Clinical data and biosamples

Clinical data were extracted from medical records. Post-transplant care followed the in-house standard of care^18,49 with adaptation of interventions and therapies to prolonged SARS-CoV-2 infection as depicted below. Biosamples were obtained for routine clinical care. Healthy donor samples were included in Fig. 4b+c as controls (n = 2 males (37 and 43 years); n = 3 females (24, 25 and 27 years)). Written informed consent was obtained from the patient and healthy donors. The study was conducted in accordance to federal guidelines, local ethics committee regulations (Albert-Ludwigs-Universität, Freiburg, Germany; vote: 322/20, 10/03, 21–1135 and 383/19) and the Declaration of Helsinki (1975).

Bronchoscopy and bronchoalveolar lavage (BAL)

Bronchoscopy and BAL were performed as surveillance bronchoscopies as usual post-transplant care or as clinically indicated. BAL samples and lymphocyte analyses were performed as described below⁵⁰.

Analysis of the lymphocyte counts

Post transplantation bronchoscopies were performed at the indicated time points (Fig. 1a) as part of the usual post-transplant care and as medically indicated¹⁸. Brochoalveolar lavage (BAL) was performed in the middle lobe segment or the lingual with 200–300 mL pre-warmed saline in 20 mL aliquots in a bronchoscopic wedge position⁵⁰. Saline solution was recovered by gentle suction and pooled in a polypropylene tube. Lavage fluid was kept on ice and processed immediately by filtering through two layers of cotton gauze. The cells were centrifuged at 500 g and then washed with phosphate-buffered saline (PBS) at 4 °C. Cell count and cell viability were assessed after staining with trypan blue, using a Thoma Chamber (Roth, Germany). For cell differentiation, cells were resuspended in PBS containing 1 % BSA, and cell smears were prepared. Air-dried smears were stained by HEMACOLORTM (E. Merck, Darmstadt, FRG) and a minimum of 300 cells were counted throughout the whole slide. To determine lymphocyte subpopulations, BAL cells were stained with the antibodies listed in Supplementary Table 3. Stained cells were measured using a CytoFLEX cytometer (Beckman Colter, Krefeld, Germany) and percentages were evaluated using CytExpert software v.2.4.0.28 (Beckman Colter, Krefeld, Germany).

Virus detection by qPCR

SARS-CoV-2 RNA testing of oropharyngeal swabs was performed using Alinity m SARS-CoV-2 assay (09N78-095, Abbott, Illinois, USA). RNA samples were extracted using the QIAamp MinElute Virus Spin kit (57704, Qiagen, Hilden, Germany). The test was performed and interpreted according to the manufacturer’s instructions, and semi-quantitative results reported in cycle threshold (Ct) values.

Serological testing

SARS-CoV-2-specific anti-nucleoprotein (N) IgG (7304, Mikrogen Diagnostik GmbH, Neuried, Germany) and anti-N/anti-S IgM (ESR400M, Serion, Germany) ELISAs were performed according to the manufacturer’s protocol. Results were evaluated semi-quantitatively as arbitrary units (AU) compared to the manufacturer’s calibrators or shown as raw values.

Cell culture, virus isolation and growth kinetics

Virus isolation and cell culture experiments with SARS-CoV-2 were performed under Biosafety Level 3 (BSL3) protocols at the Institute of Virology, Freiburg. Adherent African green monkey kidney VeroE6 cells (ATCC CRL-1586) and human lung Calu-3 cells (ATCC HTB-55), kindly provided by Markus Hoffmann (Göttingen), were cultured in 1 × Dulbecco’s modified Eagle medium (DMEM) containing 5% or 10% fetal calf serum (FCS), respectively. All cell lines were routinely tested for mycoplasma. To isolate SARS-CoV-2 from patient material, filtered throat swabs were inoculated on 2 × 10⁶ Calu-3 cells in 4 mL DMEM with 2% FCS and incubated at 37 °C and 5% CO₂ for 4–6 days until the cytopathic effect was visible. The culture supernatant was cleared and stored at − 80 °C. Virus titers were determined by plaque assay on VeroE6 cells. Mutations in the viral genomes of the initial isolation and all derived virus stocks were confirmed by next-generation sequencing.

For viral growth kinetics 1 × 10⁶ VeroE6 or Calu-3 cells were infected with a multiplicity of infection of 0.001 for 1.5 h. Cells were washed three times with PBS and overlaid with 2 mL DMEM with 2% FCS. The supernatants were collected at 8 h, 24 h, 48 h and 72 h post infection. Viral titers were determined by plaque assay on VeroE6 cells. As a control, a prototypic BA.2 isolate was used (EPI_ISL_9324096).

