Research ArticleCANCER

Growth of pancreatic cancers with hemizygous chromosomal 17p loss of MYBBP1A can be preferentially targeted by PARP inhibitors

See allHide authors and affiliations

Science Advances  04 Dec 2020:
Vol. 6, no. 49, eabc4517
DOI: 10.1126/sciadv.abc4517

Abstract

Here, we selectively target pancreatic ductal adenocarcinoma (PDAC) cells harboring a hemizygous gene essential for cell growth. MYB binding protein 1A (MYBBP1A), encoding a chromatin-bound protein, is hemizygous in most of the PDAC due to a chromosome 17p deletion that also spans TP53. We find that hemizygous MYBBP1A loss in isogenic PDAC cells promotes tumorigenesis but, paradoxically, homozygous MYBBP1A loss is associated with impaired cell growth and decreased tumorigenesis. Poly–adenosine 5′-diphosphate–ribose polymerase 1 (PARP1) interacts with MYBBP1A and displaces it from chromatin. Small molecules, such as olaparib, that trap PARP1 to chromatin are able to evict the minimal pool of chromatin-bound MYBBP1A protein in MYBBP1A hemizygous cells and impair cell growth, greater than its impact on wild-type cells. Our findings reveal how a cell essential gene with one allele lost in cancer cells can be preferentially susceptible to a specific molecular therapy, when compared to wild-type cells.

INTRODUCTION

Pancreatic ductal adenocarcinoma (PDAC) is the most common histologic type of pancreatic cancer, and its dismal outcome can be attributed to the presence of advanced or metastatic disease at the time of diagnosis, high frequency of local and distant relapse, and resistance to conventional chemotherapy and radiotherapy (1). Genomic characterization of sporadic PDAC has elucidated recurrent somatic mutations in KRAS, TP53, and CDKN2A that are collectively found in approximately 95% of PDACs, as well as less frequent mutations in SMAD4, BRCA2, ARID1A, and other genes (2). Few of these genes are considered actionable at this time. Consequently, the standard management of PDAC remains limited to surgery and chemotherapy and, in restricted circumstances, with immunotherapy, although a limited number of patients are cured or have enduring responses to treatment (3). In this study, we expand the spectrum of actionable PDAC targets by finding that a cell essential gene with one allele lost due to hemizygosity in cancer cells, Myb binding protein 1A (MYBBP1A), can be preferentially susceptible to a specific molecular therapy when compared to wild-type (WT) cells.

PDAC is associated with frequent chromosomal abnormalities, including deletion of chromosome 17p deletion [del(17p)] in up to 64% of cases (4). The chromosomal region associated with del(17p) nearly always encompasses the TP53 gene in PDAC, and the pathologic contribution of this cytogenetic feature is assumed to be due to the heterozygous loss of TP53 (5). However, recent evidence suggests that tumorigenesis associated with del(17p) has a contribution of the loss of other genes on 17p, in addition to TP53, because heterozygous deletion of mouse chromosome 11B3, syntenic to human 17p13.1, produces a greater effect on lymphoma and leukemia development than does Trp53 deletion alone (6). While cancer target discovery is largely aimed at identifying genetically dominant, oncogenic driver events, loss of genetic material including hemizygous deletion of a cell essential gene may represent a potential therapeutic target. That is, monoallelic loss of a critical gene may not impair cell fitness, as gene expression is only partially reduced (7). However, further perturbations in the expression or function of a critical gene product, e.g., by therapeutic intervention, may result in impaired growth or the death of cancer cells but not nonmalignant cells that have preserved two alleles. The rationale is supported by the unprecedented effectiveness of the anti-casein kinase 1 alpha 1 (CSNK1A1) drug lenalidomide in myelodysplastic syndrome with chromosome 5q deletion (8). Loss of chromosome 5q results in a 50% reduction in the expression of the haploinsufficient tumor suppressor gene CSNK1A1, which is critical to cell viability (9). Lenalidomide induces CSNK1A1 degradation at nonlethal doses for karyotypically intact cells but causes apoptosis of myeloid cells with 5q deletion (10). On the basis of the fact that haploinsufficient tumor suppressors are usually associated with one remaining intact allele, we reasoned that putative tumor suppressor genes on chromosome 17p may represent a synthetic vulnerability in PDACs harboring del(17p).

Poly–adenosine 5′-diphosphate (ADP)–ribose polymerase 1 (PARP1) is a nuclear enzyme responsible for the formation of poly-ADP-ribose (PAR) and is implicated in multiple cellular processes, including DNA repair, DNA replication, DNA methylation, chromatin compaction, and histone modification (11). While PARP1 catalytic activity is known to use nicotinamide adenine dinucleotide (NAD+) to generate PAR moieties, under NAD+-depleted conditions, PARP1 can directly compact nucleosomes resulting in denser chromatin structure (12). Given its essential role in multiple DNA repair pathways, the PARP1 inhibitors (PARPis) have been used as a method to selectively target cancer cells harboring mutations in BRCA1/2 and other genes involved in the homologous repair pathway for DNA damage (13). Recently, results from a large phase 3 trial demonstrated the efficacy of PARPi in PDAC harboring mutations in BRCA1/2 genes (14). However, BRCA1/2 mutations are an imperfect predictor of PARPi activity, as not all BRCA1/2 mutant cancers respond to PARPi and PARPi has demonstrated efficacy in BRCA1/2 WT cancers (15, 16). Thus, the precise mechanism of PARPi activity in cancer remains to be defined.

In our current study, we find that MYBBP1A is a critical tumor suppressor gene on chromosome 17p and provide evidence that hemizygous MYBBP1A loss in PDAC confers PARPi sensitivity. Our study supports the use of PARPi in the treatment of PDAC with or without BRCA1/2 mutations.

RESULTS

In silico analysis of functional genomic data as a basis to identify potential therapeutic targets in PDAC

To identify potential candidate genes to target therapeutically in PDAC, we focused on genes located on chromosome 17p, since deletions in this region are frequent (17). Genes critical for cell viability were determined using the Project Achilles dataset, a genome-wide CRISPR-Cas9 screen of 517 cancer cell lines (18). We also narrowed our genes of interest to putative tumor suppressor genes, since these genes would be predicted to be selectively hemizygous in PDAC cells but not in WT cells. Eight genes were at the intersection of genes located on 17p found from Project Achilles and whose expression is associated with longer overall survival in The Cancer Genome Atlas (TCGA) PDAC dataset (https://cancer.gov/tcga): MYBBP1A, METTL16, MIS12, ELP5, PELP1, RPAIN, TIM22, and TOP3A (Fig. 1A). Of these genes, MYBBP1A was selected for further evaluation as it has been implicated in carcinogenesis, although separate loss of function studies with variable means of MYBBP1A repression suggested either oncogenic or tumor-suppressive activities (1923).

Fig. 1 MYBBP1A is a cell essential gene located on 17p.

(A) Venn diagram of genes located on 17p, essential genes, and genes with positive correlation with survival in patients with PDAC. Top right: Correlation of MYBBP1A expression and survival in patients with PDAC. Bottom right: Shared genes. (B) Immunoblot of MYBBP1A and glyceraldehyde-3-phosphate dehydrogenase (GAPDH) levels after sgRNA transduction. (C) Cell growth analysis after sgControl or sgMYBBP1A transduction (means ± SEM, Student’s t test). *P < 0.01. (D) Immunoblot of MYBBP1A, GAPDH, and green fluorescent protein (GFP) after indicated single-guide RNA (sgRNA) and sgRNA-resistant MYBBP1A expression construct or GFP. Arrow denotes MYBBP1A-specific band. Asterisk denotes nonspecific band. (E) Cell growth analysis after indicated sgRNA and sgRNA-resistant MYBBP1A expression construct or GFP (means ± SEM, Student’s t test). *P < 0.05. (F) Immunoblot of cells with shControl of shMYBBP1A expression. Arrow denotes MYBBP1A-specific band. Asterisk denotes nonspecific band. (G) Cell growth analysis with shControl of shMYBBP1A expression (means ± SEM, Student’s t test). *P < 0.005. (H) Images of tumor at 30 days after transplantation. Tumor volume (n = 9 tumors for each group) was measured with calipers. (I) Tumor weights 30 days after transplantation. *P < 0.005 as determined by Student’s t test. Photo credit: A. Hsieh, University of Pennsylvania.

MYBBP1A is essential for cell growth and tumorigenesis

To test whether MYBBP1A is critical in PDAC cells, we induced insertions or deletions at the MYBBP1A gene using CRISPR-Cas9 technology, using lentivirus transduction of two single-guide RNAs (sgRNAs) targeting exons 14 and 16 of MYBBP1A (sgMYBBP1A#1 and sgMYBBP1A#2). An sgRNA targeting the ROSA allele (sgControl) was used as a control. MYBBP1A knockout (KO) was confirmed by immunoblotting showing loss of MYBBP1A protein in HS766T cells, a PDAC cell line that originally harbored two MYBBP1A alleles (Fig. 1B) (24). Cell growth over 15 days from transduction in culture was reduced threefold in cells transduced with sgRNAs targeting MYBBP1A, compared to nontargeting sgRNAs (Fig. 1C). The effect on cell growth is comparable to many other reported studies that have examined cell essential genes (25). Examination with immunofluorescence and immunoblotting of MYBBP1A KO cells over an extended time course showed that cells persisting after weeks in cell culture, after CRISPR mutagenesis, express MYBBP1A, likely due to selective outgrowth of escaper cells (fig. S1, A to C). Consistent with this, sequencing of individual clones that underwent 3 weeks of selection after transduction with sgMYBBP1A demonstrated no evidence of CRISPR mutagenesis (fig. S1D). This also indicates that cell growth was due to loss of MYBBP1A KO cells rather than a nonspecific effect of cell division rates. MYBBP1A KO cells hereafter only refers to cells that were transduced with MYBBP1A targeting sgRNAs and cultured for less than a week.

