Recombinant adeno-associated viruses (rAAV) are the most commonly used delivery vectors in gene therapy. However, naturally occurring AAVs with organ-targeting properties are very limited. In order to achieve more precise treatment goals, scientists have continuously screened for novel AAV serotypes. Currently, there are various screening systems including natural discovery, rational design, directed evolution, and deep learning aided by AI, which involves predicting variants through algorithms. Approximately 13 naturally selected AAV serotypes have been identified and exist in the human body. Rational design involves genetic engineering modifications of capsid proteins. One method is to add a peptide to the capsid protein that can bind to cell membrane surface receptors, thus enhancing tissue specificity. Another method is to inhibit capsid degradation, where proteasome inhibitors can enhance AAV transduction both in vitro and in vivo. Modifying the AAV capsid can reduce ubiquitination, allowing AAV viral particles to avoid cytoplasmic degradation mediated by proteasomes. Directed evolution is based on mimicking natural evolution, where capsids are subjected to selective pressure, resulting in genetically diverse variants with specific biological properties. Common methods for constructing AAV capsid libraries include chimeric capsids, error-prone PCR, insertion of random peptide fragments at specific sites, and random substitution of capsid surface variable regions. Utilizing AI for prediction combines bioinformatics and computational power to simulate mutant protein folding and assembly, aiming to discover new AAV variants.

Currently, the most commonly used method is rational design, where serotypes selected using this system can provide experimental data obtained from mouse experiments as the original model for AI learning. Here, we analyze several different methods in detail to help establish AAV serotype screening systems effectively.

 

CREATE System:

CREATE stands for Cre recombination-based AAV targeted evolution. This method utilizes Cre-loxP for screening recombinant AAV, enabling more effective transduction of specific cell populations expressing Cre enzyme in vivo. rAAV is the preferred vector for in vivo gene transfer in many non-dividing cells and has been proven to be safe in clinical trials. However, therapeutic applications are limited due to low transduction efficiency of target cells. For instance, when transferring target genes to the central nervous system via intravenous administration, the available vectors currently have at least an order of magnitude lower transduction efficiency in the central nervous system of mice compared to the liver. By creating an AAV capsid library and selecting variants with more ideal characteristics, the transduction efficiency of rAAV can be improved. However, there is still no clear vector capable of effectively transducing the entire central nervous system. This is due to the challenges posed by the highly selective blood-brain barrier (BBB) and cellular heterogeneity of the central nervous system in transferring genes from the vascular system to the central nervous system. Therefore, researchers have designed the CREATE system to selectively screen suitable serotypes by providing selective pressure on capsids that pass through the blood-brain barrier and selectively transduce specific cell types functionally. CREATE is a method that relies on Cre recombinase, selectively directing capsids to target cell populations expressing Cre enzyme.

CREATE utilizes the rAAV capsid genome (rAAV-cap-in-cis-lox), including the full-length AAV cap gene, AAV rep gene regulatory elements, and a Cre reversible switch. A capsid library is constructed within the rAAV-Cap-in-cis-lox framework, and the virus library is transferred to animals with specific cell populations expressing Cre enzyme. This system allows for selective amplification and restoration of the sequence of transduced target sequences. Since the rAAV-Cap-in-cis-lox genome lacks functional rep genes, researchers modified the AAV2/9 Rep-Cap plasmid by inserting termination codons into the reading frames of each capsid protein VP1-VP3 to terminate capsid protein expression. Assembly activating protein (AAP) is expressed in another reading frame within the cap gene, so these termination codons do not alter the amino acid sequence of AAP. This combination of separate rAAV-cap-in-cis-lox genomes and Rep-AAP AAV auxiliary systems efficiently produces rAAV. The CREATE selection system applies this basic principle, enabling the restoration and amplification of capsid sequences in different tissue samples expressing Cre enzyme.

Researchers inserted a 7-peptide random sequence (7-mer) between amino acids 588 and 589 (VP1 site) of the rAAV-Cap-in-cis-lox genome to establish an AAV9 mutant capsid library. To screen for AAV serotypes capable of penetrating the blood-brain barrier and effectively transducing the entire central nervous system, researchers intravenously injected the capsid library into adult GFAP-Cre mice, which specifically express Cre in astrocytes. One week later, DNA was isolated from brain and spinal cord tissues, and primers were designed to PCR amplify the genomes with Cre-mediated recombination of the restored capsid sequence order.

