Cold-adapted SARS-CoV-2 variants with different temperature sensitivity exhibit an attenuated phenotype and confer protective immunity

Abstract

As novel SARS-CoV-2 Variants of Concern emerge, the efficacy of existing vaccines against COVID-19 is declining. A possible solution to this problem lies in the development of a live attenuated vaccine potentially able of providing cross-protective activity against a wide range of SARS-CoV-2 antigenic variants. Cold-adapted (ca) SARS-CoV-2 variants, Dubrovka-ca-B4 (D-B4) and Dubrovka-ca-D2 (D-D2), were obtained after long-term passaging of the Dubrovka (D) strain in Vero cells at reduced temperatures. Virulence, immunogenicity, and protective activity of SARS-CoV-2 variants were evaluated in experiments on intranasal infection of Syrian golden hamsters (Mesocricetus auratus). In animal model infecting with ca variants, the absence of body weight loss, the significantly lower viral titer and viral RNA concentration in animal tissues, the less pronounced inflammatory lesions in animal lungs as compared with the D strain indicated the reduced virulence of the virus variant. Single intranasal immunization with D-B4 and D-D2 variants induced the production of neutralizing antibodies in hamsters and protected them from infection with the D strain and the development of severe pneumonia. It was shown that for ca SARS-CoV-2 variants, the temperature-sensitive (ts) phenotype was not obligate for virulence reduction. Indeed, the D-B4 variant, which did not possess the ts phenotype but had lost the ability to infect human lung cells Calu-3, exhibited reduced virulence in hamsters. Consequently, the potential phenotypic markers of attenuation of ca SARS-CoV-2 variants are the ca phenotype, the ts phenotype, and the change in species specificity of the virus. This study demonstrates the great potential of SARS-CoV-2 cold adaptation as a strategy to develop a live attenuated COVID-19 vaccine.

1. Introduction

The COVID-19 pandemic has become the main medical challenge of the 21st century. Almost all vaccines for the specific prevention of COVID-19 that have been or are being evaluated in clinical trials are based on replication-defective viral vectors, self-replicating RNA molecules, recombinant or native viral antigen preparations [1], [2], [3], [4]. These technological platforms allow the rapid development of safe vaccines capable of inducing a protective immune response. The development and widespread use of vaccines has significantly reduced the incidence of hospitalization and mortality from the disease [1], [4]. However, licensed vaccines generally have high production costs, include a limited number of protective viral antigens, provide a short-lived immune response, and their effectiveness declines when novel SARS-CoV-2 Variants of concern (VOC) appear [5], [6] since SARS-CoV-2 is evolving rapidly. Indeed, the Delta (B.1.617.2) VOC replaced the Alpha, Beta and Gamma variants due to increased infectivity and was less neutralized by the serum obtained from COVID-19 convalescents caused by other SARS-CoV-2 variants [7], [8], [9]. Delta VOC was replaced by highly contagious Omicron (B.1.1.529) VOC, which by February 2022 had occupied the dominant position, constituting over 95% of all the strains characterized by sequencing (https://www.gisaid.org/). The Omicron variant has several deletions in the genome and more than 30 amino acid substitutions in the S-protein, which have led not only to increased infectivity of the virus, but also the ability to evade the neutralizing antibodies of COVID-19 convalescents and vaccinated individuals [10], [11], [12], [13], [14]. Therefore, the monoclonal antibodies used in COVID-19 therapy also proved to be less effective against the Omicron variant [15]. Consequently, there is an urgent need to develop a safe and effective vaccine with cross-protective activity against a wide range of SARS-CoV-2 VOCs. Live attenuated COVID-19 vaccines may be considered as a potential highly effective strategy to fight the threat posed by SARS-CoV-2 [16], [17].

A long history of successful use of live attenuated viral vaccines has proven their ability to generate a long-lasting cellular and humoral immune response and cross-protection against different antigenic virus variants [18], [21]. However, according to the WHO data (22.07.2022), out of 169 vaccines against COVID-19, licensed or in clinical trials, only two vaccines (1.2%) were based on live attenuated strains constructed by codon deoptimization, COVI-VAC (Codagenix/Serum Institute of India, India) and MV-014-212 (Meissa Vaccines, Inc, USA) [19].

Previously, it was shown that adaptation of viruses to growth at suboptimal low temperature leads to a temperature sensitive (ts) phenotype (reducing viral replication at 37 °C or higher) associated with attenuation of virulence for a normal host [20]. In this case, the cold-adapted (ca) attenuated virus provides safe and effective protection against wild-type virus infection [20], [21]. Recently, scientific groups from Japan, Korea and Iran reported ca attenuated SARS-CoV-2 strains exhibiting a ts phenotype [22], [23], [24]. In this work, we generated two ca live attenuated SARS-CoV-2 clones with different temperature sensitivity and evaluated their virulence, immunogenicity, and protective activity in a Syrian golden hamster model of coronavirus pneumonia.

