Background
Genome-wide association studies (GWAS) have identified thousands of genetic variants associated with normal and disease traits. Most variants are located in noncoding regions of the genome and do not directly affect protein-coding sequences. A well-established mechanism by which GWAS variants modulate disease risk is through the alteration of DNA enhancers, causing changes in the expression of nearby target genes [
1,
2]. In addition to housing DNA regulatory elements, the human noncoding genome is pervasively transcribed and at least a subset of the resulting molecules are functional at the transcript level. The majority of the transcripts are long noncoding RNAs (lncRNAs), defined as RNA transcripts longer than 200 nucleotides that do not code for proteins.
Despite lacking the ability to code for proteins, lncRNAs perform a diverse array of cellular functions [
3]. In the nucleus, they can alter transcription by guiding epigenetic modifications and/or transcription factors, regulate splicing by binding splicing factors, act as scaffolds for protein complexes and promote the formation of nuclear bodies and domains [
4]. In the cytoplasm, lncRNAs have been shown to regulate RNA stability and translation, interact with proteins to affect their localization, stability, and post-translational modifications and influence cellular export and signaling pathways, among other described functions [
5]. LncRNAs display exquisite cell-type specific expression and are important in defining specific cell subpopulations and cell states [
6]. The aberrant expression of lncRNAs has been reported in various disease phenotypes, including cancer, and many have been directly implicated in disease development [
3]. However, the impact of disease risk-associated variants on lncRNA expression and function is less evident and requires further study.
We recently discovered thousands of lncRNAs transcribed from breast cancer GWAS loci and nearby regions [
7] (< 1.5 Mb). An enrichment of GWAS variants was observed in lncRNA exons but not in their introns or promoter regions, suggesting that lncRNA transcripts are important mediators of breast cancer risk. We identified 844 lncRNAs as potential GWAS target genes based on the presence of breast cancer risk variants in their exons, promoters or distal DNA regulatory elements [
7]. Expression quantitative trait loci (eQTL) analyses identified lncRNAs whose expression are associated with risk variants in breast tumors, providing additional evidence for their involvement in breast cancer development [
7]. From our findings, we expected that some of the identified lncRNAs would influence breast cell proliferation.
High-throughput pooled loss-of-function screens are a powerful strategy for identifying genes implicated in different phenotypes. CRISPR-Cas9 cutting (CRISPRko), targets DNA regions in the genome and is the most used strategy for protein-coding gene knockouts. CRISPRko will often be ineffective for lncRNAs as the cutting may not alter lncRNA stability or function. Several CRISPR-dCas9-based activation (CRISPRa) and inhibition (CRISPRi) screens have successfully been used to overexpress and knockdown mRNA and lncRNAs [
8‐
11]. However, given that lncRNA transcription is often initiated from enhancer elements encoded in DNA, it is not clear whether the observed CRISPRi/a effect is DNA or RNA mediated. To overcome these hurdles, we performed CRISPR-Cas13d RNA knockdown screens to identify the breast cancer-associated lncRNAs whose knockdown affects proliferation of normal breast and breast cancer cells.
Discussion
Although the number of lncRNAs has surpassed protein-coding genes, it is still unclear what proportion of lncRNAs are functional as opposed to transcriptional noise. High-throughput pooled CRISPR screens provide an unbiased method of identifying protein-coding and noncoding genes that function in different biological processes. CRISPR-Cas9 screens are commonly used to assess protein-coding genes for function, however they are often ineffective for lncRNAs as it is difficult to predict the impact of a Cas9-induced indel on lncRNA function. CRISPRi screens have successfully been used to identify functional lncRNAs [
8,
27,
28], but as they are often transcribed from enhancers, the observed phenotype can be a consequence of CRISPRi-induced enhancer suppression. Additional experiments are required to decipher if the phenotype is mediated by DNA or RNA.
