Lithium salts as a treatment for COVID-19: Pre-clinical outcomes.

INTRODUCTION: Identifying effective drugs for Coronavirus disease 2019 (COVID-19) is urgently needed. An efficient approach is to evaluate whether existing approved drugs have anti-SARS-CoV-2 effects. The antiviral properties of lithium salts have been studied for many years. Their anti-inflammatory and immune-potentiating effects result from the inhibition of glycogen synthase kinase-3. AIMS: To obtain pre-clinical evidence on the safety and therapeutic effects of lithium salts in the treatment of COVID-19.
RESULTS: Six different concentrations of lithium, ranging 2-12 mmol/L, were evaluated. Lithium inhibited the replication of SARS-CoV-2 virus in a dose-dependent manner with an IC50 value of 4 mmol/L. Lithium-treated wells showed a significantly higher percentage of monolayer conservation than viral control, particularly at concentrations higher than 6 mmol/L, verified through microscopic observation, the neutral red assay, and the determination of N protein in the supernatants of treated wells. Hamsters treated with lithium showed less intense disease with fewer signs. No lithium-related mortality or overt signs of toxicity were observed during the experiment. A trend of decreasing viral load in nasopharyngeal swabs and lungs was observed in treated hamsters compared to controls.
CONCLUSIONS: These results provide pre-clinical evidence of the antiviral and immunotherapeutic effects of lithium against SARS-CoV-2, which supports an advance to clinical trials on COVID-19's patients.

Introduction

SARS-CoV-2 causes a severe respiratory disease (COVID-19) characterized by significant deregulation of the immune system with over-production of pro-inflammatory cytokines (cytokine storm) [1]. It shares about 80% genetic homology with SARS-CoV causing the severe acute respiratory syndrome epidemic in 2003 [2] that began in the Chinese province of Guangdong [3]. Like other coronaviruses, its viral particles include genetic material of a single chain of positive sense ribonucleic acid (RNA) and structural proteins necessary for the invasion of host cells, which include the surface or spike glycoprotein (S), the membrane glycoprotein (M), the envelope glycoprotein (E), the nucleocapsid glycoprotein (N) and the dimeric glycohemagglutinin-esterase [4]. Once inside the cells, the infective RNA encodes structural and non-structural proteins that control assembly, transcription and viral replication [5]. The N protein supports the viral RNA and is involved in its packaging [6], [7]. The native hypo-phosphorylated form of this protein favors its binding to genetic material [8], forming RNA-protein condensates.

Coronaviruses require a host’s kinase termed glycogen synthase kinase-3 (GSK-3) activity for its replication. This enzyme phosphorylates the N glycoprotein [9], [10] reducing RNA-protein interactions, forming a looser conglomerate [6], which favors the genome transcription, an essential step in the generation of long sub-genomic and genomic mRNAs that code for structural proteins in the cell cytosol during viral replication [9]. GSK-3 is constitutively expressed in all mammalian cells [11] and is important in several biological processes [12]. At the immune level, inhibition of cytosolic and nuclear GSK-3 increases the transcription of TBX21 (also called T-box expressed in T cells, Tbet), which in turn suppresses the transcription and expression of programmed cell death protein 1 (PD-1) in activated lymphocytes, in particular, TCD8 + lymphocytes [13]. PD-1 is a cell surface receptor expressed transiently in multiple cells of the immune system, including T and B lymphocytes [14], macrophages [15], natural killer (NK) cells [16], and dendritic cells [17]. It functions as an immune checkpoint and plays an important role in the negative regulation of the immune system [18]. Blockade of PD-1 is known to increase self-renewal and proliferation of stem-like TCD8 in lymph nodes, increasing effector functions of cytolytic T lymphocytes (CTL) against infections and cancer [19], [20]. The same inhibitory effect on the expression of another inhibitory receptor LAG3 has been seen [21]. GSK-3 inhibition down-regulates the inhibitory co-receptor LAG-3 on CD4 + and CD8 + T-cells to enhance anti-viral and anti-tumor immunity [21], [22].