2. Materials and methods

2.1. Viruses

The laboratory SARS-CoV-2 Dubrovka (D) strain and its variants: Dubrovka-37 (D-37), Dubrovka-ca (D-ca), Dubrovka-ca-B4 (D-B4), and Dubrovka-ca-D2 (D-D2) (Supplementary Table S1), were isolated or generated in our laboratory previously and used in this study. The D strain (GenBank number MW514307.1, clade GR according to the GISAID classification, line B.1.1.317 according to the Pangolin classification), phylogenetically close to strain Wuhan-Hu-1 (GenBank number NC_045512.2), was obtained and characterized in summer 2020 in Moscow (Russia) by isolation in Vero cells from a clinical sample of a patient with COVID-19 [25]. Variants of D strain, D-37 and D-ca, were obtained by propagating of the D strain in Vero cells for 42 passages [26]. The D-37 variant was adapted to a constant cultivation temperature of 37 °C. The D-ca variant (cold-adapted) was gradually adapted from 37 °C to 23 °C according to the following scheme: 10 passages at a temperature of 37 °C, then the cultivation temperature was lowered by 1 °C every two passages, the final 6 passages were carried out at a temperature of 23 °C (a total of 42 passages) [26]. D-B4 and D-D2 variants were generated by three-fold cloning of the D-ca variant at 23 °C by limiting dilution method (Supplementary Fig. S1). The D strain revealed a 27 nucleotide deletion in the N-terminal region of S-gene (encodes 9 amino acids from 68 to76 a.a. – YMSLGPMVL in the S-protein) appeared already on the 1st passage and persisted with prolonged cultivation in all variants.

SARS-CoV-2 strains belonging to Delta and Omicron VOC isolated in Vero cells in the Moscow region (Russia) were used in the viral neutralization test: Podolsk strain (collection date: 2021–10-08, GenBank ID ON032860.1, Delta B.1.617.2.122); Otradnoe strain (collection date: 2022–01-25, GenBank ID ON032857.1, Omicron BA.1.1).

2.2. Cells and virus cultivation

Virus cultivation was performed on African green monkey kidney epithelial cells Vero CCL81 (ATCC), Vero cells, and human lung cancer cells Calu-3 HTB-55 (ATCC), Calu-3 cells. Cells were cultured at 37 °C in DMEM medium complemented with Earle’s balanced salt solution (PanEco, Russia), 5% fetal bovine serum (FBS) (ThermoFisher Scientific, USA), L-glutamine (300 µg/ml, PanEco), and gentamicin (40 µg/ml, PanEco) in an atmosphere of 5% CO₂. A three-day monolayer of Vero or Calu-3 cells was infected with the SARS-CoV-2 virus at different multiplicity of infection (MOI). After 60 min of viral adsorption at 37 °C, 0% FBS maintaining medium was added and cells were incubated at 23 °C to 39 °C for 3–8 days (depending on viral variant) in an atmosphere of 5% CO₂. To study the kinetics of viral replication, the culture medium was collected daily within 3–6 days after infection and stored at −80 °C until use. Vero cells were tested for the absence of mycoplasma contamination with the “Myco Real-Time” (Evrogen, Russia) kit based on real-time PCR.

2.3. Animals

4-week-old 40–50 g female Syrian golden hamsters (Mesocricetus auratus) were kept under sterile conditions for SPF-animals (Research Institute of Laboratory Animals, IBC RAS, Russia). The hamsters were screened for the absence of viruses, as well as bacterial infections and parasites following the recommendations of the European Laboratory Animal Science Association. Hamsters were arbitrarily assigned to study groups, had free access to food and sterilized tap water, and were kept on a 12-h light/dark cycle. All studies with animals were carried out in accordance with the National Research Council's Guide for the Care and Use of Laboratory Animals [27] and approved by the Mechnikov Research Institute of Vaccines and Sera Institutional Animal Care and Use Committee.

2.4. Virus titration

The SARS-CoV-2 virus titers were determined by the cytopathic effect endpoint method (CPE) in Vero cells as described earlier [26]. The virus titer was calculated as described by Ramakrishnan M.A. et al [28] and expressed as log10 TCID₅₀/ml.

2.5. Quantification of SARS-CoV-2 RNA

Real-time RT-PCR was performed as described earlier [29]. To detect viral RNA we used primers and the probe designed for the nucleocapsid N gene of the SARS-CoV-2 virus, proposed by Chan J. et al [30]. Samples obtained by successive ten-fold dilutions of synthetic oligonucleotide COVN-PC (Table S2) with known concentrations were used to construct a calibration curve.

2.6. SARS-CoV-2 genome sequencing

NEBNext® ARTIC SARS-CoV-2 library preparation kit (New England Biolabs, USA) was used to obtain a pool of amplicons for subsequent genome-wide sequencing. This kit was designed to perform the SARS-CoV-2 whole-genome sequencing based on the “SARS-CoV-2 McGill Nanopore sequencing protocol SuperScript IV_42C_ArticV3″ (dx.https://doi.org/10.17504/protocols.io.bjajkicn). Ligation Sequencing kit 1D and Native Barcoding Kit 1D (Oxford Nanopore Technologies, UK) were used to prepare the resulting pool. Nanopore sequencing was performed in a Flow Cell R9.4 using MinKNOW software (Oxford Nanopore Technologies, UK). Genome assembly was performed in Minimap2 v. 2.24 (https://github.com/lh3/minimap2).

2.7. Evaluation of the ts phenotype of SARS-CoV-2 ca variants

Vero cells were infected with ca variants of SARS-CoV-2 and the D strain at a multiplicity of infection (MOI) 0.001 or 0.00001 and incubated at 37 °C or 39 °C in an atmosphere of 5% CO₂ for 3 days. Supernatants were collected daily and stored at −80 °C until use. Viral titers and viral RNA concentration were determined in the collected samples. A 4.0 log10 or greater difference in viral titer or viral RNA concentration compared to infection with the parental D strain was considered as the ts phenotype of the virus as described by Larionova N.V. et al [31].

