High-throughput mutagenesis reveals functional determinants for DNA targeting by activation-induced deaminase

Cooperativity between Cas9 and hyperactive AID establishes broad and diversifying mutational footprints in base editors

The partnership of DNA deaminase enzymes with CRISPR-Cas nucleases is now a well-established method to enable targeted genomic base editing. However, an understanding of how Cas9 and DNA deaminases collaborate to shape base editor (BE) outcomes has been lacking. Here, we support a novel mechanistic model of base editing by deriving a range of hyperactive activation-induced deaminase (AID) base editors (hBEs) and exploiting their characteristic diversifying activity. Our model involves multiple layers of previously underappreciated cooperativity in BE steps including: (i) Cas9 binding can potentially expose both DNA strands for 'capture' by the deaminase, a feature that is enhanced by guide RNA mismatches; (ii) after strand capture, the intrinsic activity of the DNA deaminase can tune window size and base editing efficiency; (iii) Cas9 defines the boundaries of editing on each strand, with deamination blocked by Cas9 binding to either the PAM or the protospacer and (iv) non-canonical edits on the guide RNA bound strand can be further elicited by changing which strand is nicked by Cas9. Leveraging insights from our mechanistic model, we create novel hBEs that can remarkably generate simultaneous C > T and G > A transitions over >65 bp with significant potential for targeted gene diversification.

ProteinGym: Large-Scale Benchmarks for Protein Design and Fitness Prediction

Predicting the effects of mutations in proteins is critical to many applications, from understanding genetic disease to designing novel proteins that can address our most pressing challenges in climate, agriculture and healthcare. Despite a surge in machine learning-based protein models to tackle these questions, an assessment of their respective benefits is challenging due to the use of distinct, often contrived, experimental datasets, and the variable performance of models across different protein families. Addressing these challenges requires scale. To that end we introduce ProteinGym, a large-scale and holistic set of benchmarks specifically designed for protein fitness prediction and design. It encompasses both a broad collection of over 250 standardized deep mutational scanning assays, spanning millions of mutated sequences, as well as curated clinical datasets providing high-quality expert annotations about mutation effects. We devise a robust evaluation framework that combines metrics for both fitness prediction and design, factors in known limitations of the underlying experimental methods, and covers both zero-shot and supervised settings. We report the performance of a diverse set of over 70 high-performing models from various subfields (eg., alignment-based, inverse folding) into a unified benchmark suite. We open source the corresponding codebase, datasets, MSAs, structures, model predictions and develop a user-friendly website that facilitates data access and analysis.

Integrating deep mutational scanning and low-throughput mutagenesis data to predict the impact of amino acid variants

BACKGROUND

Evaluating the impact of amino acid variants has been a critical challenge for studying protein function and interpreting genomic data. High-throughput experimental methods like deep mutational scanning (DMS) can measure the effect of large numbers of variants in a target protein, but because DMS studies have not been performed on all proteins, researchers also model DMS data computationally to estimate variant impacts by predictors.

RESULTS

In this study, we extended a linear regression-based predictor to explore whether incorporating data from alanine scanning (AS), a widely used low-throughput mutagenesis method, would improve prediction results. To evaluate our model, we collected 146 AS datasets, mapping to 54 DMS datasets across 22 distinct proteins.

CONCLUSIONS

We show that improved model performance depends on the compatibility of the DMS and AS assays, and the scale of improvement is closely related to the correlation between DMS and AS results.

Controllable genome editing with split-engineered base editors

DNA deaminase enzymes play key roles in immunity and have recently been harnessed for their biotechnological applications. In base editors (BEs), the combination of DNA deaminase mutator activity with CRISPR-Cas localization confers the powerful ability to directly convert one target DNA base into another. While efforts have been made to improve targeting efficiency and precision, all BEs so far use a constitutively active DNA deaminase. The absence of regulatory control over promiscuous deaminase activity remains a major limitation to accessing the widespread potential of BEs. Here, we reveal sites that permit splitting of DNA cytosine deaminases into two inactive fragments, whose reapproximation reconstitutes activity. These findings allow for the development of split-engineered BEs (seBEs), which newly enable small-molecule control over targeted mutator activity. We show that the seBE strategy facilitates robust regulated editing with BE scaffolds containing diverse deaminases, offering a generalizable solution for temporally controlling precision genome editing.

MaveDB: an open-source platform to distribute and interpret data from multiplexed assays of variant effect

Multiplex assays of variant effect (MAVEs), such as deep mutational scans and massively parallel reporter assays, test thousands of sequence variants in a single experiment. Despite the importance of MAVE data for basic and clinical research, there is no standard resource for their discovery and distribution. Here, we present MaveDB ( https://www.mavedb.org ), a public repository for large-scale measurements of sequence variant impact, designed for interoperability with applications to interpret these datasets. We also describe the first such application, MaveVis, which retrieves, visualizes, and contextualizes variant effect maps. Together, the database and applications will empower the community to mine these powerful datasets.

