Attenuation hotspots in neurotropic human astroviruses

During the last decade, the detection of neurotropic astroviruses has increased dramatically. The MLB genogroup of astroviruses represents a genetically distinct group of zoonotic astroviruses associated with gastroenteritis and severe neurological complications in young children, the immunocompromised, and the elderly. Using different virus evolution approaches, we identified dispensable regions in the 3′ end of the capsid-coding region responsible for attenuation of MLB astroviruses in susceptible cell lines. To create recombinant viruses with identified deletions, MLB reverse genetics (RG) and replicon systems were developed. Recombinant truncated MLB viruses resulted in imbalanced RNA synthesis and strong attenuation in iPSC-derived neuronal cultures confirming the location of neurotropism determinants. This approach can be used for the development of vaccine candidates using attenuated astroviruses that infect humans, livestock animals, and poultry.


Introduction
Human astroviruses (HAstVs) belong to the genus Mamastrovirus, family Astroviridae and are 34 a common cause of gastroenteritis in children, the elderly and immunocompromised adults 1 . 35 Lately, the HAstV group of the Astroviridae family has expanded to include new groups of 36 viruses unrelated to the eight previously described classic HAstV serotypes (Fig. 1A). These 37 new human astrovirus groups are more closely related to certain animal astroviruses than to the 38 classical HAstVs, suggesting zoonotic transmission 1 . One of these groups is named MLB, after is not yet understood. It is also unclear if MLB astroviruses exploit cellular proteases other than trypsin to process the capsid protein and how this impacts the infectivity 66 of virus particles. 67 Growing evidence suggests that astroviruses are found globally, infecting a wide range of 68 species, and have the potential for recombination, rapid evolution, and can adapt to different 69 hosts 3,11-16 . Unfortunately, many astrovirus groups have remained overlooked for decades 70 because of the absence of molecular tools, such as infectious clones and replicons. Therefore, 71 developing a robust reverse genetics (RG) system for the non-classical human astroviruses is 72 essential to understand the basic biology, evolution and host-virus interplay. 73 In 1997 Matsui's group established the first RG system for the human astrovirus serotype 1 to 74 rescue infectious viral particles 17 . This system has been successfully used and has shed light 75 on multiple aspects of astrovirus replication and pathogenesis. HAstV1 RG system requires were limited to classical human astroviruses. So far, RG systems for two non-human 81 astroviruses were developed: first for the avian astrovirus by using duck astrovirus (DAstV) 82 genome of D51 strain 20 , and second for porcine astrovirus (PAstV1-GX1) 21 . Although both 83 non-human RG systems allow the recovery of infectious viral particles, these systems also rely 84 on the two cell lines.

85
It is therefore essential to develop the RG system for neurotropic astroviruses to understand the 86 molecular determinants for neurotropism and neurovirulence. Here we report the RG system 87 for two non-classical human neurotropic astroviruses that relies on a single cell line and can be 88 used to rescue and propagate MLB1 and MLB2 human astroviruses. We also developed a set 89 of detection tools as well as replicon systems for both MLB astroviruses. Using this system we 90 identified and characterized attenuation hotspots located at the 3ʹ end of the MLB genomes. In 91 the future, this RG system will deepen the understanding of molecular virology of MLB-group

96
Evolution and cell culture adaptation of neurotropic MLB astroviruses. The selection for 97 attenuated viruses through serial passaging in highly susceptible cells is a well-known approach 98 for directed evolution 22 . We, therefore, hypothesized that this strategy could be applied to 99 attenuate MLB astroviruses. Clinical MLB isolates were passaged in susceptible cell lines as 100 previously described 2 (Fig. 1B). Sequencing of a passaged clinical MLB1 isolate revealed a 101 deletion of 30 nucleotides in the 3ʹ end of the genome spanning into the coding sequence of 102 CP. A similar region was affected in the passaged MLB2 clinical isolate -a single out-of-frame 103 deletion of 5 nucleotides in the 3ʹ part of the genome (Fig. 1D).

104
Another strategy for directed virus evolution is based on co-infection of closely related virus 105 species. The better replicating "partner" can either out-compete or complement the replication 106 of another virus. To test this hypothesis, we co-infected Huh7.5.1 cells at a multiplicity of 107 infection (MOI) 0.1 with MLB1 and MLB2 viruses (Fig. 1C). This resulted in simultaneous 108 replication and propagation of both strains on passaging without detected out-competition or 109 recombination for 10 consecutive passages. Interestingly, no changes were observed in MLB2 110 genomes; however, several in-frame and out-of-frame deletions were detected in the 3ʹ part of 111 the MLB1 genome further confirming the instability of the 3ʹ region in this virus (Fig. 1D).

112
Elucidating the functional significance of the identified deletions requires MLB astrovirus 113 detection tools and would be dramatically accelerated by the establishment of a robust RG 114 system. We, therefore, aimed to create these essential tools.

115
Cell culture models and detection tools for neurotropic MLB1 and MLB2 astroviruses. 116 First, we developed a set of essential tools for specific immune detection of virus infection.

117
The folded region of MLB1 capsid protein corresponding to amino acids 61-396 of ORF2-118 encoded polyprotein possessing a C-terminal 8×His-tag ( Fig. 2A) was used for bacterial 119 expression and affinity purification, resulting in homogeneous CPNTD protein (Fig. 2B). The 120 purified recombinant protein was used for the production of highly sensitive antibodies 121 allowing the detection of ≤ 1 ng of the purified CPNTD of MLB1 (Fig. 2C). Due to 95% identity 122 between corresponding domains of MLB1 and MLB2 CPs, polyclonal antibodies were 123 expected to cross-detect capsid proteins derived from both strains. Indeed, it specifically 124 recognized capsid proteins from MLB1-and MLB2-infected cells (Fig. 2C).

