key: cord-0941635-u7ge5pem
authors: Park, Sang Ho; Siddiqi, Haley; Castro, Daniela V.; De Angelis, Anna; Oom, Aaron L.; Stoneham, Charlotte A.; Lewinski, Mary K.; Clark, Alex E.; Croker, Ben A.; Carlin, Aaron F.; Guatelli, John; Opella, Stanley J.
title: Interactions of SARS-CoV-2 envelope protein with amilorides correlate with antiviral activity
date: 2021-04-06
journal: bioRxiv
DOI: 10.1101/2021.04.06.438579
sha: 747b31873c766b9c939f0add34df33fd030f8aa6
doc_id: 941635
cord_uid: u7ge5pem

SARS-CoV-2 is the novel coronavirus that is the causative agent of COVID-19, a sometimes-lethal respiratory infection responsible for a world-wide pandemic. The envelope (E) protein, one of four structural proteins encoded in the viral genome, is a 75-residue integral membrane protein whose transmembrane domain exhibits ion channel activity and whose cytoplasmic domain participates in protein-protein interactions. These activities contribute to several aspects of the viral replication-cycle, including virion assembly, budding, release, and pathogenesis. Here, we describe the structure and dynamics of full-length SARS-CoV-2 E protein in hexadecylphosphocholine micelles by NMR spectroscopy. We also characterized its interactions with four putative ion channel inhibitors. The chemical shift index and dipolar wave plots establish that E protein consists of a long transmembrane helix (residues 8-43) and a short cytoplasmic helix (residues 53-60) connected by a complex linker that exhibits some internal mobility. The conformations of the N-terminal transmembrane domain and the C-terminal cytoplasmic domain are unaffected by truncation from the intact protein. The chemical shift perturbations of E protein spectra induced by the addition of the inhibitors demonstrate that the N-terminal region (residues 6-18) is the principal binding site. The binding affinity of the inhibitors to E protein in micelles correlates with their antiviral potency in Vero E6 cells: HMA ≈ EIPA > DMA >> Amiloride, suggesting that bulky hydrophobic groups in the 5’ position of the amiloride pyrazine ring play essential roles in binding to E protein and in antiviral activity. An N15A mutation increased the production of virus-like particles, induced significant chemical shift changes from residues in the inhibitor binding site, and abolished HMA binding, suggesting that Asn15 plays a key role in maintaining the protein conformation near the binding site. These studies provide the foundation for complete structure determination of E protein and for structure-based drug discovery targeting this protein. Author Summary The novel coronavirus SARS-CoV-2, the causative agent of the world-wide pandemic of COVID-19, has become one of the greatest threats to human health. While rapid progress has been made in the development of vaccines, drug discovery has lagged, partly due to the lack of atomic-resolution structures of the free and drug-bound forms of the viral proteins. The SARS-CoV-2 envelope (E) protein, with its multiple activities that contribute to viral replication, is widely regarded as a potential target for COVID-19 treatment. As structural information is essential for drug discovery, we established an efficient sample preparation system for biochemical and structural studies of intact full-length SARS-CoV-2 E protein and characterized its structure and dynamics. We also characterized the interactions of amilorides with specific E protein residues and correlated this with their antiviral activity during viral replication. The binding affinity of the amilorides to E protein correlated with their antiviral potency, suggesting that E protein is indeed the likely target of their antiviral activity. We found that residue asparagine15 plays an important role in maintaining the conformation of the amiloride binding site, providing molecular guidance for the design of inhibitors targeting E protein.

*Corresponding author E-mail address: sopella@ucsd.edu

Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) has garnered attention as the domain (residues 36-75). Notably, the wild-type N-terminal 39-residues are present in both the full-126 length and C-terminal truncated proteins. helices of membrane proteins [46] . When samples of the E protein constructs were prepared in >90% 238 D2O instead of ~ 90% H2O, no amide signals from residues 36-75 in the cytoplasmic domain were 239 observable in the spectra of EF or EC; by contrast, strong signals from residues 19-35 and 19-33 were 240 present in the spectra of EF and ET, respectively (Fig 3 and Fig 4C) , demonstrating that these residues 241 contribute to the stable core of its unusually long trans-membrane helix. Truncation at residue 39 242 enhances solvent exchange at residues 34 and 35 of ET, which are 5 and 6 residues distal to its C- 268 data also shows that residues 2-7, before the start of the N-terminal helix, and residues 61-75 following 269 the end of the C-terminal helix exhibit gradients of increasing motion towards the termini, although even 270 the terminal residues do not appear to be highly mobile and unstructured, as is sometimes the case in 271 this class of proteins [32, 50] .

