Enterovirus A71 adaptation to heparan sulfate comes with capsid stability tradeoff

Enterovirus; heparan sulfate; uncoating; virus adaptation; virus capsid stability 31 32 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted February 25, 2024. ; https://doi.org/10.1101/2024.02.23.581741doi: bioRxiv preprint

33 Heparan sulfates (HS) are linear, negatively charged polysaccharides connected to various 34 cell surface and extracellular matrix proteins. Expressed on a wide range of cells, they play a 35 pivotal role in various biological processes, and many viruses exploit them to attach and 36 concentrate onto cell surfaces before binding to main entry receptor 1. Despite a substantial 37 body of literature on the subject, the actual implication of HS binding on viral infections 38 remains a topic of debate. 39 Enterovirus-A71 (EV-A71) is an excellent example of the ongoing controversy regarding the 40 impact of HS receptor utilization in viral pathogenesis . This virus is a member of the 41 Picornaviridae family and the most neurotropic EV after poliovirus. It causes significant hand, 42 foot and mouth disease outbreaks, particularly in Asia -Pacific countries, and is associated 43 with severe neurological complications , notably in small children and immuno suppressed 44 patients2. The virus uses human scavenger receptor class B member 2 (SCARB2) as the main 45 receptor for internalization and uncoating3,4. Since SCARB2 is mostly localized on lysosomal 46 membrane and sparsely on plasma membrane 3,5, it seems to play only a minor role in EV -47 A71 cell attachment6. Consistently, numerous other EV A71 receptors have been described 48 in the literature, including HS3,7. When propagated in cell culture, EV -A71 rapidly acquires 49 adaptive mutations (i.e. patches of positively charged amino acids on the viral capsid) that 50 allow them to bind HS, sometimes with high avidity. These strong HS -dependent variants 51 grow efficiently in cell culture but show attenuated virulence in animal models, such as mice 52 and cynomolgus monkeys 8-11. Analysis of the differential expression of SCARB2 and HS i n 53 tissues from monkey or transgenic mice revealed little overlap. S trong HS expression was 54 detected in sinusoidal endothelial cells and vascular endothelia, where SCARB2 was not 55 detected9,10. Similarly, HS expression in the brain was mainly found in vascular endothelia 56 but SCARB2 expression was found predominantly in neuronal cells. The authors of these 57 studies concluded that binding to HS on endothelial cells in absence of SCARB2 leads to viral 58 trapping, abortive infection, and attenuation9,10. Similar observations were shown for other 59 viruses, including Murray Valley encephalitis 12, Japanese encephalitis 12, Sindbis 13, Theiler’s 60 murine encephalomyeliti14, tick-borne encephalitis15, West Nile16 and dengue17. 61 We previously isolated cell-adapted EV-A71 mutants with strong affinity for HS which 62 emerged upon passaging of intermediate HS binders derived from both patient and mouse- 63 adapted MP4 strains in cell culture 18,19. The mutants presented two amino acid changes in 64 the VP1 capsid protein: VP1-L97R mutation in the VP1 BC loop, shown to confer 65 intermediate affinity for HS together with a second ary mutation, VP1-E167G, located in the 66 VP1 EF loop, which significantly strengthened HS binding with reduction of negative 67 charges19,20. As previously observed for strong HS-binding variants, we showed that, in 68 contrast to the original mouse-adapted MP4 strain which exhibited virulence in mice, this 69 cell-adapted MP4-97R/167G double mutant was completely attenuated in mice 19. In the 70 current study, we used MP4 and MP4-97R/167G mutant as representatives of respectively, 71 weak and strong HS-binders, slow and fast-growing in cell lines and virulent and avirulent in 72 mouse models (as documented previously 19,20), to elucidate the consequence of virus 73 adaptation towards HS binding on the viral growth cycle. We demonstrated that these 74 mutations not only increase binding to HS, but also reduce capsid stability, leading to 75 improved uncoating and faster cell internalization in a HS-independent manner. Of note, 76 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted February 25, 2024. ; https://doi.org/10.1101/2024.02.23.581741doi: bioRxiv preprint another strong HS binder harboring VP1-E145Q substitution also showed decreased capsid 77 stability compared to the wildtype HS-independent variant. These data provide another 78 possible explanation for the in vivo attenuation of strong HS-binders which may originate 79 from viral trapping but also from decreased capsid stability which would be detrimental to 80 the virus in challenging environments such as the gastrointestinal tract. 81 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted February 25, 2024. ; https://doi.org/10.1101/2024.02.23.581741doi: bioRxiv preprint

82 Lysosomotropic drugs reduce infectivity of the HS-independent MP4 but enhance 83 infectivity of the strong HS-binder MP4-97R/167G 84 First, we sought to assess whether viruses displaying different dependence on HS exploit 85 different growth cycle pathways. We compared the effect of lysosomotropic drugs, namely 86 hydroxychloroquine (HCQ) and bafilomycin A1 (BAF -A1) on MP4 and MP4-97R/167G double 87 mutant. As presented in Fig. 1A , Vero cells were pre -treated for 1 hour with each drug 88 before infection. T hese drugs showed no cytotoxic eff ect at the concentrations used in the 89 assay (Fig. S1A ). Inoculation was then performed for 1 hour in presence of the drug and 90 inoculum was removed and replaced with fresh drug -free media. The number of infected 91 cells was determined at 24 hours post-infection (hpi) using immunofluorescence. To confirm 92 inhibition of endosomal acidification by these drugs, the presence or absence of acidic 93 lysosomes was assessed by immunostaining of lysosomal-associated membrane protein 1 94 (LAMP1) and by staining with lysotracker, a dye specific for acidic compartments. The 95 amount of LAMP1 and lysotracker double-positive lysosomes decreased drastically following 96 treatment with the drugs, confirming the inhibition of endosomal acidification ( Fig. 1B). The 97 effect of the drugs on viral replication was then compared for the two variants (Fig. 1C & D). 98 MP4 infectivity was significantly reduced by both drugs in a dose -dependent manner, while 99 MP4-97R/167G infectivity was in contrast enhanced. Similar results were obtained in RD 100 cells (Fig. S1B), indicating that drug effects are not cell -dependent. The different sensitivity 101 of the two variants to acidification inhibitors was more pronounced with HCQ, so we 102 performed a detailed examination of the mechanism of action with this drug. 103 To determine whether the effect of HCQ was related to the usage of HS as attachment 104 receptor, we repeated the virus inhibitory assay using cells depleted of HS by either 105 heparinase digestion ( Fig. 1E) or treatment with sodium chlorate ( Fig. 1F). The dist inct 106 sensitivity to HCQ was reproduced regardless of the presence or absence of HS on the cell 107 surface. Of note, we confirmed that as for the human strain 20, the HS-independent and HS-108 dependent MP4- derivatives both need SCARB2 to infect cells and cannot replicate in 109 SCARB2 CRISPR-Cas9 knock-out cells (Fig. S2). Altogether, these data indicate that the capsid 110 mutations change the sensitivity to HCQ, independent of the attachment receptor used. 111 MP4 enters via SCARB2-mediated and pH-dependent endocytosis, while MP4-97R/167G 112 utilizes an alternative SCARB2-dependent pathway. 113 In our previous studies, we showed that MP4 virus uses SCARB2 as entry receptor and here 114 we demonstrated that it is inhibited by HCQ. This strongly suggests that MP4 uses SCARB2 -115 mediated pH -dependent endocytosis for entry and uncoating as demonstrated for many 116 other EV-A71 variants7,21. After having excluded that HCQ effect did not affect virus binding 117 (Fig. 2A), we conducted single -cycle infection assay and showed that differential effect of 118 HCQ on MP4 and MP4-97R/167G became prominent from 4 hpi onward (Fig. 2B). To further 119 dissect which step of the viral growth cycle was differentially affected by the treatmen t, we 120 performed a time -of-addition assay. As shown in Fig. 2C, HCQ significantly los t its effect 121 when administered later than 2 hpi, confirming that the effect occurs during the early phase 122 of the viral cycle. To more specifically assess whether the drug a ffects viral entry, we 123 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted February 25, 2024. ; https://doi.org/10.1101/2024.02.23.581741doi: bioRxiv preprint transfected in vitro transcribed genomic RNA containing the nanoluciferase (Nluc) gene as 124 reporter ( Fig. 2D). RNA transfection allows to bypass receptor-mediated entry. In these 125 conditions, no difference was observed for both variant s, whether HCQ was present or not 126 (Fig. 2E). This observation also indicates that the drug does not impact genome replication. 127 In contrast, infection with infectious Nluc reporter virus reproduced the differential HCQ 128 inhibition as observed in the original non -modified viruses ( Fig. 2F & Fig. 1D ). Taken 129 together, these data indicate that MP4 enters via pH -dependent endocytosis, while MP4 -130 97R/167G entry pathway is independent on endosomal acidification. 