key: cord-0013167-3ujnmxhx authors: Rufino de Sousa, Nuno; Shen, Lei; Silcott, David; Call, Charles J; Rothfuchs, Antonio Gigliotti title: Operative and Technical Modifications to the Coriolis(®) µ Air Sampler That Improve Sample Recovery and Biosafety During Microbiological Air Sampling date: 2020-05-29 journal: Ann Work Expo Health DOI: 10.1093/annweh/wxaa053 sha: 8418dc34892b943ecec5312d03ff201a9c6adb1a doc_id: 13167 cord_uid: 3ujnmxhx Detecting infectious aerosols is central for gauging and countering airborne threats. In this regard, the Coriolis(®) µ cyclonic air sampler is a practical, commercial collector that can be used with various analysis methods to monitor pathogens in air. However, information on how to operate this unit under optimal sampling and biosafety conditions is limited. We investigated Coriolis performance in aerosol dispersal experiments with polystyrene microspheres and Bacillus globigii spores. We report inconsistent sample recovery from the collector cone due to loss of material when sampling continuously for more than 30 min. Introducing a new collector cone every 10 min improved this shortcoming. Moreover, we found that several surfaces on the device become contaminated during sampling. Adapting a high efficiency particulate air-filter system to the Coriolis prevented contamination without altering collection efficiency or tactical deployment. A Coriolis modified with these operative and technical improvements was used to collect aerosols carrying microspheres released inside a Biosafety Level-3 laboratory during simulations of microbiological spills and aerosol dispersals. In summary, we provide operative and technical solutions to the Coriolis that optimize microbiological air sampling and improve biosafety. Mycobacterium tuberculosis, measles virus, influenza virus, severe acute respiratory syndrome coronavirus-2, and other highly contagious human pathogens transmit through air either by aerosols or droplets (Riley et al., 1978; Bloch et al., 1985; Remington et al., 1985; Fennelly et al., 2004; Yang et al., 2011; Cowling et al., 2013; Patterson et al., 2017; Meselson, 2020) . These airborne pathogens pose a heavy burden on society by incurring severe clinical outcomes including death. Airborne microbes are of particular concern in enclosed and crowded environments, where occupants are readily exposed to respired air and thus at risk of inhaling infectious bioaerosols carrying viruses, bacteria, or fungi. This is well-recognized during infection with M. tuberculosis where congregate settings such as prisons, homeless shelters, slums, and refugee camps are recognized hotspots of transmission (WHO, 2009) . The ability to monitor pathogens in air is an important investment for gauging and controlling infectious disease in society. Microbiological air-sampling tools coupled to analytical methods enable detection of pathogens in air and as such improve our position to investigate the transmission of airborne pathogens and to counter airborne threats through capacitybuilding, infection control measures (Mubareka et al., 2019) . In particular, there is an outstanding need to monitor pathogens in air in critical infrastructure such as government buildings, hospitals, mass transit, and airports, during manufacturing in clean rooms and 'ready-to-eat' food preparation, to name but a few. There is also a growing interest in detection of infectious bioaerosols for point-of-care diagnostics (Fung and Mykhaylova, 2014) . Most air samplers operate by filtration, electrostatic precipitation, or inertial/gravitational collection (Fronczek and Yoon, 2015; Sharma Ghimire et al., 2019) . Inertial/gravitational-type samplers include impactors, impingers, and cyclones and are heavily used for microbiological air-sampling purposes. The Coriolis ® µ (Bertin Instruments, Montigny-le-Bretonneux, France) is a state-of-the-art, high-volume cyclone sampler that collects airborne particles into liquid. The cyclonic sampling mechanism is based on the principle of a liquid-scrubber aerosol collector with tangential impingement into liquid (Decker et al., 1969) . Briefly, the unit pulls air into a collector cone containing buffer to form a vortex that traps particles through centrifugal force onto the inner walls of the collector cone, separating particles from the air and concentrating them in the collection buffer. A diagram of the principle is available in the Bertin user manual (05027-006-DU002-F ENG) and in original reports (Carvalho et al., 2008; Gómez-Domenech et al., 2010) . The Coriolis has tactical capacity and produces a sample that is amenable to different analytical methods. It is gaining ground as a tool for microbiological air sampling where it has been successfully used to detect toxins or microbial contaminants in the food industry (Verreault et al., 2011; Viegas et al., 2012 Viegas et al., , 2014 Ahmed et al., 2013; Chang et al., 2019) , fungus and pollen in outdoor and indoor air (Carvalho et al., 2008; Gómez-Domenech et al., 2010; Chang et al., 2019) and airborne pathogens in healthcare facilities (Choukri et al., 2010; Bellanger et al., 2012; Le Gal et al., 2015; Montagna et al., 2017; Alsved et al., 2019; Montagna et al., 2019) . In the above studies the Coriolis was used mainly in conjunction with culture and nucleic acid amplification methods, although microscopy and in some cases spoligotyping and sequencing were also performed. In addition, flow cytometry, solid-phase cytometry, and immunoassays have been employed to detect aerosolized microbes or their simulants collected on the Coriolis (Langer et al., 2012; Le Goff et al., 2012; Rufino de Sousa et al., 2020) . The Coriolis has also been used to detect airborne traces of the explosive trinitrotoluene through an electrochemical assay (Caygill et al., 2013) . Liquid loss during air sampling is known to occur with traditional impingers (Willeke et al., 1998; Han and Mainelis, 2012) and has also been reported for the Coriolis (Carvalho et al., 2008; Chang et al., 2019) . During impingement, liquid loss may be coupled to the escape and reaerosolization of material from the collection media (Grinshpun et al., 1997; Riemenschneider et al., 2010; Han and Mainelis, 2012) . This raises similar concerns for the Coriolis. Indeed, unintentional reaerosolization of a microorganism during sampling may contaminate sampler parts and expose the operator to a pathogen threat. Despite its increasing use in different microbiological air-sampling applications, investigation on actual Coriolis performance and biosafety concerns during operation have not been thoroughly addressed. Herein we have evaluated the Coriolis in a series of aerosol collection experiments with microspheres and bacterial spores under controlled laboratory conditions. We report on a sampling protocol to maximize sample recovery from the unit and a high efficiency particulate air (HEPA)-filter adaptation to reduce unintentional contamination of device parts which occurs as a consequence of reaerosolization of collected material during sampling. We demonstrate the use of the modified Coriolis in the detection of aerosols generated during a simulated laboratory spill and aerosol dispersal. Lyophilized endospores of Bacillus atrophaeus var. globigii (Bg) (from ECBC Pine Bluff Arsenal Laboratories, US Army, originally given to D. Silcott) were resuspended in sterile deionized (DI) water. Quantification of Bg colony-forming units (CFUs) in stocks and aerosol samples was determined by culture on Luria-Bertani (LB) Miller agar (Sigma-Aldrich) for 1 day at 37°C. Stock solutions of Bg were diluted to 1 × 10 9 CFUs ml −1 and stored at 4°C until further use. Stock solutions of 1 µm, yellow-green (505/515) fluorescent, polystyrene FluoSpheres™ (ThermoFisher) were also prepared in DI water at 1 × 10 9 beads ml −1 , stored at 4°C and protected from light until further use. A Coriolis ® µ cyclonic air sampler (Bertin Instruments, Montigny-le-Bretonneux, France) ( Fig. 1A ) was used to collect aerosolized microparticles in controlled, aerosol dispersal experiments. A bespoke solution for integrating a HEPA-filter system to the Coriolis was conceived and assembled as illustrated ( Fig. 1B) with necessary filter component, stainless-steel metal tubing, and other fittings ( Supplementary Fig. S1 , available at Annals of Work Exposures and Health online edition) (all McMaster-Carr, Elmhurst, IL, USA). The tube fitting that connects the air intake headpiece to the HEPA filter is similar in area to the commercial tubing supplied by Bertin Instruments. Air-flow measurements were made on the Coriolis with a TSI 4040 mass flow meter (TSI, Shoreview, MN, USA) connected to the blower inlet of the Coriolis using thinwall latex tubing, stretched such that a tight seal was obtained on both the air-flow meter and the Coriolis inlet. Measurements were made at the manufacturer-specified flow rate setting of 300 L air min −1 . The flow rate measured was 298 L air min −1 without a collector cone (Fig. 1A) connected to the device. With a collector cone loaded with 15 ml connected to the Coriolis the flow rate measured was 270 L air min −1 . With the HEPA filter attached to the unit and loaded with a collector cone containing 15 ml, the flow rate at the inlet was 250 L air min −1 . Contained aerosol dispersal experiments were performed inside a large, airtight, flexible polyvinyl chloride enclosure mounted on a metal-support frame (Solo Containment, UK). The enclosure measures 270 cm (L) × 165 cm (W) × 255 cm (H) with an inner volume of 9.3 m 3 (Rufino de Sousa et al., 2020). It was assembled and kept inside a Biosafety Level (BSL)-2 laboratory. Particles were purged from the chamber before the start of experiments by drawing air into the enclosure through a HEPA filter using an Attix 30 industrial-grade vaccuum cleaner (Nilfisk, Sweden). FluoSpheres or Bg were aerosolized into the chamber using a 4-jet Baustein atomizing module (BLAM) nebulizer (CH Technologies, Westwood, NJ, USA) operated in multipass mode and producing quasi-monodisperse aerosols with a particlesize diameter range of 0.7-2.5 µm. Aerosolization was performed for 