Post-acute sequelae of SARS-CoV-2 cardiovascular symptoms are associated with trace-level cytokines that affect cardiomyocyte function

Ethics and participants

All research performed complied with the relevant ethical guidelines (committees and institutions listed below).

Participants with PASC were defined as individuals with persistent symptoms >12 weeks post infection who reported ongoing discomfort at the time of sample collection.

Participants with PASC-CVS were defined as individuals who experienced chest pain and/or chest tightness after SARS-CoV-2 infection >12 weeks post infection.

‘Recovered individuals’ were defined as those who were infected with SARS-CoV-2 but did not report persistent symptoms >12 weeks post infection.

‘Healthy individuals’ were defined as individuals who were recruited before community transmission of SARS-CoV-2 in Queensland or South Australia (that is, before the end of 2021) or seronegative individuals.

Details of the individual cohorts are provided below.

PASC cohort 1, PASC-CVS cohort 1 and Recovered cohort 1

Participants for the South Australia cohort of the study were enlisted through the Central Adelaide Health Network (CALHN), with the research protocol receiving approval from the CALHN Human Research Ethics Committee (HREC) in Adelaide, Australia (approval number 13050). The inclusion criteria included a PCR-confirmed SARS-CoV-2 infection from nasopharyngeal swabs, occurring in March and April 2020 for all participants, the capability to attend follow-up visits and the provision of voluntary informed written consent. Analyses of blood samples collected from this cohort at 12, 16 and 24 weeks post infection have been previously reported12. For the current study, blood samples were additionally obtained at approximately 44 and 68 weeks post infection. Convalescent patients were asked to complete a retrospective questionnaire detailing self-reported symptoms related to PASC (listed as Additional File 10 in ref. 12) at approximately 18 months (mean 70.4 weeks, minimum 61 weeks, maximum 74 weeks) post infection. Furthermore, clinical data were obtained to identify convalescent individuals referred to a long-COVID clinic operated by the South Australian State health service. Blood (54 ml per individual) was collected in serum separator tubes (acid citrate dextrose) or ethylenediaminetetraacetic acid (EDTA) tubes and processed for plasma isolation. A 2.5 ml blood sample for RNA-seq was collected into PAXgene tubes (762165 BD) and stored at −80 °C until processing. All participants provided written informed consent.

PASC-CVS cohort 2 and Recovered cohort 2

Individuals who reported a laboratory-confirmed case of SARS-CoV-2 infection in Queensland (as defined by nasopharyngeal swab or serology) were recruited from the COVID OZGenetics study (University of Queensland Human Research Ethics approval UQ 2020001490). Informed written consent was obtained from all participants. All recruited participants reported a SARS-CoV-2 infection between 4 February 2020 and 4 September 2020. Participants completed online self-report questionnaires developed in-house. Questionnaires included self-reported information on the date of testing positive, pre-COVID-19 activities, COVID-19 experiences and post-COVID-19 experience. A lifestyle and environmental exposure questionnaire used across multiple studies operated by the University of Queensland’s Human Studies Unit, which included a full self-report medical history, was also administered to all participants. Participants were asked to report on seven different symptoms across multiple time frames (brain fog, chest tightness, fast heart rate, fatigue, loss of ability to smell/taste, shortness of breath). Blood samples were collected in VACUETTE (Griener Bio-One) EDTA and serum separator tubes. Plasma and serum were isolated from respective tubes by centrifugation at 2,000 g for 10 min with low brake with transit times not exceeding 72 h from time of collection. Plasma fractions of 1.3 ml were stored at −80 °C. Where relevant, an aliquot of whole blood was taken from an EDTA tube and added to RNALater.

PASC-CVS cohort 3

Ethical approval to collect blood samples from individuals with PASC was granted by the Mater Misericordiae Ltd HREC (MML HREC) under project reference HREC/MML/89069 (V2) and accepted by Queensland Institute of Medical Research Berghofer HREC under project reference p3844. All participants provided written informed consent. Participants were administered a PASC survey which included date of SARS-CoV-2 infection and a description of symptoms which persisted >3 months post infection (including fatigue, shortness of breath, brain fog, heart palpitations, chest pain and loss of taste/smell) across different time frames. Blood was collected in either EDTA or heparin tubes (BD Biosciences) or BD SST Vacutainers (BD Biosciences). To isolate plasma, tubes containing blood samples were spun down at 1,500 g for 10 min and stored at −80 °C until processing.

