Indoor Dust as a Matrix for Surveillance of COVID-19

Findings.We measured SARS-CoV-2 using quantitative reverse transcription PCR (RT-qPCR), chip-based digital PCR (dPCR), and droplet digital PCR (ddPCR) in samples of bulk dust, passive surface samples, and surface swabs from rooms of individuals with COVID-19. In bulk dust, the SARS-CoV-2 viral concentration had a geometric mean value of 163 copies/mg of dust and ranged from nondetects to 23,049 copies/mg of dust (Fig. 1A). We detected SARS-CoV-2 RNA in 89% of bulk dust, 55% of surface swabs, and 21% of passive surface sampler samples (average among all three detection methods used, Kruskal-Wallis P = 0.02). The ddPCR method detected viral RNA in 97% of bulk dust samples compared to 93% for the chip-based dPCR and 76% for RT-qPCR (Kruskal-Wallis P = 0.37) (Fig. 1B). Across all sample types, the ddPCR method detected viral RNA in 60% of samples compared to 71% for the chip-based dPCR and 29% for RT-qPCR (Kruskal-Wallis P = 0.06).

• Open in new tab
• Download powerpoint

FIG 1

Heatmap displaying the geometric mean number of copies per microliter of RNA template of samples above the detection limit (samples below detection limit were excluded) (A) and the proportion of samples positive (as a percentage) (B) for bulk dust, surface swabs, and passive sampler as observed with three PCR-based methods. The variable n is the number of samples used to calculate the geometric mean (positive detects) in panel A or proportion of samples positive in panel B. For “Floor Dust,” n refers to the total number of bulk dust samples tested for SARS-COV-2, including the dust samples tested weekly over a period of 4 weeks. Items in white on the room diagram were not sampled, and items in white with a slash in the lower heatmaps were not detected. Colors on the room diagrams represent the average value among all three measurement methods.

The COVID-19 isolation rooms were treated with a chlorine-based disinfectant prior to dust collection as part of the normal cleaning process, and the disinfectant is expected to largely inactivate the virus through reactions with the viral capsid (15). The bags were stored in the laboratory at room temperature after collection. Triplicate subsamples were extracted, and viral RNA was measured immediately upon collection and once per week for 4 weeks. Viral RNA did not measurably decay over 4 weeks in the vacuum bags (regression R² = 0.009, P = 0.47) (Fig. 2A). The coefficient of variance (CV) for number of copies/mg of dust ranged from 73.5 to 313.4% within each vacuum bag when averaged across the three methods of viral detection. This large variation in viral concentration is likely due to the heterogeneous mixture in the bags (Fig. 2B to D).

• Open in new tab
• Download powerpoint

FIG 2

(A) RNA concentration of bulk dust samples (average of four bags) from initial collection to 4 weeks as measured by ddPCR, dPCR, and qPCR. Error bars shown represent the 95% confidence interval of each measurement. (B to D) Cumulative normal probability plots for each measurement method show variability of RNA concentration values for each bulk bag collected.

Experimental protocol. (i) Overview.Samples were collected from two different homes, as well as isolation rooms used to quarantine individuals who tested positive for SARS-COV-2. Bulk dust was collected from both homes, and four composite samples were collected from 30 to 50 student isolation rooms each. Surface swabs and a passive sampler collection were completed in one home. Viral RNA was measured using RT-qPCR, chip-based digital PCR, and droplet digital PCR.

(ii) House surface swabs and passive sampler collection.Surface and passive samples were collected at the end of the 10-day quarantine period from two bedrooms of individuals who tested positive for SARS-CoV-2 in house 1. Surfaces were swabbed using sterile flocked swabs (Puritan, ME, USA), and passive samplers consisting of carpet coupons and polystyrene coupons were placed on the floors of their isolation rooms. The passive samplers consisted of three cut pile carpet squares (fiber length, 10 mm), three loop pile carpet squares (fiber length, 7.5 mm), and three polystyrene squares attached to a template. All squares were 5 cm × 5 cm each. Both carpet types used were made of 100% polyethylene terephthalate (PET) fibers and a synthetic jute backing material. Fibers were specifically manufactured to contain no antimicrobial, stain, or soil resistance coatings. Swabs were dipped in autoclaved phosphate-buffered saline (PBS), and each was used on a different 10 cm × 10 cm surface area. Each wetted swab was wiped left and right across the 10 cm × 10 cm surface area, rotated 1/3 turn, and wiped to cover the surface area up and down, rotated a final 1/3 turn, and wiped in circular motions across the surface area. Swabs were placed back in the corresponding tube and resealed until extraction.