Here, we describe the use of Cas13d-mediated RNA knockdown screens to identify breast cancer-associated lncRNAs that modulate proliferation in normal breast and breast cancer cells. Cas13d has previously been used to screen circular RNAs (circRNAs), with crRNAs designed to target their back-splicing junction allowing the discrimination of circRNAs from their host mRNA [
29]. Several circRNAs were identified as important mediators of cell growth, including
circFAM120A which was shown to promote cell proliferation by preventing the translation inhibitor IGF2BP2 from binding its host mRNA,
FAM120A (and other family members [
29]). Pooled Cas13d screens have also been used to optimize crRNA design. For example, fluorescent sorting for the cell surface markers CD46, CD55 and CD71 were used to screen for the best crRNA sequences for mRNA knockdown [
30]. Using knowledge gained from these studies, Wessels et al. [
30] developed a computational pipeline for crRNA design (
cas13design), which we utilized in this study. Based on
cas13design results, we selected the top ten non-overlapping crRNAs with the highest predicted quality for each lncRNA in the screen.
Off-targeting effects are one of the major limitations with CRISPR-based technologies and one that is often overlooked. To increase the likelihood of obtaining
bona fide hits, we removed crRNAs with complementarity to genomic regions which were not the intended target. Optimally, this filtering step should be performed as part of the crRNA design prior to library synthesis. A limitation of using Cas13d in CRISPR screens is its reported collateral activity, where in addition to specifically cleaving the target RNA, it also promiscuously cleaves bystander RNAs [
31,
32]. To mitigate the consequences of this collateral activity, we individually validated the prioritized proliferation-related lncRNAs identified in our screens using multiple methods, prior to any follow-up characterization. Recently, a high-fidelity Cas13 enzyme was engineered (hfCas13d) which potentially minimizes collateral degradation of bystander RNA [
33]. We anticipate that future CRISPR screens will benefit from the improved enzyme, representing an important advancement to the field.
The majority of GWAS variants are located in noncoding regions, frequently at lncRNA exons [
7], but there is limited functional evidence implicating lncRNAs in disease risk. In this study, we identified 43 lncRNAs (39 unannotated [
7] and four annotated) whose knockdown modulated breast cell proliferation, a fundamental trait of cancer cells. One the unannotated lncRNAs,
KILR is transcribed from an intron of
KCTD1-5, using an alternative promoter of
KCTD1. As
KILR is polyadenylated, it is likely to be a product of alternative polyadenylation rather than recursive mRNA splicing. Similarly, the start of
KILR is ~ 105 kb from the
KCTD1-5 TSS, indicating that its 5’ end is post-transcriptionally processed and stabilized. Our RNA folding predictions suggest that
KILR possesses complex secondary structures at both terminal ends, which could explain how the transcript is protected from exonucleases. As
KILR also has three predicted snoRNAs within its sequence, it is possibly a SPA-lncRNA (5’ small nucleolar RNA capped and 3′ polyadenylated), where the 5’ end is stabilized by a snoRNA structure rather than an m7G cap [
34].
The
KILR breast cancer risk signal at 18q11 is colocalized with the genetic signal of the eQTL, suggesting that the risk variants can function by modulating
KILR expression. In support of this, we show that the half-life of
KILR in the presence of the risk alleles is significantly reduced as compared to that of
KILR with the protective alleles. It is likely that one or more of the risk variants disrupts
KILR secondary structure or affects the binding of a protein(s) responsible for maintaining
KILR stability. This is the first time that GWAS variants have been shown to act by directly altering the RNA stability of a lncRNA transcript reducing its expression. Mechanistically, we showed that reduced
KILR expression promoted breast cancer cell proliferation by increasing DNA replication fork speed.
KILR binds to and sequesters RPA1, suggesting that its reduced expression would increase the available pool of RPA1. Previous studies have shown that the RPA complex participates in the initiation and elongation steps of DNA replication [
35,
36] and that increased levels of RPA1 accelerate DNA replication and therefore promote cell proliferation [
37].