Lithium was the first natural GSK-3 inhibitor described [23]. In psychiatry, lithium has been used as a specific treatment for mania since 1949 [24]. The antiviral properties of lithium have been studied since 1980 [25]. Although a detailed review of lithium antiviral effects lies outside the scope of this work, some other potential mechanisms might be relevant besides the GSK-3 inhibition, as the regulation of autophagy, the inhibition of the phosphatidylinositol signaling pathway [26] and inhibition of RNA polymerases [27]. In vitro experiments have demonstrated that, at non-toxic concentrations in cell cultures, it is capable of inhibiting the replication of different viral types, including herpes viruses [25], coxsackievirus [28], [29], hepatitis C virus [30] and coronaviruses [31], [32], [33].

Clinical administration of lithium is typically in the form of salts (lithium carbonate) as oral tablets (in doses of 0.4–2.0 g/day) and its effective therapeutic range lies between 0.5 and 1.2 mmol/L in blood serum [34]. Besides its antiviral properties, lithium has immunomodulatory/anti-inflammatory effects [35], [36], [37], [38], [39]. Despite the longstanding use of lithium salts in humans, including a proof of concept in COVID-19 [40], no pre-clinical trials have been conducted.

The study of the characteristics of viruses and the evaluation of drugs with antiviral properties is carried out first in cell cultures. Vero E6 cells are African green monkey (Cercopithecus aethiops) kidney epithelial cells commonly used in virology studies. Research projects on SARS-CoV and SARS-CoV-2 have been carried out using these cells because of their high expression of angiotensin-converting enzyme 2 (ACE2) receptor and their ability to support a high rate of viral replication [41], [42], [43], [44].

The availability and use of animal models in the study of the pathogenesis of COVID-19 is also of great significance in the pre-clinical evaluation of new therapeutics. Syrian hamsters (Mesocricetus auratus) are widely used as experimental animals and, due to the high degree of homology of its ACE2 receptor with that of humans, they have been used previously in SARS studies [45], [46], and have been defined as a bio-model for SARS-CoV-2 studies [47], [48], [49], [50]. After intranasal inoculation of hamsters with SARS-CoV-2, moderate interstitial pneumonia occurs with high viral load in lungs, although only a mild and rapidly resolving disease is observed [47].

Even though most studies in hamsters have been carried out using high infectious doses (103–105 mean tissue culture infectious dose, TCID50) [48], [50], [51], it has been defined that the mean infective dose (ID50) is only 5 TCID50 [47]. The most common manifestations of the disease in hamsters are weight loss, ruffle fur, lethargy (reduced mobility), hunched back and polypnea [47], [49], [50], [51].

The present work aims to obtain pre-clinical evidence on the safety and therapeutic effects of lithium salts in the treatment of COVID-19.

Materials and methods

Reagents

Minimal Essential Medium (MEM) (Gibco, Paisley, UK), gentamicin and L-glutamine (Capricorn Scientific, Ebsdorfergrund, Germany), lithium chloride monohydrate (LiCl*H2O), sulfuric acid, sodium bicarbonate, balanced salts solution (BSS) and Tween-20 (Merck, Darmstadt, Germany), bovine serum albumin (BSA), fetal bovine serum (FBS) and horseradish peroxidase (Sigma-Aldrich, St. Louis, MO, USA), neutral red (NR) (PanReac AppliChem, Darmstadt, Germany), tetramethylbenzidine (TMB) (Sigma-Aldrich, Darmstadt, Germany), anti-SARS-CoV-2N protein monoclonal antibody (CBSS NCoV) (Center for Genetic Engineering and Biotechnology, Sancti-Spíritus, Cuba) and lithium carbonate tablets (250 mg) (MedSol Laboratories, Havana, Cuba) were used.

Cell culture, virus strain, and animal model

Vero E6 cells (ATCC CRL-1586, CICDC) cryo-preserved at − 80 °C were defrosted and then subcultured twice a week at 3 × 104 cell/cm2 in MEM supplemented with 10% FBS, 2 mmol/L of 1% L-glutamine and 7.5% sodium bicarbonate. Cells were seeded in 96-well flat-bottom plates (Nunclon Delta Surface, Roskilde, Denmark) at a density of 20 × 103 cells/well in 100 µL of supplemented MEM and grown overnight at 37 °C in 5% CO2 atmosphere, with 4% FBS. SARS-CoV-2 strain (CUT-2010–2025) isolated at CICDC from a Cuban patient nasopharyngeal swab and stored at − 80°C was used. The TCID50 was determined by end point microtitration in Vero E6 cells at 10 × 103 cells/well [52], [53].