131 HCQ differentially impacts MP4 and MP4-97R/167G uncoating 132 To further dissect the mechanism of action of HCQ on the entry of each variant, we next 133 examined the effect of the drug on viral uncoating. We generated neutral red labelled virus 134 stocks and performed neutral red uncoating assay as previo usly described 22. Briefly, viral 135 stocks were produced in presence of neutral red in the dark to induce co -encapsidation of 136 viral genome and neutral red insi de viral particles. P hotoactivation of neutral red induces 137 the dye to cross-link viral genomes to the capsid and subsequent blockade of viral 138 uncoating22. Accordingly, upon infection with neutral red labelled viruses, light exposure 139 only inactivates viruses that have not yet completed uncoating, while virus genomes already 140 released in the cytoplasm are unaffected. This allows to precisely define the timepoint of 141 viral uncoating. Cells were pre -treated with HCQ and then incubated with neutral red -142 labelled viruses for 1 hr at 37˚C for infection. Light inactivation was performed at selected 143 timepoints post-infection, and infected cells were quantified 24 h later by immunostaining 144 (Fig. 3A). In the absence of HCQ, the majority of viruses had undergone uncoating between 145 2 and 3 hpi for both variants (30 -70% of uncoated viruses for MP4 and 60 -80% for MP4 -146 97R/167G, respectively) ( Fig. 3B). In the presence of HCQ, MP4 uncoating was completely 147 blocked, even when photoactivation was performed at 4 hpi (Fig 3B, left panel). In contrast, 148 the uncoating rate was not inhibited in the presence of HCQ for MP4 -97R/167G (Fig. 3B, 149 right panel) and the final viral yield was even increased as already observed in Fig. 1D. 150 To further validate these results, we then combined fluorescent in situ hybridization (FISH) 151 of viral genomic RNA and immunofluorescence staining of viral capsid at early timepoints. 152 Full particles are characterized by colocalization of virus genomic RNA (vRNA) and viral 153 capsid (as shown in 1 hpi at 4 ˚C), while the colocalization is lost following the uncoating 154 process (Fig. 3C). Quantification of vRNA and capsid colocalization highlighted no significant 155 difference between the two variants at 2 hpi in presence or absence of HCQ ( Fig. 3E). 156 However, at 4 hpi (prior to the initiation of replication , see Fig. 3B ), MP4 uncoating 157 appeared to be inhibited by HCQ, as highlighted by a decrease of empty capsids and an 158 increase of capsid/RNA colocalization in presence of the drug (Fig. 3D & E). An opposite 159 effect was observed for MP4-97R/167G, with a reduced capsid/RNA colocalization in the 160 presence of HCQ , indicating that more viruses had undergone uncoating in the presence of 161 HCQ at this time point. Altogether these data indicate that MP4 -97R/167G can uncoat s at 162 neutral pH and that acidification is instead increasing its replicati on capacity, while MP4 163 needs acidic pH to uncoat. 164 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted February 25, 2024. ; https://doi.org/10.1101/2024.02.23.581741doi: bioRxiv preprint MP4 relies on late endosomes for uncoating, whereas MP4-97R/167G undergoes 165 uncoating in early endosomes 166 HCQ is known to inhibit endosomal acidification by accumulating in endosomes in a 167 protonated form. This accumulation leads to endosomal swelling and inhibition of fusion 168 between endosomes and lysosomes within cells, as previously described 23,24 and as shown 169 in Fig. 4A. We thus hypothesized that the two variants could exploit different entry routes, 170 which could explain the different sensitivity to HCQ. We showed that MP4 needs acidic pH 171 to uncoat and is thus expected to release its RNA in late endosomes/lysosomes. In contrast, 172 MP4-97R/167G can uncoat in absence of pH acidification and accordingly in a non-acidic 173 environment. To test this hypothesis, we infected Vero cells transiently expressing a variant 174 of small GTPase Rab5a, a protein involved in the maturation of early endosomes (EE) into 175 late endosomes (LE). This Rab5a -Q79L mutant is constitutively active (CA) and blocks LE 176 maturation (Fig. 4B). Viral capsids of MP4 and MP4 -97R/167G were observed within EE in 177 both Rab5a WT and CA -expressing cells at 0.5 hpi and 2 hpi ( Fig. S3 and 4C). However at 7 178 hpi, the percentage of cells stained for double stranded RNA (dsRNA), a marker of virus 179 replication, was significantly reduced for MP4 in Rab5a CA -expressing cells but not for MP4 -180 97R/167G (Fig. 4D). This indicates that MP4-97R/167G genomes were successfully released 181 in the cytoplasm to undergo translation and replication, even in absence of EE fusion to LE. 182 Conversely, a transition from EE to LE with a gradual pH decrease is necessary for MP4 to 183 release its genome. 184 VP1-L97R/E167G substitutions confer affinity for HS but decrease capsid stability 185 Our data highlighted that both MP4 and MP4-97R/167G enter via a SCARB2-dependent 186 pathway, localize in early endosomes at early times post -infection but exhibit di stinct 187 sensitivities to HCQ, a feature independent of their differential use of HS as attachment 188 receptor. We therefore speculated that the varying pH-dependency may be attributed to 189 differences in virion stability. We used DynaMut server 25 to assess the relative influence of 190 VP1-97R and VP1 -167G mutations on interaction networks in their respective local 191 environments (Fig. S4A) as well as the impact on overall capsid stability ( Fig. S4B). Indeed, 192 VP1-L97R mutation is predicted to reduce hydrophobic interactions between the original 193 leucine residue on this position and its neighbors VP1-245Y and VP1 -246P. Further, this 194 residue is in close proximity to VP1 -244K, thereby adding more positive charge to this site. 195 On the other hand, VP1-E167G mutation causes a loss in hydrogen bonding capacity to VP1-196 165S and reduces the net negative charge. Altogether, DynaMut analysis predicted reduced 197 interaction capabilities by the substituted residues and changes in electrostatic properties 198 within the region. This is consistent with a nalyses of vibrational entropy change , indicating 199 that the presence of two mutations result s in enhanced local dynamics, which has 200 previously been correlated with reduced capsid stability (Fig. S4B)26,27. 201 We therefore speculated that MP4-97R/167G mutant features a lower stability and could 202 bypass the need for acidic pH fo r uncoating. To test this hypothesis, we subjected these 203 variants to neutral and acidic conditions and assessed their virion structure using negative 204 staining electron microscopy (nsEM) ( Fig. 5A & B). At acidic pH, there was no notable 205 alteration in the ca psid morphology of MP4, which maintained a stable particle diameter of 206 ~31-33 nm across both pH conditions. On the other hand, for MP4 -97R/167G, both 2D 207 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted February 25, 2024. ; https://doi.org/10.1101/2024.02.23.581741doi: bioRxiv preprint images and 3D reconstructions highlighted a loss of density at the center of the viral 208 particles, as well as an expansion in size for a subset of particles (diameter ranging from 31 209 to 41 nm at pH5 versus 31 to 33 nm at pH7), indicating partial virus uncoating. We then 210 performed temperature sensitivity assay by heating viruses at different temperatures for 1 h 211 before inoculation on cells. Quantification of infected cells at 24 hpi further confirmed that 212 MP4 capsid is more resistant to higher temperature as 80% of MP4 population survived a 213 50°C thermal stress compared to only 50% for MP4-97R/167G ( Fig. 5C). Consistently, the 214 predictions of Gibbs free energy change ( ΔΔG) induced by these mutations further 215 supported that both mutations induce slight destabilization of the capsid structure, 216 regardless of pH and temperature (Table S1). 217 As MP4-97R/167G is less stable, we hypothesized that binding to SCARB2 may be sufficient 218 to trigger its capsid opening. We conducted a competitive experiment and compared the 219 infectivity of the two variants after incubation with soluble SCARB2 (sSCARB2) at neutral pH 220 for 1 hr at 37°C. We observed that MP4-97R/167G but not MP4 variant lost infectivity upon 221 pre-exposure to sSCARB2 ( Fig. 5D ). Altogether, our results confirmed that MP4-97R/167G 222 capsid is less stable and highly sensitive to thermal and acidic stresses as well as receptor 223 binding, which are sufficient triggers to initiate virus capsid disruption and subsequent viral 224 uncoating. 225 Resistance to HCQ and reduced capsid stability extend to other strong heparan sulfate-226 binding strains. 227 To check if our observations could extend to other HS-binding variants, we checked the 228 effect of mutation present at position VP1-145, a residue known to play a key role in 229 modulating viral HS-binding capacity and in vivo virulence. EV-A71 variants (sub-genogroup 230 B4) with wild-type VP1 -145E is a weak HS binder and is attenuated in mice while the cell-231 adapted VP1-145Q variant is a strong HS-bind er and virulent in mice 11. As shown in Fig. 6A, 232 HCQ enhanced infectivity of VP1-145Q variant but reduced infectivity of VP1-145E variant, 233 consistent with what we observed with the MP4- 97R/167G. Both temperature sensitivity 234 (Fig. 6B) and sSCARB2 inhibition assays (Fig 6C) also confirmed that the capsid of VP1-145E 235 variant is much more stable compared to VP1-145Q variant, in line with free energy change 236 prediction as shown in Table S1 . These data further reinforced the observation that HS-237 binding phenotype is inversely correlated with virus capsid stability. 238 239

