. Brain-wide Cas9-mediated cleavage of a gene causing familial Alzheimer's disease alleviates amyloid-related pathologies in mice. Nat Biomed Eng. 2021 Jul 26; PubMed.


Please login to recommend the paper.


  1. The best available AAV vectors used in the clinic today, AAV9 and AAVrh10, can only address a limited number of disease indications in the central nervous system. These are situations where local administration is possible (as in the striatum or in the hippocampus) or where a secreted therapeutic protein—an enzyme or an antibody—can be made by a minor fraction of modified neuronal cells and diffused throughout the CNS. High doses of vectors, potentially associated with toxicity, are needed to achieve efficacy. In cases where a genetic modification is needed in a large proportion of cells in an extended area such as the cortex, gene-transfer technologies are still suboptimal.

    The paper by Goertsen et al. is a worthwhile read as it represents a milestone toward the development of gene-transfer vectors with an appropriate safety and efficacy profile for gene therapy of neurological diseases. Building on their previously described workflow, the Gradinaru lab describes novel AAV capsid variants with enhanced pharmacological and biodistribution properties in nonhuman primates (NHPs). The approach uses transgenic mice expressing Cre recombinase in various tissues and a lox-able vector design to identify variants with positive or negative tropism for a given target organ out of a complex library. Variants of interest are then independently characterized in Marmosets.

    The group originally reported AAV9-derived variants­ AAV-PHP.B and AAV-PHB.eB­ with vastly improved performances, compared to the archetypal AAV9, for crossing the blood-brain barrier and targeting neural cells after intravenous administration. As generic tools for modifying brain cells in situ, these variants rapidly became a staples of neuroscience labs.

    Several proofs of concept of brain-wide genetic modifications have been published, involving different gene-therapy modalities. The work described by Duan et al. uses a CRISPR Cas9 system and AAV-PHP.eB to knock out the AAPswe allele in transgenic mice with pathological features of Alzheimer's disease. The paper reports a surprisingly high level of gene editing throughout the brain (around 30 percent), a decrease in Aβ-associated pathologies, and an improvement in cognitive performance. Although the quantification and precise localization of gene editing would need to be confirmed and further established, this is an exciting example of what gene therapies for an intractable neurological condition could be. If, of course, all this can be demonstrably translated into the clinic—this where the latest variants may come into play.

    The development of the original PHP.B variant toward the clinic has not been straightforward, essentially because their improved performances in rodents were not reproduced in NHPs where they behaved essentially as the parental AAV9. Here, Goertsen and colleagues selected new variants from a library built from PHP.eB, in which another variable surface loop of the capsid was randomized. Importantly, a negative selection of variants targeting the liver was also applied. The new variants obtained have enhanced tropism for the brain (markedly in mice, less so in NHPs) but interestingly, some of them are de-targeted from the liver.

    This is the primary interest of these variants, because AAV capsids, including AAV-PHP.B, mostly end up in the liver once they access the circulation—this happens not only upon IV administration, but also with intra-CSF or intramuscular delivery. Orders of magnitude more genome copies are usually detected in the liver than at the target of interest, reducing bioavailability, potentially enhancing immune responses, and creating a genotoxic risk due to spurious integration of vector genomes in hepatocytes DNA.

    The variants described by Goertsen et al. are truly of interest. Understanding how they work will be important: How do they reach the CNS, and what determined their cellular tropism there, i.e., neuron versus glia? Is Ly6B or a related protein the cellular receptor at play in BBB crossing? Is the original PHP.eB modification still needed, or would the new loop insertion work in an AAV9 background? What is the basis of escaping retention in the liver? Do they go to the spleen instead, as other de-targeted capsid mutants often do?

    Almost certainly this team is already looking for third-generation variants that can avoid the dorsal root ganglion, a source of toxicity in preclinical studies with AAV9. As others in the field are also reporting new and interesting AAV capsids for CNS and other targets using comparable approaches, I expect to see a flurry of next-gen AAV capsids becoming available and empowering AAV gene therapy in the near future.

    View all comments by Olivier Danos
  2. The remarkable success of gene therapy for fatal spinal muscular atrophy has opened up a new paradigm for treating diseases afflicting the CNS, and it is difficult to imagine a future where gene-based therapies will not be translated to more common neurodegenerative diseases such as Alzheimer’s and Parkinson’s.

    However, many neurodegenerative diseases will require diffuse delivery of genes to the large human brain, which remains a challenge. Moreover, though introduction of a single transgene into the CNS—such as that used in gene therapy for spinal muscular atrophy—is straightforward in principle, better tools are needed to harness the power of contemporary approaches such as CRISPR and base editing, which is more nuanced than a simple gene overexpression. The highlighted studies from the Gradinaru and Ip labs (Goertsen/Flytzanis/Goeden et al. and Duan/Ye et al.) advance both fronts.