In room 1, two swabs were used for the desk, two for a bedside table, and two for a computer. A passive sampler was placed on the floor by the bed for 4 days. In room 2, two swabs were used for 100-cm² areas on a computer, two for the same areas on a desk, two for the same areas on a second desk, one for the doorknob of the bedroom, two for 100-cm² areas on the bathroom counter, and two on the bathroom doorknob, one for each side of the door. A passive sampler was placed on the floor between the desk and bed for 2 days, and another was placed on the open space in the bedroom for 4 days.

(iii) House bulk dust.Bulk floor dust was collected from occupant vacuum bags of two different houses that had individuals infected with COVID-19. House 1 had floor dust collected 27 days after quarantine ended. House 2 had floor dust collected in the middle of the quarantine period.

(iv) Isolation room bulk dust.Bulk dust samples were retrieved from four different vacuum bags used to clean the isolation rooms for students with COVID-19 at Ohio State University in Columbus, OH, USA, and extracted over 4 weeks. Vacuum bags were collected by cleaning staff from rooms that were used to house students who tested positive for SARS-COV-2. One or two students would isolate in the rooms for 10 days after a positive diagnosis. The cleaning staff would vacuum and clean the rooms after the quarantine period was over and within 18 h of the students leaving the rooms. Cleaning staff would spray the room with an electrostatic sprayer containing a disinfectant (sodium dichloro-s-triazinetrione, CAS 2893-78-9) and wait at least 20 min prior to vacuuming. This disinfectant provides free chlorine (stabilized by cyanuric acid), which nonselectively oxidizes biomolecules to inactivate pathogens. It is possible the disinfectant may be depleted by reacting with other organic material and biomolecules (dead skin, etc.) and viral capsids in dust samples before impacting viral RNA (15). Cleaning staff used a Windsor Sensor XP12 vacuum (Kärcher, Denver, CO) to collect dust over a 3- to 4-week period. Each vacuum bag contained dust from approximately 30 to 50 isolation rooms as well as hallways, and potentially from surfaces in the isolation rooms if considered dusty. The isolation room flooring was vinyl composite tile, and the hallway flooring was wall to wall carpet.

(v) RNA extraction.Viral RNA was extracted from dust and surface samples. Bulk dust samples and surface swabs were extracted using a Qiagen RNeasy Powermicrobiome extraction kit procedure (Qiagen, Hilden, Germany) modified to include 10 times the recommended concentration of 2-mercaptoethanol and using phenol-based lysis. Triplicates of approximately 50 mg of dust were removed from bulk dust samples using an autoclaved spatula, and each replicate was extracted individually in a laminar flow biosafety cabinet. The spatula was flame sterilized between removing replicates. Isolation room bulk dust was extracted over a period of 4 weeks in which triplicate dust samples were extracted from the same bag after initial collection and then again each week for 4 weeks for a total of 60 samples of this type. Bulk dust from student isolation rooms was stored in sealed bags and kept at a room temperature of approximately 22.8°C with a room relative humidity that fluctuated from 15 to 30%. Dust was not sieved due to biosafety concerns. Swabs were placed directly into the lysis tubes for extraction. Carpet samples from the passive sampler were extracted using the QIAmp DSP Viral RNA minikit (Qiagen, Hilden, Germany). A 3 cm × 1 cm area was cut out of the middle of each carpet to reduce potential edge effects and vortexed for 1 min in 4,000 μl of autoclaved PBS. A total of 140 μl of this wash liquid was used in the RNA extraction. Swabs were dipped in autoclaved PBS and wiped horizontally and vertically across the polystyrene pieces on the passive sampler. All extraction sets included a blank to detect potential contamination. The RNA extract of a prepandemic dust sample collected in September 2019 was also tested and shown to be negative for SARS-CoV-2 on RT-qPCR with no amplification.