2.8. Evaluation of virulence and protective activity of SARS-CoV-2

Virulence and efficacy were assessed according to the scheme (Supplementary Fig. S2). Syrian hamsters were divided into four groups of nine animals each. Each animal received intranasally 4.0 log10 TCID₅₀ of the D strain (passage 17) or its variants, D-B4 or D-D2. When determining infectious dose we based on Sia SF et al [32]. Infectious dose in 4.0 log10 per animal was established empirically before the start of the study - this dose of the D strain reproducibly led to significant weight loss compared to uninfected animals and the development of severe lobar pneumonia. Before intranasal procedures, animals were anesthetized and held in an upright position for viral infection. The negative control group received an equivalent amount of PBS. Hamsters were observed daily and weight control was performed every 2 days. Four days after infection, 4 animals from each group were euthanized. Hamsters were randomly selected and sacrificed during experiments regardless of the weight and condition of the animals. Lung, brain, and other organ tissues were collected, homogenized in DMEM medium with gentamicin (40 µg/ml, PanEco) using a Tissue Lyser LT homogenizer (Qiagen, Netherlands), and centrifuged at 10 000 rpm for 5 min at 4 °C. Supernatants were collected to measure virus titers and viral RNA concentration. The absence of body weight loss, death, the significantly lower viral titer and viral RNA concentration in animal tissues, and the less pronounced inflammatory lesions in animal lungs as compared with the D strain indicated the reduced virulence of the virus variant, thus, the attenuated (att) phenotype. To evaluate the protective activity 21 days after immunization, each animal (5 animals per group) received 4.0 log10TCID₅₀ of the D strain intranasally. Hamsters were observed daily and weight control was performed at day 0 and day 4. Four days after infection, animals were euthanized. Animal tissues were prepared for determination of viral titer, viral RNA concentration, and histological examination as described above.

2.9. Lung histological examination.

The right hamster lung was fixed in 10% neutral buffered formalin for 24 h, dehydrated, then embedded in Histomix paraffin medium (BioVitrum, Russia) and used for histological study. The lung tissue was cut into 3–5 µm sections by Leica RM 2125 RTS rotary microtome (Leica, Germany), then stained with hematoxylin and eosin solution and embedded in Canada balsam (Sigma-Aldrich). Histological preparations were visualized under a BX 51 light microscope (Olympus, Japan).

2.10. ELISA

Determination of antibodies to SARS-CoV-2 in hamster sera was performed using BioKit ELISA reagent kit (Bioservice, Russia) according to the instruction. Inactivation of SARS-CoV-2 by UV-light was performed as described earlier [29]. For use in the ELISA, a 45 ml centrifuged clarified UV-inactivated virus-containing supernatant (≥8.5 log10 TCID₅₀/ml) was passed through a 100 kD Amikon MWCO centrifuge filter (Millipore, Ireland) at 4000 rpm. The virus preparation collected on the filter was diluted to 4.5 ml with sterile PBS (pH 7.2), achieving a 10-fold concentration of SARS-CoV-2 virions. Before use, the preparation was processed on an MSE ultrasonic disintegrator (UK) at amplitude 2 for 2 min. The native ultraviolet (UV) inactivated viral antigen (D strain) in a dilution of 1:100 was sorbed on to wells of an immunoassay plate.

Double dilutions of the sera were analyzed by ELISA, starting at a dilution of 1:50. Anti-HAMSTER IgG (H + L)-Peroxidase antibody produced in goat (Sigma-Aldrich) was used in a dilution of 1:10000 for detection of hamster antibodies. The reciprocal value of the last dilution at which the OD value of the sample was higher than the cut-off threshold in each assay was taken as the titer of SARS-CoV-2 antibodies. The OD value for the negative serum multiplied by 2 was taken as the cut-off threshold.

2.11. Viral neutralization test

Measurement of SARS-CoV-2 neutralizing antibodies was performed in Vero cells according to the protocol described earlier [25]. D (Wuhan-like), Podolsk (Delta) and Otradnoe (Omicron) SARS-CoV-2 strains were used to determine the neutralizing activity of sera for different antigenic variants of the virus. The neutralizing titer was considered the reciprocal value of the last dilution, in which no signs of CPE were detected in two or more wells.

2.12. Statistical analysis

Statistical processing was performed using Graphpad Prism v.5.03 software. The data were presented as the mean ± standard deviation (SD) and mean ± standard error (SE) on the plots. The normality was checked based on the Shapiro-Wilk Test. The differences in hamsters body weights of Syrian hamsters and log10-transformed viral titres and quantitative viral RNA in different organs between strain variants and over time were compared using two-way ANOVA followed by Tukey’s multiple-comparison test or Kruskal Wallis test with Dunn’s Multiple Comparison Test. The Holm-Bonferroni correction for multiple testing was applied for primary analysis. The differences in hamsters body weights of Syrian hamsters over time (before and after challenge)were compared using paired sample T-test. Differences were considered to be significant at p < 0.05.

2.13. Work safety requirements

All work with the SARS-CoV-2 virus was carried out under conditions of Biosafety Level-3 laboratory.

