Unnatural Selection

Genetic mutations are not uncommon. Around the globe, five million people currently suffer from congenital retinal dystrophies caused by defects in specific genes. Though these errors may only alter a single element of a person’s genetic blueprint, their effect can be devastating, often resulting in a loss of function in the photoreceptors or cells that form the retinal pigmented epithelium. To date, around 150 such defects have been identified. 

But hope is on the horizon. In recent years, scientists have developed dedicated gene-delivery vehicles – vectors – that can be used to transport functional copies of the relevant gene into the defective retinal cells, essentially supplementing the missing function. These efforts have largely focused on vectors that are based on the genomes of adeno-associated viruses (AAVs). In collaboration with Hildegard Büning, a professor of the Infection Biology of Gene Transfer at Hannover Medical School (MHH), and an international team of researchers, my team and I have succeeded in constructing vectors that can be more easily and effectively introduced into retinal cells. Until now, it has been necessary to inject the viral vectors directly under the retina – a technique that requires highly skilled experts and facilities that are available only at specialized hospitals. Moreover, when using such an approach, there is always a risk of damage to the fragile retinal tissue itself. Another drawback of this method is that each injection reaches only a relatively small fraction of the target cells.

Using animal models, my colleagues and I have injected novel AAV constructs directly into the vitreous humor – and confirmed that these novel vectors can be transported into the light-sensitive photoreceptors in the retinal tissue (1). Our theory was strengthened by studies on human retinal tissue cultured in the laboratory, as well as studies with human retinal organoids (HROs), which are multilayer 3D models of the human retina grown from human induced pluripotent stem cells (hiPSC) in culture (2). The novel AAV vectors resulted in high levels of gene expression in most major cells types of mature HROs, including photoreceptors. Finally, studies in a mouse model of achromatopsia – a complete lack of color vision – provided a first preclinical proof-of-concept for restoration of daylight vision after intravitreal gene supplementation with one of these novel vectors (1).

Origin story
So how do these novel vectors work? The key is to engineer specifically modified variants of the naturally occurring AAV serotypes commonly used so far. In our case, these vectors carry modifications in the capsid (the protein part of the viral vector) surface that endow novel properties and enable the vectors to cross multiple biological barriers and more efficiently infect (and transduce) target cells – for example, photoreceptors – in the retina. Biological complexity aside, the important point is that these novel vectors show higher efficiency across species and also work when tested on post-mortem human retina, which we can keep in culture for approximately 10 days.

To engineer modified variants, we used a procedure known as “directed evolution” – essentially mimicking what happens in nature (where variants with the most advantageous properties are chosen to overcome a given selection pressure) but under laboratory settings and conditions that accelerate the process. The aim? To select AAV variants that outperform other variants at a given task. In principle, you start with a high diversity of vector variants (aka a AAV variant library) – our library consists of five million distinct variants. Next, you screen those variants for beneficial properties using a specific selection assay (either in vitro or in vivo). 

I think it’s important to note that we did not invent directed evolution – nor are we the only group who applied it to engineer novel vectors; however, we performed the selection process in a unique way, applying very high selection pressure. 

The general (or dogmatic) view is that selection pressure used should be logically and specifically linked to the desired outcome. For our goal, this would mean applying the AAV library via intravitreal injection and then recovering vector variants (or their transcripts) from the photoreceptors after an incubation time of several days to weeks. And indeed, other researchers have done this in the past – with varying degrees of success. We decided to break with the dogma and applied the AAV library systemically (via tail vein injection) into live mice and then tried to recover AAV vector genomes from retinal cells (in particular, photoreceptors) after only a short incubation time (24 hours).

The high selection pressure is made more obvious when you explore the individual biological barriers that successful vector variants have to overcome, starting with the host immune system, systemic clearance, and the blood vessel endothelial cell barrier and retina-blood-barrier (RBB). Within the retina, vectors would need to escape from the retinal blood vessels and diffuse into the retinal tissue. If entry from the choroidal blood vessels is assumed, then vectors should move through the Bruch’s membrane, the retinal pigment epithelial cell barrier, the photoreceptor extracellular matrix, and the outer limiting membrane to finally enter the photoreceptor cells. If the entry pathway is the photoreceptor outer segment, then the connecting cilium would need to be overcome as well. If entry from the vitreal blood vessels is assumed, then vectors would need to penetrate through the inner limiting membrane, ganglion cell layer, the inner plexiform layer, the inner nuclear layer, and the outer plexiform layer to finally enter the photoreceptors at their synaptic endings. Upon cell entry, the vectors also need to traffic through and escape the endolysosomal vesicle system, uncoat and finally shuttle their genome through the nuclear membrane into the cell nucleus. 

Despite the novel delivery method (and difficult journey), we reduced our selection time window for overcoming these biological barriers to just 24 hours. 

The result was promising – surprisingly so. To be frank, I did not expect to be able to recover any AAV vector genome from retinal photoreceptors after systemic application, let alone with only 24 hours incubation time. Scientific curiosity drove me to try it – and now I am delighted I did! At least in the retinal universe, we had created a superhero vector.

From systemic to intravitreal injection – and beyond
Building on these promising preclinical results, we now aim to develop next-generation gene therapies for retinal disorders. One major goal will be to address conditions that require broad and highly efficient transduction of retinal photoreceptors, such as retinitis pigmentosa. Conventional AAV vectors used so far need to be delivered via subretinal injection in order to target photoreceptors. Such subretinal injections are challenging and carry the risk of collateral damage to the fragile retina of the affected patients. To mitigate this risk, only small volumes of vector are administered in order to detach only a small part of the retina. However, due to limited spreading of the vector out of the subretinal bleb, only a small portion of the affected retina can be treated in this way. We now plan to leverage the novel AAV vector technology for intravitreal gene therapy of retinitis pigmentosa aiming to treat a large part of the retina, without the need for potentially damaging subretinal injections. 

When using the vitreous humor as a target site, the risk for collateral damage caused by the surgery is markedly reduced, but the vector is now better exposed to the local immune system, which could result in a local immune response. However, we already have firsthand evidence that the modifications made in our novel vectors have also had a positive effect on the immune response. The novel vectors partially escape neutralization by anti-AAV antibodies – almost certainly the result of the tough selection process (1).

So, what’s next? Though our results are promising, they do not necessarily speak to whether gene therapy will ever be able to fully restore lost or damaged vision, which, of course, depends on the specific underlying pathobiology. For some inherited retinal disorders, the disease progression (for example, the degeneration of photoreceptors) is rather slow. In those cases, supplementation of a healthy copy of the diseased gene – at an sufficiently early stage – could lead to a substantial restoration of vision. But with blinding disorders, even small effects from treatments can substantially improve a patient’s life – while more significant effects allow them to have a more normal life.

Fortunately, our novel vectors are based on previously clinically-validated AAV technology – so you could say they are improved versions of already known and widely used tools. We therefore believe that our technology can be more easily translated into the clinic; in fact, I expect to see the first clinical trial within the next couple of years. 

When you work on the development of gene therapies, you always want to see your approach translated into clinical application. Success requires translational science and clinical development – and that’s why my group continues to engineer further improved vectors. We can only hope we will have similar success addressing other remaining unmet needs in the future.