Twenty-four female Syrian hamsters from the National Center for the Production of Laboratory Animals (CENPALAB, Havana, Cuba), with 4–6 weeks old reception age, 70–80 g at the beginning of the experiments were used. Sample size calculation was based on law of diminishing return (“resource equation” method), with a degree of freedom of analysis of variance of 19 [54]. After a satisfactory clinical inspection, the hamsters were housed in polycarbonate boxes 1264 C Eurostandard Type II, floor area 370 cm2 (Tecniplast, Italy) at a rate of two animals per box at CICDC’s facility. Cages were identified with the following information: species, sex, age, name of the project, group of randomization, number and physical mark for identification. The hamsters were used after one week of acclimatization.

Animals received sterile pelletized feed EMO-1002 (AlyCo®, CENPALAB, Havana, Cuba) and acidified water ad libitum. During the experimental phase, animals were kept in a CICDC Biosafety Laboratory Level 3 (BSL-3) with the required environmental conditions of temperature (20 ± 2 ºC), relative humidity between 50% and 60%, and with a 12 by 12 h light-dark cycle. All experiments with animals were carried out in conformity to the Directive 2010/63/EU of the European Parliament and of the European Union council (22 September 2010) on the protection of animals used for scientific purposes. All efforts were made to minimize animal suffering and to reduce the number of animals used. Animals were euthanized with an overdose of thiopental.

All work with viable SARS-CoV-2 was performed in BSL-3 facilities.

Antiviral assay on Vero E6

Vero E6 monolayers were inoculated with 100 TCID50 of SARS-CoV-2 strain CUT-2010–2025 in 100 µL/well of supplemented MEM and incubated for 1 h. Then, cells were washed twice with phosphate-buffered saline (PBS) solution at pH 7.3–7.5. Six different concentrations of lithium (Li+), ranging 2–12 mmol/L, were prepared from a stock solution of LiCl*H2O dissolved in MEM with 2% FBS. A volume of 100 µL/well of each concentration was added to Vero E6 monolayer using six replicates. Each assay plate contained cells, either inoculated or not, as viral or cellular controls, respectively.

Microplates were incubated for 96 h at 37ºC in 5% CO2 atmosphere. Cytopathic effects were recorded 96 h after inoculation under inverted microscope. The antiviral efficacy of lithium was evaluated spectrophotometrically using the NR uptake assay [55] with slight modifications. Briefly, after removing the supernatant, 100 µL of 0.02% NR was added to each well. After incubating cells for 1 h at room temperature (20–25ºC) in the dark, NR solution was removed and cells were washed twice with 0.05% Tween-20 in PBS. A volume of 50 µL of lysis buffer (absolute ethanol/ultrapure water/glacial acetic acid, 50:49:1, v/v/v) per well was added, and plates were incubated for 15 min. Absorbance was measured using a PR621 microplate reader (Immunoassay Center, Havana, Cuba) at 520 nm [56].

Results were expressed as a percentage (%) of inhibition of lithium over SARS-CoV-2 replication. It was calculated using the following equation: % inhibition = 100 [(As – Avc)/(Acc – Avc)], where As is the absorbance of the sample, Avc is the absorbance of the viral control, and Acc is the absorbance of the cellular control. Half-maximal inhibitory concentration (IC50) is the most widely used and informative measure of a drug’s efficacy and indicates how much drug is needed to inhibit a biological process by half. Lithium half-maximal inhibitory concentration was defined as the concentration which achieved 50% inhibition of virus-induced cytopathic effects and was determined through a non-linear regression analysis of the concentration-response curve on Fig. 3, using GraphPad Prism 8.0 software (GraphPad Inc., CA, USA) [57].

Fig. 3

Antiviral activity of lithium against SARS-CoV-2 virus in Vero E6 cells and viability of lithium-treated cells Legend: Percent of inhibition of SARS-CoV-2 replication in Vero E6 cells, under increasing concentrations of lithium, is represented with circles; black squares represent viability of lithium-treated cells. Absorbance was measured using a PR621 reader at 540 nm. Error bars represent the SEM.