3. Results

3.1. Genetic characterization of SARS-CoV-2 variants

Previously, we obtained D-37, D-ca, D-B4, and D-D2 variants of SARS-CoV-2 by long-term passaging of the Dubrovka strain in Vero cells [26] (Supplementary Table S1, Fig. S1). Their complete genome sequences (GenBank numbers ON380441.1, ON040960.1, ON059701.1, and ON040961.1, respectively) were determined. Genome analysis of the variants revealed a significant number of nucleotide substitutions, most of which were nonsynonymous (Supplementary Table S3, S4). In the genome of the D-37 variant, 7 nucleotide substitutions occurred after prolonged passaging in Vero cells at 37 °C, 5 of which led to amino acid substitutions. Following a cold adaptation in Vero cells, the genome of the D-ca variant collected 17 nucleotide substitutions, 16 of which led to amino acid substitutions. In the genomes of ca clones, D-B4 and D-D2, 16 and 20 nucleotide substitutions were determined resulting in 14 and 17 amino acid substitutions, respectively. The greatest number of nonsynonymous substitutions was localized in the S gene: 2 in the D-37 genome, while 5, 6, and 7 in the D-ca, the D-B4, and the D-D2 genomes, respectively.

3.2. Evaluation of virulence of SARS-CoV-2 variants

In hamsters infected intranasally with ca variants, D-B4 and D-D2, no body weight loss and behavioral changes were observed compared to uninfected animals. However, when infected with the D strain, there was a significant delay in the weight gain at day 2–6p.i. reaching its maximum values on day 4p.i. of 13.4% (P < 0.001) and day 6p.i. of 10,1% (P < 0.01) (Fig. 1 A, B), coupled with other clinical signs including reduced appetite, lethargy, and somnolence.

Fig. 1.

Evaluation of virulence of SARS-CoV-2 D strain and D-B4 and D-D2 variants.

Each animal received 4.0 log10 TCID₅₀ of the virus intranasally. The negative control group (K-) received an equivalent amount of PBS. Weight controls were performed every 2 days (from the 0th to the 4th day - n = 9/group, from the 6th to the 8th day - n = 5/group). Four days after infection, animals were euthanized; lungs, brain, and other animal tissues were homogenized and the viral titer and the concentration of viral RNA was measured (n = 4/group). Bars = SEM. A. Weight of the infected hamsters from day 0 to day 8p.i. B. Weight distribution of hamsters at day 4p.i. C. Concentration of viral RNA in organs of infected hamsters. Limit of detection was 3.0 log10 RNA copies/ml. D. Viral titer in lung and brain of infected hamsters. Limit of detection was 2.0 log10 TCID₅₀/ml. «n.d.» - not detected.

The replication of D-B4 and D-D2 variants in lungs, brain, and other organs of Syrian hamsters on day 4 after infection was significantly lower compared to the D strain (Fig. 1 C, D). The lowest concentration of viral RNA in the lungs was observed upon infection with the D-D2 variant, 6.5 log10 RNA copies/ml, which was 1.6 log10 lower compared to the control group (p = 0.004). The replication of D-B4 and D-D2 variants in the brain decreased more significantly, by 2.2 and 3.2 log10, respectively. In liver, heart and blood of hamsters infected by D-B4 and D-D2 variants viral RNA was not detected. Infectious viral titers in the lungs of animals on the fourth day after infection with D-B4 and D-D2 variants was 1.2 log10 lower than in the control group (p < 0.05). No infectious virus was detected in the brain upon infection with ca variants, whereas upon infection with the D strain it reached a titer of 5.0 log10 TCID₅₀/ml. It should be pointed out, that tissues homogenates were toxic to Vero cells in which titration was performed, hence, reducing the sensitivity limit to 2 log10 TCID₅₀/ml.

On the fourth day after intranasal infection, inflammatory lesions developed in the lungs of Syrian hamsters in all groups (Fig. 2, Fig 3, Fig. 4 ). However, there were significant differences in the nature, severity, and extent of inflammatory damages between the groups. Thus, when infected with the D strain, lobar interstitial pneumonia developed in hamster lungs, accompanied by pronounced inflammatory changes: formation of extensive confluent airless pneumonia foci; desquamation and death of respiratory epithelium; formation of peribronchial and perivascular lymphohistiocytic infiltration; inflammatory infiltration of interalveolar septa circulatory system disorders (marked vascular and microvascular congestion, perivascular edema, interaalveolar and interstitial edema, interaalveolar hemorrhages) (Fig. 2). In all lobes, extensive confluent airless areas and areas with reduced airiness occupying most of the slice area were observed.

Fig. 2.

Morphological characterization of hamster lungs infected with the D strain.

A - confluent foci of pneumonia; B - section of lung parenchyma with reduced airiness located peribronchiolar (black arrow indicates bronchiole, a lumen of which contains single cells of desquamated epithelium, macrophages, epithelial lining preserved, single cells with dystrophic changes; weak inflammatory infiltration of the bronchial wall; red arrow indicates pulmonary artery, diffuse inflammatory infiltrate located around it, perivascular edema; asterisks indicate areas of parenchyma with reduced airiness, thickened interalveolar septa, slotted lumen of alveoli); C - fragment of airless pneumonia focus located peribronchial (arrow - bronchial wall; asterisk - pneumonia focus, in which interalveolar septa and alveolar lumen are not defined). A - ×40, B - ×200, C - ×400. Hematoxylin and eosin staining

Fig 3.

