Integrated microfluidic cartridge for RIAMs
Our RIAMs platform has three different versions of airborne-virus monitoring systems for different application scenarios, including M-RIAMs for multiple-site sampling and on-site batch analysis, S-RIAMs for continuously real-time monitoring of airborne pathogens without any manual intervention in a specific location, and R-RIAMs for continuously sampling and analysis of bioaerosols in a large indoor environment (Fig. 1b). Inside of the RIAMs is the central biochemical assay of in-situ PCR which provides the ultra-high sensitivity for virus detection. The in-situ PCR utilizes a piece of chitosan-modified quartz filter (QF) paper for nucleic acid extraction followed by PCR amplification directly on the paper (Fig. 1c)28,29. Unlike the conventional solid-phase nucleic acid extraction methods30, which employ a bind-wash-elute protocol, the QF paper carries positive charges in the acidic condition, leading to the adsorption of negative-charged nucleic acid molecules. After a simple wash with water, PCR reagents are directly loaded to soak the QF paper completely. Then, PCR occurs within the QF paper and all the captured nucleic acids can be used for amplification without any loss. This bind-wash-amplification assay not only provides an ultra-high sensitivity for detecting viruses in aerosols, but also simplifies the structures of the integrated microfluidic cartridge for accommodating this process.
Once the central biochemistry is determined, we next designed a fully integrated and enclosed microfluidic cartridge to accommodate the assay using the “needle-plug/piston” mesoscopic design paradigm in the “3D extensible” architecture developed previously by our group (Fig. 1d)31,32. The mesoscopic design paradigm has a series of basic elements, such as IN, OUT, MIX, etc., for various basic fluidic operations (Fig. 1e). For example, the IN element represents the basic fluidic operation of loading a reagent into the microdevice, while the OUT element is for driving a solution out of the microdevice. In the process of designing this cartridge, a reaction chamber containing an embedded piece of QF paper is first designed for nucleic acid extraction and amplification on a planner plastic chip, which is sealed with a piece of pressure-sensitive membrane. By sequentially linking three IN elements to the reaction chamber, the sample loading, the washing, and the PCR reagent loading can be realized in the cartridge. An OUT element is linked to the reaction chamber as well to function as a waste reservoir. A MIX element can replace the IN element for the PCR mixture loading to realize the reconstitution of the lyophilized reagents in the device. After the initial fluid function verification, the microfluidic cartridge is further modified for injection molding in order to achieve mass production (Supplementary Fig. 1). To operate this microfluidic cartridge, we next built a compact control and detection instrument using a modular design structure. This instrument consists of a fluidic actuation module, a PCR thermal cycle module, and a four-color fluorescence detection module and can process two microfluidic cartridges at the same time (Fig. 1f). Once the cartridge is inserted into the instrument, the entire analysis process can be automatically conducted by the instrument.
Ultrasensitive detection of SARS-COV-2, Influenza A, B, and respiratory syncytial virus
We first tested the performance of the QF paper for RNA extraction and PCR amplification by using a simplified microdevice containing a chamber with a piece of QF paper. Once the RNA capture is finished on the device, the QF paper is taken out from the chamber and put into a tube for PCR amplification. Here we designed a four-plex PCR system that can simultaneously detect four common human respiratory viruses, including SARS-COV-2, Influenza A, B, and respiratory syncytial virus (RSV) (Table S1, Supplementary Fig. 2a, b). A concentration gradient of four respiratory viruses ranging from 1000 to 5 copies/mL was prepared to measure the limit of detection (LoD). The results showed that our assay achieved the ultra-low LoDs of 10, 5, 5, and 5 copies/mL for SARS-COV-2, influenza A, influenza B, and RSV, respectively (Fig. 2b), with a coefficient of determination of r2 = 0.72 for SARS-COV-2, r2 = 0.78 for Influenza A, r2 = 0.87 for Influenza B, and r2 = 0.71 for RSV (Fig. 2c). We also verified that there are no cross-reactions or non-specific amplifications in the four-plex PCR system using a series of mixture samples (Fig. 2d).
Next, we test the sensitivity of the assay conducted in the cartridge using the compact instrument. The analytical process is shown as follows (Fig. 2a and Supplementary Video. 1): 1-mL sample in lysis buffer is first driven through the modified QF paper, where RNA is captured, by the actuation plunger in the instrument. After that, 800-µL of DEPC water is washed through the paper followed by the loading of 50-µL PCR mixture into the chamber. Finally, the OUT-waste container is closed and the thermal cycling is conducted in the chamber. The amplification curves can be read out through the fluorescence intensity emitted by the QF paper during the amplification reaction. We prepared gradient samples containing varying amounts of pseudoviruses of these four respiratory viruses ranging from 1000 copies down to zero. The on-cartridge results showed that our system achieved a remarkable sensitivity with a LoD of 10 copies/mL for all four respiratory viruses (Fig. 2e, and Supplementary Fig. 2c). We also tested the detection of the Omicron variant of SARS-CoV-2 using the RIAMs, showing that 20 copies/mL of Omicron could be detected with a good linear correlation between Ct values and RNA concentration (Fig. 2f and Supplementary Fig. 2d). With further optimization of the primers, this sensitivity can be improved to 10 copies/mL.
