The two body donors of fatal cases with COVID–19 were a 53 years old female and a 62 years old male, respectively. Both patients had progressively decreased lymphocytes with elevated serum IL–6 and C reactive protein (CRP) levels in the late stage of disease (Supplementary Table 1), which was consistent with recent report1,2. The gross anatomy of the lung showed moderate bilateral pleural effusion and pleural adhesion in the two patients. The hepatization of lung tissues was observed on the cut-surface of the collapsed and consolidated lungs. The microscopic manifestation of the lung injury was consistent with diffuse alveolar damage (DAD). Alveolar cavities were filled with a large number of macrophages with scattered neutrophils and lymphocytes (Fig. 1a). The massive serous (Fig. 1b) and fibrinoid exudate in the alveolar spaces were shown by the Masson staining (Fig. 1c, d). The acidic mucopolysaccharides from a large amount of mucinous secretion were observed by the Alcian blue-periodic acid-Schiff (AB-PAS) staining in the bronchi and bronchioles, terminal bronchioles and pulmonary alveoli (Fig. 1e, f). A lot of mucus in the distal respiratory tract lined by mucous cells were shown, reminiscent of the morphology of mucoid adenocarcinoma (Fig. 1g). The peribronchiolar metaplasia (PBM) with interstitial fibrous hyperplasia but without invasive growth of atypical cells was observed. The mucous plug with fibrinous exudate in the alveoli and terminal bronchioles formed the cribriform pattern. The bronchial phlegm combined with epithelial detachment and fibrinoid exudate was visible (Fig. 1h).
The hyaline membranes and widened alveolar walls with collagen fibers proliferation and lymphocyte infiltration were observed in alveoli occasionally (Extended Data Fig. 1a, b). Focal or patchy hemorrhage with fibrinous exudate were seen in the alveolar cavities and interstitial tissues (Extended Data Fig. 1c, d). The broken alveolar walls flushed by huge hemorrhagic effusion formed the “blood lake”. The endothelial cells of small pulmonary arteries were swollen and shed (Extended Data Fig. 1e). Mixed thrombi were present in small veins (Extended Data Fig. 1f).
Intensive loss of bronchiole and alveolar epithelial cells was remarkable (Fig. 2a, b) while abundant swollen and degenerated alveolar cells desquamated in the alveoli (Fig. 2c, d). Patchy type II pneumocytes proliferated with atypical changes, including enlarged nuclei, clearing of nuclear chromatin, prominent nucleoli and suspected viral inclusions (Fig. 2e, f). The notable proliferation of type Ⅱ alveolar epithelial cells resembled the morphological changes of atypical adenomatous hyperplasia, in situ adenocarcinomas, or even invasive adenocarcinoma. Thickened alveolar walls and widened interstitial tissues were accompanied by lymphocyte infiltration and fibroblast proliferation (Fig. 2g, h).
Notably, the alveolar macrophages significantly increased and filled in a part of the alveolar cavities with scattered neutrophils and lymphocytes. CD68, one of the scavenger receptors, is a well-documented specific surface marker of macrophages4. The alveolar macrophages were presented in diverse forms, including aggregation in small clusters (Fig. 3a, b), diffused distribution (Fig. 3c), single macrophage exhibiting intracytoplasmic phagocytosis, spherical acidophilic hyaline bodies or hemophagocytic phenomenon (Fig. 3d, e), and multinucleated giant cells (Fig. 3f). Furthermore, using immunohistochemistry approach, we examined several chemokine and inflammatory cytokines secreted by alveolar macrophages including IL–6, IL-10 and TNFα with specific antibodies. IL–6 and TNFα were moderately expressed in macrophages (Fig. 3g, i), while the expression of IL–10 was strong (Fig. 3h). Besides, extensive and strong expression of Programmed Death-Ligand 1 (PD-L1) was observed (Fig. 3j).
In general, the degree of infiltration of lymphocytes into the pulmonary tissues was much inferior to that of macrophages, although some focal lymphocyte infiltrations were present in lungs (Extended Data Fig.2a). CD20-positive B lymphocytes (Extended Data Fig.2b) accounted for a large majority of the lymphocytes whereas a few CD3-positive T lymphocytes (Extended Data Fig.2c) made up a small proportion including CD4-strong positive for helper T cells (Extended Data Fig.2d) and CD8-positive for cytotoxic T cells (Extended Data Fig.2e). Among the inflammatory infiltrating cells, no CD56-positive NK/T cells (Extended Data Fig.2f) were detected. Neither Programed cell death protein–1 (PD–1) nor PD-L1 proteins were shown on the surface of lymphocytes (Extended Data Fig.2g, h). None of the lymphocytes were proven virus-infected by using immunohistochemistry of Rp3-NP specific antibodies (Extended Data Fig.2i).
