Polyvalent Nanobody Structure Designed for Boosting SARS-CoV-2 Inhibition

Abstract

Coronavirus transmission and mutations have brought intensive challenges on pandemic control and disease treatment. Developing robust and versatile antiviral drugs for viral neutralization is highly desired. Here, we created a new polyvalent nanobody (Nb) structure that shows the effective inhibition of SARS-CoV-2 infections. Our polyvalent Nb structure, called “PNS”, is achieved by first conjugating single-stranded DNA (ssDNA) and the receptor-binding domain (RBD)-targeting Nb with retained binding ability to SARS-CoV-2 spike protein and then coalescing tne ssDNA–Nb conjugates around a gold nanoparticie (AuNP) via DNA hybridization with a desired Nb density that offers spatial pattern-matching with that of the Nb binding sites on the trimeric spike. The surface plasmon resonance (SPR) assays show that the PNS binds the SARS-CoV-2 trimeric spike proteins with a ~1000-fold improvement in affinity than that of monomeric Nbs. Furthermore, our viral entry inhibition assays using the PNS against SARS-CoV-2 WA/2020 and two recent variants of interest (BQ1.1 and XBB) show an over 400-fold enhancement in viral inhibition compared to free Nbs. Our PNS strategy built on a new DNA–protein conjugation chemistry provides a facile approach to developing robust virus inhibitors by using a corresponding virus-targeting Nb with a desired Nb density.

Graphical Abstract

INTRODUCTION

Coronaviruses belong to a virus genus that envelopes with positive-sense single-stranded RNA, which can infect host cells for viral genome replication. Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) has caused huge medical and financial burdens on people around the world due to its transmissible, infectious, and lethal respiratory illness features.^1,2 During the viral invasion process, SARS-CoV-2 binds to a host cell through the interaction between human angiotensin-converting enzyme 2 (hACE2) with spike receptor-binding domain (RBD) on viral membranes.^3,4 Blocking the interaction between RBD and ACE2 works as a general strategy offered by therapeutic neutralizing antibodies (nAbs).⁵ However, producing nAbs for therapeutics is very costly and time-consuming and they are not stable in ambient conditions.^6,7 Importantly, it may induce unwanted antibody-dependent enhancement (ADE) of infection,^8,9 as was the case for dengue virus vaccine, in which nAbs can cause increased viral infectivity in vivo. Therefore, developing more robust and versatile virus-neutralizing agents is highly desired in combating existing and emerging threats by viruses.

The nanobody (Nb) is known as a variable domain from an antibody’s heavy chain only. Nbs received increased attention as SARS-CoV-2 neutralization reagents^10–13 because of their excellent stability and cost-effective production. Despite their advantages, individual Nbs are hundreds of folds less effective than therapeutic nAbs.¹⁴ To fulfill the purpose of building much stronger neutralizing Nbs, a multivalency strategy that fuses two or three Nbs with an amino acid linker was developed through the traditional protein biosynthesis pathway.^14–17 In these studies, only dimeric or trimeric Nb structures^14–17 were constructed as limited by the high complexity of designing protein sequence and modulating protein for higher structure orders. To overcome the challenges to engineering polyvalent-Nb structures (PNS) that could consist of many individual Nbs working together with synergy, we have developed a new strategy for constructing PNS by decorating Nbs on gold nanoparticles (AuNPs) with a desired Nb density via DNA hybridization. DNA is known as the genetic information carrier and has also exhibited distinct powers in the field of bionanotechnology. In combination of programmability and unique Watson–Crick base pairing, DNA molecules have been utilized in programming nanomaterial synthesis,^18,19 assembling desired nano-objects,^20–22 and building ultrasensitive biosensors.^23–25 Synchronized with phosphoramidite chemistry development and advancement, DNA oligonucleotides can be synthesized robustly and cost effectively with many desired chemical attachments at any positions, such as the amino, biotin, or dibenzocyclooctyl group that can be used to cross-link with other molecular moieties including proteins.^26–28 Additionally, when a number of complementary ssDNAs are attached to a AuNP surface via thiol–gold linkage,^29,30 PNS would be possibly constructed through DNA hybridization as templated on a AuNP with desired Nb densities. Importantly, after being attached to AuNP surfaces, DNA molecules show greatly improved thermodynamic stability compared to free DNA.³¹

In this study, we have developed click chemistry to conjugate an ssDNA to an RBD-targeting Nb at a selective site with a quantitative yield. Forming such a conjugation does not compromise the Nb’s binding capacity. We further constructed a PNS templated on AuNPs with a desired Nb density. The SPR assays show a ~1000-fold improvement in PNS-spike binding strength compared to the interaction between an individual Nb with spike, which is attributed to the polyvalent, pattern-matching interaction offered by the PNS. Moreover, our viral entry inhibition assays demonstrate an over 400-fold enhancement of the PNS in inhibiting virus entry compared to individual Nbs, proving the benefit of deploying PNS for blocking virus infections.

