Khabou et al., Biotechnol Bioeng. 2016 Dec;113(12):2712-2724. doi: 10.1002/bit.26031

Insight into the mechanisms of enhanced retinal transduction by the engineered AAV2 capsid variant -7m8.

PubMed link



Manipulating viral capsid surface to create more efficient gene therapy vectors

The use of adeno-associated virus (AAV)-derived vectors that carry therapeutic DNA has become a widely used strategy to treat genetic diseases. The 3D structure of these vectors’ capsid is responsible for the efficiency of several early-steps in gene delivery, such as binding to cell surface receptors, cell entry, and intracellular trafficking. These early infection steps determine transduction efficiency and ultimately the therapeutic benefits in treated patients. For all of these reasons, designing highly efficient vectors is critical for success of gene therapies in the clinic.


However, the specific regions of the viral capsid involved in each step remain not fully understood for the wide variety of naturally occurring variants and serotypes of AAV: Which amino acids interact with the virus’s primary and secondary receptors? What are these receptors? Can these amino acids be mutated to improve capsid properties, such as better interaction with its receptor or immune system escape? By which amino acids can we replace them?


Rational design of AAV capsids using specific mutations of surface amino acids can improve viral infectivity, thereby allowing higher therapeutic efficiency. As an alternative, a combinatorial process can be used to screen millions of variants with random capsid mutations to circumvent our lack of knowledge of viral structure-activity relationships. The library of capsid variants is screened in vitro or in vivo for a specific required property, such as being able to reach the photoreceptors, located in deepest layers of the retina, which is located at the back of the eye1. These cells are indeed primary target cells in gene therapies that aim to fight blindness2. Using this strategy, we previously described a variant called ‘AAV2-7m8’, that out of the millions of screened capsids was the most efficient in gene delivery to deep retinal layers. AAV2-7m8 can indeed reach rodent and monkey photoreceptors when delivered into the vitreous, the gel-like structure that fills the eye (Figure 1). The creation of this new AAV allowed treatment of mouse models of human retinal degenerative disorders, namely Retinitis Pigmentosa3,4 and X-linked retinoschisis1.



Figure 1: Creation of new viruses that can reach deep retinal layers.

(1) To obtain an AAV vector that can transduce deep retinal layers from the vitreous, millions of AAV capsid variants were created using molecular biology. Random genetic mutations included amino acids substitutions or insertions at specific capsid location. (2) The newly synthesized capsids were injected into the eyes of mice expressing a fluorescent protein in their photoreceptors. Photoreceptors were then sorted and the best vector that could reach them was isolated. Steps (1) and (2) were repeated multiple times (3 rounds) to enrich AAV mutants before converging to the final, successful capsid variant dominating the screen. (3) Characterization of the successful variant showed that it contained a 10-amino acid insertion, referred to as ‘7m8’ (sequence: ‘LALGETTRPA’)1. Abbreviations: RPE; Retinal Pigmented Epithelium.


However, the mechanisms by which AAV2-7m8 achieves deeper retinal transduction remained unclear. We thus sought to better understand this newly identified capsid, AAV2-7m85. When 3D structures of AAV2 and AAV2-7m8 (Figure 2) are compared we see that the interaction between two arginine residues is disrupted on AAV2-7m8 capsid, while this same region has previously been shown to be involved in AAV2 binding to its primary receptor, heparan sulfate proteoglycan (HSPG)6,7. Besides, this specific glycan is highly abundant in the inner limiting membrane (ILM), a physical barrier to AAV penetration at the surface of the retina8 (Figure 2). AAV2-7m8’s binding to HSPG being reduced1, it is likely that AAV2-7m8 is less captured in the ILM and thus better penetrates retinal tissue compared to AAV2 (Figure 2), whose affinity to HSPG is high.



Figure 2: Mechanism for enhanced retinal transduction properties of AAV2-7m8.

(1) Molecular modeling of the capsids at the site of insertion of 7m8 peptide. On AAV2 capsid surface, two specific arginines (R) interact with each other and are involved in binding to HSPG. 7m8 insertion disrupts this interaction and leads to a reduced HSPG binding phenotype. (2) Proposed mechanism for the consequences of 7m8 insertion on retinal tissue penetration properties. The inner limiting membrane (ILM) is a physical barrier to AAV penetration in the retina. It is composed of various proteoglycans produced by Müller glial cells’ endfeet (light purple). AAV2 capsids remain retained at the ILM because of its high affinity for HSPG, highly abundant in the ILM. AAV2-7m8, however, likely more efficiently crosses the ILM, thereby allowing better retinal penetration.


We then wondered the role of 7m8 peptide. We inserted it onto other serotype capsids to investigate if this specific peptide influences viral properties aside from HSPG binding and retinal penetration. We thus grafted 7m8 onto capsids that interact with primary receptors other than HSPG, namely AAV5, AAV8 and AAV9 capsids. We produced Green Fluorescent Protein (GFP) encoding vectors with each of these capsids and injected them in mouse eyes to study their retinal transduction (Figure 3). We found different behaviors resulting from the addition of this peptide: insertion of 7m8 reduced AAV5 infectivity while it did not induce obvious changes when inserted onto the AAV8 capsid (Figure 3). Only AAV9 visibly benefited from this insertion (Figure 3) as previously observed with AAV2-7m8. Thus, 7m8 by itself does not seem to improve AAV properties, but rather in conjunction with surrounding amino acids.