Overexpression of
KILR mimics the reported effects of RPA1 knockdown on cancer cell growth [
38]. In line with this, RPA1 deficiency has been shown to cause spontaneous DSBs and apoptosis [
39]. Breast cancer cells partially depleted of RPA1 by siRNA treatment also become over-sensitive to DNA damage [
40]. Indeed, relative expression of RPA is a predictor of response to chemotherapy in many cancers [
41]. In breast cancer, RPA has also been linked with tumor aggressiveness and a decrease in overall survival [
40]. Attempts to inhibit the RPA complex with synthetic molecules have resulted in cell death via apoptosis and has been established as a novel class of broad spectrum anticancer agents (RPAis [
41]). The most promising first generation RPAi (TDRL-551) increases the efficacy of platinum-based chemotherapy in ovarian cancer [
42]. Although this class of drugs was successful in preclinical studies, the RPAis explored so far presented chemical liabilities that could hinder their clinical use. RNA-focused therapy that interferes with cell proliferation and apoptosis has been cited as a promising avenue for cancer treatment [
43]. We suggest that
KILR could be used as an endogenous replacement of chemical RPAis or in combination therapy, if second generation synthetic RPAis prove to be safe.
We showed that overexpression of
KILR in normal breast cells did not induce apoptosis. Normal cells are not subject to replication stress, which is often detected in highly-proliferative cancer cells and thus are less dependent on RPA availability. We observed a pool of free RPA1 in the normal breast cells which we hypothesized was sufficient to maintain DNA replication even after
KILR overexpression. This idea is supported by the fact that after IR exposure, normal breast cells overexpressing
KILR form RPA1 puncta independent of
KILR, suggesting that free RPA1 molecules can aggregate (likely at DSBs) in response to DNA damage. In cancer cells, the combination of replication stress and defective DNA damage repair results in replication catastrophe and cell death. RPA is critical to prevent this from happening as the exhaustion of free RPA1 leads to the accumulation of unprotected ssDNA and subsequent DSBs [
44]. Further understanding the different outcomes of altering
KILR expression in normal breast versus cancer cells will be important to determine the clinical relevance of
KILR.
Methods
Cell lines and culture
MCF7, MDAMB231, T47D, MCF10A, Hs578T and HEK293 cell lines were obtained from ATCC and grown according to their guidelines. MDAMB361 cells (ATCC) were grown in DMEM (Gibco Invitrogen) with 20% fetal bovine serum (FBS; Gibco Invitrogen) and 1% antibiotics (Gibco Invitrogen). B80T5 cells (a gift from Roger Reddel; CMRI, Australia) were grown in RPMI (Gibco Invitrogen) with 10% FBS and 1% antibiotics. K5+/K19 + cells [
12] were grown in 1:1 MEM α (Gibco Invitrogen) and Ham’s F-12 Nutrient Mix (Gibco Invitrogen) with 1% FBS, 10mM HEPES, 1 µg/ml bovine pancreatic insulin, 1 µg/ml hydrocortisone, 50 µg/ml epidermal growth factor (Sigma Aldrich), 10 mg/ml transferrin, 100 µM β-estradiol, 2 mM glutamine, 2.6 ng/ml sodium selenite, 1 ng/ml cholera toxin (Sigma Aldrich), 6.5 ng/ml triiodothyronine, 100 µM ethanolamine, 35 µg/ml BPE, 10 µg/ml gentamicin, 10 µg/ml ascorbic acid, 15 µg/ml hygromycin B. All cell lines were tested for mycoplasma contamination and verified by short tandem repeats (STR) profiling.