RESULTS

DNA–Nanobody Conjugation.

To connect an ssDNA to the spike RBD-targeting Nb (ref 32) with a high conjugation yield as well as keeping the Nb’s binding ability, we developed a robust, site-selective modification strategy, as illustrated in Figure 1 and Figure S1 that involves dibenzaocyclooctyl (DBCO) cooper-free click reaction,³³ as inspired by the success of azide modification on Gly-His tagged protein.³⁴ As illustrated in Figure 1a, our conjugation reaction consists of two steps. The Gly-His tag on the N-terminal of a Nb is first acylated by 4-methoxyphenyl 2-azidoacetate,³⁴ resulting in an azide group attachment to the N terminal. Then, DBCO-modified ssDNA (synthesized by IDTDNA Inc.) is mixed with the azide–Nb to form ssDNA–Nb conjugates through a click reaction. The successful production of the DNA–Nb conjugate was characterized by both native polyacrylamide gel electrophoresis (PAGE) and sodium dodecyl-sulfate polyacrylamide gel electrophoresis (SDS-PAGE) (Figure 1b and Figure S2), which shows that the DNA–Nb conjugate with a larger molecular weight (MW) migrates more slowly than the components ssDNA and Nb. When increasing the amount of Nb used in the conjugation reaction to a 1:1 molar ratio with the ssDNA, 93.6% of the ssDNA has been linked to Nb (Figure 1b middle lane). Furthermore, we used MALDI-Mass analysis to confirm the two-step conjugation reactions. Shown in Figure 1c, after an acylation reaction, the Nb’s MW increases from 14,524 to 14,610, indicating the addition of the azide tag to the Nb. Upon the conjugation of an ssDNA (MW: 6671), the Nb’s MW further increases to 21,268, which is equivalent to the MW of one Nb plus one ssDNA molecule. The PAGE and MALDI-Mass characterizations prove that our DNA–protein conjugation reaction can produce ssDNA-Nb conjugates at a selective Nb site with a quantitative yield. In the reported studies on protein and DNA conjugation, nickel ion domain conjugation³⁵ suffers from its stability in biological conditions, while enzyme-controlled immobilization³⁶ is constrained by its low conjugation efficiency. The SMCC strategy utilizes the N-hydroxysuccinimide (NHS) ester on SMCC to react with primary amines on both N-terminal and lysine residues of a protein in a more random fashion,³⁷ which may compromise the protein function. The DNA–Nb conjugation method developed by us successfully achieved a quantitative reaction yield at a specific site without disturbing the spike RBD-binding capability of the Nb.

Figure 1.

Site-selective nanobody (Nb) conjugation with single-stranded DNA (ssDNA). (a) Schematic of the chemical reactions for conjugating a Nb with DBCO-modified ssDNA through click chemistry. (b) Native polyacrylamide gel electrophoresis (PAGE) analysis of the Nb–DNA conjugate. Left lane: ssDNA only; middle lane: DNA–Nb conjugate product formed with a 1:1 molar ratio; right lane: DNA–Nb conjugate product formed with a 1:0.4 molar ratio. The PAGE was stained with gel red. (c) MALDI MS analysis of the products after each chemical reaction.

Construction of the PNS.

To create a PNS built on AuNPs, we used the thiol–gold reaction to functionalize a 13 nm AuNP surface with ssDNA probes that in turn can hybridize with the ssDNA attached to the Nb with a complementary sequence (Figure 2a). To ensure the formation of a dsDNA between Nb-attached ssDNA and AuNP-tethered complementary DNA at 37 °C (a normal body temperature), a 15 bp DNA sequence was used with a melting temperature (T_m) of 59.6 °C as calculated by using the IDTDNA OligoAnalyzer tool. A T5 spacer sequence was also added to both ssDNAs to eliminate the steric hindrance effect that could possibly impair the formation of PNS. Upon DNA hybridization, Nbs will be anchored on a AuNP surface to form a PNS. The electrophoretic light scattering (ELS) was used to characterize the AuNP surface properties for monitoring the whole process of forming PNS. Shown in Figure 2b, the zeta potential of AuNPs decreases from −1.88 ± 0.93 to −16.13 ± 0.88 mV after surface modification by ssDNA.