Comparative genomic hybridization identified PDAC cell lines PANC1, ASPC1, and BXPC3 as being hemizygous for MYBBP1A (24). To test whether loss of MYBBP1A KO could similarly reduce growth in PDAC cells that harbor one MYBBP1A allele, we transduced PANC1, ASPC1, and BXPC3 cells with sgRNA targeting MYBBP1A (fig. S2A). All PDAC cell lines showed threefold reduced cell growth upon KO of the remaining MYBBP1A allele (fig. S2B). To determine whether the defect in cell growth was attributable specifically to MYBBP1A loss, we used an sgRNA-resistant MYBBP1A expression construct that lacks the protospacer adjacent motif (PAM) sequence targeted by sgMYBBP1A#1. Expression of the sgRNA-resistant MYBBP1A rescued protein expression and cell growth defects (Fig. 1, D and E, and fig. S2, C and D). Consistent with MYBBP1A encoding a cell critical function, we found no evidence of any PDAC harboring homozygous loss of MYBBP1A among 167 samples in the TCGA database (fig. S2E).

The impaired cell growth after MYBBP1A homozygous loss was further studied by gene silencing using two doxycycline (dox)–inducible short hairpin RNAs (shRNAs) targeting MYBBP1A (shMYBBP1A), compared to a nontargeting shRNA (shControl). Dox treatment of shMYBBP1A HS766T cells reduced MYBBP1A protein by at least 90% and decreased cell growth, relative to shControl cells (Fig. 1, F and G). These findings were also found in shRNA-treated PANC1, ASPC1, and BXPC3 cells (fig. S2, F and G). Since anchorage-independent growth is a hallmark of carcinogenesis, we examined the MYBBP1A knockdown (KD) with a soft agar assay. Upon MYBBP1A KD, HS766T cells demonstrated decreased total colony number and size that was more marked than the effect on cell growth (fig. S3, A and B).

To test whether MYBBP1A is required for tumor growth in vivo, we performed injections of HS766T cells transduced with control sgRNA or sgRNA-targeting MYBBP1A in bilateral flanks of nonobese diabetic severe combined immunodeficient gamma (NSG) mice. Notably, MYBBP1A KO cells were associated with a nearly fourfold reduction in tumor weight and volume, compared to HS766T cells transduced with sgControl (Fig. 1, H and I, and fig. S3C). Together, these studies in various PDAC genetic backgrounds show that MYBBP1A is critical for cell growth in vitro and for tumor growth in vivo.

MYBBP1A haploinsufficiency can contribute to HS766T PDAC cell tumorigenesis

As expected, hemizygous MYBBP1A PDAC tumors from the TCGA database exhibit a roughly 50% decrease in MYBBP1A mRNA levels compared to PDAC tumors with both WT alleles (fig. S4A). We used the dox-inducible shRNA system to assay cell phenotypes with varying levels of MYBBP1A repression in HS766T PDAC cells, where both arms of chromosome 17p are intact (24). Dox doses sufficient to reduce MYBBP1A levels by approximately 50% in shMYBBP1A cells increased cell growth (fig. S4, B and C) and soft agar colony formation (fig. S4, E to G) in shMYBPP1A cells, compared to shControl cells. By contrast, when MYBBP1A expression was reduced by more than 90%, it decreased cell growth by 90% and colony formation by 75% (fig. S4, B and D).

To test the impact of MYBBP1A hemizygosity directly, we engineered a dox-inducible Cas9 HS766T cell line that stably expresses an sgRNA targeting the first start codon of MYBBP1A, to facilitate the homology-directed insertion of a transcriptional stop element and blasticidin-resistance cassette flanked by loxP sites (Fig. 2A and fig. S5A). Correct targeting of the MYBBP1A locus was confirmed by polymerase chain reaction (PCR) (fig. S5B), and the generated cells expressed about half of the MYBBP1A mRNA and protein compared to WT cells (Fig. 2, B and C). Cell growth (Fig. 2D), soft agar colony formation (fig. S5, C to E), and in vivo tumor growth (Fig. 2, F and G) were increased in the two hemizygous MYBBP1A KO clones tested, compared to WT cells. Expression of Cre recombinase, to excise the transcriptional stop cassette in hemizygous MYBBP1A cells, fully restored MYBBP1A expression (Fig. 2, B and C) and similarly restored cell growth (Fig. 2D) and tumorigenic phenotypes (Fig. 2, F and G) to that of MYBBP1A WT cells. No evidence of metastatic disease was found in xenograft models after 1 month, the time being limited by the tumor load in the WT xenografts, and an in vitro scratch test showed no difference in cell migration between MYBBP1A WT and hemizygous cells (fig. S5F).

Fig. 2 Hemizygous MYBBP1A is not associated with decreased cell growth and promotes HS766T tumorigenesis.

(A) Schematic of construction of hemizygous MYBBP1A cells. (B) qPCR of MYBBP1A hemizygous cell lines (means ± SEM, Student’s t test). *P < 0.05. (C) Immunoblot of MYBBP1A from nuclear lysates isolated from indicated cells. (D) Cell growth analysis by hemocytometry (means ± SEM, Student’s t test). *P < 0.05. (E) Tumor volume (n = 9 for WT, n = 8 for −/+ #1, −/+ #2, and +/fl#1 lines). *P < 0.005 as determined by Student’s t test. (F) Images of tumor at 30 days after transplantation. (G) Tumor weights 30 days after transplantation. *P < 0.005 as determined by Student’s t test. (H) WT HS766T cells or MYBBP1A hemizygous HS766T cells were transduced with sgMYBBP1A#1 or sgControl and implanted into NSG mice. Tumor volume (n = 5 tumors for each group) was measured with calipers until day 30 when all mice began receiving intraperitoneal injections of dox (2.5 mg/kg). Tumor volume was measured with calipers until day 45. Photo credit: A. Hsieh, University of Pennsylvania.

To test whether the targeting of the remaining allele of MYBBP1A hemizygotic cells could arrest cell growth in established tumors in vivo, we introduced an sgRNA targeting the MYBBP1A locus at exon 14 (sgMYBBP1A#1) or a nontargeting sgRNA into either HS766T MYBBP1A WT or hemizygous cells. As both HS766T WT and hemizygous mutant cells both harbor a MYBBP1A locus susceptible to sgMYBBP1A#1, reintroduction of Cas9 with dox exposure should mutate both MYBBP1A alleles in cells harboring sgMYBBP1A#1 (fig. S6A). The engineered cells were implanted into mice and allowed to form tumors for 30 days before administration of dox for 15 days. Consistent with the loss of MYBBP1A, both WT and hemizygous MYBBP1A tumors that were continuously exposed to dox that harbored sgMYBBP1A#1 showed decreased tumor growth (Fig. 2H) and smaller tumors compared to WT or hemizygous MYBBP1A tumors that expressed sgControl (fig. S6, B and C). Collectively, the results demonstrate that inducing MYBBP1A hemizygosity increases the tumorigenesis of transformed cells and, most notably, that the subsequent targeting of the remaining allele after tumor formation can reduce tumor growth.

Loss of chromatin-bound MYBBP1A is associated with impaired cell growth

Biochemical fractionation of PDAC HS766T cells showed that MYBBP1A protein is present in both nuclear soluble and chromatin fractions (Fig. 3A). However, in MYBBP1A hemizygous cells, there was complete MYBBP1A loss from the nuclear soluble fraction, while levels in the chromatin fraction were preserved (Fig. 3B). Thus, total nuclear MYBBP1A levels are normally in excess of that required to saturate its target sites in chromatin. In accordance, chromatin immunoprecipitation sequencing (ChIP-seq) analysis of MYBBP1A WT and MYBBP1A hemizygous clones demonstrated that MYBBP1A peaks were largely preserved in cells with reduced expression of MYBBP1A (Fig. 3C and fig. S7A). The data suggest that the chromatin-bound pool of MYBBP1A is associated with cell growth.

Fig. 3 MYBBP1A is preserved on the chromatin in MYBBP1A hemizygous cells.