After visualizing the cytopathic effect, the supernatants were collected to detect and compare the amount of viral proteins. A standardized sandwich ELISA was performed in polypropylene 96-well plates, coated with 10 µg/mL of CBSS NCoV for capture. Sensitization was performed in a humid chamber at 37ºC for 2 h, with 100 µL/well of Na2CO3-NaHCO3 (0.05 mol/L, pH 9.6). Plates were washed four times with BSS-Tween 20 1X that contains NaH2PO4 (0.6 g/L), NaCl (4 g/L), KCl (0.2 g/L) and Na2HPO4 * 12 H2O (1.9 g/L) at pH of 7.2–7.4% and 0.31% of Tween-20. Plates were blocked with BSA and dried at room temperature for 24 h.

N protein of SARS-CoV-2 was used as positive control. The supernatant of non-infected Vero E6 was used as negative control. BSS that contains NaCl (100 g/L), NaH2PO4 (15 g/L), KCl (5 g/L) and Na2HPO4 * 12 H2O (47.5 g/L) with 1.25% of BSS-Tween-20 was used as a washing solution. CBSS NCoV conjugated to horseradish peroxidase was used for detection. TMB at 1:100 was used as chromogenic substrate.

Six replicates of samples were added to plates (100 µL /well) and incubated for 24 h at 37 °C in the dark in humid chamber. Plates were washed three times. Peroxidase conjugate was added to wells followed by incubation at 37 °C for 1 h in the dark, and then plates were washed. Revealing reaction was performed by adding 100 µL of TMB and incubating for 20 min. Reaction was stopped with 50 µL/well of a 4 mol/L sulfuric acid solution. Absorbance was measured using a PR621 reader at 450 nm. Cut-off value was established as the mean of the OD values of the negative controls plus three standard deviations (SDs). Samples with an absorbance value greater than or equal to the cut-off value were considered positive for the test.

Toxicity controls were set up in parallel on uninfected cells in every experiment.

Syrian hamsters and lithium carbonate administration

Animals were randomized into five groups: negative and positive controls (n = 4 each); lithium control (n = 4); and two groups of infected animals treated with lithium, one on the same day of the infection (T-0) and the other on the day before (T-1) (n = 6 each).

Lithium carbonate was administrated daily, intragastrically, at 15 mg/Kg of weight in 200 µL of saline solution to all animals, except those in control groups who received 200 µL of intragastric saline. The determination of lithium in the serum of treated hamsters was performed using a Roche/Hitachi Cobas C system (Cobas 6000). The blood samples were allowed to clot, and the sera were separated by centrifugation at 3000 rpm for 5 min and stored at − 20 °C until the analysis was performed. Lithium-control animals were checked daily for any signs suggestive of toxicity. Studied signs were lethargy/inactivity, seizures, and gastrointestinal signs (diarrhea).

Hamsters were anesthetized using a ketamine/xylazine cocktail and inoculated via intranasal instillation of 100 TCID50 of SARS-CoV-2 in 50 µL (25 µL in each nostril). Signs of the disease were monitored daily for all animals. Studied signs were lethargy, stooping and polypnea/abdominal breathing. Lethargy was defined as a clear state of reduced responsiveness and spontaneous activity; stooping was determined when the animal moved with hunched back or rested upright with the back hunched. To simplify the analysis of these signs, a value of one point was assigned to each of the signs observed per animal per day, making a maximum of three points per animal per day (lethargy, stooping and polypnea/abdominal breathing). Every animal was placed into an individual cage for 3 min. During this time, an observer blinded to the experimental conditions of the study, scored the hamsters according to the presence (=1) or absence (=0) of the evaluated sign (Fig. 5). All data were collected by observers blinded to the protocol.

Fig. 5

Assessment of COVID-19 signs in SARS-CoV-2-infected hamsters Legend: Mean and error of cumulative signs for each group. Black squares represent infected hamsters treated with lithium (T-1 and T-0 groups were combined); black circles represent infected untreated hamsters. Error bars represent the SEM.

Two animals of each group were euthanized on the fifth day post-infection. Nasal swabs followed by reverse transcriptase-polymerase chain reaction (RT-PCR) and culture on Vero E6 cells on days 2, 4, 5, and 7 were performed. The presence of the virus in the nasopharynx and lungs was determined. A histopathologic study of the lungs tissue was carried out. The left lung was fixed in 10% formalin before paraffin embedding. Tissue Section (3 µm) were analyzed after staining with hematoxylin and eosin (H:E) and scored blindly for lung damage by an expert pathologist. To facilitate the analysis, a value between 0 and 4 points was assigned to lung injure.

