A comparison of Remdesivir versus gold cluster in COVID-19 animal model: A better therapeutic outcome of gold cluster.

While gold compound have been approved for Rheumatoid arthritis treatment as it well suppresses inflammatory cytokines of patients, no such treatment is currently available for COVID-19 treatment in vivo . We firstly disclose gold cluster yields better therapeutic outcome than Remdesivir in COVID-19 hamster treatments as it is armed with direct inhibition viral replication and intrinsic suppression inflammatory cytokines expression. Crystal data reveals that Au (I), released from gold cluster (GA), covalently binds thiolate of Cys145 of SARS-CoV-2 Mpro. GA directly decreases SARS-CoV-2 viral replication and intrinsically down-regulates NFκB pathway therefore significantly inhibiting expression of inflammatory cytokines in cells. The inflammatory cytokines in GA-treated COVID-19 transgenic mice are found to be significantly lower than that of control mice. When COVID-19 golden hamsters are treated by GA, the lung inflammatory cytokines levels are significantly lower than that of Remdesivir. The pathological results show that GA treatment significantly reduce lung inflammatory injuries when compared to that of Remdesivir-treated COVID-19 hamsters.

Introduction

To date, over 5 Million people have succumbed to COVID-19 infections worldwide, with little sign of this global pandemic being swiftly brought under control. Despite tremendous global efforts in identifying a suitable drug against COVID-19, no drug has been proven to effectively treat COVID-19 infections. In comparison of biological agents, chemical drugs are of unique advantages in dealing with the COVID-19 pandemic: they are easily produced in large scale with low cost, thus satisfying the huge number of COVID-19 patients in low income countries. Importantly, chemicals allow for efficient handling, storing and distributing to patients living in environments unsuitable for biological agents. Several traditional chemicals are currently repurposed for COVID-19 treatment [1], [2], [3], [4], [5], [6], [7]. For example, Remdesivir was shown to directly inhibit SARS-CoV-2 replication but failed to intrinsically suppress inflammatory cytokines expression in patients [3], [7]. In comparison, Ruxolitinib or Acalabrutinib intrinsically suppressed inflammatory cytokines expression, but failed to directly inhibit virus replication [4], [6].

We firstly propose that single compound combination of directly inhibition viral replication and intrinsically suppression inflammatory cytokines expression should yield better therapeutic results in COVID-19 treatment. The SARS-CoV-2 Mpro has been used as critical drug target for COVID-19 treatment as it plays a key role in SARS-CoV-2 replication, and organic compounds have developed to yield a Michael adduct with Cys145 of the catalytic dyad of Mpro [8], [9]. We speculated that Au (I) ion would yield a Michael adduct with Cys145 of Mpro as previous reports showed that the Au (I) ion, released from gold compound, inactivates Echinococcus Granulosus Thioredoxin Glutathione Reductase via covalently bind with Cys519 and Cys573 [10], [11]. While gold compound, like Auranofin (AF), have previously been approved for Rheumatoid arthritis (RA) treatment as it well suppresses inflammatory cytokines of patients [12], [13], no such treatment is currently available for COVID-19 treatment in vivo. Here, we propose a novel compound, gold cluster, for COVID-19 treatment via directly inhibition of SARS-CoV-2 replication and intrinsically suppression of inflammation cytokines expression. We set out to investigate if this gold cluster possesses therapeutic properties in mitigating the effects of COVID-19 infections.

Results

Gold compound (AF or GA) firstly associates with the catalytic domain of SARS-CoV-2 Mpro and finally produce Au-Mpro adduct

