SARS-CoV-2 infection and persistence in the human body and brain at autopsy – Nature

Autopsies

Autopsies were carried out and tissues were collected as previously described³⁵ in the National Cancer Institute’s Laboratory of Pathology at the National Institutes of Health Clinical Center following consent of the legal next of kin. Autopsy patients in this cohort were unvaccinated against SARS-CoV-2. Tissues preserved for histopathologic analysis and special staining were dissected fresh at the time of autopsy, placed into tissue cassettes, fixed for 24 h in neutral-buffered formalin, and then transferred to 70% ethanol for 48 h before impregnation with paraffin.

Measurement of IgG and IgM antibodies to N and spike protein of SARS-CoV-2

Fluid-phase luciferase immunoprecipitation system assays were used to study IgG and IgM antibody response to SARS-CoV-2. For IgG luciferase immunoprecipitation system measurements, Renilla luciferase–nucleocapsid and Gaussia luciferase–spike protein extracts were used with protein A/G beads (Protein A/G UltraLink Resin, Thermo Fisher) as the IgG capture reagent as previously described with microtitre filter plates³⁶. For IgM measurements, anti-human IgM goat agarose beads (Sigma) were substituted as the capture reagent using both the microfilter plate and microtube format³⁷. The IgM immunoprecipitation assays were carried out in 1.5-ml microfuge tube format containing 1 µl serum or plasma, Renilla luciferase–N (10 million light unit input per tube) or Gaussia luciferase–spike protein (40 million light input per tube) and buffer A (20 mM Tris, pH 7.5, 150 mM NaCl, 5 mM MgCl₂, 0.1% Triton X-100) to a total volume of 100 µl. After mixing, the tubes were incubated at room temperature for 1 h. Next, 10 µl of the anti-human IgM agarose bead suspension was added to each tube for a further 60 min, and tubes were placed on a rotating wheel at 4 °C. The samples were then washed by brief centrifugation to collect the bead pellet at room temperature three times with 1.5 ml buffer A and once with 1.5 ml PBS. After the final wash, the beads were mixed with coelenterazine substrate (100 µl) and light units were measured in a tube luminometer. Known seronegative and seropositive samples for IgG and IgM antibodies to the N and spike proteins were used for assigning seropositive cutoff values and for standardization.

SARS-CoV-2 RNA quantification of tissues and body fluids

Total RNA was extracted from RNAlater (Invitrogen)-preserved tissues and body fluids collected at autopsy using the RNeasy Mini, RNeasy Fibrous Tissue Mini, RNeasy Lipid Tissue Mini and QIAamp Viral RNA Mini Kits (Qiagen) according to the manufacturer’s protocols. Upstream tissue processing and subsequent RNA quantification have been described previously³⁵. The QX200 AutoDG Droplet Digital PCR System (Bio-Rad) was used to detect and quantify SARS-CoV-2 RNA in technical replicates of 5.5 µl RNA for fluids and up to 550 ng RNA for tissues. Raw data were collected using QuantaSoft version 1.7.4.0917 and analysed using QuantaSoft Analysis Pro version 1.0.596. Results were then normalized to copies of N1, N2 and RP per millilitre of sample input for fluids and per nanogram of RNA concentration input for tissues. Samples had to be positive for the human RNase P (RP) gene at the manufacturer’s limit of detection (LOD) of ≥0.2 copies per microlitre and ≥4 positive droplets per well to ensure RNA extraction was successful and be reported. For samples to be considered positive for SARS-CoV-2 N1 or N2 genes, the technical replicates needed to have an average at or above the manufacturer’s LOD of ≥0.1 copies per microlitre and ≥2 positive droplets per well. More than 60 control autopsy tissues from uninfected individuals, representing all organs collected for COVID-19 autopsy cases, were used to validate the manufacturer’s emergency use authorization published LOD for nasopharyngeal swabs for tissues (Supplementary Data 1e). ddPCR data for P3 (ref. ¹⁷) as well as a portion of the oral cavity³⁵ have been reported previously.

Subgenomic RNA analysis of ddPCR positive tissues

Tissues that tested positive for one or both SARS-CoV-2 N gene targets through ddPCR had RNA submitted for subgenomic RNA analysis. Briefly, 5 µl of sample RNA was added to a one-step real-time RT–qPCR assay targeting subgenomic RNA of the envelope (E) gene (forward primer 5′-CGATCTCTTGTAGATCTGTTCTC-3′; reverse primer 5′-ATATTGCAGCAGTACGCACACA-3′; probe 5′-FAM-ACACTAGCCATCCTTACTGCGCTTCG-ZEN-IBHQ-3′)³⁸ using the Rotor-Gene probe kit (Qiagen) according to instructions of the manufacturer. In each run, standard dilutions of counted RNA standards were run in parallel to calculate copy numbers in the samples. The LOD for this assay was determined to be <40 Cq (Supplementary Data 1) using 40 control autopsy tissues from uninfected individuals, representing all organs collected for COVID-19 autopsy cases.