1 min. Previous tests established a steady concentration of particles in the enclosure 8 min after completing the aerosolization cycle on the BLAM (Fig. 2C) and (data not shown); air sampling was therefore routinely initiated 8 min after finishing the aerosolization cycle. In certain experiments, a Lighthouse Handheld 3016 particle counter (Lighthouse Worldwide Solutions, Medford, OR, USA) or an IBAC fluorescent particle counter (FLIR Systems, Orlando, FL, USA) was used to measure the decay of aerosolized FluoSpheres inside the enclosure. The Coriolis was placed on a table 30 cm away from the BLAM aerosol port with the collector part of the unit approximately 100 cm from the ground. As per the manufacturer's recommendation, the collector cones were filled with 15 ml DI water and kept sealed. After completion of the aerosolization cycle, a collector cone was unsealed, loaded onto the Coriolis and the sampler operated at the manufacturer-designated flow rate of 300 L air min −1 . Collection on the Coriolis was performed using the same cone for the entire sampling duration and referred to as standard sampling. In this setting an injection port was used to manually replenish the collector cone back to 15 ml of DI water after every 10 min of sampling. Alternatively, a new collector cone containing 15 ml of DI water was replaced after every 10 min of sampling, referred to as cumulative sampling. At the end of experiments, collector cones were placed in a pass box where the surface of the cones were disinfected with 70% ethanol before being removed from the chamber. The entire chamber was then decontaminated with 35% hydrogen peroxide vapor using a BQ-50 unit (Bioquell, UK). Before starting this procedure, the removable metal piping was disassembled from the body of the Coriolis and the Coriolis operated in decontamination mode. Metal piping was cleaned in mild detergent and autoclaved. Laboratory spill and aerosol dispersal simulations were performed inside a suite of the BSL-3 facility at Biomedicum, Karolinska Institutet, Solna, Sweden. The suite dimensions measure 1000 cm (L) × 300 cm (W) × 270 cm (H). The facility operates at approximately 20-27 air-changes-per-hour (ACH) depending on the number of microbiological safety cabinets in simultaneous use. Safety cabinets in the suite were inactive during our experiments, lowering forced ventilation parameters to about 18 ACH. Our experiments were done before the BSL-3 was opened to users and pathogens introduced to the facility. For spill simulations, the Coriolis was placed approximately 10 cm from the planned spill site and rested either on top of a working bench (90 cm from the ground) or just above the ground (30 cm). A large microbiological spill was simulated by tumbling a container carrying 0.5 l of FluoSpheres (2 × 10 6 beads ml −1 in DI water, 1 × 10 9 beads in total) from the same working bench resting the Coriolis. For aerosol dispersal experiments, the Coriolis was placed on the working bench 90 cm from the ground. The Coriolis was located 900 cm in front of the BLAM, which was supported on a tripod 100 cm from the ground. The BLAM was loaded with FluoSpheres (1 × 10 9 beads ml −1 ). Aerosolization was performed for 1 min, releasing a maximum of 1 × Figure 2 . Clearance of aerosolized microspheres from air during Coriolis sampling. (A-C) FluoSpheres (1 µm, 1 × 10 9 ) were aerosolized inside the aerosol chamber. Coriolis was turned on or left off (dashed line) and particle counters used to record microspheres in the air. (A) Fluorescent particle counts recorded on an IBAC sensor with Coriolis on. Particle counts recorded on a Lighthouse particle counter with Coriolis on (B) and at steady state with Coriolis off (C). 10 9 microspheres into the room. In both simulations, air sampling was performed on the Coriolis with accompanying HEPA-filter modification (Fig. 1B) for 1 h using the cumulative sampling method described above. Researchers donned full-body Tyvek ® 500 Labo overalls, 3M ® FFP3D respirators, safety goggles, and nitrile gloves (all VWR), providing all-around contact protection from aerosol exposure. At the end of experiments, working surfaces were cleaned with mild detergent and the entire BSL-3 suite was decontaminated with 35% hydrogen peroxide vapor using a BQ-50 unit. At the end of a collection cycle on the Coriolis, the aerosol chamber was purged from airborne particles as described above. The Coriolis was then swabbed using sterile cotton swabs (VWR) at the designated points P1-P11 in Fig. 1C ,D. Sample was extracted from swabs by breaking the cotton-end of the swab into a sterile microcentrifuge tube. 0.5 ml phosphate buffered saline-0.05% Tween-80 was added to the tube, the tube was sealed, incubated at room temperature for 2 min and vortexed for 1 min. The cotton-end of the swab was then aseptically removed from the tube using a pair of sterile tweezers and the sample plated on LB agar for quantification of Bg CFUs. The particle clearance rate for aerosolized FluoSpheres in the aerosol chamber was obtained by dividing the inner volume of