Healthy cohort 1

Unexposed healthy participants were recruited from South Australia (CALHN HREC in Adelaide, Australia; approval number 13050). Healthy controls had no respiratory disease, no positive COVID-19 PCR test in 2020 and 2021, no known significant systemic diseases and negative anti-spike and anti-RBD serology.

Healthy cohort 2

Unexposed healthy participants were recruited from the Red Cross in Queensland (University of Queensland Human Research Ethics approval UQ 2020001490 and Bellberry Human Ethics Research Committee 2015-12-817-A-6). Alternatively, healthy volunteers were recruited from the Mater Research Institute (MML HREC 17/MHS/78). Informed consent was obtained from all participants. Blood was collected in either EDTA or heparin tubes (BD Biosciences) or BD SST Vacutainers (BD Biosciences). To isolate plasma, tubes containing blood samples were spun down at 2000 g for 10 min and stored at −80 °C until processing.

Human cardiomyocyte differentiation

All hiPSC studies were carried out with consent from The University of Queensland’s Institutional HREC (HREC number 2015001434). WTC-11 hiPSC cells (Gladstone Institute of Cardiovascular Disease, University of California, San Francisco)44,45 were maintained in mTeSR Plus medium with supplement (Stem Cell Technologies, catalogue number 05825) and cultured on Vitronectin XF (Stem Cell Technologies, catalogue number 07180) coated plates (Nunc, catalogue number 150318) at 37 °C with 5% CO2. HiPSC-derived cardiomyocytes were generated according to a modified monolayer platform protocol as previously described28,29,30,31,46. Briefly, on day −1 of differentiation, hiPSC were dissociated using 0.5% EDTA, plated onto Vitronectin XF-coated flasks (Nunc, catalogue number 156367) at a density of 1.12 × 105 cells per cm2 and cultured overnight in mTeSR Plus medium supplemented with 10 µM Y-27362 ROCK inhibitor (Stem Cell Technologies, catalogue number 72308). Differentiation was induced on day 0 by first washing cells with phosphate-buffered saline (PBS) when the monolayer reached approximately 80% confluence, then changing the culture medium to Roswell Park Memorial Institute (RPMI) 1640 medium (ThermoFisher, catalogue number 11875119) containing 3 μM CHIR99021 (Stem Cell Technologies, catalogue number 72054), 500 μg ml−1 bovine serum albumin (BSA, Sigma Aldrich, catalogue number A9418), and 213 μg ml−1 ascorbic acid (Sigma Aldrich, catalogue number A8960). After 3 days of culture, the medium was replaced with RPMI 1640 containing 5 μM XAV-939 (Stem Cell Technologies, catalogue number 72674), 500 μg ml−1 BSA and 213 μg ml−1 ascorbic acid. On day 5, the medium was changed to RPMI 1640 containing 500 μg ml−1 BSA and 213 μg ml−1 ascorbic acid without additional supplements. From day 7 and onward, the cultures were fed every other day with RPMI 1640 containing 2% B27 supplement with insulin (Life Technologies Australia, catalogue number 17504001). Starting on days 9 and 11 of differentiation, spontaneous beating cardiomyocytes were typically observed. The purity of hiPSC-CM cultures was confirmed by flow cytometry using a phycoerythrin-conjugated sarcomeric α-actinin antibody (Miltenyi Biotec Australia Pty) or phycoerythrin-conjugated anti-human isotype (IgG) control (Miltenyi Biotec Australia Pty).

RNA extraction

RNA extraction and genomic DNA elimination from whole blood was carried out using the PAXgene Blood RNA kit (762164, Qiagen) as per the manufacturer’s instructions (cohort 1). Alternatively, RNA was extracted using the Ribopure Blood RNA Kit (Invitrogen) (cohort 2). Cardiomyocyte RNA was extracted using the Qiagen RNeasy Plus Kit according to the manufacturer’s instructions.

