key: cord-0933125-c381k2rj
authors: Bidon, Miya K.; Khelashvili, George; Straus, Marco R.; Tang, Tiffany; Carten, Juliana Debrito; Weinstein, Harel; Whittaker, Gary R.; Daniel, Susan
title: The interaction of calcium ions with specific residues in the SARS-CoV fusion peptide and the regulation of viral infectivity
date: 2022-03-04
journal: bioRxiv
DOI: 10.1101/2022.03.03.482731
sha: 03e58851ec70312c7ebf989db83d77fd40e56f3b
doc_id: 933125
cord_uid: c381k2rj

Viral envelope fusion with the host cell membrane is dependent on a specific viral fusion peptide (FP) or loop, which becomes exposed during virus entry to drive the process of membrane fusion. In coronaviruses, the FP is a highly conserved domain that sits in the center of spike protein and in SARS-CoV, is adjacent to the S2’ proteolytic cleavage site. This peptide contains a hydrophobic LLF motif, as well as several conserved negatively charged amino acids that interact with Ca2+ ions to promote membrane fusion. In this work we perform a systematic mutagenesis study of the negatively charged amino acids within the SARS-CoV fusion peptide (FP1/FP2) and combine this with molecular dynamics simulations to define the membrane interactions that regulate virus infectivity. We show that the E801/D802 amino acid pair in the SARS-CoV FP is predicted to bind to one Ca2+ ion to promote FP-membrane interaction, with a second Ca2+ ion likely pairing residue D812 with either E821 or D825. The D812/D821 residue pair promotes membrane interaction, whereas the D821/D825 is inhibitory to membrane insertion. Taken together, our results demonstrate the dynamic nature of the coronavirus FP region that likely facilitates its interactions with and insertion into the host cell membrane. Author Summary Coronaviruses have reemerged as a highly pathogenic virus family through the rise of SARS-CoV, MERS-CoV, and more recently, SARS-CoV-2. As more transmissible variants of SARS-CoV-2 arise, it is imperative that we understand the mechanisms of CoV viral entry to enable the development of effective therapeutics. Recent reviews have suggested the repurposing of FDA-approved calcium channel blockers to treat infection by coronaviruses; however, calcium’s method of action on viral-host cell fusion events is unknown. We have found that increased calcium availability leads to increased viral infection across the CoV family, suggesting that calcium is involved in mediating the interaction between the viral fusion peptide and the host cell membrane. As such, we hypothesize that the highly conserved fusion peptide interacts directly with calcium and this interaction is required for viral entry and infection. Through mutagenesis studies of specific negatively charged residues in the fusion peptide, we have identified residues that impact viral infectivity. We have also compared the infectivity of wild-type and mutant CoV pseudoparticles in calcium-rich or -depleted environments using chelating drugs. Our data mirrors the residue coordination observed SARS-CoV-2, as both between SARS-CoV and SARS-CoV-2 FPs bind to two calcium ions. These results demonstrate the importance of Ca2+ for CoV FP function during viral entry and opens the possibility of utilizing FDA-approved calcium-blocking drugs as a treatment for COVID-19.

contains a hydrophobic LLF motif, as well as several conserved negatively charged amino acids 23 that interact with Ca2+ ions to promote membrane fusion. In this work we perform a systematic 24 mutagenesis study of the negatively charged amino acids within the SARS-CoV fusion peptide 25 (FP1/FP2) and combine this with molecular dynamics simulations to define the membrane 26 interactions that regulate virus infectivity. We show that the E801/D802 amino acid pair in the 27 SARS-CoV FP is predicted to bind to one Ca 2+ ion to promote FP-membrane interaction, with a 28 second Ca 2+ ion likely pairing residue D812 with either E821 or D825. The D812/D821 residue pair 29 promotes membrane interaction, whereas the D821/D825 is inhibitory to membrane insertion. coronaviruses as one of the top five emerging pathogens likely to cause major epidemics due to 59 the minimal countermeasures that exist [7] [8] [9] . While vaccines are one strategy that has proven 60 successful during this SARS-CoV-2 pandemic, there is an outstanding need for antiviral drugs as 61 (FP) segment that is liberated upon cleavage [18] . S2' cleavage allows the S2 domain to undergo 84 a large conformational change that positions the FP for insertion into the host cell membrane 85 [10] . FP membrane insertion is a key step in commencing the process of merging the host cell 86 membrane with the viral envelope, resulting in the transfer of the viral genome into the host cell. 87

Given the importance of the FP in initiating membrane fusion, it is not surprising to find 88 significant conservation of the amino acid sequence in this region of the S protein across many 89

CoVs [19] . Due to its high degree of conservation, the FP stands out as a potential antiviral target. 90

Thus, work carried out by our team and others has focused on understanding the structure and 91 function of the FP. This work has identified conserved residues that interact with calcium ions 92 (Ca 2+ ) and mediate the interactions between the host cell membrane and the FP that lead to viral 93 infection [19] [20] [21] [22] [23] . 94