the chamber (9300 l) by the amount of time needed to reach a 4-log reduction of the starting material of FluoSpheres aerosolized in the chamber (i.e. 99.99% clearance). The amount of time needed to reach a 4-log reduction was calculated by fitting a onephase decay curve to the particle counts measured in the chamber over time using designated particle counters (IBAC sensor and Lighthouse particle counter). The particle clearance rate for the Coriolis was expressed in L air min −1 . FluoSpheres collected on the Coriolis were quantified on a flow cytometer using CountBright™ Absolute Counting Beads (ThermoFisher). Briefly, a defined number of CountBright beads were added to 1 ml of sample, samples were acquired on a FACS Calibur (BD Biosciences) and analyzed on FlowJo (BD Biosciences). The number of FluoSpheres in the sample was calculated with CountBright™ beads according to the instructions of the manufacturer (ThermoFisher). The significance of differences in data group means was analyzed using Student's t test or analysis of variance where appropriate, with a cutoff of P < 0.05. Rapid decay of aerosolized particles from air during Coriolis sampling A robust air sampler must be able to effectively collect particles from air while also generating a sample that is amenable to downstream analysis. For the Coriolis (Fig. 1A) , we investigated particle collection indirectly by measuring the unit's particle clearance rate, i.e. the rate at which the Coriolis removes airborne particles from air. Fluorescent polystyrene (1 µm) FluoSpheres were aerosolized in the aerosol chamber in the presence of the Coriolis and airborne particles measured in real time with an IBAC sensor and a Lighthouse particle counter, respectively. When the Coriolis was turned on the number of airborne particles recorded in the chamber began to steadily decay and decreased about 1.5 orders of magnitude in 2 h ( Fig. 2A,B) . On the contrary, particle decay was not observed when the Coriolis was left off (Fig. 2C) . With this information the effective clearance rate for the aerosolized microspheres was calculated to approximately 35 L air min −1 . This number was reached with the help of either particle counter. Additional sizediscriminating particle analysis with the Lighthouse particle counter showed that the Coriolis struggled with particles smaller than 0.3 µm (Fig. 2B ), in line with technical specifications reported by the manufacturer (Bertin Instruments). A cumulative sampling method that improves sample recovery during Coriolis air sampling We aerosolized FluoSpheres in the aerosol chamber and investigated their collection on the Coriolis over time to see if this process could be improved. Collection was performed every 10 min for a total of 2 h. The sample volume on the collector cone was manually replenished to 15 ml every 10 min as recommended by the manufacturer. Indeed, although the Coriolis is reported by the manufacturer to be able to collect material for up to 6 h, we found that sample recovery was inconsistent after 30 min of sampling (Fig. 3A) . We hypothesized that the collected material might be escaping the collector cone over time as a consequence of cyclonic sampling. Hence, we modified the sampling procedure by replacing the collector cone after every 10-min cycle with a new one; analyzing each cone individually and cumulatively adding the quantification obtained from each time point to the detection curve. At the end of the 2-h sampling interval we found that this cumulative sampling protocol lead to a 50% improvement of collection compared with the standard, manufacturer-recommended sampling with manual liquid replenishment (Fig. 3A) . Analysis of material recovered from individual time points during cumulative sampling showed that approximately 95% of the FluoSpheres recovered during the 2-h sampling interval were collected within the first 60 min (Fig. 3B) . Still, an appreciable amount of microspheres were collected by the Coriolis during the remaining 60 min of sampling. Overall, our observations suggest that the cumulative sampling protocol may be especially useful for long-term sampling applications. Given that sample recovery decreased with increasing sampling time, we asked whether sampled material redistributed to other parts of the Coriolis during operation. To investigate this, we decided to swab various surfaces of the Coriolis after air sampling to see if any parts other than the collector cone became positive after collection. We chose to aerosolize Bg spores, the Anthrax simulant, since we have successfully extracted Bg by surface swabs in the past (Rufino de Sousa et al., 2020). Thus, we aerosolized Bg spores in the chamber and used the Coriolis to collect Bg bioaerosols. We then investigated regrowth of Bg from different parts of the Coriolis, more specifically, the collector cone and parts P1-P11 according to the schematics in Fig. 1C . We found that many surfaces on the device exposed to air flow were contaminated with Bg. Substantial regrowth of Bg was obtained from the headpiece inlet (P1) where air