Introduction
SARS-CoV-2 replicates primarily in the upper respiratory tract, where it is detected at high concentrations before and within the first days after symptoms onset and for several days or even weeks, depending on disease severity and host immune status [
1‐
3]. A mean duration of SARS-CoV-2 RNA shedding of 17 days has been reported, with no live virus detected beyond 9 days [
4]. High viral load is also detected in asymptomatic patients [
5‐
7]. ACE2, the main receptor of SARS-CoV-2, is expressed specifically within the motile cilia of upper airway epithelial cells [
6]; expression gradually decreases when moving to the lower airways [
8,
9], which may explain why viral load is generally higher in the nose than in the oral cavity or lungs [
10]. Severe disease is thought to be due to a migration of the infection to the lungs and a dysregulated innate immune response leading to an excessive inflammatory reactions causing tissue damage and, ultimately, multi-organ failure [
11,
12]. Efficient early control of viral replication in the upper respiratory tract could hamper dissemination to the lower tract and thus effect a more rapid resolution of infection.
Despite viral tropism for the upper respiratory tract, rhinitis is not often associated with COVID-19 [
13], suggesting a limited inflammatory response at this site. The fact that asymptomatic infections are commonly observed also argues against symptoms driven by the local inflammatory response. In contrast, anosmia (usually combined with ageusia) is frequently reported, in particular by patients with mild disease and in the absence of other local symptoms, and can persist after resolution of other symptoms [
13]. The mechanism involved in loss of olfactory functions is largely unknown and may be the consequence of a SARS-CoV-2-specific local inflammatory response [
14].
A better understanding of the determinants that may effectively control the early replication of the virus in the nose may help defining strategies to rapidly reduce viral load and therefore potentially limit transmission and disease complications. As for many respiratory infections, evaluation of the mucosal as well as the systemic response have shown the disconnection between them [
15,
16]. Although, the nature of the excessive inflammatory response associated with severe disease is well described in blood and lungs [
11,
17‐
19], only a few studies have assessed the early local response at the primary site of infection in the upper respiratory tract [
20‐
26]. These studies focus mainly on severe disease underestimating mild disease that represents the majority of infected patients and therefore the main source of transmission. SARS-CoV-2 induces a robust antiviral innate response, characterized by an increase in gene transcripts encoding interferons, interferon-stimulated proteins, and pro-inflammatory cytokines and reduced expression of genes involved in metabolic pathways. To date, neither the kinetics of local innate responses nor their relationship with viral shedding and symptom burden have been described. It is also unclear whether a robust local response may prevent infection or limit transmission. We report a longitudinal analysis of local and blood cytokine responses, viral load, antibody response, and symptom burden in patients with mild COVID-19.
Methods
Study Design, Setting, and Population
This prospective observational cohort study was nested within “Understanding Covid,” an ongoing, single-center prospective observational study conducted at the Geneva University Hospitals (HUG), Switzerland, to examine early virological and early and late immunological responses among SARS-CoV-2-infected adults and children, with inclusion no more than two days after diagnosis, and their household contacts. The study was approved by the Geneva Cantonal Ethics Commission (2020-00516); written informed consent is required. Further details on the Understanding Covid study are available in the online supplement. For this nested study, additional inclusion criteria were a positive RT-PCR test at the visits that were predefined in the protocol, age ≥ 18 years and the ability to undergo a nasal wash (NW). Visits were performed by house call unless patients were hospitalized and took place from inclusion until symptom resolution, with a maximum of 4 visits roughly 7 days apart. Symptoms were reported by patients at time of inclusion and at each visit. The following symptoms were solicited: fatigue, fever, subjective fever, chills, headache, myalgia (systemic); difficulty breathing, cough, sore throat (respiratory); anosmia and ageusia. Additionally, six healthy controls without symptoms and with negative SARS-CoV-2 PCR and serology testing were included for one-time NW and serum collection.
Nasal Wash
Nasal wash was performed according to the institutional protocol [
27,
28] and adapted from previous publications [
29,
30]. Briefly, 3 ml of NaCl 0.9% are injected in the nose for NW and regurgitated. 2 ml are then retrieved and mixed with 1 ml of DMEM with 2 × protease inhibitor (Thermo scientific, ref 78437) for protein preservation. A concomitant oropharyngeal swab was also collected as part of the main study.