(A) Subcellular fractionation of HS766T cells transduced with sgControl of sgMYBBP1A. (B) Subcellular fractionation of WT HS766T or MYBBP1A hemizygotic clones. (C) Heatmap view of MYBBP1A ChIP-seq detected in WT HS766T or MYBBP1A hemizygotic cells. (D) Gene ontology of MYBBP1A target genes in HS766T cells. (E) Overlap of MYBBP1A target ChIP-seq peaks in HS766T, PANC1, ASPC1, BXPC3, and BJ fibroblasts. (F) Diagram of CRISPR-Cas9 editing of FKBPF36V domain into the MYBBP1A locus. Arrows indicate primers used for PCR genotyping. (G) Immunoblot from PANC1/HA/FKBPF36V/MYBBP1A cells treated with dTAG13 or dimethyl sulfoxide (DMSO). (H) qPCR of PANC1/HA/FKBPF36V/MYBBP1A cells treated with dTAG13 or DMSO (means ± SEM, Student’s t test). *P < 0.05.

MYBBP1A ChIP-seq peaks in HS766T cells demonstrate that MYBBP1A is predominantly localized to the promoter of genes (fig. S7, B and C). Gene ontology analysis of the 219 MYBBP1A ChIP-seq peaks in HS766T MYBBP1A WT cells shows that MYBBP1A is bound to genes involved in processes such as cell cycle, DNA transcription, and positive regulation of circadian rhythm (Fig. 3D). Validation of the ChIP-seq identified peaks was performed by ChIP–quantitative PCR (qPCR) of 10 sites, which all demonstrated loss of ChIP-qPCR signal in MYBBP1A KO cells (fig. S7, B and D). Spearman analysis of MYBBP1A peaks also assessed in AsPC1, BxPC3, and PANC1 cells demonstrated that MYBBP1A binds to many more peaks in the other cell lines. While examination of MYBBP1A expression across available cell lines showed no consistent correlation between MYBBP1A protein level and ChIP-seq peaks (fig. S6F), 98% of the ChIP-seq peaks in HS766T were found in the other lines (Fig. 3E and fig. S7D).

MYBBP1A ChIP-seq in primary human BJ fibroblasts also revealed that 99% of MYBBP1A genomic binding sites in HS766T cells were found in BJ fibroblasts (Fig. 3E). MYBBP1A has been associated with transcriptional repression and heterochromatin (21). By integrating MYBBP1A ChIP-seq data with other studies (26, 27), MYBBP1A peaks in PANC1 cells typically reside in broad areas exhibiting modest levels of H3K9me3, H3K27me3, and H3K36me3 (fig. S8A), which can reflect areas that are generally repressed but may have a low level of basal activity (28). In addition, PANC1 MYBBP1A peaks are localized to H3K4me3 sites that are devoid of H3K4me1 (fig. S8A). In BJ fibroblasts, MYBBP1A localized with focal areas of H3K9me3, H3K27me3, and H3K36me3, as well as of H3K4me3 and broad levels of H3K4me1 (fig. S8B). Together, these findings reveal that MYBBP1A generally binds not only to chromatin harboring repressive chromatin features but also with H3K4me3, reminiscent of bivalent domains, which can represent a poised state for activity (29).

To study the consequences of immediate and selective loss of MYBBP1A on the gene expression of its target genes, we used the degradation TAG (dTAG) system. It comprises a small-molecule FKBP12F36V degradation (dTAG13) that induces dimerization of a FKBP12F36V-tagged protein with the Cereblon E3 ligase complex, in turn inducing a rapid, proteasome-dependent degradation of the target protein (30). Using CRISPR-Cas9, an sgRNA targeting the promoter region of MYBBP1A was used to facilitate the homologous recombination of a gene block with FKBP12F36V and a hemagglutinin (HA) tag to be expressed in-frame in PANC1 cells, which harbor a single MYBBP1A allele (Fig. 3F) (31). Successful insertion of FKBP12F36V and HA tag was confirmed by PCR and the detection of a 170-kDa MYBBP1A species by Western blot using either anti-HA or anti-MYBBP1A antibodies, corresponding to the expected size of the modified protein product (fig. S9, A to C). Treatment of FKBP12F36V/MYBBP1A PANC1 cells with dTAG13 resulted in MYBBP1A degradation in a dose-dependent manner (Fig. 3G). Notably, the expression of MYBBP1A target genes was found to be down-regulated after 24 hours of acute degradation of MYBBP1A (Fig. 3H). In summary, we found that MYBBP1A binds and up-regulates the transcription of genes involved in basic cellular processes. Furthermore, the localization of MYBBP1A at its target genes at the chromatin is lost in homozygous MYBBP1A KO cells but preserved in hemizygous MYBBP1A KO cells.

MYBBP1A interacts with PARP1

The preservation of chromatin-bound MYBBP1A in heterozygous KO cells and its absence in homozygous KO cells suggests that the MYBBP1A-bound chromatin is associated with cell essential functions. We hypothesized that targeting the preserved pool of chromatin-bound MYBBP1A in hemizygous MYBBP1A cells would mimic the detrimental cell growth effects of a complete MYBBP1A KO. Accordingly, we analyzed prior work to identify MYBBP1A interactors that could modulate its chromatin binding and focused on PARP1, as it regulates the chromatin localization of other proteins (3234). Reciprocal immunoprecipitation from whole-cell lysates detected an interaction between endogenous PARP1 and MYBBP1A in PANC1 and HS766T cells (Fig. 4A and fig. S10A), which was not affected by ethidium bromide, deoxyribonuclease (DNase), or ribonuclease (RNase) and, thus, appears to be independent of RNA or DNA (Fig. 4B). Although PARP1 is predominantly localized to the nucleoplasm, it also colocalizes with MYBBP1A in the nucleolus (Fig. 4C, arrows). In vitro pulldown assays using purified, recombinant PARP1 and MYBBP1A demonstrated a direct interaction between the two proteins (Fig. 4D and fig. S10B). Immunoprecipitation of PARP1 mutants and exogenous MYBBP1A showed that the ZNF (zinc finger) and BRCT (BRCA1 C Terminus) domains of PARP1 are required for its interaction with MYBBP1A (fig. S10C). Examination of PARP1 ChIP-seq from MCF10A (35), a breast cancer cell line, with MYBBP1A ChIP-seq peaks from HS766T, PANC1, ASPC1, BXPC3, and BJ fibroblast cells revealed significant overlap between PARP1 and MYBBP1A bound genomic sites (fig. S11, A and B). Together, we identify an interaction between MYBBP1A and PARP1 and their colocalization in the genome.

Fig. 4 PARP1 interacts and disrupts MYBBP1A interaction with polynucleosome.

(A) Endogenous MYBBP1A and PARP1 coimmunoprecipitation performed with MYBBP1A and PARP1 antibodies from HS766T whole-cell lysates. Arrow marks MYBBP1A-specific band. Asterisk denotes nonspecific band. IP, immunoprecipitation; IB, immunoblotting; IgG, immunoglobulin G. (B) Nuclear lysates were prepared from PANC1 transiently transfected with FLAG/MYBBP1A and immunoprecipitated with Flag beads in the presence of RNase, DNAse, or ethidium bromide (EtBr). NT, no treatment. (C) Immunofluorescence of WT HS766T cells. Scale bars, 20 μm. (D) Flag immunoprecipitation was performed recombinant Flag-tagged MYBBP1A and PARP1 purified from Escherichia coli. (E) Schematic of experimental design of assessing PARP1 effects on MYBBP1A binding to polynucleosome in vitro. (F and G) Purified MYBBP1A protein was incubated with PARP1 protein and then assessed for polynucleosome binding as shown in (E).

PARP1 impairs MYBBP1A interaction with polynucleosomes in vitro

PARP1 has pleiotropic roles in chromatin, where it can directly compact chromatin structure and regulate recruitment of proteins via enzymatic and nonenzymatic activities (12, 36). To test whether PARP1 directly modulates MYBBP1A interaction with chromatin, we designed an in vitro pulldown assay to assess polynucleosome binding (Fig. 4E). Bacterially purified, Flag-tagged MYBBP1A was immobilized onto anti-Flag beads and incubated with or without purified PARP1. The beads were then incubated with purified polynucleosomes (fig. S12A). As observed by the diminished recovery of histone H3 in the pulldown assay, we found that PARP1 can disrupt MYBBP1A nucleosome binding (Fig. 4F, lanes 1 to 4). Electromobility shift assay of the same reactions demonstrated that polynucleosomes remained intact during our assay conditions (fig. S12B). We repeated the nucleosome binding experiment with immobilized MYBBP1A that underwent extensive high-salt washes with 500 mM NaCl, to remove NAD+ after incubation with PARP1, and found that even in the absence of NAD+, which is an essential cofactor for PAR anabolism, PARP1 could still disrupt MYBBP1A nucleosome binding (Fig. 4G, lanes 1 to 4). Thus, PARP1 can disrupt MYBBP1A interaction with polynucleosomes in the apparent absence of NAD+.