In our previous studies, we showed that for both SARS-CoV and MERS-CoV, calcium depletion 95 in cell culture leads to a significant drop in viral infection [19, 20] . We connected this decrease in 96 infectivity to a defect in membrane fusion, observing that syncytia formation during cell-to-cell 97 fusion between S-and ACE2 receptor-expressing VeroE6 cells also drops in calcium-depleted 98 media. We then sought to connect this observed impact on membrane fusion with the molecular 99 scale features of S that may interact directly with calcium ions. Within the S2 domain of S, the 100 FP sequence contains highly conserved charged amino acids that flank the hydrophobic residues 101 immunoblots displayed full length, uncleaved (S0), and cleaved S2 subunit (S2) SARS-CoV S species 150 migrating at 180kDa and 80kDa, respectively (Fig 2A) . SARS-CoV S mutants containing the single 151 mutations D802A, D812A, E821A, D825A, and D830A have comparable steady-state levels of full-152 length protein in comparison to the wild-type protein, indicating that these mutations in the FP 153 did not impair the synthesis or trafficking of S (Fig 2A) . Additionally, proteolytic cleavage of the 154 D802A, D812A, E821A, D825A, and D830A mutants occurred following treatment with TPCK-155 trypsin, indicating that these mutants were able to be primed for downstream fusion evens. We 156 observed a higher molecular weight species running above the full-length S protein (Fig 2, (Higher 157 order S)). We have determined that heating samples at 95°C with 5mM DTT for 10 minutes 158 resolves this higher molecular weight band (data not shown), indicating that this species likely 159 results from spike-based protein-protein interactions. 160

We were unable to detect steady-state levels of the E801A mutant on the spike 161 immunoblots. To further probe the nature of this residue's importance, we substituted in the 162 larger, nonpolar methionine (E801M), the polar and uncharged glutamine (E801Q), positively 163 charged lysine (E801K) or negatively charged aspartic acid (E801D). None of these mutations, 164

including the charge mimetic E801D, restored the steady-state levels of the S protein to that of 165 wild-type, indicating the specific requirement of glutamic acid in this region of the FP (Fig 2B) . 166 167

To functionally test the ability of S protein mutants to induce cell-cell fusion, we 169 performed a syncytia assay. To do this we transiently expressed the WT and mutated S proteins 170 in VeroE6 cells, which are kidney epithelial cells that express the ACE2 receptor [26] . 24 hours 171 following transfection, we induced cell-cell fusion by treating the cells with trypsin to cleave the 172 FP at the S2' site (R797). Syncytia formation was visualized by immunofluorescence using the 173 fluorescently labeled SARS-CoV S antibody and DAPI stain to identify multinucleated S-expressing 174 cells that had fused. Syncytia were quantified by counting every group of fused cells that had a 175 minimum of 4 nuclei. As expected, VeroE6 cells expressing the WT S protein readily formed 176 syncytia (Fig 3A and B) . Conversely, cells expressing the D802A, D812A, E821A, D825A, and 177 D830A mutants all exhibited very few syncytia, indicating a defect in cell-cell fusion (Fig 3A and  178 B; Fig S1) . 179 180 (Fig 4A) . 195

As expected, we did not detect the E801A mutant in the SARS-CoV pseudoparticles, likely due to 196 its low cellular expression or synthesis. 197

After confirming the incorporation of the wild-type and mutant spike proteins into the 198 SARS-CoV pseudoparticles, we tested the infectivity of the pseudoparticles. As we have 199 demonstrated previously, exposure of VeroE6 cells to WT SARS CoV pseudoparticles results in a 200 robust luminescent signal 72 hours post infection, indicating successful viral entry (Fig 4B) . 201 E821A-and D825A-containing pseudoparticles did not exhibit a significant change in infectivity 202 compared to wild-type particles, suggesting these individual residues do not have a major effect 203 on FP function when calcium is present (Fig 4B) . In contrast, the D802A, D812A, and D830A 204 mutant pseudoparticles all showed a pronounced decrease in infectivity, with the D812A and 205 D830A mutations resulting in essentially noninfectious particles. These results indicate that 206 mutations in these three highly conserved negatively charged residues in the S protein's FP 207 significantly decrease SARS-CoV pseudoparticle infectivity in calcium-containing conditions. 208

We next tested the effect of depleting extracellular calcium on the pseudoparticles' ability to 209 infect VeroE6 cells using the chelator ethylene glycol-bis (ß-aminoethyl ether)-N,N,N'N'-210 tetraacetic acid (EGTA). We have previously studied the effects of calcium on viral infectivity and 211 adopted similar calcium depletion methods in this study [19, 20] . Briefly, we treated VeroE6 cells 212 with 50 µM EGTA prior to and during infection. Following incubation with the chelating agent, 213

VeroE6 cells were inoculated with WT or mutant S protein-containing pseudoparticles and 214 incubated for 72 hours. Cells were then lysed and luminescence was quantified as a measure of 215 pseudoparticle infectivity (Fig 4B) . In support of our previous work [19] , depletion of extracellular 216 calcium had a marked effect on WT pseudoparticle infectivity, decreasing it to approximately half 217 that of cells infected with WT pseudoparticles in the absence of EGTA (Fig 4B) . Cells exposed to 218 pseudoparticles containing either D802A or the D825A mutations also exhibited a further drop 219 in infectivity, with the D825A mutation having the greatest additional reduction in infectivity (Fig  220   4B ). The depletion of extracellular calcium had an additive effect on the decrease in infectivity in 221 the context of specific FP mutations (D802A, D825A). These results suggest that these negatively 222 charged residues in the FP are affected either directly or indirectly by loss of extracellular calcium. 223