enters the device (Fig. 4A ). Bacilli were also readily detected at the initial tubing after the collector cone (P2) and importantly, at the air outlet (P11) (Fig. 4A) , suggesting that bacteria may become deposited on the unit's fan as well. This raises overall concerns regarding user safety, deposition of contaminants on the fan over time and the validity of samples analyzed on the unit. In an attempt to improve device function and tactical deployment, we tailored and adapted a HEPA-filter system to the Coriolis (Fig. 1B,D and Supplementary Fig. S1 , available at Annals of Work Exposures and Health online) and repeated Bg aerosol sampling. The HEPA filter significantly reduced Bg contamination from the piping (P7-P10) going into the body of the Coriolis (Fig. 4A) . Spiking the collector cone with Bg confirmed that at least part of this contamination originated from reaerosolization of bacilli in the collector cone, as the pattern of Bg deposition on the Coriolis was similar (Fig. 1A) or with the HEPA-filter adaptation (Fig. 1B) . Surface swabs were obtained from positions P1-P11 on the Coriolis (Fig. 1C,D) . Regrowth of Bg from surface swabs and the collector cone was quantified on LB agar and graphed as total number of CFUs. (B) Bg spores (1 × 10 8 CFUs) were loaded directly into the collector cone, the Coriolis was turned on for 1 h and Bg regrowth investigated as in (A). (C and D) FluoSpheres (1 µm, 1 × 10 9 ) were aerosolized inside the aerosol chamber. Coriolis with the HEPA modification was turned on (dashed line) while particle counters were used to record airborne microspheres in the air. (C) Fluorescent particle counts recorded on an IBAC sensor. (D) Particle counts recorded on a Lighthouse particle counter. Data from five experimental repeats shown. Dots represent individual measurements. Bar graphs depict average of CFUs obtained. Error bars show standard error of the mean. * denotes statistically significant differences between Coriolis with and without the HEPA-filter modification. after aerosol dispersal (Fig. 4A) and spiking (Fig. 4B) . Reducing buffer volume on the collector cone from 15 to 5 ml produced the same result (data not shown), suggesting that reaerosolization was independent of the buffer volume in the cone. Importantly, there was no bacillary regrowth from the air outlet (P11) of the HEPA-modified Coriolis when the collector cone was spiked with Bg (Fig. 4B) , indicating that the fan was protected from contamination in the presence of the HEPA filter. Bg could be detected on the outlet (P11) of the HEPA-modified unit after aerosol sampling (Fig. 4A) . Since the HEPA filter prevents access to the outlet (P11) during spiking, detection here must be due to the high concentration of Bg aerosols inside the chamber, promoting deposition of bacilli onto the outside surface of the outlet, rather than a contamination coming from inside the unit. Lastly, introducing the HEPA filter did not negatively impact on the effective clearance rate of the Coriolis (Fig. 4C,D) . We used the Coriolis with the above operative and technical modifications to investigate aerosols generated during simulated microbiological accidents in a laboratory work place. We simulated spills and aerosol dispersals in a functional BSL-3 infrastructure before it was opened to users. To simulate a large microbiological spill, we dropped a container with 0.5 l of DI water carrying a total of 1 × 10 9 (1 µm) FluoSpheres over the edge of a designated working surface in the BSL-3. A particle counter was used to record particle dispersal from the spill. Airborne FluoSpheres where collected on the Coriolis and quantified by flow cytometry. The impaction of liquid on the ground was accompanied by a detectable peak of 0.3, 0.5, and 1 µm particles close to the ground (Fig. 5A) . To our surprise, the number of particles generated by this large liquid impaction was only marginally above baseline-particle counts in the room. Levels returned to steady state about 10 min after the spill and were altogether undetectable when measured at the height of the working bench from which the spill was generated (Fig. 5A) . Despite the generation of few airborne particles from this simulation, it was nevertheless possible to use the Coriolis in conjunction with flow cytometry to detect FluoSpheres aerosolized from the spill (Fig. 5B ). Next, we tested Coriolis sampling during aerosol dispersal of the same amount of FluoSpheres but on a BLAM aerosol generator. A similar peak with albeit much higher particle counts was observed concomitant with the aerosolization of these microspheres on the BLAM (Fig. 6A) . In line, significant numbers of FluoSpheres were detected by Coriolis air sampling (Fig. 6B ) and followed the general decay of particles in the BSL-3 suite due to forced ventilation. When central ventilation in the suite was intentionally turned off and the experiment repeated, airborne particle counts were increased further and remained elevated for the duration of the experiment without returning to baseline (Fig. 6C) . Consistent with elevated