We next examined the impact of PARPi olaparib and veliparib on the interaction of MYBBP1A with polynucleosomes. Several studies have demonstrated that in addition to blocking PARP1 catalytic activity, selective PARPi such as olaparib, but not veliparib, can prevent PARP1 disassociation from chromatin, referred to as PARP1 “trapping” (33, 34). PARP1 trapping has been recently described to be essential for PARPi-mediated toxicity and is independent from PARPi’s ability to inhibit PARP1 catalytic activity (35). We found that while both olaparib and veliparib blocked parylation by PARP1 on MYBBP1A, only veliparib prevented PARP1 from inhibiting MYBBP1A binding to the polynucleosomes (Fig. 4F, lanes 5 to 8). Given olaparib’s ability to induce PARP1 trapping while veliparib cannot, we hypothesized that if excess PARP1 that was not already bound to MYBBP1A was removed from the nucleosome binding reaction, then olaparib would act similarly to veliparib and inhibit PARP1-mediated displacement of MYBBP1A from chromatin. In support of this hypothesis, when excess non-MYBBP1A interacting PARP1 was removed by high-salt washes, both olaparib and veliparib reversed the effects of PARP1’s ability to disrupt MYBBP1A interaction with polynucleosomes (Fig. 4G, lanes 5 to 8). Last, to examine whether PARP1 could evict MYBBP1A from nucleosomes after MYBBP1A and nucleosomes were allowed to form a stable complex, we performed binding assays with PARP1 and preformed MYBBP1A-polynucleosomes complexes (fig. S12C) and found that PARP1 could still displace MYBBP1A interactions with polynucleosomes (fig. S12D). The result suggests that PARP1 actively displaces MYBBP1A from polynucleosomes. In summary, PARP inhibitors that induce PARP1 trapping impair MYBBP1A interaction with polynucleosomes in vitro.

PARPi decreases MYBBP1A chromatin localization in vivo

We next examined whether the ability to trap PARP1 on chromatin by PARPi could modulate MYBBP1A nucleolar localization. While PARP1 localizes at both the nucleoplasm and nucleolus, upon exposure to olaparib and niraparib, PARP1 translocated to areas of more intense 4′,6-diamidino-2-phenylindole (DAPI) stain (Fig. 5A), which is consistent with their known PARP1 trapping activity (37). Veliparib lacks PARP1 trapping activity (38), and thus, as expected, no changes in PARP1 localization were observed when cells were exposed to veliparib (Fig. 5A). Olaparib and niraparib caused a redistribution of MYBBP1A from the nucleolus (Fig. 5A, arrowheads) to the nucleoplasm. The effect correlates with the nucleosome binding assay, demonstrating that PARP1 trapping by olaparib impairs MYBBP1A interaction with polynucleosomes (Fig. 4F, lanes 5 and 6). By contrast, veliparib, which does not cause PARP1 trapping and does not impair MYBBP1A interaction with polynucleosomes in vitro (Fig. 4F, lanes 7 to 8), did not cause observable changes to MYBBP1A localization (Fig. 5A). In addition, no changes in localization were observed in the localization of the nucleolar protein, fibrillarin (fig. S8A), demonstrating a specific effect of olaparib and niraparib on MYBBP1A localization.

Fig. 5 PARPi that causes PARP1 trapping disrupts MYBBP1A chromatin localization.

(A) Immunofluorescence of MYBBP1A and PARP1 after 3 hours after PARPi treatment in HS766T cells. DNA was visualized with costaining with DAPI. Scale bars, 5 μm. (B) SMT was performed on PANC1 cells stably expressing Halo-tagged H2B, MYBBP1A, or PARP1. Spatial distribution of H2B, MYBBP1A, and PARP1 molecules based on either low or high diffusion coefficient over a single nucleus is shown. (C) PANC1 cells stably expressing H2B, MYBBP1A, or PARP1 were treated with DMSO, olaparib, or niraparib for 3 hours. Logarithmic distribution of diffusion coefficients of H2B, MYBBP1A, and PARP1 under indicated treatments is shown. (D) PANC1 cells were treated with DMSO or indicated PARPi for 6 hours and fractionated to chromatin and nuclear soluble compartments. (E) ChIP of MYBBP1A was performed from cross-linked chromatin of HS766T cells with DMSO or indicated PARPi for 6 hours (means ± SEM, Student’s t test). *P < 0.05. (F) ChIP of PARP1 was performed from cross-linked chromatin of HS766T cells with DMSO or indicated PARPi for 6 hours (means ± SEM, Student’s t test). *P < 0.05.

Single-molecule tracking (SMT) allows the monitoring of real-time protein and chromatin binding dynamics in living cells, with a high spatiotemporal resolution (39). The measure of diffusion coefficients (in square micrometers per second) can discern proteins in a chromatin bound state (low diffusion coefficients) from proteins undergoing free diffusion (high diffusion coefficients) (40). To assess the chromatin-bound status of MYBBP1A visualized in the different nuclear subregions, we generated cell lines stably expressing HALO-tagged MYBBP1A. Visualization of Halo-tagged MYBBP1A molecules, based on their high or low diffusion coefficients, demonstrated that MYBBP1A enrichment at the nucleolus, as expected, generally exhibited low diffusion, reflecting a chromatin bound state (Fig. 5B, blue dots). By contrast, high diffusion coefficient MYBBP1A molecules were more enriched outside of the nucleolus (Fig. 5B, red dots). HALO-tagged histone 2B (H2B) exhibits low diffusion coefficients throughout the nucleus, showing that low diffusion proteins are not unique to the nucleolar compartment (Fig. 5B). While PARP1 also localizes to the nucleolus by immunofluorescence microscopy, PARP1-HALO exhibited low and high diffusion coefficients throughout the nuclei, including at the nucleolus (Fig. 5B). We hypothesized that PARP inhibitors that induce PARP1 chromatin trapping would increase the proportion of MYBBP1A molecules that show free diffusion, by having elicited release of MYBBP1A from the chromatin. In only 3 hours of olaparib treatment, the fraction of free-diffusing MYBBP1A molecules is increased, with a reciprocal decrease in chromatin-bound MYBBP1A molecules (Fig. 5C and fig. S8B). By contrast, there was an increase in chromatin-bound PARP1 molecules and decrease in free-diffusing PARP1 molecule after 3 hours of olaparib treatment, which is consistent with olaparib-mediated PARP1 trapping on chromatin (Fig. 5C and fig. S8B). Neither solvent nor veliparib, which does not induce PARP1 chromatin trapping, changed MYBBP1A or PARP1 mobility. The SMT technology thereby shows that PARP1 trapping triggers MYBBP1A disassociation from chromatin dynamically in living cells.

To corroborate these findings, we performed chromatin and nuclear soluble protein isolation by biochemical fractionation of HS766T cells, with and without PARPi. As expected from prior work (38), increased chromatin-bound PARP1 was observed in cells treated with olaparib and niraparib (Fig. 5D). However, the pool of chromatin-bound MYBBP1A was depleted in cells treated with olaparib and niraparib but not with veliparib (Fig. 5D). Furthermore, the effect of olaparib and niraparib on MYBBP1A chromatin localization was found to be dose dependent (fig. S8, C to E). The negative regulation of MYBBP1A chromatin localization by PARP1 trapping is supported by the observation that gene silencing of PARP1 increases MYBBP1A chromatin localization (fig. S8F). Furthermore, ChIP-qPCR in HS766T cells treated with PARPi demonstrated that olaparib and niraparib, but not veliparib, induced loss of MYBBP1A and a gain of PARP1 at MYBBP1A-bound loci (Fig. 5, E and F). Together, these results indicate that PARP inhibitors that induce the trapping of PARP1 on chromatin can displace MYBBP1A from chromatin.

MYBBP1A hemizygous tumors are growth sensitive to PARP trapping

Niraparib and olaparib cause a significant reduction of the MYBBP1A chromatin pool in MYBBP1A hemizygous cells but not WT HS766T cells (Fig. 6A). Thus, niraparib and olaparib may be therapeutic agents in MYBBP1A hemizygous cells, because these cells have a reduced reservoir of total MYBBP1A protein that can access the chromatin. In support of the chromatin-bound MYBBP1A fraction being critical for cell viability, olaparib had a lower concentration where 50% of an effect on cell viability is observed (median inhibitory concentration (IC50)] in MYBBP1A hemizygous cells, compared to WT HS766T cells (Fig. 6B). The differences in olaparib sensitivity enabled us to find that at selective doses, olaparib can impair cell growth in MYBBP1A hemizygous cells but not in MYBBP1A WT cells (Fig. 6C). Olaparib also reduced anchorage-independent growth of isogenic MYBBP1A hemizygous cells, but not of WT cells (fig. S14, A to C). To examine whether olaparib could similarly impair cell growth in other PDAC cell lines with reduced MYBBP1A expression, we used PANC1 and ASPC1 cells that stably express shRNAs targeting MYBBP1A from a dox-inducible promoter. Using a dose of dox sufficient to reduce MYBBP1A levels so that only the chromatin-bound MYBBP1A remained (fig. S14D), we found that PANC1 and ASPC1 cells had a reduced IC50 to olaparib (fig. S14, E and F). Chromatin fractionation of PANC1 and ASPC1 cells with reduced MYBBP1A showed that olaparib could effectively reduce the remaining chromatin-bound MYBBP1A in these cells (fig. S15, A and B).

Fig. 6 Hemizygous MYBBP1A cells are more susceptible to PARPi trapping.