240 Acidic pH is an important trigger for viral uncoating and many enveloped or non-enveloped 241 viruses, including influenza A virus28,29, human adenovirus30, foot-and-mouth disease virus31, 242 Semliki forest virus 32, that enter the cell through pH-dependent endocytosis 33. Similarly, for 243 EV-A71, binding to SCARB2 and subsequent endosomal acidification are required for 244 uncoating4,7. In this study, we provide new insights on the impact of mutations in the VP1 245 capsid protein leading to strong HS affinity, on the uncoating process of EV-A71. We show 246 that, unlike the mouse-adapted EV-A71 MP4 strain, the MP4-97R/167G -derived double 247 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted February 25, 2024. ; https://doi.org/10.1101/2024.02.23.581741doi: bioRxiv preprint mutant, which has a high affinity for HS, does not require acidification for uncoating and can 248 release its genome under neutral or weakly acidic environment of early endosomes. 249 To demonstrate the importance of acidification on both MP4 and MP4-97R/167G variant 250 uncoating, we u sed two lysosomotropic drugs, namely BAF-A1 and HCQ, that increase 251 endosomal pH by distinct means. On one hand, BAF-A1 inhibits the vacuolar H+ ATPase (V-252 ATPase), preventing the acidification process and thereby elevating the endosomal pH 34-38. 253 On the other hand, HCQ, a less toxic derivative of the antimalarial drug chloroquine, acts as 254 a weak base that can be protonated and trapped in the acidic environment of cellular 255 organelles39-41. In addition to this effect, HCQ can impact other cellular pathways, such as 256 autophagy, a cellular process that has been demonstrated to be induced by EV-A71 to 257 create a favorable environment for its replication 42,43. However, we show here that the 258 differential effects on MP4 and MP4-97R/167G occur during the uncoating process rather 259 than in later stages of the cycle, such as virus genome replication 40. Our results thus 260 underline that, despite their distinct modes of action, both HCQ and BAF-A1 influenced virus 261 entry through their effect on endocytic compartments, as both compounds ultimately 262 inhibit the reduction in endosomal pH levels. In addition, the fact that differential sensitivity 263 to HCQ was retained even in cells devoid of HS at their surface by treatment with 264 heparinase or sodium chlorate pointed out that this pH-independent mode of entry of MP4-265 97R/167G is not linked to the use of HS as an attachment receptor. Interestingly, our 266 experiments using the Rab5a CA to block the transition of EE to acidic LE44,45, indicate that 267 binding to SCARB2 is sufficient to trigger MP4 -97R/167G genome release into the cytosol 268 even in the near neutral pH of the EE (pH ∼6.0 to 6.5), whereas MP4 requires the acidity of 269 LE and/or lysosomes (pH ∼5.0 to 5.5) to uncoat efficiently 46. These observations led us to 270 hypothesize that the two variants exhibit intrinsic differences in capsid stability. We 271 conducted various tests to compare how each variant reacted to heat, sSCARB2 and low pH, 272 and found that MP4 -97R/167G was more sensitive to all these conditions. Particularly, we 273 used nsEM to study the properties of viral particles and observed an expansion of MP4-274 97R/167G capsid fo llowing the exposure to pH 5 . T hese data, plus the virus structural 275 dynamics prediction and free energy change computation , all indicate that MP4-97R/167G 276 presents reduced capsid stability compared to MP4. 277 One strength of our study lies in the fact that we were able to extend these results to 278 another HS-dependent EV-A71 strain, VP1-145Q variant. Given that these two HS-279 dependent variants share the characteristic of having incorporated a less acidic amino acid 280 within the VP1 capsid protein, we hypothesized that an increase in positive charges within 281 the capsid not only enhances affinity for HS but also alters capsid stability, consequently 282 impacting the virus entry mechanism. In the same line, a thermostable EV-A71 variant (VP1-283 K162E, change of a basic to an acidic residue) isolated from serial passages at higher 284 temperatures was shown to be less efficient at uncoating with poorer cell infectivity but 285 more virulent in mice 47. Interestingly, this variant showed a more expanded conformation 286 compared to the original non -mutated virus47. While the thermostable variant showed no 287 difference in bi nding to SCARB2 receptor, the binding affinity to heparin was greatly 288 reduced, an observation consistent with what we noticed for MP4. Additional experiments 289 with cell-adapted, HS-binding viruses will help to define whether HS -binding is always 290 associated with a loss of virion stability and whether these findings could event extend to 291 other group of viruses . Interestingly, mutations conferring similar in vitro phenotypes were 292 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted February 25, 2024. ; https://doi.org/10.1101/2024.02.23.581741doi: bioRxiv preprint observed for other enteroviruses such as rhinovirus A16 (RV-A16) and coxsackievirus B3 (CV-293 B3). For RV-A16, capsid mutations conferring resistance to endosomal acidification 294 inhibitors also abrogated the need for acidic pH for uncoating . More importantly, these 295 mutations were also associated with higher sensitivity to low pH, high temperatures, and 296 binding to soluble receptors48. For CV-B3, a fast-growing variant was shown to exhibit faster 297 genome release and destabilized capsid, and this led to attenuated virulence in mice 49. This 298 existing literature aligns with our findings, suggesting that alterations in capsid stability due 299 to amino acid changes may significantly impact various aspects of the virion and its life cycle, 300 including sensitivity to environmental factors, receptor interactions, and infection rates. 301 In the light of these published studies and our data, we propose the following model 302 depicting the relationship between HS -binding, capsid stability and viral fitness in vitro and 303 in vivo (schematized in Fig. 7). Viruses undergo mutations and positive selection to adapt to 304 different environments50,51. EV-A71 can take advantage of the high pla sticity of its capsid to 305 optimize its fitness upon environmental changes. Many strains adapt to use HS in vitro due 306 to the abundant expression of this attachment receptor in cell lines. To do so, they usually 307 acquire additional positively charged amino aci ds within outward -facing VP1 domains 308 proximal to the capsid 5 -fold axis. In addition to help the virus to attach on the cell surface 309 and find the SCARB2, we show here that these mutations concomitantly decrease virion 310 stability. This further contributes to higher multiplication in cell lines by triggering uncoating 311 rapidly after internalisation, within EE, without the need of acidic pH. Interestingly in our 312 experiments, acidification inhibitors improved rather than inhibited viral fitness of MP4-313 97R/167G. Although additional experiments are required to define the mechanism behind 314 this observation, it could occur via the protection of virions that have not yet uncoated 315 when endosome s fuse with lysosome s. In this context, absence of acidification would 316 improve the chance of these virions to release their genome in the cytoplasm, while 317 exposure to acidic pH would induce viral opening within the late endosomes. The situation 318 may differ significantly in vivo as strong HS binders are attenuated. Koike and colleague s 319 have demonstrated that strong binding to HS induces virus trapping in vivo 9-11,19. Our data 320 suggest that the associated decreased stability may further contribute to viral attenuation. 321 To be virulent in vivo, a non-enveloped virus must have a sufficiently stable capsid to resist 322 unfavourable environmental conditions, both during dissemination within a host and during 323 transmission between hosts. This last point is particularly important for EV -A71, which is 324 transmitted via the fecal -oral route and must therefore resist the acidic pH of the 325 gastrointestinal tract before reaching the intestinal mucosa, its main multiplication site. 326 Capsid stability thus ensures that virus genome release occurs only in a proper environment, 327 but in turn renders the virus dependent on both SCARB2 and acidic pH for uncoating4,7. 328 Of note, although strong HS-binder are clearly attenuated in mice, the situation in humans is 329 more puzzling as strains with Q or G at position VP1-145 have been associated with severe 330 neurological cases and the methodology used in those studies excluded emergence of 331 mutation during cell culture 52,53. Furthermore, we previously showed that intermediate HS -332 binding affinity can result in increased virulence even in mouse models 19. It is still unclear if 333 there is a lack of trapping and/or limited impact on viral stability under these conditions. To 334 conclude, to reach optimal fitness in v itro and in vivo , the virus needs to find the correct 335 balance between HS binding and capsid stability (Fig. 7 ). Our study improves the current 336 knowledge on the mechanism behind the in vivo attenuation of cell culture-adapted viruses. 337 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted February 25, 2024. ; https://doi.org/10.1101/2024.02.23.581741doi: bioRxiv preprint It also opens the doors to new antiviral strategies targeting endosomal acidification as well 338 new principles for vaccine design based on attenuated -acid -independent variants to help 339 combat EV-A71. 340 341