The gold cluster was synthesized by a chemical method (SI 1 , Fig. S1). We firstly study if gold compound (AF or GA) associates with hydrophobic domain surround the catalytic dyad of Mpro via molecular dynamic (MD) simulations (SI 2). In this theory study, the crystal structure of Mpro was obtained from the PDB database (PDB 6LU7). The AF and GA molecular structure were performed by using GaussView 5.0 and Gaussian 09 software packages. As shown in Fig. 1A, there is one hydrophobic cavity with its volume over 500 Å3 in each Mpro monomer, which surrounds the catalytic site of Cys145 (green or purple region of Mpro monomer). In this work, we utilized AF or GA as ligand to complex with Mpro. For the AF-Mpro or GA-Mpro complexing system, we studied the root-mean-square deviation (RMSD) to learn the binding stability. Obviously, the AF-Mpro or GA-Mpro reached binding equilibrium at about 10 ns, and the binding conformation remained stable afterwards (Fig. S2). The average binding energy of AF-Mpro and GA-Mpro was ~50.47 kcal/mol. and ~722.94 kcal/mol., respectively (Fig. S3). The GA (yellow) and AF (orange) bind to the hydrophobic cavity around the Cys145, and this makes Au atom very close to S of Cys145 (Fig. 1B). In AF-Mpro system, Van del Vaals force plays main role to keep system stable where the Glu166 and Gln189 of Mpro interact with AF via hydrogen bonds. While the electrostatic force contributes the main interaction in GA-Mpro system in Fig. 1B, Arg4, Arg40, Lys137 and Arg188 of Mpro formed four salt bridges with glutathione of GA (with yellow S atoms).

Fig. 1

Theory simulations and SPR studies of AF and GA complex with Mpro, ESI-Mass spectra of final Au-Mpro adduct. (A) The molecular structure of GA, AF, and SARS-CoV-2 Mpro. (B) Molecular dynamic simulations of GA and AF complex with Mpro. Both gold compounds stably bind the hydrophobic domain surround the catalytic dyad of Mpro, GA-Mpro (binding energy ~722.94 kcal/mol.) is more stable than AF-Mpro (binding energy ~50.47 kcal/mol.). (C) SPR studies of GA and AF association with Mpro, KD of GA is ~ 14.8 nM and that of AF is ~ 49.1 µM. (D) ESI-Mass spectra of Au-Mpro adduct which is extracted from GA treated HEK293F cells (left), ESI-Mass spectra of Au-Mpro adduct which is purified from AF incubated Mpro solution (right). Note that molecule weight of strep-Mpro is 36,116 Da and Apo Mpro molecular weight is 33,849 Da. Mass data revealed Au ion bind Mpro monomer.

The experimental association affinity between gold compound (AF or GA) and Mpro was further studied by surface plasma resonance (SPR) method (SI 3). After Mpro was fixed in chip, serial dose of gold compound (AF or GA) was introduced into solution and the time dependent optical signal were tracked. As shown in Fig. 1 C, the KD of GA-Mpro and AF-Mpro is ~14.8 nM and ~49.1 µM, respectively. This SPR data implied that GA is with stronger affinity with Mpro when compared with that of AF, and this experimental result matches the aforementioned molecular dynamic simulations where the average binding energy of AF-Mpro was ~50.47 kcal/mol. and that of GA-Mpro is the ~722.94 kcal/mol. Although theory and SPR studies revealed gold compound can tightly bind the Mpro, we do not know the final product form of GA-Mpro and AF-Mpro. To clarify this issue, strep tagged-Mpro (strep-Mpro) were expressed in HEK293F cell and high concentration of GA were introduced in cell culture media, this would produce final Au-Mpro adduct in cytoplasm (SI 4). AF were directly introduced to none strep tagged-Mpro (Mpro in Apo form) solution to get final Au-Mpro adducts as HEK293F cell could not bear toxicity of high concentration AF in cell culture media (SI 4). After gold compound treated Mpro was purified, ESI mass spectra was used to check the final product of GA-Mpro or AF-Mpro. In Fig. 1D, both GA and AF treated Mpro finally produce Au-Mpro adduct where one Au atom is added to one Mpro. These studies verified that GA or AF firstly associated with Mpro and finally produce Au-Mpro adduct.

Au (I), released from gold compound in buffer solution, covalently binds the thiolate of Cys145 of Mpro via experimental crystal structure studies