(A) HS766T WT or MYBBP1A hemizygous cells were treated with indicated PARPi for 6 hours and then fractionated. (B) Cell viability of cells treated at indicated concentrations of olaparib was measured by CellTiter-Glo assay (means ± SEM). (C) Cell growth analysis of HS766T WT or MYBBP1A hemizygous cells with or without 10 μM olaparib (means ± SEM, Student’s t test). *P < 0.05. (D) Tumor volume (n = 5 tumors for each group) were measured with caliper after injection with HS766T WT or MYBBP1A hemizygous cells into flanks of NSG mice. Oral lavage was performed with olaparib. (E) Images of tumor at 45 days after transplantation. (F) Tumor weights 45 days after transplantation. *P < 0.01 as determined by Student’s t test. ns, not significant. (G) Working model of this study. In MYBBP1A WT cells, there is an abundance of chromatin-bound MYBBP1A protein and a reserve of nuclear soluble MYBBP1A protein. In hemizygous MYBBP1A cells, loss of MYBBP1A expression results in the preservation of chromatin-bound MYBBP1A. PARPi that trap PARP1 to chromatin are able to evict MYBBP1A proteins from chromatin, resulting in phenotypic loss of MYBBP1A function. Photo credit: A. Hsieh, University of Pennsylvania.

To test whether olaparib can selectively reduce tumor formation in vivo, we treated mice that harbored tumors, 30 days after injection with either WT HS766T cells or MYBBP1A hemizygous HS766T cells, with olaparib or vehicle via daily oral lavage in mice (Fig. 6D). While olaparib minimally reduced in vivo growth of MYBBP1A WT tumors, there was a threefold reduction in tumor size of MYBBP1A hemizygous tumors that harbor reduced MYBBP1A levels (Fig. 6, D to F, and fig. S16). Our results demonstrate a selective efficacy of PARP1-trapping agents on cell growth and in reducing the growth of existing PDAC tumors with a MYBBP1A hemizygous genotype (Fig. 6G).

DISCUSSION

Extensive genomic characterization of PDAC and the recognition of multiple oncogenic pathways that drive PDAC pathogenesis can suggest targeted approaches to cancer treatment (11). While chromosomal abnormalities are a hallmark of cancer, targeting functionalities lost by these mutations have been challenging (41). In this study, we used functional genomic data from TCGA and DepMap to identify MYBBP1A as a candidate targetable gene that is frequently hemizygous due to 17p loss in PDAC. We find that while MYBBP1A KO impairs cell growth, hemizygous MYBBP1A increased tumorigenesis. We further demonstrate that PARPIs such as olaparib, which induce PARP1 trapping, demonstrate selective toxicity in cells hemizygous for MYBBP1A. This offers a therapeutic window for selective PARP inhibitors in 17p-deleted PDAC, because MYBBP1A WT cells have a greater pool of MYBBP1A protein that can maintain chromatin MYBBP1A levels necessary for necessary cell growth in the presence of olaparib. Our findings may have broad ranging implications, given that del(17p) occurs in up to 64% of PDAC (4).

Prior studies demonstrate conflicting roles for MYBBP1A in cancer (42). Specifically, loss of MYBBP1A was associated with decreased, increased, or unchanged growth of cancer cell types (20, 23, 4244). Although it is possible that MYBBP1A may have cancer type–specific functions, the discrepancies in past studies can also be reconciled by the variable repression of MYBBP1A protein, as noted in Mori et al. (21). Our study uses multiple redundant loss-of-function systems including RNA interference (RNAi), shRNA, and CRISPR mutagenesis in multiple PDAC cell lines to demonstrate that complete loss of MYBBP1A compromises cell growth and impairs tumorigenesis in vivo and in vitro. Paradoxically, using an adjustable shRNA-targeting MYBBP1A or using CRISPR mutagenesis to create MYBBP1A hemizygous cells, we find that partial loss of MYBBP1A expression can enhance tumorigenesis. Therefore, our study demonstrates that MYBBP1A gene dosage has distinct consequences on cell growth and can resolve the discrepant findings in past studies.

MYBBP1A has also been demonstrated to either promote or impair cell migration (20, 42). The lack of any changes to cell migration in MYBBP1A hemizygous cells appears at odds with these prior studies, but this can be reconciled by the degree of MYBBP1A KD performed in prior studies as both prior studies achieved 80 to 90% reduction of MYBBP1A levels compared to the 50% reduction of MYBBP1A protein in MYBBP1A hemizygous cells. Further work will be needed to dissect the exact role of MYBBP1A in cell migration, but our work highlights the need to use careful titration of MYBBP1A expression to fully understand MYBBP1A’s role in cell migration.

The role of MYBBP1A in chromatin and how its loss contributes to impaired growth remains to be elucidated. A prior study reported a role for MYBBP1A in repressing the transcription factors of its gene targets (42). However, the gene programs of transcription factors repressed by MYBBP1A, such as peroxisome proliferator–activated receptor gamma coactivator, RelA/p65, and period circadian regulator 2, are not associated with impaired cell growth (4547). Given its nucleolar localization, MYBBP1A has also been thought to play a role in the biogenesis of ribosomal RNAs (rRNAs), in part, by up-regulating rRNA transcription (48). Our study reveals that MYBBP1A binds to genes that are involved in basic cellular processes and that rRNA genes constitute only a minority of genes regulated by MYBBP1A. We further show that immediate loss of MYBBP1A decreases the transcription of MYBBP1A target genes, which is consistent with prior studies, demonstrating that MYBBP1A can up-regulate transcription of target genes (48). Additional functions for MYBBP1A in governing cell growth remain under investigation.

PARP inhibitors were previously shown to be synthetically lethal in cancers with mutations in BRCA or other homologous repair pathway genes by blocking alternate DNA repair pathways (12). However, multiple clinical trials in breast and ovarian cancer have demonstrated an efficacy of PARP inhibitors in cancers that are not known to harbor DNA repair gene mutations (1315). This suggests the existence of non-DNA repair mechanisms of PARP inhibitor–dependent cell killing. It was recently shown that PARPi-dependent loss of DDX21, an RNA helicase that promotes ribosomal DNA transcription, can induce breast cancer cell death (49). We highlight a new mode of PARPi-mediated cancer killing, which is selective to 17p-deleted cancer cells and is dependent on both the inhibition of PARP1 catalytic activity and PARP1 trapping. Our data suggest that del(17p) may be predictive of particular PARP inhibitor efficacy and supports the use of partial 17p loss as a biomarker for selecting patients who may benefit from particular PARP inhibitors.

Expanding the use of PARP inhibitors beyond current BRCA clinical indications has the potential to significantly affect patients with PDAC, since BRCA and homologous repair pathway gene mutations are relevant but relatively uncommon in PDAC (50). Given that MYBBP1A appears to be critical across diverse cell types, our results suggest that PARP inhibitors may have activity in 17p-deleted cancers other than PDAC (18). For instance, PARP inhibitors have clinically significant efficacy in unselected breast and ovarian cancers, and coincidentally, both diseases are frequently associated with loss of heterozygosity of 17p. Because del(17p) is also associated with treatment resistance and poor prognosis across cancer types, our findings suggest that particular PARP inhibitors, but not all of them, may fulfill a critical need of a tolerable and effective therapy for challenging malignancies. Notably, our results predict that only PARP inhibitors that trap PARP1 and inhibit poly-ADP ribosylation will be efficacious in 17p-deleted cancers, which indicates that careful contextual selection of PARP inhibitors is necessary in future trial designs.

MATERIALS AND METHODS

Cell lines and genome editing

PANC1, HS766T, BxPC3, AsPC1, and BJ fibroblasts cells were obtained from the American Type Culture Collection. Cells were confirmed to be free of Mycoplasma infection. PANC1 and HS766T cells were cultured in Dulbecco’s modified Eagle’s medium (DMEM) supplemented with 10% fetal bovine serum in a 5% CO2 incubator at 37°C. BxPC3 and AsPC1 cells were cultured in RPMI 1640 medium supplemented with 10% fetal bovine serum in a 5% CO2 incubator at 37°C. Hemizygotic MYBBP1A HS766T cells were established by initially creating a dox-inducible Cas9 expression HS766T cell line by transducing HS766T cells with TLCV2 (Addgene, #87360) and selecting with puromycin. This cell line was then transduced with LRCherry2.1 (Addgene, #108099) harboring sgRNA targeting the 5′ untranslated region (5′UTR) of MYBBP1A. Cells harboring mCherry signal were isolated by flow cytometry. This cell line was then pulsed with dox for 48 hours at 2 μg/ml and electroporated with a gene block containing 500–base pair (bp) homology arms from either regions of the transcriptional start site of MYBBP1A, followed by a blasticidin resistance gene and SV40 termination sequence flanked by loxP sites. Individual blasticidin-resistant clones were isolated and genotyped by primers provided in table S1. To remove the blasticidin resistance and SV40 termination sequence, MYBBP1A hemizygotic cells were infected with lentiviral-Cre (Addgene, #86805) with a multiplicity of infection (MOI) of 200. PANC1 cells stably expressing Halo-tagged H2B, MYBBP1A, and PARP1 were created by transducing PANC1 cells with an MOI of 0.3. dTAG/HA/MYBBP1A PANC1 cells established by initially creating a constitutively Cas9-expressing cell line by transducing PANC1 cells with LentiCRISPRv2 (Addgene, #52961) and selecting with puromycin. This cell line was then cotransfected with LRCherry2.1 (Addgene, #108099) harboring sgRNA targeting the 5′UTR of MYBBP1A and a PCR-amplified gene block containing 500-bp homology arms from either regions of the transcriptional start site of MYBBP1A, followed by a blasticidin resistance gene and FKBPF36V/HA sequence. Individual blasticidin-resistant clones were isolated and genotyped by primers provided in table S1.

