Peptides derived from the SARS-CoV-2 receptor binding motif bind to ACE2 but do not block ACE2-mediated host cell entry or pro-inflammatory cytokine induction

SARS-CoV-2 viral attachment and entry into host cells is mediated by a direct interaction between viral spike glycoproteins and membrane bound angiotensin-converting enzyme 2 (ACE2). The receptor binding motif (RBM), located within the S1 subunit of the spike protein, incorporates the majority of known ACE2 contact residues responsible for high affinity binding and associated virulence. Observation of existing crystal structures of the SARS-CoV-2 receptor binding domain (SRBD)–ACE2 interface, combined with peptide array screening, allowed us to define a series of linear native RBM-derived peptides that were selected as potential antiviral decoy sequences with the aim of directly binding ACE2 and attenuating viral cell entry. RBM1 (16mer): S443KVGGNYNYLYRLFRK458, RBM2A (25mer): E484GFNCYFPLQSYGFQPTNGVGYQPY508, RBM2B (20mer): F456NCYFPLQSYGFQPTNGVGY505 and RBM2A-Sc (25mer): NYGLQGSPFGYQETPYPFCNFVQYG. Data from fluorescence polarisation experiments suggested direct binding between RBM peptides and ACE2, with binding affinities ranging from the high nM to low μM range (Kd = 0.207–1.206 μM). However, the RBM peptides demonstrated only modest effects in preventing SRBD internalisation and showed no antiviral activity in a spike protein trimer neutralisation assay. The RBM peptides also failed to suppress S1-protein mediated inflammation in an endogenously expressing ACE2 human cell line. We conclude that linear native RBM-derived peptides are unable to outcompete viral spike protein for binding to ACE2 and therefore represent a suboptimal approach to inhibiting SARS-CoV-2 viral cell entry. These findings reinforce the notion that larger biologics (such as soluble ACE2, ‘miniproteins’, nanobodies and antibodies) are likely better suited as SARS-CoV-2 cell-entry inhibitors than short-sequence linear peptides.