RNA-seq

Whole blood—PASC, PASC-CVS, Recovered and Healthy cohort 1

Total RNA was converted to strand-specific Illumina compatible sequencing libraries using the Nugen Universal Plus Total RNA-seq library kit from Tecan as per the manufacturer’s instructions (MO1523 v2) using 12 cycles of PCR amplification for the final libraries. An Anydeplete probe mix targeting both human ribosomal and adult globin transcripts (HBA1, HBA2, HBB, HBD) was used to deplete these transcripts. Sequencing of the library pool (2 × 150 bp paired-end reads) was performed using 2 lanes of an S4 flowcell on an Illumina Novaseq 6000. RNA-seq raw read quality was evaluated with FastQC (v0.12.1)47 before quality control with Trimmomatic (v0.39)48. Reads that passed all quality control steps were then aligned to the GRCh38 human genome with HISAT2 (v2.2.1)49. A gene count matrix was generated with FeatureCounts (v2.0.6, union model)50 and the Ensembl v109 annotation, which was imported into R version 4.2 for further analysis. Genes with 51 with the Molecular Signatures Database (MSigDB) R package (msigdbr v7.4.1) and blood transcriptional modules defined by Li et al.22 as an alternative to pathway-based analyses. Data visualization was performed with ggplot2 v2.3.3. Before UMAP analysis surrogate variable analysis was applied with the sva R package v3.38.

Whole blood—PASC-CVS and Recovered cohort 2

Library preparation and sequencing was performed at the University of Queensland Sequencing Facility. RNA-seq libraries were prepared using the Illumina stranded total RNA prep ligation with Ribo-Zero plus kit (Illumina, 20040529) and IDT for Illumina RNA UD Indexes (illumina, 20040554) according to the standard manufacturer’s protocol (Illumina, document number 1000000124514 v03, June 2022). The libraries were quantified on the Perkin Elmer LabChip GX Touch with the DNA High Sensitivity Reagent kit (Perkin Elmer, CLS760672). Libraries were pooled in equimolar ratios. Sequencing was performed using the Illumina NovaSeq 6000. The library pool was diluted and denatured according to the standard NovaSeq protocol and sequenced to generate paired-end 152 bp reads using a NovaSeq 6000 SP reagent kit v1.5 (300 cycles) (Illumina, 20028400). Transcriptomic analysis was performed as described for cohort 1.

Cardiomyocytes

Library preparation and sequencing was performed at the University of Queensland Sequencing Facility. RNA-seq libraries were generated using the Illumina Stranded mRNA Library Prep Ligation kit (illumina, 20040534) and IDT for Illumina RNA UD Indexes (illumina, 20040553) according to the standard manufacturer’s protocol. Libraries were pooled in equimolar ratios. Sequencing was performed using the Illumina NovaSeq 6000. The library pool was diluted and denatured according to the standard NovaSeq protocol and sequenced to generate paired-end 102 bp reads using a 100 cycle NovaSeq reagent Kit v1.5 (illumina, 20028401). After sequencing, fastq files were generated using bcl2fastq2 (v2.20.0.422), which included trimming the first cycle of each insert read due to expected low diversity. RNA-seq raw read quality was evaluated with FastQC (v0.12.1)47 before quality control with Trimmomatic v0.3948. Reads that passed all quality control steps were then aligned to the GRCh38 human genome with HISAT2 (v2.2.1)49. A gene count matrix was generated with FeatureCounts (v2.0.6, union model)50 and the Ensembl v109 annotation. The count matrix was then imported into MATLAB (vR2022a) for further analysis. The histogram of raw log(counts) was used to determine the filter threshold for count of gene expression. Any gene expression falling below the filter threshold was taken as zero. Batch effect removal was performed using ComBat factor analysis in MATLAB with the model matrix defined using ‘batch’ and ‘class’ as regressor variables. Bioconductor differential expression analysis was performed using the rnadeseq.m function in MATLAB on the batch-corrected gene expression counts. GSEA was carried out with the fgsea R package (v1.22)51 as described above.

