SARS-CoV-2 can induce a life-threatening systemic inflammatory state by triggering various cytokines, chemokines, and hyperactive immune cells. Our study revealed that the N protein, rather than the S protein, of SARS-CoV-2 can trigger lung epithelial A549 cells to express various cytokines, namely, IP-10, RANTES, IL-16, MIP-1α, basic FGF, eotaxin, IL-15, PDGF-BB, TRAIL, VEGF-A, and IL-5. These may initiate subsequent cytokine cascades and promote systemic inflammatory syndromes. We also found that patients with COVID-19 showed significantly high levels of serum CTACK, basic FGF, GRO-α, IL-1α, IL-1RA, IL-2Rα, IL-9, IL-15, IL-16, IL-18, IP-10, M-CSF, MIF, MIG, RANTES, SCGF-β, SDF-1α, TNF-α, TNF-β, VEGF, PDGF-BB, TRAIL, β-NGF, eotaxin, GM-CSF, IFN-α2, INF-γ, and MCP-1. High serum IP-10 and M-CSF levels accompanied by low IL-16, TNF-β, and PDGF-BB levels may be associated with the development of pneumonia. In contrast, high serum IP-10 levels may be associated with rapid virus elimination in patients with COVID-19.

In our study, the lung epithelial A549 cells displayed a greater immune response to N-Bac compared with that of monocytic THP-1 cells. This finding suggests that cytokine cascades initiate in the lung epithelial cells and not in the circulating monocytes of patients with COVID-19, which may explain the occurrence of severe COVID-19 in patients with pneumonia and milder forms in those without pneumonia. IP-10, which recruits activated T helper 1 (Th1) cells for host defense against lung intracellular pathogens, and IL-16, a lymphocyte chemoattractant factor, are released by both bronchial and alveolar epithelial cells [14, 15]. The blockade of eotaxin and IL-16 causes 70% inhibition of eosinophil chemotactic activity to the lung in pulmonary disorders [16]. Collectively, these studies suggest that bronchial and alveolar epithelial cells secrete cytokines, such as IP-10, IL-16, or eotaxin, leading to subsequent pulmonary disorders, consistent with our in vitro experiments.

Stimulation of THP-1 cells with S-bac or N-bac increased the expression levels of CD-86 and IL-6 in THP-1 cells, while those of CD-204 and TGF-β remained unchanged, suggesting that both S-Bac and N-Bac stimulations induce THP-1 cells toward M1 polarization. Most COVID-19 studies have focused on the role of the S protein of SARS-CoV-2 in COVID-19 pathoetiology. Previous in silico studies predicted that cell surface Toll-like receptors (TLRs), especially Toll-like receptor 4 (TLR4), are likely to be involved in recognizing molecular patterns, probably SARS-CoV-2 S protein, to induce inflammatory responses [17, 18]. Another study found that the S protein can activate TLR4 and induce IL-1β production in THP-1 cells [19]. These studies only suggest that the S protein may induce IL-1β production by activating TLR4 in macrophages. Whether S protein-induced IL-1β production represents the large proportion of whole IL-1β production in macrophages and IL-1β production affects cytokine activation in COVID-19 patients remains unknown. Moreover, the SARS-CoV-2 S protein can only prime inflammasome formation and release of mature IL-1β in macrophages derived from patients with COVID-19 but not in macrophages from healthy SARS-CoV-2 naive individuals [20]. Therefore, IL-1β may not be a key cytokine in the pathoetiology of patients first time contracting COVID-19. The N protein interacts directly with the Nucleotide-binding oligomerization domain leucine-rich repeat, and pyrin domain containing 3 (NLRP3) inflammasome can promote IL-1β and IL-6 activation and cause subsequent lung injury in mouse models [21]. Furthermore, the N protein can promote the activation of Nuclear factor κB (NF-κB) signaling by enhancing the association between TGF-beta-activated kinase 1 and IκB kinase complex [22] and function as a pathogen-associated molecular pattern to directly bind to TLR2 and activate NF-κB and mitogen-activated protein kinase signaling in endothelial cells [23]. Therefore, the N protein can induce pro-inflammatory cytokines through promoting the activation of NF-κB signaling and NLRP3 inflammasome. However, these studies could not determine which protein is the important virulence factor in cytokine activation and subsequent lung injury. In contrary, our study revealed that N protein, rather than S protein, can trigger A549 cells to express considerably high levels of various cytokines to promote a cytokine storm. Moreover, S protein is a leading target antigen in the development of COVID-19 vaccine; however, nonsynonymous mutations developed in the S protein could create SARS-CoV-2 variants as the epidemic progressed [24, 25]. These variants reduced the effectiveness of current S protein recombinant vaccines and contributed to the continuation of the COVID-19 pandemic. In contrast, the N gene is more conserved and stable, with 90% amino acid homology and fewer mutations over time [26,27,28,29], hence is an appropriate target in the development of new generation medicine or COVID-19 vaccines. Due to the conserved nature of the N protein, despite the continuous emergence of new variants of SARS-CoV-2, the results of our in vitro study may still be applicable to other strains.

