Deep learning models to predict the editing efficiencies and outcomes of diverse base editors.

Applications of base editing are frequently restricted by the requirement for a protospacer adjacent motif (PAM), and selecting the optimal base editor (BE) and single-guide RNA pair (sgRNA) for a given target can be difficult. To select for BEs and sgRNAs without extensive experimental work, we systematically compared the editing windows, outcomes and preferred motifs for seven BEs, including two cytosine BEs, two adenine BEs and three C•G to G•C BEs at thousands of target sequences. We also evaluated nine Cas9 variants that recognize different PAM sequences and developed a deep learning model, DeepCas9variants, for predicting which variants function most efficiently at sites with a given target sequence. We then develop a computational model, DeepBE, that predicts editing efficiencies and outcomes of 63 BEs that were generated by incorporating nine Cas9 variants as nickase domains into the seven BE variants. The predicted median efficiencies of BEs with DeepBE-based design were 2.9- to 20-fold higher than those of rationally designed SpCas9-containing BEs.

Massively parallel evaluation and computational prediction of the activities and specificities of 17 small Cas9s.

Recently, various small Cas9 orthologs and variants have been reported for use in in vivo delivery applications. Although small Cas9s are particularly suited for this purpose, selecting the most optimal small Cas9 for use at a specific target sequence continues to be challenging. Here, to this end, we have systematically compared the activities of 17 small Cas9s for thousands of target sequences. For each small Cas9, we have characterized the protospacer adjacent motif and determined optimal single guide RNA expression formats and scaffold sequence. High-throughput comparative analyses revealed distinct high- and low-activity groups of small Cas9s. We also developed DeepSmallCas9, a set of computational models predicting the activities of the small Cas9s at matched and mismatched target sequences. Together, this analysis and these computational models provide a useful guide for researchers to select the most suitable small Cas9 for specific applications.

Generation of a more efficient prime editor 2 by addition of the Rad51 DNA-binding domain.

Although prime editing is a promising genome editing method, the efficiency of prime editor 2 (PE2) is often insufficient. Here we generate a more efficient variant of PE2, named hyPE2, by adding the Rad51 DNA-binding domain. When tested at endogenous sites, hyPE2 shows a median of 1.5- or 1.4- fold (range, 0.99- to 2.6-fold) higher efficiencies than PE2; furthermore, at sites where PE2-induced prime editing is very inefficient (efficiency < 1%), hyPE2 enables prime editing with efficiencies ranging from 1.1% to 2.9% at up to 34% of target sequences, potentially facilitating prime editing applications.

Sequence-specific prediction of the efficiencies of adenine and cytosine base editors.

Base editors, including adenine base editors (ABEs)1 and cytosine base editors (CBEs)2,3, are widely used to induce point mutations. However, determining whether a specific nucleotide in its genomic context can be edited requires time-consuming experiments. Furthermore, when the editable window contains multiple target nucleotides, various genotypic products can be generated. To develop computational tools to predict base-editing efficiency and outcome product frequencies, we first evaluated the efficiencies of an ABE and a CBE and the outcome product frequencies at 13,504 and 14,157 target sequences, respectively, in human cells. We found that there were only modest asymmetric correlations between the activities of the base editors and Cas9 at the same targets. Using deep-learning-based computational modeling, we built tools to predict the efficiencies and outcome frequencies of ABE- and CBE-directed editing at any target sequence, with Pearson correlations ranging from 0.50 to 0.95. These tools and results will facilitate modeling and therapeutic correction of genetic diseases by base editing.

Efficient genome editing by FACS enrichment of paired D10A Cas9 nickases coupled with fluorescent proteins.