Evaluation of the neutralizing capacity of therapeutic monoclonal antibodies

Neutralizing antibody titers were determined by a plaque reduction assay. Therefore, serial monoclonal antibodies dilutions were incubated with 100 plaque-forming units (pfu) of the SARS-CoV-2 isolates for 1 h. The mixture was dispersed on VeroE6 cells in a 12-well format, and cells were overlaid with 0.6% oxoid-agar for 72 h at 37 °C. Fixed cells were stained with 0.1% Crystal violet. The number of plaques was compared with an untreated control without serum or antibodies. To evaluate the neutralizing capacity and determine the neutralizing titer 50 (NT₅₀), a non-linear fit least squares regression (constraints: bottom constant equal to 0 and upper constant equal to 100) was performed. Besides the patient isolates and the prototypic BA.2 isolate, the Muc-IMB-1 isolate (lineage B.1) was used as a control (EPI_ISL_406862 Germany/BavPat1/2020)^51,52, kindly provided by Roman Woelfel, Bundeswehr Institute of Microbiology.

Whole-genome sequencing

For SARS-CoV-2 sequencing, the NEBNext ARTIC SARS-CoV-2 FS Library Prep Kit (E7658L, NEB, Frankfurt am Main, Germany) was used. Briefly, cDNA was generated from the RNA of oropharyngeal swabs (57704, QIAamp MinElute Virus Spin kit, Qiagen) or from cell culture supernatants (R1035, Quick-RNA Viral Kit, Zymo Research). The viral genome was then amplified by PCR with primers tiling the entire viral genome. Subsequently, indexed paired-end libraries for Illumina sequencing were prepared. Normalized and pooled sequencing libraries were denatured with 0.2 N NaOH and sequenced on an Illumina MiSeq instrument using the 300-cycle MiSeq Reagent Kit v2 (MS-102-2002, Illumina).

The de-multiplexed raw reads were subjected to a custom Galaxy pipeline, which is based on SARS-CoV-2 analysis pipelines on usegalaxy.eu⁵² The raw reads were pre-processed with fastp⁵³ and mapped to the SARS-CoV-2 Wuhan-Hu-1 reference genome (Genbank: NC_045512) using BWA-MEM⁵⁴. Primer sequences were trimmed with ivar trim (https://andersen-lab.github.io/ivar/html/manualpage.html). Variants (SNPs and INDELs) were called with the ultrasensitive variant caller LoFreq⁵⁵, demanding a minimum base quality of 30 and a coverage of at least 20-fold. Afterwards, the called variants were filtered based on a minimum variant frequency of 10% and on the support of strand bias. The effects of the mutations were automatically annotated in the vcf files with SnpEff⁵⁶. Finally, consensus sequences were constructed by bcftools (v.1.10)⁵⁷. Regions with low coverage or variant frequencies between 0.3 and 0.7 were masked with Ns. Final consensus sequences have been deposited in the GISAID database (www.gisaid.org) (Supplementary Table 4). Raw data has been deposited in the European Nucleotide Archive (ENA) under the study accession number: PRJEB71389.

Phylogenetic analysis

All Omicron BA.2 variant sequences generated in Freiburg, Germany between the 01.03.2022 and 10.11.2022 were included in the analysis (Supplementary Table 4). Then, a maximum-likelihood phylogenetic tree was constructed based on SARS-CoV-2 full genome consensus sequences. Therefore, sequences were aligned with MAFFT (v7.45)⁵⁸ and a tree was constructed with IQ-Tree (v2.1.2)⁵⁹. The best-fitting substitution model was automatically determined, and the tree was calculated with 1000 bootstrap replicates. Branch support was approximated using the Shimodaira–Hasegawa [SH]-aLRT method (1000 replicates). The tree was rooted to the reference sequence NC_045512. To visualize the phylogenetic tree, a custom R script was written utilizing the ggtree (v2.2.4)⁶⁰, treeio (v1.12.0)⁶¹ and ggplot2 (v3.3.3) packages.

Mutational analysis

An in-house R script was used to plot the variant frequencies that were detected by LoFreq as a heatmap (pheatmap package v1.0.12). The script is publicly available (github.com/jonas-fuchs/SARS-CoV-2-analyses, v.1.1 https://doi.org/10.5281/zenodo.7692398) and has also been implemented as a Galaxy tool (Variant Frequency Plot on usegalaxy.eu). Percentage of substitutions, substitution type and amino acid effect of novel viral mutations with a variant frequency > 50% compared to Wuhan-Hu-1 were calculated on the basis of the annotated VCF files.