In order to examine molecular structural basis of gold compound inactive Mpro in vitro/vivo, we determined the SARS-CoV-2 Mpro experimental crystal structures in the Mpro crystal incubated with AF (AF incubating form), Mpro crystal incubated with GA (GA incubating form), and Mpro crystal only (the native form), see details in SI 5 , Fig. S4, and Table S1. In Fig. 2A, the Mpro molecular structures treated with AF or GA are highly similar, and share most features of the crystal structures of the apo SARS-CoV-2 Mpro determined recently [8], [9]. However, crystal structural analysis showed that the densities of two Au (I) ions were found clearly to be very close to the thiol residues of Cys145 and Cys156 (Fig. 2B). The position of two Au (I) ions was confirmed by applying the anomalous difference Fourier maps, two Au (I) ions are defined as Au(I) 1 and Au (I) 2, respectively. In Figs. 2C and S5, the Au-S bond length is 2.3 Å in Mpro, such short bond length confirms that AF or GA can release Au (I) ions which covalently bind to the thiolate of Cys145 and Cys156 of Mpro. The thiolate of Cys519 and Cys573 residues of Echinococcus Granulosus enzyme covalently interacts with Au (I) ions, released from gold compounds in solution, were previously reported [10] , which supports Au (I) irreversibly interact with thiolate of Cys145 of the catalytic dyad of Mpro in this study. Further temperature factor analysis shows that the occupancy of Au (I) ion is partial and the occupancy factors is about 33% for Cys145 and 11% for Cys156, indicating that the Au (I) ions released from AF or GA are gradually bind the thiolate of Cys145 and Cys156 of Mpro molecules in solution.

Fig. 2

The X-ray crystal structure of AF and GA treated Mpro in Au-S covalent bind state. (A) Surface presentation of the Mpro homodimer in Au-S bound state with Chain A and Chain B shown in green and violet, respectively. (B) The presentation of one Mpro monomer with Domain I-III shown in light blue, light pink and pale cyan. Enlarged views of the Au (I)-S bound sites. Anomalous difference Fourier maps (blue mesh, contoured at 5 sigma) are shown for Au (I) 1 and Au (I) 2. Residues His41, Cys145 and Cys156 are shown in sticks and two Au (I) ions are shown in spheres. (C) Comparison of Au (I)-Cys145 bound state with the native state of Mpro. Superposition of crystal structures of AF treated (purple), GA treated (yellow) and native Mpro (blue). The catalytic pocket of native and Au (I)-S bound Mpro in surface presentation and the surrounding residues shown in sticks. (D) DFT calculation of interaction between Au (I) ions and Cys145 and Cys156 of Mpro, respectively. Left panels show protein catalytic pockets consisting of amino acids and one Au (I) ion. Right panels represent geometrically relaxed structures for the catalytic pockets encapsulating the Au (I), all Au−N atomic distances (in Å) within 5 Å are labeled with the corresponding distances. C, N, O, S, and Au atoms are displayed in gray, blue, red, pink, and yellow, respectively.

Although the Mpro monomer contains 12 Cys residues (Cys16, Cys22, Cys38, Cys44, Cys85, Cys117, Cys128, Cys145, Cys156, Cys160, Cys265, Cys300), only Cys145 and Cys156 specifically bind to the Au (I). To further verify the covalently binding of Au (I) to the S atom of Cys145 and Cys156, we calculated the interaction energies between Au and Mpro protein using density functional theory (SI 6) method. In Fig. 2D, our analysis showed that the bond dissociation energies (E BD’s) between Au (I) and Cys145 are ~ 46.1 kcal/mol. and that of Au (I) and Cys156 are ~ 26.5 kcal/mol. The larger E BD value strongly suggests that the Au ion covalently bind to Cys145 and lock the active pocket of Mpro, thus efficiently inhibiting catalytic activity.

Gold compound (AF or GA) inhibits SARS-CoV-2 Mpro activity, suppress SARS-CoV-2 replication, and block inflammatory cytokines expression in cell assays

To exam whether AF or GA effectively inhibits Mpro activity, we determined the IC50 of AF or GA using a previously reported method [14]. Mpro activity was measured using a fluorescence resonance energy transfer (FRET) assay. To this end, a fluorescence labeled substrate, (EDNAS-Glu)-Ser-Ala-Thr-Leu-Gln-Ser-Gly-Leu-Ala-(Lys-DABCYL)-Ser, derived from the auto-cleavage sequence of the viral protease was chemically modified for enzyme activity assay (SI 7). As shown in Fig. 3A and B, the IC50 of AF was ~0.46 µM, and the IC50 of GA was ~3.3 µM (count by gold element).