    Using a Cre-based screening and sequential engineering, Goertsen/Flytzanis/Goeden et al. identified AAV-based capsid variants that show remarkably widespread gene delivery in a nonhuman primate (marmoset) brain, while showing reduced targeting to non-CNS organs such as the liver. Moreover, one of the capsids (CAP-B10) showed high specificity for marmoset neurons, implying that we should be able to target different cellular populations in larger brains—for example neurons v/s glia—depending on what is needed for the specific disease. While the original screen was in mice, results translated well to marmosets, giving hope that capsids tested in nonhuman primates might also work in humans. This is important because testing efficacy of viral capsids in human brains will be challenging in a clinical setting. By achieving diffuse delivery in a larger brain, this work gets us one step closer to the goal of diffuse brain delivery in humans.

    Duan/Ye et al. use a creative, CRISPR-based strategy to silence the “Swedish” mutation in the APP gene that causes a form of familial Alzheimer’s disease. Their AAV-based approach targets and degrades only the mutant APP allele, keeping the wild-type allele intact. They show that silencing the mutant allele in vivo with a single AAV injection alleviates neuropathology, behavior, and electrophysiologic alterations in a transgenic Alzheimer’s mouse model. They also use a single vector with a smaller Cas enzyme (SaCas9) that carried all the required CRISPR components in a single AAV, which is likely going to be the administration modality for most future CNS CRISPR therapies.

    Indeed, an ongoing Phase I/I trial by Editas for an eye disease also uses the same SaCas9. Though applicable to a small fraction of Alzheimer’s patients carrying the familial Swedish mutation, this study opens the door for similar “personalized” CRISPR-based approaches for other Alzheimer’s mutations, which is potentially transformative for these unfortunate patients. Along with other gene-editing approaches that broadly target Alzheimer’s, for instance a strategy we are pursuing, CRISPR-based therapeutics offer some optimism against a disease where nothing seems to work (Sun et al., 2019).


    . CRISPR/Cas9 editing of APP C-terminus attenuates β-cleavage and promotes α-cleavage. Nat Commun. 2019 Jan 3;10(1):53. PubMed.

    View all comments by Subhojit Roy
  3. The paper by Duan and colleagues describes the use and evaluation of CRISPR/Cas9 to disrupt the Swedish mutation of the APP gene. The authors have carefully tested this gene therapeutic strategy both via intracerebral delivery in 5xFAD and APP/PS1-transgenic mice and via AAV-mediated systemic delivery in 5xFAD mice. Interestingly, they found that both administration routes resulted in less brain pathology as compared to control mice. This work represents a clear step forward from our previous study, in which the same strategy was mainly evaluated in patient fibroblasts, and it is an important contribution to the development of CRISPR/Cas-based treatment strategies for dominantly inherited forms of neurodegenerative disorders (György et al., 2018). 

    The success of gene therapies for CNS disorders, based on CRISPR/Cas9 and other strategies, will to a large extent depend on the development of AAV capsids with improved properties. The PHP.eB subtype used by Duan et al. features widespread CNS transduction in rodents, but targets both neurons and glial cells. Moreover, it also transduces organs outside the brain—such as the liver—and does not effectively pass the blood-brain barrier in nonhuman primates. By deploying a synthetic library of sequence variants of a key region of the capsid, followed by a screen of such variants in cre-based transgenic mice, Goertsen and colleagues were able to identify PHP.eB versions with a preserved ability to cross the blood-brain barrier, and with tropism for neurons and avoidance for cell types outside the brain. One variant in particular with such properties, AAV.CAP-B10, was also found to efficiently transduce the marmoset brain upon peripheral administration.

    This elegant piece of work from the researchers who previously generated PHP.eB represents an important step forward to enable future AAV-based systemic therapies for various brain disorders (Chan et al., 2017). 


    . CRISPR/Cas9 Mediated Disruption of the Swedish APP Allele as a Therapeutic Approach for Early-Onset Alzheimer's Disease. Mol Ther Nucleic Acids. 2018 Jun 1;11:429-440. Epub 2018 Mar 16 PubMed.

    . Engineered AAVs for efficient noninvasive gene delivery to the central and peripheral nervous systems. Nat Neurosci. 2017 Aug;20(8):1172-1179. Epub 2017 Jun 26 PubMed.

    View all comments by Martin Ingelsson
  4. Two complementary and joint studies from the groups of Viviana Gradinaru at Caltech and Nancy Ip at Hong Kong University highlight the potential of gene therapy for neurological disorders, in particular for familial Alzheimer’s disease (FAD). Combined, they take us one step closer to the day when targeted gene editing may allow resolution of genetic defects associated with FAD and other inherited neurological disorders.