We then examined the effect of intracellular calcium depletion on pseudoparticle infectivity 224 by using the cell-permeable calcium chelator 1,2-bis(2-aminophenoxy)ethane-N,N,N'N'-225 tetraacetic acid tetrakis (BAPTA-AM). We have previously optimized the concentrations of 226 BAPTA-AM so as not to significantly impact cell viability [19, 20] . VeroE6 cells were treated with 227 50 µM BAPTA-AM briefly prior to and during pseudoparticle infection, then harvested and 228 assayed for luminescence (Fig 4C) . Interestingly, depletion of intracellular calcium rendered the 229 WT pseudoparticles noninfectious; similar results were observed across the WT and all mutant 230 pseudoparticles tested; none of the pseudoparticles were infectious. In general, these data 231 suggest that depletion of intracellular calcium exerts a more severe effect on pseudoparticle 232 infectivity and point to the endosomal pathway as the primary route of pseudoparticle entry. 233

From our infectivity data we deduced that the FP likely uses multiple negatively charged 234 residues to bind to multiple calcium ions. This is supported by the further decrease in infectivity 235 of Spike E802A-and D825A-containing pseudoparticles when extracellular calcium is depleted. 236

We hypothesize that mutating those residues individually does not lead to a complete loss in 237 pseudoparticle infectivity, since other charged residues in the FP can compensate by binding 238 calcium; however, removal of calcium from the medium mimics the loss of those additional 239 residues, leading to the FP's inability to bind calcium and a loss in infectivity. 240 241

To test the hypothesis that the FP utilizes alternative residues to mediate calcium binding, (Fig 5A) . 254 We first confirmed that these mutants were synthesized, trafficked to the plasma 255 membrane, and were able to be cleaved by trypsin, as we had done with the single mutants (Fig  256   5B ). We also tested the functionality of these S mutants using the previously described syncytia 257 assay. VeroE6 cells transiently expressing either a double mutant (E821A/D825A, E821A/D830A, 258 D825A/D830A) or the triple mutant (E821A/D825A/D830A) all displayed fewer syncytia 259 compared to WT S-expressing cells (Fig 5C and D) . 260 We then generated pseudoparticles containing the double and triple mutants and 261 assayed the incorporation of the various mutants into the particles by immunoblotting for the 262 spike protein. The double mutant E821/D830A-and D825A/D830A-containing pseudoparticles 263 incorporated roughly equal amounts the S protein, while the triple mutant pseudoparticles 264 (E821A/D825A/D830A) showed slightly reduced protein levels (Fig 5E) . The E821A/D825A mutant 265 particles exhibited significantly decreased protein levels indicating that although this mutant 266 could be expressed and trafficked to the plasma membrane, its incorporation into 267 pseudoparticles was diminished. We acknowledge the general low protein levels of the double 268 and triple S mutants in our pseudoparticles as being a potential confounding variable. 269

Nevertheless, we infected VeroE6 cells with the WT and mutant S protein pseudoparticles to 270 investigate if multiple mutations in negatively charged residues of the FP caused a more 271 pronounced decrease in infectivity. As was previously demonstrated, WT pseudoparticles are 272 able to infect VeroE6 cells, with subsequent depletion of extra-or intra-cellular calcium causing 273 a partial and then complete loss of infectivity, respectively (Fig 5F and G) . Infectivity was 274 dramatically reduced for all mutants in comparison to WT pseudoparticles, irrespective of extra-275 and intracellular calcium depletion. (Fig 5F and G) . These data support the hypothesis that the 276 SARS-CoV FP contains multiple sites of calcium binding, which when mutated result in a 277 nonfunctional spike protein irrespective of calcium levels. Moreover, simultaneous binding of Ca 2+ ions to the SARS-CoV-2 FP residue pairs equivalent to 300 E801/D802 and D812/E821 produced the peptide conformations prone to membrane 301 penetration. In contrast, conformations that stabilized Ca 2+ binding to residues equivalent to the 302 E821/D825 pair did not enable sustained bilayer insertion. 303

