Presumed SARS-CoV-2 Viral Particles in the Human Retina of Patients With COVID-19

Key Points

Question  Is the SARS-CoV-2 virus present in the human retina?

Findings  In this case series of 3 individuals, S and N COVID-19 proteins were seen by immunofluorescence microscopy within endothelial cells close to the capillary flame and cells of the inner and the outer nuclear layers. At the perinuclear region of these cells, it was possible to observe by transmission electron microscopy double-membrane vacuoles that were consistent with the virus, presumably containing COVID-19 viral particles.

Meaning  The present observations show presumed SARS-CoV-2 viral particles may reach the various layers of the human retina and also could be associated with this infection’s ocular clinical manifestations.

Importance  The presence of the SARS-CoV-2 virus in the retina of deceased patients with COVID-19 has been suggested through real-time reverse polymerase chain reaction and immunological methods to detect its main proteins. The eye has shown abnormalities associated with COVID-19 infection, and retinal changes were presumed to be associated with secondary microvascular and immunological changes.

Objective  To demonstrate the presence of presumed SARS-CoV-2 viral particles and its relevant proteins in the eyes of patients with COVID-19.

Design, Setting, and Participants  The retina from enucleated eyes of patients with confirmed COVID-19 infection were submitted to immunofluorescence and transmission electron microscopy processing at a hospital in São Paulo, Brazil, from June 23 to July 2, 2020. After obtaining written consent from the patients’ families, enucleation was performed in patients deceased with confirmed SARS-CoV-2 infection. All patients were in the intensive care unit, received mechanical ventilation, and had severe pulmonary involvement by COVID-19.

Main Outcomes and Measures  Presence of presumed SARS-CoV-2 viral particles by immunofluorescence and transmission electron microscopy processing.

Results  Three patients who died of COVID-19 were analyzed. Two patients were men, and 1 was a woman. The age at death ranged from 69 to 78 years. Presumed S and N COVID-19 proteins were seen by immunofluorescence microscopy within endothelial cells close to the capillary flame and cells of the inner and the outer nuclear layers. At the perinuclear region of these cells, it was possible to observe by transmission electron microscopy double-membrane vacuoles that are consistent with the virus, presumably containing COVID-19 viral particles.

Conclusions and Relevance  The present observations show presumed SARS-CoV-2 viral particles in various layers of the human retina, suggesting that they may be involved in some of the infection’s ocular clinical manifestations.

SARS-CoV-2 is an enveloped positive-sense RNA coronavirus belonging to the Coronaviridae family, and its cellular entry depends mainly on the binding of S protein^1,2 to angiotensin-converting enzyme 2, a specific cellular receptor located at the surface of the host cells.^3,4 The SARS-CoV-2 viral particles’ presence in the retina of deceased patients with COVID-19 has been suggested through the real-time polymerase chain reaction (PCR) and immunological methods to detect its main proteins. The eye is affected by COVID-19 infection,^5,6 and retinal changes were attributed to secondary microvascular and immunological changes. The aim of this study was to detect the presence of presumed SARS-CoV-2 viral particles in the retina of individuals who died of COVID-19 using fluorescence microcopy of tissues immunostained for S1 and nucleocapsid proteins and transmission electron microscopy of thin sections.

This investigational study was approved by the ethical and research committee at the Federal University of São Paulo in São Paulo, Brazil, and all patients’ representatives agreed to participate through written consent applied after the patient’s death. They were informed of the procedure and potential benefits and risks, and no compensation was received for agreeing to participate. Detailed demographic, medical history, concomitant events, medication history, hospitalization details, and laboratory test results were obtained. Data on race were not obtained. All patients had nasal swab PCR confirming SARS-CoV-2 infection according either to the CDC 2019-Novel Coronavirus (2019-nCoV) Real-Time RT-PCR Diagnostic Panel or to the Charité Protocol: SARS-CoV-2.⁷ All patients were hospitalized and died in the Hospital Municipal de Barueri Dr Francisco Moran, and enucleation was performed at this hospital from June 23 to July 2, 2020. The laboratory analysis was conducted in the Cell Structure Laboratory of the Instituto de Biofísica Carlos Chagas Filho, Federal University of Rio de Janeiro.