SARS-CoV-2 RNA Quantification
Nucleic acids from NW and oropharyngeal samples were extracted individually from 400 µL of each specimen, spiked with 20 µL of standardized canine distemper virus as internal control [
31], using the NucliSens eMAG extraction (BioMérieux, France) kit, according to the manufacturer’s instructions, and eluted in 50 µL. Quantitative real-time polymerase chain reaction (qRT-PCR) assays were performed using the one-step Eurogentec RT-PCR Kit (Qiagen, Hombrechtikon, Switzerland) in a QuantStudio 5 instrument (Applied Biosystems) and SARS-CoV-2 RNA was quantified with the Charité qRT-PCR protocol [
32] using in vitro transcribed RNA for quantification [
33] and reported as number of copies/ml of NW collected.
Cytokine Measurement
A Luminex assay (Magnetic Luminex assay, R&D Systems) was used to calculate the concentration of 23 markers in cryopreserved plasma and NW. For plasma, assays were performed according to the supplier’s instructions. For NW, a last incubation step of 20 min with PFA 4% (followed by 4 washes) was added before reading to inactivate any remaining live virus. Briefly, beads conjugated to the analyte-specific capture antibodies, samples, controls, and standards were incubated at room temperature for 2 h. Biotinylated detector antibodies and R-phycoerythrin-conjugated streptavidin were subsequently added. The mean fluorescence intensity of each marker was read on the Bio-Plex 200 array reader (Bio-Rad Laboratories) using the Luminex xMAP Technology (Luminex Corporation). Five-parameter logistic regression curve (Bio-Plex Manager 6.0) was used to calculate sample concentrations. List of markers tested: CD40L, GM-CSF, Granzyme B, IFN-alpha, IFN-gamma, IL-1-beta, IL-1-alpha, IL-1Ra, IL-2, IL-4, IL-6, IL-8, IL-10, IL-12p70, IL-13, IL-17A, IL-33, CXCL10, CCL2, CCL3, CCL4, PD-L1, and TNF-alpha. Samples whose measurement was below the detection limit were assigned a value corresponding to 50% of the last standard dilution value.
Antibody Characterization by ELISA
IgG and IgA antibodies directed against the S1 domain of the spike protein of SARS-CoV-2 were determined using a commercially available kit (Euroimmun AG, Lübeck, Germany, #EI 2606-9601 G and EI 2606-9601 A) according to the manufacturer's instructions. All nasal wash samples were tested at 1:2 dilutions, while for serum only samples outside linear range were diluted 1:5 or 1:10. In serum, S1-specific IgG and IgA were positive when > 0.8 OD ratio [
34], while for nasal wash S1-specific IgA was positive when > 0.4 OD
450 (highest value of the control group). Total IgA measurements in nasal wash were performed using a Thermo Fisher IgA human ELISA kit (pre-coated) and all samples were tested at 1:100 dilutions.
Statistical Analysis
Categorical variables were described by counts and percentages. Continuous variables were expressed as median and interquartile range (IQR). Correlation analyses were performed using the Pearson test. Cytokine data were categorized by defined time interval post onset of symptoms and differences in intervals between patients and healthy controls were calculated using the Mann–Whitney test. Principal components analysis was performed in R version 3.6.2 using the prcomp function of the stats package. Data were standardized by the scale function in R. The first two principal components were retained since the first two eigen values were higher or close to 1. The first two components explained 75.45% and 12.11% of the variance, respectively. A bootstrap procedure was used to check the robustness of each retained principal component. Fifty-thousand re-sampling with replacement were done. For each re-sampling, the same PCA was conducted. The frequency of the number of retained components (eigen value > 1) over the 50,000 re-sampling was 100% for PC1 and 49.19% for PC2.