Plasmid constructs
To generate a Cas13d-NLS expression vector (pLXTRC311/NLS-EF1a-RxCas13d-2 A-EGFP-blast; abbreviated to Cas13d), the Cas13d-NLS cassette was PCR-amplified from the pXR001_EF1a-CasRx-2 A-EGFP (Addgene #109,049) plasmid and cloned into pLX_TRC311-NLS-Cas13b-NES-P2A-Blast-eGFP. CRISPR-Cas13d crRNAs were cloned into BsmBI-digested pLentiRNAGuide_001 vector (Addgene #138,150) and CRISPRa gRNAs into BsmBI-digested pXPR502 vector (Addgene #96,923). For overexpression, full-length KILR was amplified from T47D cDNA using the KAPA HiFi PCR Kit (Kapa Biosystems) and KCTD1-5 cDNA was synthesized by IDT. PCR products were cloned into the doxycycline-inducible plasmid pCW57-MCS1-2 A-MCS2 (Addgene #71,782), which was modified by adding bGHpolyA between the MluI and BamHI restriction sites. RPA1 cDNA was amplified from pCMV-N-FLAG-RPA1 (Sinobiological, HG15561-NF) and cloned into pLX-TRC311-Blasticidin. For pull-down assays, full-length KILR cDNA was cloned into pGEM-T (Promega). All constructs were confirmed by Sanger sequencing at the Australian Genome Research Facility (AGRF). The primers, crRNAs and gRNAs sequences are provided in Supplementary Table 4.
Generation of stable cell lines
Lentiviral plasmids were co-transfected with VSV-G envelope plasmid, pMD2.G (Addgene #12,259) and packaging plasmid psPAX2 (Addgene #12,260) into HEK293 using FuGENE 6 transfection reagent (Promega). Culture supernatant containing lentiviral particles was harvested after 24–48 h incubation and passed through a 0.45 μm filter. Lentivirus was concentrated by centrifuging at 10,000 rpm at 4 °C for 16-24 h, resuspended in RPMI 1640 medium with 10% FBS, aliquoted and stored at -80 °C. Breast cells were transduced at a high multiplicity of infection (MOI) with either Cas13d or CRISPRa (dCas9-VP64; Addgene #61,425) lentivirus by spinoculation at 2,500 rpm for 1.5 h at room temperature. To increase transduction efficiency, 5–8 µg/ml of polybrene (Sigma Aldrich) was supplemented in the media. Forty-eight hours post-transduction, cells were stabilized with 10–15 µg/ml blasticidin (Thermo Fisher Scientific) for two weeks and then maintained at 5–10 µg/ml blasticidin. Cas13d-expressing cells with high GFP were further purified by fluorescent activated cell sorting (FACS; FACSAria™ III Cell Sorter; BD Biosciences).
CRISPR-Cas13 guide RNA (crRNA) library design
Cas13 crRNAs were designed using the basic algorithm in the
cas13design tool [
30] (
https://cas13design.nygenome.org) and further filtered to improve library quality. The following steps were followed: (1) Only the pool of high-quality guides (top quartile of quality scores) was considered for further analysis, unless step 4 is activated (2). From the high-quality guides, we selected those with no overlap, to increase gene coverage (3). According to the quality scores provided by
cas13design, the top ten guides that meet the above criteria were selected per transcript (4). For transcripts with less than five guides after filtering, we relaxed some of the criteria (e.g. allowing guides with quality scores in the third quartile or with 5–10 nucleotides overlap with each other, in this order) (5). We then re-run steps 2–3 for this subset and re-enter step 4 if necessary (6). When all transcripts have 5–10 guides that pass the quality filtering described above, we stopped reiterating (7). We removed redundancy in the library and added the required flanking sequences before sending the library to be synthesised (see below) [
8]. Blast alignments were used to remove guides with off-targets to either the reference human genome (hg38) or transcriptome (Gencode v.36). All crRNAs matching any region outside the target gene with up to two mismatches were considered as off-targets and removed from the
in silico library.