Researchers cloned the entire cap library into the rAAV-Cap-in-cis-lox genome and randomly selected 13 clones for sequencing. All tested sequences recovered from the GFAP1 plasmid library were unique. A second virus library (GFAP2) was generated from the GFAP1 plasmid library, and an additional round of screening was conducted in GFAP1-Cre mice. Following the second screening, several mutants were significantly enriched and showed enhanced central nervous system transduction effects. Researchers further targeted the most significantly enriched mutant AAV-PHP.B for in vivo validation, encoding a 7-mer sequence as TLAVPFK.

GMP mRNA
Check out our AAV Capsid Service to expedite your research
CREATE System Screening Strategy

Fig1: CREATE System Screening Strategy

Efficient transduction of AAV-PHP.B
Fig2: Efficient transduction of AAV-PHP.B-mediated genes throughout the entire central nervous system following intravenous injection in adult mice.
AAV-PHP.B Transduces
Figure 3: AAV-PHP.B Transduces Various Central Nervous System Cell Types More Effectively than AAV9

CREATE system’s important advantage is that it introduces selective pressure for capsid screening, effectively mediating intracellular transport, and converting the single-stranded viral genome to the double-stranded DNA form required for long-term transduction (only double-stranded DNA genomes can serve as substrates for Cre). After two rounds of in vivo selection, additional selective pressure for functional capsids identified AAV-PHP.A and AAV-PHP.B variants in individual assays. Theoretically, CREATE can be applied to all tissues capable of expressing Cre enzyme to develop new AAV capsid variants.

 

TRACER System:

The TRACER system, abbreviated from Tropism Redirection of AAV by Cell-type-specific Expression of RNA, is a serum type screening method developed for the central nervous system. However, considering that currently engineered capsids capable of penetrating the blood-brain barrier and targeting the central nervous system have only been validated at the level of rodents, there is still a long way to go for gene therapy delivery with the screened serotypes. In order to overcome the limitations of animal experiments, researchers have developed the TRACER platform, which can rapidly screen capsids with blood-brain barrier penetration and central nervous system targeting in non-transgenic animals.

High-throughput mutagenesis and directed evolution of AAV capsids were proposed in 2003, which promoted the development of AAV serotype research due to the simple viral genome structure, low recombination of wild-type AAV assembly, and high correlation between the genome and capsid. Early designs of AAV directed evolution were strictly designed for in vitro cell culture and used co-infection with helper adenoviruses to enhance transduction efficiency. Over the past decade, functional AAV libraries based on cell-specific sorting, in vivo helper virus infection, or cell-specific Cre-lox systems have provided avenues for AAV capsid variant screening and the identification of new serotypes. For example, two variants of AAV9, PHP.B, and PHP.eB, have been shown to have excellent transduction ability in the brains of C57BL/6 mice after systemic injection. However, subsequent studies have shown that these characteristics cannot be translated to other laboratory mouse strains or non-human primates (NHPs). Additionally, Cre-dependent AAV library screening methods strictly rely on transgenic animals to restore the specific transduction ability of AAV, limiting the use of clinically relevant animals (such as NHPs). The TRACER system, on the other hand, is a platform for RNA capsid library screening expressed in a cell type-specific manner, eliminating the restriction on transgenic animals.