Targeted genome editing by clustered regularly interspaced short palindromic repeats (CRISPR-Cas9) raised concerns over off-target effects. The use of double-nicking strategy using paired Cas9 nickase has been developed to minimize off-target effects. However, it was reported that the efficiency of paired nickases were comparable or lower than that of either corresponding nuclease alone. Recently, we conducted a systematic comparison of the efficiencies of several paired Cas9 with their corresponding Cas9 nucleases and showed that paired D10A Cas9 nickases are sometimes more efficient than individual nucleases for gene disruption. However, sometimes the designed paired Cas9 nickases exhibited significantly lower mutation frequencies than nucleases, hampering the generation of cells containing paired Cas9 nickase-induced mutations. Here we implemented IRES peptide-conjugation of fluorescent protein to Cas9 nickase and subjected for fluorescence-activated cell sorting. The sorted cell populations are highly enriched with cells containing paired Cas9 nickase-induced mutations, by a factor of up to 40-fold as compared with the unsorted population. Furthermore, gene-disrupted single cell clones using paired nickases followed by FACS sorting strategy were generated highly efficiently, without compromising with its low off-target effects. We envision that our fluorescent protein coupled paired nickase-mediated gene disruption, facilitating efficient and highly specific genome editing in medical research.

Deep learning improves prediction of CRISPR-Cpf1 guide RNA activity.

We present two algorithms to predict the activity of AsCpf1 guide RNAs. Indel frequencies for 15,000 target sequences were used in a deep-learning framework based on a convolutional neural network to train Seq-deepCpf1. We then incorporated chromatin accessibility information to create the better-performing DeepCpf1 algorithm for cell lines for which such information is available and show that both algorithms outperform previous machine learning algorithms on our own and published data sets.

Constriction of the mitochondrial inner compartment is a priming event for mitochondrial division.

Mitochondrial division is critical for the maintenance and regulation of mitochondrial function, quality and distribution. This process is controlled by cytosolic actin-based constriction machinery and dynamin-related protein 1 (Drp1) on mitochondrial outer membrane (OMM). Although mitochondrial physiology, including oxidative phosphorylation, is also important for efficient mitochondrial division, morphological alterations of the mitochondrial inner-membrane (IMM) have not been clearly elucidated. Here we report spontaneous and repetitive constriction of mitochondrial inner compartment (CoMIC) associated with subsequent division in neurons. Although CoMIC is potentiated by inhibition of Drp1 and occurs at the potential division spots contacting the endoplasmic reticulum, it appears on IMM independently of OMM. Intra-mitochondrial influx of Ca2+ induces and potentiates CoMIC, and leads to K+-mediated mitochondrial bulging and depolarization. Synergistically, optic atrophy 1 (Opa1) also regulates CoMIC via controlling Mic60-mediated OMM-IMM tethering. Therefore, we propose that CoMIC is a priming event for efficient mitochondrial division.

In vivo high-throughput profiling of CRISPR-Cpf1 activity.

CRISPR from Prevotella and Francisella 1 (Cpf1) is an effector endonuclease of the class 2 CRISPR-Cas (clustered regularly interspaced short palindromic repeats-CRISPR-associated proteins) gene editing system. We developed a method for evaluating Cpf1 activity, based on target sequence composition in mammalian cells, in a high-throughput manner. A library of >11,000 target sequence and guide RNA pairs was delivered into human cells using lentiviral vectors. Subsequent delivery of Cpf1 into this cell library induced insertions and deletions (indels) at the integrated synthetic target sequences, which allowed en masse evaluation of Cpf1 activity by using deep sequencing. With this approach, we determined protospacer-adjacent motif sequences of two Cpf1 nucleases, one from Acidaminococcus sp. BV3L6 (hereafter referred to as AsCpf1) and the other from Lachnospiraceae bacterium ND2006 (hereafter referred to as LbCpf1). We also defined target-sequence-dependent activity profiles of AsCpf1, which enabled the development of a web tool that predicts the indel frequencies for given target sequences (http://big.hanyang.ac.kr/cindel). Both the Cpf1 characterization profile and the in vivo high-throughput evaluation method will greatly facilitate Cpf1-based genome editing.

Generation of ΔF508-CFTR T84 cell lines by CRISPR/Cas9-mediated genome editing.