(vi) Viral detection. (a) RT-qPCR.The viral detection assay targeted the N1 gene using the IDT SARS-CoV-2 (2019-nCoV) CDC qPCR probe assay (Integrated DNA Technologies, Inc., Coralville, IA, USA). This assay uses the 2019-nCoV_N1 forward primer (GAC CCC AAA ATC AGC GAA AT), the 2019-nCov_N1 reverse primer (TCT GGT TAC TGC CAG TTG AAT CTG), and the 2019-nCoV_N1 probe (6-carboxyfluorescein [FAM]-ACC CCG CAT/ZEN/TAC GTT TGG ACC-3' Iowa Black FQ [3IABkFQ]). Direct one-step real-time qPCR amplification of cDNA was performed using qScript XLT one-step RT-qPCR ToughMix (Quanta BioSciences, Gaithersburg, MD, USA). Each well contained 5 μl of RNA template, 10 μl of qScript XLT one-step RT-qPCR ToughMix, 1.5 μl of the IDT SARS-CoV-2 forward and reverse primers at 500 nM and probe at 125 nM, and 3.5 μl of sterile deionized (DI) water. The 2019-nCoV plasmid control 10-fold serial dilutions were used as a standard curve to calculate the number of copies per microliter of RNA template, based on plasmid quantification determined by dPCR (see below) (Integrated DNA Technologies, Inc., Coralville, IA, USA). Cycling parameters were set following the instructions supplied by the CDC for qScript XLT one-step RT-qPCR ToughMix (23). Cycling parameters consisted of 10 min at 50°C for 1 cycle, 3 min at 95°C for 1 cycle, and 3 s at 95°C followed by 30 s at 55°C for 50 cycles. Seven no-template controls were tested, and no amplification occurred.

A subset of 10% of samples were tested for inhibition. The RNA template was spiked with positive plasmid control to test for a reduction in signal due to the presence of inhibitors. The spike concentration was 100 times the highest sample concentration determined by qPCR. Inhibition was indicated if there was a delay in expected amplification. Each sample type was tested for inhibition: bulk dust, swab, and passive sampler. No inhibition was detected in any of the sample types except for the carpet wash from the passive sampler, where the inhibition delayed amplification by 1.45 cycles. Diluting these samples by 10-fold to reduce inhibition would place these samples below the detection limit of 2.3 copies per μl of RNA.

(b) Chip-based dPCR.Digital PCR was performed using the QuantStudio 3D Digital (QS3D) PCR system (Applied Biosystems, Forest City, CA) that utilizes a chip-based technology. This system uses a QuantStudio 3D Digital PCR Chip Adapter kit for the ProFlex Flat Block thermal cycler equipped with a tilt base, which holds the chips (version 2) in place during thermocycling. cDNA was first reverse transcribed from RNA samples using the iScript cDNA Synthesis kit (Bio-Rad, Hercules, CA) according to the recommended reaction protocol on the ProFlex PCR system (Applied Biosystems, Forest City, CA). RNA was detected and quantified using the N1 assay described above. Each reaction was prepared as a 15-μl volume consisting of 2.00 μl of water, 7.25 μl of QuantStudio 3D Digital PCR Master Mix v2 (Applied Biosystems, Forest City, CA), forward and reverse primers at 500 nM, probe at 125 nM, and 5 μl of RNA extract. A portion (14.5 μl) of the solution was transferred into the sample loading port of the loading blade and then loaded onto the chip. Immersion fluid was used to cover the surface of the chip. The chip was then sealed, and additional immersion fluid was added to fill the chip case. Thermal cycling consisted of 10 min at 96°C, 39 cycles of 60°C for 2 min followed by 98°C for 30 s, and finally 60°C for 2 min. The cover temperature was set at 70°C, and the reaction volume was set at 1 nl. Each experiment included one negative control and one N1 positive control (2019-nCoV_N_Positive Control; Integrated DNA Technologies, Inc., Coralville, IA, USA). Chips were then removed and imaged using the QuantStudio 3D digital PCR instrument following thermal cycling. Manual thresholding and quantification were performed using the QuantStudio 3D AnalysisSuit Software. The 95% limit of blank for the N1 assay was determined to be 1.45 gene copies per μl of reaction mixture using seven replicates of negative controls. Inhibition was not assessed for dPCR.