From these MD simulations of the SARS-CoV FP constructs, we were able to predict the 304 phenotypes for each construct (active or inactive). The mode of Ca 2+ ion association predicted to 305 be inhibitory for membrane insertion (E821-D825) is shown in red, and the modes of Ca 2+ 306 association predicted to facilitate membrane insertion are depicted in green boxes (Fig 6) . In 307 those constructs with single mutations that were predicted to not have fusion activity (i.e., 308 E801A, D802A, D812A, and D830A) we showed that only the Ca 2+ -coordination mode involving 309 the E821/D825 pair persists, but not the D812/E821 pair of residues (Fig 6) . Conversely, in the 310 SARS-CoV FP single mutants that maintained WT-like fusion activity (E821A and D825A), we did 311 not observe the E821/D825 residues predicted to participate in calcium binding. In these 312 function-preserving mutants we identified additional modes of Ca 2+ binding that are enhanced in 313 comparison to the WT system: E801/D802 in mutant E821A, or E801/D830 in mutant D825A. For 314 the SARS-CoV FP constructs with multiple mutations, our MD trajectories revealed an overall 315 reduced Ca 2+ binding ability (Fig 6 and Fig S3) , consistent with our experimental findings that (i) 316 these constructs are severely defective in the cell-cell fusion (Fig 5C and D) and infectivity assays, 317

and (ii) their function does not depend on the levels of either extracellular or intracellular Ca 2+ 318 (Fig 5F and G) . 319 320

Overall, the above computational results reveal that the Ca 2+ binding patterns of SARS-322

CoV FP are very similar to those of SARS-CoV-2 FP. On this basis, membrane insertion of the SARS-323

CoV FP could be expected to be enhanced by the modes of Ca 2+ binding involving the E801/D802 324 and D812/E821 pairs of residues, and to be reduced by the ones involving the E821/D825 pair. 325

To test this premise, we carried out MD simulations of two models of the WT SARS-CoV FP 326 spontaneously associating with the lipid membrane (see Methods). In one, the peptide was 327 interacting with 2 Ca 2+ ions at the E801/D802 and D812/E821 sites (Model 1), and in the second, 328 2 Ca 2+ ions were bound to the E801/D802 and E821/D825 pairs (Model 2). Each structure was 329 simulated in 36 independent replicates, each run for ~0.9-1.0µs (see Methods). was inserted into the membrane if the z-distance between its Cα atom and the second carbon 336 atom in the tail of a POPC lipid (atom C22 in CHARMM36 notation) was <5Å. As shown in Fig 7A,  337 in Model 1, the N-terminal FP1 segment of the fusion peptide shows strong propensity for bilayer 338 insertion, while in Model 2 the insertion is minimal (Fig 7B) . 339

The detailed analysis of the individual trajectories in the Model 1 set revealed two distinct 340 modes of bilayer penetration for SARS-CoV FP, similar to our findings for the SARS-CoV-2 FP. Thus, 341 the Model 1 construct penetrates the bilayer either with its N-terminal LLF motif (Fig 7C) or with 342 the more centrally located hydrophobic F815-M816 segment (Fig 7D) . Interestingly, the two 343 insertion modes appear to alternate which Ca 2+ ion is neighboring the inserted portion. Thus, 344 when the LLF is inserted, the Ca 2+ ion associated with the neighboring E801/D802 residues is 345 bound to the membrane while the other Ca 2+ binding site (D812/E821) is situated away from the 346 membrane surface (see snapshot in Fig 7C) . In case of the F815-M816 insertion, the position of 347 the Ca 2+ binding loci with respect to the membrane is reversed -the one associated with the 348 D812/E821 pair is membrane-bound, while the E801/D802 pair is located farther from the bilayer 349 (snapshot in Fig 7D) . We also note that in both cases, the remaining anionic residues in the 350 peptide (i.e. the ones not engaged with the Ca 2+ ions) are either solvent exposed (D830) or 351 engaged with electro-neutralizing interactions with neighboring basic residues (D825/R829, Fig  352   7C To interrogate the requirement of the negatively charged residues in the SARS CoV fusion 372 peptide for its function, we first made single charge-to alanine substitutions in those residues 373 (Fig 1) . Following transient expression of these mutants in Vero6 cells, we confirmed that 374 mutants D802A, D812A, E821A, D825A, and D830A were synthesized, accumulated to levels 375 comparable to the WT protein, and were trafficked to the plasma membrane (Fig 2) . We further 376 confirmed the cell surface localization of these same FP mutants by treating VeroE6 cells 377 expressing these mutants with trypsin to assess their cleavage, after which we retrieved the FPs 378 using a cell surface biotinylation assay (Fig 2) . The majority of the FP mutants assayed exhibited 379 cleavage following treatment with trypsin, indicating the accessibility of the S1/S2 cleavage site. 380

The E801A mutant was not detected on our spike immunoblots and further attempts to 381 understand this occurrence through additional substitutions did not result in the detection of this 382 mutant. We hypothesize that the absence of glutamate at position 801 in the fusion peptide 383 causes a loss in protein stability, though further work is needed to determine the significance of 384 this residue in the fusion peptide. 385