Discussion
Cytokines were found locally in the upper respiratory during mild SARS-CoV-2 infection. Local cytokine responses persisted for up to two weeks after symptom onset; they correlated with viral load, irrespective of the time since symptom onset, but not with blood cytokine levels or specific symptoms. Our results are consistent with the transcriptional responses measured in nasopharyngeal specimens collected during SARS-CoV-2 screening, which show a positive correlation between gene expression and viral load [
20,
22,
23,
25]. The presence of high levels of CXCL10 at the transcription and protein level, together with the detection of an interferon-stimulated gene signature [
22,
23], definitively show that the upper respiratory tract is an important antiviral immune site during mild disease. Although an early IFN signal in the nose has been reported at the transcriptional level [
21,
25,
26], we did not find a detectable increase in IFN-alpha or IFN-gamma proteins in NW, even at early time points after infection. This could be due to the low sensitivity of the assay used, a dilution effect due to the nasal wash procedure or evidence for active inhibition of interferon release by SARS-CoV-2 [
37], as suggested by the relatively low concentration of interferons detected in serum of COVID-19 patients [
38] or in vitro [
39]. In addition, in silico analysis of the transcriptome led to the suggestion that macrophages, neutrophils, and activated dendritic cells are recruited in the nasopharyngeal epithelium of infected patients [
23]. The fact that we detected more IL-8 and CCL2, as well as GM-CSF in NW support this finding but a more detailed analysis of the recruited cells in the compartment is needed, especially compared with the lungs [
11,
40].
We observed no correlation between NW and serum cytokine concentrations. This discordance is well known for respiratory viral infections [
15,
16] and underscores the likelihood that other sites such as the oral cavity also contribute to the overall host response [
41]. It has the potential for misleading conclusions when assessments of SARS-CoV-2 immunopathology focus exclusively on the blood compartment, underestimating the role of the mucosal immune system especially early in infection. Serum cytokines therefore do not fully reflect viral responses at primary sites of infection.
In contrast to reports assessing single time points, this study provides a longitudinal analysis of the immune response in the upper respiratory tract in the course of mild disease, which represents the vast majority of cases worldwide. We show that local cytokine production is sustained in most patients until 2–3 weeks after disease onset, even when symptoms have subsided. Infectious particles are generally detected within the first week of infection, but viral RNA may be detected much later. Immune detection of viral RNA itself may thus be the main trigger of innate response in the epithelium, as previously shown for SARS-CoV [
42]. Nonetheless, although our sample size remains limited, this sustained response does not lead to specific symptoms in the upper respiratory tract, but could rather be important in the induction of SARS-CoV-2-specific mucosal immunity during mild disease. We could not link the presence of high levels of CXCL10 or IL-6, indicative of a robust antiviral response, to loss of olfactory function, which may therefore rely on other mechanisms [
43,
44].
The intensity of the local inflammatory response after mild SARS-CoV-2 infection seems overall less pronounced than that reported for other viral respiratory infections such as influenza [
22,
29], respiratory syncytial virus, or rhinoviruses [
45‐
47]. In addition to potential inhibition of the interferon pathway, an impairment of inflammasome/IL-1 beta induction as compared to other viruses was also a proposed mechanism [
22]. In line with these results, we did not detect a significant change in IL-1Ra or IL-1-beta in NW. This relatively moderate antiviral response to SARS-CoV-2 may explain the high viral load in the upper respiratory tract and higher transmission rates.
We observed that only some patients had S1-specific IgA in the nasal cavity while almost none had IgG, the magnitude of the response was lower than the one measured in serum. These heterogeneity of the S1-specific IgA levels at the mucosa has been recently reported by others [
48‐
50]. IgA is important for early neutralization of SARS-CoV-2 in blood and saliva [
41]. Whether these specific local IgA plays a role in preventing severe disease early during infection requires further studies.
Our study had limitations. The size of the cohort is small, mainly due to the limited access to infected patients at the end of the first wave of COVID-19 in Geneva and practical constraints linked to performing the procedure during home visits at multiple time points. Patients had mild disease so our data should not be extrapolated to more severe disease. Enrolled participants were already RT-PCR positive at inclusion, at a median of 5 DPOS, so we could not capture early antiviral response during incubation time before symptoms onset. We could not reliably isolate RNA or phenotype the cells present in the NW samples due to the low number of cells recovered. The relatively low sensitivity of the assay used to measure cytokine and the dilution of the samples due to the nasal wash procedure may have limited our ability to detect cytokine production locally. Lack of correlation with symptoms should be interpreted with caution as symptoms were mainly self-reported and could therefore be subjective, although we believe that lack of correlation with anosmia remains solid despite limited sample size.
Acknowledgements
The authors would like to thank Irene Sabater Royo, Romain Burquier, and Adriana Renzoni for technical support, Adrien Calame, Lena Mazza, Cecilia Schweblin, Mathilde Bellon, and Diego Andrey for recruitment support, volunteers of the “Understanding COVID-19 study,” and those providing NW in the healthy control group.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.