(c) ddPCR.Droplet digital PCR was performed using the Bio-Rad QX200 system along with a C1000 Touch thermal cycler (Bio-Rad, Hercules, CA). SARS-CoV-2 RNA was detected and quantified using the N1 assay previously described. Inhibition was assessed by spiking a subset of sample extracts (n = 17) with bovine respiratory syncytial virus (BRSV) RNA extracted directly from a live attenuated bovine vaccine (Inforce 3 cattle vaccine; Zoetis, Parsippany-Troy Hills, NJ) using a Qiagen PowerViral AllPrep DNA/RNA kit (Hilden, Germany). BRSV RNA was detected and quantified using an assay targeting the nucleoprotein gene with forward primer (GCA ATG CTG CAG GAC TAG GTA TAA T), reverse primer (ACA CTG TAA TTG ATG ACC CCA TTC T), and probe (FAM-ACC AAG ACT/ZEN/TGT ATG ATG CTG CCA AAG CA-3IABkFQ) (Integrated DNA Technologies, Inc., Coralville, IA, USA) (24). Each N1 ddPCR reaction mixture was prepared as a 22-μl volume consisting of 5.45 μl of water, 5.45 μl of one-step RT-ddPCR Supermix (Bio-Rad, Hercules, CA), 2.1 μl of reverse transcriptase, 1.05 μl of dithiothreitol, forward and reverse primers at 1,000 nM, probe at 250 nM, and 5 μl of RNA extract. BRSV wells were prepared in the same manner except that forward and reverse primers at 900 nM and probe at 250 nM were used. A volume of 20 μl of each reaction mixture was passed into droplet generation. Thermal cycling was performed with reverse transcription for 60 min at 50°C, followed by 10 min at 95°C, 40 cycles of 95°C for 30 s followed by 59°C for 1 min, and finally 98°C for 10 min. Each ddPCR experiment included two no-template controls each for BRSV and N1 and two positive controls each for BRSV (RNA and molecular water) and N1 (2019-nCoV_N_Positive control; Integrated DNA Technologies, Inc., Coralville, IA, USA). Manual thresholding and quantification were performed in QuantaSoft version 1.7.4 (Bio-Rad, Hercules, CA) so that no-template controls yielded no positive droplets.

The 95% limit of detection for the N1 assay was determined to be 3.3 gene copies per ddPCR reaction using a 10-replicate dilution series of synthetic SARS-CoV-2 RNA control material (catalog no. MT188340; Twist Bioscience, San Francisco, CA) with a cumulative Gaussian distribution fit to the observed proportion of the replicates positive along the dilution gradient. There was no evidence of inhibition as no difference was observed in the quantification of BRSV RNA in sample extracts compared to the BRSV positive controls (two-tailed t test, P = 0.19).

Statistical and data analyses.Our goal was to compare measurement of SARS-CoV-2 in bulk dust, on surface swabs, and on a passive sampler using three different measurement methods. Each vacuum bag of dust was sampled and extracted in triplicate at each time point (immediately after collection and 1, 2, 3, and 4 weeks after collection). All three detection methods (qPCR, dPCR, and ddPCR) analyzed the same sample extractions for all sample types. Differences in positivity rates among detection methods and sample types were assessed using the Kruskal-Wallis H test. Detection limit information is described above for each detection method. The geometric mean was reported for quantification of SARS-CoV-2 RNA present in samples using each method due to the logarithmic nature of PCR-based data. Potential RNA decay over the 4-week time period was evaluated in bulk dust with a regression analysis on the ddPCR data transformed with the natural logarithm. The data set is available at https://doi.org/10.5061/dryad.3n5tb2rg1.