The Development of a Novel Nanobody Therapeutic for SARS-CoV-2

Combating the COVID-19 pandemic requires potent and low-cost therapeutics. We identified a novel series of single-domain antibodies (i.e., nanobody), Nanosota-1, from a camelid nanobody phage display library. Structural data showed that Nanosota-1 bound to the oft-hidden receptor-binding domain (RBD) of SARS-CoV-2 spike protein, blocking out viral receptor ACE2. The lead drug possessing an Fc tag (Nanosota-1C-Fc) bound to SARS-CoV-2 RBD with a Kd of 15.7picomolar (~3000 times more tightly than ACE2 did) and inhibited SARS-CoV-2 infection with an ND50 of 0.16microgram/milliliter (~6000 times more potently than ACE2 did). Administered at a single dose, Nanosota-1C-Fc demonstrated preventive and therapeutic efficacy in hamsters subjected to SARS-CoV-2 infection. Unlike conventional antibody drugs, Nanosota-1C-Fc was produced at high yields in bacteria and had exceptional thermostability. Pharmacokinetic analysis of Nanosota-1C-Fc documented a greater than 10-day in vivo half-life efficacy and high tissue bioavailability. Nanosota-1C-Fc is a potentially effective and realistic solution to the COVID-19 pandemic.


Introduction
The ability of the Nanosota-1 drugs to neutralize SARS-CoV-2 infection in vitro 171 was investigated nex. Both a SARS-CoV-2 pseudovirus entry assay and authentic SARS-172 CoV-2 infection assay were performed (Fig. 3). For the pseudovirus entry assay, 173 retroviruses pseudotyped with SARS-CoV-2 spike protein (i.e., SARS-CoV-2 174 pseudoviruses) were used to enter human ACE2-expressing HEK293T cells in the 175 presence of an inhibitor. The efficacy of the inhibitor was expressed as the concentration 176 capable of neutralizing 50% of the entry efficiency (i.e., 50% Neutralizing Dose or 177 ND 50 ). Nanosota-1C-Fc had an ND 50 for the SARS-CoV-2 pseudovirus of 0.27 µg/ml, 178 which was ~10 times more potent than monovalent Nanosota-1C (2.52 µg/ml) and over 179 100 times more potent than ACE2 (44.8 µg/ml) (Fig. 3A). Additionally, Nanosota-1 180 drugs potently neutralized SARS-CoV-2 pseudovirus bearing the D614G mutation in the 181 SARS-CoV-2 spike protein (Fig. S5), which has become prevalent in many strains (25). 182 For the authentic virus infection assay, live SARS-CoV-2 was used to infect Vero cells in 183 the presence of an inhibitor. Efficacy of the inhibitor was described as the concentration 184 capable of reducing the number of virus plaques by 50% (i.e., ND 50 ). Nanosota-1C-Fc 185 had an ND 50 of 0.16 µg/ml, which was ~20 times more potent than monovalent 186 Nanosota-1C (3.23 µg/ml) and ~6000 times more potent than ACE2 (980 µg/ml) ( inoculation. In addition to an untreated control group, three groups of animals were 192 injected with a single dose of Nanosota-1C-Fc: (i) 24 hours pre-challenge at 20 mg/kg 193 . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted November 17, 2020. ; https://doi.org/10.1101/2020.11.17.386532 doi: bioRxiv preprint body weight, (ii) 4 hours post-challenge at 20 mg/kg, and (iii) 4 hours post-challenge at 194 10 mg/kg. As previously validated in this model (26), body weight, tissue pathology and 195 virus titers in nasal swabs were used as metrics of therapeutic efficacy. In the untreated 196 control group, weight loss was precipitously starting on day 1 post-challenge with the 197 lowest weight recorded on day 6 (Fig. 4A). Nasal virus titers were high on day 1 and 198 remained high on day 5 before a decline (Fig. S7). Pathology analysis on tissues collected 199 on day 10 revealed moderate hyperplasia in the bronchial tubes (i.e., bronchioloalveolar Nanosota-1C-Fc 24-hours pre-challenge were protected from SARS-CoV-2, as 204 evidenced by the metrics of no weight loss, no bronchioloalveolar hyperplasia, and 205 significantly reduced nasal virus titers (Fig. 4, Fig. S7). When administered 4 hours post-206 challenge, Nanosota-1C-Fc also effectively protected hamsters from SARS-CoV-2 207 infections at either dosage (20 or 10 mg/kg), as evidenced by the favorable therapeutic 208 metrics (Fig. 4, Fig. S7). Overall, Nanosota-1C-Fc was effective at combating SARS-209 CoV-2 infections both preventively and therapeutically. 210