We next tested the fusion competency of the single FP mutants we had created using a 386 syncytia assay, which utilizes fusion events in cells transiently expressing the FP as a readout of 387 FP fusion activity. All of the single FP charged-to-alanine mutants the were detected on the spike 388 immunoblot (E801A, D802A, D812A, E821A, D825A, and D830A) exhibited a pronounced fusion 389 defect, as evidenced by the low number of fused VeroE6 cells, or syncytia, that were observed 390 (Fig 3) . Taken together, these data suggest that the highly conserved, negatively-charged 391 residues within the FP individually contribute in a non-redundant manner to the function of the 392 fusion peptide. However, due to the limitations and variability of the syncytia assay, we chose to 393 use SARS-CoV2 pseudoparticles to mimic a more in vivo-like system to examine the functionality 394 of the various FP mutants. Successful pseudoparticle entry into host cells results in the 395 integration of the luciferase reporter gene into the cellular genome. Luminescence can therefore 396 be used as a readout of pseudoparticle infectivity. We first confirmed the incorporation of the 397 WT and single-charged-to-alanine mutant FPs into the pseudoparticles (Fig 4A) . Nearly all FP 398 mutants generated (E801A, D802A, D812A, E821A, D825A, and D830A) were incorporated in the 399 SARS-CoV2 pseudoparticles; the E801A mutant was not detected. 400

We then infected VeroE6 cells using the SARS-CoV2 pseudoparticles we had generated 401 and measured luminescence as a readout of infectivity and a proxy for viral entry. Introduction 402 of WT pseudoparticles into VeroE6 cells results in a robust luminescence signal, indicating 403 successful viral entry and fusion competency of the viral particles when calcium levels are 404 unperturbed (Fig 4B and C) . Infections with E821A and D825A-containing pseudoparticles at 405 physiological levels of calcium also resulted in luminescence levels comparable to WT-containing 406 pseudoparticles, suggesting that these residues are not required for FP function. To the contrary, 407 pseudoparticles containing the D802A, D812A, or D830A mutations were unable to infect VeroE6 408 cells, resulting in aa significant drop in luminescence. These data indicate that when intra-and 409 extra-cellular calcium levels remain unchanged, loss of an individual negative charges at positions 410 D802, D812, or D830 is sufficient to cause a substantial decrease in infectivity, likely due to a 411 defect in FP-mediated viral fusion and subsequent viral entry. 412

We then proceeded to test the requirement of extracellular calcium on the infectivity of 413 our FP-containing pseudoparticles by treating the cells with EGTA, a calcium-preferring chelator, 414 prior to infection. Removal of extracellular calcium resulted in a significant drop in the infectivity 415 of WT FP-containing particles, which is consistent with the known requirement of calcium for 416 SARS CoV2 viral entry (Fig 4B) . Interestingly, pseudoparticles containing either the D802A or 417 D825A mutant showed a further reduction in infectivity when the extracellular calcium was 418 depleted. This suggests that multiple negatively charged residues in the FP are involved in calcium 419 binding and while loss of a single negative charge may not be sufficient to completely disrupt 420 infectivity, multiple "hits" are. Thus, removal of extracellular calcium mimics the loss of additional 421 electrostatic interactions needed for FP function, resulting in a further decrease in infectivity. 422

Pseudoparticles containing either the D812A or D830A FP mutant were essentially non-infectious 423 in the presence or absence of extracellular calcium. These results implicate residues D812A and 424 D830A in FP function; however, their specific roles as they relate to calcium cannot be teased 425 apart in this system. 426

We also depleted intracellular calcium levels using the cell permeable calcium chelator 427 BAPTA-AM, in order to test the contribution of intracellular (endosomal) during pseudoparticle 428 infectivity. 429

Following treatment with BAPTA-AM, WT FP-containing pseudoparticles are no longer 431 infectious (Fig 4C) , indicating that intracellular calcium, specifically in the endosome, contributes 432 to pseudoparticle entry (Fig 4B and C) . We propose that chelation with EGTA likely removed 433 the majority of extracellular calcium causing a 50% reduction in infectivity of WT FP-containing 434 pseudoparticles; however, it is known that SARS-CoV enters cells via two pathways: the plasma 435 membrane and endosomal pathways. Hence, the partial reduction in viral infectivity in the 436 presence of EGTA and then the complete loss of infectivity in the presence of BAPTA-AM may 437 reflect the SARS-CoV pseudoparticles' usage of these two pathways as well. It is important for 438 us to acknowledge that we cannot rule out the potential impact that depletion of intracellular 439 calcium may have on the integration, expression, or synthesis of the reporter transgene in the 440 pseudoparticle. However, we performed cell viability assays to optimize the concentrations of 441 both chelators used in this study and are confident that treatments with EGTA and BAPTA-AM 442 did not induce toxicity in the VeroE6 cells. 443

Given that the results from our infectivity assay suggested the involvement of multiple 444 calcium-binding residues in the fusion peptide, we then generated double mutants (E821A, 445 E825A; E821A, D830A; E825A, D830A) and a triple mutant (E821A:E825A;D830A) of the FP (Fig  446   6A) . We validated the expression, synthesis, and cell surface localization of these mutants, as 447 well as their ability to be cleaved by trypsin (Fig 6B) . As with the single mutants, we observed a 448 fusion defect in VeroE6 cells transiently expressing these mutant constructs, evidenced by the 449 few number of syncytia formed in comparison to cells expressing the WT FP (Fig 6D) . When then 450 assayed the fusion activity of these mutants using the previously described pseudoparticle 451 infectivity assay. The double mutants E821A; D830A and E825A; D830A and a triple mutant 452 E821A; E825A; D830A were all incorporated into the pseudoparticles; the E821A; E825A mutant 453 was not (Fig 6C) . If the FP binds calcium using more than one negatively charged residue, then 454 pseudoparticles containing the double or triple charge-to alanine mutants should exhibit a 455 further decrease in infectivity. In comparison to WT FP-containing pseudoparticles, the FP double 456 mutants E821A; D830A and E825A; D830A and the triple mutant E821A;E825A;D830A we assayed 457 showed a complete loss in infectivity. Because the FP double mutant E821A, E825A was not 458 incorporated into the pseudoparticles, we could not fully assess the impact of these mutations 459 on FP function; however, the low number of syncytia observed in cells expressing this mutant 460 suggests these residues are important for FP function. In summary, these results support the 461 hypothesis that the SARS-CoV2 fusion peptide requires multiple negatively charged residues to 462 bind to calcium during viral entry. Specifically, our data implicate FP residues D802 and D825 in 463 coordinating the FP's interaction with calcium. 464