We carefully examined the heart and kidney in the two donor bodies. No obvious gross abnormalities were observed. Nevertheless, microscopical abnormalities were observable in both organs. Multifocal myocardial degeneration was present in the heart, together with myocardial atrophy and interstitial fibrous tissue hyperplasia (Extended Data Fig.3a). A few CD20-positive B cells and CD3-positive T cells were scattered (Extended Data Fig. 3b, c). In the kidneys, normal renal structures were retained. However, the fibrotic glomeruli and edematous tubular epitheliums (Extended Data Fig.3d) were focally present with a small amount of infiltrating B (Extended Data Fig.3e) and T lymphocytes (Extended Data Fig.3f). It is worth noting that no viral particles were found in parenchymal cells in both heart and kidney.
Next, we examined lymph nodes and other lymphoid organs. Notably, lymph nodes in the pulmonary hilum were swollen whereas splenic volume was slightly reduced with shrunken capsule in the two cases. Morphological changes of the pulmonary hilum lymph nodes were characterized by an obvious dilation of cortical sinuses with numerous macrophages (Extended Data Fig.4e). In the spleen, the lymphocytes in the white pulp were slightly reduced with infiltration of macrophages in the red pulp (Extended Data Fig.4g). Notably, the hyperplastic type II alveolar epithelial cells, alveolar macrophages, macrophages in the pulmonary hilum lymph nodes and spleen were all infected by SARS-CoV–2 whereas no obvious viral infection was found in lymphocytes and mesenchymal cells (Extended Data Fig.4a-h).
An important finding in the present work was the infections of gastrointestinal mucosa cells (Extended Data Fig.4i) and spermatogenic testicular cells (Extended Data Fig.4k) by SARS-CoV–2 without obvious histological abnormalities. In addition, the intestinal epithelium cells, submucosa ganglion cells, spermatogenic Sertoli and Leydig cells were all infected by SARS-CoV–2 (Extended Data Fig.4j, l). Scrutiny of pathological sections of esophagus, breasts, muscles, stomach, thyroid, bladder and adrenal glands showed no obvious abnormalities or SARS-CoV–2 infection.
The pathological investigations of severe patients are pivotal for the understanding of pathogenesis of COVID–19 and assessment of clinical treatments. Lungs are the main damaged organ in severe COVID–19 patients due to the ARDS, similar to the situation in SARS outbreak of 2003. In this regard, the common features of lung injuries constituting the main pathological abnormalities somehow mimicked those in SARS, including: (1) extensive impairment of type I alveolar epithelial cells and atypical hyperplasia of type II alveolar cells; (2) formation of hyaline membrane, focal hemorrhage, exudation and pulmonary edema; (3) pulmonary consolidation with infiltration of macrophages, lymphocytes as well plasma cells; (4) endothelial injury and thrombosis in small vessels and microvascular. Thus, like SARS-CoV, SARS-CoV–2 was capable of triggering the pathogenesis and resulting in severe dysfunction of ventilation and gas exchange obstruction in patients5–9.
However, the pathology of lungs with SARS-CoV–2 infection also exhibited some distinct features as compared to that found in SARS patients. The hyaline membranes in alveoli, which constituted major anatomical abnormalities leading to gas exchange obstruction in SARS, were uncommon in COVID–19. On the other hand, we observed mucous plugs in all respiratory tracts, terminal bronchioles and pulmonary alveoli in COVID–19, and this was neither described in SARS5,7–11 nor in the recently reported autopsy studies on COVID–19 patients12,13. Another unique feature of COVID–19 was the excessive mucus secretion with serous and fibrinous exudation, which could aggravate the dysfunction of ventilation. These findings suggested the existence of different pathogenic mechanisms responsible for the hypoxemia between COVID–19 and SARS patients. We found the hyperplasia and peribronchiolar metaplasia of mucosal epithelium, a phenomenon which might result from the inflammation- induced pulmonary tissue reparatory processes or even proliferative reaction of cells originated from bronchioles and terminal bronchioles. We assume that the mucus aggregation in distal respiratory tracts by peribronchiolar metaplasia of mucosal epithelium as a result of inflammation-induced reparatory changes should play a part in the sputum suction failure in very severe COVID–19 patients as previously reported12.