crRNA library generation
The oligonucleotides for the crRNA library were synthesized by Genscript. The sequences were collectively amplified with primers that generated 40 bp homologies with the pLentiRNAGuide_001 vector digested with BsmBI and XhoI. PCR was performed using Q5 High-Fidelity DNA Polymerase (New England Biolabs; NEB) for 20 cycles. The amplified crRNA library was then gel purified and assembled into BsmBI/Xhol-digested pLentiRNAGuide_001 using NEBuilder HiFi DNA Assembly master mix (NEB). The assembled plasmids were purified and concentrated by isopropanol precipitation. Three hundred nanograms of purified plasmids were electroporated into 25 µl of Endura electrocompetent cells (Lucigen) according to the manufacturer’s instructions. The electroporated cells were recovered in recovery medium (Lucigen) for 1 h and then plated on Terrific Broth (TB) agar plates with 100 µg/ml ampicillin at 37 °C for 16 h. The resulting colonies were scraped and harvested in bulk at a coverage of more than 500 colonies per crRNA. The library plasmids were extracted using the NucleoBond Xtra Maxi EF Kit (Macherey-Nagel) to avoid endotoxin contamination. Library quality was assessed by next-generation sequencing.
Pooled CRISPR-Cas13d proliferation screens
K5+/K19+, MCF7 and MDAMB231 cells stably expressing Cas13d were transduced with the crRNA library at an MOI of 0.3 to obtain 1000 cells/crRNA (three biological replicates per cell line). Twenty-four hours post-infection, cells were selected using 1–2 µg/ml puromycin and then maintained with 1–2 µg/ml puromycin and 10 µg/ml blasticidin throughout the screen to ensure crRNA and Cas13d expression. At 21 days post-infection, gDNA was extracted from the cells using the Quick-DNA Midiprep Plus Kit (Zymo Research), and one-step PCR was performed to amplify and add barcodes to the integrated crRNA sequences. PCR products were gel purified and sequenced by next-generation sequencing (20 M reads/replicate). Quality control using FastQC v.0.11.8 (
https://www.bioinformatics.babraham.ac.uk/projects/fastqc) was performed on the sequenced libraries and abundance estimation of all crRNAs using BBduk v.2019 (
https://sourceforge.net/projects/bbmap) on Java v.1.8.192. Read counts were obtained for all crRNAs using MAGeCK v.0.5.9.4 run on Python v.3.6.1 and hits were called using MAGeCK test. A false discovery rate (FDR) threshold of 0.3 was applied to recover true hits in every cell line.
Quantitative real-time PCR (qPCR)
Total RNA was extracted using TRIzol (Thermo Fisher Scientific). Complementary DNA (cDNA) was synthesized from RNA samples using SuperScript IV (Thermo Fisher Scientific). qPCR was performed using TaqMan assays (Thermo Fisher Scientific) or Syto9 incorporation into PCR-amplified products. Primers are listed in Supplementary Table 4.
Cell proliferation assays
Cell proliferation was monitored using the IncuCyte live cell imaging system (Essen Bioscience). Cells were seeded at 2-3 × 104 cells per well in 24-well plates and imaged using a 10x objective lens every 3 h over 4–7 days. Imaging was performed in an incubator maintained at 37 °C under a 5% CO2 atmosphere. Cell confluence in each well was measured using IncuCyte ZOOM 2016 A software and the data analyzed using GraphPad Prism.
Random amplification of cDNA ends (RACE)
5’ and 3’ RACE was performed using the GeneRacer kit (Thermo Fisher Scientific), following the manufacturer’s protocol. The purified PCR products were cloned into the pCR4-TOPO TA vector (Thermo Fisher Scientific) and identified by Sanger sequencing. Primers are listed in Supplementary Table 4.
KILR secondary structure prediction and motif annotation
RNAfold, part of the Vienna package v.2.0 [
45] was used for secondary structure predictions based on the
KILR RNA transcript sequence (Supplementary File 1). The minimum free energy structure, based on the Turner model of 2004, was considered representative of
KILR. The modeling temperature was defined as 37
oC and isolated base pairs were avoided. The ALU elements that form the IRAlu structure of
KILR were characterized based on Dfam v.3.6 [
46] predictions. The machine learning algorithm implemented in snoReport v.2.0 [
47] was used to identify snoRNA-like sequences in
KILR. Other motifs such as the SIRLOIN nuclear localization sequence [
19] and BORG-like motifs were sourced from the literature.
