Massively parallel CRISPR off-target detection enables rapid off-target prediction model building.

BACKGROUND: CRISPR (clustered regularly interspaced short palindromic repeats) genome editing holds tremendous potential in clinical translation. However, the off-target effect has always been a major concern. METHODS: Here, we have developed a novel sensitive and specific off-target detection method, AID-seq (adaptor-mediated off-target identification by sequencing), that can comprehensively and faithfully detect the low-frequency off targets generated by different CRISPR nucleases (including Cas9 and Cas12a). FINDINGS: Based on AID-seq, we developed a pooled strategy to simultaneously identify the on/off targets of multiple gRNAs, as well as using mixed human and human papillomavirus (HPV) genomes to screen the most efficient and safe targets from 416 HPV gRNA candidates for antiviral therapy. Moreover, we used the pooled strategy with 2,069 single-guide RNAs (sgRNAs) at a pool size of about 500 to profile the properties of our newly discovered CRISPR, FrCas9. Importantly, we successfully built an off-target detection model using these off-target data via the CRISPR-Net deep learning method (area under the receiver operating characteristic curve [AUROC] = 0.97, area under the precision recall curve [AUPRC] = 0.29). CONCLUSIONS: To our knowledge, AID-seq is the most sensitive and specific in vitro off-target detection method to date. And the pooled AID-seq strategy can be used as a rapid and high-throughput platform to select the best sgRNAs and characterize the properties of new CRISPRs. FUNDING: This work was supported by The National Natural Science Foundation of China (grant nos. 32171465 and 82102392), the General Program of Natural Science Foundation of Guangdong Province of China (grant no. 2021A1515012438), Guangdong Basic and Applied Basic Research Foundation (grant no. 2020A1515110170), and the National Ten Thousand Plan-Young Top Talents of China (grant no. 80000-41180002).

Gene knock-out chain reaction enables high disruption efficiency of HPV18 E6/E7 genes in cervical cancer cells.

A genome editing tool targeting the high-risk human papillomavirus (HPV) oncogene is a promising therapeutic strategy to treat HPV-related cervical cancer. To improve gene knockout efficiency, we developed a gene knockout chain reaction (GKCR) method for continually generating mutagenic disruptions and used this method to disrupt the HPV18 E6 and E7 genes. We verified that the GKCR Cas9/guide RNA (gRNA) cassettes could integrated into the targeted loci via homology-independent targeted insertion (HITI). The qPCR results revealed that the GKCR method enabled a relatively higher Cas9/gRNA cassette insertion rate than a control method (the common CRISPR-Cas9 strategy). Tracking of Indels by DEcomposition (TIDE) assay results showed that the GKCR method produced a significantly higher percentage of insertions or deletions (indels) in the HPV18 E6 and E7 genes. Furthermore, by targeting the HPV18 E6/E7 oncogenes, we found that the GKCR method significantly upregulated the P53/RB proteins and inhibited the proliferation and motility of HeLa cells. The GKCR method significantly improved the gene knockout efficiency of the HPV18 E6/E7 oncogenes, which might provide new insights into treatment of HPV infection and related cervical cancer.

Non-homologous dsODN increases the mutagenic effects of CRISPR-Cas9 to disrupt oncogene E7 in HPV positive cells.