Virus isolation from select postmortem tissues

Select tissues with high viral RNA levels through ddPCR and subgenomic RNA RT-qPCR measuring across a broad range of 16 to <35 Cq underwent virus isolation to prove the presence of infectious virus. Virus isolation was carried out on tissues by homogenizing the tissue in 1 ml DMEM and inoculating Vero E6 cells in a 24-well plate with 250 µl of cleared homogenate and a 1:10 dilution thereof. Plates were centrifuged for 30 min at 1,000 r.p.m. and incubated for 30 min at 37 °C and 5% CO₂. The inoculum was then removed and replaced with 500 µl DMEM containing 2% FBS, 50 U ml⁻¹ penicillin and 50 μg ml⁻¹ streptomycin. Six days after inoculation, the cytopathic effect was scored. A blind passage of samples in which no cytopathic effect was present was carried out according to the same method. Additional virus isolation from P38 thalamus and hypothalamus was carried out using Vero E6-TMPRSS2-T2A-ACE2 (catalogue no. NR-54970, BEI Resources) grown in DMEM containing 10% FBS, 50 U ml⁻¹ penicillin, 50 μg ml⁻¹ streptomycin and 10 μg ml⁻¹ puromycin. Virus isolation was carried out as described for other tissues on Vero E6 cells, without selection antibiotics. Tissue homogenate from flash-frozen specimens and supernatants from plates were analysed using RT–qPCR for SARS-CoV-2 E gene subgenomic RNA (described above) or genomic RNA as previously described³⁹ to rule out other causes for the cytopathic effect. Cell lines were not authenticated in house, but were confirmed to be free of mycoplasma contamination.

Virus sequencing

Five early cases (P18, P19, P27 and P38) and one late case (P33) with multiple body site tissues containing subgenomic RNA levels ≤31 Cq were selected for HT-SGS as previously described²². Presence of variants of SARS-CoV-2 was analysed within and between tissues.

Supernatant from the virus isolation plates of thalamus of P38 were sequenced using short-read, whole-genome sequencing. Total RNA was depleted of rRNA using Ribo-Zero+ following the manufacturer’s protocol (Illumina). Cleaned RNA was eluted in water and sequencing libraries were prepared following the Kapa RNA HyperPrep kit according to the manufacturer’s protocol (Roche Sequencing Solutions). Briefly, 10 µl of depleted RNA was used as a template for fragmentation (65 °C for 1 min) and first-strand synthesis. To facilitate multiplexing, adapter ligation was carried out with KAPA Unique Dual-Indexed Adapters, and samples were enriched for adapter-ligated product using KAPA HiFi HotStart Ready mix and a range of 9–19 PCR amplification cycles based on SARS-CoV-2 Ct values and total RNA starting inputs. Pools consisting of 1–6 sample libraries were used for hybrid-capture virus enrichment using myBaits Expert Virus SARS-CoV-2 panel following the manufacturer’s manual, version 5.01, with a range of 12–18 cycles of post-capture PCR amplification (Arbor Biosciences). Purified, enriched libraries were quantified on a CFX96 Real-Time System (Bio-Rad) using Kapa Library Quantification kit (Roche Sequencing Solutions). Libraries were diluted to 2 nM stock, pooled together as needed in equimolar concentrations and sequenced on the MiSeq (Illumina) generating 2 × 150-bp paired-end reads. Raw sequence reads were trimmed of Illumina adapter sequence using Cutadapt version 1.12 (ref. ⁴⁰) and then trimmed and filtered for quality using the fastq_quality_trimmer and fastq_quality_filter tools from the FASTX-Toolkit 0.0.14 (Hannon Lab, CSHL). Reads were then mapped to the SARS-CoV-2 2019-nCoV/USA-WA1/2020 genome (MN985325.1) using Bowtie2 version 2.2.9 (ref. ⁴¹) with parameters -local -no-mixed -X 1500. PCR duplicates were removed using picard MarkDuplicates, version 2.26.10 (Broad Institute).

SARS-CoV-2 RNA ISH

Chromogenic ISH detection was carried out using the manual RNAScope 2.5 HD assay (catalogue no. 322310, Advanced Cell Diagnostics) with a modified pretreatment protocol. Briefly, formalin-fixed and paraffin-embedded (FFPE) tissue sections were cut at 7 μm, air dried overnight, and baked for 1–2 h at 60 °C. The FFPE tissue sections were deparaffinized, dehydrated and then treated with pretreat 1 for 15 min at room temperature. The slides were boiled with pretreatment reagent for 15 min, digested with protease at 40 °C for 20 min, and then hybridized for 2 h at 40 °C with probe-V-nCov2019-S (catalogue no. 848561, Advanced Cell Diagnostics) or probe-V-nCoV-N (catalogue no. 846081, Advanced Cell Diagnostics)⁴². In addition, probe-Hs-PPIB (peptidylprolyl isomerase B, catalogue no. 313901, Advanced Cell Diagnostics) and probe-dapB (catalogue no. 310043, Advanced Cell Diagnostics) were used as a positive and negative control, respectively. Subsequent amplification was carried out according to the original protocol. Detection of specific probe-binding sites was visualized with RNAScope 2.5 HD Reagent Kit-BROWN chromogenic labels (Advanced Cell Diagnostics). The slides were counterstained with haematoxylin and coverslipped.

