The Structure of an Infectious Human Polyomavirus and Its Interactions with Cellular Receptors

摘要

class="head no_bottom_margin" id="sec1title">IntroductionBK polyomavirus (BKV) is an opportunistic pathogen that causes severe diseases in the immunosuppressed (). Initial infection may occur through a respiratory or oral route, and is either asymptomatic or causes a mild respiratory illness (). A persistent infection is then established in the kidney and urinary tract, and 80%–90% of the general population are thought to be seropositive for the most prevalent BKV genotypes (, , , ). In patients undergoing immunosuppression following kidney or bone marrow transplantation, lytic replication can occur, causing polyomavirus-associated nephropathy (PVAN) and hemorrhagic cystitis (, , ). Up to 10% of kidney transplant patients experience PVAN, with as many as 90% of these eventually losing their graft (). The prevalence of BKV-associated disease is rising as the number of transplants increases (), and there is also mounting evidence that BKV, similar to Merkel cell polyomavirus, may be carcinogenic in humans (, , , , href="#bib13" rid="bib13" class=" bibr popnode">Frascà et al., 2015, href="#bib61" rid="bib61" class=" bibr popnode">Tillou and Doerfler, 2014, href="#bib12" rid="bib12" class=" bibr popnode">Feng et al., 2008). No effective antivirals specifically targeting BKV are available (href="#bib5" rid="bib5" class=" bibr popnode">Bennett et al., 2012, href="#bib2" rid="bib2" class=" bibr popnode">Ambalathingal et al., 2017), and so there is increasing interest in the development of targeted therapeutics. High-resolution structural information for the infectious BKV virion and its interaction with cellular receptors would be a useful resource in designing and evaluating such therapies.Our current understanding of polyomavirus structure comes largely from pioneering work by Caspar (href="#bib17" rid="bib17" class=" bibr popnode">Griffith et al., 1992, href="#bib52" rid="bib52" class=" bibr popnode">Rayment et al., 1982) and Harrison (href="#bib57" rid="bib57" class=" bibr popnode">Stehle et al., 1996, href="#bib59" rid="bib59" class=" bibr popnode">Stehle and Harrison, 1996, href="#bib33" rid="bib33" class=" bibr popnode">Liddington et al., 1991) on SV40 (which infects Rhesus monkeys), and murine polyomavirus (PyV). Crystal structures revealed that polyomavirus capsids consist of 360 copies of the major capsid protein VP1, which form 72 “pentons,” each consisting of a ring of five β barrel-containing VP1 monomers (href="#bib45" rid="bib45" class=" bibr popnode">Nilsson et al., 2005). Together, these form a T = 7d lattice, with a C-terminal arm from each VP1 making interactions with neighboring pentons to stabilize the capsid shell. Essentially, polyomaviruses position pentameric capsomers in what would be the pentavalent and hexavalent positions of a true quasi-equivalent lattice (href="#bib19" rid="bib19" class=" bibr popnode">Harrison, 2017, href="#bib7" rid="bib7" class=" bibr popnode">Caspar and Klug, 1962). The asymmetric unit of these capsids therefore comprises six quasi-equivalent VP1 chains, five of which form the pentons in a hexavalent position, with the remaining VP1 chain from five asymmetric units occupying a pentavalent position. A single copy of the minor capsid protein VP2, or its N-terminal truncated variant VP3, is incorporated within each penton (href="#bib9" rid="bib9" class=" bibr popnode">Chen et al., 1998, href="#bib22" rid="bib22" class=" bibr popnode">Hurdiss et al., 2016).Subsequently, there has been a shift toward using isolated VP1 pentons for both structural and functional studies. Binding experiments using such pentons, which cannot assemble into virus-like particles (VLPs), indicated that binding of free pentons to cells requires the presence of sialylated glycans on the host cell membrane (href="#bib37" rid="bib37" class=" bibr popnode">Maginnis et al., 2013, href="#bib40" rid="bib40" class=" bibr popnode">Neu et al., 2013a). Crystallographic structures of putative receptor ligands soaked into penton crystals have revealed the molecular mechanisms of receptor recognition (href="#bib43" rid="bib43" class=" bibr popnode">Neu et al., 2010, href="#bib41" rid="bib41" class=" bibr popnode">Neu et al., 2012, href="#bib40" rid="bib40" class=" bibr popnode">Neu et al., 2013a, href="#bib42" rid="bib42" class=" bibr popnode">Neu et al., 2013b, href="#bib58" rid="bib58" class=" bibr popnode">Stehle and Harrison, 1997). However, the major BKV receptors GD1b and GT1b (href="#bib35" rid="bib35" class=" bibr popnode">Low et al., 2006) cannot be soaked into crystals, presumably as a result of steric restrictions (href="#bib40" rid="bib40" class=" bibr popnode">Neu et al., 2013a). Furthermore, crystallization of isolated pentons requires the removal of the N and C termini of VP1. Therefore, such structures do not allow us to understand how receptors are engaged by the native virion, and penton structures may not capture the full range of surface features which could be targeted by therapeutics. Indeed, recent reports suggest BKV VLPs, but not pentons, can use glycosaminoglycans (GAGs) to attach to target cells (href="#bib15" rid="bib15" class=" bibr popnode">Geoghegan et al., 2017), suggesting that GAGs bind in regions where different capsomers interact, which are missing from capsomer structures. Cryoelectron microscopy (cryo-EM) is potentially well suited to determining the structures of intact virion:receptor complexes, but no EM structure to date has sufficient resolution to elucidate the interactions stabilizing the polyomavirus capsid, or to identify bound ligands (current structures range from 8 to 25 Å [href="#bib22" rid="bib22" class=" bibr popnode">Hurdiss et al., 2016, href="#bib32" rid="bib32" class=" bibr popnode">Li et al., 2015, href="#bib45" rid="bib45" class=" bibr popnode">Nilsson et al., 2005, href="#bib56" rid="bib56" class=" bibr popnode">Shen et al., 2011, href="#bib17" rid="bib17" class=" bibr popnode">Griffith et al., 1992]).Here, we use high-resolution cryo-EM to determine the structure of BKV at 3.8 Å resolution. This resolution allows us to highlight differences between this human pathogen and those that infect simian and murine hosts, which include differences in how both the C-terminal arms and disulfide bonding stabilize the capsid. The importance of disulfide bonds is confirmed by a lower-resolution structure of BKV in reducing conditions in which the density we ascribe to disulfide bonds disappears. We also use a simple method analogous to crystallographic soaking, to determine the structures of BKV bound to receptors: the oligosaccharide moiety of ganglioside GT1b (at 3.4 Å resolution) and the model GAG heparin (at 3.6 Å resolution). Neither receptor causes a conformational change in the capsid, presumably functioning cooperatively to increase the avidity of attachment to the host cell surface.