Nanosota-1C-Fc is stable in vitro and in vivo with excellent bioavailability 211
With the lead drug Nanosota-1C-Fc demonstrating therapeutic efficacy in vivo, 212 we characterized other parameters important to its clinical translation. First, we expressed 213 Nanosota-1C-Fc in bacteria for all the experiments carried out in the current study (Fig.  214 5A). After purification on protein A column and gel filtration, the purity of Nanosota- 1C-215 Fc was nearly 100%. With no optimization, the expression yield reached 40 mg/L of 216 . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted November 17, 2020. ; https://doi.org/10.1101/2020.11.17.386532 doi: bioRxiv preprint bacterial culture. Second, we investigated the in vitro stability of Nanosota-1C-Fc 217 incubated at four temperatures (-80 o C, 4 o C, 25 o C or 37 o C) for one week and then 218 measured the remaining SARS-CoV-2 RBD-binding capacity by ELISA (Fig. 5B). With 219 -80 o C as a baseline, Nanosota-1C-Fc retained nearly all of its RBD-binding capacity at 220 the temperatures surveyed. Third, we measured the in vivo stability of Nanosota-1C-Fc 221 (Fig. 5C). Nanosota-1C-Fc was injected into mice via tail vein. Sera were obtained at 222 different time points and measured for their SARS-CoV-2 RBD-binding capacity by 223 ELISA. Nanosota-1C-Fc retained most of its RBD-binding capability after 10 days in 224 vivo. Antithetically, Nanosota-1C was stable for only several hours in vivo. (Fig. S8A). 225 Last, we examined the biodistribution of Nanosota-1C-Fc in mice (Fig. 5D). Nanosota-226 1C-Fc was radiolabeled with zirocinium-89 and injected systemically into mice. Tissues 227 were collected at various time points and biodistribution of Nanosota-1C-Fc was 228 quantified by scintillation counter. After three days, Nanosota-1C-Fc remained at high 229 levels in the blood, lung, heart, kidney, liver and spleen, all of which are targets for 230 SARS-CoV-2 (27); moreover, it remained at low levels in the intestine, muscle and 231 bones. In contrast, Nanosota-1C had poor biodistribution documenting high renal 232 clearance (Fig. S8B). Overall, our findings suggest that Nanosota-1C-Fc is potent SARS-233 CoV-2 therapeutic with translational values applicable to the world's vast population. 234 235

Discussion 236
Nanobody therapeutics derived from camelid antibodies potentially offer a 237 realistic solution to the COVID-19 pandemic compared to conventional antibodies. 238 Currently, there have only been a few reports of nanobody drugs that specifically target 239 . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted November 17, 2020. ; https://doi.org/10.1101/2020.11.17.386532 doi: bioRxiv preprint SARS-CoV-2 (12, 13). Those reported were developed against SARS-CoV-2 RBD, 240 either blocking out ACE2 or locking the RBD in the closed inactive state on the spike 241 protein (12, 13). None of the nanobodies have been evaluated in animal models for their 242 anti-SARS-CoV-2 therapeutic efficacy. From our novel library, we developed a series of 243 nanobody drugs, named Nanosota-1, that specifically target the SARS-CoV-2 RBD. Two 244 rounds of affinity maturation yielded Nanosota-1C which bound to the RBD with high 245 affinity. Addition of an Fc tag to make a bivalent construct with increased molecular 246 weight and picomolar RBD-binding affinity resulted in the best performing drug 247 Nanosota-1C-Fc. Our structural and biochemical data showed that binding of Nanosota-248 1C to the RBD blocked virus binding to viral receptor ACE2. A unique feature of the 249 SARS-CoV-2 spike protein is that it is present in two different conformations, an RBD-250 up open conformation for receptor binding and an RBD-down closed conformation for 251 immune evasion (20, 22, 23). Due to its small size as well as its ideal binding site on the 252 RBD, Nanosota-1 can bind to the spike protein in both conformations. In contrast, ACE2 253 can only bind to the spike protein in its open conformation. Thus, Nanosota-1 drugs are 254 ideal RBD-targeting therapeutics -they can chase down and inhibit SARS-CoV-2 viral 255 particles whether they are infecting cells or hiding from immune surveillance. As a result 256 of this unique property, both Nanosota-1C and Nanosota-1C-Fc exhibited a profound 257 therapeutic effect in vitro against SARS-CoV-2 pseudovirus and authentic SARS-CoV-2. 258 Nanosota-1C-Fc was also found to be the first anti-SARS-CoV-2 camelid nanobody-259 based therapeutic reported in the literature to demonstrate efficacy in an animal model. 260 Additionally, Nanosota-1C-Fc was the first anti-SARS-CoV-2 nanobody to have been 261 characterized for ease of production and purification, in vitro and in vivo stabilities, and 262 . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted November 17, 2020. ; https://doi.org/10.1101/2020.11.17.386532 doi: bioRxiv preprint biodistribution. These features are critical for the implementation of Nanosota-1C-Fc as a 263