Statistical analyses

Statistical analysis was performed in GraphPad Prism 8.0 software (GraphPad Inc., CA, USA). Six replicates of all cells experiments were made and results were reported as the mean ± SD. T-tests were used to assess studies with two groups, and ANOVA was used to analyze studies with > 2 groups. A statistical significance was considered present when p < 0.05.

Results

Lithium treatment directly inhibits SARS-CoV-2 replication on Vero E6

Initially, to examine the effect of lithium salts on SARS-CoV-2, Vero E6 cells were cultured as monolayers in the presence or absence of lithium chloride monohydrate. Cells treated with lithium showed a significantly less cytopathic effect and more cellular viability than viral controls, as determined by inverted microscopic observation ( Fig. 1) and NR uptake assay ( Fig. 2), respectively. Maximal expression of cytopathic effect was detected on viral control wells at 96 h after inoculation. Cytopathic effect was characterized by cell monolayer destruction, with blank spaces, multinucleated giant cell (syncytium) formation, and cell rounding. Lithium-treated wells showed a significantly higher percentage of monolayer conservation than viral control, particularly at concentrations higher than 6 mmol/L (Fig. 1).

Fig. 1

Inverted microscope (100x) observation of the cytopathic effect of SARS-CoV-2 on Vero E6 monolayer treated with different concentrations of lithium (mmol/L) Legend: Numbers in the upper left corner of each photo indicate the lithium concentration in mmol/L; Cellular C. is the Vero E6 monolayer cultured under normal conditions; Virus C. is the Vero E6 monolayer infected with 100 TCID of SARS-CoV-2.

Fig. 2

Neutral red uptake assay in Vero E6 cells infected with SARS-CoV-2, treated with different concentrations of lithium (mmol/L) Legend: NR absorption expressed as optical density (OD) correlates with cellular viability. Infected-treated wells under increasing concentrations of lithium are shown. Cc represents the cellular control; Vc represents the viral control (infected-untreated). Absorbance was measured using a PR621 reader at 540 nm. Error bars represent the standard error of the mean (SEM). Lithium inhibited the replication of SARS-CoV-2 virus in a dose-dependent manner with an IC50 value of 4 mmol/L (Fig. 3). Viability of non-infected Vero E6 cells in the presence of lithium are shown.

To further confirm the efficacy of lithium inhibition of SARS-CoV-2 replication in Vero E6 cells, an ELISA was performed to detect viral N protein in the harvested supernatant of a reduced sample of wells and not all lithium concentrations were evaluated. Determination of N protein in treated wells matched the uninfected cellular control observing a significant downgrading (p < 0.001). Fig. 4 represents the mean OD in the supernatant of treated cells, according to concentrations evaluated. This semi-quantitative method allowed us to compare concentrations of the N protein of SARS-CoV-2 in supernatants of treated and untreated infected-cells. Overall, direct anti-SARS-CoV-2 effect of lithium, evaluated by cytopathic effects, cell-viability and absence of viral replication was demonstrated.

Fig. 4

Concentration of N protein of SARS-CoV-2 in the supernatant of culture Vero E6 treated with lithium. Legend: A standardized sandwich ELISA was performed to determine the N protein of SARS-CoV-2 in the harvested supernatant of treated wells. Absorbance was measured using a PR621 reader at 450 nm. Cut off value was set up at 0.129.

Lithium treatment affects SARS-CoV-2 infection decreasing the appearance and intensity of signs of COVID-19 in Syrian hamsters

Clinical signs of COVID-19 were markedly reduced in both groups of Syrian hamsters treated with lithium. There were no significant differences between this groups and the signs of both were presented together in Fig. 5. Absence of signs prevailed in four animals, another four presented signs only for one day, one animal for two days, and none had polypnea. Meanwhile, all animals in the viral control group showed lethargy and stooping between days 3–5 post-infection, and two of them had polypnea on the fifth day ( Table 1). Three animals died after anesthesia on the first day (two in T-1 and one in T-0) and were excluded from statistical analysis.

Table 1

COVID-19 behavior in Syrian hamsters treated or not with lithium carbonate.