342 Chemical reagents 343 Chemical reagents used in this study were listed as following: hydroxychloroquine (Tocris), 344 bafilomycin-A1 (InvivoGen), sodium chlorate ( NaClO3, Sigma -Aldrich), neutral red (Sigma 345 Aldrich) and LysoTracker Deep Red (Thermo Fisher Scientific). 346 Cell lines and virus 347 Vero (monkey kidney; ATCC CCL-81) and human rhabdomyosarcoma cells (RD; ATCC no.: 348 CCL-136) were propagated in Dulbecco’s Modified Eagle Medium (DMEM) and GlutaMAX 349 (31966021, Thermo Fisher Scientific) containing 10% fetal bovine serum (FBS). RD -SCARB2-350 KO20 and RD-ΔEXT1+hSCARB254 cells were maintained in DMEM supplemented with 10 351 µg/ml puromycin ( 58-58-2, Invivo Gen). All infected cells were maintained in media 352 supplemented with 2.5% FBS. All cells were maintained at 37˚C in 5% CO 2. Viruses used in 353 this study including MP4, MP4 -97R/167G ( Genbank accession number: JN544419 ; 354 subgenogroup C2), IEQ (EV -A71 VP1-145Q variant; Genbank accession number: AF316321; 355 subgenogroup B4 ) and IEE (EV -A71 VP1 -145E variant) strains were prepared as previousl y 356 described11,19. For Nluc reporter virus, Nluc gene was inserted between 5’ UTR and VP4 of 357 the virus as previously described 55. Viruses were generated in RD-ΔEXT1+hSCARB2 cells54, 358 propagated for an additional passage and used as working stocks. All virus stocks were 359 sequenced for confirmation (Fasteris) prior to experiments. 360 Plasmids 361 Plasmids encoding eGFP-Rab5a and eGFP-Rab5a Q79L are kind gifts from Pierre-Yves Lozach 362 (University Claude Bernard Lyon 1). Both IEQ and IEE plasmids ( Genbank accession number: 363 JN544419: AF316321; subgenogroup B4 ) strains are kind gifts from Yoke Fun Chan 364 (University of Malaya). 365 Virus inhibitory assay and time-of-addition assay 366 For virus inhibitory assay, cells were pre-treated either with drugs for 1 hour at 37 ˚C. 367 Viruses (MOI 0.1) were inoculated onto cells in presence of drugs for 1 hour at 37 ˚C. Upon 368 infection, inocula were removed and cells were rinsed thoroughly with phosphate buffered 369 saline buffer (PBS) before incubated with fresh media up to 24 hpi. For time -of-addition 370 assay, HCQ was either pretreated ( -1hpi), introduced during virus infection (0hpi) or 371 introduced onto cells at post -infection (1, 2 and 3hpi) for 1hr. After incubation, cells were 372 rinsed with PBS before loaded with maintenance media and incubate up to 24 hpi. For both 373 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted February 25, 2024. ; https://doi.org/10.1101/2024.02.23.581741doi: bioRxiv preprint assays, infected cells were fixed for immunofluorescence staining for virus-positiv e cells 374 detection. 375 Virus binding and replication assay 376 All the experiments were done on Vero cells seeded in 96 wells plate. For virus binding assay, 377 cells were incubated with 1 × 108 RNA copy number/ml virus for 1 hour at 4 ˚C. The inocula 378 were removed an d rinsed with cold PBS twice, and then subjected to cell lysis for RNA 379 extraction and qRT -PCR quantitation. For virus replication assay, cell monolayers were 380 incubated with virus for 1 hour at 37 ˚C. The inocula were removed, rinsed with PBS, and 381 then furt her incubated up to 24 hpi at 37 ˚C. Infected cells were lysed and viral RNA was 382 quantified by qRT-PCR. 383 Heparan sulfate removal 384 Both enzymatic and chemical methods, heparinase assay and sodium chlorate (NaClO 3) 385 treatment respectively, were used to cleave H S from cell surface. For heparinase assay, the 386 cells were first rinsed with PBS and then incubated with 3.5 mIU/ml of heparinase III 387 (AmsBio) diluted in 0.1 M sodium acetate pH 7.0, 1 mM calcium acetate and 0.2% BSA for 1 388 hour at 37 ˚C. Meanwhile, mock -treated cells were incubated with heparinase buffer. Upon 389 incubation, cells were washed twice prior to virus infection. For NaClO 3 treatment, cells 390 were propagated in presence of 30 mM NaClO3 at least one passage before experiment. The 391 cells were also pre -seeded in media supplemented with NaClO 3 and incubated overnight 392 before the experiment. 393 Immunofluorescence and confocal imaging 394 To detect infected cells, cells were fixed with absolute methanol (Sigma Aldrich) at room 395 temperature for 10 min and then incubated with blocking buffer consisting of 5% BSA 396 (PanReac Applichem) and 0.05% TritonX -100 (PanReac Applichem) for 20 min. Fixed cells 397 were first incubated with anti -EV-A71 capsid monoclonal antibody MAB979 (1:1000; Sigma) 398 for 1 hour at 37 ˚C and then with Alexa Fluor 488-conjugated secondary antibodies (1:2000; 399 Thermo Fisher Scientific) dissolved in DAPI solution for 1 hour at 37˚C. To detect dsRNA, 400 infected cells were fixed with 4% paraformaldehyde (Santa Cruz) and incubated with 401 blocking buffer, cells were inc ubated with anti -dsRNA monoclonal antibody J2 (1:500; 402 Scicons) for 1 hour at 37 ˚C and then with Alexa Fluor 594 -conjugated secondary antibodies 403 (1:2000; Thermo Fisher Scientific) dissolved in DAPI solution for 1 hour at 37˚C. Stained cells 404 were acquired using ImageXpress Pico (Molecular Devices) and percentages of positive cells 405 were determined using CellReporterXpress software. For confocal imaging, 406 immunofluorescence staining was performed with EE and lysosomes were stained using 407 anti-EEA1 (1:100; Santa Cruz) and anti -LAMP1 (1:100; Cell Signalling), respectively, for 1 408 hour at 37 ˚C and then with Alexa Fluor 488 -conjugated secondary antibodies (1:200) 409 dissolved in DAPI solution for 1 hour at 37˚C. The stained slides were mou nted under 410 coverslip (Hecht Assistent) with Fluoromount G mounting medium (Southern Biotech) and 411 analyzed using Zeiss LSM 800 confocal microscopy. 412 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted February 25, 2024. ; https://doi.org/10.1101/2024.02.23.581741doi: bioRxiv preprint RNAscope FISH detection and colocalization experiments 413 For FISH, cells were seeded on Nunc LabTek II chamber slides (Thermo Fisher Scientific) and 414 fixed with 4% paraformaldehyde. To detect viral RNA in infected cells, fixed cells were 415 processed for RNAscope FISH using RNAscope Multiplex Fluorescent V2 assay (Biotechne) 416 according to manufacturer’s protocol. In b rief, the cells were hybridized with V -EV71-C1 417 probe (Biotechne) at 40°C for 2 hours and then the signals were revealed using TSA Vivid 418 570 kit (Tocris). The slides were then incubated with blocking buffer and incubated with 419 MAB979 (1:100) at 4 °C overnight, followed by incubation with Alexa Fluor 488 -conjugated 420 secondary antibodies (1:200) at room temperature for 30min. After incubated with DAPI for 421 30s, the stained slides were mounted under coverslip with mounting medium and analyzed 422 using Zeiss LSM 800 confocal microscopy. Images were acquired and analysed using ZEN 3.2 423 software. 424 Luciferase assay 425 Enterovirus-A71 nanoluciferase (Nluc) reporter particles were used to study virus replication 426 bypassing cell entry in presence and absence of drug. Briefly, the r eporter virus plasmid was 427 linearized and in vitro transcribed to generate RNA using T7 RiboMax Express Large Scale 428 RNA Production System (Promega). Transcribed RNA was purified using RNeasy Mini Kit 429 (Qiagen) and then transfected in RD cells using Lipofecta mine 2000 (Thermo Fisher 430 Scientific). At certain timepoints, cell supernatants were harvested for luciferase activity 431 detection using Nano-Glo Luciferase Assay System kit (Promega) on Glomax Multi-Detection 432 System (Promega). 433 Neutral red uncoating assay 434 To generate neutral red (NR)-labelled viruses, virus stocks were propagated in cells in 435 presence of 5 µg/ml neutral red (Aldrich). The virus stocks were harvested at 3 dpi and 436 titered. For uncoating assay, NR -labelled viruses were infected at 37 ˚C for 1 hr i n the dark 437 then washed twice with PBS and loaded with FluoroBrite DMEM (Thermo Fisher Scientific) 438 supplemented with 2.5% FBS. At certain timepoints, infected cells were exposed to light for 439 30 min and then allowed to incubate up to 24 hpi. Infected cells w ere analysed using 440 immunofluorescence as stated earlier. 