In silico analysis of PFN1 related to amyotrophic lateral sclerosis

Profilin 1 (PFN1) protein plays key roles in neuronal growth and differentiation, membrane trafficking, and regulation of the actin cytoskeleton. Four natural variants of PFN1 were described as related to ALS, the most common adult-onset motor neuron disorder. However, the pathological mechanism of PFN1 in ALS is not yet completely understood. The goal of this work is to thoroughly analyze the effects of the ALS-related mutations on PFN1 structure and function using computational simulations. Here, PhD-SNP, PMUT, PolyPhen-2, SIFT, SNAP, SNPS&GO, SAAP, nsSNPAnalyzer, SNPeffect4.0 and I-Mutant2.0 were used to predict the functional and stability effects of PFN1 mutations. ConSurf was used for the evolutionary conservation analysis, and GROMACS was used to perform the MD simulations. The mutations C71G, M114T, and G118V, but not E117G, were predicted as deleterious by most of the functional prediction algorithms that were used. The stability prediction indicated that the ALS-related mutations could destabilize PFN1. The ConSurf analysis indicated that the mutation C71G, M114T, E117G, and G118V occur in highly conserved positions. The MD results indicated that the studied mutations could affect the PFN1 flexibility at the actin and PLP-binding domains, and consequently, their intermolecular interactions. It may be therefore related to the functional impairment of PFN1 upon C71G, M114T, E117G and G118V mutations, and their involvement in ALS development. We also developed a database, SNPMOL (http://www.snpmol.org/), containing the results presented on this paper for biologists and clinicians to exploit PFN1 and its natural variants.

In silico analysis of the V66M variant of human BDNF in psychiatric disorders: An approach to precision medicine

Brain-derived neurotrophic factor (BDNF) plays an important role in neurogenesis and synapse formation. The V66M is the most prevalent BDNF mutation in humans and impairs the function and distribution of BDNF. This mutation is related to several psychiatric disorders. The pro-region of BDNF, particularly position 66 and its adjacent residues, are determinant for the intracellular sorting and activity-dependent secretion of BDNF. However, it has not yet been fully elucidated. The present study aims to analyze the effects of the V66M mutation on BDNF structure and function. Here, we applied nine algorithms, including SIFT and PolyPhen-2, for functional and stability prediction of the V66M mutation. The complete theoretical model of BNDF was generated by Rosetta and validated by PROCHECK, RAMPAGE, ProSa, QMEAN and Verify-3D algorithms. Structural alignment was performed using TM-align. Phylogenetic analysis was performed using the ConSurf server. Molecular dynamics (MD) simulations were performed and analyzed using the GROMACS 2018.2 package. The V66M mutation was predicted as deleterious by PolyPhen-2 and SIFT in addition to being predicted as destabilizing by I-Mutant. According to SNPeffect, the V66M mutation does not affect protein aggregation, amyloid propensity, and chaperone binding. The complete theoretical structure of BDNF proved to be a reliable model. Phylogenetic analysis indicated that the V66M mutation of BDNF occurs at a non-conserved position of the protein. MD analyses indicated that the V66M mutation does not affect the BDNF flexibility and surface-to-volume ratio, but affects the BDNF essential motions, hydrogen-bonding and secondary structure particularly at its pre and pro-domain, which are crucial for its activity and distribution. Thus, considering that these parameters are determinant for protein interactions and, consequently, protein function; the alterations observed throughout the MD analyses may be related to the functional impairment of BDNF upon V66M mutation, as well as its involvement in psychiatric disorders.

Harnessing natural DNA modifying activities for editing of the genome and epigenome

The introduction of site-specific DNA modifications to the genome or epigenome presents great opportunities for manipulating biological systems. Such changes are now possible through the combination of DNA-modifying enzymes with targeting modules, including dCas9, that can localize the enzymes to specific sites. In this review, we take a DNA modifying enzyme-centric view of recent advances. We highlight the variety of natural DNA-modifying enzymes-including DNA methyltransferases, oxygenases, deaminases, and glycosylases-that can be used for targeted editing and discuss how insights into the structure and function of these enzymes has further expanded editing potential by introducing enzyme variants with altered activities or by improving spatiotemporal control of modifications.

Mutations in human AID differentially affect its ability to deaminate cytidine and 5-methylcytidine in ssDNA substrates in vitro

Activation-induced cytidine deaminase (AID) is known for its established role in antibody production. AID induces the diversification of antibodies by deaminating deoxycytidine (C) within immunoglobulin genes. The capacity of AID to deaminate 5-methyldeoxycytidine (5 mC) and/or 5-hydroxymethyldeoxycytidine (5 hmC), and consequently AID involvement in active DNA demethylation, is not fully resolved. For instance, structural determinants of AID activity on different substrates remain to be identified. To better understand the latter issue, we tested how mutations in human AID (hAID) influence its ability to deaminate C, 5 mC, and 5 hmC in vitro. We showed that each of the selected mutations differentially affects hAID’s ability to deaminate C and 5 mC. At the same time, we did not observe hAID activity on 5 hmC. Surprisingly, we found that the N51A hAID mutant, with no detectable activity on C, efficiently deaminated 5 mC, which may suggest different requirements for C and 5 mC deamination. Homology modeling and molecular dynamics simulations revealed that the pattern of enzyme-substrate recognition is one of the important factors determining enzyme activity on C and 5 mC. Consequently, we have proposed mechanisms that explain why wild type hAID more efficiently deaminates C than 5 mC in vitro and why 5 hmC is not deaminated.