Design Scheme of Cell-Specific RNA Library Screening
Figure 4: Design Scheme of Cell-Specific RNA Library Screening
In the absence of helper viruses, AAV genome transcription is inhibited. In order to generate a sequence library capable of transcriptional evolution driven by RNA, researchers inserted a non-AAV promoter upstream of the Cap gene, allowing specific expression in cells while retaining the minimum regulatory elements required for capsid protein expression. The research determined that the minimum Rep fragment starting from nucleotide 1700, containing the AAV P40 promoter and splice donor and acceptor sequences, was sufficient to efficiently produce the virus, although the yield was lower than that of wild-type AAV. A series of in vitro experiments were conducted to test for potential interference between the P40 promoter and non-AAV promoters. Three promoters were tested in the AAV9 TRACER vector: the cytomegalovirus (CMV) enhancer-chicken β-actin (CAG) promoter, the neuron-specific human synapsin 1 (SYN) promoter, and the astrocyte-specific gfaABC1D (GFAP) promoter. Constructs were built, viruses were produced, and virus particles were added to HEK293T cells and primary mouse brain cells. The results showed that RNA expression driven by SYN and GFAP promoters was strongly inhibited in non-central nervous system HEK293T cells (170-fold and 400-fold lower, respectively) but normal in transcription in primary mouse brain cells. Similar trends were observed in experiments using GFP vectors lacking the P40 sequence, suggesting that the presence of the P40 element has almost no effect on the regulation of upstream concatenated promoters used for biological screening. Since the TRACER structure lacks a full-length rep sequence, during virus production, REP proteins are provided by separate plasmids.
NGS Sequencing Analysis of C57BL6 Mouse TRACER Library
Figure 5: NGS Sequencing Analysis of C57BL/6 Mouse TRACER Library

Dependence on Cre enzyme for viral DNA directed evolution strategies may be affected by viral virulence, and there is a risk of detecting sequences that are not expressed and screening out false-positive serotypes. In the TRACER system, RNA expression is used to screen for positive serotypes. In addition, the use of cell-type-specific promoters reduces the recovery of capsids from unwanted cell types, enriching serotypes with high neuronal or glial mutations after only two rounds of screening.

The TRACER system can be used for any promoter sequence less than 2 kb in length, meeting the size limits of AAV packaging, and can be of great value when selective expression from rare cell types is required. In this study, NGS analysis data related to the syn promoter highly predicted the characteristics of individual capsids and identified multiple variants with high specificity in neurons. In contrast, in experiments with GFAP-initiated expression, capsids with high enrichment scores showed low consistency in in vivo and in vitro expression in the brain and could not efficiently transduce GFAP-positive astrocytes.

In this experiment, the high success rate of individual candidate capsids (10/10 candidate capsids, at least 10 times higher than AAV9) demonstrated the robustness of the TRACER system’s cell type-specific promoter NGS platform. Variations from different sequence clusters showed some differences in cell tropism and mouse strain limitations. Clearly, AAV employs multiple mechanisms to break through the blood-brain barrier. Brain tissue examination showed very similar enrichment effects for all candidate serotypes in specific brain regions, such as the hippocampal CA2 region, cerebellum, or thalamus. Although significant progress has been made in studying 7-mer peptide insertion for AAV blood-brain barrier penetration, the design of other variable regions of the AAV capsid domain may also contribute to virus tropism research.

 

DELIVER System:

The DELIVER system, abbreviated from Directed evolution of AAV capsids leveraging in vivo expression of transgene RNA, is a platform similar to TRACER, both of which utilize RNA screening strategies. The DELIVER system involves the screening of multiple capsid libraries combined with in vivo transcription reactions to find targeted capsids obtained through directed evolution in tissues and animal models.

DELIVER System
Gene editing offers hope for treating many human diseases, but delivering gene-modifying drugs to large tissues (such as skeletal muscles, accounting for about 40% of body weight, distributed throughout the body) poses challenges. Recombinant AAV generated from natural capsids is intercepted by the liver after systemic administration, thus limiting the efficiency of drug transduction to other organs.
DELIVER Screening of Muscle
Figure 6: DELIVER Screening of Muscle-Targeting AAV Capsid Variants Containing RGD Motif
DELIVER is a new serotype screening method designed for muscle tissue, aiming to find high-quality serotypes targeting muscles and develop drugs for treating muscle diseases. Using this method, researchers identified a class of capsid variants containing the RGD motif, which exhibited good muscle transduction efficiency in mice and non-human primates (NHPs) after intravenous injection.

The DELIVER screening process is also based on RNA screening, with similar design principles, but with the use of muscle-specific promoters. In this study, a random 7-peptide sequence was inserted between amino acids 588 and 589 in the high variability region VIII of the AAV9 capsid, ensuring exposure of the inserted sequence on the capsid surface. Under the regulation of cell-type-specific mammalian promoters, each variant was also packaged with a transgene encoding its own capsid sequence, allowing for normal expression in tissues after production in HEK293 cells and in vivo delivery.