COVID-19 therapeutic. 264
When evaluating the anti-SARS-CoV-2 potency of the nanobody therapeutics, we 265 used recombinant ACE2 as a comparison. Recombinant ACE2 was selected because 266 Nanosota-1 series directly compete with cell-surface ACE2 for the same binding site on 267 the RBD. Our study showed that compared with ACE2, the best performing drug 268 Nanosota-1C-Fc bound to the RBD ~3000 fold more strongly, blocking out ACE2 269 binding to the RBD. Furthermore, compared with ACE2, Nanosota-1C-Fc inhibited 270 SARS-CoV-2 pseudovirus entry ~100 fold more effectively and inhibited authentic 271 SARS-CoV-2 infections ~6000 fold more effectively. Note that recombinant ACE2 has 272 been shown to be a potent anti-SARS-CoV-2 inhibitor (28) and is currently undergoing 273 clinical trials in Europe as an anti-COVID-19 drug. Compared with ACE2, the much 274 higher anti-SARS-CoV-2 potency of Nanosota-1C-Fc was due to both its much higher 275 RBD-binding affinity and its better access to the oft-hidden RBD in the spike protein. of the best non-primate models available for studying anti-SARS-CoV-2 therapeutic 280 efficacy, but it is limited by a short virus infection window; hence, repeated dosing was 281 not evaluated. As a result, we were only able to dose the mice once via intraperitoneal 282 injection. Because SARS-CoV-2 is fast acting in hamsters, the time points and dosages 283 for drug administration in hamsters are difficult to directly translate to humans. Our 284 supporting data document that Nanosota-1c-Fc is easy to produce in bacteria and has 285 . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted November 17, 2020. ; https://doi.org/10.1101/2020.11.17.386532 doi: bioRxiv preprint excellent bioavailability and pharmacokinetics when administered intravenously in mice. 286 This suggests that Nanosota-1C-Fc may have therapeutic potential when administered 287 intraperitoneal, intravenous or even intramuscular. These parameters will need to be 288 determined in future studies in anticipation of clinical trials. Overall, Nanosota-1C-Fc 289 has proven to be an effective therapeutic in the model that we currently have available. 290 How can the novel nanobody therapeutics help to end the COVID-19 pandemic? 291 populations. Second, we also learned from our in vivo study that Nanosota-1C-Fc can 296 potentially be used to treat SARS-CoV-2 infections, thus, saving lives and alleviating 297 symptoms in infected patients in the clinical setting. Third, though ephemeral in nature 298 given its short half-life and rapid clearance from the blood, Nanosota-1C could be used 299 as an inhaler to treat infections in the respiratory tracts (10) or as an oral drug to treat 300 infections in the intestines (11). Overall, the novel series of Nanosota-1 therapeutics can 301 help minimize the mortality and morbidity of SARS-CoV-2 infections and help restore 302 the economy and daily human activities. Given the wide distribution of SARS-CoV-2 in 303 the world, large quantities of anti-SARS-CoV-2 therapeutics would need to be 304 manufactured to provide for the world's populations. This is only feasible with easy to 305 produce and scalable molecules, such as Nanosota-1 drugs, that are produced at high 306 yields and have long in vitro and in vivo half-life. Therefore, if further validated in 307 clinical trials, Nanosota-1 therapeutics can provide a realistic and effective solution to 308 . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted November 17, 2020. ; https://doi.org/10.1101/2020.11.17.386532 doi: bioRxiv preprint help end the COVID-19 global pandemic. 309 . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted November 17, 2020. ; https://doi.org/10.1101/2020.11.17.386532 doi: bioRxiv preprint