Morphological characteristics of hamster lungs infected with the D-B4 variant.

A, B - arrows indicate foci of alveolitis; C - peribronchial foci of alveolitis: parenchymal airiness is reduced, the alveolar lumen is narrowed, interalveolar septa are thickened due to inflammatory infiltration, microvessel fullness. A, B - ×40, C - ×200. Hematoxylin and eosin staining.

Fig. 4.

Morphological characteristics of hamster lungs infected with the D-D2 variant.

A - in the lower left corner of the microphotograph there are areas of parenchyma with reduced airiness (alveolitis foci); B - alveolitis foci (black arrow) bordered by alveoli without inflammatory changes (red arrow); C - fragment of histological section of a lung lobe of a healthy animal. B, C - ×40, B - ×200. Hematoxylin and eosin staining.

When hamsters were infected with D-B4 and D-D2 variants, a morphological picture of focal interstitial pneumonia was observed in the lungs, the prevalence of which was significantly lower than in the D strain group (Fig 3, Fig. 4). The histological structure of the lungs in the × 10 objective field of view corresponded to that in hamsters of the control group. The lumens of all bronchi and bronchioles were free, containing single desquamated epitheliocytes. Areas of alveolitis occupied a small part of the slice area and contained small airless foci, which were often located around the vessels. In the respiratory tract, there were no signs of circulatory disturbances (edema, interaalveolar hemorrhages) and pronounced damage to interalveolar septa. No pathological changes were found in the lungs of one of the two hamsters infected with the D-D2 variant (Fig. 4 C).

No pathological changes were detected in the lungs of uninfected animals (Fig. 5 ).

Fig. 5.

Morphological characterization of the lung of an uninfected hamster. A – a histological section of a lung lobe: bronchi of different generations, bronchioles, pulmonary arteries veins, acini profiles; lumen of bronchi and bronchioles free, organ parenchyma are airy; B - bronchiole lumen free, epithelial lining preserved, bronchiole wall contains single lymphocytes; alveoli, alveolar passages, and sacs are uniformly airy, interalveolar septa are thin, the interstitium is scanty, vessels and capillaries are moderately full of blood. A - ×40, B - ×200. Hematoxylin and eosin staining.

3.3. Identification of possible markers of SARS-CoV-2 attenuation in vitro.

Since the markers of attenuation of cold-adapted viruses are the ca and ts phenotype [20], we evaluated the presence of these markers in D-B4 and D-D2 variants. D-B4 and D-D2 variants possessed the ca phenotype because they replicated efficiently in Vero cells at 23 °C, reaching a titer of 6.0 to 7.0 log10 TCID₅₀/ml on day 5, whereas the parental D strain and the D-37 variant did not replicate under these conditions (Fig. 6 A).

Fig. 6.

Identification of possible markers of SARS-CoV-2 attenuation in vitro.

Vero cells were infected with the D strain and D-B4 and D-D2 variants. Supernatants were collected daily and viral titers or viral RNA concentration were determined. Limit of detection was 1.0 log10 TCID₅₀/ml and 3.0 log10 RNA copies/ml. Mean values from two independent experiments. Bars show the difference between two individual values. «n.d.» - not detected. A. Replication kinetics of SARS-CoV-2 variants at 23 °C in Vero cells. MOI = 0.001. B. Viral titer in Vero cells at day 3p.i. at different MOI at 39 °C. C. Viral RNA concentration in infected Vero cells at 37 °C. MOI = 0.001. D. Viral RNA concentration in infected Vero cells at 39 °C. MOI = 0.001.

At 37 °C, the replication of D-B4 and D-D2 variants in Vero cells was comparable to that of the D strain (Fig. 6 C). At 39 °C, the D strain and the D-B4 variant replicated efficiently, as indicated by the increase in viral RNA concentration from day 0 to day 3p.i. (Fig. 6 D). The D-D2 variant did not replicate at 39 °C, exhibiting a distinct ts phenotype (Fig. 6 B, D). The D-B4 variant exhibited weakly expressed signs of ts phenotype only at low MOI (0.00001) and temperature 39 °C, the difference in viral titer with the D strain was 3.0 log10 at day three p.i. (Fig. 6 B).

Since both ca variants of the virus, D-B4 and D-D2, had reduced virulence for hamsters regardless of the presence of ts phenotype, we hypothesized that not only ca phenotype and temperature sensitivity are possible markers of virus attenuation. To assess changes in species and tissue specificity, we examined the replication of SARS-CoV-2 variants in Calu-3 human lung cancer cells and Vero monkey kidney cells. D-37 and D-B4 variants did not replicate in Calu-3 cells, whereas D-D2 replicated, but slower compared to the D strain (Fig. 7 A, B).

Fig. 7.

Replication of SARS-CoV-2 variants in Calu-3 cells at 37 °C.

Calu-3 cells were infected with the virus variants at MOI = 0.001. Supernatants were collected daily and viral titers or viral RNA concentration were determined. Limit of detection was 1.0 log10 TCID₅₀/ml and 3.0 log10 RNA copies/ml. Mean values from three independent experiments. Bars = SEM. A. Viral RNA concentration from day 1 to day 4p.i. B. Viral titer at day 3p.i.