Plasmid construction

sgRNA vectors. Oligos (see table 1) harboring sgRNA sequence targeting ROSA or exons 14 and 16 of MYBBP1A (see table 1) were annealed by incubating at 95°C and then cooled slowly to room temperature. Annealed oligos were ligated with T4 DNA ligase [New England Biolabs (NEB), M0202T] into the Bsm BI restriction site of LentiCRISPRv2 (Addgene, #52961).

shRNA vectors. Oligos (see table 1) harboring shRNA sequence targeting MYBBP1A 3′UTR or 5′UTR were annealed by incubating at 95°C and then cooled slowly to room temperature. Annealed oligos were ligated with T4 DNA ligase into the Age I and Eco RI site of Tet-pLKO-puro vector (Addgene, #21915). Control shRNA was obtained from Addgene (Addgene, #47541).

Mammalian expression vectors. sgRNA-resistant MYBBP1A construct was created by PCR mutagenesis with primers (see table 1) that harbor silent mutations in the PAM sequence and seed region using pFRT/TO/HIS/FLAG/HA-MYBBP1A (Addgene, #38084) as the template. Mutagenesis was confirmed by Sanger sequencing, and the sgRNA-resistant MYBBP1A complementary DNA (cDNA) was cloned into the Bam HI site of pWTP/green fluorescent protein (GFP) (our laboratory) by PCR. For expression of Flag-tagged MYBBP1A, cDNA of MYBBP1A was PCR-amplified and cloned into the Bam HI site of pENTR1A (Thermo Fisher Scientific) vector using In-Fusion HD (Takara Bio). MYBBP1A cDNA was then subcloned into pLenti6.2-ccdb-3×Flag-V5 (Addgene, #87072) by Gateway cloning. Halo-tagged expression constructs were made by PCR-amplifying MYBBP1A and PARP1 cDNA and cloning into Eco RI and Kpn I–digested pCMV-FOXA1-HaloTag construct using T4 DNA ligase (NEB M0202T). The HaloTag-fused MYBBP1A or PARP1 was PCR-amplified and assembled using Gibson Assembly Master Mix kit (NEB, E2611L) with Eco RI (NEB)–digested TETO-FUW-OCT4 (Addgene, plasmid #20323). pCMV_Tag5a/PARP1 deletion mutant expression constructs were a gift from J. Mendell.

Bacterial expression vectors. MYBBP1A cDNA was amplified by PCR and cloned into the Ssp I site of pET/His6/Sumo vector (Addgene, #29659).

Lentiviral construction. Human embryonic kidney–293 T cells were seeded at 90% confluency in 10-cm dishes with fresh media 24 hours before transfection. Cells were transfected with lentiviral expression vector, psPAX2 (Addgene, #12260), and pMD2.6 (Addgene, #12259) using FuGENE 6 (Promega E2691) at a ratio of 1 μg of DNA to 3 μl of FuGENE 6. Media were collected every 24 hours up to 72 hours after transfection. Cell debris was removed by centrifugation at 2000g for 10 min at 4°C and passaged through a 0.45-μm filter (Millipore, SLHV033RS). Viral particles were concentrated by ultracentrifugation at 25,000 rpm in an SW-32 swinging bucket rotor (Beckman Coulter). Viral particles were resuspended in 150 μl of DMEM high glucose and snap-frozen with liquid nitrogen and stored at −80°C.

Cell growth assay. Cell growth was measured by hemocytometry with 0.4% trypan blue (Thermo Fisher Scientific, catalog no. 15250061) staining. Approximately 50,000 cells were plated in triplicates in 24-well plates. Cells were counted twice using a hemocytometer the following day to calculate total cell number, and the total cell number was determined again at 3, 6, 9, 12, and 15 days after the initial cell count.

Cell viability assay. Cell viability was assessed by CellTiter-Glo Luminescent Cell Viability Assay (Promega, catalog no. G7572). Cells were plated in 24-well plates at 50,000 cells per well and treated with PARPi at indicated concentrations for 96 hours. Cells were lysed with CellTiter-Glo Luminescent Cell Viability Assay reagent, and luminescence was read using the Envision plate reader. The percentage of cell growth was calculated relative to dimethyl sulfoxide (DMSO)–treated cells.

Soft agar assay. Cells were suspended in 1 ml of DMEM supplemented with 10% fetal bovine serum containing 0.3% agarose and plated in triplicate on a firm 0.6% agarose base in six-well plates (3000 cells per well) as described previously. Cells were then placed in a 37°C and 5% CO2 incubator. Colonies of cells were allowed to grow over the course of 3 weeks, and then, colonies were stained with 0.05% crystal violet. Images were obtained using a Nikon Eclipse TE2000-U microscope controlled by Nikon Elements software.

Migration scratch assay. Cells were plated at 95% confluency in 24-well plates coated with Matrigel (BD Biosciences, catalog no. 356234). A P1000 tip was used to produce a scratch along premarked lines. Cells were washed once with phosphate-buffered saline (PBS) before receiving fresh media and placed in a 37°C and 5% CO2 incubator.

Plates were placed under an inverted microscope for photographing at 0 and 24 hours. Images were obtained using a Nikon Eclipse TE2000-U microscope controlled by Nikon Elements software.

Xenograft assay. WT or CRISPR-modified HS766T cells were trypsinized and washed twice with PBS and resuspended in equal volume of Matrigel (BD Biosciences, catalog no. 356234) and DMEM. For establishment of each tumor, 106 cells in 100-μl final volume were injected into the flanks of 8-week-old female NSG mice. For evaluation of MYBBP1A KO on established tumors, dox (2.5 mg/kg; Sigma-Aldrich, catalog no. D9891) was injected intraperitoneally after transplantation for 15 days from days 30 to 45. For evaluation of PARPi on tumor growth, olaparib (50 mg/kg per day; Selleckchem, catalog no. S1060) was administered by oral lavage for 15 days from days 24 to 40. Tumors were measured with a caliper every 3 days after transplantation (tumor volume = width2 × length × 0.523). Animals were maintained in accordance with the institutional guidelines (University of Pennsylvania).

Chromatin immunoprecipitation. HS766T cells were plated at 80% confluency 1 day before cross-linking in 150-cm plates with 20 ml of medium. Cells were cross-linked directly in the cell culture plate with 540 μl of 37% formaldehyde (Millipore, F8775) for 10 min at room temperature. Cross-linking reaction was quenched with addition of 1440 μl of 2 M glycine for 5 min at room temperature. Cells were washed three times with ice-cold PBS and then collected by centrifugation at 300g. For isolation of chromatin, cells were lysed with lysis buffer [10 mM Hepes (pH 7.4), 10 mM KCl, and 0.05% NP-40] with protease inhibitor cocktail (Roche) for 30 min, and nuclei were collected by centrifugation at 600g for 10 min. Nuclear lysis was performed with low-salt buffer [10 mM tris-HCl (pH 7.4) 0.2 mM MgCl2, and 1% Triton X-100) with protease inhibitor cocktail for 30 min, and chromatin was collected by centrifugation at 5000g for 10 min. Chromatin was resuspended in NP-40 lysis buffer [50 mM tris-HCl (pH 8.0), 150 mM NaCl, and 1% NP-40) and sonicated with S220 Focused Ultrasonicator (Covaris) with settings: incident peak power: 140, duty cycle: 5%, and burst cycle: 200 for optimal chromatin size of 200 to 500 bp in length. Cell debris was then pelleted by centrifugation at 8000g for 10 min at 4°C. Supernatant was then used for immunoprecipitation with anti-MYBBP1A (Sigma-Aldrich, catalog no. HPA005466), anti-PARP1 (Active Motif, catalog no. 39559), and rabbit immunoglobulin G (IgG) (Abcam, catalog no. ab37415) antibodies at 4°C on a rotating platform overnight. Dynabeads Protein G (Thermo Fisher Scientific) was added and incubated for an additional 2 hours at 4°C on a rotating platform. Immunoprecipitated chromatin was washed three times with NP-40 lysis buffer, two times with LiCl buffer [0.5 M LiCl, 1% deoxycholic acid, and 100 mM tris-HCl (pH 9.0)], one time with TE buffer [10 mM tris-HCl (pH 8.0) and 1 mM EDTA]. Chromatin was eluted with elution buffer [50 mM tris-HCl (pH 8.0), 10 mM EDTA, and 1% SDS) at 50°C for 15 min and reversed cross-linked at 65°C overnight. Samples were diluted twofold and treated with 8 μl of RNase A (10 mg/ml; Roche, catalog no. 10109169001) for 2 hours at 37°C, followed by stock Proteinase K (4 μl 20 mg/ml ; Roche, catalog no. 03115828001) at 55°C for 2 hours. DNA precipitation was performed with phenol-chloroform and resuspended in TE, and DNA concentration was measured by Qubit double-stranded DNA HS Assay (Thermo Fisher Scientific, catalog no. Q32854).