Acknowledgements 310
The development of Nanosota-1 drugs and the animal testing were supported by 311 funding from the University of Minnesota (to F.L.), NIH grants R01AI157975 (to F.L., (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted November 17, 2020. ; https://doi.org/10.1101/2020.11.17.386532 doi: bioRxiv preprint

Ethics statement 326
This study was performed in strict accordance with the recommendations in the 327 Guide for the Care and Use of Laboratory Animals of the National Institutes of Health. Lenti-CMV vector (Vigene Biosciences) with an N-terminal tissue plasminogen activator 344 (tPA) signal peptide and a C-terminal human IgG4 Fc tag or His tag. The ACE2 345 ectodomain (residues 1-615) was constructed in the same way except that its own signal 346 peptide was used. Nanosota-1A, -1B and -1C were each cloned into PADL22c vector 347 . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted November 17, 2020. ; https://doi.org/10.1101/2020.11.17.386532 doi: bioRxiv preprint (Lucigen) with a N-terminal PelB leader sequence and C-terminal His tag and HA tag. 348

Nanosota-1C-Fc was cloned into pET42b vector (Novagen) with a C-terminal human 349
IgG 1 Fc tag. SARS-CoV-2 (US_WA-1 isolate) from CDC (Atlanta) was used throughout 350 the study. All experiments involving infectious SARS-CoV-2 were conducted at the 351 University of Texas Medical Branch and University of Iowa in approved biosafety level 3 352

Construction of camelid nanobody phage display library 355
The camelid nanobody phage display library was constructed as previously 356 described (29, 30). Briefly, total mRNA was isolated from B cells from the spleen, bone 357 marrow and blood of over a dozen non-immunized llamas and alpacas. cDNA was 358 prepared from the mRNA. The cDNA was then used in nested PCR reactions to construct 359 the DNA for the library. The first PCR reaction was to amplify the gene fragments 360 encoding the variable domain of the nanobody. The second PCR reaction (PCR2) was 361 used to add restriction sites (SFI-I), a PelB leader sequence, a His 6 tag, and a HA tag. The 362 PCR2 product was digested with SFI-I (New England Biolabs) and then was ligated with 363 SFI-I-digested PADL22c vector. The ligated product was transformed via electroporation 364 into TG1 E. coli (Lucigen). Aliquots of cells were spread onto 2YT agar plates 365 supplemented with ampicillin and glucose, incubated at 30°C overnight, and then scraped 366 into 2YT media. After centrifugation, the cell pellet was suspended into 50% glycerol 367 and stored at -80°C. The library size was 7. 5 × 10 10 . To display nanobodies on phages, 368 aliquots of the TG1 E. coli bank were inoculated into 2YT media, grown to early 369 logarithmic phase, and infected with M13K07 helper phage. 370 . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted November 17, 2020. ; https://doi.org/10.1101/2020.11.17.386532 doi: bioRxiv preprint

371
Camelid nanobody library screening 372 The above camelid nanobody phage display library was used in the bio-panning 373 as previously described (31). Briefly, four rounds of panning were performed to obtain 374 the SARS-CoV-2 RBD-targeting nanobodies with high RBD-binding affinity. The 375 amounts of the RBD antigen used in coating the immune tubes in each round were 75 µg, 376 50 µg, 25 µg, and 10 µg, respectively. The retained phages were eluted using 1 ml 100 377 mM triethylamine and neutralized with 500 µl 1 M Tris-HCl pH 7.5. The eluted phages 378 were amplified in TG1 E. coli and rescued with M13K07 helper phage. The eluted 379 phages from round 4 were used to infect ss320 E. coli. Single colonies were picked into 380 2YT media and nanobody expressions were induced with 1 mM IPTG. The supernatants 381 were subjected to ELISA for selection of strong binders (described below). The strong 382 binders were then expressed and purified (described below) and subjected to SARS-CoV-383 2 pseudovirus entry assay for selection of anti-SARS-CoV-2 efficacy (described below). 384 The lead nanobody after initial screening was named Nanosota-1A. 385 386