3.4. Immunogenicity and efficacy of SARS-CoV-2 variants.

Twenty-one days after a single intranasal immunization of hamsters with the D strain or D-B4 or D-D2 variants, the IgG titers to structural SARS-CoV-2 antigens reached 51200 ± 31353, 8960 ± 3505 or 20480 ± 7011, respectively, in animals sera (Fig. 8 A). The virus-neutralizing activity of hamsters sera after immunization with the D-B4 or D-D2 variants (neutralizing antibodies titer 1012 ± 740 and 1408 ± 701) was comparable to sera of the group of animals infected with the D strain (2304 ± 572) (Fig. 8 B).

Fig. 8.

Humoral immune response 21 days after single immunization of hamsters with SARS-CoV-2 variants.

Hamsters received a dose of 4.0 log10 TCID₅₀ per animal with the D strain or variants D-B4 or D-D2 intranasally. The negative control group received an equivalent amount of PBS. After 21 days, animal blood was collected, and total and neutralizing antibodies to SARS-CoV-2 were determined in sera (n = 5/group; bars = SEM). A. Total IgG titer to SARS-CoV-2 by ELISA. B. Viral neutralizing antibody titer against Wuhan-like D strain. C. Viral neutralizing antibody titer against Podolsk strain (Delta) and Otradnoe strain (Omicron).

Additionally, the virus neutralizing activity of the obtained sera was studied against virus strains related to Delta (Podolsk strain) and Omicron (Otradnoe strain) VOC. The neutralizing activity of hamster sera immunized with the D-D2 variant was reduced by 2.6 times in relation to the Delta variant (p = 0.03) and by more than 60 times in relation to the Omicron variant (p = 0.0014) (Fig. 8 C) compared to Wuhan-like strain D.

Single intranasal immunization with D-B4 or D-D2 variants fully protected hamsters from infection with the D strain as evidenced by the absence of viral replication in lungs and brain (Fig. 9 A, B). Remarkably, the hamsters immunized with the D strain did not develop sterile immunity, as viral RNA and infectious virus titer 3.45 ± 2.9 log10 TCID₅₀/ml were detected in the lungs. In the organs of non-immunized animals, viral RNA was detected with the highest virus load in the lungs, 7.83 ± 7.7 log10 RNA copies/ml and 5.55 ± 5.5 log10 TCID₅₀/ml.

Fig. 9.

Efficacy of intranasal immunization with SARS-CoV-2 variants.

Hamsters immunized with the D strain or D-B4 or D-D2 variants and the negative control group were challenged with the D strain at a dose of 4.0 log10 TCID₅₀ per animal after 21 days p.i. Weight control was performed at days 1 and 4. Four days after challenge, animals were euthanized; lungs and brain were homogenized and viral titer and viral RNA concentration were determined (n = 5/group; bars = SEM). Limit of detection was 2.0 log10 TCID₅₀/ml and 3.0 log10 RNA copies/ml. A. Virus titer in lung. B. Concentration of viral RNA in lungs and brain. C. Weight change of the infected hamsters.

After experimental infection with the D strain, immunized animals gained weight 4 days after infection on average from 5.8 to 7.6 g, while non-immunized animals lost 5.1 g (Fig. 9 C).

On the fourth day after infection, hamster lungs developed inflammatory changes expressed in different degrees. In the control group of unimmunized hamsters, the morphological picture corresponded to lobar viral interstitial pneumonia in the exudation phase, and extensive confluent airless areas and areas with reduced airiness were observed (Supplementary Fig. S3). The lumen of bronchi and bronchioles located in pneumonia foci were predominantly free, some of them contained small layers of the desquamated epithelium. Adjacent parenchyma represented airless fields, in which alveoli lumen was not determined, and interalveolar septa were destroyed due to inflammatory infiltrate, edema, and death of respiratory epithelium. Alveolar capillaries were in a state of acute profundity, dilated, and stasis of erythrocytes was noted in their lumens.

The severity of inflammatory changes in the lungs of immunized hamsters on day four after infection with the D strain was significantly lower than in the control group (Supplementary Fig. S4, S5, S6). The histological structure of the lungs in the majority of the objective field of view (×10) was consistent with the normal structure. However, histological sections of all hamster lung lobes identified small areas of pulmonary parenchyma with inflammatory changes, usually located peribronchial correspondingly to focal alveolitis in the recovery stage. At visual assessment foci of alveolitis in hamsters immunized by D-B4 or D-D2 variants occupied small areas of lung slices (Supplementary Fig. S5, S6).

4. Discussion

Live viral vaccines included in the national immunization schedules made it possible to eliminate smallpox on a global scale and bring such diseases as polio, measles, rubella, and mumps to the brink of eradication [21]. Live attenuated vaccines are similar to natural infectious agents; they induce strong, cross-protective, and long-lasting immune response [21]. High immunological efficacy coupled with relatively simple and low-cost manufacturing of live viral vaccines could potentially legitimate their use [21]. The currently used mRNA vaccines and adenovirus vectored vaccines encode only the spike protein, therefore limiting the immune response against only this viral antigen [1]. However, live attenuated vaccines can induce immunity to several structural and non-structural viral proteins, enhancing the chances of protection [21], [33]. While novel SARS-CoV-2 VOCs evade the immune response produced by approved vaccines [7], [8], [9], [10], [11], [12], [14], [15] development of vaccines against COVID-19 based on live attenuated viruses seems particularly attractive, as they activate all branches of the host immune system: humoral, innate, and cellular [16], [17]. Additionally, intranasal administration of live vaccine may induce of mucosal immunity, stimulate IgA production, which is capable not only neutralize the virus at the entry gate of infection but also limit the spread of the respiratory virus [35], [36], [37], [38], [39].