ChIP-qPCR. Isolated DNA from ChIP was diluted to 0.1 ng/ml. qPCR reactions were performed in 384-well optical plates with 2 μl of DNA sample, 10 mM forward and reverse primer, and 5 μl of Power SYBR Green PCR Master Mix (Thermo Fisher Scientific, #4367659), as well as nuclease-free water (Invitrogen, catalog no. 10977-015) in a total volume of 10 μl. qPCR was performed with QuantStudio 7 Flex (Applied Biosystems), using the following thermal cycler protocol: 50°C for 4 min, 95°C for 10 min, followed by 40 cycles of 95°C for 15 s, then 54°C for 15 s, and then 72°C for 45 s, with a dissociation curve generated to verify that a single PCR product was generated. qPCR results for each fraction were normalized to the input sample.

Preparation of ChIP-seq libraries. Fifty nanograms of each biological replicate was used for library preparation using the NEBNext Ultra DNA Library Prep Kit for Illumina (NEB, E7370S) and amplified using 10 cycles of PCR using NEBNext Multiplex Oligos for Illumina (NEB, E7335S andE7500S). DNA was size-selected using Agencourt AmpureXP beads (Beckman Coulter, A63881), following the protocol in the NEBNext Ultra Kit for a 200- to 600-bp average insert size. Library fragment size distribution was assessed on a Bioanalyzer 2100 instrument (Agilent Technologies), using the DNA 1000 Kit (Agilent Technologies, 5067-1504). Quantification of libraries was performed using the NEBNext Quant Kit for Illumina (NEB, E7630L).

ChIP-seq analysis. MYBBP1A and input ChIP-seq data were aligned using bowtie2 (v2.3.4.1) with parameters -p 16 --local -X 1000. Alignments were quality filtered using samtools (v1.1) view with a MAPQ filter of 5 (-q 5). Four lanes of NextSeq data per sample were merged using samtools merge with sorting on tag ID, and PCR and optical duplicates were removed using PICARD (v2.21.3-SNAPSHOT) MarkDuplicates with parameter REMOVE_DUPLICATES = True. Peaks were called using MACS2 (v2.1.0) with parameters -f BAM -g hs -s 37 -q 0.01 and using the input sample as a control (-c). Peak output was filtered at a false discovery rate of <1%, and BED files of filtered peaks were pooled across replicates, merging overlapping peak regions. ENCODE blacklisted regions from https://sites.google.com/site/anshulkundaje/projects/blacklists were then removed using bedtools intersect -wa -v. Tracks for data visualization were created in the following way: Genome maps were created using BEDtools genomeCoverageBed -bg, and then, samples were adjusted for library size by multiplying all bin counts by the RPM; maps were then subtracted chIP-input, sorted using UC Santa Cruz Genome Browser tools’ bedSort and converted to bigWig format using UCSC Genome Browser tools’ bedGraphToBigWig. Correlation matrices were assessed by pooling peaks for all samples under consideration, merging overlapping regions, and measuring area under the curves (RPM-adjusted, length-adjusted ChIP-input tag density) for each sample over the pooled set and then calculating Spearman rank correlations between pairs of samples [in R, cor(matrix, method = “spearman”)]. Heatmaps were generated by assessing vectors (length = 100 U) of input-subtracted, RPM-adjusted ChIP tag density as described above across a 5-kb (contrast of HS766T WT versus Mybbp1a hemizygous samples) or 4-kb window (PANC1 and BJ); data are plotted in descending order vertically by Mybbp1a WT total enrichment. Histone PTM (post-translational modification) data used for comparison heatmaps are Pash-mapped BED files of aligned tags downloaded from the National Center for Biotechnology Information Gene Expression Omnibus (GEO) and pooled using cat; GEO accessions are GSM1127060, GSM1127076, and GSM958163 for H3K27ac; GSM817234, GSM941717, and GSM958164 for H3K4me1; GSM817235, GSM941718, and GSM958158 for H3K4me3; GSM817236 and GSM817239 for H3K9me3; GSM817237, GSM817240, and GSM958154 for H3K27me3; GSM817238, GSM817241, and GSM958149 for H3K36me3; GSM817246 and GSM817247 for Input; and GSE93040 for PARP1.

Chromatin fractionation. HS766T and PANC1 cells that were 80% confluent were trypsinized and washed three times with PBS. Cells were collected by centrifugation at 250g for 5 min at 4°C and fractionated using the Subcellular Protein Fractionation Kit for cultured cells (Thermo Fisher Scientific, catalog no. 78840). Fractions were either used for immunoprecipitation or denatured with NuPAGE LDS Sample Buffer and heated to 95°C for 10 min.

Immunoprecipitation. HS766T and PANC1 cells seeded at 90% confluency in 15-cm dishes were washed with ice-cold PBS, scrapped off the plate, and pelleted by centrifugation at 300g for 5 min at 4°C. For immunoprecipitation of whole-cell lysates, cells were lysed with NP-40 lysis buffer with protease inhibitor cocktail and briefly sonicated in polystyrene tubes using a Bioruptor UCD-200 [power HI (Diagenode) with settings: cycles of 30-s on/30-s off] for five cycles and cell debris removed by centrifugation at 13,000g for 10 min at 4°C. For immunoprecipitation of nuclear or chromatin fractions, fractions were obtained as above (see “Chromatin fractionation” section). Immunoprecipitation using anti-MYBBP1A (Sigma-Aldrich, catalog no. HPA005466), anti-Flag (Sigma-Aldrich, catalog no. F3165), and rabbit IgG (Abcam, catalog no. ab37415) antibodies at 4°C on a rotating platform overnight. Dynabeads Protein G (Thermo Fisher Scientific) was added and incubated for an additional 2 hours at 4°C on a rotating platform. Immunoprecipitated protein was eluted by applying NuPAGE LDS Sample Buffer to beads and heated to 95°C for 10 min.

Immunofluorescence. HS766T cells were grown on plates coated with poly-l-lysine (Millipore, catalog no. P8920). After washing three times with PBS, cells were fixed in 4% paraformaldehyde (Electron Microscopy Sciences, catalog no. 15714) in PBS for 15 min at room temperature. Fixed cells were washed three times with PBS and permeabilized with 0.1% Triton X-100 in PBS for 5 min. Cells were washed twice with PBS and blocked with 4% donkey serum (Sigma-Aldrich, D9663) in PBS for 1 hour at room temperature. Primary antibodies were diluted in 4% donkey serum in PBS with the following concentrations: anti-MYBBP1A (1:200; Sigma-Aldrich, catalog no. HPA005466), anti-MYBBP1A (1:200; Novus Biologicals, catalog no. H00010514-B01P), anti-PARP1 (1:200; Active Motif, catalog no. 39559), anti-PAR (1:200; Millipore, catalog no. MABC547), and anti-fibrillarin (1:500; Abcam, catalog no. ab166630). Cells were washed three times with PBS and then incubated with Alexa Fluor 488– or Alexa Fluor 594–conjugated secondary antibodies raised in donkey (Thermo Fisher Scientific) at a 1:1000 dilution in PBS for 1 hour at room temperature. Samples were then washed three times with PBS and counterstained with DAPI (1 mg/ml; Thermo Fisher Scientific, D1306) in PBS for 10 min. Cells were washed two times with PBS and visualized with a Nikon Eclipse TE2000-U microscope controlled by Nikon Elements software and equipped with the appropriate filters.

Single-molecule live-cell imaging. All single-molecule live-cell imaging was carried out on a Nanoimager S from Oxford Nanoimaging Limited, comporting a temperature and humidity controlled chamber, a scientific complementary metal oxide semiconductor camera with a 2.3-electron rms read noise at standard scan, a 100×, 1.49–numerical aperture oil-immersion objective and a 561-nm green laser. Images were acquired with the included Nanoimager software. A total of 30,000 PANC1 cells were cultured in Lab-Tek II chambered eight-well plates (Lab-Tek, 155049), transfected with the appropriate tetracycline-inducible lentivirus when necessary [without rtTA (tetracycline-dependent transactivator) or dox to keep low levels of expression] and then incubated at 37°C and 5% CO2 for 3 days. Before imaging, cells were treated with 5 nM Janelia Fluor 549 HaloTag ligand (a gift from L. Lavis laboratory) for 15 min. Cells were subsequently washed three times in PBS at 37°C, and phenol red–free low-glucose medium was added to each well. All imaging was carried out under HILO (highly inclined and laminated optical) conditions Tokunaga et al. (51). For imaging experiments, one frame was acquired with 100 ms of exposure time (10 Hz) to measure the intensity of fluorescence of the nuclei. For FastSPT experiments, 5000 frames were acquired with an exposure of 10 ms (100 Hz). Acquisition was realized over a field of 50 μm by 80 μm with a depth of 0.2 μm.