RNA stability assays
MDAMB361 cells were treated with 10 µg/ml actinomycin D (Sigma-Aldrich) to block transcription then harvested at 0, 3, 4, 8 and 12 h post-treatment. qPCR was performed using a TaqMan™ Genotyping Assay (rs4555225 G/C; Thermo Fisher Scientific). A cyclin-dependent kinase inhibitor 2 A (CDKN2A) TaqMan probe (Thermo Fisher Scientific) was used as an internal control. Linear regression analysis (GraphPad Prism) was used to estimate the decay rate of KILR with or without the risk alleles. The half-life was calculated by the equation t1/2 = ln(2)/kdecay.
Estrogen induction
Cells were treated with 10 nM fulvestrant (Sigma-Aldrich) for 48 h before the media was removed and replaced with media containing either 10 nM 17β-estradiol (Sigma-Aldrich) or DMSO (as vehicle control) for 24 h. Cells were harvested with TRIzol and assessed for induction of gene expression by qPCR.
Cell fractionation
T47D cells were first separated into nuclear and cytoplasmic fractions using hypotonic lysis buffer (HLB, 10 mM Tris, 10 mM NaCl, 3 mM MgCl2, 0.3% NP-40, and 10% glycerol, pH 7.5) and centrifuging at 1000 g for 3 min. RNA in the cytoplasmic fraction was precipitated using RNA precipitation solution (RPS, 0.15 M sodium acetate and 95% ethanol). For the precipitated nuclear fraction, Modified Wuarin-Schibler buffer (MWS, 10 mM Tris-HCl, 4 mM EDTA, 0.3 M NaCl, 1 M urea, 1% NP-40) was added to further separate the RNA into the chromatin and nucleoplasmic fractions by spinning at 1000 g for 3 min. The supernatant containing the nucleoplasmic RNA was then precipitated by RPS. The remaining pellet was the chromatin fraction. RNA from the three different cellular fractions was extracted using TRIzol. qPCR was performed to detect RNA in each fraction, with RSP14 (Ribosomal protein S14), U2snRNA (U2 spliceosomal RNA) and NEAT1 (nuclear enriched abundant transcript 1) serving as positive controls for RNA fractionated into the cytoplasmic, nuclear and chromatin compartments, respectively. Primers are listed in Supplementary Table 4.
Apoptosis assays
For CRISPRa, cells were transduced with gRNAs and selected with 3 µg/ml puromycin and 10 µg/ml blasticidin for three days. After culturing for further 3–5 days, cells were then trypsinized, fixed and immunostained with the Alexa Fluor 488 Annexin V/Dead Cell Apoptosis Kit (Thermo Fisher Scientific), according to the manufacturer’s protocol. For Tet-On overexpression, cells were transduced with lentivirus at a low MOI. Twenty-four hours post-transduction, the cells were treated with 1–3 µg/ml of puromycin for 4 days. After induction by 1 µg/ml doxycycline hyclate (Sigma-Aldrich) for 3 days, cells were trypsinized, fixed and immunostained with the apoptosis kit according to the manufacturer’s protocol. The percentage of apoptotic cells was assessed by FACS.
RNA in vitro transcription and RNA-protein pull-down
The pGEMT-KILR construct was linearized with NotI then in vitro transcribed using the HiScribe T7 Quick High Yield RNA synthesis kit (NEB) according to the manufacturer’s instructions. LacZ RNA, produced from NotI-linearized pEF-ENTR-LacZ (Addgene #17,430), was used as a negative control. RNA pull down was performed using the Pierce Magnetic RNA-Protein Pull-Down Kit (Thermo Fisher Scientific). Briefly, the in vitro transcribed RNAs were purified by TRIzol extraction, labeled with biotinylated cytidine bisphosphate, and incubated with cell lysates. After overnight incubation at 16 °C, the RNA-protein complexes were captured with streptavidin beads and proteins were identified by mass spectrometry.