441 Virus infection in Rab5a-transfected cells 442 Vero cells (1.5 × 106) were transfected with 25µ g of Rab5a-eGFP plasmids using 443 Lipofectamine 3000 (Thermo Fisher Scientific). The next day, transfected cells were 444 harvested, resuspended in buffer ( PBS, 2nM EDTA, 1% BSA ), and subjected to fluorescence -445 activated flow cytometry (FACS) on S3 Cell Sorter (Biorad). EGFP -positive cells were sorted, 446 collected and then further propagated at least one day before virus i nfection. For virus 447 infection, cells were infected with virus (MOI 1.5) for 1 hour at 37 ˚C. The inocula were 448 removed, rinsed with PBS, and cells were further incubated up to 7hpi at 37 ˚C. Cells were 449 then stained with anti-dsRNA as described above. 450 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted February 25, 2024. ; https://doi.org/10.1101/2024.02.23.581741doi: bioRxiv preprint Temperature sensitivity assay and shSCARB2 inhibition assay 451 Viruses (MOI 0.5) were incubated at different temperatures (4°C, 37°C, 45°C, 50°C and 55°C) 452 for 1hr. Upon incubation, viruses were immediately transferred onto ice for cooling down 453 before inoculated onto cells for 1hr at 37 °C. Cells were washed and allowed to incubate in 454 maintenance media up to 24hpi before virus -positive cell detection using 455 immunofluorescence. For SCARB2 inhibition assay, viruses were incubated with 1 µg of 456 soluble recombinant human SCARB2 -FC chimera protein (bio -techne) at 37 °C for 1hr. The 457 mixture was then inoculated onto cells at 37 °C for 1hr. Upon incubation, cells were was hed, 458 and allowed to incubate in maintenance media up to 7hpi before lysed the cells for viral 459 RNA quantitation. 460 Electron microscopy (EM) 461 For structural analyses, virus stocks were first inactivated by formaldehyde treatment. 462 Formaldehyde at 100 µg/ml final concentration was added to the vir us stock and incubated 463 at 37°C for 3 days. Inactivated viruses were purified through 30% sucrose cushion at 32,000 464 rpm in SW32 Ti rotor (Beckman Coulter) for 14 hr at 4˚C, followed by sedimentation through 465 a discontinuous 20-45% (w/v) sucrose at SW41 Ti rotor (Beckman Coulter) for 12 hr at 4 ˚C. 466 The purified stocks were then subjected to HiPrep 16/60 Sephacryl S -500 HR column (Sigma 467 Aldrich) with 25 mM Tris -HCl + 150 mM NaCl (pH 7.5) as the running buffer. Fractions 468 corresponding to EV A71 particles were pooled and concentrated to 0.3 – 1.1 mg/mL using 469 Amicon Ultra centrifugal filter units with 100 kDa cutoff (Millipore Sigma). For pH -based 470 assays we prepared Tris -Acetate-based buffers at pH 5 and pH 7.5. The buffers compris ed 471 150 mM NaCl and a 100 mM mix of Tris base and acetic acid at the ratio necessary to reach 472 the desired pH. Each EV A71 variant was diluted to 100 µg/ml in the two buffer and 473 incubated for 30 minutes. Following incubation, the samples were applied onto ne gative 474 stain EM grids (Cat # CF300 -Cu-50, Electron Microscopy Sciences). Prior to sample 475 application the grids were glow discharged for 30 seconds. 2% solution of uranyl formate 476 was used for staining. The grids were imaged on a Talos L120C G2 microscope (T hermo 477 Fisher Scientific) running at 120 kV and featuring the CETA 4k camera. EPU software from 478 Thermo Fisher Scientific was used for data acquisition, and all data processing was 479 performed in the cryoSPARC package 56. Each dataset comprised 100 -200 micrographs, and 480 2’000-10’000 virus-corresponding particles. Particles were extracted from micrographs and 481 subjected to 2D classification. 3D reconstruction was performed using Ab initio al gorithm 482 with icosahedral symmetry imposed. 483 Computational analysis of virus capsid protein structure stability 484 To assess the virus capsid protein structure stability, EV-A71 crystal structures with PDB ID 485 of 3J22 and 4AED were used for MP4 and VP1 -145 variants, respectively. I -mutant 2.0 486 server57 was used to predict the free energy stability change upon introduction of mutation 487 into virus capsid VP1 protein. Visualization of mutational effects on interatomic interactions 488 and prediction of molecule flexibility were performed on DynaMut server25. 489 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted February 25, 2024. ; https://doi.org/10.1101/2024.02.23.581741doi: bioRxiv preprint Schematic diagram and statistical analysis 490 All schematic diagrams and illustrations were created via BioRender.com. All data and 491 statistical analyses were generated using GraphPad Prism 9. All drug treatment experiments 492 were analy zed with one -way and two -way ANOVA. For dose -dependent inhibitory assay, 493 area under curve (AUC) was calculated and analyzed using one -way ANOVA. Degree of 494 colocalization of virus capsid and vRNA in individual cells was measured using Mander’s 495 overlap coefficient calculation in ZEN 3.2 software. Data were presented as mean ± SEM. *p 496 < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. and not significant (n.s.). 497 498

499 This work was funded in part by the Swiss national foundation (Grant N° 310030_184777 500 to CT) and by the University of Geneva (Salary to HKT). We would like to thank Prof Satoshi 501 Koike and Dr Kyousuke Kobayashi from Tokyo Metropolitan Institute of Medical Science, 502 Japan for providing RD-ΔEXT1+hSCARB2 cells, Prof Jen-Ren Wang from National Cheng Kung 503 University, Taiwan for providing infectious clone plasmids EV -A71/MP4, Prof Pierre-Yves 504 Lozach from Université Claude Bernard Lyon 1 for providing plasmids encoding eGFP-Rab5a 505 and eGFP-Rab5a Q79L, Prof Yoke Fun Chan from University o f Malaya for providing IEQ and 506 IEE infectious clone plasmids. We would also like to acknowledge Jessica Swanson, Dr 507 Natalie Kingston and Prof Nicola Stonehouse for giving advice and guidance about virus 508 purification. Electron microscopy data was collected at the Interdisciplinary Centre for 509 Electron Microscopy (CIME) at EPFL with assistance from Davide Demurtas, PhD. Electron 510 microscopy data was processed using the computational infrastructure provided by the IT 511 department of the School of Life Sciences (SV -IT) at EPFL. The authors express sincere 512 gratitude to the CIME and SV-IT personnel for their contribution. 513 514 Conflict of interest 515 The authors declare that they have no conflict of interest. 516 517 Figure legends 518 Fig 1. Lysosomotropic drugs inhibit infection by MP4 but not by MP4-97R/167G. (A) 519 Schematic illustration of the virus inhibitory assay workflow. Cells were pre -treated with 520 lysosomotropic drugs and infected in presence of the drug. After inoculum removal, 521 infected cells were cultured in drug -free media and infected cells were stained by 522 immunofluorescence (IF) with anti -VP2 Ab. ( B) Inhibition of endosomal acidification 523 confirmed with lysotracker staining (red). Lysosomes are in green (anti -LAMP1 Ab) and 524 nuclei in blue (DAPI). Representative IF images ar e shown (scale bar, 10 µm). C) Dose 525 response assay in i nfected Vero cells. Results are shown as % of virus -positive cells relative 526 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted February 25, 2024. ; https://doi.org/10.1101/2024.02.23.581741doi: bioRxiv preprint to nontreated control. Statistical significance (one -way ANOVA) between treated and 527 untreated virus or between treated MP4 an d MP4-97R/167G was calculated based on the 528 AUC. (D) Representative IF staining of EV -A71 (anti -VP2 in green) 24 hpi of Vero cells in 529 presence of 25 µM HCQ or 250 nM BAF -A1 (scale bar, 300 µm). (E & F ) HCQ effect in Vero 530 cells pre-treated or not with heparinase III (hepIII) (E) or sodium chlorate (NaClO3) as in A (F). 531 Statistical significance (two -way ANOVA) was calculated for each virus between each 532 condition. In D to F, mean and S.E.M of biological triplicates are shown. *p < 0.05, **p < 533 0.01, ***p < 0.001, ****p < 0.0001. 534 Fig 2. HCQ targets viral entry. (A) Virus binding assay in Vero cells in presence of 25 µM HCQ. 535 (B) Single -cycle replication kinetic in nontreated and HCQ -treated Vero cells. At each 536 timepoint, cell lysates were collected, and viral RNA copy numbers were quantitated using 537 RT-qPCR (C) Time -of-addition assay in Vero cells treated with HCQ starting at different 538 timepoints. Infected cells were quantitated 24 hpi by IF . (D) Schematic diagram of Vero cells 539 pre-treated with HCQ and subsequently subjected to transfection of in vitro RNA transcript 540 or infection with EV-A71 nanoluciferase (Nluc) reporter viruses. At the indicated timepoints, 541 cell supernatants were collected, and luciferase activity was measured. (E & F ) Results are 542 expressed in % relative light unit (RLU) of treated versus nontreated virus at indicated 543 timepoints. The mean and S.E.M from biological triplicates are shown. Statistical significance 544 was calculated using two -way ANOVA, comparing treated and untreated control. *p <0.05, 545 **p < 0.01, ***p < 0.001, ****p < 0.0001. 546 Fig 3. HCQ delays the uncoating of MP4. (A) Schematic illustration of the neutral red assay 547 workflow. Vero cells were pre -treated with or without HCQ for 1hr. Neutral red -labelled 548 viruses were allowe d for cell infection at 37°C for 1hr. Upon infection, the inoculum was 549 removed and replaced with fresh media. Infected cells were exposed to light for 30 min at 550 different timepoints and further incubated up to 24hpi for IF staining (B) Effect of light 551 inactivation on replication of neutral red -labelled MP4 (left panel) or MP4 -97R/167G (right 552 panel). Results are plotted as % of virus -positive cells relative to non -treated dark control . 