    In the first study, Yangyang Duan, from the Ip lab, and co-workers set out to investigate the potential benefits of AAV-mediated CRISPR-Cas9 delivery to the brain in two transgenic mouse models of AD, the 5xFAD and APP/PS1 mice. They began by designing an AAV9 vector simultaneously carrying Staphylococcus aureus Cas9 and one of two single-guide RNAs targeting two amino acid replacements in the amyloid precursor protein (APPswe) harbored in 5xFAD mice. After demonstrating that this AAV9 construct efficiently and specifically edited mutant APPswe (but not WT APP) in vitro, they injected it into the hippocampi of 3-month-old 5xFAD mice and examined their brains when they were 6 months old, an age at which they develop abundant brain amyloid deposits, gliosis, synapse impairment and loss. Interestingly, they found that disruption of APPswe reduced brain loads of both soluble and insoluble Aβ, reduced gliosis, and preserved synapse density (assessed by immunohistochemistry) and function (assessed by hippocampal LTP measurements) compared to mice that did not undergo allele editing. Similar beneficial actions were found in experiments with 9-month-old APP/PS1 mice, which already showed clear signs of neuropathology at the onset of the experiment.

    Most interestingly, they then performed experiments employing an engineered AAV vector (AAV-PHP.eB) that is able to cross the blood-brain barrier. Using this approach, they showed that systemic delivery of the Cas9-sgRNA system resulted in efficient brain editing of APPswe in mice. In 5xFAD mice, this resulted in reductions in brain amyloid load and microgliosis and, remarkably, rescued some behavioral abnormalities presented by such mice.

    While this study clearly demonstrates the potential of brain gene editing for monogenic neurological disorders, an issue that needs to be taken into consideration in the particular case of AD is that a small proportion (maybe as little as 3 percent) of total AD cases are of clear autosomal-dominant inheritance. This limits the applicability of the current approach to familial AD, meaning that chances appear slim that a similar approach could be developed that would comprise a universal therapy for the vast majority of sporadic AD cases. On the other hand, it should be emphasized that familial AD typically affects people at very young ages, often in their fifth or even fourth decades of life, and shows a very rapid progression. Once the results of genetic testing are complete, patients with mutations in APP, presenilin 1 (PS1), or presenilin 2 (PS2) essentially face an irrevocable sentence of AD. To these patients, therefore, the possibility of genetic targeting of disease-causing mutated alleles may represent a strong hope of evading that life sentence.

    Another aspect that remains to be determined is whether the benefits of systemic AAV-mediated editing of brain genes are persistent or transient. It would be interesting to perform follow-up studies to determine whether the progression of AD neuropathology and functional deficits are stalled by AAV-Cas9 treatment as mice age further, or whether another round of treatment may be necessary. Finally, although Duan et al. have shown that working memory and anxiety-type behavior are improved by gene editing in 5xFAD mice, it would have been nice to see the potential beneficial effects in measures of hippocampus-dependent spatial orientation.

    In the second study, co-lead authors Goertsen, Flytzanis, and Goeden (from the Gradinaru lab) and co-workers have paved new ground for the development of safe and effective viral-mediated gene therapies in the central nervous system. AAV-mediated gene therapy approaches have been under intensive investigation for CNS disorders for several years now. Although impressive cases of success have emerged, approaches are often hampered by challenges in vector delivery (e.g., requiring intrathecal or intracranial administration), low efficiency in transduction (often related to the need for parenchymal infusion, which limits diffusion of the viral construct), lack of cellular specificity, the possibility of systemic undesirable side-effects. Beginning with both naturally occurring serotypes and previously engineered AAV vectors, and using methods of directed evolution and AAV engineering, the current study generated a very large number of changes in AAV capsid, and screened them for optimal CNS targeting and minimal targeting of peripheral organs in mice.

    Among thousands of capsids screened, a few were quite effective in targeting the brain while at the same time avoiding peripheral organs when administered systemically in mice. One of them, AAV.CAP-B10, was particularly effective in transducing brain cells while exhibiting virtually no transduction in the liver (which is highly targeted by systemic administration of previously used AAV vectors). Of particular interest was the comparison between AAV.CAP-B10 and AAV-PHP.eB, which was previously developed to cross the blood-brain barrier and was used in the study discussed above. While systemic administration of both AAV.CAP-B10 and AAV-PHP.eB in mice effectively targeted the brain, the latter significantly targeted the liver. In addition, greater cellular specificity was also achieved with AAV-CAP-B10, which largely transduced neurons (and not glia).

    Goertsen et al. further showed that systemic administration of AAV.CAP-B10 efficiently targeted the brain of marmosets, a relevant finding in terms of the translational potential of this approach in primates. Overall, AAV.CAP-B10 appears as a powerful tool to expand the repertoire of viral-mediated therapeutic approaches in the central nervous system while minimizing peripheral side-effects. It would be interesting to determine, for example, if use of AAV.CAP-B10 could improve efficiency of the Cas9-mediated gene-disruption approach used in the study by Duan et al., while at the same time abrogating peripheral targeting. On the other hand, in light of the increasing recognition of the role of glial cells (notably microglia and astrocytes) in the pathogenesis of AD and of the growing number of glial proteins that are risk factors for AD, it remains to be determined whether enhanced specificity toward neurons would represent an advantage or a drawback of AAV.CAP-B10.

    View all comments by Sergio Ferreira

Make a Comment

To make a comment you must login or register.