OBJECTIVES: To provide a simple method to make a stable ΔF508-CFTR-expressing T84 cell line that can be used as an efficient screening model system for ΔF508-CFTR rescue. RESULTS: CFTR knockout cell lines were generated by Cas9 with a single-guide RNA (sgRNA) targeting exon 1 of the CFTR genome, which produced indels that abolished CFTR protein expressions. Next, stable ΔF508-CFTR expression was achieved by genome integration of ΔF508-CFTR via the lentivirus infection system. Finally, we showed functional rescue of ΔF508-CFTR not only by growing the cells at a low temperature, but also incubating with VX-809, a ΔF508-CFTR corrector, in the established T84 cells expressing ΔF508-CFTR. CONCLUSIONS: This cell system provides an appropriate screening platform for rescue of ΔF508-CFTR, especially related to protein folding, escaped from endoplasmic-reticulum-associated protein degradation, and membrane transport.

In Situ Pluripotency Factor Expression Promotes Functional Recovery From Cerebral Ischemia.

Recovery from ischemic tissue injury can be promoted by cell proliferation and neovascularization. Transient expression of four pluripotency factors (Pou5f1, Sox2, Myc, and Klf4) has been used to convert cell types but never been tested as a means to promote functional recovery from ischemic injury. Here we aimed to determine whether transient in situ pluripotency factor expression can improve neurobehavioral function. Cerebral ischemia was induced by transient bilateral common carotid artery occlusion, after which the four pluripotency factors were expressed through either doxycycline administration into the lateral ventricle in transgenic mice in which the four factors are expressed in a doxycycline-inducible manner. Histologic evaluation showed that this transient expression induced the proliferative generation of astrocytes and/or neural progenitors, but not neurons or glial scar, and increased neovascularization with upregulation of angiogenic factors. Furthermore, in vivo pluripotency factor expression caused neuroprotective effects such as increased numbers of mature neurons and levels of synaptic markers in the striatum. Dysplasia or tumor development was not observed. Importantly, neurobehavioral evaluations such as rotarod and ladder walking tests showed that the expression of the four factors dramatically promoted functional restoration from ischemic injury. These results provide a basis for novel therapeutic modality development for cerebral ischemia.

Diabetic Mesenchymal Stem Cells Are Ineffective for Improving Limb Ischemia Due to Their Impaired Angiogenic Capability.

The purpose of this study was to investigate the effects of diabetes on mesenchymal stem cells (MSCs) in terms of their angiogenic and therapeutic potential for repairing tissue ischemia. We culture-isolated MSCs from streptozotocin-induced diabetic rats (D-MSCs) and compared their proliferation, differentiation, and angiogenic effects with those from normal rats (N-MSCs). The angiogenic effects of MSCs were evaluated by real-time PCR, in vitro tube formation assay, and transplantation of the MSCs into a hindlimb ischemia model followed by laser Doppler perfusion imaging. The number of MSCs derived from diabetic rats was smaller, and their proliferation rate was slower than N-MSCs. Upon induction of differentiation, the osteogenic and angiogenic differentiation of D-MSCs were aberrant compared to N-MSCs. The expression of angiogenic factors was lower in D-MSCs than N-MSCs. D-MSCs cocultured with endothelial cells resulted in decreased tube formation compared to N-MSCs. D-MSCs were ineffective to improve hindlimb ischemia and showed lower capillary density and angiogenic gene expression in ischemic limbs than N-MSCs. D-MSCs have defective proliferation and angiogenic activities and are ineffective for repairing hindlimb ischemia. Newer measures are needed before MSCs can be employed as a source for autologous cell therapy.

Surrogate reporter-based enrichment of cells containing RNA-guided Cas9 nuclease-induced mutations.

RNA-guided endonucleases (RGENs), which are based on the clustered, regularly interspaced, short palindromic repeat (CRISPR)-CRISPR-associated (Cas) system, have recently emerged as a simple and efficient tool for genome editing. However, the activities of prepared RGENs are sometimes low, hampering the generation of cells containing RGEN-induced mutations. Here we report efficient methods to enrich cells containing RGEN-induced mutations by using surrogate reporters. HEK293T cells are cotransfected with the reporter plasmid, a plasmid encoding Cas9 and a plasmid encoding crRNA and tracrRNA, and subjected to flow cytometric sorting, magnetic separation or hygromycin selection. The selected cell populations are highly enriched with cells containing RGEN-induced mutations, by a factor of up to 11-fold as compared with the unselected population. The fold enrichment tends to be high when RGEN activity is low. We envision that these reporters will facilitate the use of RGEN in a wide range of biomedical research.