M-RIAMs for ultrasensitive virus-laden aerosol monitoring with sub-single molecule spatial resolution
To analyze airborne viruses in aerosols, we employ a cyclone aerosol sampler with a high air flow rate of 400 lpm to collect aerosol particles27. As the air swirls in the cyclone pipe, the particles hit the pipe wall due to the centrifugal force generated by the swirling motion and fall into the bottom collection tube containing virus lysis buffer (Supplementary Fig. 3a). By using polystyrene latex (PSL) microspheres with various diameters, we confirmed that the sampler is capable of collecting particles as small as 200 nm (Supplementary Fig. 3b). When the diameter exceeds 2 µm, the collection efficiency can reach nearly 100% with a cutoff diameter of 0.8 µm (Supplementary Fig. 3c).
Here we proposed a multi-site sampling RIAMs (M-RIAMs) containing multiple aerosol samplers for the application where the aerosols can be collected in several locations and then analyzed in a batch. Since the samples must be transported from the sampling to the analysis site, we designed that the sample container of the cartridge can be directly used as the collection tube in the sampler. As a result, once the aerosol collection is completed and the sample container is capped at the sampling site, there is no need to open the cap again in the rest of the analysis to guarantee the safety of the operator and to eliminate the risk of sample cross-contamination (Fig. 3a). Under a 30-minute sampling time, the system can achieve an ultrasensitive detection of airborne viruses with a sub-single molecule spatial resolution of 0.83 copies/m3, which is the highest spatial resolution reported so far. We investigated whether the detection results could be affected by unknown airborne impurities in different environments. We conducted the aerosol collections for 30 minutes in several locations, including public restroom, offices, corridors, and parking lots. Then, we added SARS-CoV-2 pseudovirus to the concentration of 50 copies/mL and analyzed in the cartridges. We found that only the samples collected from parking lots showed some false negative results (Fig. 3b), which can be resolved by adding a piece of filter cotton to the air inlet of the cyclone sampler (Fig. 3c).
Large-scale analysis of COVID-19 environmental samples using M-RIAMs
We conducted a comprehensive evaluation of the M-RIAMs with aerosol and surface swabbing samples collected from COVID-19 patient wards in Peking University First Hospital (Fig. 3d). We first collected aerosol samples using the cyclone aerosol sampler before and after the enhanced ventilation in the same patient ward. We then analyzed these samples and found that all four samples were positive before the enhanced ventilation, while only one out of four tests remained positive after ventilation (Fig. 3e), suggesting that the aerosol sample analysis can effectively reflect the risk of aerosol transmission. Next, we collected various environmental samples in wards to assess the effectiveness of the aerosol analysis compared with the conventional surface swabbing method for the assessment of infection risks. In each room, the aerosol samples and six surface swabbing samples from the bedside table, the patient collar, the bedding, the patient skin, the medical monitor, and the pager were repeatedly collected 5 times at 1-hour intervals. Ultimately, we tested a total of 210 environmental samples from five wards on different days. From the violin plots of Ct values of those collected samples, we noticed that the median Ct values of the positive aerosol samples was smaller than those of surface swabbing samples (typically 1–2 units lower than those of the swabbing samples) (Fig. 3f). In addition, we found that indoor aerosol samples showed the highest positivity rate of 60% (Fig. 3g), suggesting that aerosol monitoring could be a more sensitive tool for detecting environmental viruses than surface swabbing method. The aerosol samples can achieve the full coverage of the space and are not affected by sampling site bias, allowing a more objective assessment of the infection risks.
In addition, we looked into the positive detection rates of the samples collected on day 1 and 3 of the same ward and found that the positive detection rate of the aerosols matched the patient's condition very well. As the patient's condition was improved, the positive detection rate dropped from 100–20%, while other environmental samples did not show a similar discernible pattern (Fig. 3h). Furthermore, we found that the aerosol samples collected from the wards with severe COVID-19 patients have a higher positive detection rate than those from the wards with mild COVID-19 patients, while this correlation was not observed in other environmental samples (Fig. 3i). These results highlight the potential of the aerosol monitoring for accurately reflecting the patients' disease status (in terms of individual viral shedding).