The initial assessment of DELIVER involved the rigorous selection of capsid variants in C57BL/6J mice, expressing under the cytomegalovirus (CMV) promoter. Differences in enrichment at the DNA and mRNA levels were observed across different tissues compared to the viral library. Unlike selection based on the library’s genomic DNA, fewer novel serotypes were selected based on mRNA expression, indicating that only a small fraction of capsid variants entering target cells could transduce these cells and express their encoded genes. The study found that capsid variants with higher abundance in the viral library upon injection were more enriched at the genomic DNA level in the liver, but there was almost no correlation between the abundance of each variant in the viral library and the transgenic mRNA level in the liver. Similarly, the abundance of each identified capsid variant in the liver and muscle was evaluated at the genomic DNA and transgenic mRNA levels, with no correlation observed between DNA and mRNA. Subsequently, muscle-specific promoters were used to enhance the tissue-targeting selectivity of capsid variants. Skeletal and cardiac muscles contain several different cell types, but AAV capsids can effectively deliver transgenes to both cell types, particularly muscle fibers and cardiomyocytes. When injected into the skeletal and cardiac muscles of mice, virus libraries driven by the CK8 or MHCK7 promoter showed higher levels of transgenic mRNA compared to those expressed under CMV, and two rounds of in vivo evolution were conducted to select multiple capsid variants expressed by the MHCK7 promoter in different muscle tissues of C57BL/6J mice.

In the study of the DELIVER screening system, researchers applied an in vivo directed evolution strategy for capsids in mice and NHPs, identifying effective muscle-targeting serotype capsid variants in different species. The study found that the RGD motif and the amino acid sequence surrounding it play an important role in the transduction efficiency and targeting selection of capsids. The most muscle-targeting capsid variants identified during the screening process all contained the RGD motif, but the biological characteristics of the amino acids around this motif varied among different species. For example, MyoAAV 2A was the optimal variant identified in mice, showing very effective muscle transduction in mice. However, after systemic administration, MyoAAV 2A could not efficiently transduce muscle tissue in NHPs. All variants identified from NHPs showed superior muscle transduction efficiency when tested after systemic delivery compared to their efficiency in mice. Interestingly, the most optimal capsid variants selected from NHPs also showed very effective muscle transduction in mice. The results of the study indicate that using different capsid libraries and rigorous screening strategies for in vivo directed evolution is an effective screening method that can produce the most effective tissue-targeting capsid variants across species. However, there are also questions about which variants are most effective for human applications. Although it is expected that serotypes selected in NHPs will perform well in humans, further optimization of AAV’s transduction effects in the human body is required considering the evolutionary relationships between these species. In summary, the DELIVER system provides a highly adaptable platform for identifying capsid variants targeting any tissue or cell type in the body, greatly expanding the clinical applications of this vector system in various fields and disciplines. Applying DELIVER to the screening of novel AAV serotypes in other tissues and organ systems will have a profound impact on accelerating the development and translation of gene therapy and genomic medicine approaches.

GMP mRNA
Check out our AAV Packaging Service (NHP Grade) to expedite your research
About PackGene

PackGene Biotech is a world-leading CRO and CDMO, excelling in AAV vectors, mRNA, plasmid DNA, and lentiviral vector solutions. Our comprehensive offerings span from vector design and construction to AAV, lentivirus, and mRNA services. With a sharp focus on early-stage drug discovery, preclinical development, and cell and gene therapy trials, we deliver cost-effective, dependable, and scalable production solutions. Leveraging our groundbreaking π-alpha 293 AAV high-yield platform, we amplify AAV production by up to 10-fold, yielding up to 1e+17vg per batch to meet diverse commercial and clinical project needs. Moreover, our tailored mRNA and LNP products and services cater to every stage of drug and vaccine development, from research to GMP production, providing a seamless, end-to-end solution.

More Articles

Related Services

icon16

AAV Packaging Services

READ MORE

icon16

AAV Packaging Service (NHP)

READ MORE

icon16

AAV Packaging Service (HT)

READ MORE