Eyes were enucleated within 2 hours after death. The anterior and posterior segments were separated at the ora serrata and then fixed in a freshly prepared solution in a 0.1 M cacodylate buffer; pH 7.2 of glutaraldehyde, 2.5%; and formaldehyde, 4%, for at least 30 days before further processing, as described below. Sections were made nasally adjacent to the optic nerve in the transition between the posterior pole and the midperiphery.

Ultrastructure analysis of numerous retinal and choroid sections obtained from 3 patients were analyzed. For localization of the nucleocapsid and the S1 proteins by immunofluorescence microscopy, previously fixed tissue fragments were dehydrated in ethanol, 50% to 70%, for 10 minutes and infiltrated 1:1 (ethanol, 70%:resin) for 1 hour, 4 changes of 20 minutes in pure resin and the samples were included in LR White resin, and the blocks were thermally cured at 55 °C for 24 hours. The blocks were cut into 1-nm sections with the Leica-Reichert-Ultracut (Reichert-Jung), collected in sheets precoated with 0.02 mg/mL poly-L-lysine (molecular weight, 350 000; Sigma Chemical Company), and dried in an oven at 55 °C. Antigenic recovery was performed with 10mM of citrate buffer for 5 minutes. The slides were incubated with 150mM of ammonium chloride and bovine serum albumin, 3% (Sigma), in Tris-buffered saline−Tween, 0.01%, for 30 minutes, respectively; rabbit SARS-CoV-2 spike S1 (GTX632604; GeneTex) and anti-SARS/SARS-CoV-2 coronavirus nucleocapsid (GTX635678; GeneTex) were diluted 1:100. Both antibodies are commercially available with low cost. Subsequently, the samples were incubated in the presence of Alexa Fluor 568 (Thermo Fisher) secondary antibodies diluted 1:100. The nucleus was labeled by incubation with Hoechst (Thermo Fisher) (1:2000). Observations were carried out in a confocal laser scanning microscope (Zeiss).

For transmission electron microscopy, sections of the retina were dehydrated and placed into crescent concentration of acetone (50%, 70%, 90%, and 100%) and infiltrated in 1:1 of 1:2 Epon (EMS) and resin (EMS) (acetone:resin) and pure resin for 12 hours, respectively, and included in the resin for ultrathin sections. For visualization, they were stained with uranyl acetate and lead citrate and observed in a HT7800 RuliTEM (Hitachi-Hightec) transmission electron microscope.

Three eyes from 3 patients were evaluated. All patients had been in the intensive care unit, received mechanical ventilation, and had severe pulmonary involvement from COVID-19, with more than 50% of the lungs showing multifocal pulmonary ground-glass opacities in chest tomography imaging and were receiving heparin, antibiotics, and corticosteroids. Patient 1 was a man in his 70s, with high blood pressure, diabetes, and pulmonary emphysema. At the time of death, he had 30 days of progressive illness after a positive nasal swab PCR result. Both eyes had no clinically relevant anterior segment finding. We were unable to obtain visual acuity or intraocular pressure levels. Fundus examination was documented (Figure 1). The right eye showed vitreous hemorrhage, and the left eye presented normal optic disc with sharp margins, associated with a temporal subretinal hemorrhage as well as increased vessel tortuosity (Figure 1). Patient 2, a woman in her 70s, had previous diagnoses of high blood pressure, diabetes, and dyslipidemia and died a week after a positive nasal swab PCR result. Patient 3, a man in his 60s, had diagnoses of high blood pressure, alcohol misuse, and chronic kidney failure; was undergoing dialysis; and had a positive nasal swab PCR result from almost 2 months earlier. Patients 2 and 3 had no ophthalmological evaluation because they died shortly after admission. The macroscopic examination showed absence of inflammatory lesions or other pathological changes in all patients.