S-RIAMs: design and performance evaluation for real-time virus-laden aerosol monitoring
We next developed a stationary real-time RIAMs (S-RIAMs) for continuously monitoring of airborne pathogens without any manual intervention in a specific location. The S-RIAMs consists of four modules: a cyclone bioaerosol sampler, an automated sampling unit, a loading tray which can store up to 8 cartridges, and a control and detection system (Fig. 4a, b and Supplementary Fig. 4). The S-RIAMs can be modified and assembled on a robotic chassis to form the roaming real-time RIAMs (R-RIAMs) for continuously sampling and analysis of bioaerosols in a large indoor environment (Fig. 4c and Supplementary Fig. 5). Unlike the M-RIAMs which collects aerosols into a sample tube, the automated sampling unit in the S-RIAMs and R-RIAMs can quantitatively inject the sample liquid from the collection tube of the sampler into the sample container of the cartridge via a needle using a peristaltic pump (Fig. 4d). Once the sample is loaded, the loading tray can move the cartridge to the multiplex PCR detection system for the virus detection (Supplementary Video. 2). We verified the adsorption of nucleic acids on the surface of the silicone tubing and other components used for liquid transfer within S-RIAMs and R-RIAMs. Results showed that the Ct values of these samples did not change significantly, indicating no nucleic acid adsorption within the sampling system (Fig. 4e).
We then evaluated the capability of the S-RIAMs for monitoring the airborne transmission of SARS-CoV-2 using a Whole-Body Inhalation Exposure System (WIES), which can simulate real-world aerosol environment by generating consistent aerosol particles through the compression of a medical nebulizer (Fig. 5a and Supplementary Fig. 6a). Using the S-RIAMs, we successfully detected pseudovirus samples with a gradient of concentrations of SARS-CoV-2 generated by WIES with a coefficient of determination r² = 0.81, validating the reliability of the S-RIAMs in aerosol monitoring (Fig. 5b).
In response to the rapid spread of SARS-CoV-2 in aerosols, countries around the world have implemented mandatory measures in the past few years, such as wearing masks in public places and environmental disinfection33. Here, we validated the effectiveness of these measures. We first tested the efficacy of medical masks and N95 respirators in curtailing the aerosol transmission of SARS-CoV-2 by affixing the masks at the aerosol container outlet of the WIES (Supplementary Fig. 6b and 6c). The results proved that both the N95 respirator showed a higher blocking efficacy than that of the medical mask (Fig. 5c). In addition, we also proved the efficacies of the measures, such as the environmental decontamination using potent oxidizing agents and the enhanced indoor ventilation in stopping the spread of COVID-19 (Supplementary Fig. 6d), which in turn demonstrated that our system can perform the high-performance monitoring of airborne viruses.
Real-world deployment of S-RIAMs for ultrasensitive aerosol surveillance of SARS-CoV-2 and RSV
To validate the applicability of the S-RIAMs for monitoring SARS-CoV-2 and other respiratory aerosol viruses in real-world settings, we deployed the S-RIAMs in several real-world locations, including the office workplace, the centralized dormitories, and the pediatric wards with RSV-positive infants in the hospital. During the COVID-19 epidemic in China at the end of 2022, we deployed the S-RIAMs in an office space and configured the system to run continuously for 6 consecutive days from 10 a.m. to 6 p.m with a monitoring interval of 1 hour. In total, 48 aerosol samples were collected and analyzed. From the plot of summarized Ct values, we observed that the mean Ct values of the positive samples in each day initially decreased and then increased, reaching a minimum on the fourth day. The change of the positive detection rate followed a similar pattern, highlighting the capability of the S-RIAMs to continuously and ultra-sensitively detect SARS-CoV-2 in aerosols (Fig. 5d). Next, we conducted the aerosol monitoring around infected individuals within the student dormitory affected by the COVID-19 pandemic. The results showed the clear increasing trend on day 1 and 4, indicating a gradual increase in the number of infected people (Fig. 5e). The results demonstrated that the monitoring of airborne SARS-CoV-2 can reflect the spread and the infection rate in a specific area.
Other than SARS-CoV-2, RSV primarily causes respiratory infections in infants and young children, especially those under two years of age34. Due to the issue of the uncomfortable sampling via nasopharyngeal swab for infants, we believe the aerosol testing could be used as a non-invasive monitoring method of RSV infections. We totally tested 10 aerosol samples from 5 neonatal RSV-positive wards using the S-RIAMs at the Second Affiliated Hospital of Wenzhou Medical University and the all-positive results demonstrated it is possible to monitor the infection spread of RSV. Overall, the monitoring results of S-RIAMs in these real-world settings convincingly demonstrate its capability for highly sensitive and continuous airborne virus detection, enabling the accurate assessment of the viral risk in a specific location.