prostatic hyperplasia. PATIENTS AND METHODS: Patients with acute urinary retention due to prostatic hyperplasia were catheterized and treated alpha1-adrenergic receptor blocker monotherapy. After two weeks later, patients were put on trial without catheter. 52 patients who were unsuccessful trial without catheter administered combination therapy with dutasteride and alpha1-adrenergic receptor blocker. We use criteria that voiding urine volume over 100 ml and post-void residual urine volume below 100 ml in deciding whether catheter should be removed. RESULTS: 33 (63.5%) men did not require re-catheterization within 7 months after combination therapy. The successful rate of Performance Status (PS) 0-1 group was significantly superior to that of PS 2-4 group. CONCLUSIONS: PS 0-1 men catheterized for AUR can void more successfully after catheter removal than PS 2-4 men if treated with combination therapy.

SNap Bond, a Crucial Hydrogen Bond Between Ser in Helix 3 and Asn in Helix 4, Regulates the Structural Dynamics of Heliorhodopsin.

Heliorhodopsin (HeR) is a new rhodopsin family discovered in 2018 through functional metagenomic analysis. Similar to microbial rhodopsins, HeR has an all-trans retinal chromophore, and its photoisomerization to the 13-cis form triggers a relatively slow photocycle with sequential intermediate states (K, M, and O intermediates). The O intermediate has a relatively long lifetime and is a putative active state for transferring signals or regulating enzymatic reactions. Although the first discovered HeR, 48C12, was found in bacteria and the second HeR (TaHeR) was found in archaea, their key amino acid residues and molecular architectures have been recognized to be well conserved. Nevertheless, the rise and decay kinetics of the O intermediate are faster in 48C12 than in TaHeR. Here, using a new infrared spectroscopic technique with quantum cascade lasers, we clarified that the hydrogen bond between transmembrane helices (TM) 3 and 4 is essential for the altered O kinetics (Ser112 and Asn138 in 48C12). Interconverting mutants of 48C12 and TaHeR clearly revealed that the hydrogen bond is important for regulating the dynamics of the O intermediate. Overall, our study sheds light on the importance of the hydrogen bond between TM3 and TM4 in heliorhodopsins, similar to the DC gate in channelrhodopsins.

The downstream atpE cistron is efficiently translated via its own cis-element in partially overlapping atpB-atpE dicistronic mRNAs in chloroplasts.

The chloroplast atpB and atpE genes encode subunits ß and ε of the ATP synthase, respectively. They are co-transcribed as dicistronic mRNAs in flowering plants. An unusual feature is an overlap (AUGA) of the atpB stop codon (UGA) with the atpE start codon (AUG). Hence, atpE translation has been believed to depend on atpB translation (i.e. translational coupling). Using an in vitro translation system from tobacco chloroplasts, we showed that both atpB and atpE cistrons are translated from the tobacco dicistronic mRNA, and that the efficiency of atpB translation is higher than that of atpE translation. When the atpB 5'-UTR was replaced with lower efficiency 5'-UTRs, atpE translation was higher than atpB translation. Removal of the entire atpB 5'-UTR arrested atpB translation but atpE translation still proceeded. Introduction of a premature stop codon in the atpB cistron did not abolish atpE translation. These results indicate that atpE translation is independent of atpB translation. Mutation analysis showed that the atpE cistron possesses its own cis-element(s) for translation, located ~25 nt upstream from the start codon.

An organellar maturase associates with multiple group II introns.

Bacterial group II introns encode maturase proteins required for splicing. In organelles of photosynthetic land plants, most of the group II introns have lost the reading frames for maturases. Here, we show that the plastidial maturase MatK not only interacts with its encoding intron within trnK-UUU, but also with six additional group II introns, all belonging to intron subclass IIA. Mapping analyses of RNA binding sites revealed MatK to recognize multiple regions within the trnK intron. Organellar group II introns are considered to be the ancestors of nuclear spliceosomal introns. That MatK associates with multiple intron ligands makes it an attractive model for an early trans-acting nuclear splicing activity.

PBRM1-deficient PBAF complexes target aberrant genomic loci to activate the NF-κB pathway in clear cell renal cell carcinoma.

PBRM1 encodes an accessory subunit of the PBAF SWI/SNF chromatin remodeller, and the inactivation of PBRM1 is a frequent event in kidney cancer. However, the impact of PBRM1 loss on chromatin remodelling is not well examined. Here we show that, in VHL-deficient renal tumours, PBRM1 deficiency results in ectopic PBAF complexes that localize to de novo genomic loci, activating the pro-tumourigenic NF-κB pathway. PBRM1-deficient PBAF complexes retain the association between SMARCA4 and ARID2, but have loosely tethered BRD7. The PBAF complexes redistribute from promoter proximal regions to distal enhancers containing NF-κB motifs, heightening NF-κB activity in PBRM1-deficient models and clinical samples. The ATPase function of SMARCA4 maintains chromatin occupancy of pre-existing and newly acquired RELA specific to PBRM1 loss, activating downstream target gene expression. Proteasome inhibitor bortezomib abrogates RELA occupancy, suppresses NF-κB activation and delays growth of PBRM1-deficient tumours. In conclusion, PBRM1 safeguards the chromatin by repressing aberrant liberation of pro-tumourigenic NF-κB target genes by residual PBRM1-deficient PBAF complexes.