and steady detection of particles in the room when central ventilation was inactivated, the Coriolis collected an elevated, steady number of FluoSpheres in the air (Fig. 6D) . Overall, these spill and aerosol dispersal experiments reinforce the capacity of the Coriolis to enable detection of aerosolized microparticles from settings where these particles are present in not only high but also low amounts. The first line in protective measures against microbiological airborne threats is the ability to detect the pathogen in air, as it enables the mounting of adequate countermeasures in the next step, such as treatment, containment, or disinfection, which greatly limits human exposure, prevents illness, and save lives. This requires air-sampling tools that can be used in conjunction with analysis methods to rapidly detect microbes in air. In this regard, portable, tactical collectors are particularly useful in infection control as they can be widely distributed throughout society for air surveillance and research purposes. The Coriolis ® µ is a commercially available, fieldable solution for air sampling that uses cyclonic technology to collect airborne particles directly into 15 ml buffer. Our study presents operative and technical modifications to the Coriolis to circumvent caveats during sampling and to improve its deployment. Building on the increasing number of applications for this sampler, we reveal its suitability in collecting aerosols generated through simulated spills or experimental aerosolization in a BSL-3 laboratory. Information on pathogen quanta in aerosols or droplets is incomplete even for pathogens that undisputedly transmit through air. Efficient air-sampling tools coupled to sensitive analysis methods are needed to bridge this gap. Quantification of M. tuberculosis in bioaerosols suggests that patients with active tuberculosis expectorate a low number of culturable bacilli (Fennelly et al., 2004 (Fennelly et al., , 2012 Patterson et al., 2017) . Pathogen load in air is thus assumed to be limited for M. tuberculosis. This may not be true for other airborne pathogens, but for air-sampling purposes, low pathogen load is generally assumed. Successful detection of low pathogen numbers may require both continuous monitoring and collecting a large volume of air. The cumulative sampling protocol for the Coriolis improves sample recovery especially during prolonged collection and may thus be useful in this setting. Replacing collector cones instead of replenishing liquid to an existing cone does not impact significantly on the overall workload for manual sampling. Sample collection time needs to be the same regardless of protocol. Cumulative sampling generates more collector cones for analysis, which could increase handling time and analysis cost depending on the method but is an acceptable trade-off for improving detection and reducing cross-contamination. The Coriolis is a high-volume, portable sampler that delivers a liquid sample compatible with culture and many other analysis methods. Unlike the AGI and SKC Biosampler impingers which also collect into liquid, the Coriolis does not require a pump and operates at a much higher flow rate. The advantages of the Coriolis are weighed against its high cost of purchase, energy consumption, and noise during operation. Several studies have used the Coriolis together with impingers or impactors. The comparative data generated have for the most part revealed concordance in detection between the Coriolis and other collectors (Carvalho et al., 2008; Gómez-Domenech et al., 2010; Bellanger et al., 2012; Langer et al., 2012; Ahmed et al., 2013; Chang et al., 2019) . These studies did not formally investigate particle collection efficiency (Ladhani et al., 2017; Kim et al., 2018) . Indeed, useful, bona fide metrics for the latter are difficult to obtain. Important previous work Figure 6 . Aerosol release inside a BSL-3 laboratory. FluoSpheres (1 µm, 1 × 10 9 ) were aerosolized inside a BSL-3 suite in the presence (A and B) or absence of forced ventilation (C and D). A Lighthouse particle counter was used to record airborne particles in the room. Coriolis air sampling was done with a HEPA-modified unit using the cumulative sampling method. has circumvented this by measuring relative sampling efficiency against the SKC BioSampler. Several different air samplers were benchmarked in this way and the Coriolis performed well against the other devices in sampling several aerosol-test agents including Bg and fluorescent microspheres (Dybwad et al., 2014) . In our investigation of Coriolis performance, we evaluated the rate at which airborne microparticles were removed from the air during active sampling and report an effective clearance rate of approximately 35 L air min −1 . This is the rate at which the Coriolis clears 1 µm particles from air and not a direct measure of particle collection on the device. Still, this value could be used to estimate operation time in a given volume. Coriolis collection greatly suffered from the harsh nature of cyclonic sampling as we observed