Virus and cells

Vero E6 cells were maintained in Dulbecco’s modified Eagle’s medium (DMEM, Gibco) supplemented with 10% fetal calf serum (FCS), HEPES, sodium bicabonate, penicillin (final concentration 100 IU/mL) and streptomycin (final concentration 100 IU/mL) at 37°C in a humidified CO₂ incubator. SARS-CoV-2 (isolate BetaCoV/Munich/BavPat1/2020; European Virus Archive Global #026V-03883; kindly provided by Dr. C. Drosten) was propagated on Vero E6 cells in Opti-MEM I (1X) + GlutaMAX (Gibco), supplemented with penicillin (final concentration 100 IU/mL) and streptomycin (final concentration 100 IU/mL) at 37°C in a humidified CO₂ incubator. Stocks were produced by infecting cells at a multiplicity of infection (MOI) of 0.01 and incubating the cells for 72 hours. The culture supernatant was cleared by centrifugation and stored in aliquots at −80°C. Stock titers were determined by preparing 10-fold serial dilutions in Opti-MEM I (1X) + GlutaMAX. Aliquots of each dilution were added to monolayers of 2E04 VeroE6 cells in the same medium in a 96-well plate. Plates were incubated at 37°C for 5 days and then examined for cytopathic effect. The TCID50 was calculated according to the method of Spearman & Kärber. All work with infectious SARS-CoV and SARS-CoV-2 was performed in a Class II Biosafety Cabinet under BSL-3 conditions at Erasmus University Medical Center.

Organoid-derived human airway culture secretions

Organoid-derived human airway culture secretions were harvested from cultures that had been differentiated at air-liquid interphase for 3 weeks as described by Lamers et al. [13]. Secretions could be harvested by pipetting using a P1000 tip and were not diluted. Secretions were stored at -80°C until use. Ten-fold dilutions of virus stock containing 1.21E06 TCID50/ml were made in Opti-MEM I (1X) + GlutaMAX. Next, 25 μl of each virus dilution was mixed with 25 μl of airway culture secretions. Virus was inactivated by adding 50 μl of 2X Laemmli buffer (BioRad) and incubating at 95°C for 10 minutes.

Collection and treatment of patient material samples

Nasopharyngeal swabs from COVID-19 patients were stored in universal transport medium (UTM; contains bovine serum albumin) after collection. Next, they were centrifuged at 15,000 g for 3 min to pellet down cell debris (termed ‘swab pellet’). The swabs were then washed twice with PBS to remove excessive albumin and fixed in 80% acetone (termed ‘swab supernatant’). Sputum from COVID-19 patients was collected and diluted in UTM. Alternatively, sputum was diluted in medium after collection and a few droplets were pipetted on glass slides, dried and fixed in 80% acetone.

The nasopharyngeal and throat swabs and sputum samples were obtained from different patients. Samples of sputum deposited on glass slides were obtained from one single patient.

Proteins present in patient nasopharyngeal and throat swabs or sputum samples in transport medium were first precipitated with acetone-TCA to remove excessive albumin according to [14]. Briefly, 40 μl of the sample was mixed with 400 μl acetone and 1% TCA and left overnight at -20°C. Proteins were pelleted, washed once with ice-cold acetone and left to dry for 5 min. The protein pellet was then resuspended in 40 μl 50 mM Tris/HCl, 4 M urea (pH 8.2) and diluted with 160 μl 50 mM Tris/HCl (pH 8.2).

Cellular human and viral material in sputum deposited on glass slides was lysed in 50 μl 2% SDS dissolved in 50 mM Tris/HCl (pH 8.2) followed by sonication in a Bioruptor Pico (Diagenode). Proteins were digested using the SP3 protocol as described below.