The sinusoidal waves that fit best to the magnitudes and signs of the measured RDCs as a 273 function of residue number have a periodicity of 3.6 residues per turn, proving with a very high level of 274 confidence that the protein has segments of regular a-helix secondary structure [48, 49, 51]. The 275 addition or subtraction of a single residue at either end of the helical segments significantly degrades 276 the quality of the fit, providing a clear demarcation of the length of the helical segments. The different wave for the core region of the transmembrane helix (residues 19-34) is best fit by a sine wave with a 280 somewhat smaller amplitude than for the rest of the long helical region (residues 8-18 and 35-43),

suggesting that the 36-residue helix is not completely uniform throughout its length.

To assess the orientation of the C-terminal helix and possible interactions of the cytoplasmic 

(EF) (Fig 5A) . Although there is evidence that residues 2-5 are affected by drug binding, the most 310 strongly perturbed signals are associated with residues 6-18 at the N-terminal end of the long helix and 311 extending to the core portion distinguished by its resistance to H/D exchange and broadening by 312 manganese ions, as well as the reduced amplitude of its dipolar wave. Qualitatively, the data in Fig 5   313 confirm that HMA interacts with the N-terminal domain of E protein.

The EF and ET constructs were designed to include all N-terminal residues, and these data show 315 that the binding site definitely includes residues 6, 7, an 8, and likely residues 2, 3, 4 and 5, none of 316 which were present in the previously studied constructs, and extends to residue 18. Nearly all of the 317 residues that constitute the binding site belong to the highly regular helix, until it abruptly changes tilt 318 angles at residue 18, the start of the core region. The residues between Ser6 and Leu18 are perturbed 319 by binding HMA and undergo facile H/D exchange: resistance to H/D exchange starts with residue 18.

Notably, signals from four hydrophilic residues (Glu8, Thr9, Thr11, and Asn15) as well as Ile13 are 321 most perturbed by HMA binding. Smaller CSPs observed in the C-terminal region of ET were not 322 present with EF, which may be due to non-specific HMA binding to the unnatural exposed C-terminal 323 region of ET. Titration of ET with HMA results in gradual chemical shift changes (data not shown), 324 demonstrating that binding occurs in fast exchange on the timescales defined by the chemical shift 325 differences. 

To compare the binding sites and affinities, we added increasing amounts of amiloride and two 337 amiloride derivatives, dimethyl amiloride (DMA) and ethyl isopropyl amiloride (EIPA), to samples of full-338 length E protein (EF) and monitored their two-dimensional 1 H/ 15 N HSQC spectra (Fig 6) . Notably, the 339 same residues of EF were affected by all of the amiloride derivatives albeit with different magnitudes of 340 CSPs, indicating that they all utilize the same binding site but with different binding affinities. No and the magnitudes of its CSPs lie between those of amiloride and HMA (Fig 6B and 6E or aromatic substituents at the 5' pyrazine ring (EIPA and HMA) showed the strongest inhibition, with 357 sub-micromolar IC50 values, while the compounds with smaller substituents were less effective 358 inhibitors (Fig 7) . The similarity of the trends for inhibition of replication and of binding to E protein 359 suggests that this protein may very well be a target for the antiviral activity of amiloride compounds. Of Fig 7) . We observed microscopically that EIPA and HMA decreased the number of cells in 381 each infected focus in the monolayer (Fig 8C and 8D ). This effect was especially striking for HMA; most The N15A and V25F mutations in SARS-CoV-2 E protein increased their expression compared to the 414 wild-type protein in HEK293T cell lysates ( Fig 9A) . Similar to the T16A mutation in IBV E protein, the 415 N15A mutation in SARS-CoV-2 E protein increased VLP production by approximately 40% compared to 416 the wild-type E protein, while the V25F mutation decreased VLP production by 60% compared to wild-417 type E protein, similar to the effect of the A26F mutation on the IBV E protein (Fig 9B and 9C ).

The mutations T16A and A26F in the IBV E protein have been shown to affect its oligomeric state 419 [52] . However, the analogous mutations in the SARS-CoV-2 E protein do not appear to affect its oligomerization in vitro under our experimental conditions; in PFO-PAGE, both of these mutant E 421 proteins ran as oligomers with only slightly different migration patterns compared to the wild-type 422 protein ( Fig 9D) . As expected, both of the mutant proteins ran as monomers in SDS-PAGE with their 423 apparent molecular weights similar to that of the wild-type protein (data not shown). 