LEGENDplex assay

The LEGENDplex Human Anti-Virus Response Panel (IL-1β, IL-6, IL-8, IL-10, IL-12p70, IFN-α2, IFN-β, IFN-λ1, IFN-λ2/3, IFN-γ, TNF, IP-10 and GM-CSF; BioLegend) assay was performed on patient plasma samples as per the manufacturer’s instructions.

ELISA

The Human Monocyte Chemotactic Protein 1 (MCP-1) ELISA Kit (MyBioSource) assay was performed on patient plasma samples as per the manufacturer’s instructions.

Immunostorm chip and cardiac chip

The chips were prepared and used as previously described24. For the immunostorm chip, the nanopillar surface of the chip was functionalized with a cocktail of antibodies against IL-1β (AF-201, polyclonal), IL-6 (MAB9540, clone 973132), IL-12p70 (MAB611R, clone 24945R) and MCP-1 (AF-279, polyclonal) obtained from R&D Systems. Surface-enhanced Raman scattering nanotags were conjugated with antibodies against IL-1β (MAB601, clone 2805), IL-6 (AF-206, polyclonal), IL-12p70 (MAB611, clone 24945), IL-12p70 (MAB219, clone 24910) or MCP-1 (MAB679, clone 23007) obtained from R&D Systems. For patient sample analysis using the immunostorm chip, 10 µl of diluted human plasma (1:10 dilution with 1× PBS) was used.

For the cardiac chip, the nanopillar surface was conjugated with antibodies against BNP (ab236101, polyclonal, Abcam), ANP (MA5-37730, polyclonal, Invitrogen), MIP-1β (PA5-34509, polyclonal, Invitrogen), cTnI (MA1-22700, polyclonal, Invitrogen) and CK-MB (PA5-28920, polyclonal, Invitrogen). Surface-enhanced Raman scattering nanotags were conjugated with antibodies against BNP (ab236101, polyclonal, Abcam), ANP (MA5-37730, polyclonal, Invitrogen), MIP-1β (PA5-34509, polyclonal, Invitrogen), cTnI (MA1-20112, clone 16A11, Invitrogen) and CK-MB (PA5-28920, polyclonal, Invitrogen). For patient sample analysis using the cardiac chip, 10 µl of diluted human plasma (1:50 dilution with 1× PBS) was used.

Correlation analysis

Correlations between plasma cytokine levels and time since infection were performed using RSTHDA/R Web52.

Proteomics

Proteomic sample preparation

About 2 µl of nondepleted whole plasma was added to 98 µl of 50 mM Tris–HCl buffer pH 8, 6 M guanidine hydrochloride and 20 mM dithiothreitol and incubated for 30 min at room temperature. Cysteines were alkylated by the addition of acrylamide to a final concentration of 50 mM and incubation at 30 °C for 1 h with shaking. Excess acrylamide was quenched with the addition of dithiothreitol to a final additional concentration of 10 mM. Samples were transferred to 10 kDa amicon cut-off filter columns and were washed three times with the addition of 500 µl of 50 mM NH4HCO3 and centrifugation at 14,000 g for 20 min. Following the washes, an additional 150 µl of NH4HCO3 and 2 µg of sequencing grade porcine trypsin were added, and samples were digested at 37 °C for 16 h in an enclosed humidified incubator. Digested peptides were eluted by centrifugation at 10,000 g for 10 min and a final wash of 50 µl of 50 mM NH4HCO3. Peptides were desalted and concentrated with C18 ZipTips (Millipore), dried down with vacuum centrifugation and resuspended in 20 µl of 0.1% formic acid.