To analyze the global number of sequences that harbor the single or combinatorial spike mutations (S:K356T, S:L368I, S:T385I), GISAID was accessed with the outbreakinfo R package on the 13.02.2023, including all sequences between the 12.01.2021 and 24.10.2022 (~ 6.9 million sequences)⁶².

BA.2 consensus mutations were calculated on the basis of 4992 BA.2 sequences from Baden-Wuerttemberg, Germany, between April and December 2022 (GISAID Identifier: EPI_SET_230216es, https://doi.org/10.55876/gis8.230216es). Mutational profiles were determined with covsonar v.1.1.8 (https://github.com/rki-mf1/covsonar) and a lightweight Python script (https://github.com/jonas-fuchs/covsonar_con_mut).

Visualization of the spike protein structure

The EM structures were accessed from the protein data bank (7XIX, 7XB0, 7X1M, 7L7E) and visualized with UCSF ChimeraX version: 1.1 (2020-09-09).

Peripheral blood mononuclear cell (PBMC) isolation

PBMCs were isolated from venous blood samples and BAL by lymphocyte separation medium density gradients (Pancoll separation medium, PAN Biotech GmbH) and either stored at − 80 °C until further processing or immediately used for further analyses.

Peptides and tetramers for SARS-CoV-2-specific T cell analysis

Peptides were produced with an unmodified N-terminus and an amidated C-terminus with standard Fmoc chemistry and exhibited > 70% purity (Genaxxon Bioscience, Ulm, Germany) (Supplementary Table 5). For tetramer generation, first peptides were loaded on biotinylated HLA class I easYmers (HLA-A*01:01 (1001-01), HLA-A*03:01 (1016-01), immunAware, Hørsholm, Denmark). Subsequently, the peptide-loaded HLA class I easYmers were tetramerized by conjugation to allophycocyanin- (APC, BD Biosciences, Heidelberg, Germany) or phycoerythrin- (PE, Agilent, California, US) coupled streptavidin according to the manufacturer’s instructions (https://immunaware.com/wp-content/uploads/2024/08/HLA-easYmers%C2%AE-Full-protocol-1.pdf).

Detection of SARS-CoV-2-specific CD8+ T cells in blood and BAL samples

SARS-CoV-2-specific (A*03/S_378-386, A*03/N_361-369, A*01/S_865-873, A*01/ORF3a_207-215, A*01/ORF1ab_4163-4172) CD8+ T cells in blood samples were identified by magnetic bead-based enrichment^63,64. To this end, 8×10⁶ to 10 x 10⁶ PBMCs were stained with APC-conjugated peptide-loaded HLA class I tetramers for 30 min. Subsequently, PBMCs were washed with PBS including 0.5 % BSA and 2 mM EDTA, and incubated with magnetic anti-APC microbeads (Miltenyi Biotec, Bergisch Gladbach, Germany) for 20 min at 4 °C. Afterwards, the cells were washed again and then subjected to positive selection using MACS technology (Miltenyi Biotec, Bergisch Gladbach, Germany) according to the manufacturer’s instructions (https://static.miltenyibiotec.com/asset/150655405641/document_rcctpg55vh435a1ffh7gfbac14?content-disposition=inline). SARS-CoV-2 spike-specific (A*03/S_378-386, A*01/S_865-873) CD8+ T cells in BAL samples were detected by ex vivo staining with fluorophore-coupled peptide-loaded HLA class I tetramers. All samples were analyzed by flow cytometry using the antibodies listed in Supplementary Table 3. After fixation of the cells in 2% paraformaldehyde (PFA, Sigma, Taufkirchen, Germany), analyses were performed on a CytoFLEX (Beckman Colter, Krefeld, Germany) with CytExpert Software v.2.3.0.84. Data were further analyzed using FlowJo v.10.7.1 (Treestar).

