Visualizing in deceased COVID-19 patients how SARS-CoV-2 attacks the respiratory and olfactory mucosae but spares the olfactory bulb

Abstract

Anosmia, the loss of smell, is a common and often the sole symptom of COVID-19. The onset of the sequence of pathobiological events leading to olfactory dysfunction remains obscure. Here, we have developed a postmortem bedside surgical procedure to harvest endoscopically samples of respiratory and olfactory mucosae and whole olfactory bulbs. Our cohort of 85 cases included COVID-19 patients who died a few days after infection with SARS-CoV-2, enabling us to catch the virus while it was still replicating. We found that sustentacular cells are the major target cell type in the olfactory mucosa. We failed to find evidence for infection of olfactory sensory neurons, and the parenchyma of the olfactory bulb is spared as well. Thus, SARS-CoV-2 does not appear to be a neurotropic virus. We postulate that transient insufficient support from sustentacular cells triggers transient olfactory dysfunction in COVID-19. Olfactory sensory neurons would become affected without getting infected.

Keywords: B.1.1.7; COVID-19; SARS-CoV-2; UGT2A1; coronavirus; leptomeninges; olfactory bulb; olfactory receptor; olfactory sensory neuron; sustentacular cell.

Graphical abstract

Figure 1

Postmortem bedside surgical procedure for tissue sample harvesting (A) Immediately after the death of a hospitalized COVID-19 patient or a control patient, the ENT team was contacted by phone, and they harvested samples bedside using an adapted endoscopic endonasal transcribriform approach. (B) Cohort of 68 COVID-19 patients. Time variables are expressed as median and interquartile range Q1–Q3. All other variables are expressed as percentages. (C–E) Endoscopic images (left) taken prior to harvesting samples of RM (C) and OM (D), and whole OBs (E). Illustrations indicate the location of the resected tissue (marked in blue) on a coronal section (middle) and a parasagittal (C) or midsagittal (D and E) section (right). Frozen frames were collected from the video of the procedure performed in the right nasal cavity of COVID #33 with a 4 mm 0° endoscope. IT, inferior turbinate; MT, middle turbinate; NS, nasal septum; OC, olfactory cleft. The 3D head models show the corresponding planes. See also Figure S1, Figure S3, and S4.

Figure S1

Patient, disease, and procedure characteristics of the cohort of 68 COVID-19 patients, 2 convalescent patients, and 15 control patients, related to Figure 1 Convalescent cases COVID #3 and COVID #66 are listed separately because these patients recovered from COVID-19 and died of other causes in a hospital several months after recovering. Continuous variables (time variables and body mass index) are expressed as median (interquartile range: Q1–Q3), and count variables are expressed as percentages. SOFA, Sequential Organ Failure Assessment; APACHE II, Acute Physiology And Chronic Health Evaluation II; P/F ratio, arterial-to-inspired oxygen ratio (P_aO2/F_IO2); ICU, intensive care unit.

Figure 2

Respiratory mucosa, olfactory mucosa, and olfactory bulb in control patients Brightfield images of hematoxylin and eosin-stained sections (top), confocal images of sections stained fluorescently with RNAscope and IHC (middle), and schematics of the main cell types with the genes (italics) and proteins (roman) used as markers (bottom). (A) RM of control #12. FOXJ1 is a marker for ciliated cells and EPCAM for epithelial cells. The MUC5AC-IR signal labels goblet cells and identifies blobs of secreted mucus. (B) OM of control #4. OR5A1 is one of the 389 OR genes in the human genome. OMP is a marker for mature OSNs, and KRT8 for sustentacular cells. (C) OB of control #15. SSTR2A is a leptomeningeal marker, and TUBB3 a neuronal and axonal marker. In the schematic, the pia mater is depicted as a thin gray line surrounding the OB, the dura mater as a thick light-blue line, and the arachnoid as a brown spider web-like structure between the pia mater and the dura mater. Axons of OSNs course through holes of the cribriform plate and synapse with three second-order neurons in the OB. DAPI served as nuclear stain. The schematics show the main cell types that we studied.