Genome editing tools targeting high-risk human papillomavirus (HPV) oncogene could be a promising therapeutic strategy for the treatment of HPV-related cervical cancer. We aimed to improve the editing efficiency and detect off-target effects concurrently for the clinical translation strategy by using CRISPR-Cas9 system co-transfected with 34nt non-homologous double-stranded oligodeoxynucleotide (dsODN). We firstly tested this strategy on targeting the Green Fluorescent Protein (GFP) gene, of which the expression is easily observed. Our results showed that the GFP+ cells were significantly decreased when using GFP-sgRNAs with dsODN, compared to using GFP-sgRNAs without donors. By PCR and Sanger sequencing, we verified the dsODN integration into the break sites of the GFP gene. And by amplicon sequencing, we observed that the indels% of the targeted site on the GFP gene was increased by using GFP-sgRNAs with dsODN. Next, we went on to target the HPV18 E7 oncogene by using single E7-sgRNA and multiplexed E7-sgRNAs respectively. Whenever using single sgRNA or multiplexed sgRNAs, the mRNA expression of HPV18 E7 oncogene was significantly decreased when adding E7-sgRNAs with dsODN, compared to E7-sgRNAs without donor. And the indels% of the targeted sites on the HPV18 E7 gene was markedly increased by adding dsODN with E7-sgRNAs. Finally, we performed GUIDE-Seq to verify that the integrated dsODN could serve as the marker to detect off-target effects in using single or multiplexed two sgRNAs. And we detected fewer on-target reads and off-target sites in multiplexes compared to the single sgRNAs when targeting the GFP and the HPV18 E7 genes. Together, CRISPR-Cas9 system co-transfected with 34nt dsODN concurrently improved the editing efficiency and monitored off-target effects, which might provide new insights in the treatment of HPV infections and related cervical cancer.

The comparison of ZFNs, TALENs, and SpCas9 by GUIDE-seq in HPV-targeted gene therapy.

Zinc-finger nucleases (ZFNs), transcription activator-like endonucleases (TALENs), and CRISPR-associated Cas9 endonucleases are three major generations of genome editing tools. However, no parallel comparison about the efficiencies and off-target activity of the three nucleases has been reported, which is critical for the final clinical decision. We for the first time developed the genome-wide unbiased identification of double-stranded breaks enabled by sequencing (GUIDE-seq) method in ZFNs and TALENs with novel bioinformatics algorithms to evaluate the off-targets. By targeting human papillomavirus 16 (HPV16), we compared the performance of ZFNs, TALENs, and SpCas9 in vivo. Our data showed that ZFNs with similar targets could generate distinct massive off-targets (287-1,856), and the specificity could be reversely correlated with the counts of middle "G" in zinc finger proteins (ZFPs). We also compared the TALENs with different N-terminal domains (wild-type [WT]/αN/ßN) and G recognition modules (NN/NH) and found the design (αN or NN) to improve the efficiency of TALEN inevitably increased off-targets. Finally, our results showed that SpCas9 was more efficient and specific than ZFNs and TALENs. Specifically, SpCas9 had fewer off-target counts in URR (SpCas9, n = 0; TALEN, n = 1; ZFN, n = 287), E6 (SpCas9, n = 0; TALEN, n = 7), and E7 (SpCas9, n = 4; TALEN, n = 36). Taken together, we suggest that for HPV gene therapies, SpCas9 is a more efficient and safer genome editing tool. Our off-target data could be used to improve the design of ZFNs and TALENs, and the universal in vivo off-target detection pipeline for three generations of artificial nucleases provided useful tools for genome engineering-based gene therapy.

DeepHBV: a deep learning model to predict hepatitis B virus (HBV) integration sites.

BACKGROUND: The hepatitis B virus (HBV) is one of the main causes of viral hepatitis and liver cancer. HBV integration is one of the key steps in the virus-promoted malignant transformation. RESULTS: An attention-based deep learning model, DeepHBV, was developed to predict HBV integration sites. By learning local genomic features automatically, DeepHBV was trained and tested using HBV integration site data from the dsVIS database. Initially, DeepHBV showed an AUROC of 0.6363 and an AUPR of 0.5471 for the dataset. The integration of genomic features of repeat peaks and TCGA Pan-Cancer peaks significantly improved model performance, with AUROCs of 0.8378 and 0.9430 and AUPRs of 0.7535 and 0.9310, respectively. The transcription factor binding sites (TFBS) were significantly enriched near the genomic positions that were considered. The binding sites of the AR-halfsite, Arnt, Atf1, bHLHE40, bHLHE41, BMAL1, CLOCK, c-Myc, COUP-TFII, E2A, EBF1, Erra, and Foxo3 were highlighted by DeepHBV in both the dsVIS and VISDB datasets, revealing a novel integration preference for HBV. CONCLUSIONS: DeepHBV is a useful tool for predicting HBV integration sites, revealing novel insights into HBV integration-related carcinogenesis.