One of the basic principles of developing live attenuated vaccines is to maintain a balance between attenuating the virulence of the vaccine strain and maintaining the ability to induce a protective immune response. In this regard, the safety and protective activity profile of the SARS-CoV-2 ca variants we obtained is the focus of attention.

In experimentally infected Syrian hamsters, D-B4 and D-D2 variants demonstrated a decrease in virulence according to all parameters analyzed: the animal body weight gain was not slowed down, the concentration of viral RNA and content of infectious virus in organs was significantly lower than when infected with the D strain, the animals retained activity and a good appetite. Histological examination of the lungs of hamsters infected with ca variants revealed relatively reduced focal inflammatory changes, whereas lobar interstitial pneumonia with extensive lesion area developed in hamsters infected with the virulent D strain. According to the previous works, attenuated SARS-CoV and SARS-CoV-2 can slightly replicate in the lungs of model animals and cause inflammatory changes with minimal histopathological damages [17], [22], [34], [40], [41]. Even limited virus replication in the lungs of model animals is certainly undesirable, but this does not exclude the possibility of clinical application of attenuated vaccine. The lowest replication efficiency in vivo was observed in the temperature sensitive D-D2 variant. Furthermore, no infectious virus was detected in the brains of Syrian hamsters infected with the ca variants, whereas when infected with the virulent D strain, the viral titer reached 5.0 log10 TCID₅₀/ml. Given the known neurovirulence of SARS-CoV-2 for humans [42], the reduction in the replication efficiency of ca variants of the virus in the hamster brain when administered intranasally reduces the probability of neurological damages in vivo.

Previously, S.H. Seo et al., S. Okamura et al. and M. Abdoli et al reported cold-adapted and temperature sensitive SARS-CoV-2 strains as vaccine candidates [22], [23], [24]. In the present study, we have shown that for ca variants of SARS-CoV-2, the ts phenotype is not mandatory for virulence reduction. Both non-ts D-B4 variant and ts D-D2 variant exhibited att phenotype for hamsters. Compared to parental strain D and the D-B4 variant, the D-D2 variant lost the ability to replicate at 39 °C (ts phenotype). Non-ts D-B4 lost the ability to infect human lung cells of Calu-3 compared to the D strain and the D-D2 variant. Thus, along with the ca and ts phenotype, the altered species specificity of the virus is a probable phenotypic marker of virus attenuation for Syrian hamsters.

Single intranasal immunization with D-B4 and D-D2 variants induced the production of neutralizing antibodies in Syrian hamsters and protected them from infection with the D strain and consequent development of severe pneumonia. This is consistent with the results of similar studies [22], [23], [24], [34] revealing high protective activity of attenuated SARS-CoV-2 strains by intranasal immunization of Syrian hamsters or hACE-2 (K18-hACE2) transgenic mice.

However, in the present study, the efficacy of immunization with the ca variants has only been studied against infection with the homologous parental D strain closely related to Wuhan-Hu-1 (GenBank number NC_045512.2). The decrease in the neutralizing activity of serums of hamsters immunized with the Wuhan-like D-D2 variant in relation to the heterologous Omicron variant (Fig. 8 C) reflects only the strength of the humoral immunity. The question of whether such immunization can protect against later emerging VOCs, such as Delta (B.1.617.2) and Omicron (B.1.1.529), as well as variants that will inevitably appear in the future, remains open. The potential for the cross-protective activity of live vaccines is determined by the fact that reinfection with SARS-CoV-2 is rare (absolute rate 0%-1.1%) [43], [44]. Pulliam JRC et al. found no evidence of increased risk of reinfection associated with circulating Beta-(B.1.351) or Delta-(B.1.617.2) variants in South Africa, and showed that the risk of SARS-CoV-2 reinfection increased slightly at the peak of fourth wave (period of Omicron variant dominance) [45]. In a retrospective study conducted in Italy (follow-up period to mid-February 2022), 729 cases of reinfection were identified among 119,266 previously infected patients, representing 0.61%. Although the marked increase of the reinfection rates during the Omicron wave is concerning, the risk of secondary severe disease or death remained close to zero [44]. Reinfections have become common since emergence of Omicron and further evolution of this variant [46], [47]. A study conducted in Qatar shows that SARS-CoV-2 infection with the original virus or pre-Omicron variants elicited less than 60% protection against reinfection with Omicron subvariants [48]. According to Chemaitelly H. et al effectiveness of pre-Omicron primary infection against Omicron reinfection was 38.1% (95% CI: 36.3–39.8%) and declined with time since primary infection. The effectiveness of primary infection against severe, critical, or fatal COVID-19 reinfection was 97.3% (95% CI: 94.9–98.6%), irrespective of the variant of primary infection or reinfection, and with no evidence for waning [49]. The extremely low proportion of severe and lethal COVID-19 reinfection cases gives hope that live attenuated COVID-19 vaccines will be able to provide cross-protection against different VOCs.