553 Mean and S.E.M of biological triplicates are shown. Statistical significa nces (two -way 554 ANOVA) were calculated between treated and nontreated conditions. (C) Schematic 555 illustration of virus uncoating monitored with the combinational use of RNA -FISH to detect 556 EV-A71 RNA (red) and IF with anti -VP2 Ab to detect the viral capsid (green). Co -staining 557 highlights intact viruses in yellow while empty capsids and free RNA are in green and red, 558 respectively. Representative images (scale, 20 µm) of MP4 and MP4 -97R/167G binding after 559 1hr at 4°C (C, right panel) and of vRNA (red) and capsids (green) with and without HCQ 560 treatment at 4hpi ( D). Arrows: empty capsid . (E) Co-localization of capsid and vRNA in 561 individual cells at 2 hpi and 4 hpi analysed using Mander’s overlap coefficient (n = 32 562 individual cells from two independent experiments). Statistical comparison (unpaired t-test) 563 of untreated and treated groups. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. 564 Fig 4. MP4-97R/167G uncoats from early endosomes. (A) Nontreated and HCQ-treated 565 Vero cells were stained with anti-EEA-1 antibody (green) to label early endosomes, and DAPI 566 (blue) to label cell nuclei. (B) Schematic representation of endosomal route upon 567 overexpression of Rab5a WT or CA mutant. (C) Viral capsids (anti-VP2 Ab) localise in early 568 endosomes, 2 hpi of Vero cells transiently expressing Rab5a (in red) WT and CA. (D) % of 569 cells stained positive with the anti-dsRNA J2 Ab in FACS-sorted Rab5a-eGFP-expressing cells 570 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted February 25, 2024. ; https://doi.org/10.1101/2024.02.23.581741doi: bioRxiv preprint 7 hpi. Results and statistical significance (two-way ANOVA) are expressed relative to Rab5a 571 WT-expressing cells. Mean and S.E.M from triplicates are shown. ***p < 0.001. In B and D, 572 white boxes are enlarged in the right panel. Scale bar: 20 µm. 573 Fig 5. MP4 displayed stronger capsid stability and reduced sensitivity to acidification and 574 high temperatures. (A) nsEM analysis of MP4 and MP4 -97R/167G incubated at pH7 and 575 pH5. Representative raw micrographs are shown in each case. ( B) Representative 2D class 576 averages generated from datasets shown in panel A (box size = 54nm; left) and the overlay 577 of the corresponding 3D maps (right). Grey and orange shade indicates virus particle 578 reconstructions at pH7 and pH5, respectively. (C) Temperature sensitivity assay. Infected 579 Vero cells were quantitated by i mmunostaining with an anti -VP2 Ab at 24 hpi after 1 hr 580 incubation at increasing temperatures. Results are shown as % of virus -positive cells relative 581 to 4°C treated control. Error bars indicate mean and S.E.M from biological triplicates. (D) For 582 sSCARB2 in hibition assay, viruses were incubated 1h at 37°C with 1 µg of soluble SCARB2 583 (sSCARB2) before infection of Vero cells. I nfected Vero cells were quantitated by 584 immunostaining with an anti-VP2 Ab at 24 hpi. Results are shown as % of virus -positive cells 585 relative to nontreated controls. Statistically significance was calculated with two -way 586 ANOVA. ***p < 0.001, ****p < 0.0001. 587 Fig 6. Heparan-sulfate-binding VP1-145Q variant exhibits resistance to HCQ and higher 588 sensitivity to sSCARB2 inhibition and thermal stress. (A) Virus inhibitory assay with VP1-145 589 variants were performed with 25 µg HCQ on Vero cells. (B) For temperature sensitivity 590 assays, VP1 -145 variants were incubated at increasing temperature for 1hr before 591 inoculated onto Vero cells. (C) For sSCARB2 inhibition assay, VP1 -145 variants were 592 incubated 1 h at 37°C with 1 µg of soluble SCARB2 (sSCARB2) before infection of Vero cells. 593 Infected cells were quantitated by immunostaining with anti -VP2 Ab at 24 hpi. Results are 594 shown as % of virus -positive cells relative to nontreated control (A & C) or 4°C treated 595 control (B). Mean and S.E.M of biological triplicates are shown. Statistically significant 596 differences (two-way ANOVA) are shown. **p < 0.01, ***p < 0.001, ****p < 0.0001. 597 Fig 7. Seesaw model depicting the interplay between capsid mutations, heparan sulfate-598 binding, capsid stability as well as the resulting fitness changes in both in vitro and in vivo 599 settings. Viruses undergo continuous mutations to optimize fitness across diverse 600 environments. In cell culture, they adapt to attain an ' in vitro advantage' by decreasing 601 capsid stability while acquiring HS-binding capacity, consequently enhancing their infectivity. 602 Conversely, during human infection, viruses adapt to secure an ' in vivo advantage' by 603 bolstering capsid stability, relinquishing HS -binding capacity, and thereby evading viral 604 trapping and resisting environmental stresses. 605 Fig S1. Lysosomotropic drugs nontoxic dose-range and differential inhibition of HS-606 dependent and independent variants. (A) Cytotoxicity effect of lysosomotropic drugs 607 evaluated with LDH and MTT assays. RD and Vero cells were treated with a range of 608 different concentrations of HCQ or BAF -A1 for 2 hr. At 24 hours post -treatment, cell 609 supernatants and lysates were collected for LDH assay and MTT assay, respectively, to 610 determine cytotoxicity effect (n =2). (B) Dose response assay with HCQ and BAF -A1 on RD 611 cells were performed exactly like in Vero cells ( Fig.1). Infected cells (stained with anti -VP2 612 Ab) were quantitated at 24 hpi after treatment with increasing drug concentrations. Results 613 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted February 25, 2024. ; https://doi.org/10.1101/2024.02.23.581741doi: bioRxiv preprint are shown as % of virus-positive cells relative to nontreated control. AUC was calculated and 614 statistical significance (one -way ANOVA) between treated and untreated virus or between 615 treated MP4 and MP4 -97R/167G are shown. Mean and S.E.M of biological triplicates are 616 shown. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. 617 Fig S2. Both EV-A71 variants are strictly dependent on SCARB2 for infection. Virus infection 618 was performed on RD WT and RD ΔSCARB2 cells. Cells were lysed, and viral RNA copy 619 numbers were quantitated at 24 hpi using RT -qPCR. Results are expressed as % Virus RNA 620 copy number relative to RD WT cells (set to 100%). M ean and S.E.M of biological triplicates 621 are shown. ****p < 0.0001. 622 Fig S3. Viruses were detected in early endosomes at 30 mpi. Vero cells transiently 623 expressing Rab5a -eGFP were infected with MP4 and MP4 -97R/167G and fixed at 30 mpi. 624 Colocalization of viruses was imaged with Rab5a in green and virus capsid (VP2) in red. 625 Magnified area was highlighted in white box and displayed at left bottom of merged image. 626 Fig S4. Visual presentation of DynaMut prediction of virus mutations on capsid amino acid 627 interactions and protein stability. Prediction of changes in amino acid int eractions and 628 capsid stability induced by the VP1 -L97R and VP1 -E167G capsid mutations were performed 629 using crystal structure of full assembled capsid on DynaMut server. ( A) Interatomic 630 interactions displayed and compared between WT and mutant capsid struct ures. VP1 -97 631 and VP1-167 residues are labelled in light green and represented as sticks together with the 632 surrounding interaction residues. Changes in interactions are highlighted on both WT and 633 mutant structures with red asterisks (*) ( B) VP1 -L97R and VP1 -E167G mutations decrease 634 capsid stability. Computation of the vibrational entropy change (ΔΔS Vib) between WT and 635 mutants. Amino acids in red indicate an increase of molecule flexibility. 636 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted February 25, 2024. ; https://doi.org/10.1101/2024.02.23.581741doi: bioRxiv preprint Table S1: 637 638 Mutation pH Temperature (°C) Predicted free Gibbs energy change value (ΔΔG) VP1-L97R 7 25 -0.65 (Destabilizing) 5 25 -0.70 (Destabilizing) 7 55 -0.51 (Destabilizing) VP1-E167G 7 25 -1.15 (Destabilizing) 5 25 -1.10 (Destabilizing) 7 55 -0.74 (Destabilizing) VP1-E145Q 7 25 -0.80 (Destabilizing) 5 25 -0.89 (Destabilizing) 7 55 -0.71 (Destabilizing) 639 Predicted Gibbs free energy change value (ΔΔG) was computed using I-mutant 2 server with 640 calculation formula and indication of protein structure stabilization as shown below. 641 642 Predicted Gibbs free energy change value (ΔΔG): ΔG (new protein) - ΔG (WT) in kcal/mol. 643 ΔΔG 0: Stabilizing mutation 645 646 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted February 25, 2024. ; https://doi.org/10.1101/2024.02.23.581741doi: bioRxiv preprint