Taiwan has been able to contain the pandemic at the time of study. Most of our enrolled patients were classified as imported cases (n = 25; 92.59%) and had gone abroad for education or tourism; most of them had relatively mild illnesses. Besides, in Taiwan at that time, all confirmed COVID-19 patients could only stop quarantine after three negative COVID-19 tests. Due to the long communicability period of COVID-19 and the relatively mild disease severity of confirmed patients in our hospital, all our patients recovered before showing three negative COVID-19 tests. Therefore, the length of quarantine in our study corresponded to the time needed for virus eradication rather than disease severity.

Patients with pneumonia expressed a considerably higher titer of anti-SARS-CoV-2 S-RBD IgG than those without. In previous studies, anti-S IgM and IgG titers remarkably correlated with the viral load and disease severity in patients with COVID-19 [30, 31]. The anti-S antibody response developed considerably faster with higher titers in patients who eventually died of SARS [32]. The anti-S IgG antibody-activated inflammatory macrophages and cytokines, such as MCP-1 and IL-8, cause severe lung injury in SARS-CoV-2-infected macaques [33]. Collectively, these studies suggest that increased anti-S IgG production may correlate with a robust inflammatory response and cause severe pulmonary injury in SARS-CoV-2 infection. Furthermore, patients with shorter hospitalization duration in our study had significantly higher anti-N IgG levels than the remaining patients. Previous studies found that SARS-CoV-2 N protein was highly immunogenic [34, 35], indicating that anti-N IgG may play a role in eliminating SARS-CoV-2 in patients with COVID-19.

IP-10 may play an important role in initiating cytokine cascades and final cytokine expression in patients with COVID-19. The MIG and IP-10/CXCR3 axis plays a crucial role in recruiting various immune cells, including T lymphocytes, natural killer cells, and macrophages, to damaged or inflamed tissues [36, 37]. The immune cell population in the lungs of patients with COVID-19 comprises a considerable proportion of T cells and monocytes, compared with patients with primary pneumonia infection without COVID-19 [38]. While T cells and monocytes are relatively rare in healthy lungs [39], their accumulation is assumed to be recruited by locally produced chemoattractant proteins, such as IP-10 and MIG. Besides, higher serum IP-10 levels are associated with a higher risk of severe Mycoplasma pneumoniae pneumonia in children [40] and higher risk of death in patients with ARDS [41]. These findings suggest that IP-10 is prone to development of pneumonia and tissue damage during inflammation. Moreover, in a neuroadapted John Howard Mueller strain of mouse hepatitis virus, IP-10 is responsible for viral suppression after central nervous system inoculation [42]. Collectively, increased expression of IP-10 from lung epithelial cells recruited immune cells, including T lymphocytes, thereby exacerbating immune reaction and organ damage, causing severe pneumonia, and eliminating viral loads, resulting in a shorter time of viral shedding.

In our study, high serum M-CSF levels were associated with the development of pneumonia in patients with COVID-19. M-CSF is a necessary growth factor for recruiting and expanding lung monocytes. It also leads to the transition of monocytes to macrophages during infection [43, 44] and contributes to tissue repair during inflammation [44]. Macrophages differentiated in the presence of M-CSF, adenosine, and PGE2 induced the downregulation of inflammatory mediators and upregulation of growth factors [44]. In contrast to IP-10, M-CSF levels were elevated in COVID-19 pneumonia patients, which reduced tissue damage during inflammation.

IL-1RA is a receptor for pro-inflammatory cytokines, specific to the activity of both IL-1α and IL-1β [45]. The blockade of IL-1 in patients with COVID-19 considerably improved survival and shortened hospital stay [46]. However, treatment with canakinumab, an IL-1β receptor inhibitor, did not reduce the need for intermittent mandatory ventilation or mortality of patients with COVID-19 [47]. These findings suggest that IL-1RA and IL-1α, rather than IL-1β, may play a role in SARS-CoV-2 infection, consistent with our findings.

The expression of vascular endothelial growth factor (VEGF) was also increased in our study patients and lung epithelial A549 cells triggered by SARS-CoV-2 N protein. In a previous clinical trial involving 26 patients with severe COVID-19, bevacizumab, an anti-VEGF neutralizing antibody, plus standard care improved the PaO2/FiO2 ratio after 24 h. By day 28, 92% patients demonstrate oxygen-support improvement, 65% patients were discharged, and none show worsen oxygen support or death [48, 49]. Collectively, these data suggest that VEGF-induced vascular changes, including angiogenesis, alteration of vascular permeability, and inflammation, may cause life-threatening defects in patients with severe COVID-19.