Mass spectrometry
Samples underwent on-bead processing with 5 mM DTT at 60 °C for 30 min then alkylated with 20 mM IAA for 10 min at room temperature in the dark. Proteins were digested with trypsin overnight at 37 °C, then centrifuged at 20,000xg for 10 min to pellet the beads. The supernatants were acidified with trifluoroacetic acid, dried on a Speedvac, then reconstituted in 0.1% formic acid (FA) for LCMS analysis. Samples were loaded onto a Thermo Acclaim PepMap 100 trap column for 5 min at a flow rate of 10 µl/min with 95% Solvent A (0.1% FA in water) and separated on a Thermo PepMap100 analytical column equipped on a Thermo Ultimate 3000 LC interfaced with Thermo Exactive HF-X mass spectrometer. Peptides were resolved using a linear gradient of 5% solvent B (0.1% FA in 80% acetonitrile) to 40% solvent B over 48 min at a flow rate of 1.5 µl/min, followed by column washing and equilibration for a total run time of 65 min. Mass spectrometry data was acquired in positive ion mode. Precursor spectra (350–1400 m/z) were acquired on orbitrap at a resolution of 60,000. The AGC target was set to 3E6 with a maximum ion injection time of 30 ms. Top 20 precursors were selected for fragmentation in each cycle and fragment spectra were acquired in orbitrap at a resolution of 15,000 with stepped collision energies of 28, 30 and 32. The AGC target was 1E5, with a maximum ion injection time of 45 ms. The isolation window was set to 1.2 m/z. Dynamic exclusion was set to 30 s and precursors with charge states from 2 to 7 were selected for fragmentation. MS/MS data were searched against the reviewed Uniprot human database using Sequest HT on the Thermo Proteome Discoverer software (v.2.2). An FDR of 1% was used to filter peptide spectrum matches (PSMs). Carbamidomethylation of cysteines was set as a fixed modification, while oxidation of methionine, deamidation of glutamine and asparagine were set as dynamic modifications. Protein abundance was based on intensity of the parent ions and data were normalized based on total peptide amount. Five biological replicates were independently analysed for statistical significance, calculated using a t-test for summed abundance based ratios. Only proteins with at least five identified peptides, log2 [fold-change] (over LacZ) > 2.0 and p-value < 0.05 were considered. The resulting metrics were combined and the fold-change averaged across the replicates to obtain the final ranking of KILR protein partners.
RNA immunoprecipitation (RIP)
MCF7 cells were cross-linked with 1% formaldehyde at 37oC for 10 min, quenched with 2 M glycine and centrifuged for 2 min at 100 g. Cell pellets were resuspended in lysis buffer (1.28 M sucrose, 40 mM Tris-HCl pH 7.5, 20 mM MgCl2, 4% Triton X-100, 200U RNase inhibitor, protease inhibitor cocktail), sonicated ten times for 10 s at 70% duty cycle (Branson SLPt) and clarified by centrifugation. For IP, protein A Dynabeads beads (Thermo Fisher Scientific) were pre-bound with IgG (Cell Signaling; 2729) or anti-RPA1/RPA70 antibody (Abcam; ab79398) at 4oC for 4 h, then incubated with lysates at 4oC overnight. The magnetic bead-protein/RNA complexes were collected and washed five times (50 mM HEPES pH 7.5, 0.4 M NaCl, 1 mM EDTA, 1 mM DTT, 1% Triton X-100, 10% glycerol, 200U RNase inhibitor) and RNA was recovered by TRIzol (Thermo Fisher Scientific) extraction and DNase treatment (NEB).