Figure S2

Negative control stainings, related to Figures 3, 5, and 7 (A) Confocal image of a section of the respiratory mucosa of COVID #63. A negative control for the RNAscope protocol was performed on a section adjacent to the section of which a confocal image is shown in Figure 3L. The specificity of the densely packed SARS-CoV-2-orf1ab-sense puncta in cells lining gland ducts in the lamina propria in Figure 3L is demonstrated by the absence of puncta for the dapB gene of Bacillus subtilis in any of the three Opal channels (3-plex). (B) Confocal image of a section of the respiratory mucosa of COVID #51. The SARS-CoV-2-S probe gives densely packed red puncta. The SARS-CoV-2003-S probe, specific for S of SARS-CoV (now known as SARS-CoV-1) causing an outbreak in 2002–2004, gives no green puncta. (C) Confocal image of a section of the respiratory mucosa of control #12. This negative control for the SARS-CoV-2-M probe and the nucleocapsid antibody reveals no red puncta or blue IR signal. KRT8 marks epithelial cells in the respiratory epithelium and the lamina propria. (D) Confocal image of a section of the respiratory mucosa of control #12. This negative control for the SARS-CoV-2-N-sense and SARS-CoV-2-orf1ab probes reveals no red or blue puncta. Ciliated cells harbor FOXJ1 puncta. (E) Confocal image of a section of the olfactory mucosa of control #15. This negative control for the SARS-CoV-2-S and SARS-CoV-2-orf1ab-sense probes reveals no red or blue puncta. TUBB3 marks OSNs in the olfactory epithelium and labels OSN axon bundles in the lamina propria. (F) Confocal image of a section of the olfactory bulb of control #15. This negative control for the SARS-CoV-2-N and SARS-CoV-2-S-sense probes reveals no red or blue puncta. TUBB3 marks axons and olfactory bulb neurons. DAPI served as nuclear stain.

Figure 3

Infection of the respiratory mucosa by SARS-CoV-2 (A–C) Confocal images of sections through the RM of control #12. FOXJ1 and KRT7 are markers for ciliated cells, and KRT5/6 for basal cells (A). EPCAM is a marker for epithelial cells in the RE and the LP, MUC5AC for goblet cells in the RE, and cells lining a transversely cut gland duct in the LP (B). The ACE2-IR signal caps the surface of the RE, and TMPRSS2 puncta cluster with FOXJ1 puncta in ciliated cells (C). (D–L) Confocal images of sections through the RM of COVID #7, #27, #51, #39, #29, and #63. Widespread nucleocapsid-IR signal occurs apically within the RE, and orf1ab-sense puncta reflect ongoing viral replication (D). The dashed square in (D) is the area magnified in (E). N-sense puncta cluster with orf1ab-sense puncta in nucleocapsid-IR cells (F) and with orf1ab puncta (G). Ciliated cells harbor N-sense, M, and FOXJ1 puncta (H). An individual ciliated cell harbors orf1ab-sense and FOXJ1 puncta (I). Ciliated cells harbor S-sense and FOXJ1 puncta and contain KRT7-IR signal (J). KRT8-IR cells lining gland ducts in the LP harbor N puncta (K) or orf1ab-sense puncta (L). DAPI served as nuclear stain. See also Figure S2.

Figure S3

Swimmer plot of the 30 informative COVID-19 cases, related to Figure 1 The definition of “informative case” is based on the detection of SARS-CoV-2 RNA in the respiratory mucosa by the RNAscope platform of ultrasensitive single-molecule fluorescence in situ RNA hybridization. The vertical axis shows the pseudonyms of the cases, with red labels indicating cases with ongoing viral replication at the time of death. The horizontal axis shows the period in days starting from the time the nasopharyngeal swab was taken that led to the diagnosis of COVID-19 by PCR (indicated by a test tube at day 0) until the time of death (indicated by a vertical stop line at the end of a bar). Hospitalization in a COVID-19 unit is indicated by the start of the orange bar (ICU, Intensive Care Unit) or the blue bar (ward).