Figure 3: Effect of 7m8 peptide insertion onto multiple AAV serotype capsids.

We inserted 7m8 on other AAV capsids and produced GFP vectors for each capsid. After intraocular injections, we performed eye fundus imaging to detect GFP fluorescence, which intensity correlates with retinal transduction efficiency of the vector. 7m8 had different effects on AAV properties as it led to reduction of AAV5 efficiency, while it had no effect on AAV8. AAV9, however, benefited from 7m8 peptide insertion, as previously shown with AAV21.



Lastly, we focused our analysis on the two successful variants that benefited from 7m8 peptide insertion: AAV2-7m8 and AAV9-7m8. We studied the infectivity of these vectors in cultured cells in vitro. To address this question, we quantified the amount of AAV genome copies inside the cells, 24 hours after infection: we found that more genome copies are found in the case of modified AAVs compared to unmodified AAVs (Figure 4). This result suggests that there is an improved cell entry mechanism for both 7m8-inserted capsids. We also studied the subcellular localization of AAV and AAV-7m8 capsids using immunostainings against the viral particles. We found that when cells are infected with the two 7m8-modified vectors, more viral particles are found inside cells compared to infection with parental, unmodified AAVs (Figure 4.2).

How cell entry is increased remains however unclear. 7m8 insertion reduces AAV2 binding to its primary receptor HSPG, whereas for AAV9 peptide insertion does not impact binding to its primary receptor, galactose, as shown by our galactose binding competition assays5. We can hypothesize that interactions with other co-receptors could be enhanced, thereby allowing better internalization of AAV-7m8 capsids. Mass spectrometry could help identify new co-receptors that would specifically interact with 7m8-inserted vectors, compared to parental capsids.




Figure 4: Effect of 7m8 peptide insertion on the infection efficiency of AAV2-7m8 and AAV9-7m8 capsids.

(1) The relative amount of AAV genome copies found inside the cells in vitro was calculated using quantitative PCR. (2) The subcellular localization of AAV was analyzed using immunofluorescence labeling of the viral particles. The outer dashed line represents cell membrane and green spots show cytosolic AAV particles, the inner dashed line represents nuclear membrane and magenta spots indicate AAV particles in the nucleus. The total amount of AAV2-7m8 particles that enter the cells is higher than with AAV2, in both cytosolic and nuclear compartments.


Altogether our data show that 7m8 peptide insertion significantly enhances AAV2 and AAV9 cell entry and retinal transduction efficiency. AAV2-7m8 has already been used successfully in multiple mouse models of blindness1,3,4. This laboratory-created virus will likely be used in the future to treat patients that suffer from blinding retinal diseases, such as Retinitis Pigmentosa.



1. Dalkara, D, Byrne, LC, Klimczak, RR, Visel, M, Yin, L, Merigan, WH, et al. (2013). In vivo-directed evolution of a new adeno-associated virus for therapeutic outer retinal gene delivery from the vitreous. Sci. Transl. Med. 5: 189ra76.
2. Wright, AF, Chakarova, CF, El-aziz, MMA and Bhattacharya, SS (2010). Photoreceptor degeneration : genetic and mechanistic dissection of a complex trait. Nat. Rev. Genet. 11: 273–284.
3. Byrne, LC, Dalkara, D, Luna, G, Fisher, SK, Clérin, E, Sahel, J, et al. (2015). Viral-mediated RdCVF and RdCVFL expression protects cone and rod photoreceptors in retinal degeneration. JCI. 125: 105–116.
4. Macé, E, Caplette, R, Marre, O, Sengupta, A, Chaffiol, A, Barbe, P, et al. (2014). Targeting channelrhodopsin-2 to ON-bipolar cells with vitreally administered AAV restores ON and OFF visual responses in blind mice. Mol. Ther. 23: 7–16.
5. Khabou, H, Desrosiers, M, Winckler, C, Fouquet, S, Auregan, G, Bemelmans, AP, et al. (2016). Insight into the mechanisms of enhanced retinal transduction by the engineered AAV2 capsid variant -7m8. Biotechnol. Bioeng. 12: 2712–2724.
6. Opie, SR, Warrington, KH, Agbandje-McKenna, M, Zolotukhin, S and Muzyczka, N (2003). Identification of amino acid residues in the capsid proteins of adeno-associated virus type 2 that contribute to heparan sulfate proteoglycan binding. J. Virol. 77: 6995–7006.
7. Kern, a, Schmidt, K, Leder, C, Müller, OJ, Wobus, CE, Lieth, CW Von Der, et al. (2003). Identification of a Heparin-Binding Motif on Adeno-Associated Virus Type 2 Capsids Identification of a Heparin-Binding Motif on Adeno-Associated Virus Type 2 Capsids. J. Virol. 77: 11072–11081.
8. Dalkara, D, Kolstad, KD, Caporale, N, Visel, M, Klimczak, RR, Schaffer, D V, et al. (2009). Inner limiting membrane barriers to AAV-mediated retinal transduction from the vitreous. Mol. Ther. 17: 2096–2102.