Introduction
The ongoing COVID19 (SARS-CoV-2) pandemic has resulted in > 235 million confirmed cases and > 4.8 million deaths globally since its outbreak in 2019 [1]. SARS-CoV-2 is a positive sense single stranded RNA virus classified within the betacoronavirus genus, primarily entering the host via epithelial cells within the upper respiratory tract [2]. Host cell invasion of SARS-CoV-2, similar to SARS-CoV-1, relies upon the direct interaction between its spike glycoprotein (S) and the host cell membrane bound angiotensin converting enzyme 2 (ACE2) [3,4]. Key contact residues within the receptor binding domain (RBD) of SARS-CoV-2 (located within the S1 subunit of the S) play a critical role in mediating high affinity, low nM binding to ACE2. This association drives high viral infection rates and consequential pathogenic severity [5][6][7][8]. Neutralising antibodies raised against the RBD, soluble recombinant ACE2 (sACE2), and other related proteins have demonstrated a clear ability to block the RBD-ACE2 proteinprotein interaction (PPI) and attenuate viral entry of both SARS-CoV-1 and SARS-CoV-2 strains [9]. These findings highlight the potential of disrupting the RBD-ACE2 protein-protein interaction (PPI) as a promising antiviral therapeutic approach.
Therapies capable of selectively disrupting PPIs represent a highly efficacious approach to treating certain diseases, with several demonstrating clear therapeutic benefit clinically [10]. However, effectively disrupting a PPI is highly challenging. The featureless interface of a PPI typically spans a large surface area (1,500-3,000 Å 2 ), consisting of multiple non-continuous binding sites (also referred to as 'hot spots') and no obvious deep binding pocket(s) [11]. This often renders traditional drug discovery approaches, such as competitive small molecules, a suboptimal approach to disrupting a given PPI due to their inability to exploit large surface areas (< 1,000 Å 2 ) [12]. Accordingly, development of PPI inhibitors has shifted focus to larger biologic molecules such as peptides (known as decoy, disruptor or interference peptides) [13]. Decoy peptides are typically designed from native structures containing a given PPI hot spot (s), meaning they possess high selectivity and affinity for their target(s), whilst constituting a low immunogenic risk [14,15].
Research aimed at developing viral entry peptide inhibitors by competitively targeting the SARS-CoV-2 -ACE2 molecular interaction, has largely focussed on the antiviral potential of ACE2 sequence derived peptides/peptidomimetics [9]. These studies utilised existing co-crystal structures of ACE2 in complex with SARS-CoV-2 S RBD , to design peptides that contain known RBD contact residues found within the α1 helix of the ACE2 binding interface [16][17][18]. Resulting antiviral efficacy of these ACE2 peptides varies considerably, depending largely on their size, stability, and secondary structure(s). Conversely, few studies have investigated peptide mimics derived from the SARS-CoV-2 S RBD sequence as potential antiviral PPI decoys. Notably, co-crystal structures of the SARS-CoV-2 RBD-ACE2 PPI highlight that all known contact residues within ACE2 are incorporated within the receptor binding motif (RBM, N437 -Y508) of the S1 subunit [7]. Considering the dense location of these known ACE2-interacting hot spots clusters (Fig 1), the RBM represents a significant region for the rational development of short sequence RBD-ACE2 decoy peptides. Previous work utilising a SARS-CoV-1 spike RBM derived hexapeptide (Y438-K-Y-R-Y-L443) demonstrated a modest ability to directly bind ACE2 (K d = 46 μM) and inhibit coronavirus NL63 cell entry and subsequent viral replication [19]. Previous work utilising a SARS-CoV-1 spike RBM derived hexapeptide (Y438-K-Y-R-Y-L443) demonstrated a modest ability to directly bind ACE2 (K d = 46 μM) and inhibit coronavirus NL63 cell entry and subsequent viral replication [19]. Although this peptide was not tested against SARS-CoV-1 viral entry, these findings highlight the possibility for a SARS-CoV-2 RBM-derived peptide to represent a potentially promising approach to the development of novel antiviral decoy peptides. However, as the spike protein corresponding to SARS-CoV-2 has been shown to bind ACE2 10 times more tightly than SARS-CoV-1, the binding affinity of SARS-CoV-2 derived peptide(s) will need to be significantly higher in order to be competitive against the spike protein [20].
In a unique approach, we have combined peptide array screening and existing co-crystal structures of the SARS-CoV-2 RBD-ACE2 PPI to select a novel set of peptides with potential anti-viral activity. We report on studies that evaluate the peptides' ability to (i) bind ACE2 directly, (ii) disrupt spike protein mediated internalisation, and (iii) inhibit spike protein induced pro-inflammatory cytokine production in human coronary microvascular epithelial cells (CMEC).

Solid phase peptide synthesis (SPPS) and peptide modifications
Peptides were synthesised on a 0.1 mmol scale using either a Biotage Initiator + Alstra (Biotage) or CEM Liberty Blue microwave assisted peptide synthesiser, with a TentaGel1 S RAM resin (0.24 mmol/g). Coupling of Fmoc-protected amino acids (5 eq, 0.2 M in DMF) and unnatural/orthogonally protected amino acids (2.5 eq, 0.1 M in DMF) was achieved by treatment with DIC (5 eq, 0.5 M in DMF) and Oxyma Pure (5 eq, 0.5 M in DMF) at 90˚C for 2 minutes. β-branched amino acids and those following unnatural amino acids were double coupled. Deprotection was achieved by treatment with morpholine (20 % in DMF with 5 % formic acid, 4 mL) at 90˚C for 1 minute. The resin was washed with DMF between deprotection and coupling (4 x 4 mL), and after coupling (2 x 4 mL).
Peptides requiring N-terminal acetylation were treated on-resin with acetic anhydride (3 eq), DIPEA (4.5 eq) and DMF (7 mL for 0.1 mmol of resin) for 20 minutes with agitation. The resin was then washed with DMF (3 x 5 mL) and DCM (3 x 5 mL) prior to peptide cleavage and global deprotection.
Peptides containing Fmoc-Lys (Alloc)-OH were treated on-resin to selectively remove the Alloc protecting group. Pd (PPh 3 ) 4 (0.25 eq) and phenylsilane (25 eq) were pre-mixed in DCM (2 mL for 0.1 mmol of resin) and added to the resin, leaving for 4 h with agitation. The resin was then washed with DMF (2 x 5 mL) and DCM (2 x 5 mL) prior to peptide cleavage and global deprotection.
Peptides requiring a C-terminal fluorescent label were treated on-resin with FITC (2 eq), DIPEA (8 eq) and DMF (4 mL for 0.1 mmol of resin) following alloc deprotection of an orthogonally protected Lys. The resin was then washed with DMF (2 x 5 mL) and DCM (2 x 5 mL) prior to peptide cleavage and global deprotection.