sample loss and reaerosolization from the collector cone. Both could contribute to misleading analysis results. Reaerosolization was responsible at least in part for disseminating collected particles to various other parts of the sampler. This could lead to cross-contamination of samples and also inadvertently expose the user to pathogens during microbiological air sampling. Introduction of a cumulative sampling protocol and a HEPA-filter adaptation to the device improved these shortcomings. The metal fittings are robust and tolerate routine autoclaving and disinfection while the HEPA filter is amiable to hydrogen peroxide favor decontamination. We estimate the loading capacity of the filter to be approximately 63 g of dust. Under normal, intermittent use in laboratory environments, even at the highest flow rate of operation the risk of excessive particle buildup on the filter media would be very low. Out of an abundance of caution we recommend replacing the HEPA filter once per year and opening the filter casing under containment conditions or after first completing decontamination with e.g. hydrogen peroxide vapor. We have recently used the HEPA-modified Coriolis with the cumulative sampling protocol in our aerosol chamber to investigate the performance of a portable electrostatic air sampler for tuberculosis (Rufino de Sousa et al., 2020) . A similar HEPA-modified Coriolis has also been used in a clinical, experimental setting to quantify M. tuberculosis from human bioaerosols (Patterson et al., 2017) . In the current study, we provide a thorough presentation of these technical and operational improvements to the Coriolis and supply details for assembly of the HEPA filter so that others can benefit from this adaptation. The HEPA-modified Coriolis operates otherwise like the standard, commercially available unit. We observed a small increase in Coriolis particle clearance rate upon mounting the HEPA filter. This is probably due to the HEPA filter trapping airborne particles that would otherwise be subject to continuous recirculation through the device during operation. Despite biosafety and other regulatory precautions in place, the research laboratory remains an indoor environment where infections are acquired, albeit unintentionally, due to accidental exposure (Sulkin and Pike, 1951; Pike et al., 1965; Pike, 1976) . We thought it interesting to employ the Coriolis with improvements in the assessment of simulated incidents in the laboratory coupled to microbiological exposure. Here, collectors such as the Coriolis may bring important insight on exposure that may impact on future biosafety regulations and recommendations. In this context, following a substantial microbiological spill, it is generally recommended by biosafety delegates that personnel should vacate the room for 20-30 min due to the risk of exposure to aerosols (WHO, 2004) . Cleaning and decontamination procedures are consequently delayed although robust experimental support for this risk assessment is missing. Using the Coriolis and particle counters, we show that a simulated spill with a large, concentrated volume of microspheres does not generate a significant number of aerosol particles in the environment. Because few airborne particles were generated in the spill, it might not have been a preferred simulation to study Coriolis performance. Nonetheless, it gave insight into an important and common biosafety issue related to microbiological exposure in the laboratory work environment. Our data thus suggest that infection control measures can be applied immediately after a large microbiological spill since the risk of aerosol dissemination and exposure to the user in this condition is negligible. It is unclear to what degree dust particulates from a surface could be reaerosolized during a spill to potentiate airborne particles and microbial exposure. Bacillus anthracis spores have been reported to be reaerosolized from contaminated office surfaces under conditions of low personnel activity (Weis et al., 2002) . A review of historical data on tuberculosis transmission highlights the risk of dust-borne M. tuberculosis in spreading the infection (Martinez et al., 2019) . A premise for our recommendation of immediate decontamination is therefore that it be performed on laboratory surfaces that are otherwise kept clean and accumulation of dust minimized. In the unique setting of our BSL-3 infrastructure with forced ventilation, we also used the Coriolis to investigate detection of aerosols carrying microspheres aerosolized on a BLAM, a Collison-type nebulizer. Collison nebulizers are readily used to experimentally infect laboratory animals through the aerosol route (May, 1973; Roy and Pitt, 2012) . This experiment thus simulates a potential incident in an animal BSL-3 or aerobiology laboratory with ensuing infectious bioaerosol dispersal. Even though forced ventilation returned particle counts in the room to background levels within 15 min after the simulated incident, microspheres could be detected with the aid of the Coriolis up to 1 h after aerosol