In vitro expansion of SARS-CoV-2-specific T cells and intracellular IFN-γ staining

For expansion and stimulation of SARS-CoV-2-specific T cells, pre-described minimal epitopes or for the HLA types best matching, predicted epitopes were used (Supplementary Table 5). For in vitro expansion of SARS-CoV-2-specific CD8+ T cells, 20 % of the PBMCs were first stimulated with a pool of optimal epitopes (10 µg/mL) for 1 h at 37 °C⁶⁵. Subsequently, the cells were washed, and the remaining, unstimulated PBMCs were added in RPMI medium supplemented with interleukin-2 (IL-2; 20 U/mL, StemCell Technologies). For expansion of SARS-CoV-2-specific CD4+ T cells, 2 x 10⁶ PBMCs were stimulated with a pool of 3 to 4 peptides (10 µM) and anti-CD28 monoclonal antibody (0.5 µg/mL, Pharmingen Becton Dickinson, Heidelberg, Germany). SARS-CoV-2-specific CD4+ and CD8+ T cells were expanded for 14 days in complete RPMI cell culture medium. On day 14, PBMCs were restimulated with individual peptides, left untreated as a negative control or stimulated with phorbol 12-myristate 13-acetate (PMA) and ionomycin as a positive control in the presence of brefeldin A (GolgiPlug, 0.5 µl/mL; BD Biosciences, Heidelberg, Germany) and IL-2. After 5 h of incubation at 37 °C, surface and intracellular IFN-y staining was performed. T cell responses were determined by subtracting the signal detected in unstimulated samples from stimulated samples and subsequently applying a cut-off of 0.01 %. For flow cytometry analysis, the antibodies listed in Supplementary Table 3 were used. After fixation of the cells in 2% PFA, analyses were performed on a FACSCanto system (BD Biosciences) with the FACSDiva software v.10.6.2 (BD Biosciences). Data were further analyzed using FlowJo v.10.7.1 (Treestar).

In vitro expansion of SARS-CoV-2-specific CD8+ T cells and analyses of effector function

A*03/S378-386- and A*03/N361-369-specific CD8+ T cells were expanded by stimulation of 1.5 x 10⁶ PBMCs with the respective peptide (10 µM) and anti-CD28 monoclonal antibody (0.5 µg/mL, Pharmingen Becton Dickinson, Heidelberg, Germany). SARS-CoV-2-specific T cells were expanded for 14 days in complete RPMI cell culture medium. On day 14, expanded SARS-CoV-2-specific CD8+ T cells were restimulated with the respective peptide, left untreated as a negative control or stimulated with PMA and ionomycin as a positive control⁶⁶. After 1 h of incubation at 37 °C, brefeldin A (GolgiPlug, 0.5 µl/mL) and monensin (GolgiStop, 0.5 µl/mL) (both BD Biosciences, Heidelberg, Germany) were added, and cells were incubated for an additional 4 h at 37 °C. After in total 5 h of incubation, surface and intracellular cytokine staining was performed. In addition to the restimulation, expanded cells were stained with peptide-loaded HLA class I tetramers to assess the functionality of the expanded cells. For flow cytometry analysis, the antibodies listed in Supplementary Table 3 were used. After fixation of the cells in 2 % PFA (Sigma, Taufkirchen, Germany), analyses were performed on a CytoFLEX (Beckman Colter, Krefeld, Germany) with CytExpert Software v.2.3.0.84 and a FACSCanto system (BD Biosciences, Heidelberg, Germany) with the FACSDiva software v.10.6.2 (BD Biosciences, Heidelberg, Germany). Data were further analyzed using FlowJo v.10.7.1 (Treestar).

Detection of CD4+ T cells in blood and BAL samples

For the detection and differentiation of donor and recipient CD4+ T cells, PBMCs and BAL samples were stained ex vivo with the antibodies listed in Supplementary Table 3. After fixation of the cells in 2% PFA (Sigma, Taufkirchen, Germany), analyses were performed on LSRFortessa with FACSDiva software v.10.6.2 (BD Biosciences, Heidelberg, Germany). Data were further analyzed using FlowJo v.10.7.1 (Treestar).

Alignment of patient SARS-CoV-2 sequence to T cell epitope sequences

To analyze whether T cell epitopes are affected by viral mutation, amino acid sequences of CD8+ and CD4+ T cell epitopes, which have been either described or predicted to be restricted by the HLA types of lung donor and recipient, were mapped to SARS-CoV-2 sequences isolated from the lung transplant recipient during the course of infection. These sequence analyses were performed in Geneious Prime 2022.0.2 Clustal Omega 1.2.2 alignment with default settings⁶⁷.

Data analysis

Data analyses, statistics and plotting were performed with GraphPad Prism v8.4.2, R Studio (R version 4.2.1) or Python 3.9.

Statistics & reproducibility

No statistical method was used to predetermine the sample size. No data were excluded from the analyses. The experiments were not randomized. The Investigators were not blinded to allocation during experiments and outcome assessment.

Reporting summary

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