To better understand how the various charged residues in the SARS-CoV FP coordinate 465 calcium binding, we undertook molecular dynamic (MD) simulations with the FP and calcium 466 ions. The MD simulations of the Ca 2+ -loaded peptide with the membrane illuminated the way 467

Ca 2+ binding to FP binding may affect its function. Exploring the probability of each FP residue 468 interacting with the membrane in the identified Ca 2+ -binding modes reveals the preferred mode 469 of peptide insertion to be with the N-terminal end interacting with a calcium ion coordinated by 470 residues E801 and D802 (Fig 7) . Ca 2+ binding site located near the N-terminus of the peptide. This site (E801/D802) is made 480 accessible following enzymatic cleavage and is inserted into the membrane via the insertion of 481 the juxtaposed hydrophobic segment (the LLF motif). In this process, all anionic residues of the 482 peptide are either engaged with Ca 2+ ions, with neighboring basic residues, or remain solvent 483 exposed away from the membrane. Based on the preferred modes of Ca 2+ -loaded FP interaction 484 with the membrane, we propose that the binding of Ca 2+ following S protein cleavage at the S2' 485 site creates an energetically feasible membrane insertion process (Fig 7A) (Fig 8) . Thus, as shown in Figs 6 and 7, mutations that creates a high propensity for the 512 Ca 2+ binding mode involving the E821/D825 pair, stabilize peptide conformations that are non-513 productive for membrane penetration and are found here to inactivate the FP (Fig 6) . Without the reinforcements, we may lose the optimal structural conformation of the S protein 525 upon receptor binding to expose the S2' cleavage site to promote successful proteolytic cleavage 526 [30] . 527

More broadly, the E801 and D830 residues are conserved within the CoV family and the D812 528 residue is conserved within betacoronaviruses [18, 19] . The implication is that calcium 529 interactions are a conserved mechanism that serves to better position the FP for membrane 530 insertion. Different coronaviruses exhibit different requirements for calcium; MERS-CoV binds 531 to one Ca 2+ ion in its FP1 domain [20], thus, it is important to investigate the role of Ca 2+ and FP 532 interactions across the CoV family. The conservation of calcium-binding residues in the FP of 533 many coronaviruses suggests that the CoV fusion mechanisms can be potential targets for broad-534 spectrum antiviral drugs [35] [36] [37] . Repurposing FDA-approved calcium channel blocking (CCB) 535 drugs to inhibit CoV entry, particularly for SARS-CoV-2, is one option worth exploring. Recent 536 studies have shown that the CCB felodipine is a potential candidate to inhibit SARS-CoV-2 entry 537 [35] . CCBs can target conserved viral functions, providing a rapid solution to address new and 538 future SARS-CoV-2 variants. It will be important to identify the mechanisms of CCBs CoV 539 inhibition, as they may directly inhibit a viral target or indirectly inhibit viral entry by affecting 540 host cell processes 541

In this study, we elucidate the relationship between highly conserved residues in the SARS 542

CoV FP and the critical role they have in coordinating calcium binding to facilitate viral entry. 543

Interestingly, SARS-CoV-2 variants have arisen as part of Clade 20A that contain a D839G or 544 D839Y mutation (E821 equivalent in SARS-CoV), with this mutation predicted to affect FP-Ca 2+ 545 interactions [38] [39] [40] . To date, it is not known if there is any selective advantage to the virus 546 conferred by this mutation or whether the emergence of these variants simply represents a 547 founder effect. 548

Regarding to the role calcium plays during SARS-CoV entry, our data points to the necessity 549 of intra-and extra-cellular calcium during infection. Loss of extracellular calcium results in a 50% 550 reduction in PP infectivity, while the loss of intracellular calcium rendered pseudoparticles 551 noninfectious. Together with the MD simulations, we propose a model of SARS-CoV viral entry 552 mediated, in part, by calcium. In this model, upon S2' cleavage the spike protein's FP is exposed 553 and stabilized by binding to 2 calcium ions through electrostatic interactions. We believe that the 554 residues that likely mediate these essential interactions with calcium are E801/D802 and 555 D812/E821. Stabilization of the FP is critical for host membrane insertion, given the unstructured 556 and flexible FP2 loop in this peptide. We hypothesize that calcium interaction helps stabilize the 557 FP structure prior to membrane insertion, thus in the absence of a single negatively charged 558 residue, the FP can compensate through alternative residues to bind to calcium. Overall, more 559 studies are needed to fully understand the role calcium plays in CoV FP function and its effects 560 on viral pathogenesis. For the simulations in water, one copy of the peptide (wild-type or a mutant) was embedded 675 in a rectangular solution box and ionized using VMD tools ("Add Solvation Box" and "Add Ions", 676 respectively) [43] . The box of dimensions ~90 Å x 80 Å x 82 Å included a Na + Clionic solution as 677 well as 2 Ca 2+ ions, and ~18000 water molecules. The total number of atoms in the system was 678 ~54,540. 679