Peptide cleavage and global deprotection
Peptides were cleaved from the resin using a cocktail of TFA (95%), TIPS (2.5%) and H 2 O (2.5%) for 3 hours with agitation. Peptides containing Cysteine (Cys) were cleaved from the resin using a cocktail containing TFA (94%), Ethane dithiol (EDT 2.5%), H 2 O (2.5%) and TIPS (1%) for 3 hours with agitation. The resin was subsequently filtered and the TFA evaporated using a stream of N 2 , the peptide precipitated with cold Et 2 O and centrifuged (4500 rpm for 5 minutes). Peptides were dissolved in a mixture of H 2 O and MeCN with 0.1% TFA and lyophilized on a Christ Alpha 2-4 LO plus freeze dryer.

Peptide purification [S2-S9 Figs]
Crude peptides were purified by reverse-phase high-performance liquid chromatography (RP-HPLC) using either an Agilent Technologies 1260 Infinity RP-HPLC system or a Dionex RP-HPLC system with Dionex P680 pumps and a Dionex UVD170U UV-vis detector, each with a Phenomenex Gemini column (5 mm C18, 250 x 21.2 mm). Purified peptides were analysed on a Shimadzu RP-HPLC system with Shimadzu LC-20AT pumps, a Shimadzu SIL20A autosampler and a Shimadzu SPD-20A UV-vis detector using a Phenomenex Aeris column (5 mm C18, 100 Å, 150 x 10 mm). Peptides were eluted with linear gradients at column-dependent flow rates (1 ml/min for the Aeris, 10

Fluorescent polarisation (FP)
FP assays were performed as described previously [16]. Briefly, direct binding of purified human ACE2-Fc (Q18 -S470) recombinant protein (GenScript) with fluorescently labelled RBM-derived peptides (FITC) was measured at an excitation/emission wavelength of 485/ 535nm using a Mithras LB 940 plate reader (Berthold Technologies). RBM-FITC peptides were incubated at a final concentration of 500 nM with increasing concentrations of ACE2 protein (0.012 μM-6 μM) in PBS, and binding affinity (K D ) was calculated by nonlinear regression analysis of dose-response curves via GraphPad prism software (8.0).

Culture and treatment of A549 cells
Human ACE2 stable expressing A549 Alveolar Type II Lung Epithelial cells were cultured as previously described [16]. A549 cells were pre-treated for 30 minutes with RBM-derived peptides at 5 μM (final concentration, diluted in media), followed by 0.1 μM (final concentration) incubation with S RBD -His protein (SARS-CoV-2 spike protein receptor binding domain, Gen-Script) for a further 3 hours.

Luminescence internalisation assay
Transiently transfected HEK293T cells were washed and maintained in complete HBSS media (ThermoFisher) supplemented with 0.035% Sodium Bicarbonate (ThermoFisher) and 0.38% HEPES. HEK293T cells were then co-incubated for 2 hours with RBM-derived peptides (0.5, 5 or 50 μM, final concentrations) and 0.1 μM (final concentration) His-tagged SARS-CoV-2 Spike protein RBD (GenScript) or vehicle. Following 2-hour incubation period, HEK293T cells were further incubated for 10 minutes at 37˚C with the nLuc substrate Nanoglo (Promega) prior to measurement of total luminescence with the BMG Clariostar lab tech plate reader. The internalisation of membrane bound nLuc-ACE2, induced by S RBD protein, is indirectly detected by a reduction in luminescence signal.

