Adenovirus-host interactions : implications for tropism and therapy

Human adenoviruses (HAdVs) are common viruses often associated withgastrointestinal, ocular and respiratory infections. They can infect a widevariety of cells, both dividing and non-dividing. HAdVs attach to and infecttarget cells through interactions with cellular receptors. It has also beenshown that HAdVs can use soluble host components in body fluids forindirect binding to target cells, a feature that enables the usage of new typesof receptors resulting in a more efficient HAdV infection. We thereforeevaluated the influence of soluble components from four different bodyfluids on HAdV infection of epithelial cells, representing the respiratory andocular tropism of most HAdVs. We found that plasma, saliva, and tear fluidpromote binding and infection of HAdV-5 (species C) and that plasmapromotes infection of HAdV-31 (species A). Further binding and infectionexperiments identified coagulation factor IX (FIX) and X (FX) as thecomponents of plasma responsible for increase of HAdV-5 infection whileFIX alone mediates increase of HAdV-31 infection. We found that as little as1% of the physiological concentration of these factors is required to facilitatemaximum binding. The effect of coagulation factors on HAdV infection was thereafterextended to include all species A HAdVs: HAdV-12, -18 and -31. Species AHAdVs normally cause infections involving the airways and/or the intestine.These infections are often mild but species A HAdVs in general, and HAdV-31 in particular, have been shown to cause severe and life-threateninginfections in immunocompromised patients. We show here that FIXefficiently increase HAdV-18 and -31 (but not HAdV-12) binding andinfection of human epithelial cells, representing the respiratory andgastrointestinal tropism. FIX was shown to interact with the hexon proteinof HAdV-31 and surface plasmon resonance analysis revealed that theHAdV-31:FIX interaction is slightly stronger than that of the HAdV-5:FIX/FX interactions, but more interestingly, the half-lives of theseinteractions are profoundly different. By performing binding and infectionexperiments using cells expressing specific glycosaminoglycans (GAGs) and ivGAG-cleaving enzymes we found that the HAdV-31:FIX and HAdV-5:FIX/FX complexes bind to heparan sulfate-containing GAGs on targetcells, but we could also see a difference in GAG dependence and specificitybetween these complexes.We conclude that the use of coagulation factors might be of moreimportance than previously recognized and that this may affect not only theliver tropism seen when administering adenovirus vectors into thecirculation but also regulate primary infections by wild-type viruses of theirnatural target cells. We also believe that our findings may contribute tobetter design of HAdV-based vectors for gene and cancer therapy and thatthe interaction between the HAdV-31 hexon and FIX may serve as a targetfor antiviral treatment. HAdV vectors are mainly based on HAdV-5 and several problems haverecently become evident when using these vectors. Major challenges withHAdV-5 based vectors include pre-existing neutralizing antibodies, pooraccess to the receptor CAR (coxsackie and adenovirus receptor), and offtarget effects to the liver due to interactions with coagulation factors. Theneed for new HAdV vectors devoid of these problems is evident.HAdV-52 is one of only three HAdVs that are equipped with two differentfiber proteins, one long and one short. We show here, by means of bindingand infection experiments, that HAdV-52 can use CAR as a cellular receptor,but that most of the binding is dependent on sialic acid-containingglycoproteins. Flow cytometry, ELISA and surface plasmon resonanceanalyses revealed that the terminal knob domain of the long fiber (52LFK)binds to CAR, and the knob domain of the short fiber (52SFK) binds tosialylated glycoproteins. X-ray crystallographic analysis of 52SFK in complexwith sialic acid revealed a new sialic acid binding site compared to otherknown adenovirus:glycan interactions. Moreover, glycan array analysisidentified α2,8-linked oligosialic acid, mimicking the naturally occurringpolysialic acid (PSia), as a potential sialic acid-containing glycan receptor for52SFK. ELISA and surface plasmon resonance confirmed the ability of52SFK to interact with PSia. Flow cytometry analysis also showed a fivefold vincrease in binding of 52SFK to PSia-expressing cells compared to controlcells. X-ray crystallographic analysis of 52SFK in complex with oligo-PSiarevealed engagement at the non-reducing end of oligo-PSia to the canonicalsialic acid-binding site, but also suggested the presence of a 'steering rim'consisting of positively charged amino acids contributing to the contact bylong-range electrostatic interactions. PSia is nearly absent on cells in healthy adults but can be expressed inhigh amounts on several types of cancers including: glioma, neuroblastomaand lung cancer. We show here that the short fiber of HAdV-52 bindsspecifically to PSia. Taking into account that HAdV-52 has a supposedly lowseroprevalence and is incapable of interacting with coagulation factors webelieve that HAdV-52 based vectors can be useful for treatment of cancertypes with elevated PSia expression