Sera from hamsters immunized by the D strain and the D-D2 variant (both are Wuhan-like) poorly neutralized Omicron BA.1-like virus in vitro (Fig. 8 C). The results obtained are in good agreement with the ability of Omicron-like viruses to evade the neutralizing antibodies of pre-Omicron COVID-19 convalescents and vaccinated individuals [10], [11], [12], [13], [14]. This indicates a potentially low efficacy of the humoral immune response against the Omicron-like virus when immunizing with D strain derived attenuated virus. Consequently, the development of temperature sensitive Omicron-like strain is a promising research direction that would allow obtaining vaccine strains that effectively induce both cellular and humoral immunity against emerging virus variants.

The obtained ca variants, D-B4 and D-D2, contain a large number of mutations in protein sequences compared with the parental D strain and among themselves (Supplementary Table S4) determining phenotypic differences between the variants. Switching species tropism and/or tissue specificity of the D-B4 and D-D2 variants is consistent with identified amino acid substitutions in the S-protein responsible for binding to host receptors. In addition, recently it has been shown that S-protein can have active and inactive states at different temperatures. In particular, the receptor binding motif (RBM) on the receptor binding domain (RBD) S1 is temperature sensitive and loses its ability to bind to the ACE2 receptor at 40 °C [50]. Consequently, it cannot be excluded that conformational changes of S protein of ca SARS-CoV-2 mutants are associated not only with host switch but also with a cold adaptation of the virus. This is indirectly evidenced by the fact that the D-37 variant, which is unable to replicate at 23 °C in Vero cells and has lost its ability to infect human Calu-3 cells, has only 2 amino acid substitutions in the S protein, whereas the ca D-B4 and D-D2 variants have 6 and 7, respectively. The role of detected amino acid substitutions in non-structural proteins nsp3, nsp4, nsp5, nsp6, nsp14, and nsp16, found in the D-D2 variant, but absent in the non-ts D-B4 variant (Supplementary Table S4), could potentially determine ts phenotype of the D-D2 variant. However, further studies are needed to reliably determine the role of these and other mutations.

It is very difficult to establish infectivity, reactogenicity, immunogenicity (including the absence of the hyper-attenuation effect) and efficacy of the obtained virus variants for humans without conducting studies on higher primates or clinical trials. Thus, the results we obtained on the Syrian hamster model are the first step towards understanding the potential of SARS-CoV-2 cold adaptation in the production of live vaccines against COVID-19.

Currently, the majority of the global adult population has markers for COVID-19 or has been vaccinated. Therefore, a strategy for the safe use of live attenuated vaccines against COVID-19 in adults may involve “booster” immunization of only individuals with markers of new coronavirus infection and/or previously immunized with inactivated vaccine. Such a vaccination schedule is applied in several countries, where the WHO recommendation includes administering at least 1 dose of inactivated polio vaccine followed by a series of immunizations with live oral polio vaccine to avoid the risk of reversion to virulence [51]. When considering the possible clinical application of live attenuated vaccine, it is also important to take into account the low susceptibility of children to SARS-CoV-2 variants phylogenetically close to the Wuhan virus [52] suggesting the avirulence for children of ca attenuated SARS-CoV-2 strains.

4.1. Limitations of research

1. The lack of objective, quantitative histopathology scoring. Histological examination of the lungs was carried out only for qualitative morphological characteristics of pathological changes.

2. Relatively small number of animals per group (4 or 5) used in experiments for evaluation of virulence, immunogenicity and efficacy of cold-adapted virus variants.

5. Conclusions

In the present study, attenuated for Syrian hamsters cold-adapted SARS-CoV-2 variants were obtained. The variants had different temperature sensitivity and provided protective immunity against infection with a virulent strain upon single intranasal immunization. The virulence and immunogenicity data of ca variants of SARS-CoV-2 obtained in an animal model are difficult to directly extrapolate to humans without clinical studies. Preclinical testing in animal models only provides an approximation to understanding the safety of clinical use and human susceptibility to live vaccine strains. However, despite the limitations of the present study noted above, the results revealed a potential for cold-adaptation of SARS-CoV-2 as a strategy for development a live attenuated vaccine against COVID-19.

Ethics statement

All applicable international, national, and/or institutional guidelines for the care and use of animals, including the Guide for the Care and Use of Laboratory Animals [27], were followed. This study was approved by the Medical Ethics Review Committee of the I. I. Mechnikov Research Institute of Vaccines and Sera (Ethics Committee Decision No 2 dated May 24, 2021).

Funding

The study was carried out with the financial support of the Ministry of Science and Higher Education of the Russian Federation (Theme No. FGFS-2022-0004) using equipment of the Collective Usage Center “I. I. Mechnikov NIIVS”, Moscow, Russia (Agreement No. 075-11-2021-676 dated 28.07.2021). Genetic characterization of SARS-CoV-2 variants was performed with the financial support of the Russian Foundation for Basic Research, No. 20–04-60079.

Author contributions

All authors contributed to the study’s conception and design. Experimental work, data collection, artwork and analysis were performed by EF, AG, EK, DS, OP, AP, and RS. Histology was performed by AP, GT and VK. The manuscript was written by EF, YA, and revised by IL, OS, and VZ.

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.

Appendix A. Supplementary data

The following are the Supplementary data to this article:

Data availability

Data will be made available on request.