RNA-fluorescence in situ hybridization (FISH) and immunofluorescence (IF)
For RNA FISH, 5 days post-transduction, CRISPRa and Tet-on KILR overexpressing cells grown on coverslips were fixed in 4% formaldehyde for 10 min followed by permeabilization in 70% ethanol overnight at 4 °C. Cells were then stained for 16 h with 125 nM of a custom KILR Stellaris RNA-FISH probe set labelled with Quasar 570 fluorophore (LGC Biosearch Technologies) according to the manufacturer’s instructions. For IF, CRISPRa, CRISPR-Cas13, Tet-On and RPA1-KILR overexpressing cells were challenged with or without 6 Gy gamma irradiation followed by 6 h of incubation. The cells were then treated with CSK buffer (10 mM PIPES, 100 mM NaCl, 300 mM sucrose, 3 mM MgCl2, 1.4% Triton X-100) to remove the cytoplasm, followed by fixation in 4% formaldehyde for 10 min and permeabilization with 0.5% Triton X-100 for 15 min. The cells were incubated with antibodies against RPA1/RPA70 (Abcam, ab79398, 1:250), γH2AX (Abcam, ab2893, 1:1000) or RAD51 (GeneTex, GTX70230, 1:500). Coverslips were mounted onto slides using ProLong Glass antifade medium containing NucBlue nuclear counterstain (Thermo Fisher Scientific). Images were acquired with the DeltaVision Deconvolution microscope (GE Healthcare) using a 60x objective lens and analyzed with ImageJ software. A minimum of 100 cells per sample were analyzed.
DNA fiber assays
CRISPR-Cas13d or Tet-On overexpressing cells were sequentially pulse-labeled with 50 µM 5-chloro-2′-deoxyuridine (CldU, Sigma-Aldrich) and 250 µM 5-iodo-2′-deoxyuridine (IdU, Sigma-Aldrich) for 25 min each, followed by treatment with or without 4 mM hydroxyurea (Sigma-Aldrich) for 4 h. Labeled cells were then washed and harvested in phosphate-buffered saline. Cell lysis, DNA spreading, denaturation and immunostaining were performed as described previously [
48]. The slides were stained overnight at 4 °C with anti-BrdU (Abcam, ab6326, 1:300) for CldU tracks and anti-BrdU (BD Biosciences, 347,583, 1:50) to detect IdU tracks. After washing three times, the slides were stained for 1 h at 37 °C with Alexa Fluor 488-labeled chicken anti-rat IgG (Invitrogen, A21470, 1:300) and Alexa Fluor 546-labeled goat anti-mouse IgG (Invitrogen, A11030, 1:300) secondary antibodies, respectively. Slides were visualized using the DeltaVision Deconvolution microscope with a 40x objective lens and analyzed with ImageJ software. A minimum of 150 fibers per sample were analyzed.
MCF7 cells (3 × 103) were seeded in 6-well plates followed by the treatment with vehicle, cisplatin (0–30 µM) or olaparib (0–25 µM) for 14 days. The colonies were fixed with 0.05% crystal violet for 30 min. The quantification of crystal violet intensity was measured after destaining colonies by Sorenson’s buffer (0.1 M sodium citrate in 50% ethanol, pH 4.2) by a PowerWave HT Microplate Spectrophometer (BioTek, USA) at 590 nM absorbance.
Western blotting
Cell pellets were lysed in RIPA buffer (50 mM Tris-HCl pH 8.0, 150 mM NaCl, 1% IGEPAL CA-630, 0.5% sodium deoxycholate, 0.1% sodium dodecyl sulfate (SDS), 1 mM 1,4-dithiothreitol (DTT), protease inhibitor cocktail) and clarified by centrifugation to remove cell debris. Proteins were separated by SDS-polyacrylamide gel electrophoresis, electroblotted onto PVDF membranes by semi-dry transfer (Bio-Rad) and blocked in 1% casein blocking buffer (Bio-Rad). Antibodies detecting RPA1/RPA70 (Abcam, ab79398, 1:1000) or Actin (Cell Signaling; 1:20,000) were incubated overnight. For detection, horseradish peroxidase (HRP)-coupled secondary antibodies were used (Cell Signaling). Detected proteins were visualized with enhanced chemiluminescence substrate (Pierce) and the iBright gel documentation system (Thermo Fisher Scientific).
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.