Cardiac microvascular endothelial cell culture and stimulation
Human coronary microvascular endothelial (CMEC) cells were cultured in endothelial growth medium MV2 (Promocell) until~80% confluency. CMEC were serum starved 2 hours prior to any treatment(s) in DMEM containing 0.5% FBS. Following serum starving, cells were stimulated with 0.66 μg/mL (1 μg/10 6 cells) full-length recombinant SARS-CoV-2 S1 protein subunit (Cambridge Bioscience) for 5 hours. RBM peptides were pre-treated at 10 μM final concentration for 1 hour prior to S1 protein stimulation. Control cells were exposed to DMEM containing 0.5% FBS. Vehicle only control cells were pre-treated with 0.01% DMSO for 1 hour (same as RBM peptides). DMSO and RBM peptide treatments were diluted to appropriate concentrations in DMEM containing 0.5% FBS.

Real-time reverse-transcription polymerase chain reaction
Total RNA was isolated using QIAzol Lysis Reagent (Qiagen) according to manufacturer's instructions and diluted in nuclease-free dH 2 O (Ambion/Life Technologies). cDNA was generated from total RNA using the High-Capacity cDNA Reverse Transcription Kits (Applied Biosytems). Real-time polymerase chain reaction (PCR) was performed via a 7900HT Fast Real-Time PCR System (Applied Biosystems), using the SyBR Green Master Mix (Applied Biosystems) and human primers specific to (i) GAPDH, (ii) IL-1β, (iii) IL-6, (iv) MCP-1 and (v) VCAM (Eurofins Genomics). Relative gene expression was calculated by the 2-ΔΔCt cycle threshold method as previously described [27].

Statistical analysis
Unless stated otherwise, all groups were analysed via a one-way ANOVA test. In the event a one-way ANOVA test measured statistical significance, follow up Dunnett's or Tukey's multiple comparison analysis was utilised to assess significance amongst groups. Where data was represented as MEAN ± SEM from N � 3 independent experiments, results were determined significant by a p value < 0.05. All statistical analyses were carried out using GraphPad Prism 8.0 software

PLOS ONE
SARS-CoV-2 RBM-derived peptides represent a suboptimal approach to blocking ACE2-mediated viral cell entry the dark binding spots). Serial truncation of the N-and C-terminus of peptide S438 -K462 (Fig 2) suggested that (i) removal of S438 -D442 resulted in little reduction in ACE2 binding signal vs. control peptide (Fig 2A), and (ii) removal of S459 -K462 led to a noticeable increase in ACE2 binding signal (i.e., observable increase in peptide spot intensity, Fig 2B). As no known ACE2 contact residues were contained within these truncated sequences (determined through observation of known co-crystal structures), the S443 -K458 16mer truncated peptide was selected for follow up assessment in biophysical and cellular assays-now referred to as receptor binding motif peptide 1 (RBM1) (Fig 2C). Truncation analysis of peptide E484 -Y508 (Fig 3) suggested that removal of E484, G485 and Q506 -Y508 resulted in no observable reduction in ACE2 binding vs. control peptide (Fig 3A and 3B). Although none of these residues are known direct ACE2 contact residues, E484 remains a residue of interest due to E484-specific point mutations (e.g., E484K) being found to increase SARS-CoV-2 infectivity [28]. Consequently, E484 -Y508 (RBM2A, 25mer) and F486 -Y505 (RBM2B, 20mer) peptides were selected for biophysical and cellular in vitro assessment (Fig 3C and 3D).
To confirm target engagement of RBM-derived peptides ( Table 2) with ACE2, FITClabelled peptides were co-incubated with increasing concentrations of purified recombinant human ACE2-Fc (Q18 -S740, peptidase domain) protein (final concentration range: 0.012 nM-6 μM, Fig 4). All three peptides demonstrated direct binding to ACE2, with binding affinities (Kd) in the high nM to low μM range (RBM1 K d = 0.207 μM, RBM2A Kd = 1.13 μM, RBM2B Kd = 1.206 μM, N = 3, Fig 4A). No binding was observed with GST negative protein control, suggesting RBM peptides bound selectively to ACE2 (Fig 4B-4D). Although the RBM2A-Sc (RBM2A scrambled peptide negative control) appeared to bind to ACE2 in a dosedependent manner (Fig 4C) saturation of ACE2 binding was not achieved and the data not interpreted via non-linear aggression analysis (GraphPad Prism 8.0). In total, the FP data suggested that we were successful in designing short sequence linear RBM-derived peptides capable of directly binding ACE2. Previous work using the peptide array approach has reliably generated highly-selective decoy peptides against pathological PPIs in a broad context of disease-indications, including heart failure (HSP20 -PDE4D) [29,30], schizophrenia (DISC1 -FBWX7) [31], and cancer (c-Raf-PDE8A) [21,32,33]. All identified RBM peptides discovered in this study were tested as potential SARS-CoV-2 viral cell-entry inhibitors.