Figure 2.

Synthesis and characterization of DNA engineered PNS. (a) Schematic of PNS formation engineered with DNA hybridization on an AuNP surface. (b) Change of AuNP surface charge in zeta potential is characterized using electrophoretic light scattering (ELS). (c) A transmission electron microscopy (TEM) image and (d) a UV–vis absorbance spectrum of the PNS in PBS buffer. (e) Front view of the spike trimer (light purple, surface) in complex with Nbs (blue, cartoon) emphasizing the positioning of the Nbs deep into the grooves between adjacent RBD domains. (f) Top view of the spike trimer (light purple, surface) in complex with Nbs (blue, cartoon) reveals the distance between adjacent interfaces, which perfectly aligns with the distance between the Nbs on the gold nanoparticles.

Since the DNA molecule is negatively charged, the decreased zeta potential demonstrates that the probe DNA was successfully attached on the AuNP surface as a standard protocol.^38,39 The zeta potential of AuNPs further decreases to −25.85 ± 0.89 mV after Nbs were attached to the AuNP surface via DNA hybridization considering that the Nb is also negatively charged with an isoelectric point of p_I = 6.2. To calculate the number of Nbs attached on each AuNP, we measured UV–vis absorbance or the fluorescence intensity of the whole and unbound Nbs in the solution before and after the immobilization (Figures S3 and S4 and Table S3), which shows an average of 32 Nbs attached to a AuNP. We then utilized transmission electron microscopy (TEM) imagining to visualize the PNS, which shows that the Nbs are displayed as dotted corona (Figure 2c and Figure S5) around a gold core (Figure S6). Additionally, we performed a UV–vis absorbance scan from 400 to 800 nm that yields a sharp surface plasmon resonance peak at 524 nm, proving a good monodispersity of the PNS in the buffer (Figure 2d). Based on the surface area (~1962 nm²) of a DNA-coated 13 nm AuNP, the Nb density on the AuNP is estimated to be 16,310/μm² based on our experimental data. Structural analysis of the spike–Nb complex reveals that the Nbs are positioned between adjacent RBD domains with an average distance of ~7.1 nm between the interfaces (Figure 2e,f), which is possibly the approximate average separation between two Nbs on the 13 nm AuNP. The N-terminals of the Nb are buried deep into the spike protein, suggesting that the binding of the Nb can effectively block any interaction with ACE2. In summary, the Nb density obtained on the AuNP offers possible spatial pattern-matching to the predominant Nb binding sites on a trimeric spike, paving the application of the PNS to virus inhibition.

Binding between PNS and SARS-CoV-2 Spikes.

Next, we performed a surface plasmon resonance (SPR) study to confirm whether the PNS can improve the spike binding affinity compared to solitary Nbs (Figure 3a). SPR sensor-grams show that the binding strength of the Nb to the spikes increases by ~1 × 10³-fold when the same Nbs are patterned on the AuNP to form PNS (Figure 3b,c). We also observed that ssDNA–Nb exhibits a similar binding strength to that of Nb–spike interaction, further confirming that attaching ssDNA to the Nb via a N-terminal site-selective linkage does not compromise the binding between Nb and the spike (Figure S7). Additionally, the probe DNA-modified AuNP showed minimal SPR response (Figure S8), proving that the improvement in Nb–spike binding affinity results specifically from the polyvalent, pattern-matching interaction offered by our PNS, not the presence of ssDNA-coated AuNP. Our SPR data also quantify the association rate constant (K_a) of the PNS–spike reaction as 4.14 × 10⁸ M⁻¹ S^1– and the dissociation rate constant (K_d) as 3.18 × 10⁻³ S^1–.

Figure 3.

Surface plasmon resonance (SPR) assay. (a) Scheme of the SPR assay using free Nbs or PNS. Trimeric spike proteins are immobilized on an SPR chip. SPR sensorgrams of (b) free Nb–spike or (c) PNS–spike binding kinetics. The Nb or PNS samples of different concentrations were injected for 120 s followed by dissociation for 240 s.

Compared to the free Nbs, the K_a of PNS has increased 465-fold. We speculate that these increases are attributed to the cooperative interactions between multiple Nbs patterned on a AuNP that are spatially matched with the trimeric spike proteins. Additionally, our SPR assay shows that the PNS does not interact with an HIV-1 GP120 protein that has a similar trimeric structure as that of SARS-CoV-2 spike protein (Figure S9), proving the specificity of our PNS against the spikes of SARS-CoV-2 and its variants. The resulting improvement in the binding avidity, in turn, led to much higher antiviral potency using our PNS, as validated in our subsequent assays.