Image 1

Legend: C+ represents the positive control (infected-untreated); T-0 represents infected-treated with lithium starting on the same day of the infection; T-1 represents infected-treated with lithium starting on the day before the infection. The arrow represents gradation of the intensity of signs, from absence to maximum intensity. L, lethargy; S, stooping; P, polypnea; PI, days post-infection; Nd, not done.

No signs of the disease were found in animals of negative control group (Fig. 5). No lithium-related mortality or overt signs of toxicity were observed during the experiment. Loss of weight with referred infective dose was not observed (data not shown).

In the histopathological evaluation of the lungs five days after infection, large leukocyte condensates were observed with loss of normal histology of the organ of the infected-untreated controls, with a high percentage of lung involvement. Animals treated with lithium were characterized by small inflammatory condensates and peribronchiolar infiltrates ( Fig. 6), which, according to clinical manifestations, did not have great repercussions on the dynamics of the organism. An inflammatory condensate of large magnitude was observed only in one treated animal (Table 1).

Fig. 6

Histopathologic study of the lungs tissue of infected Syrian hamsters Legend: Tissue Section (3 µm) staining with hematoxylin and eosin A-C. Infected-untreated hamster (40x-400x): large leukocyte condensates, high percentage of lung injury; D-F: infected-treated with lithium (40x-400x), smaller inflammatory condensates and peribronchiolar infiltrates.

Samples obtained by nasopharyngeal swab, nasopharyngeal lavage, or lung maceration after euthanasia were seeded in Vero E6 cell cultures to confirm the infection. A RT-PCR was also performed to compare the viral load in each animal. Between days 2 and 4 post-infection, there was a trend towards a lower viral load (higher Ct value including three with a value higher than 35, considering negative at day 4) in hamsters treated with lithium over viral controls. This was also observed in the analysis of the lungs of euthanized hamsters on the fifth day post-infection. However, an inconclusive dispersion was observed in terms of viral load of swabbing and nasopharyngeal lavages on the fifth day post-infection ( Fig. 7). In our study, the intensity of clinical signs of COVID-19 in Syrian hamsters did not directly correlate with the viral load detected in each animal. At 11 days, all animals became negative by RT-PCR (data not shown). Overall, the effects of lithium on SARS-CoV-2 infection in hamsters, particularly at signs and lungs histology levels (regarding signs and lung injure), were demonstrated.

Fig. 7

RT-PCR for SARS-CoV-2 in infected Syrian hamsters Legend: NF, nasopharyngeal; NFL, NF lavage. Horizontal lines represent the mean of each group (continuous for viral control; discontinuous for lithium-treated). Cut off value of Ct was established at 35.

Discussion

SARS-CoV-2 causes a severe respiratory disease (COVID-19) that remains a global crisis [1]. Any new therapeutic approaches will be of importance and potentially could save lives. In this paper, we provide evidence that lithium salts have ablative effects in vitro on the infection of cells and in a Syrian hamster model of infection. We found that lithium inhibited the replication of SARS-CoV-2 in Vero E6 cells in a concentration-dependent manner. The Vero E6 monolayer was conserved as verified by microscopic observation and a NR uptake assay. Cell viability is directly proportional to NR absorption. Fig. 2 clearly shows a superior uptake of NR (expressed as OD) of lithium-treated wells compared to the viral control (infected-untreated). This is consistent with an important antiviral effect of lithium, as show in Fig. 3. Further, using an ELISA we showed that lithium effectively reduced viral titers when comparing the OD of the supernatants of treated wells with the OD of the supernatants of viral control cells (p < 0.001).

Previous studies have shown that lithium salts can reduce coronaviruses infection when used at concentrations ranging from 5 to 25 mmol/L [31], [33], [58], [59]. Tested concentrations in our study correlates with antiviral evaluation led by Harrison et al. on the avian coronavirus infectious bronchitis virus, a work that demonstrated that 5 mmol/L of LiCl cause a 50% reduction in virus titers [33]. Liu et al. [10] showed that GSK-3 inhibitors, including lithium, inhibit SARS-CoV-2 N protein phosphorylation, an essential step in viral replication. LiCl inhibited N phosphorylation with IC50 ∼10 mM in 293 T cells, above the concentration needed to inhibit viral replication in Vero E6 cells in our study. Liu et al. also demonstrated that low doses of the highly selective GSK-3 inhibitor CHIR99021 reduced viral titers over 15-fold in the supernatant from SARS-CoV-2–infected human lung epithelium-derived cell line Calu-3.

