One image at a time: using cryoEM to unpick how viruses infect cells

Author: Dr Alan Parker, 25th April 2019

Over the last few years a wave of virus-based therapies has progressed through late stage clinical evaluation and to (or near to) the market place. Therapeutic viruses are now being harnessed to correct a diverse array of genetic disorders from Severe Combined Immunodeficiencies (StrimvelisTM), to Retinal Dystrophies (e.g. LuxternaTM), Spinal Muscular Atrophy (ZolgensmaTM) and Haemophilia B (Fidanacogene elaparvovec). It’s fair to say that we are now entering an era where the use of virus-based therapeutics is becoming mainstream, and where advanced gene and cell therapies begin to deliver their immense promise.

The transient nature of transgene expression mediated by adenoviral (Ad) based vectors, coupled with their inherent immunogenicity, means they have limited utility as agents to for treating genetic disorders where lifelong correction of a mutated gene is required. However, these same qualities make Ad well suited as cancer treatments (oncolytic virotherapies), where the goal is targeted and selective (cancer) cell killing with local, tumour-selective expression of immunostimulatory agents, rather than life long gene augmentation or correction.

Many labs worldwide are researching means to tailor Ad based virotherapies into useful anti-cancer agents. The best studied and understood Ad vector, both clinically and experimentally, is the species C, Ad5. For cancer applications, intravenous delivery of the therapeutic is a pre-requisite for targeting the milieu of disseminated metastases. In this context, Ad5 based virotherapies have limited ability to target and infect tumours without significant refinement. This largely due to “off target” sequestration of therapeutic virions in the liver and spleen, the ubiquitous nature of the primary Ad5 receptor, hCAR, and high levels of pre-existing anti-Ad5 immunity in the general population. Combined, these limitations result in rapid and efficient elimination of vector from the blood, and potentially can promote dose-limiting toxicities.

Despite the over-reliance on Ad5, the phylogenic tree of adenovirus is diverse, with 57 individual human serotypes described to date, split over 7 species, A-G. The limitations of Ad5 as a cancer therapeutic have prompted increasing interest in evaluating the natural diversity that exists within the adenovirus family, with many serotypes largely unstudied. As anti-cancer agents, attention has focussed on adenoviruses from species B, with two virotherapies based wholly or partially on species B Ads are undergoing clinical investigation.

Enadenotucirev, developed by PsiOxus is a chimera of two species B adenoviruses, Ad11 and Ad3, whilst Oncos-102, developed by TargoVax is an Ad5 based oncolytic that is modified to use the entry pathways of Ad3, by pseudotyping the Ad3 fiber knob protein onto the oncolytic Ad5 based vector. There therefore exists significant translational interest in the mechanisms by which species B Ads infect cells.

In 2011, Desmoglein-2 (DSG-2) was identified as a receptor for Ad3 and Ad111. Ad3 was shown to utilise DSG-2 exclusively, whilst Ad11 can engage both DSG-2 and CD46 for cell entry. The identification of DSG-2 as an entry receptor highlighted a potential pitfall with their use as anticancer agents – where shedding of partially formed viral particles called penton dodecahedrons from infected tumour cells appeared to promote epithelial to mesenchymal cell transition – a trait that increases tumour aggression and promotes metastasis.

Given the renewed interest in species B based oncolytic virotherapies, it remains critically important to fully and completely define the nature of these interactions at the molecular level, in order to manipulate them to engineer safer and more efficacious virotherapies. The “gold standard” method for defining protein: protein interactions is to crystallise interacting proteins in complex and generate high resolution structure based on x-ray diffraction. This proved difficult for DSG-2 in complex with the Ad3 fiber knob protein, due to the low affinity nature of the interaction, which thwarted attempts to co-complex the proteins.

In this month’s Nature Communications, Vassal-Stermann et al. describe how they overcame this limitation to produce the first accurate structure of Ad3 fiber in complex with DSG-22. Rather than using traditional crystallography and x-ray diffraction-based technologies, the team employed cryoelectron microscopy (cryoEM), analysing hundreds of images of Ad3kn in complex with DSG-2 to produce two structures of the virus: receptor complex at a resolution of 3.5Å and 3.8Å.

The models produced demonstrate that Ad3 can bind to DSG-2 in a stoichiometry of 1:1 or a 1:2 (Ad3kn: DSG2). Surprisingly, the interaction with DSG-2 was found to different to those previously solved for the interaction with CAR or CD46, with the contact residues found to be at the head of the fiber protein, rather than to lower or intermediate portion of the fiber knob protein.  The structures elucidated also allowed the identification of critical DSG-2 binding residues within the Ad3kn protein, particularly a key residue at Asp261, which the authors successfully engineered to ablate interactions with DSG-2.

These findings advance our understanding of how Desmoglein-2 utilising virotherapies infect cells, allowing scientists to develop better strategies for their manipulation and engineering into efficient anti-cancer agents of the future.


1) Wang H, Li ZY, Liu Y, Persson J, Beyer I, Möller T, Koyuncu D, Drescher MR, Strauss R, Zhang XB, Wahl JK 3rd, Urban N, Drescher C, Hemminki A, Fender P, Lieber A. Desmoglein 2 is a receptor for adenovirus serotypes 3, 7, 11 and 14. Nat Med. 2011 Jan;17(1):96-104.

2) Vassal-Stermann E, Effantin G, Zubieta C, Burmeister W, Iseni F, Wang H, Lieber A, Schoehn G, Fender P. CryoEM structure of adenovirus type 3 fibre with desmoglein 2 shows an unusual mode of receptor engagement. Nat Commun. 2019 Mar 12;10(1):1181.