125
When passaging clinical isolates, we noticed that the Huh7.5.1 cell line supports MLB      (Table 1). A distinct cluster of 184 mutations at the C-terminal end of CP was identified in the second experiment (Table 1) 185 suggesting instability of RG MLB1 genomes during longer virus passaging.

186
To put these mutations and previously identified deletions (Fig. 1D) in the context of naturally   infectious virus release (Fig. 6E). 274 Next, we measured the levels of intracellular virus RNA in infected neurons. Consistent with 275 lower infectivity (Fig. 6B, D, E), the virus-specific RNA transcripts were significantly reduced 276 in MLB2-∆5 OF (Fig. 6F). Surprisingly, the intracellular virus RNA levels were increased in  Finally, to examine if MLB1 and MLB2 mutants have altered specific infectivity, we analyzed 284 the input Huh7.5.1-derived virus stocks for the presence of virus-specific RNA per number of 285 infectious particles. As expected, for all deletion mutants this ratio was significantly altered 286 when compared to the wild-type viruses (Fig. 6G). represents an opportunity to study differential genomic stability in closely related viruses.

308
The processing of capsid polyprotein, virion assembly and maturation in MLB viruses is 309 regulated by several cellular proteases, such as caspases. What is the role of C-terminal 310 deletions in the context of ORF2-encoded polyprotein? The removal of the significant part of 311 the acidic CP portion could affect CP localization, trafficking, particle formation and potential 312 pro-viral roles that are yet to be characterized for astroviruses. In infected cells, we observed a 313 major large polyprotein precursor and several smaller products (Fig. 2C, 3F), suggesting 314 intracellular cleavage of CP. The analysis of media-derived samples revealed a prevalence of 315 smaller CP cleavage products of about 55 kDa (Fig. 3G), indicating that CP is cleaved by 316 additional unknown intracellular and/or extracellular proteases. The C-terminal cleavage of CP 317 in MLB2 is likely to be regulated by cellular caspases (Fig. 4F), similar to classical human 318 astroviruses 7 . Despite the presence of the predicted cleavage sites (Fig. 4C), the processing of MLB1 capsid is not sensitive to cellular caspase inhibitors (Fig. 4F), resembling another 320 neurotropic astrovirus, VA1 29 . In contrast to neurotropic VA1 and MLB groups of 321 astroviruses, the infectivity of classical astroviruses strongly depends on exogenous trypsin 322 activation 2,29 . Since trypsin is a gut-specific enzyme, this may also explain the extra- (ii) sensitivity to caspase inhibition was observed for MLB2, but not MLB1 (Fig. 4E), (iii) 333 increased RNA replication in neurons and more significant increase of replicon activity was a 334 hallmark for MLB1 deletion mutants (Fig. 5E, 6F). It was logical to expect that the 3ʹ 335 attenuation in these two closely related MLB viruses could have differences in associated RNA 336 and protein-related effects. All deletion mutants identified in MLB1 were mapped to the 337 predicted structured stem-loops that could be beneficial for RdRp processivity in susceptible RNA replication-translation-packaging and resulted in lower infectious virus particle release 343 and infectivity in neurons (Fig. 6C-E). In contrast, attenuation of MLB2 virus with a small 5-344 nucleotide out-of-frame deletion resulted in modest differences in replicon activity (Fig. 5E) 345 and ten-fold decreased RNA levels in neuronal infection (Fig. 6F), suggesting that differences 346 in cleaved C-terminal part of CP (Fig. 1D) could play a major role in attenuation of MLB2.

347
This, however, does not exclude associated RNA defects (Fig. 5E, 6F  The selection for deletion-prone viruses through serial passaging in highly susceptible cells is  by TEV protease cleavage site and C-terminal 8×His-tag (Fig. 1A).

394
To create RG clones for MLB1 and MLB2, the 5ʹ and 3ʹ terminal consensus sequences 2 were 395 used to design specific primers to amplify MLB1 and MLB2 full-length genomes using  Purification of His-tagged CP NTD and generation of CP-specific antibody. 415 The MLB1 CPNTD protein was produced in Rosetta 2 (DE3) cells (Novagen) cultured in 2×YT 416 media with overnight expression at 18 °C induced with 0.4 mM IPTG. The protein was purified 417 first by immobilized metal affinity chromatography using PureCube Ni-NTA resin and then by 418 affinity chromatography using amylose resin (NEB). N-terminal MBP fusion tag was removed by the cleavage with TEV protease (produced in-house). The MLB1 CPNTD protein was further 420 purified by heparin chromatography using HiTrap Heparin HP 5 ml column (Cytiva) and, 421 finally, by size exclusion chromatography using a Superdex 200 16/600 column (Cytiva).

422
Protein solution in 50 mM Na-phosphate pH 7.4 , 300 mM NaCl, 5% glycerol was concentrated 423 to 2 mg/ml and used for immunisation.

424
Antibody against CPNTD was generated in rabbit using 5-dose 88-day immunisation protocol.

425
Sera was used for CPNTD-specific affinity purification, followed by purification of specific IgG 426 fraction (BioServUK Ltd).    settings. Only predicted cleavage sites with probability score of >0.7 were considered.

582
Statistical analyses. Data were graphed and analyzed using GraphPad Prism and MS Excel.