The system was equilibrated with NAMD version 2.13 [44] following a multi-step protocol 680 during which the backbone atoms of the SARS-CoV FP as well as Ca 2+ ions in the solution were 681 first harmonically constrained and subsequently gradually released in four steps (totaling ~3ns), 682 changing the restrain force constants kF from 1, to 0.5, to 0.1 kcal/ (mol Å 2 ), and 0 kcal/ (mol Å 2 ). 683

These simulations implemented all option for rigidbonds, 1fs (for kF 1, 0.5, and 0.1 kcal/ (mol Å 2 )) 684 or 2fs (for kF of 0) integration time-step, PME for electrostatics interactions [45] , and were carried 685 out in NPT ensemble under isotropic pressure coupling conditions, at a temperature of 310 K. 686

The Nose-Hoover Langevin piston algorithm [46] was used to control the target P = 1 atm 687 pressure with the "LangevinPistonPeriod" set to 200 fs and "LangevinPistonDecay" set to 50 fs. 688

The van der Waals interactions were calculated applying a cutoff distance of 12 Å and switching 689 the potential from 10 Å. 690

After this initial equilibration phase, the velocities of all atoms in the system were reset and 691 ensemble MD runs were initiated with OpenMM version 7.4 [47] during which the system was 692 simulated in 18 independent replicates, each for 640ns (i.e., cumulative time of ~11.5 µs for each 693 FP construct). These runs implemented PME for electrostatic interactions and were performed 694 at 310K temperature under NVT ensemble. In addition, 4fs time-step was used, with hydrogen 695 mass repartitioning and with "friction" parameter set to 1.0/picosecond. Additional parameters 696 for these runs included: "EwaldErrorTolerance" 0.0005, "rigidwater" True, and 697 "ConstraintTolerance" 0.000001. The van der Waals interactions were calculated applying a 698 cutoff distance of 12 Å and switching the potential from 10 Å. Na + Clsalt concentration), each system was equilibrated with NAMD version 2.13 following the 707 same multi-step protocol described above during which the backbone atoms of the FP as well as 708 the Ca 2+ ions were first harmonically constrained and subsequently gradually released in four 709 steps. After this phase, the velocities of all atoms of the system were reset, and ensemble MD 710 runs were initiated with OpenMM version 7.4. Each system was simulated in 18 independent 711 replicates, each ran for ~ 1 µs (i.e., cumulative time of ~18 µs for each FP-membrane complex). 712

These runs implemented PME for electrostatic interactions and were performed at 298K 713 temperature under NPT ensemble using semi-isotropic pressure coupling, with 4fs time-steps, 714 using hydrogen mass repartitioning and with "friction" parameter set to 1.0/picosecond. 715

Additional parameters for these runs included: "EwaldErrorTolerance" 0.0005, "rigidwater" True, 716

and "ConstraintTolerance" 0.000001. The van der Waals interactions were calculated applying a 717 cutoff distance of 12 Å and switching the potential from 10 Å. 718 pre-fusion FP domain (from PDB: 5XLR), and summarizes the data obtained from our work, 812

highlighting the role of key charged residues within the fusion peptide. We depict the insertion 813 of the FP1 domain with E801/D802-Ca 2+ (green) needed for membrane association and with D812 814 able to coordinate Ca 2+ and pair with either E821 to promote membrane interaction (green) or 815 with D825 to inhibit membrane interaction (red). 

The Molecular Biology of Coronaviruses

The Molecular Biology of Coronaviruses

SARS: the first pandemic of the 21st century

Middle East respiratory 828 syndrome coronavirus (MERS-CoV): announcement of the Coronavirus Study Group

A new coronavirus associated with 832 human respiratory disease in China

A pneumonia outbreak associated 836 with a new coronavirus of probable bat origin

World Health Organization methodology to prioritize 840 emerging infectious diseases in need of research and development

The emergence of SARS, MERS and novel SARS-2 843 coronaviruses in the 21st century

The COVID-19 epidemic

Coronavirus membrane fusion 848 mechanism offers a potential target for antiviral development

SARS and MERS: recent insights into 852 emerging coronaviruses

Mechanisms of coronavirus cell entry 855 mediated by the viral spike protein

Physiological and molecular triggers for SARS-CoV membrane fusion 858 and entry into host cells

Animal models for SARS-CoV-2. Current Opinion in Virology

Host cell proteases: Critical determinants of coronavirus tropism and 863 pathogenesis