Affinity maturation 387
Affinity maturation of Nanosota-1A was performed as previously described (32). 388 Briefly, mutations were introduced into the whole gene of Nanosota-1A using error-prone 389 PCR. Two rounds of error-prone PCR were performed using the GeneMorph II Random 390 Mutagenesis Kit (Agilent Technologies). The PCR product was cloned into the PADL22c 391 vector and transformed via electroporation into the TG1 E. coli. The library size was 6 x 392 10 8 . Three rounds of bio-panning were performed using 25 ng, 10 ng and 2 ng RBD-Fc, 393 . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted November 17, 2020. ; https://doi.org/10.1101/2020.11.17.386532 doi: bioRxiv preprint respectively. The strongest binder after affinity maturation was named Nanosota-1B. A 394 second round of affinity maturation was performed in the same way as the first round, 395 except that three rounds of bio-panning were performed using 10 ng, 2 ng and 0.5 ng 396 RBD-Fc, respectively. The strongest binder after the second round of affinity maturation 397 was named Nanosota-1C. (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted November 17, 2020. ; https://doi.org/10.1101/2020.11.17.386532 doi: bioRxiv preprint (http://www.addgene.org/protocols/plko/). The proteins were secreted to cell culture 417 media, harvested, and purified on either Ni-NTA column (for His-tagged proteins) or 418 protein A column (for Fc-tagged protein) and then on Superdex200 gel filtration column 419 as previously described (20). (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted November 17, 2020. ; https://doi.org/10.1101/2020.11.17.386532 doi: bioRxiv preprint soaked briefly in 50 mM MnCl 2 , 50 mM MES pH 6.0, 25% (W/V) PEG 4000 and 30% 440 ethylene glycol before being flash-frozen in liquid nitrogen. X-ray diffraction data were 441 collected at the Advanced Photon Source beamline 24-ID-E. The structure was 442 determined by molecular replacement using the structures of SARS-CoV-2 RBD (PDB 443 6M0J) and another nanobody (PDB 6QX4) as the search templates. Structure data and 444 refinement statistics are shown in Table S1. After one-hour incubation at room temperature, the bound proteins were eluted using 462 . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted November 17, 2020. incubated together at 37°C for 6 hours, the medium was changed to fresh medium, 481 followed by incubation of another 60 hours. Cells were then washed with PBS buffer and 482 lysed. Aliquots of cell lysates were transferred to plates, followed by the addition of 483 luciferase substrate. Relative light units (RLUs) were measured using an EnSpire plate 484 . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted November 17, 2020. ; https://doi.org/10.1101/2020.11.17.386532 doi: bioRxiv preprint reader (PerkinElmer). The efficacy of the drug was expressed as the concentration 485 capable of neutralizing 50% of the entry efficiency (Neutralizing Dose 50 or ND 50 ). 486

SARS-CoV-2 plaque reduction neutralization test 488
The potency of Nanosota-1 drugs in neutralizing authentic SARS-CoV-2 489 infections was evaluated using a SARS-CoV-2 plaque reduction neutralization test 490 indicates that we can detect an effect size of 1.6 with a power of .80 (alpha = .05 one-505 tailed). Four groups of hamsters (n=6 each randomly assigned) were treated with 506 Nanosota-1C-Fc via intraperitoneal injection at one of the following time points and 507 . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted November 17, 2020. ; https://doi.org/10.1101/2020.11.17.386532 doi: bioRxiv preprint dosages: (1) 24 hours pre-challenge at 20 mg/kg body weight of hamsters; (2) 4 hours 508 post-challenge at 20 mg/kg body weight of hamsters; (3) 4 hours post-challenge at 10 509 mg/kg body weight of hamsters. Hamsters in the control (negative) group were 510 administered PBS buffer 24 hours pre-challenge. An additional group was tested for a 511 different hypothesis and the data were not included in the current study. Body weights 512 were collected daily beginning prior to challenge. Nasal swabs were collected prior to 513 challenge and additionally 1 day, 2 days, 3 days, 5 days and 10 days post-challenge for 514 quantitative real-time RT-PCR (nasal swabs collected on day 2 and day 3 were lost due to 515 Hurricane Laura). Hamsters were humanely euthanized 10 days post-challenge via 516 overexposure to CO 2 . The lungs and bronchial tubes were collected and fixed in formalin 517 for histopathological analysis. This experiment was performed in accordance with the 518 guidelines set by the Institutional Animal Care and Use Committee at the University of 519 Texas Medical Branch (UTMB). 520 521