Our findings are consistent with and supported by the recent in vitro studies of Shapira et al. [60], who tested three active compounds with GSK3β inhibitory effects and demonstrated their ability to reduce SARS-CoV-2 infection in various cell lines, with a significant 2-log reduction 48 h after infection and treatment.

In vitro non-toxic concentrations of lithium chloride have been set up between 15 and 50 mmol/L [31], [58], [59]. Our study evaluated concentrations up to 12 mmol/L of Li+. Although we did not perform a full toxicity assay, we assessed the non-toxicity of used concentrations in parallel.

As coronaviruses require host GSK-3 activity for their replication [10], [19], its inhibition represents a direct antiviral mechanism of lithium. Besides, lithium can potentiate CTL activity through PD-1 and LAG3 inhibition, favoring indirect antiviral activity. Although it is difficult to translate pre-clinical results into actual clinical treatment in patients, lithium has advantages over other antiviral drugs. In addition to its antiviral activity, lithium has anti-inflammatory effects by decreasing the production/expression of inflammatory-associated mediators [37] such as prostaglandin E2, cyclooxygenase-2 [61] and various pro-inflammatory cytokines, including interleukin (IL) 1 [39], IL-6 [39], [62] and tumor necrosis factor-alpha (TNF-α) [63].

With these results on cellular cultures, we decided to assess the therapeutic effect and safety of lithium on Syrian hamsters infected with SARS-CoV-2. It was determined to use 100 TCID50 as infective dose that is 20 times higher than the average infective dose of SARS-CoV-2 for this animal model [47]. This allowed us to observe the appearance of a moderate/severe disease in these animals. As the most frequent signs of COVID-19 described in Syrian hamsters are loss of weight, stooping, lethargy and polypnea [47], [49], [50], [51], it was decided to evaluate them on a daily basis. Surprisingly, no significant loss of weight was observed in infected animals. This may be due to the low infective dose used compared to other studies [47], [49], [50], [51]. We hypothesized that the higher the infective dose, the greater the weight loss. We found that hamsters treated with lithium showed a less intense disease with fewer signs. Some animals treated with lithium did not show any signs of the disease. The presence of signs in the viral control group (infected untreated) was significantly higher than those treated with lithium. This is more evident in Fig. 5, since although the n in the groups treated with lithium was significantly higher (n = 9) than in the untreated group (n = 4), the average number of signs of hamsters per day, was higher in the group of the infected-untreated animals. Even two animals of the viral control group showed polypnea, a sign that reflects the greater severity of the disease.

As a worsening peak has been described on the fifth day post-infection in this animal model, followed by rapid improvement [50], it was therefore decided to euthanize two animals in each group to compare the viral load and histological lesions in the lungs.

We evaluated in vivo the antiviral and immunotherapeutic potential of lithium. Despite the fact that between days 2 and 4 post-infection a trend towards a decrease in viral load was observed in treated hamsters compared to controls, on the fifth day post-infection an inconclusive dispersion was noted in the Ct value of the swab and nasopharyngeal lavage samples. However, viral load in lungs at fifth day post-infection was also discreetly inferior in treated animals.

Finally, the histopathologic study of the lungs showed less tissue damage in those treated animals compared to the untreated ones, which is consistent with the decreased amount of signs of the disease found in the study group. Given that the freedom of analysis in this case [54] was below 10, these findings should be interpreted carefully and further investigations are required with larger sample size.

The assessment of the immunotherapeutic effect of lithium salts in animal models allows us not only to evaluate their antiviral capacity, but also other properties such its immune-potentiating and anti-inflammatory effects.

Due to the rapid evolution of the disease in Syrian hamsters and the high replication rate of SARS-CoV-2 in this animal model [49], [50], we decided to explore whether there were differences between hamsters treated 24 h before infection in comparison to those that started treatment the same day of infection. There were no significant differences between the groups treated with lithium. This may be due to the rapid intestinal absorption of lithium [34] that does not require starting treatment from a day before in this animal model.