PNS in SARS-CoV-2 Inhibition.

The elevated spike binding affinity by the PNS shall potentially enhance the Nb’s inhibition capability against SARS-CoV-2 spike-mediated infection. To determine whether our PNS could also block viral infection, we performed a pseudoviral entry inhibition assay. To determine the ability of PNS to inhibit SARS-CoV-2 viral entry, we used an HIV pseudotype system barring the SARS-CoV-2 spike protein. Using BHK21-hACE2 cells, viral entry was determined by the bioluminescence produced by luciferases encoded in the pseudoviral genome (Figure 4a). Our cryo-EM imaging showed that PNS can bind to the intact virus (Figure 4b) to form a PNS–virus complex on the viral surface (Figure 4c and Figure S10). We speculate that spike proteins on the outer viral surface may get lost during virus purification as well as cryo-EM sample preparation as observed before.⁴⁰ Thus, it is expected that not many PNS appear around each virus particle under TEM imaging. Additionally, we performed dynamic light scattering (DLS) analysis to confirm the positive interaction of PNS with SARS-CoV-2 virus in a free solution, in which although still not ideal, the virus can maintain the integrity of its surface spike proteins better than the cryo-EM condition. As shown in Figure S11, mixing excessive PNS with SARS-CoV-2 virus has indeed resulted in forming complexes of bigger size than PNS and virus particle. At the same time, the peak indicating free virus particles disappears after mixed with PNS, proving that all the virus particles are complexed with the PNS. Interestingly, we found that the 400 nM free Nb can only impede 14% WA/2020 SARS-CoV-2 viral entry but the PNS has an IC₅₀ of 0.92 ± 0.06 nM, representing an over 400-fold enhancement in antiviral inhibition (Figure 4d,e and Figure S12). Our assays also confirm that AuNPs alone do not exhibit much of an antiviral effect (Figure 4f). We synthesized control PNS constructs to explore the influence of nanoparticle size and Nb density on PNS antiviral efficacy. More specifically, we used 13 nm AuNP to make another PNS construct (called PNS-L) with a more sparse Nb coverage (Nb density = 10,703 Nb/μm², see Table S4 for detailed calculations). We also used 5 nm (called PNS-5) and 20 nm (called PNS-20) AuNPs (see TEM images in Figure S13) to make two PNS constructs of different AuNP sizes with slightly denser Nb coverages (Nb density = 20,937 Nb/μm² for PNS-5 and 27,679 Nb/μm² for PNS-20, see Table S4 for detailed calculations).

Figure 4.

Neutralization of SARS-CoV-2 viral entry by the PNS. (a) Schematic of blocking the invading of pseudotyped SARS-CoV-2 virus by PNS. The degree of neutralization was determined using luminescence released by luciferase encoded in the pseudoviral genome. (b) Cryo-EM image of SARS-CoV-2. (c) Cryo-EM image of the virus–PNS complex. SARS-CoV-2 inhibition efficacy of (d) DNA–Nb conjugates, (e) PNS, and (f) DNA-coated AuNPs.

Our data in Figure S14 shows that the viral neutralization efficacy at 1 nM PNS-L dropped by ~36% when PNS-L that carries a smaller Nb density was used compared to the PNS carrying an optimal Nb density offering a possible matching pattern with spike protein. Our data fUrther prove the importance of the pattern/spacing matching between Nbs on the AuNP and RBDs on the virus for achieving a higher antiviral efficacy. Our assay also shows that in comparison with PNS’s antiviral performance at the respective concentrations, 2 nM PNS-5 built on 5 nm AuNP core has a 77% reduction in antiviral efficacy while 0.5 nM PNS-20 with the 20 nm AuNP core showed twofold antiviral efficacy improvement (Figure S15). We speculate that such a difference in antiviral efficacy is attributed to the fact that PNS-20 is physically bigger so it can offer a better physical separation between host cells and PNS-bound viruses. However, we also need to point out that larger AuNPs are not as stable as smaller ones and cost more to manufacture. Thus, using 13 nm AuNP-based PNS is considered collectively more beneficial to antiviral applications. Compared to the cell medium, PNS at its IC₅₀ concentration exhibits very minimal cytotoxicity (Figure S16), which attributes to the exposed spherical nucleic acid (SNA) area, as recently reported.⁴¹ To confirm whether our PNS can inhibit the viral entry of different SARS-CoV-2 variants, we performed additional pseudovirus entry assays using the BQ1.1 and XBB variants of SARS-CoV-2. Our data show that the PNS also effectively neutralizes BQ1.1 or XBB with similar IC₅₀ of 0.66 ± 0.10 and 0.74 ± 0.18 nM for the two variants, respectively. This proves the broad neutralization capability of the PNS (Figure 5 and Figure S12).