Representative images obtained by transmission electron microscopy are shown in Figure 2 and Figure 3. The presence of particles between with a diameter varying from 60 to 70 nm was seen in the perinuclear region of cells of the inner nuclear layer (INL) (Figure 2B, inset), especially in double-membrane reticular structures recently characterized as derived from the changes in the endoplasmic reticulum induced by SARS-CoV-2⁸ (Figure 2B, inset and white arrowhead). They were also seen in capillary endothelial cells present in the INL where images showing surface projections resembling those seen in isolated and well-preserved virus particles are observed (Figure 2B, inset and black arrowheads, and Figure 2C, white arrowheads). Electrodense granularity within the particle was also observed (Figure 2C, asterisk). Capillary endothelial cells present in INL also had the presence of virus particles (Figure 3A, white arrowheads). In the outer nuclear layer, numerous viral particles were found in the perinuclear (Figure 3B-D, white arrowheads) and cytoplasmic regions (Figure 3D). In all viral particles, it is possible to see electrodense granularity (Figure 3, asterisk).

One classic way to determine the virus’ presence within the cells is the immunocytochemical approach using specific antibodies that recognize 2 essential proteins of the SARS-CoV-2 viral particles: S protein, exposed on the virus surface, and nucleocapsid protein, located within the virus. Using this approach applied to semithin sections of tissue embedded in LR White resin, we could visualize (Figure 4 and Figure 5) labeling indicative of the presence of the nucleocapsid protein (Figure 4A, red spots) in the cytoplasmic ganglion cell layer (Figure 4A, inset) and perinuclear region (Figure 4A, inset), as well as labeling for nucleocapsid proteins (Figure 4B-D).

Labeling with antibodies that recognize the S1 protein was mainly observed (Figure 5A-C, red spots) in the INL, and it is possible to verify by orthogonal X-Y cut that the labeling is located intracellularly, in the perinuclear region, and close to the nucleus (cyan; Figure 5A, inset and red spots). The S1 protein labeling was observed mainly (Figure 5B and C, inset and red spots) throughout the sensory tissue of the retina, ganglion cell layer, inner plexiform layer, INL, outer plexiform layer, and outer nuclear layer, as well as in the retinal pigment epithelium and choroid. The labeling for both antibodies were precise, and no labeling was observed when the sections were incubated only in the presence of the secondary-labeled antibody (controls shown in the eFigure in the Supplement). Cell nuclei were labeled with DNA intercalating compound (cyan) (350 nm excitation/461 nm emission). Under this condition, there is a pale autofluorescence of some structures of the tissue itself, such as retinal pigment epithelium melanin granules, external and internal segments of photoreceptors, and erythrocytes, which did not compromise the observation of the presence of specific proteins.

It is now clear that after the initial infection in the respiratory system, the virus can spread throughout the whole body, reaching different tissues and organs. Current literature shows the eye may be involved in COVID-19 infection, and many retinal changes have been reported. It is unknown if these changes could be secondary to the virus’ presence in the retina or microvascular and immunological secondary changes or coincident to an infection that has affected hundreds of millions of people worldwide.^14,15 One study demonstrated the detection of SARS-CoV-2 RNA through the performance of reverse-transcription PCR in human retinal biopsies of deceased patients with COVID-19.⁹ In addition, there are recent studies showing angiotensin-converting enzyme 2 expression in eye structures¹⁰ in human-induced pluripotent stem cell–derived retinal organoids and in healthy eye tissues,^11,12 findings that may indicate the possible susceptibility of the various cells to SARS-CoV-2.

Our observations show the presence of presumed SARS-CoV-2 viral particles in the retina of 3 patients who died of COVID-19, using 2 approaches. The direct visualization of presumed virus particles in thin sections of the retina was examined by transmission electron microscopy.