647 1. Cagno, V., Tseligka, E.D., Jones, S.T., and Tapparel, C. (2019). Heparan Sulfate 648 Proteoglycans and Viral Attachment: True Receptors or Adaptation Bias? Viruses 11. 649 10.3390/v11070596. 650 2. Tee, H.K., Zainol, M.I., Sam, I.C., and Chan, Y.F. (2021). Recent advances in the 651 understanding of enterovirus A71 infection: a focus on neuropathogenesis. Expert 652 Rev Anti Infect Ther 19, 733-747. 10.1080/14787210.2021.1851194. 653 3. Kobayashi, K., and Koike, S. (2020). Cellular receptors for enterovirus A71. J Biomed 654 Sci 27, 23. 10.1186/s12929-020-0615-9. 655 4. Yamayoshi, S., Ohka, S., Fujii, K., and Koike, S. (2013). Functional comparison of 656 SCARB2 and PSGL1 as receptors for enterovirus 71. J Virol 87, 3335-334 7. 657 10.1128/jvi.02070-12. 658 5. Kuronita, T., Eskelinen, E.L., Fujita, H., Saftig, P., Himeno, M., and Tanaka, Y. (2002). A 659 role for the lysosomal membrane protein LGP85 in the biogenesis and maintenance 660 of endosomal and lysosomal morphology. J Cell Sci 115, 4117-4131. 661 10.1242/jcs.00075. 662 6. Collins, K.P., Witta, S., Coy, J.W., Pang, Y., and Gustafson, D.L. (2021). Lysosomal 663 Biogenesis and Implications for Hydroxychloroquine Disposition. J Pharmacol Exp 664 Ther 376, 294-305. 10.1124/jpet.120.000309. 665 7. Dang, M., Wang, X., Wang, Q., Wang, Y., Lin, J., Sun, Y., Li, X., Zhang, L., Lou, Z., Wang, 666 J., and Rao, Z. (2014). Molecular mechanism of SCARB2-mediated attachment and 667 uncoating of EV71. Protein & Cell 5, 692-703. 10.1007/s13238-014-0087-3. 668 8. Chua, B.H., Phuektes, P., Sanders, S.A., Nicholls, P.K., and McMinn, P.C. (2008). The 669 molecular basis of mouse adaptation by human enterovirus 71. J Gen Virol 89, 1622-670 1632. 10.1099/vir.0.83676-0. 671 9. Fujii, K., Sudaka, Y., Takashino, A., Kobayashi, K., Kataoka, C., Suzuki, T., Iwata -672 Yoshikawa, N., Kotani, O., Ami, Y., Shimizu, H., et al. (2018). VP1 Amino Acid Residue 673 145 of Enterovirus 71 Is a Key Residue for Its Receptor Attachment and Resistance to 674 Neutralizing Antibody during Cynomolgus Monkey Infection. J Virol 92. 675 10.1128/jvi.00682-18. 676 10. Kobayashi, K., Sudaka, Y., Takashino, A., Imura, A., Fujii, K., and Koike, S. (2018). 677 Amino Acid Variation at VP1-145 of Enterovirus 71 Determines Attachment Receptor 678 Usage and Neurovirulence in Human Scavenger Receptor B2 Transgenic Mice. J Virol 679 92. 10.1128/jvi.00681-18. 680 11. Tee, H.K., Tan, C.W., Yogarajah, T., Lee, M.H.P., Chai, H.J., Hanapi, N.A., Yusof, S.R., 681 Ong, K.C., Lee, V.S., Sam, I.C., and Chan, Y.F. (2019). Electrostatic interactions at the 682 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted February 25, 2024. ; https://doi.org/10.1101/2024.02.23.581741doi: bioRxiv preprint five-fold axis alter heparin-binding phenotype and drive enterovirus A71 virulence in 683 mice. PLoS Pathog 15, e1007863. 10.1371/journal.ppat.1007863. 684 12. Lee, E., and Lobigs, M. (2002). Mechanism of virulence attenuation of 685 glycosaminoglycan-binding variants of Japanese encephalitis virus and Murray Valley 686 encephalitis virus. J Virol 76, 4901-4911. 10.1128/jvi.76.10.4901-4911.2002. 687 13. Byrnes, A.P., and Griffin, D.E. (2000). Large-plaque mutants of Sindbis virus show 688 reduced binding to heparan sulfate, heightened viremia, and slower clearance from 689 the circulation. J Virol 74, 644-651. 10.1128/jvi.74.2.644-651.2000. 690 14. Reddi, H.V., and Lipton, H.L. (2002). Heparan sulfate mediates infection of high-691 neurovirulence Theiler's viruses. J Virol 76, 8400-8407. 10.1128/jvi.76.16.8400-692 8407.2002. 693 15. Mandl, C.W., Kroschewski, H., Allison, S.L., Kofler, R., Holzmann, H., Meixner, T., and 694 Heinz, F.X. (2001). Adaptation of tick-borne encephalitis virus to BHK-21 cells results 695 in the formation of multiple heparan sulfate binding sites in the envelope protein 696 and attenuation in vivo. J Virol 75, 5627-5637. 10.1128/jvi.75.12.5627-5637.2001. 697 16. Lee, E., Hall, R.A., and Lobigs, M. (2004). Common E protein determinants for 698 attenuation of glycosaminoglycan-binding variants of Japanese encephalitis and 699 West Nile viruses. J Virol 78, 8271-8280. 10.1128/jvi.78.15.8271-8280.2004. 700 17. Chen, Y., Maguire, T., Hileman, R.E., Fromm, J.R., Esko, J.D., Linhardt, R.J., and Marks, 701 R.M. (1997). Dengue virus infectivity depends on envelope protein binding to target 702 cell heparan sulfate. Nat Med 3, 866-871. 10.1038/nm0897-866. 703 18. Cordey, S., Petty, T.J., Schibler, M., Martinez, Y., Gerlach, D., van Belle, S., Turin, L., 704 Zdobnov, E., Kaiser, L., and Tapparel, C. (2012). Identification of site-specific 705 adaptations conferring increased neural cell tropism during human enterovirus 71 706 infection. PLoS Pathog 8, e1002826. 10.1371/journal.ppat.1002826. 707 19. Weng, K.F., Tee, H.K., Tseligka, E.D., Cagno, V., Mathez, G., Rosset, S., Nagamine, 708 C.M., Sarnow, P., Kirkegaard, K., and Tapparel, C. (2023). Variant enterovirus A71 709 found in immune-suppressed patient binds to heparan sulfate and exhibits 710 neurotropism in B-cell-depleted mice. Cell Rep 42, 112389. 711 10.1016/j.celrep.2023.112389. 712 20. Tseligka, E.D., Sobo, K., Stoppini, L., Cagno, V., Abdul, F., Piuz, I., Meylan, P., Huang, S., 713 Constant, S., and Tapparel, C. (2018). A VP1 mutation acquired during an enterovirus 714 71 disseminated infection confers heparan sulfate binding ability and modulates ex 715 vivo tropism. PLoS Pathog 14, e1007190. 10.1371/journal.ppat.1007190. 716 21. Chen, P., Song, Z., Qi, Y., Feng, X., Xu, N., Sun, Y., Wu, X., Yao, X., Mao, Q., Li, X., et al. 717 (2012). Molecular determinants of enterovirus 71 viral entry: cleft around GLN -172 718 on VP1 protein interacts with variable region on scavenge receptor B 2. J Biol Chem 719 287, 6406-6420. 10.1074/jbc.M111.301622. 720 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted February 25, 2024. ; https://doi.org/10.1101/2024.02.23.581741doi: bioRxiv preprint 22. Brandenburg, B., Lee, L.Y., Lakadamyali, M., Rust, M.J., Zhuang, X., and Hogle, J.M. 721 (2007). Imaging Poliovirus Entry in Live Cells. PLOS Biology 5, e183. 722 10.1371/journal.pbio.0050183. 723 23. Baradaran Eftekhari, R., Maghsoudnia, N., and Dorkoosh, F.A. (2020). Chloroquine: a 724 brand-new scenario for an old drug. Expert Opin Drug Deliv 17, 275-277. 725 10.1080/17425247.2020.1716729. 726 24. Mauthe, M., Orhon, I., Rocchi, C., Zhou, X., Luhr, M., Hijlkema, K.J., Coppes, R.P., 727 Engedal, N., Mari, M., and Reggiori, F. (2018). Chloroquine inhibits autophagic flux by 728 decreasing autophagosome-lysosome fusion. Autophagy 14, 1435-1455. 729 10.1080/15548627.2018.1474314. 730 25. Rodrigues, C.H., Pires, D.E., and Ascher, D.B. (2018). DynaMut: predicting the impact 731 of mutations on protein conformation, flexibility and stability. Nucleic Acids Res 46, 732 W350-w355. 10.1093/nar/gky300. 733 26. Panja, A.S., Maiti, S., and Bandyopadhyay, B. (2020). Protein stability governed by its 734 structural plasticity is inferred by physicochemical factors and salt bridges. Scientific 735 Reports 10, 1822. 10.1038/s41598-020-58825-7. 736 27. Chand, G.B., Banerjee, A., and Azad, G.K. (2020). Identification of novel mutations in 737 RNA-dependent RNA polymerases of SARS-CoV-2 and their implications on its 738 protein structure. PeerJ 8, e9492. 10.7717/peerj.9492. 739 28. White, J., Kartenbeck, J., and Helenius, A. (1982). Membrane fusion activity of 740 influenza virus. Embo j 1, 217-222. 10.1002/j.1460-2075.1982.tb01150.x. 741 29. Yoshimura, A., Kuroda, K., Kawasaki, K., Yamashina, S., Maeda, T., and Ohnishi, S. 742 (1982). Infectious cell entry mechanism of influenza virus. J Virol 43, 284 -293. 743 10.1128/jvi.43.1.284-293.1982. 744 30. Greber, U.F., Willetts, M., Webster, P., and Helenius, A. (1993). Stepwise dismantling 745 of adenovirus 2 during entry into cells. Cell 75, 477 -486. 10.1016/0092-746 8674(93)90382-z. 747 31. Vázquez-Calvo, Á., Caridi, F., Rodriguez-Pulido, M., Borrego, B., Sáiz, M., Sobrino, F., 748 and Martín-Acebes, M.A. (2012). Modulation of foot-and-mouth disease virus pH 749 threshold for uncoating correlates with differential sensitivity to inhibition of cellular 750 Rab GTPases and decreases infectivity in vivo. J Gen Virol 93, 2382-2386. 751 10.1099/vir.0.045419-0. 752 32. White, J., Kartenbeck, J., and Helenius, A. (1980). Fusion of Semliki forest virus with 753 the plasma membrane can be induced by low pH. J Cell Biol 87, 264-272. 754 10.1083/jcb.87.1.264. 755 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted February 25, 2024. ; https://doi.org/10.1101/2024.02.23.581741doi: bioRxiv preprint 33. Ravindran, M.S., and Tsai, B. (2016). Viruses Utilize Cellular Cues in Distinct 756 Combination to Undergo Systematic Priming and Uncoating. PLOS Pathogens 12, 757 e1005467. 10.1371/journal.ppat.1005467. 758 34. Yoshimori, T., Yamamoto, A., Moriyama, Y., Futai, M., and Tashiro, Y. (1991). 759 Bafilomycin A1, a specific inhibitor of vacuolar-type H(+)-ATPase, inhibits 760 acidification and protein degradation in lysosomes of cultured cells. Journal of 761 Biological Chemistry 266, 17707-17712. https://doi.org/10.1016/S0021-762 9258(19)47429-2. 763 35. Redmann, M., Benavides, G.A., Berryhill, T.F., Wani, W.Y., Ouyang, X., Johnson, M.S., 764 Ravi, S., Barnes, S., Darley-Usmar, V.M., and Zhang, J. (2017). Inhibition of autophagy 765 with bafilomycin and chloroquine decreases mitochondrial quality and bioenergetic 766 function in primary neurons. Redox Biology 11, 73-81. 767 https://doi.org/10.1016/j.redox.2016.11.004. 768 36. Wang, R., Wang, J., Hassan, A., Lee, C.-H., Xie, X.-S., and Li, X. (2021). Molecular basis 769 of V-ATPase inhibition by bafilomycin A1. Nature Communications 12, 1782. 770 10.1038/s41467-021-22111-5. 771 37. Yamamoto, A., Tagawa, Y., Yoshimori, T., Moriyama, Y., Masaki, R., and Tashiro, Y. 772 (1998). Bafilomycin A1 prevents maturation of autophagic vacuoles by inhibiting 773 fusion between autophagosomes and lysosomes in rat hepatoma cell line, H-4- II-E 774 cells. Cell Struct Funct 23, 33-42. 