Sample preparation for MS

A 90% confluent T75 flask of VeroE6 was infected at a MOI of 0.3 and incubated for 24 hours at 37°C in a humidified CO₂ incubator. Next, cells were collected by scraping and the medium was removed after centrifuging at 400 g for 5 min. Cells were lysed in 2X Laemmli buffer (final concentration; Bio-Rad) and boiled at 95°C for 20 min to inactivate the virus. Proteins were reduced and alkylated with DTT (Sigma) and IAA (Sigma) and precipitated using chloroform/methanol [15]. The protein pellet was then dissolved in 100 μl of a 50 mM Tris/HCl buffer (pH 8.0) with 2 M urea. Proteins were quantified using the BCA protein kit (ThermoFisher Scientific / Pierce, #23225); peptides were quantified with a quantitative colorimetric peptide assay (ThermoFisher Scientific / Pierce, #23275). Fifty μg of protein was digested with 1 μg trypsin (Thermo) overnight at room temperature. The peptide digest was cleaned on a 50 mg tC18 Sep-Pak cartridge (Waters) and the peptides were eluted with 2 ml acetonitrile/water (1:1) with 0.05% TFA.

Alternatively, proteins were digested with trypsin using the SP3 protocol [16], with minor modifications. Briefly, proteins in 30 μl Laemmli buffer were reduced for 30 min at 50°C with 5 mM DTT and alkylated with 10 mM IAA. A slurry of 10 μg of Sera-Mag speedbeads (GE Healtcare) in 20 μl milliQ/ethanol (1:1, vol/vol) was added to the solution and mixed for 10 min at RT. Using a magnetic rack, the beads were immobilized and washed three times with 100 μl 80% ethanol. 1 μg trypsin and 0.5 μg Lys-C in 100 μl 50 mM Tris/HCl pH 8.3 were added to the beads and the sample was incubated overnight at 37°C. The tryptic digest was then acidified with TFA and desalted using a StageTip. Peptides were eluted with 100 μl 40% acetonitrile and 0.1% formic acid and dried using a Speedvac. Before analysis by LC-MS peptides were dissolved in 20 μl 2% acetonitrile / 0.1% formic acid.

For PRM measurements, peptide samples with concentrations ranging from 0 to 25 ng/μl were prepared from SARS-CoV-2 infected VeroE6 cell lysates. For global proteomics, peptides were fractionated off-line using high pH reversed-phase (ThermoFisher / Pierce, #84868) into four fractions.

Synthetic AQUA peptide analogs containing a heavy stable isotope labeled C-terminal Arginine (R10) residue were purchased from Thermo.

LC-MS

Peptide mixtures were trapped on a 2 cm x 100 μm Pepmap C18 column (ThermoFisher Scientific, #164564) and separated on an in-house packed 50 cm x 75 μm capillary column with 1.9 μm Reprosil-Pur C18 beads (Dr. Maisch) at a flow rate of 250 nL/min on an EASY-nLC 1200 (ThermoFisher Scientific), using a linear gradient of 0–32% acetonitrile (in 0.1% formic acid) during 60 or 90 min. The eluate was directly sprayed into the mass spectrometer by means of electrospray ionization (ESI).

For targeted proteomics, a parallel reaction monitoring regime (PRM) was used to select for a set of previously selected peptides on an Orbitrap Eclipse Tribrid mass spectrometer (ThermoFisher Scientific) operating in positive mode and running Tune version 3.3. Precursors were selected in the quadrupole with an isolation width of 0.7 m/z and fragmented with HCD using 30% collision energy (CE). MS2 spectra were recorded in profile mode in the Orbitrap at 30,000 resolution. The maximum injection time was set to dynamic with a minimum of 9 points measured across the chromatographic peak. See S2 Fig for the isolation list. For global DDA proteomics, data were recorded on an Orbitrap Fusion Lumos Tribrid mass spectrometer (ThermoFisher Scientific) in data dependent acquisition (DDA) mode. Full MS1 scans were recorded in the range of 375–1,400 m/z at 120,000 resolution. Fragmentation of peptides with charges 2–5 was performed using HCD. The collision energy was set at 30% and previously fragmented peptides were excluded for 60 seconds. The resolution of tandem mass spectra (MS2) was set at 30,000 and automatic gain control (AGC) was set to 5E4 and the maximum injection time (IT) was set to 50 ms. MS2 spectra were recorded in centroid mode. The sequence of sampling was blanks first and then in order of increasing peptide input amounts to avoid any contamination of previous samples.