Fig. 3

AF and GA inhibit Mpro activity, suppress SARS-CoV-2 replication, and inactivate the NFκB pathway and suppress inflammatory cytokines expression in cells. (A) IC50 of AF is ~ 0.46 µM. (B) IC50 of GA is ~ 3.3 µM count by gold element. (C) EC50 of AF is ~ 0.83 µM. (D) EC50 of GA is ~ 12.52 µM count by gold element. (E) Low dose of AF (0.6 µM) and GA (20 µM count by gold element) significantly inhibit IL-6, IL-1β, TNF-α inflammatory cytokines expression in RAW264.7 macrophages via western blot method (unpaired t-test, ***p < 0.001, **p < 0.01, *p < 0.05). (F) Low dose of AF (0.08 µM) and GA (10 µM count by gold element) significantly suppress NFκB activation, thus inhibiting IL-6, IL-1β, TNF-α inflammatory cytokine expression in human 16HBE bronchial epithelial cells via western blot method (unpaired t-test, ***p < 0.001, **p < 0.01, *p < 0.05).

To assess whether GA inhibits Mpro activity in mammalian cells, we transiently transfected HEK293F cells with a plasmid of strep-tagged SARS-CoV-2 Mpro gene. The Mpro gene was expressed 24 h in HEK293F cells, then GA was added to the culture medium at a final concentration of 500 µM, and cell cultured for an additional 24 h. After cell harvest, SARS-CoV-2 Mpro extracted from GA-treated HEK293F cells was purified and analyzed for enzyme activity (SI 7). As shown in Fig. S6, the GA treated-SARS-CoV-2 Mpro activity was reduced to approx. 60% of control SARS-CoV-2 Mpro. ICP-MASS results indicated that when Mpro was expressed in GA-treated mammalian cells, gold can bound to SARS-CoV-2 Mpro (Fig. S6). We failed to obtain similar results for AF in cell assays as AF exhibited strong cytotoxicity when HEK293F cell cultured with AF. We measured the EC50 of AF or GA to evaluate if they inhibit SARS-CoV-2 replication in Vero cells, using a recently described method [5] (SI 7). As shown in Fig. 3C and D, the EC50 of AF was approx. 0.83 µM and EC50 of GA approx. 12.52 µM (count by gold element). However, for Vero cell the CC50 value of AF is ~ 2.27 µM and that of GA approx. 1.11 mM (Fig. S7).

COVID-19 viral infections are characterized by infections of bronchial epithelial cells, resulting in activation of inflammatory cytokine gene expression via the NFκB pathway, and these cytokines will in turn activate macrophages, which as a result acquire an inflammatory status to profoundly produce inflammatory cytokines [15], [16]. Recent reports revealed that among versatile inflammatory cytokines, the IL-6, IL-1β, TNF-α play key roles in inflammation development of modest and severe infections [16], [17]. We firstly assessed whether GA or AF inactivate the NFκB pathway and suppress inflammatory cytokine expression levels in inflammatory human 16HBE bronchial epithelial cells (SI 7). To test whether AF or GA inactivate the NFκB pathway in macrophage cells, and as a result down-regulate expression of IL-6, IL-1β, TNF-α, RAW264.7 macrophage was incubated with AF or GA at different concentrations for 24 h. As shown in Fig. 3E and F, the low dose of AF (0.6 µM) or GA (20 µM, count by gold element) significantly suppressed IL-6, IL-1β and TNF-α expression levels in RAW264.7 macrophage cells. For human bronchial epithelial cells, AF (0.08 µM) and GA (10 µM, count by gold element) significantly inhibited phosphorylation of IKK, IκB, p65, thus significantly inhibiting IL-6, IL-1β, and TNF-α inflammatory cytokine expression. Jue et al. had demonstrated that the Cys179 of IKKβ plays a critical role in activation of NFκB pathway, and the anti-inflammatory activity of gold compound may depend on modification of thiolate of Cys179 via Au ion released from gold compounds [18]. For 16HBE cell the CC50 of AF is ~ 0.63 µM and that of GA approx. 1.06 mM, for RAW cell the CC50 of AF is ~ 2.63 µM, and that of GA approx. 1.44 mM (Fig. S7). This implied that GA has better safety in COVID-19 treatment.