Figure S4

Results of rapid antigen tests and Ct values of PCR tests on nasopharyngeal swabs taken postmortem and preprocedurally, related to Figure 1 (A) Contingency table comparing rapid antigen test (RAT) results with informative versus non-informative classification based on RNAscope staining. RATs were performed starting with COVID #9. In 27 of the 30 informative cases, a RAT was performed, and in 24 of these (89%) the RAT was scored positive. In 34 of 38 the non-informative cases, a RAT was performed, and in 27 of these (79%) the RAT was scored negative. (B) Ct-values of PCR tests ranked from low ( = high viral load) to high ( = low viral load) and RAT results from 9 informative cases (median 18.2) and 2 non-informative cases, COVID #58 and COVID #52.

Figure 4

Infection of the respiratory mucosa with B.1.1.7/Alpha versus non-B.1.1.7/non-Alpha (A) Confocal fluorescence images of sections through the RM of COVID #60, infected with a non-Alpha lineage (top), and of COVID #68, infected with Alpha (bottom). Several ciliated cells harboring FOXJ1 puncta are diffusely filled with nucleocapsid-IR signal, and some contain orf1ab-sense puncta (top) or S-sense puncta (bottom). DAPI served as nuclear stain. (B and C) BaseScope chromogenic stainings with a mix of two probes designed for wt S and the HV deletion (B), or a mix of two probes designed for wt orf1ab and the SGF deletion (C). The presence of teal and absence of red precipitate is diagnostic for infection with a non-Alpha lineage (B and C, top). Conversely, the presence of red and absence of teal precipitate is consistent with infection with Alpha (B and C, bottom).

Figure 5

Infection of the olfactory mucosa by SARS-CoV-2 (A–C) Confocal images of sections through the OM of control #4 (A and B) and non-infected OM of COVID #22 (C). Four OSNs harbor puncta for a pool of probes for four OR genes, and KRT8-IR sustentacular cells harbor TMPRSS2 puncta (A). ACE2-IR crests face the lumen of the nasal cavity (B). ANO2 is a marker for mature OSNs, GPX3 for sustentacular cells, and KRT5/6 for basal cells (C). (D–J) Confocal images of sections through the OM of COVID #8. Nucleocapsid-IR signal occurs in a few sustentacular cells but not in TUBB3-IR OSNs, and uninfected sustentacular cells harbor puncta for UGT2A1, a gene encoding an UDP glucuronosyltransferase (D). N puncta are densely packed in several sustentacular cells, which are low on or devoid of UGT2A1 puncta and ERMN-IR signal, whereas uninfected sustentacular cells harbor UGT2A1 puncta and are capped with ERMN-IR signal (E). Uninfected sustentacular cells harbor GPX3 puncta but not an individual sustentacular cell diffusely filled with nucleocapsid-IR signal and harboring orf1ab-sense puncta (F). Numerous OSNs in the middle layer harbor puncta for the probe pool, and the apical layer of KRT8-IR sustentacular cells harbors S-sense puncta (G and H). The dashed square in (G) is the area magnified in (H). S-sense puncta occur in the apical layer of KRT8-IR sustentacular cells, but not in OSNs harboring puncta for the probe pool (I) and not in an individual OR7C1⁺ OSN and in surrounding TUBB3-IR cells (J). DAPI served as nuclear stain. See also Figures S2 and S5.