Mass spectrometry

Proteomic samples were analysed in randomized injection order by data-independent acquisition (DIA) using a ZenoTOF 7600 mass spectrometer (SCIEX) coupled to a Waters M-Class ultra performance liquid chromatography system. About 400 ng of sample was injected onto a Waters nanoEase M/Z HSS T3 C18 column (300 µm, 150 mm, 1.8 µm, 100 Å). Peptides were separated with buffer A (0.1% formic acid in water) and buffer B (0.1% formic acid in acetonitrile). Chromatography was performed at 5 µl min−1 with the column at 40 °C, with the following liquid chromatography program: 0–0.5 min, 1% B; 0.5–0.6 min, linear gradient to 5% B; 0.6–22 min, linear gradient to 35% B; 22–23 min, linear gradient to 60% B; 23–23.4 min, linear gradient to 90% B; held at 90% B for 3 min; and re-equilibration in 1% B for 1.6 min. Eluted peptides were directly analysed on a ZenoTOF 7600 instrument (SCIEX) using an OptiFlow Micro/MicroCal source. Curtain gas, 35 p.s.i.; collision activated dissociation (CAD) gas, 7; gas 1, 20 p.s.i.; gas 2, 15 p.s.i.; source temperature, 150 °C; spray voltage, 5,000 V; declustering potential (DP), 80 V; collision energy (CE), 10 V. MS1 spectra were acquired at m/z 400–1,500 (0.1 s) followed by DIA with 65 variable windows across an m/z range of 400–850 with fragment data acquired across m/z of 140–1,750 (0.013 s) with Zeno pulsing on and threshold set to 100,000 counts per second. Variable window sizes were generated using the SWATH Variable Window Calculator (SCIEX), based on a representative pooled plasma sample. Parameters included 1 Da window overlap and a minimum window width of 5 Da. Dynamic collision energy was automatically assigned by the Analyst software (SCIEX, v3.9.9.3339) based on variable window m/z ranges.

Data analysis

A spectral library was generated by processing all raw DIA files with DIA-NN (v1.8)53. A library-free search was performed using a FASTA file containing 20,428 human proteins (Uniprot reviewed, downloaded 7 December 2023) and common mass spectrometry contaminants. Default settings for a library-free search were used with the following changes: propionamide was set as a fixed modification on cysteine residues; N-term M excision was not included; minimum peptide length, 4; precursor charge range, 2–5; protein inference, Genes; quantification strategy, Robust LC. The spectral library generated by DIA-NN was converted into a PeakView (SCIEX) compatible format using a custom-designed script. The PeakView (v2.2, SCIEX) SWATH MicroApp was used to determine the abundance of peptide fragments, peptides and proteins as previously described54. Protein abundances were recalculated using a 1% false discovery rate cut-off. Normalization was performed to total human protein abundance in each sample. No data imputation was used, and proteins with missing values were not discarded in the analysis.

Treatment of hiPSC cardiomyocytes

Differentiated cardiomyocytes were replated on day 15 of differentiation for contractility recording using the CardioExcyte 96 system (Nanion Technologies). Cells were seeded at a density of 50,000 cells per well onto Vitronectin XF-coated NSP-96 plates (Nanion Technologies) and transferred to the CardioExcyte 96 platform for baseline monitoring. Upon stabilization of baseline readings for human plasma experiments, plasma was diluted to 1:40 in media (RPMI (Thermo Fisher Scientific) with B27 Supplement (Thermo Fisher Scientific)) then transferred to the hiPSC-derived cardiomyocytes for 48 h of treatment in either the presence or absence of 100 ng ml−1 of dexamethasone (Sigma-Aldrich). For cytokine spiking experiments, recombinant cytokines IL-12, IL-1β, IL-6 and MCP-1 (Abcam) were diluted in media at the stipulated concentrations. Contractility recordings performed with CardioExcyte 96 were obtained with 1 ms time resolution and 1 kHz sampling rate. All experiments were conducted under physiological conditions using a built-in incubation chamber in the system (37 °C, 5% CO2 and 80% humidity). The CardioExcyte NSP-96 plates use two gold electrodes on a rigid surface in each well to study physiological changes in contractility via continuous impedance measurements. The readouts for detection are amplitude, beat rate, upstroke velocity and relaxation velocity. For recombinant cytokine experiments, data are displayed as raw data, while experiments using patient plasma show data as relative to a standard media control treatment.

Statistics

Data were analysed on Graphpad v9.0.2 (Dotmatics). Normal distribution of data was assessed with the Shapiro–Wilk test. Statistical significance was determined with a Kruskal–Wallis test with Dunn’s multiple comparison test, Mann–Whitney U test or Welch analysis of variance (ANOVA) test and Dunnett’s multiple comparison test where P P values for multiple testing were made using the Benjamini–Hochberg procedure.

Reporting summary

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