SMT analysis: Tracking algorithm. Imaging was performed over a depth of 200 nm; hence, molecules motions are tracked in three dimensions (3D) but projected on a plane. For each individual-imaged nucleus, a movie is generated as a TIF stack, and analyzed by the MATLAB-based SLIMfast script Teves et al. (52), a modified version of MTT Sergé et al. (53). As in Chen et al. (40), we defined a maximal expected diffusion coefficient Dmax of 3 μm2/s−1, defining the maximal distance (dm) between two consecutive frames for a particle to be considered as the same object. For each imaged nucleus, SLIMfast gives an output in the form of a .txt file consisting of a series of successive (x and y) coordinates and times of detection, corresponding to the displacement of each individual molecule in the nucleus. For analysis purposes, the output SLIMfast .txt files was reorganized by a homemade MATLAB script in .csv format with the following data order: iteration (1 − n, with n number of rows of the csv table), frame (the frame on which each single molecule was detected), time (t = frame × exposure), trajectory (ID number of the trajectory), x and y (2D coordinates of the molecule in micrometers). Each imaged individual nucleus, thus, has its corresponding .csv file.

Diffusion coefficients. Diffusion coefficients were measured as in Liu et al. (55). Briefly, the first four points of each T-MSD curve corresponding to each trajectory were fitted with a linear distribution to estimate the diffusion coefficient (Eq. 1, Michalet)MSD=4·D·tlag+offset(1)where D is the diffusion coefficient and tlag is the time between the two positions of the molecule used to calculate the displacement. The offset is due to the limited localization precision inherent to localization-based microscopy methods (~14 nm for our experiments). We set a coefficient of determination R2 ≥ 0.8 to ensure the good quality of the fitting performed to estimate D. Since the distribution of D follows a log-normal distribution Nandi et al. (55), the log10(D) was used for a proper visualization and fitting of the Gaussian bimodal distribution.

Immunoblot. All protein samples were quantified by BCA (bicinchoninic acid assay) assay (Thermo Fisher Scientific, #23227). Protein samples were denatured with 4× NuPAGE LDS Sample Buffer and 10× NuPAGE Sample Reducing Agent (Thermo Fisher Scientific, NP0009) at 95°C for 10 min. Samples were loaded in NuPAGE Novex 4 to 12% bis-tris protein gels (NP0335) and run using NuPAGE Running Buffers (NP0001 and NP0002). Wet transfer to polyvinylidene difluoride membranes at a constant of 300 mA for 3 hours was performed using NuPAGE Transfer Buffer (NP0006). Membranes were blocked for 1 hour in 5% nonfat dairy milk in PBS-T (137 mM NaCl, 2.7 mM KCl, 10 mM Na2HPO4, 1.8 mM KH2PO4, and 0.1% Tween 20). Primary antibodies were diluted in 3% milk/PBS-T at the following concentrations: anti-MYBBP1A (1:300; Sigma-Aldrich, catalog no. HPA005466), anti-PARP1 (1:1000; Active Motif, catalog no. 39559), anti-PAR (1:000; Millipore, catalog no. MABC547), and anti-fibrillarin (1:1000; Abcam, catalog no. ab166630). anti–glyceraldehyde-3-phosphate dehydrogenase (GAPDH) (1:1000; Santa Cruz Biotechnology, sc-365062), and anti–histone H3 (1:3000; Abcam, ab1791), anti-Flag (1:1000; Sigma-Aldrich, catalog no. F3165), anti-myc (1:1000; Cell Signaling Technology, catalog no. 2276), anti-Halo (1:1000; Millipore, catalog no. G9211). Horseradish peroxidase–conjugated secondary antibodies (Santa Cruz Biotechnology, sc-2004 and sc-2005) were diluted 1:10,000 in 3% milk/PBS-T and developed with Super Signal West Pico Chemiluminescent Substrate (Thermo Fisher Scientific, #34080). Blots were visualized with an Amersham Imager 600 (GE Healthcare Life Sciences).

RNA isolation and reverse transcription PCR. Total RNA was isolated from cells with TRIzol (Invitrogen), and reverse transcription was performed using the High-Capacity cDNA Reverse Transcription Kit (Thermo Fisher Scientific, #4368814).

Genomic DNA isolation. Genomic DNA was isolated from cells with GeneJET Genomic DNA Purification Kit (Thermo Fisher Scientific, catalog no. K0721).

Small interfering RNA transfection. HS766T cells were plated at 80% confluency in six-well plates 1 day before transfection. Silencer Select small interfering RNAs (siRNAs) from Thermo Fisher Scientific were used as follows: negative control no. 1 siRNA (catalog no. 4390843) and PARP1 siRNA (catalog no. 4390824). Transfections were performed using 10 nM siRNA with Lipofectamine RNAiMAX (Thermo Fisher Scientific, catalog no. 13778150).

Bacterial recombinant protein. pET/His6/Sumo/MYBBP1A was transformed into Rosetta 2(DE3)pLysS competent cells (Novagen). Recombinant protein was induced with 1 mM isopropyl-β-d-thiogalactopyranoside at 18°C. Bacteria were sonicated with the following settings: an intensity of 7 and cycles of 30-s on /30-s off for three cycles in NPI-10 buffer (50 mM NaH2PO4, 300 mM NaCl, and 10 mM imidazole) with protease inhibitor cocktail and benzonase (Millipore, catalog no. E1014) at 4°C. Insoluble material was removed by centrifugation at 13,000g for 10 min at 4°C. Soluble lysate was then applied to gravity columns packed with Ni–nitrilotriacetic acid agarose resin (Thermo Fisher Scientific, catalog no. R90101) and washed with NPI-20 buffer (50 mM NaH2PO4, 300 mM NaCl, and 20 mM imidazole). Bound proteins were eluted with on-column cleavage with tobacco etch virus (TEV) protease (NEB, catalog no. P8112S). Purified protein was dialyzed using Slide-A-Lyzer Dialysis cassette (Thermo Fisher Scientific, catalog no. 66012) to remove TEV protease against PBS. Concentration and purity of purified protein were determined by electrophoresis using bovine serum albumin (BSA) standards (Pierce) with Coomassie staining. Purified PARP1 protein was purchased from Trevigen (catalog no. 4668-02 K-01).

Nucleosome binding assay. Bacterial purified MYBBP1A protein was immobilized onto anti-Flag magnetic beads (Millipore, catalog no. M8823) and incubated with PARP1 protein in the presence of 20 mM NAD (Trevigen, catalog no. 4684-096-02) and 50 mM tris-HCl (pH 8.0) and 25 mM MgCl2 for 30 min at 4°C. Purified Hela polynucleosomes (EpiCypher, catalog no. 16-0003) were added to the reaction either directly or after washing with high-salt buffer (500 mM NaCl and 50 mM NaH2PO4) three times. Reactions were incubated for 1 hour at 4°C and then washed with high-salt buffer three times and eluted off with anti-Flag Magnetic Beads with 1× NuPAGE LDS Sample Buffer. Nucleosome binding was assessed by immunoblot.

Electromobility shift assay. Purified polynucleosomes (EpiCypher, catalog no. 16-0003) were incubated with recombinant proteins in binding buffer [10 mM tris-HCl (pH 7.5), 1 mM MgCl2, 10 mM ZnCl2,1 mM dithiothreitol, 50 mM KCl, BSA (3 mg/ml), and 5% glycerol] at room temperature for 30 min and separated on 4% nondenaturing polyacrylamide gels run in 0.5× tris-borate-EDTA. Gels were stained with ethidium bromide and visualized under ultraviolet light.

SUPPLEMENTARY MATERIALS

Supplementary material for this article is available at http://advances.sciencemag.org/cgi/content/full/6/49/eabc4517/DC1

https://creativecommons.org/licenses/by-nc/4.0/

This is an open-access article distributed under the terms of the Creative Commons Attribution-NonCommercial license, which permits use, distribution, and reproduction in any medium, so long as the resultant use is not for commercial advantage and provided the original work is properly cited.

REFERENCES AND NOTES

Acknowledgments: We thank J. Mendell for PARP1 deletion constructs. We thank J. Shi and laboratory members for providing Cas9-CRISPR plasmids and assistance with CRISPR mutagenesis. We thank members of the Zaret laboratory for discussions. Funding: A.H. was supported by the American Society of Clinical Oncology Young Investigator Award and the NIH training grant (T32 DK007066). The research was supported by funding from the Institute for Regenerative Medicine, the Abramson Cancer Center, and the Pancreatic Cancer Research Center at UPenn to K.Z. Author contributions: A.H., D.H., and K.Z. conceived the project. A.H., D.H., A.K.R., and K.Z. designed the experiments. A.H., J.L., and J.R.P. performed the experiments. A.H., D.H., J.L., G.D., A.K.R., and K.Z. analyzed data. A.H., D.H., and K.Z. wrote the manuscript with input from all authors. A.H. and K.Z. obtained funding. K.Z. supervised the project. Competing interests: The authors declare that they have no competing interests. Data and materials availability: All data needed to evaluate the conclusions in the paper are present in the paper and/or the Supplementary Materials. Genome sequencing results are available on GEO under accession number GSE149291. Additional data related to this paper may be requested from the authors.
View Abstract

Stay Connected to Science Advances

Navigate This Article