The presumed SARS-CoV-2 viral particles appeared as small vesicles, with a mean diameter of 70 nm, located near the cells’ nucleus and is associated with a complex membrane network known as the viral factory¹³ involving the endoplasmic reticulum–Golgi complex system. The immunolocalization by fluorescence microscopy of 2 major proteins of the virus are (1) the spike protein (S), projected from the viral envelope and is crucial for virus internalization and antigenicity, and (2) the nucleocapsid protein N. Both proteins were seen in various regions of the retina, including the ganglion cell layer, inner plexiform layer, INL, outer plexiform layer, and outer nuclear layer, as well as the pigment epithelium and choroid. Serial optical sections obtained using a confocal laser scanning microscopy showed that labeling was localized in the perinuclear region, in close agreement with the electron microscopy observations.

Study limitations were small number of cases caused by difficulty in obtaining the eyes and lack of proven pathogenic viral effect in the retina.

In conclusion, our present observations point to the presence of presumed SARS-CoV-2 viral particles in the retina of patients who died of COVID-19. Two lines of evidence support this conclusion. First, transmission electron microscopy of thin sections showed the presence of presumed virus particles with a morphology similar to that observed in cell cultures experimentally infected with the virus. Here it is important to emphasize that the images obtained do not reveal the virus structure with the same quality of those obtained in cell cultures. This is most probably owing to dealing with tissues that were fixed hours after a patient death. Second, the characteristic S and N viral proteins were visualized using immunofluorescence microscopy. Thus, the findings are in close agreement with previous recent reports showing the presence of the S1 protein in the neurosensory retina. Given that these findings support the presence of viruses that presumably are SARS-CoV-2 in the retina, research should assess whether retinal changes are related only to secondary microvascular and immunological changes, coincidence to a very prevalent infection or the virus’ direct presence, or a combination of these. The findings may help to elucidate the virus’ pathophysiological mechanisms, which may allow a better understanding of the sequelae of the disease and may direct some avenues of future research.

Corresponding Author: Alléxya A. A. Marcos, MD, Federal University of São Paulo-UNIFESP, R. Botucatu, 822 Vila Clementino, São Paulo, São Paulo 04023-062, Brazil (allexya.affonso@gmail.com).

Accepted for Publication: June 14, 2021.

Published Online: July 29, 2021. doi:10.1001/jamaophthalmol.2021.2795

Author Contributions: Drs Belfort and de Sousa had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

Concept and design: Araújo-Silva, Marcos, Branco, Roque, Schor, Nascimento, de Souza, Belfort.

Acquisition, analysis, or interpretation of data: Araújo-Silva, Marcos, Marinho, Branco, Romano, Matuoka, Farah, Burnier, Moraes, Tierno, Sakamoto, de Souza, Belfort.

Drafting of the manuscript: Araújo-Silva, Marcos, Marinho, Branco, Matuoka, Sakamoto, Nascimento, Belfort.

Critical revision of the manuscript for important intellectual content: Araújo-Silva, Marcos, Branco, Roque, Romano, Farah, Burnier, Moraes, Tierno, Schor, de Souza, Belfort.

Statistical analysis: Araújo-Silva.

Obtained funding: de Souza, Belfort.

Administrative, technical, or material support: Marcos, Marinho, Branco, Matuoka, Moraes, Tierno, Sakamoto, Belfort.

Supervision: Branco, Romano, Farah, Burnier, Schor, Nascimento, de Souza, Belfort.

Conflict of Interest Disclosures: None reported.

Funding/Support: The work has received sponsorship from Conselho Nacional de Pesquisa, Instituto da Visão, and CAPES (Coordenação de Aperfeiçoamento de Pessoal de Nível Superior).

Role of the Funder/Sponsor: The work has received sponsorship from Conselho Nacional de Pesquisa, Instituto da Visão, and CAPES (Coordenação de Aperfeiçoamento de Pessoal de Nível Superior) for the following phases of the study: design and conduct and collection, management, analysis, and interpretation of the data.