To correlate viral load detected between ddPCR and ISH, the interventricular septum of 16 cases spanning a four-log range of SARS-CoV-2 N copies per nanogram of RNA through ddPCR underwent ISH and subsequent quantification using image analysis. The interventricular septum stained slides were digitalized using a NanoZoomer XR Digital Pathology system (Hamamatsu, Hamamatsu City, Japan) at 40× magnification. Digitalized images were automatically analysed using Visiopharm software v2021.09.02 (Visiopharm, Hørsholm, Denmark). A training set was used to configure the algorithm and identify SARS-CoV-2 RNA signals. In brief, 3,3′-diaminobenzidine (DAB) dots of positive signals were identified using a Bayesian classifier trained on pre-processing steps. We randomly selected 30 regions of interest per slide and calculated the median of positive cells.

SARS-CoV-2 multiplex IF

FFPE CNS sections were deparaffinized, rehydrated and subjected to 0.01 M citrate buffer antigen retrieval for 20 min at 120 °C. Slides were then rinsed briefly in deionized water, washed with PBS and subsequently incubated in a 5% milk (catalogue no. 1706404, Bio-Rad), 5% normal donkey serum, 0.3 M glycine and 0.1% Triton X-100 blocking solution made up in PBS for 30 min. Primary antibodies to SARS-CoV-2 N protein 1 (NP1, 1:1,000, custom made GenScript U864YFA140-4/CB2093)^43,44,45 and neuronal nuclear protein (NeuN, 1:200, catalogue no. MAB377, Chemicon) or transmembrane protein 119 (TMEM119, 1:1,000, catalogue no. MAB130313, R&D Systems) were diluted in blocking serum and applied to slides overnight at 4 °C. The following day, slides were washed extensively with PBS to remove any detergent and freshly made True Black Plus solution (1:40 in PBS, catalogue no. 23014, Biotium) was applied for 14 min. Slides were again extensively washed and then species-specific secondary conjugates (1:500, catalogue nos. A-21206 and A-21203, Thermo Fisher) were applied for 1 h at room temperature. Following PBS wash, Hoechst 33342 was applied for 10 min (1:2,000, catalogue no. H3570, Thermo Fisher) to label nuclei. Slides were coverslipped with Prolong Gold (catalogue no. P36930, Thermo Fisher).

SARS-CoV-2 chromogenic IHC

Chromogenic IHC was carried out on various ddPCR positive CNS and lung tissues and negative pre-pandemic control cases demonstrating relative expression of the target protein between infected and control samples. Briefly, 5 µm FFPE tissue sections were incubated at 60 °C for 2 h, deparaffinized in xylene, and hydrated in serial alcohol solutions to distilled water. Heat antigen retrieval was carried out using a pressure cooker (DAKO) by submerging slides in 1× pH 6 citrate buffer for 20 min. Endogenous enzyme activity was quenched with 3% hydrogen peroxide containing sodium azide for 10 min with additional 10% non-fat dry milk (Bio-Rad) for 20 min to prevent nonspecific binding. Tissue sections were then incubated with polyclonal SARS or SARS-CoV-2 N antibody (1:500, custom made, GenScript U864YFA140-4/CB2093, 0.447 mg ml⁻¹)^43,44,45 for 1 h at room temperature. Negative controls were congruently stained on subsequent sections following the same protocol replacing the primary antibody with a rabbit IgG control antibody (0.5 mg ml⁻¹, catalogue no. I-1000-5, Vector Laboratories). The antigen–antibody reaction was detected with Dako Envision+Rb polymer detection system (DAKO) and visualized with DAB chromogen. Sections were lightly counterstained with haematoxylin, dehydrated in graded alcohols, cleared in xylene, mounted and coverslipped.

Statistical analysis

Correlations between two continuous variables were assessed using Spearman’s rank correlation coefficient (ρ). Fisher’s z-transformation was used for the calculations of 95% CI and P values. To compare ddPCR levels between tissue types (respiratory versus non-respiratory), we used linear mixed models with compound symmetry correlation structure to account for repeated measures within each subject. Standard residual diagnoses were used to check model assumptions. Log-transformations were used when needed. To log₁₀-transform ddPCR values, 0 values were replaced with a small positive random number according to the detection limits. Logistic regression models were used to generate ROC curves. Optimal cutoff values were selected by treating sensitivity and specificity as equally important. SAS version 9.4 was used for all analyses. All P values are two-sided and reported without adjustment for multiple comparisons.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.