Data analysis

Mass spectrometry data were analyzed using Mascot v 2.6.2 within the Proteome Discoverer v 2.3 (PD, ThermoFisher Scientific) framework or with MaxQuant v 1.6.10.43 (www.maxquant.org), all with standard settings (note: fragment tolerance set to 20 ppm). Raw data recorded on the Orbitrap Eclipse with the FAIMS option were first converted into mzXML format using the FAIMS MzXML Generator software tool (Coon Lab) before MaxQuant analysis. PRM data were analyzed with Skyline (skyline.ms). Spectra and chromatograms were visualized in PD 2.3, Skyline or the PDV proteomics viewer (pdv.zhang-lab.org). The Skyline output was converted to ridgeline plots using in-house developed software. Calibration curves were generated for several endogenous tryptic peptides of SARS-CoV-2 proteins from infected Vero E6 cell samples. The Skyline settings were: regression fit: bilinear through zero; regression weighting: none; normalization method: none; LOD calculation method: bilinear turning point: mac LOQ bias: [left empty]; max LOQ CV: [left empty]; qualitative ion ratio threshold: [left empty]). For global proteome analyses the UniprotKB SARS2 database (https://covid-19.uniprot.org/; 14 entries; May 2020) was concatenated with the UniprotKB database, taxonomy Chlorocebus (African green monkey) or taxonomy Homo sapiens (version Oct 2019).

Ethics statement

The Institutional Review Board of both the Jeroen Bosch Hospital and the Star-shl Medical Diagnostic Center approved this study. The boards approve anonymous use of remnant biospecimens for scientific purposes. All patients were informed of the possibility that residual samples could be used anonymously for research purposes with right of refusal. All samples were anonymized before they arrived at the Erasmus MC Proteomics Center for further analysis.

Detection of SARS-CoV-2 peptides in clinical specimens

As a proof-of-concept experiment we then applied this targeted proteomics technology to detect SARS-CoV-2 proteins in samples from COVID-19 patients. Several different types of patient samples were collected and provided to us by the Erasmus MC diagnostic department. Since all viral infectivity in these clinical specimens needs to be abolished according to established protocols in an BSL-3 facility before they can be further processed, the condition of the starting material was not optimized for subsequent proteomics. Notably, some clinical samples contained high amounts of contaminants such as detergents, albumin, etc. Sputum diluted in viral transport medium deposited on glass slides and then simply fixed in 80% acetone turned out to be the sample type that was most compatible with the subsequent proteomics workflow. Apparently, the relatively simple background matrix composition combined with a sample preparation protocol that does not involve the addition of detergents or albumin offers a substantial advantage for proteomics workflows.

For PRM, we focused on the SARS-CoV-2 Nucleocapsid tryptic peptide AYNVTQAFGR, since this peptide was found to be one of the most prominent and responsive peptides in the SARS2-CoV-2 infected Vero E6 cell lysate. Also, this amino acid sequence is unique to SARS-CoV-2, even in comparison to SARS-CoV.