Half-life of Nanosota-1 drugs in mice 522
Male C57BL/6 mice (3 to 4 weeks old) (Envigo) were intravenously injected (tail-523 vein) with Nanosota-1C or Nanosota-1C-Fc (100 µg in 100 µl PBS buffer). At varying 524 time points, mice were euthanized and whole blood was collected. Then sera were 525 prepared through centrifugation of the whole blood at 1500xg for 10 min. The sera were 526 then subjected to ELISA for evaluation of their SARS-CoV-2 RBD-binding capability. 527 528 Biodistribution of Nanosota-1 drugs in mice 529 . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted November 17, 2020. ; https://doi.org/10.1101/2020.11.17.386532 doi: bioRxiv preprint To evaluate the in vivo biodistribution of Nanosota-1C-Fc and Nanosota-1C, the 530 nanobodies were labeled with Zirconium-89 [ 89 Zr] and injected into male C57BL/6 mice 531 (5 to 6 weeks old) (Envigo). Briefly, the nanobodies were first conjugated to the 532 bifunctional chelator p-SCN-Bn-Deferoxamine (DFO, Macrocyclic) as previously 533 (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted November 17, 2020. ; https://doi.org/10.1101/2020.11.17.386532 doi: bioRxiv preprint . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted November 17, 2020. ; https://doi.org/10.1101/2020.11.17.386532 doi: bioRxiv preprint Table 1. Binding affinities between Nanosota-1 drugs and SARS-CoV-2 RBD as 629 measured using surface plasmon resonance. The previously determined binding 630 affinity between human ACE2 and RBD is shown as a comparison (20). . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted November 17, 2020. ; https://doi.org/10.1101/2020.11.17.386532 doi: bioRxiv preprint Retroviruses pseudotyped with SARS-CoV-2 spike protein (i.e., SARS-CoV-2 655 pseudoviruses) were used to enter HEK293T cells expressing human ACE2 in the 656 presence of the inhibitor at various concentrations. Entry efficiency was characterized via 657 . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted November 17, 2020. ; https://doi.org/10.1101/2020.11.17.386532 doi: bioRxiv preprint a luciferase signal indicating successful cell entry. Data are the mean ± SEM (n = 4). 658 Nonlinear regression was performed using a log (inhibitor) versus normalized response 659 curve and a variable slope model (R 2 > 0.95 for all curves). The efficacy of each inhibitor 660 was expressed as the 50% Neutralizing Dose or ND 50 . The assay was repeated three times 661 (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted November 17, 2020. ; https://doi.org/10.1101/2020.11.17.386532 doi: bioRxiv preprint Data are the mean ± SEM (n = 6). ANOVA on group as a between-group factor and day 681 (1-10) as a within-group factor revealed significant differences between the control group 682 and each of the following groups: 24 hour pre-challenge (20 mg/kg) group (F(1, 10)  (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted November 17, 2020. ; https://doi.org/10.1101/2020.11.17.386532 doi: bioRxiv preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted November 17, 2020. ; https://doi.org/10.1101/2020.11.17.386532 doi: bioRxiv preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted November 17, 2020. ; https://doi.org/10.1101/2020.11.17.386532 doi: bioRxiv preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted November 17, 2020. ; https://doi.org/10.1101/2020.11.17.386532 doi: bioRxiv preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted November 17, 2020. ; https://doi.org/10.1101/2020.11.17.386532 doi: bioRxiv preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted November 17, 2020. ; https://doi.org/10.1101/2020.11.17.386532 doi: bioRxiv preprint . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted November 17, 2020. ; https://doi.org/10.1101/2020.11.17.386532 doi: bioRxiv preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted November 17, 2020. ; https://doi.org/10.1101/2020.11.17.386532 doi: bioRxiv preprint Figure S3. Measurement of the binding affinities between Nanosota-1 drugs and 760 SARS-CoV-2 RBD by surface plasmon resonance assay using Biacore. Purified 761 recombinant SARS-CoV-2 RBD was covalently immobilized on a sensor chip through its 762 amine groups. Purified recombinant nanobodies flowed over the RBD individually at one 763 of five different concentrations. The resulting data were fit to a 1:1 binding model and the 764 value of K d was calculated for each nanobody. The assay was repeated three times 765 (biological replication: new aliquots of proteins and new sensor chips were used for each 766 repeat). 767 768 . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted November 17, 2020. ; https://doi.org/10.1101/2020.11.17.386532 doi: bioRxiv preprint Figure S4. Binding interactions between Nanosota-1 drugs and SARS-CoV-2 RBD.