DeepEBV: a deep learning model to predict Epstein-Barr virus (EBV) integration sites.

MOTIVATION: Epstein-Barr virus (EBV) is one of the most prevalent DNA oncogenic viruses. The integration of EBV into the host genome has been reported to play an important role in cancer development. The preference of EBV integration showed strong dependence on the local genomic environment, which enables the prediction of EBV integration sites. RESULTS: An attention-based deep learning model, DeepEBV, was developed to predict EBV integration sites by learning local genomic features automatically. First, DeepEBV was trained and tested using the data from the dsVIS database. The results showed that DeepEBV with EBV integration sequences plus Repeat peaks and 2-fold data augmentation performed the best on the training dataset. Furthermore, the performance of the model was validated in an independent dataset. In addition, the motifs of DNA-binding proteins could influence the selection preference of viral insertional mutagenesis. Furthermore, the results showed that DeepEBV can predict EBV integration hotspot genes accurately. In summary, DeepEBV is a robust, accurate and explainable deep learning model, providing novel insights into EBV integration preferences and mechanisms. AVAILABILITYAND IMPLEMENTATION: DeepEBV is available as open-source software and can be downloaded from https://github.com/JiuxingLiang/DeepEBV.git. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

DeepHPV: a deep learning model to predict human papillomavirus integration sites.

Human papillomavirus (HPV) integrating into human genome is the main cause of cervical carcinogenesis. HPV integration selection preference shows strong dependence on local genomic environment. Due to this theory, it is possible to predict HPV integration sites. However, a published bioinformatic tool is not available to date. Thus, we developed an attention-based deep learning model DeepHPV to predict HPV integration sites by learning environment features automatically. In total, 3608 known HPV integration sites were applied to train the model, and 584 reviewed HPV integration sites were used as the testing dataset. DeepHPV showed an area under the receiver-operating characteristic (AUROC) of 0.6336 and an area under the precision recall (AUPR) of 0.5670. Adding RepeatMasker and TCGA Pan Cancer peaks improved the model performance to 0.8464 and 0.8501 in AUROC and 0.7985 and 0.8106 in AUPR, respectively. Next, we tested these trained models on independent database VISDB and found the model adding TCGA Pan Cancer performed better (AUROC: 0.7175, AUPR: 0.6284) than the model adding RepeatMasker peaks (AUROC: 0.6102, AUPR: 0.5577). Moreover, we introduced attention mechanism in DeepHPV and enriched the transcription factor binding sites including BHLHA15, CHR, COUP-TFII, DMRTA2, E2A, HIC1, INR, NPAS, Nr5a2, RARa, SCL, Snail1, Sox10, Sox3, Sox4, Sox6, STAT6, Tbet, Tbx5, TEAD, Tgif2, ZNF189, ZNF416 near attention intensive sites. Together, DeepHPV is a robust and explainable deep learning model, providing new insights into HPV integration preference and mechanism. Availability: DeepHPV is available as an open-source software and can be downloaded from https://github.com/JiuxingLiang/DeepHPV.git, Contact: huzheng1998@163.com, liangjiuxing@m.scnu.edu.cn, lizheyzy@163.com.

An effective vaginal gel to deliver CRISPR/Cas9 system encapsulated in poly (ß-amino ester) nanoparticles for vaginal gene therapy.