Unlike other drugs that did not replicate in Syrian hamsters the antiviral results obtained in cell cultures [64], [65], lithium showed, although slight, a tendency to decrease the viral load in nasopharyngeal swabs and lungs. Although it must be said that the infective dose used in those studies was higher than ours, it has been shown that a low-dose challenge inoculum induced a comparable viral load and disease pathology compared with higher viral dose challenges [66].

A dose of 15 mg/kg of body weight of lithium carbonate per day was chosen for treatment. This dosing regime is equivalent to the usual daily dose in humans, about 1000 mg/day [34] and it yielded a serum lithium level of 0.91 ± 0.27 mmol/L, within the human therapeutic range. Hence, the data obtained from the current study could be extrapolated to clinical scenarios. For estimation of the toxicity of a drug, the mean lethal dose (LD50) is a commonly used pharmacological parameter and is determined in laboratory animals, preferably in mice and rats. For lithium carbonate the LD50 in rats is ∼700 mg/kg after oral administration [67]. In our search we did not find the LD50 for Syrian hamsters and its determination is beyond the scope of this work. The lithium regimen used is markedly below the LD50 for lithium application in rats, and also below the usual doses in pre-clinical toxicology studies [68], [69].

Taking into account the pre-clinical results shown here, it seems that the immunomodulatory/anti-inflammatory properties of lithium play a greater role in the improvement of subjects affected by COVID-19. This gains relevance as the inflammatory host response drives much of the pathology of the SARS CoV-2 infection. A reduced incidence of COVID-19 in patients taking lithium compared to the general population was demonstrated after the analysis of clinical data from four major health systems [10], [40]. Unfortunately, severity and mortality data were not obtained and compared between the two groups of infected patients, since modulation of the inflammatory response by lithium may also contribute to reduce the risk of COVID-19 severity and mortality [70].

Spuch et al. [40] treated six patients with severe COVID-19 with lithium carbonate. The authors found that lithium carbonate reduced plasma C-reactive protein levels and increased lymphocyte numbers, reducing the neutrophil-lymphocyte ratio. All patients treated with lithium carbonate improved their clinical status and no side effects related to lithium carbonate treatment were reported. These preliminary data suggest that lithium treatment reduces the inflammatory state of patients with COVID-19, improving both inflammatory activity and the immune response. Because of the limitations of this study such as the small number of cases and the lack of randomization, these findings should be interpreted carefully and deeper investigation is required in a cohort with larger sample size.

Four cases of lithium toxicity in COVID-19 patients under long-term lithium treatment have been reported [71], [72]. A recent study stablished a 1% incidence of lithium intoxication in chronically treated patients [73]. They assessed the incidence, clinical course and renal function based in a retrospective cohort study, and concluded that lithium intoxication seems rare and can be safely managed in most cases. Also, lithium intoxication is associated with prolonged use, and short period treatment could minimize the events associated with toxicity [74].

Considering the world’s current situation as a consequence of the SARS-CoV-2 pandemic, we believe that the information provided by this study supports the performance of clinical trials for the lithium treatment of patients with COVID-19.

Limitations of study

Our study has the limitation of have not determined pro-inflammatory markers (i.e. C-reactive protein and IL-6) in Syrian hamsters treated or not with lithium.

Conclusions

Lithium directly inhibits the replication of SARS-CoV-2 in Vero E6 cells in a concentration-dependent manner. It was effective in treating Syrian hamsters with COVID-19, evidenced by a decrease in viral load and the appearance of signs of the disease and/or its intensity. These results reaffirm the possible therapeutic effect of lithium salts and support further clinical trials in patients with COVID-19.

Authors contributions

OP, OST, and ENR designed and led the study.ENR, OST, NGS, ROH, JMEP, MGM, AFQ, YIH, DMA, MTZD and IAH performed the virological assays and experimental work, and contributed to analysis of the results and preparation of figures. ML, LRP, RRP, LMR, IGV, VEG, MRA, IGR, AMS, LAC and YB supported and contributed to the study design and analysis. All authors revised the manuscript and approved submission.

Ethics statement

Animal experimental procedures were approved by the Ethics Committee of CICDC and performed in compliance with national and international guidelines for care and humane use of animals, and in conformity to the Directive 2010/63/EU of the European Parliament and of the European Union council (22 September 2010) on the protection of animals used for scientific purposes.

Conflict of interest statement

The authors declare no conflict of interest.