Of particular note, we found the alveolar macrophages with SARS-CoV–2 infection were expressing ACE2, a well-established receptor for both SARS-CoV and SARS-CoV–2 (Extended Data Fig.5). It was reported that SARS-CoV could occasionally be identified in the alveolar macrophages9. In COVID–19 patients, the extraordinary aggregation and activation of these macrophages could occupy a central position in pathogenesis of the very severe “inflammatory factor storm” or “cytokine storm”. Therefore, the spectacular infiltration and activation of alveolar macrophages in COVID–19, especially among patients with severe and critical stages of ARDS, might represent the shift of classically activated phenotype (M1) to alternatively activated phenotype (M2) of alveolar macrophages, whereas this shifted property of alveolar macrophages could contribute to the inflammatory injuries and fibrosis of respiratory tracts14. To further address the issue of accumulation of macrophages in lung tissues and to explore the potential function of macrophages in response to SARS-CoV–2, we incubated purified and Fc-tagged spike proteins (S protein), which contains the receptor binding domain (RBD) responsible for the entry of SARS-CoV–2 into the host cells15, with white blood cell samples from six healthy donors. The possible location of the S protein on the surface of these white blood cells was examined by flow cytometry analysis. To our surprise, the S protein interacted with CD68-expression monocytes/macrophages but not with T or B lymphocytes, suggesting a direct viral infection of the macrophage/monocytes. We then determined the expression of ACE2 on the surface of macrophages. Indeed, an expression pattern similar to the binding of S protein by monocytes/macrophages was observed (Fig. 4b). These findings highlighted the role of macrophages as direct host cells of SARS-CoV–2 and potential drivers of “cytokine storm syndrome” in COVID–19.
Additionally, an elevated serum IL–6 was observed in the two cases in this study and also in some other very recent reports16. These features were similar to the pathogenesis of “cytokine storm syndrome” in patients with hemophagocytic lymphohistocytosis (HLH) or macrophage activation syndrome (MAS)17,18. The blockage of cytokine storm using anti-IL–6 or IL–6R antibody, such as Tocilizumab, has promising therapeutic effects and clinical practice in the treatment of MAS or HLH. Therefore, our data are in support of the beneficial use of anti-IL- 6/IL–6R antibody for the inhibition of alveolar macrophage activation as well as inflammatory injuries in COVID–19 patients. Recently, the Tocilizumab therapy has been recommended in the Guideline of Diagnosis and Treatment of COVID–19 (version 7) by the National Health Commission.
The fact that the known ACE2-exressing cells19–21, including type II alveolar epithelial cells, alveolar macrophages, intestinal epithelial cells and spermatogenic cells, were all found infected by SARS-CoV–2 infection suggests the necessarily of clinical tests of SARS-CoV–2 in feces samples and the blockade of possible fecal-oral transmission22. Infected submucosa ganglion cells in small intestine were never reported before. Whether it could be the host cells for long-term coexistence of virus or not remains to be investigated. It is worth noting that remarkable viral infection persisted even at the end stage of COVID–19, when the viremia was well passed in the great majority of patients. Under this circumstance, use of specific anti-viral therapy should be encouraged. Recently, our group identified the convalescent plasma (CP) from recovered COVID–19 to be a specific and effective therapy for this disease, especially in severe cases, since the overwhelming majority of patients presented high neutralizing antibody titers against SARS-CoV–2 and the preliminary results of a phase I trial of CP showed very promising effects (Duan K, ZHANG XX, YANG XM et al, submitted).
Still, some issues remain to be addressed in future studies: first, what are the molecular and cellular mechanism underlying the infection of alveolar macrophages of SARS-CoV–2 should be illustrated so that a deeper understanding of the pathogenic role of viral infection and the mechanism for its escape from immune reaction can be achieved. These studies may accelerate smart drug and vaccine design targeting vulnerabilities of viral proliferation; Second, in the two cases studied here and in some other recent reports, there is a remarkable reduction of both CD4 and CD8 cells in the peripheral blood in COVID–19 patients. A graded decrease of T cells was found with increase clinical severity of COVID–19. Intriguingly, there is a negative correlation between the extent of T lymphocytopenia and increased IL–6 and Il–8 levels in the serum. The causal relationship between these two phenomena should be addressed; Third, in this study, no ACE2-expression was found on the surface of T cells, which may eliminate the possibility of a direct toxic effect of SARS-CoV–2 on distinct subsets of T cell population. However, only a small number of T lymphocytes were observed in the inflammatory lung tissues. This situation seems to be a paradox to the initial assumption that the severe T cell reduction could be ascribed to a tremendous infiltration of T cells into damaged lung tissues in response to the effect of IL–6 and other cytokines. The detailed mechanism of T cell depletion in severe COVID–19 certainly requires in-depth study in the future either among patients or in experimental animal models.