Proteases and variants: context matters for SARS-CoV-2 867 entry assays. Current Opinion in Virology

Fusion of Enveloped Viruses in Endosomes

Characterization of a highly conserved domain 872 within the severe acute respiratory syndrome coronavirus spike protein S2 domain with 873 characteristics of a viral fusion peptide

The SARS-CoV Fusion Peptide Forms an 877

Extended Bipartite Fusion Platform that Perturbs Membrane Order in a Calcium-Dependent 878 Manner

Ca(2+) Ions Promote Fusion of 882

Middle East Respiratory Syndrome Coronavirus with Host Cells and Increase Infectivity

Epub 2020/04/17

Calcium Ions Directly 886 Interact with the Ebola Virus Fusion Peptide To Promote Structure-Function Changes That 887 Enhance Infection

SARS-CoV-2 Fusion Peptide has a Greater Membrane Perturbating Effect 891 than SARS-CoV with Highly Specific Dependence on Ca2+

Ca<sup>2+</sup>-dependent mechanism 894 of membrane insertion and destabilization by the SARS-CoV-2 fusion peptide

SARS-CoV fusion peptides induce 897 membrane surface ordering and curvature

Fusion Peptide of SARS-CoV-2 Spike Rearranges into 900 a Wedge Inserted in Bilayered Micelles

SARS coronavirus entry into host cells 903 through a novel clathrin-and caveolae-independent endocytic pathway

Production of Pseudotyped 907 Particles to Study Highly Pathogenic Coronaviruses in a Biosafety Level 2 Setting

Proteolytic Activation of 910 SARS-CoV-2 Spike at the S1/S2 Boundary: Potential Role of Proteases beyond Furin

SARS-coronavirus spike S2 domain flanked by cysteine 914 residues C822 and C833 is important for activation of membrane fusion

The spike protein of SARS-CoV--a target for vaccine 918 and therapeutic development

Bat-to-human: spike features determining 'host jump' of 922 coronaviruses SARS-CoV, MERS-CoV, and beyond

Rubella Virus: First Calcium-Requiring Viral Fusion Protein

Binding of SARS-CoV-2 Fusion Peptide to Host tmpr and 927 Plasma Membrane

Distinct conformational states of 930 SARS-CoV-2 spike protein

FDA approved calcium channel blockers 932 inhibit SARS-CoV-2 infectivity in epithelial lung cells

Danta CC. Calcium Channel Blockers: A Possible Potential Therapeutic Strategy for the 937 Treatment of Alzheimer's Dementia Patients with SARS-CoV-2 Infection

Massive dissemination 941 of a SARS-CoV-2 Spike Y839 variant in Portugal

Tracking Changes 944 in SARS-CoV-2 Spike: Evidence that D614G Increases Infectivity of the COVID-19 Virus

The Impact of Mutations in SARS-CoV-2 Spike on 947 Viral Infectivity and Antigenicity

Activation of the SARS coronavirus spike protein via 949 sequential proteolytic cleavage at two distinct sites. Proceedings of the National Academy of 950 Sciences

PROPKA3: Consistent Treatment of 952 Internal and Surface Residues in Empirical pKa Predictions

VMD: Visual molecular dynamics

Scalable molecular 957 dynamics with NAMD

A smooth particle mesh 959 Ewald method

The Nose-Hoover thermostat

OpenMM 7: 964 Rapid development of high performance algorithms for molecular dynamics

CHARMM36m: an 967 improved force field for folded and intrinsically disordered proteins

Simulations of Anionic Lipid Membranes: 970 Development of Interaction-Specific Ion Parameters and Validation Using NMR Data. The 971

Fusion activity of the SARS-CoV wild-type and mutant S-expressing cells

Immunofluorescence images of VeroE6 cells expressing WT or mutant (single/double/triple) S 1003 constructs. Following transfection, cells were treated with trypsin to cleave SARS-CoV S proteins 1004 and induce syncytium formation. Syncytia were visualized using a SARS-CoV S antibody (green) 1005 and nuclei appear in blue (DAPI)

In a particular trajectory, residue pairs simultaneously 1012 engaging with the bound Ca 2+ are denoted by red or blue rectangles, whereas instances of Ca 2+ 1013 ion associating with a single acidic residue is depicted with grey-striped rectangle. The 1014 trajectories in which simultaneous binding of two Ca 2+ ions to different pairs of residues were 1015 observed are highlighted in green. The simulations in which a single Ca 2+ ion was bound to a

 The tables show,  1020 for each construct, acidic residues implicated in Ca 2+ binding in 18 independent atomistic MD 1021 simulations (each 640ns in length). In a particular trajectory, residue pairs simultaneously 1022 engaging with the bound Ca 2+ are denoted by red or blue rectangles, whereas instances of Ca 2+ 1023 ion associating with a single acidic residue is depicted with grey-striped rectangle. The 1024 trajectories in which simultaneous binding of two Ca 2+ ions to different pairs of residues were 1025 observed are highlighted in green. The simulations in which a single Ca 2+ ion was bound to a pair 1026 of acidic residues are shown in yellow. 1027 1028