In order to unambiguously confirm the presence of Nucleocapsid peptides we compared the chromatogram patterns of AYNVTQAFGR with those of a variant of this peptide that contains a heavy isotope labeled C-terminal Arginine. This synthetic AQUA peptide was spiked in all patient samples and co-elutes with the corresponding (non-labeled) endogenous peptide in LC-MS because of its similar biophysical properties. For a selection of clinical samples the reconstructed Skyline PRM chromatograms of the target peptides are shown in Fig 3A (sputum) and 3B (swab). For all samples, the endogenous peptides are shown in the lower panels, while the corresponding AQUA counterpart peptides are shown in the upper panels. The similarities in both fragment ion chromatogram pattern and elution time confirm the presence of SARS-CoV-2 proteins in all sputum samples and sample ‘swab supernatant 4’.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. PRM fragment ion chromatograms of SARS-CoV-2 Nucleocapsid and VME1 tryptic peptides VAGDSFAAYSR and AYNVTQAFGR in representative A) sputum specimens and B) throat swab specimens of COVID-19 patients.

Chromatograms for each of the Top6 fragment ions are shown in different colors. The upper panels show the fragment ion chromatograms of the corresponding synthetic AQUA peptides VAGDSFAAYS[R] (m/z 605.79) and AYNVTQAFG[R] (m/z 568.79). See S4 Fig for additional clinical specimens.

https://doi.org/10.1371/journal.pone.0259165.g003

To investigate the relationship between amounts of viral RNA as detected by PCR methods and protein abundances determined by mass spectrometric methods, we collected two cohorts of clinical specimens with known PCR Ct values ranging from the 12 to >30. These samples were nasopharyngeal Eswabs, Aptima or Sigma swabs from individuals who had tested positive for COVID-19 in regular diagnostic assays. The viral material in these swabs was first inactivated in 80% acetone Swabs and similar proteomics sample preparation procedures were followed as for the sputum samples described earlier. For several target tryptic peptides of SARS-CoV-2 proteins, AQUA peptide counterparts were included in the samples as spike-in. Relative protein abundances were defined by the sum of the AUCs for all fragment ion chromatograms for every peptide of each viral protein detected in a sample.

For the first patient cohort, the mass spectrometry data are shown in Fig 4A and S7 File. For all specimens with E-Gen CT value <20 relatively high mass spectral peak intensities were observed for various target peptides in our PRM assay. Also, for several specimens with Ct values in the low 20s viral protein could still be unambiguously detected. For example, in sample #5 peptide GQGVPINTNSSPDDQIGYYR was identified by eight highly mass accurate fragment ions (Fig 4B). The correlation between the PCR Ct values and the summed mass spectrometry intensities is shown in Fig 4C. There is a clear inverse relationship between these sample characteristics, with a threshold value for detection by targeted mass spectrometry around Ct value 22. For some samples with high Ct values no SARS-CoV-2 peptides could be detected.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4. PRM data of clinical specimens of COVID-19 patients (cohort 1).

A) Total AUCs of SARS-CoV-2 target peptide fragment ion chromatograms (upper panels show the spiked-in AQUA peptide signals; if no AQUA peptide counterpart was available upper panels are left empty). The color shading of the bars indicate the relative AUCs of the different fragment ions. B) Fragment ion chromatograms for various SARS-CoV-2 target peptides in one representative clinical specimen. C) Comparison of AUCs versus PCR Ct values for clinical specimens. Data points in grey represent samples in which no target peptides were detected by PRM.

https://doi.org/10.1371/journal.pone.0259165.g004

The second sample cohort consisted of 15 nasopharyngeal swabs from individuals who had tested positive for COVID-19. Since the two sample cohorts were collected and analyzed at different COVID-19 diagnostic testing sites, the results cannot be directly compared to one another. For this reason, we treat them separately here.

Viral samples were collected with different types of collection kits, such as Eswabs, Aptima and Sigma swabs, and the same sample preparation procedures were followed as for the sputum samples described earlier.