Figure 5.

Neutralization of SARS-CoV-2 BQ 1.1. and XBB variants by the PNS. Efficacy of (a) SARS-CoV-2 BQ 1.1 or (b) XBB by the PNS shows an IC₅₀ of 0.66 ± 0.10 or 0.74 ± 0.18 nM.

DISCUSSION

The challenges for COVID-19 therapeutics have historically been evidenced by repeated tragedies during previous epidemics and pandemics. Therapeutic nAbs have been used as an alternate type of virus inhibitor, but their production is very costly and time-consuming. They are not stable in ambient conditions and thus require low-temperature conditions for storage and transportation, which have greatly increased the cost and difficulty in reagent distribution, particularly to low-resource places. Antiviral reagents made of DNA aptamers are easy to synthesize and can be readily attached to a DNA nanostructure (via DNA hybridization) or AuNP (via thiol–gold chemistry) to enable multivalent interactions with virus particles. Recent in vitro studies have suggested a great potential of using aptamers to combat the SARS-CoV-2 infections.^{24,40,42–44} However, aptamers’ binding affinity to their protein targets is normally lower than that of many Abs or Nbs. Moreover, a nucleic acid-based viral inhibitor needs additional chemical modifications and/or polymer coating for achieving similar biostability levels, as Abs or Nbs have, to enable a possible use in clinics. In this study, we have developed a new covalent coupling reaction for DNA–Nb conjugation, in which a single azide group-modified Nb was site-selectively connected with an ssDNA through click reaction. Such DNA–Nb conjugation with a quantitative yield lays a foundation for successful creation of a AuNP-templated polyvalent–Nb structure (PNS) with a desired Nb density. Thus, PNS offers multivalent, pattern-matching interactions between AuNP surface-patterned Nbs and viral surface spike proteins, resulting in >1000-fold improvement on Nb-spike binding affinity, as shown by our SPR assay. The viral entry inhibition assays demonstrate a >400-fold antiviral efficacy enhancement of the same monomeric Nbs using our PNS, while both free Nb and AuNPs show very weak neutralization activities. Furthermore, we also demonstrated a great performance of the PNS in inhibiting the invasion of recent SARS-CoV-2 variants of interest (BQ1.1 and XBB), which eliminates the neutralization escape concerns suffered by therapeutic nAbs. Since the Nbs used here were initially developed against the spike of WA/2020 SARS-CoV-2, its binding affinity against the spikes of later SARS-CoV-2 variants could be reduced, potentially leading to a neutralization escape. However, PNS enables high virus binding affinity of the same Nb through multivalent, pattern-matching interactions that compensate for the reduced binding strength by individual Nbs. Additionally, the AuNP scaffold can serve as a physically bulky barrier to enhance the blockage of viral–host cell interactions. It is also known that many viruses first interact with negatively charged glycosaminoglycans (GAGs) on the host cell surface before invasion.^45–49 Thus, the PNS not only relies on specific multivalent interactions for binding SARS-CoV-2 but also electrostatically traps and isolates virions from the host cell plasma membrane and GAGs through the negative charges of the AuNP scaffold.

In summary, we have developed a AuNP-templated polyvalent–Nb structure that offers a versatile platform for effective inhibition of SARS-CoV-2 infections. Our experiments clearly demonstrate the advantage of using multivalent and pattern-matching interactions for increasing binding avidity between the nanostructured Nbs and the SARS-CoV-2 spikes. Our polyvalent–Nb structure design can be adapted to combat other viruses by using virus-specific Nb and matched viral surface antigen spacing. We foresee that our PNS can be formulated as a nasal spray and applied as a preventive reagent to bind aerosolized respiratory viruses, such as SARS-CoV-2 and influenza. Additionally, our PNS can be systematically administered as a therapeutic reagent to capture and neutralize target viruses in circulation. Targeting moieties can be attached to the AuNP for directing the captured viruses to kidney for clearance or to immune cells for stimulating Abs’ production.