BACKGROUND: Gene therapy has held promises for treating specific genetic diseases. However, the key to clinical application depends on effective gene delivery. METHODS: Using a large animal model, we developed two pharmaceutical formulations for gene delivery in the pigs' vagina, which were made up of poly (ß-amino ester) (PBAE)-plasmid polyplex nanoparticles (NPs) based two gel materials, modified montmorillonite (mMMT) and hectorite (HTT). FINDINGS: By conducting flow cytometry of the cervical cells, we found that PBAE-GFP-NPs-mMMT gel was more efficient than PBAE-GFP-NPs-HTT gel in delivering exogenous DNA intravaginally. Next, we designed specific CRISPR/SpCas9 sgRNAs targeting porcine endogenous retroviruses (PERVs) and evaluated the genome editing efficacy in vivo. We discovered that PERV copy number in vaginal epithelium could be significantly reduced by the local delivery of the PBAE-SpCas9/sgRNA NPs-mMMT gel. Comparable genome editing results were also obtained by high-fidelity version of SpCas9, SpCas9-HF1 and eSpCas9, in the mMMT gel. Further, we confirmed that the expression of topically delivered SpCas9 was limited to the vagina/cervix and did not diffuse to nearby organs, which was relatively safe with low toxicity. INTERPRETATION: Our data suggested that the PBAE-NPs mMMT vaginal gel is an effective preparation for local gene therapy, yielding insights into novel therapeutic approaches to sexually transmitted disease in the genital tract. FUNDING: This work was supported by the National Science and Technology Major Project of the Ministry of science and technology of China (No. 2018ZX10301402); the National Natural Science Foundation of China (81761148025, 81871473 and 81402158); Guangzhou Science and Technology Programme (No. 201704020093); National Ten Thousand Plan-Young Top Talents of China, Fundamental Research Funds for the Central Universities (17ykzd15 and 19ykyjs07); Three Big Constructions-Supercomputing Application Cultivation Projects sponsored by National Supercomputer Center In Guangzhou; the National Research FFoundation (NRF) South Africa under BRICS Multilateral Joint Call for Proposals; grant 17-54-80078 from the Russian Foundation for Basic Research.

Risk stratification of cervical lesions using capture sequencing and machine learning method based on HPV and human integrated genomic profiles.

From initial human papillomavirus (HPV) infection and precursor stages, the development of cervical cancer takes decades. High-sensitivity HPV DNA testing is currently recommended as primary screening method for cervical cancer, whereas better triage methodologies are encouraged to provide accurate risk management for HPV-positive women. Given that virus-driven genomic variation accumulates during cervical carcinogenesis, we designed a 39 Mb custom capture panel targeting 17 HPV types and 522 mutant genes related to cervical cancer. Using capture-based next-generation sequencing, HPV integration status, somatic mutation and copy number variation were analyzed on 34 paired samples, including 10 cases of HPV infection (HPV+), 10 cases of cervical intraepithelial neoplasia (CIN) grade and 14 cases of CIN2+ (CIN2: n = 1; CIN2-3: n = 3; CIN3: n = 9; squamous cell carcinoma: n = 1). Finally, the machine learning algorithm (Random Forest) was applied to build the risk stratification model for cervical precursor lesions based on CIN2+ enriched biomarkers. Generally, HPV integration events (11 in HPV+, 25 in CIN1 and 56 in CIN2+), non-synonymous mutations (2 in CIN1, 12 in CIN2+) and copy number variations (19.1 in HPV+, 29.4 in CIN1 and 127 in CIN2+) increased from HPV+ to CIN2+. Interestingly, 'common' deletion of mitochondrial chromosome was significantly observed in CIN2+ (P = 0.009). Together, CIN2+ enriched biomarkers, classified as HPV information, mutation, amplification, deletion and mitochondrial change, successfully predicted CIN2+ with average accuracy probability score of 0.814, and amplification and deletion ranked as the most important features. Our custom capture sequencing combined with machine learning method effectively stratified the risk of cervical lesions and provided valuable integrated triage strategies.