10.1247/csf.23.33. 775 38. Mauvezin, C., and Neufeld, T.P. (2015). Bafilomycin A1 disrupts autophagic flux by 776 inhibiting both V-ATPase-dependent acidification and Ca-P60A/SERCA-dependent 777 autophagosome-lysosome fusion. Autophagy 11, 1437-1438. 778 10.1080/15548627.2015.1066957. 779 39. Pillat, M.M., Krüger, A., Guimarães, L.M.F., Lameu, C., de Souza, E.E., Wrenger, C., 780 and Ulrich, H. (2020). Insights in Chloroquine Action: Perspectives and Implications in 781 Malaria and COVID-19. Cytometry A 97, 872-881. 10.1002/cyto.a.24190. 782 40. Schrezenmeier, E., and Dörner, T. (2020). Mechanisms of action of 783 hydroxychloroquine and chloroquine: implications for rheumatology. Nature 784 Reviews Rheumatology 16, 155-166. 10.1038/s41584-020-0372-x. 785 41. Tian, A. -L., Wu, Q., Liu, P., Zhao, L., Martins, I., Kepp, O., Leduc, M., and Kroemer, G. 786 (2021). Lysosomotropic agents including azithromycin, chloroquine and 787 hydroxychloroquine activate the integrated stress response. Cell Death & Disease 12, 788 6. 10.1038/s41419-020-03324-w. 789 42. Xu, R., Ji, Z., Xu, C., and Zhu, J. (2018). The clinical value of using chloroquine or 790 hydroxychloroquine as autophagy inhibitors in the treatment of cancers: A 791 systematic review and meta-analysis. Medicine (Baltimore) 97, e12912. 792 10.1097/md.0000000000012912. 793 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted February 25, 2024. ; https://doi.org/10.1101/2024.02.23.581741doi: bioRxiv preprint 43. Huang, S.C., Chang, C.L., Wang, P.S., Tsai, Y., and Liu, H.S. (2009). Enterovirus 71-794 induced autophagy detected in vitro and in vivo promotes viral replication. J Med 795 Virol 81, 1241-1252. 10.1002/jmv.21502. 796 44. Bucci, C., Parton, R.G., Mather, I.H., Stunnenberg, H., Simons, K., Hoflack, B., and 797 Zerial, M. (1992). The small GTPase rab5 functions as a regulatory factor in the early 798 endocytic pathway. Cell 70, 715-728. 10.1016/0092-8674(92)90306-w. 799 45. Serio, G., Margaria, V., Jensen, S., Oldani, A., Bartek, J., Bussolino, F., and Lanzetti, L. 800 (2011). Small GTPase Rab5 participates in chromosome congression and regulates 801 localization of the centromere-associated protein CENP-F to kinetochores. 802 Proceedings of the National Academy of Sciences 108, 17337-17342. 803 doi:10.1073/pnas.1103516108. 804 46. Mercer, J., Schelhaas, M., and Helenius, A. (2010). Virus Entry by Endocytosis. Annual 805 Review of Biochemistry 79, 803-833. 10.1146/annurev-biochem-060208-104626. 806 47. Catching, A., Te Yeh, M., Bianco, S., Capponi, S., and Andino, R. (2023). A tradeoff 807 between enterovirus A71 particle stability and cell entry. Nat Commun 14, 7450. 808 10.1038/s41467-023-43029-0. 809 48. Murer, L., Petkidis, A., Vallet, T., Vignuzzi, M., and Greber, U.F. (2022). Chemical 810 Evolution of Rhinovirus Identifies Capsid-Destabilizing Mutations Driving Low- pH-811 Independent Genome Uncoating. J Virol 96, e0106021. 10.1128/jvi.01060-21. 812 49. Lanahan, M.R., Maples, R.W., and Pfeiffer, J.K. (2021). Tradeoffs for a viral mutant 813 with enhanced replication speed. Proceedings of the National Academy of Sciences 814 118, e2105288118. doi:10.1073/pnas.2105288118. 815 50. LaTourrette, K., and Garcia-Ruiz, H. (2022). Determinants of Virus Variation, 816 Evolution, and Host Adaptation. Pathogens 11. 10.3390/pathogens11091039. 817 51. Andino, R., and Domingo, E. (2015). Viral quasispecies. Virology 479-480, 46- 51. 818 https://doi.org/10.1016/j.virol.2015.03.022. 819 52. Chang, S.C., Li, W.C., Chen, G.W., Tsao, K.C., Huang, C.G., Huang, Y.C., Chiu, C.H., Kuo, 820 C.Y., Tsai, K.N., Shih, S.R., and Lin, T.Y. (2012). Genetic characterization of enterovirus 821 71 isolated from patients with severe disease by comparative analysis of complete 822 genomes. J Med Virol 84, 931-939. 10.1002/jmv.23287. 823 53. Liu, Y., Fu, C., Wu, S., Chen, X., Shi, Y., Zhou, B., Zhang, L., Zhang, F., Wang, Z., Zhang, 824 Y., et al. (2014). A novel finding for enterovirus virulence from the capsid protein VP1 825 of EV71 circulating in mainland China. Virus Genes 48, 260-272. 10.1007/s11262-826 014-1035-2. 827 54. Kobayashi, K., Mizuta, K., and Koike, S. (2020). Heparan sulfate attachment receptor 828 is a major selection factor for attenuated enterovirus 71 mutants during cell culture 829 adaptation. PLoS Pathog 16, e1008428. 10.1371/journal.ppat.1008428. 830 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted February 25, 2024. ; https://doi.org/10.1101/2024.02.23.581741doi: bioRxiv preprint 55. Tan, C.W., Tee, H.K., Lee, M.H., Sam, I.C., and Chan, Y.F. (2016). Enterovirus A71 831 DNA-Launched Infectious Clone as a Robust Reverse Genetic Tool. PLoS One 11, 832 e0162771. 10.1371/journal.pone.0162771. 833 56. Punjani, A., Rubinstein, J.L., Fleet, D.J., and Brubaker, M.A. (2017). cryoSPARC: 834 algorithms for rapid unsupervised cryo-EM structure determination. Nature Methods 835 14, 290-296. 10.1038/nmeth.4169. 836 57. Capriotti, E., Fariselli, P., and Casadio, R. (2005). I-Mutant2.0: predicting stability 837 changes upon mutation from the protein sequence or structure. Nucleic Acids Res 33, 838 W306-310. 10.1093/nar/gki375. 839 840 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted February 25, 2024. ; https://doi.org/10.1101/2024.02.23.581741doi: bioRxiv preprint Fig 1 A Pre-treatment with drug Infection with drug for 1 hr Immunofluorescence detection B Nontreated 25 µM HCQ 250 nM BAF-A1 E F D Nonntreated 25 µM HCQ MP4 MP4-97R/167G 250 nM BAF-A1 C .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted February 25, 2024. ; https://doi.org/10.1101/2024.02.23.581741doi: bioRxiv preprint Fig 2 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted February 25, 2024. ; https://doi.org/10.1101/2024.02.23.581741doi: bioRxiv preprint Fig 3 Infection with/ without HCQ Light inactivation Immunofluorescence detection A B C E 1 hpi at 4˚C MP4-97R/167G MP4 D Virus capsid vRNA Merge MP4 MP4/97R/167G HCQ Nontreated HCQ Nontreated Pretreatment with HCQ Virus infection at 37°C with HCQ Virus –positive cell detection -1 0 1 2 3 4 24 hpi Light inactivation Wash and replace with fresh media .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted February 25, 2024. ; https://doi.org/10.1101/2024.02.23.581741doi: bioRxiv preprint Fig 4 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted February 25, 2024. ; https://doi.org/10.1101/2024.02.23.581741doi: bioRxiv preprint Fig 5 A pH 7.5 pH 5 MP4-97R/167G Particle diameter range: 31-33nm Particle diameter range: 31-41nm pH 5 pH 7.5 MP4 Particle diameter range: 31-33nm Particle diameter range: 31-33nm MP4 MP4-97R/167G pH7.5 pH5 B C D .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted February 25, 2024. ; https://doi.org/10.1101/2024.02.23.581741doi: bioRxiv preprint Fig 6 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted February 25, 2024. ; https://doi.org/10.1101/2024.02.23.581741doi: bioRxiv preprint Capsid stability Fig 7 HS binding affinity In vivo advantage In vitro advantage Resist to acidic pH, temperature... pH-independent entry Facilitate uncoating upon SCARB2 binding Avoid trapping by heparan sulfate → Improve dissemination and transmission → Improve multiplication Binding to heparan sulfate .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted February 25, 2024. ; https://doi.org/10.1101/2024.02.23.581741doi: bioRxiv preprint Fig S1 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted February 25, 2024. ; https://doi.org/10.1101/2024.02.23.581741doi: bioRxiv preprint Fig S2 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted February 25, 2024. ; https://doi.org/10.1101/2024.02.23.581741doi: bioRxiv preprint Fig S3 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted February 25, 2024. ; https://doi.org/10.1101/2024.02.23.581741doi: bioRxiv preprint Fig S4 A Hydrogen bonds Water-mediated hydrogen bonds Weak hydrogen bonds Water-mediated weak hydrogen bonds Halogen bonds Ionic interactions Metal complex interactions Aromatic contacts Hydrophobic contacts Carbonyl contacts B VP1-L97R VP1-E167G ΔΔSVib ENCoM: 0.055 kcal.mol-1.K-1 (Increase of molecule flexibility) ΔΔSVib ENCoM: 0.278 kcal.mol-1.K-1 (Increase of molecule flexibility) GLU167 SER165 LEU169 ARG85 TRP171 TRP171 ARG85 GLY167 GLN172 SER165 LEU169 ARG97 PRO246 TYR245 LYS244 GLY99 LEU95 LEU95 LEU97 PRO246 TYR245 LYS244 GLY99 SER243 WT L97R/167G VP1-167 VP1-97 * * * * * * * * .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted February 25, 2024. ; https://doi.org/10.1101/2024.02.23.581741doi: bioRxiv preprint Table S1 Mutation pH Temperature (°C) Predicted free Gibbs energy change value (ΔΔG) VP1-L97R 7 25 -0.65 (Destabilizing) 5 25 -0.70 (Destabilizing) 7 55 -0.51 (Destabilizing) VP1-E167G 7 25 -1.15 (Destabilizing) 5 25 -1.10 (Destabilizing) 7 55 -0.74 (Destabilizing) 7 25 -0.80 (Destabilizing) VP1-E145Q 5 25 -0.89 (Destabilizing) 7 55 -0.71 (Destabilizing) Predicted Gibbs free energy change value (ΔΔG) was computed using I-mutant 2 server with calculation formula and indication of protein structure stabilization as shown below. Predicted Gibbs free energy change value (ΔΔG): ΔG (new protein) - ΔG (WT) in kcal/mol. ΔΔG < 0: destabilizing mutation .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted February 25, 2024. ; https://doi.org/10.1101/2024.02.23.581741doi: bioRxiv preprint