GA decreases inflammatory cytokine level in lungs and protects lungs from inflammatory injury in COVID-19 transgenic mouse model

To evaluate the safety of administering AF or GA into COVID-19 animal model, we first tested the toxicity of AF and GA. The reported mice intraperitoneal LD50 for AF was approx. 33.8 mg/kg.bw [19], and that for GA more than 1000 mg/kg.bw (count by gold element, all mice survived, see SI 8). The SD rat intraperitoneal LD50 for AF was found to be approx. 25.5 mg/kg.bw [19], while that for GA approx. 288 mg/kg.bw (count by gold element, SI 8 ). Together, these animal toxicity and aforementioned cell toxicity data (Fig. S8) strongly suggested that GA is safer for mice/rats treatment than AF. We therefore continued investigation of GA in a COVID-19 transgenic mouse. To evaluate whether GA inhibit lung inflammation injury, the Ad5-hACE2-transduced mice were generated according to recently published methods [20]. Briefly, BALB/c mice were randomly divided into three groups, namely GA (GA treated COVID-19 mice), NS (0.9% NaCl treated COVID-19 mice), Mock (0.9% NaCl treated control mice), all mice were anesthetized with pentasorbital sodium and transduced intranasally with 2.5 × 108 FFU of Ad5-ACE2 in 50 μL DMEM (SI 9). Five days post transduction, the mice in GA group received a dose of 15 mg/kg.bw (count by gold element) via intraperitoneal injection (i.p.). For mice in NS group, an equivalent volume of normal saline (0.9% NaCl) via intraperitoneal injection (i.p.) as vehicle. For mice in Mock group, an equivalent volume of normal saline (0.9% NaCl) via intraperitoneal injection (i.p.) without SARS-CoV-2 infection as a control. One hour after GA or normal saline treatment, mice in GA or NS group were infected intranasally using SARS-CoV-2 (1 ×105 PFU) in a total volume of 50 μL DMEM. After virus infection at day 0, mice received GA or NS treatment for further three times as shown in Fig. 4A. All mice were euthanized at day 3 post infection, and several parameters were measured, including body weight loss, histopathological change in lung tissues, level of SARS-CoV-2 spikes in lung, and levels of key inflammatory cytokines (IL-6, IL-1β, TNF-α) in lungs.

Fig. 4

The GA protects lung injury and suppresses inflammatory cytokines and SARS-CoV-2 spike in lungs of COVID-19 mice. (A) Schematic diagram of GA treated COVID-19 mice. (B) Weight loss of infected mice treated either with GA or NS. (C) Histopathological scores of lung injury in SARS-CoV-2 infected mice. (D) Representative Hematoxylin-eosin (HE) staining of lungs from mice harvested at day 4 post infection. (E) The fluorescence intensity of IL-6, IL-1β, TNF-α, and SARS-CoV-2 spike in lungs of NS and GA treated COVID-19 mice, scale bar ~50 µm. Normalized fluorescence intensity of IL-6, IL-1β, TNF-α, and SARS-CoV-2 spike (unpaired t-test, *p < 0.05).

As shown in Fig. 4, infection of mouse model with SARS-CoV-2 resulted in a number of phenotypes, including obvious body weight loss and severe bronchopneumonia and interstitial pneumonia and infiltration of lymphocytes within alveolar. GA-treated COVID-19 mice are with good body weight in comparison of NS-treated COVID-19 mice (Fig. 4B). The histopathological changes in mice lung tissues were assessed by grading the injuries in accordance with the International Harmonization of Nomenclature and Diagnostic Criteria (INHAND) scoring standard. As shown in Fig. 4C, the GA-treated mice significantly reduced histopathological scores (~1.8) compared with that of NS-treated mice (~3.0). We further evaluated the therapeutic effects of GA using histopathological analysis of mouse lung tissues. As shown in Fig. 4D, mice infected with SARS-CoV-2 showed severe lung inflammation following treatment with NS, the alveolar septum, bronchus, bronchioles and perivascular interstitium were significantly widened, along with an infiltration of higher numbers of lymphocytes and a small number of neutrophils. In addition, a small number of lymphocytes and exfoliated epithelial cells localized in the lumen of local bronchioles following NS treatment. Treatment with GA significantly abrogated lung inflammation in SARS-CoV-2 infected mice, the local alveolar septum, bronchi, bronchiole and perivascular interstitial widening significantly decreased. Although we still observed mild lymphocytic infiltration, the mucosal epithelium of bronchus and bronchioles was intact, and we failed to observe foreign bodies in the lumen in lung of GA treated mice.