770
(A) Binding interactions between SARS-CoV-2 RBD, Nanosota-1C, and ACE2 as 771 evaluated using a protein pull-down assay. Various concentrations of Nanosota-1C and a 772 constant concentration of ACE2 (all His tagged) were combined in different molar ratios.

773
SARS-CoV-2 RBD (Fc tagged) was used to pull down Nanosota-1C and ACE2. A 774 western blot was used to detect the presence of Nanosota-1C and ACE2 following pull 775 down by SARS-CoV-2 RBD. The assay was repeated three times (biological replication: 776 new aliquots of proteins were used for each repeat). (B) Binding interactions between 777 SARS-CoV-2 RBD, Nanosota-1C, and ACE2 as examined using gel filtration 778 chromatography. Nanosota-1C, ACE2 and SARS-CoV-2 RBD (all His tagged) were 779 mixed together in solution (both Nanosota-1C and ACE2 in molar excess of SARS-CoV-780 2 RBD) and purified using gel filtration chromatography. Protein components in each of 781 the gel filtration chromatography peaks were analyzed with SDS-PAGE and stained by 782 Coomassie blue. The assay was repeated three times (biological replication: new aliquots 783 of proteins were used for each repeat). 784 785 . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted November 17, 2020. ; https://doi.org/10.1101/2020.11.17.386532 doi: bioRxiv preprint Figure S5. Neutralization of SARS-CoV-2 pseudovirus, which contains the D614G 787 mutation in the spike protein, by Nanosota-1 drugs. The procedure was the same as 788 described in Fig. 3A, except that the mutant spike protein replaced the wild type spike 789 protein. The assay was repeated three times (biological replication: new aliquots of 790 pseudoviruses and cells were used for each repeat). 791 792 . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted November 17, 2020. ; https://doi.org/10.1101/2020.11.17.386532 doi: bioRxiv preprint regression was performed using a log (inhibitor) versus normalized response curve and a 796 variable slope model (R 2 > 0.95 for all curves). The assay was repeated twice (biological 797 replication: new aliquots of virus particles and cells were used for each repeat). 798 799 . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted November 17, 2020. ; https://doi.org/10.1101/2020.11.17.386532 doi: bioRxiv preprint Figure S7. Additional data on the efficacy of Nanosota-1 drugs in protecting 801 hamsters from SARS-CoV-2 infections. Nasal swabs were collected from each hamster 802 on days 1, 2, 3, 5, 7, and 10. Nasal swab samples from day 2 and day 3 were lost due to 803 Hurricane Laura. qRT-PCR was performed to determine the virus loads in each of the 804 samples. The qRT-PCR results are displayed on a log scale (since qRT-PCR amplifies 805 signals on a log scale). Data are the mean ± SEM (n = 6). Missing data from one animal 806 in the 4-hour post-challenge (10mg/kg) group on Day 7 were replaced by the average of 807 that animal's days 5 and 10 data. ANOVA analysis using group as a between-group 808 factor and day (1, 5, 7, and 10) as a within-group factor revealed significant differences 809 between the control group and each of the following groups: 24 hour pre-challenge (20 810 mg/kg) group (F(1, 10) = 6.02, p = .017, effect size η p 2 = .38), 4 hour post-challenge (20 811 mg/kg) group (F(1, 10) = 5.38, p = .037, η p 2 = .31), and 4 hour post-challenge (10 mg/kg) 812 group (F(1, 10) = 3.40, p = .048, η p 2 = .25). All p-values are one-tailed for directional 813 tests. 814 815 . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted November 17, 2020. ; https://doi.org/10.1101/2020.11.17.386532 doi: bioRxiv preprint 816 Figure S8. Pharmacokinetics of Nanosota-1C. In vivo stability and biodistribution of 817 Nanosota-1C were measured in the same way as described in Fig. 5C and Fig. 5D, 818 respectively, except that time points for Nanosota-1C differed from those for Nanosota-819 1C-Fc due to pharmacokinetic differences of the small molecular weight nanobody 820 versus the larger Fc tagged nanobody. 821 822 . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted November 17, 2020. ; https://doi.org/10.1101/2020.11.17.386532 doi: bioRxiv preprint