Figure S5

Infection of the olfactory mucosa by SARS-CoV-2, related to Figure 5 Confocal images of sections through the olfactory mucosa of COVID #8 (A and B), COVID #7 (C-H), COVID #57 (I), and COVID #25 (J). (A and B) SARS-CoV-2-N puncta occur throughout the apical-basal width of the olfactory epithelium. KRT8-IR signal labels a patch of uninfected sustentacular cells (stippled line) and cells lining gland ducts in the lamina propria (A). Infected sustentacular cells are low on or negative for UGT2A1 puncta and ERMN-IR signal, in contrast to uninfected sustentacular cells (B). (C-E) SARS-CoV-2-N puncta occur throughout the apical layer of sustentacular cells, in a mutually exclusive manner with CNGA2 puncta and TUBB3-IR signal (C and D) or GNAL puncta (E) in the middle layer of OSNs. (F) Sustentacular cells harbor SOX2 puncta across their apical-basal width. The stippled lines outline two sustentacular cells harboring perinuclear SARS-CoV-2-orf1ab-sense puncta reflecting ongoing viral replication. (G and H) Three infected sustentacular cells harbor densely packed SARS-CoV-2-N puncta, reflecting a high viral load. The stippled lines in H outline two sustentacular cells that are in the plane of focus of this confocal image: their KRT18-IR signal is depleted, in contrast to the strong KRT18-IR signal in nearby uninfected sustentacular cells. (I) Remnants of TUBB3-IR OSNs do not contain nucleocapsid-IR signal. (J) A patch of disintegrating olfactory epithelium containing nucleocapsid-IR signal is flanked by two areas of olfactory epithelium that do not contain nucleocapsid-IR signal but contain numerous TUBB3-IR OSNs.

Figure 6

Spatial whole-transcriptome profiling of the olfactory epithelium in COVID-19 (A) Scanned image of a section through the OM of COVID #8. RNAscope for S was combined with IHC for KRT8/18 and pan-KRT. Four AOIs in the OE are indicated, with a low viral load (13 and 14) or a high viral load (20 and 7). SYTO 83 served as nuclear stain. (B) Magnification of AOI 13 and AOI 7. (C) Boxplots of nucleus counts of AOIs with low versus high viral load, with the numbers of the four AOIs shown in (A). NS, not significant. (D) Boxplots of log₂ normalized expression counts for orf1ab (p = 7.96e-07). (E) Regression curve between normalized expression counts for S (x axis) and orf1ab (y axis). (F) Volcano plot with the magnitude expressed as log₂ FC (x axis) and significance expressed as –log₁₀ of the unadjusted p value (y axis) of differential expression of 9,262 genes (dots) in WTA data between the ten ORF1ab High AOIs and the seven ORF1ab Low AOIs. Dashed vertical lines represent a |log₂ FC| of one. Genes of interest are labeled. Dots with gray outlines are the 26 OR genes. Dots with p values with a false discovery rate (FDR) <5% are shown in red. (G) Log₂ FC values of markers for sustentacular cells (bottom) and OSNs (middle) and of OR genes (top). The asterisk indicates the average of the 26 differentially expressed OR genes. (H) Boxplots of log₂ normalized expression counts for the 26 differentially expressed OR genes (26 OR) and the 8 OSN markers (OSN).

Figure 7

The parenchyma of the olfactory bulb is spared from infection (A–D) Confocal images of sagittal sections through an OB of COVID #16. The SSTR2A-IR signal labels leptomeninges, and the TUBB3-IR signal labels OSN axons and OB neurons. N puncta are not visible in the tiled confocal image (A), with the single asterisk indicating a position in this section (B), and double asterisks and triple asterisks positions in adjacent sections (C) and (D). N puncta occur within the side of pia mater abutting the OB (B). N puncta are dispersed over an area of the pia mater containing abundant nucleocapsid-IR signal (C). PECAM1 labels endothelial cells of a blood vessel cut obliquely, and S puncta occur within its lumen (D). (E) Sagittal section through an OB of COVID #7. N puncta occur within a swath of the pia mater that is partially detached. (F) Sagittal section through an OB of COVID #27. The pia mater contains densely packed M puncta. PECAM1 puncta occur in endothelial cells of a blood vessel within the pia mater and blood vessels within the OB. (G) Section through the leptomeninges including the transition zone to dura mater of COVID #60 shows scattered N and S puncta. DAPI served as nuclear stain. See also Figure S2.