RBM peptides do not prevent viral internalisation
To assess the ability of RBM-derived peptides to inhibit SARS-CoV-2 spike protein-mediated internalisation, the peptides were initially tested against the spike protein receptor binding
It is worth noting that, even though (i) cells were pre-treated with RBM peptides prior to the application of S RBD (giving the peptides time to bind ACE2 without direct competition with S RBD ), and (ii) the concentration of RBM peptides was 5-to 500-fold higher than S RBD (100 nM), RBM peptides (alone or in combination) were unable to significantly abolish S RBD internalisation. These findings suggest that RBM1-2B are unable to effectively outcompete S RBD protein binding to ACE2.
Next, we attempted to evaluate the RBM peptides against the full-length spike trimer protein (Wuhan-Hu-1 strain) in an ACE2 overexpressing HEK293 pseudovirus (PsV) neutralisation assay (Fig 6). As expected, and previously demonstrated [16], soluble ACE2 protein completely blocked spike PsV cell entry at 0.5 μM. However, pre-treatment with RBM1, RBM2B or combined RBM1 and RBM2B did not inhibit PsV cell entry, even at a

PLOS ONE
SARS-CoV-2 RBM-derived peptides represent a suboptimal approach to blocking ACE2-mediated viral cell entry concentration of 300 μM. These data suggest that the spike trimer can outcompete RBMderived peptides for binding to ACE2, resulting in a lack of inhibitory activity.
Our findings, outlined above, strengthen the hypothesis that linear short sequence native RBM1-2B peptides represent a suboptimal approach to inhibiting ACE2-mediated SARS-CoV-2 host cell entry. Whether rational chemical modifications to the current structure of RBM1-2B peptides would significantly improve ACE2 binding affinity (and subsequent anti-viral activity), as seen with double stapled α1 helix ACE2-derived peptides vs. their respective linear analogues [18], remains unknown. It is worth noting that RBM peptides lack helical properties and are therefore not suitable candidates for peptide stapling. As such, other approaches (e.g., peptide cyclisation) should be considered as a potential way of improving RBM-peptides structureactivity-relationship. Furthermore, analysis of global genome sequencing datasets detailing identified mutations within the receptor binding domain have identified variants that possess higher ACE2 binding affinity (e.g., Y453F and N501Y) compared with the original Wuhan-1 strain [34,35]. Resultingly, these mutations could be leveraged in future efforts aimed at selecting peptide sequences that possess higher ACE2 binding affinity. However, with picomolar SARS-CoV-2 'miniproteins' demonstrating enhanced molecular interactions with ACE2 and potent anti-viral activity against SARS-CoV-2 [36], our data (and others like it [16][17][18]) further indicate that larger molecules with significantly higher ACE2 binding affinity represent a more apt approach to competitively blocking SARS-CoV-2 cell entry.

RBM peptides do not inhibit S1 protein induced inflammation
Finally, to determine whether our linear RBM peptides had the ability to perturb SARS-CoV-2 mediated inflammation (directly associated with the cytokine storm [37]) we tested them in a