Next, we measured SARS-CoV-2 spike and the key inflammatory cytokines in lung of the mice using immuno-fluorescent imaging. As shown in Fig. 4E, the SARS-CoV-2 spike, IL-6, IL-1β, TNF-α expression level in the lung tissues of GA-treated infected mice were significantly lower than those found for NS-treated infected mice. Together, these results clearly demonstrated that GA inhibits virus replication (count by SARS-CoV-2 spike), while also suppressing inflammatory cytokine expression, thus protecting the lungs of infected mice from inflammation injury.

Via nasal dropping administration, GA shows better therapy outcome than Remdesivir in COVID-19 hamster model

Remdesivir was approved to treat COVID-19 in clinical in 2020. In this study, GA is used to compare with Remdesivir to see which one is with better outcome in COVID-19 hamster treatment. A golden Syrian hamster model was generated according to recently published paper [21]. Briefly, golden hamsters were randomly divided into five groups, and Hamsters were then infected intranasally using SARS-CoV-2 (1 ×105 PFU) in a total volume of 50 μL DMEM (SI 10). One hour after SARS-CoV-2 infection, hamsters in NS group received normal saline (0.9% NaCl), hamsters in Remdesivir group received a dose of 25 mg/kg.bw, and hamsters in first GA group received a dose of 5 mg/kg.bw and in second GA group received a dose of 10 mg/kg.bw (all count by gold element). Mock group are control hamsters. All hamsters are treated by intranasally dropping administration. After virus infection at day 0, hamsters received GA or Remdesivir or NS treatment for further 4 times as shown in Fig. 5A. All hamsters were euthanized at day 4 post infection, and several parameters were measured, including body weight loss, pathological change in lung tissues, level of SARS-CoV-2 spikes in lung, and levels of key inflammatory cytokines (IL-6, IL-1β, TNF-α) in lungs.

Fig. 5

The GA and Remdesivir protect lung injury and suppresses inflammatory cytokines and SARS-CoV-2 spike in lungs of COVID-19 hamsters. (A) Schematic diagram of GA or Remdesivir treated COVID-19 hamsters. (B) Weight loss of infected hamsters treated with GA or Remdesivir. (C) Histopathological scores of lung injury of GA or Remdesivir treated SARS-CoV-2 infected hamsters. (D) Representative Hematoxylin-eosin (HE) staining of lungs from hamster harvested at day 4 post infection, scale bar ~50 µm. (E) The fluorescence intensity of IL-6, IL-1β, TNF-α, and SARS-CoV-2 spike distributed in lungs of GA or Remdesivir treated COVID-19 hamster, scale bar ~50 µm. Normalized fluorescence intensity of IL-6, IL-1β, TNF-α, and SARS-CoV-2 spike (unpaired t-test, ***p < 0.001, **p < 0.01).

GA or Remdesivir treated COVID-19 hamsters are with body weight loss when compared to Mock hamsters (Fig. 5B). The histopathological changes in lung tissues are key index to assess the therapy effects of GA or Remdesivir in COVID-19 golden hamster. The lungs of hamsters were assessed by grading the injuries in accordance with the International Harmonization of Nomenclature and Diagnostic Criteria (INHAND) scoring standard. As shown in Fig. 5C and D, the average histopathological score of virus-infected hamsters in NS group was approx. 3, the alveolar septum, bronchus, and perivascular interstitium were significantly widened, along with an infiltration of lymphocytes and neutrophils. For hamsters infected by SARS-CoV-2, treatment of Remdesivir in dose of 25 mg/kg.bw got histopathological scores approx. 2.6, briefly the alveolar septum, bronchus, and perivascular interstitium were obviously widened, along with an infiltration of some of lymphocytes and neutrophils. Treatment of GA in dose of 5 mg/kg.bw got pathological scores, approx. 2.3, and treatment with GA in dose of 10 mg/kg.bw pathological score was approx. 2.3. GA treatment significantly decreased lung injury in comparison of Remdesivir treated SARS-CoV-2 infected hamsters. The local alveolar septum, bronchi, and perivascular interstitial widening were significantly decreased, along with an infiltration of smaller numbers of lymphocytes and neutrophils. According histopathological results, GA in dose of 5 mg/kg.bw or 10 mg/kg.bw is with better therapy outcome than Remdesivir in dose of 25 mg/kg.bw.