Dr. Yang et al. had reported that the expression levels of IP-10, MCP-3, HGF, MIG, and MIP-1α are significantly higher in critically ill patients, followed by severe and then the moderate patients [50]. The study conducted by Dr. Yang focused on the cytokines those may relate to the development of severe or critical diseases. Among the patients they enrolled, 22% were critical illness patients, 50% were severe disease patients, and 28% were moderated disease patients. No patients with mild disease were enrolled. Although serum levels of IP-10, MCP-3, HGF, MIG, and MIP-1α were all higher in critical and severe disease patients than healthy control, the serum levels of MCP-3, HGF, MIG, and MIP-1α were no difference between patients with moderate disease and healthy control. These were indicated that these cytokines were markers to predict severe and critical disease progression, but not moderate disease. On the contrary, patients enrolled in our study were relatively non-illness. Among them, 3.7% were critical patients, 14.8% were severe disease patients, 44.4% were moderate disease patients, and 37% were mild disease patients. Considering the substantial influence of lung epithelial cells on the initiation of cytokine cascades in our in vitro study, as well as the impact of pneumonia on the clinical outcomes of COVID-19 patients, the clinical segment of our study is crafted to investigate the correlation between cytokine expression and the development of pneumonia. Our findings revealed that serum levels of IP-10 and M-CSF were significantly elevated in patients with pneumonia (critical, severe, and moderate cases) compared with those without (mild cases).

Several cell membrane-based biomaterials derived from various types of cells have been developed. These membrane-based biomaterials, rich in biologically active proteins and phospholipids, are designed to treat inflammation, tumors, or autoimmune diseases by regulating immune cell function, exerting enzyme-like activity, or neutralizing cytokines [51, 52]. Applying this cell membrane-based biomaterials platform to the potential therapeutic targets found in our study, such as lung epithelial cells, N proteins, IP-10, or other cytokines, will help develop new treatment strategies to prevent the progression of severe COVID-19.

This study has some limitations. First, we used A549 human lung carcinoma epithelial cells, not primary lung epithelial cells to evaluate the cytokine response triggered by SARS-CoV-2 proteins. Primary cells are the gold standard for studying cell behavior in vitro. However, the utilization of primary cells may face obstacles, such as challenges associated with in vitro isolation and cultivation, and loss of phenotype over extended periods in culture. Human primary lung epithelial cells lose their phenotype and capacity over a period of 1–2 weeks when cultured in vitro [53, 54]. Cell lines are generally easier to cultivate compared with primary cells, exhibit a rapid proliferation rate and extended lifespan, and retain their phenotype when maintained in culture. Therefore, the human lung adenocarcinoma cell line A549 is extensively used in lung cell biology. However, despite exhibiting similar responses to viral infection compared to primary cells, A549 cells show limited cytokine response [55]. The suitability of using A549 cells in COVID-19 studies is also a matter of discussion due to the low levels of angiotensin-converting enzyme 2 (ACE2) expression [56]. However, although the A549 cells is less sensitive to SARS-CoV-2 infection, the expression of ACE2 in A549 cells has been well documented [56]. Indeed, several studies have successfully utilized A549 cells to assess the impact of SARS-CoV-2 infection [56,57,58]. Moreover, to gain better understanding of the cytokine storm and chronic autoimmune symptoms caused by SARS-CoV-2 infection, Dr. Wang et al. identified autoantigens from A549 cells that are strongly tied to diverse immune symptoms of COVID-19 [58]. By comparing the autoantigens they discovered in A549 cells with previously collected proteomic and transcriptomic data related to SARS-CoV-2 infection found in the Coronascape database, they found that out of the 348 autoantigen proteins identified in A549 cells, 291 of them (83.6%) had previously been documented as having changes in cells or patient tissues during SARS-CoV-2 infection in earlier scientific literatures. This finding suggests that, despite being less sensitive to SARS-CoV-2 infection when compared to primary lung epithelial cells, A549 cells are still appropriate for evaluating how cells respond to SARS-CoV-2 infection. Second, our study is a cross-sectional design study. Correlations between the expression of some cytokines and clinical characteristics of COVID-19 patients were found. However, correlation does not necessarily imply causation. Many confounding factors such as the rapidity of diagnosis, recognition of disease progression, and secondary complications both directly from COVID-19 and indirectly from its treatment might contribute to the observed associations. Further longitudinal studies remain warranted to provide stronger evidence regarding the relationship between cytokine levels and disease progression. Third, although we found that the N protein may trigger cytokine release in COVID-19 patients, the underlying molecular mechanisms driving this response were not explored; therefore, further investigations remain warranted.

In conclusion, our study suggested that the N protein of SARS-CoV-2 can preferentially trigger lung epithelial cells over macrophages to express IP-10 and other pro-inflammatory cytokines, thereby initiating cytokine cascades. In patients with COVID-19 and pneumonia, IP-10 may play a role in inflammation and virus elimination. Further studies are warranted to validate our findings.