Effects of N15A and V25F mutations on structure and HMA binding of E protein

The N15A mutation results in significant chemical shift perturbations of resonances from residues 435 throughout the N-terminal region of E protein, especially for the signals from Ser6, Glu7, Leu12, and 436 Ser16 (Fig 10A and 10C ). In contrast, only minor perturbations were observed for signals from residues 437 adjacent to the mutation site in the V25F mutant E protein (Fig 10B and 10D ). Since no significant 438 differences were observed among the circular dichroism spectra from wild-type E protein and these two (Fig 11A and 11C ). By contrast, HMA binding was not affected by the V25F mutation, since the 447 chemical shifts of signals from residues near the HMA binding site were unchanged and their CSPs 448 were identical to those observed for wild-type E protein (Fig 11B and 11D ). Based on these results, it appears that Asn15 is essential for maintaining the conformation of E protein required for binding HMA. 

Nothing could be done without the preparation of isotopically labeled E protein samples suitable for 493 NMR spectroscopy. This formidable barrier required the design and implementation of a novel bacterial 494 expression and purification system (Fig 2) . There are three notable aspects to our approach: 1) The NaCl, pH 7.8) (Fig 2C lane 3) . The column was washed with five-bed volumes of HPC binding buffer 701 and then 10-bed volumes of HPC washing buffer (0.05% HPC, 20 mM HEPES, 500 mM NaCl, 20 mM 702 imidazole, pH 7.8) (Fig 2C lane 4) . The KSI-E protein fusion proteins were eluted with two-bed volumes 703 of HPC elution buffer (0.05% HPC, 20 mM HEPES, 500 mM NaCl, 500 mM imidazole, pH 7.8).

The fractions containing the fusion protein were pooled and dialyzed overnight against thrombin 705 cleavage buffer (20 mM HEPES, 50 mM NaCl, 1 mM EDTA, pH 7.8) in a 10 kDa MW cutoff dialysis 706 membrane (www.spectrumchemical.com). Approximately 50 mg of uniformly 15 N labeled KSI-E protein 707 fusion protein was obtained from 1L of culture (Fig 2C lane 5) . 10 units of high-purity thrombin 708 (www.mpbio.com) per mg of fusion protein were added to the dialyzed solution and incubated overnight 709 at room temperature with gentle rotation (Fig 2C lane 6) . Importantly, thrombin retains its specificity and 

GAPDH (www.genetex.com, #GTX627408). Primary antibodies were detected using horseradish masses were estimated using a commercial protein standard (www.thermofisher.com, PageRulePlus).

Chemiluminescence was detected using a Bio-Rad Chemi Doc imaging system and analyzed using 811 Bio-Rad Image Lab v5.1 software. Densitometry was performed using the Image Lab software 812 (www.bio-rad.com) and statistical significance was determined with Welch's t-test.

Backbone NMR resonances of full-length E protein was deposited in the BRMB (accession number: 

respiratory syndrome coronavirus that lacks the E gene is attenuated in vitro and in vivo

Exceptional flexibility in the sequence requirements for 886 coronavirus small envelope protein function

Absence of E protein arrests 889 transmissible gastroenteritis coronavirus maturation in the secretory pathway

Relating structure and function of viral membrane-spanning miniproteins

Novel method for evaluation of the 1079 oligomeric structure of membrane proteins

Phage-induced alignment of 1085 membrane proteins enables the measurement and structural analysis of residual dipolar couplings with 1086 dipolar waves and lambda-maps

NMRPipe: a 1089 multidimensional spectral processing system based on UNIX pipes

NMR View: A computer program for the visualization and 1092 analysis of NMR data

The cytoplasmic tail of the severe acute respiratory 1095 syndrome coronavirus spike protein contains a novel endoplasmic reticulum retrieval signal that binds 1096 COPI and promotes interaction with membrane protein

Identification of a Golgi complex-targeting signal in the 1099 cytoplasmic tail of the severe acute respiratory syndrome coronavirus envelope protein

Three-dimensional structure of 1103 the transmembrane domain of Vpu from HIV-1 in aligned phospholipid bicelles

Opella 1 *

Division of Infectious Diseases and Global Public Health, Department of Medicine

USA 1163 1164 Short title: Interactions of SARS-CoV-2 E with amilorides correlate with antiviral activity

DNA sequence of intact full-length E protein with codons optimized for expression in E. coli. The 1182

BamHI and SacI sites were inserted for GST-S6 Fig. Expanded region of 1 H/ 15 N IPAP spectra of full-length E protein (EF) (residues 1-75) in HPC 1219 micelles. (A) Isotropic sample. (B) Weakly aligned sample in the presence of Y21M fd bacteriophage at 20 1220 mg/mL. Residue numbers and 1 JNH couplings are

The spectrum of the ET sample was obtained at 35 o C on a Bruker 900 MHz spectrometer using a home-built