Next, we measured the level of SARS-CoV-2 spikes and inflammatory cytokines in lung of the hamsters using immuno-fluorescent imaging. As shown in Fig. 5E, the SARS-CoV-2 spike expression level in GA and Remdesivir treated COVID-19 hamster was significantly lower than those found in NS group, and the SARS-CoV-2 spike level in lung of Remdesivir group is significantly higher than that of GA group. For inflammatory cytokine level in lung of virus infected hamsters, IL-6, IL-1β, TNF-α of Remdesivir or GA-treated hamsters were significantly lower than those found for NS-treated hamsters, and the IL-6, IL-1β, TNF-α in lung of Remdesivir group is significantly higher than that of GA group. These immune-imaging results clearly demonstrated that GA intrinsically suppresses SARS-CoV-2 spike level and inflammatory cytokine expression in lungs of COVID-19 hamsters, the GA is with better outcome in suppression inflammatory cytokines expression when compared with Remdesivir.

The tissue distribution of Au and histopathological studies of GA treated mice and golden hamsters

BALB/c mice (15 mg/kg.bw count by gold element and treat via intraperitoneal injection) and golden hamster (10 mg Au/kg.bw via intranasally dropping administration) were treated as description (SI 11). GA treatment resulted in none measurable side effects. Neither movement, out-looking, sleeping, nor eating behavior appeared to be affected. When we stained tissue sections with Hematoxylin-eosin (HE), we could not detect any pathological changes in GA-treated mice or GA-treated hamsters in Fig. S8, indicating that our treatment dose of GA was safe. Next, we analyzed Au element distribution in lung, brain, liver, spleen, heart, and kidney using an ICP-MASS approach. As shown in Tables S2 and S3. The distribution of gold in lungs, hearts, livers, kidneys, brains, and spleens can be beneficial for COVID-19 treatment, as it potentially inhibits SARS-CoV-2 replication and direct suppressing the expression of inflammatory cytokines therein [15], [16], [17]. The Au element concentration in the lung of mice (intraperitoneal injection) and hamster (intranasally dropping) was approx. ~78.14 µg/g and ~416.12 µg/g, respectively. This result showed that GA is well adsorbed by lung after intranasally dropping method, which is a good way to deliver GA into lung for inhibit virus replication and directly suppress inflammation injury therein.

Conclusion

In summary, our findings show that GA represents a promising therapeutic compound that combines intrinsically suppression of inflammatory cytokines expression with directly inhibition of SARS-CoV-2 replication in COVID-19 transgenic mice and golden hamsters. Via intranasally dropping, GA (5 mg/kg.bw or 10 mg/kg.bw, all count by gold element) and Remdesivir (25 mg/kg.bw) GA significantly suppress the lung injury when compared with Remdesivir in COVID-19 hamsters. According immuno-fluorescent observations, GA significantly suppress the lung inflammatory cytokines expression, this might be attributed to GA directly inactivate NFkB pathway and further reduce inflammatory cytokines expression level in lung. The key factor that need to be considered when designing a drug for COVID-19 treatment is its toxicity. Our study showed that GA is characterized by very mild cytotoxicity and very low mice/rat acute toxicity. These features render GA an ideal candidate for safe and effective COVID-19 treatment.

Author contributions

X. G. conceived the project. H. Z., F. Y., C. Z., Z. D., W. Z., Y. G., Y. Q., F. G. performed cell/animal studies and enzyme activity. Y. G., J. F., B. H., P. C., B. S. performed the sample preparation, characterization and data collection. W. N., L. Z., X. G., X. G. performed theory calculation. Y. G., Y. D., J. Q., X. L., and H. J. processed data analysis. W.T., F. Y., J. Z. performed animal studies. X. G. wrote the manuscript; all authors discussed and commented on the results and the manuscript.

Declaration of Competing Interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.