Positive mass spectrometric detection was observed for all specimens with E-Gen CT values <24 (Fig 5), although the absolute summed intensities of target peptide fragments varied widely. Individual peptide identifications and quantifications are visualized in (Fig 5A and 5B and S8 File) and compared to corresponding AQUA peptide counterparts if applicable. Clearly, the highest summed AUCs values were observed for samples 1, 8 and 14, which have very low PCR CT values and thus a relatively high amount of viral material. Strikingly, for several patient samples different sets of target peptides were more pronouncedly detected, although the exact same sample preparation protocols were followed for all samples. This may reflect the heterogeneity of the samples, possibly leading to quite diverse outcomes of protein digestion procedures. For example, in sample #6 several relatively long peptides are highly abundant, while some shorter ones are virtually absent. Whether or not the differences in detected target peptide sets may reflect the status of the viral particle (e.g. active particles versus aggregated, non-assembled proteins from viral residue) is the subject of current research.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 5. PRM data of clinical specimens of COVID-19 patients (cohort 2).

A) Total AUCs of SARS-CoV-2 target peptide fragment ion chromatograms (upper panels show the spiked-in AQUA peptide signals; if no AQUA peptide counterpart was available upper panels are left empty). The color shading of the bars indicate the relative AUCs of the different fragment ions. B) Selection of PRM results for several target peptides in samples 2, 4 and 7. See main text for explanation. Retention times for the identical peptides in different samples may slightly differ as a result of small variations in LC gradients and chromatography setup. C) Comparison of AUCs versus PCR Ct values for clinical specimens. Data points in grey represent samples in which no target peptides were detected by PRM.

https://doi.org/10.1371/journal.pone.0259165.g005

Overall, there is an inverse relationship (R² = 0.733) between ¹⁰Log transformed summed AUCs of the fragment ion chromatograms and the E-Gen Ct values from PCR assays on the same samples (Fig 5D), which makes sense because of the logarithmic nature of the Ct value scale and which reflects the amounts of virus RNA and proteins present. Obviously, the number of data points is only limited in our case and the strength of this relationship is expected to become stronger with an increasing number of data points.

Two target peptides were detected in a sample with a Ct value of 23.2 (Fig 5C; sample #7, peptides GPEQTQGNFGDQELIR and DQVILLNK. Sample #2 is shown for comparative purposes). Strikingly, for one specimen with a Ct value of 29.3 a positive detection was reported for at least one tryptic peptide of VME1 (Fig 5C; sample #4, peptide EITVATSR). When the contaminating peak that was incorrectly assigned as the b3 fragment ion by Skyline is removed from the chromatograms, the pattern closely resembles that of its AQUA counterpart peptide.

Finally, we tested two different experimental procedures to obtain higher sensitivity and to decrease the overall LC-MS analysis time. To increase the measurement sensitivity, high pH reversed phase fractionation was applied to tryptic digests of clinical specimens. Fractionated peptides were collected in eight fractions, which were separately analyzed by PRM MS. This leaded in many cases to improved peptide detection and higher quantitation values, as exemplified in Fig 6A for several representative target peptides (left panels: unfractionated digests, right panels: fractionated digests). Peptide abundances were up to five times higher in the fractionated samples, while absolute quantitation based on comparison to estimated spiked-in amounts of AQUA counterpart peptides revealed that SARS-CoV-2 peptides could be detected in the low to mid-attomolar range. Shorter LC-MS gradient (20 min) resulted in overall slightly less identifications and quantitation results. Still, extremely low abundant target peptides could be reliably identified and quantified, despite the increased presence of contaminating peaks that are most likely the result of more crowded mass spectra (Fig 6B).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 6.

A) Comparison of one-shot versus high pH fractionation LC-MS PRM for several target peptides. For peptide GFYQTSNFR in Sample #7 the normalized peak area would correspond to the low attomolar range. B) Example of a positive target peptide identification in a 20 min gradient LC-MS run of a sample of high Ct value.

https://doi.org/10.1371/journal.pone.0259165.g006

In conclusion, more sensitivity could be obtained by fractionation of tryptic digests prior to PRM analysis, although at the costs of longer analysis times. Shorter LC gradients were used to decrease the overall sample analysis time. While some peptides fell below the detection limit, the far majority of target peptides could still be reliably identified and quantified, also in samples of relatively high Ct values.