Functions and Malfunctions of Mammalian DNA-Cytosine Deaminases

The AID/APOBEC family enzymes convert cytosines in single-stranded DNA to uracils, causing base substitutions and strand breaks. They are induced by cytokines produced during the body's inflammatory response to infections, and they help combat infections through diverse mechanisms. AID is essential for the maturation of antibodies and causes mutations and deletions in antibody genes through somatic hypermutation (SHM) and class-switch recombination (CSR) processes. One member of the APOBEC family, APOBEC1, edits mRNA for a protein involved in lipid transport. Members of the APOBEC3 subfamily in humans (APOBEC3A, APOBEC3B, APOBEC3C, APOBEC3D, APOBEC3F, APOBEC3G, and APOBEC3H) inhibit infections of viruses such as HIV-1, HBV, and HCV, and retrotransposition of endogenous retroelements through mutagenic and nonmutagenic mechanisms. There is emerging consensus that these enzymes can cause mutations in the cellular genome at replication forks or within transcription bubbles depending on the physiological state of the cell and the phase of the cell cycle during which they are expressed. We describe here the state of knowledge about the structures of these enzymes, regulation of their expression, and both the advantageous and deleterious consequences of their expression, including carcinogenesis. We highlight similarities among them and present a holistic view of their regulation and function.

Probing RNA recognition by human ADAR2 using a high-throughput mutagenesis method

Adenosine deamination is one of the most prevalent post-transcriptional modifications in mRNA. In humans, ADAR1 and ADAR2 catalyze this modification and their malfunction correlates with disease. Recently our laboratory reported crystal structures of the human ADAR2 deaminase domain bound to duplex RNA revealing a protein loop that binds the RNA on the 5′ side of the modification site. This 5′ binding loop appears to be one contributor to substrate specificity differences between ADAR family members. In this study, we endeavored to reveal detailed structure–activity relationships in this loop to advance our understanding of RNA recognition by ADAR2. To achieve this goal, we established a high-throughput mutagenesis approach which allows rapid screening of ADAR variants in single yeast cells and provides quantitative evaluation for enzymatic activity. Using this approach, we determined the importance of specific amino acids at 19 different positions in the ADAR2 5′ binding loop and revealed six residues that provide essential structural elements supporting the fold of the loop and key RNA-binding functional groups. This work provided new insight into RNA recognition by ADAR2 and established a new tool for defining structure–function relationships in ADAR reactions.

The power of multiplexed functional analysis of genetic variants

New technologies have recently enabled saturation mutagenesis and functional analysis of nearly all possible variants of regulatory elements or proteins of interest in single experiments. Here we discuss the past, present, and future of such multiplexed (functional) assays for variant effects (MAVEs). MAVEs provide detailed insight into sequence-function relationships, and they may prove critical for the prospective clinical interpretation of genetic variants.

Antibody diversification caused by disrupted mismatch repair and promiscuous DNA polymerases

The enzyme activation-induced deaminase (AID) targets the immunoglobulin loci in activated B cells and creates DNA mutations in the antigen-binding variable region and DNA breaks in the switch region through processes known, respectively, as somatic hypermutation and class switch recombination. AID deaminates cytosine to uracil in DNA to create a U:G mismatch. During somatic hypermutation, the MutSα complex binds to the mismatch, and the error-prone DNA polymerase η generates mutations at A and T bases. During class switch recombination, both MutSα and MutLα complexes bind to the mismatch, resulting in double-strand break formation and end-joining. This review is centered on the mechanisms of how the MMR pathway is commandeered by B cells to generate antibody diversity.

DNA cytosine and methylcytosine deamination by APOBEC3B: enhancing methylcytosine deamination by engineering APOBEC3B

APOBEC (apolipoprotein B mRNA-editing enzyme catalytic polypeptide-like) is a family of enzymes that deaminates cytosine (C) to uracil (U) on nucleic acid. APOBEC3B (A3B) functions in innate immunity against intrinsic and invading retroelements and viruses. A3B can also induce genomic DNA mutations to cause cancer. A3B contains two cytosine deaminase domains (CD1, CD2), and there are conflicting reports about whether both domains are active. Here we demonstrate that only CD2 of A3B (A3BCD2) has C deamination activity. We also reveal that both A3B and A3BCD2 can deaminate methylcytosine (mC). Guided by structural and functional analysis, we successfully engineered A3BCD2 to gain over two orders of magnitude higher activity for mC deamination. Important determinants that contribute to the activity and selectivity for mC deamination have been identified, which reveals that multiple elements, rather than single ones, contribute to the mC deamination activity and selectivity in A3BCD2 and possibly other APOBECs.