release. In the absence of forced ventilation, particle numbers remained high and elevated for the entire duration of the experiment. These experiments reveal the importance of ventilation in limiting transmission of infectious bioaerosols. They also indicate that the risk of exposure remains for at least 1 h after a bona fine aerosolization, even in the presence of forced ventilation. Thus, our simulations show that an accident with an aerosol generator introduces a much higher risk for occupational exposure compared with a large (0.5 l) microbiological spill. The field of aerobiology has been hampered by the lack of tactical (fieldable) units for microbiological air sampling. Units such as the Coriolis are helping to fill this gap by providing a useful tool for the study and quantification of infectious bioaerosols. Our simple operative and technical modifications to the Coriolis should add to its biosafe deployment and promote continued investigation on human transmission and exposure to airborne pathogens. Supplementary data are available at Annals of Work Exposures and Health online. This work was funded by the Bill and Melinda Gates Foundation (grant number OPP1118552), Karolinska Innovations AB, and Karolinska Institutet, all to A.G.R. The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication. Air samplings in a Campylobacter jejuni positive laying hen flock Sources of airborne norovirus in hospital outbreaks Contribution of a cyclonic-based liquid air collector for detecting Aspergillus fumigatus by QPCR in air samples Measles outbreak in a pediatric practice: airborne transmission in an office setting Performance of the Coriolis air sampler, a high-volume aerosol-collection system for quantification of airborne spores and pollen grains Electrochemical detection of TNT at cobalt phthalocyanine mediated screenprinted electrodes and application to detection of airborne vapours Collection efficiency of liquid-based samplers for fungi in indoor air Quantification and spread of Pneumocystis jirovecii in the surrounding air of patients with Pneumocystis pneumonia Aerosol transmission is an important mode of influenza A virus spread Advances in large-volume air sampling Comparative testing and evaluation of nine different air samplers: end-to-end sampling efficiencies as specific performance measurements for bioaerosol applications Variability of infectious aerosols produced during coughing by patients with pulmonary tuberculosis Coughgenerated aerosols of Mycobacterium tuberculosis: a new method to study infectiousness Biosensors for monitoring airborne pathogens Analysis of airborne biomarkers for point-of-care diagnostics Evaluation of the efficiency of the Coriolis air sampler for pollen detection in South Europe Effect of impaction, bounce and reaerosolization on the collection efficiency of impingers Investigation of inherent and latent internal losses in liquid-based bioaerosol samplers Comparison of labmade electrostatic rod-type sampler with single stage viable impactor for identification of indoor airborne bacteria Sampling and detection of airborne influenza virus towards point-of-care applications Rapid quantification of bioaerosols containing L. pneumophila by Coriolis ® µ air sampler and chemiluminescence antibody microarrays Pneumocystis jirovecii in the air surrounding patients with Pneumocystis pulmonary colonization A new combination of microbial indicators for monitoring composting bioaerosols Detection, survival and infectious potential of Mycobacterium tuberculosis in the environment: a review of the evidence and epidemiological implications The Collison nebulizer: description, performance and application Droplets and aerosols in the transmission of SARS-CoV-2 Evaluation of Legionella air contamination in healthcare facilities by different sampling methods: an Italian multicenter study Evaluation of air contamination in orthopaedic operating theatres in hospitals in southern Italy: the IMPACT Project Bioaerosols and transmission, a diverse and growing community of practice. Front Public Health Detection of Mycobacterium tuberculosis bacilli in bio-aerosols from untreated TB patients Laboratory-associated infections: summary and analysis of 3921 cases Continuing importance of laboratory-acquired infections Airborne transmission of measles in a physician's office Characterization of reaerosolization from impingers in an effort to improve airborne virus sampling Airborne spread of measles in a suburban elementary school Infectious disease aerobiology: aerosol challenge methods A fieldable electrostatic air sampler enabling tuberculosis detection in bioaerosols Linking the conventional and emerging detection techniques for ambient bioaerosols: a review Survey of laboratory-acquired infections Detection of airborne lactococcal bacteriophages in cheese manufacturing plants Accessing indoor fungal contamination using conventional and molecular methods in Portuguese poultries Occupational exposure to Aflatoxin (AFB1) in poultry production Secondary aerosolization of viable Bacillus anthracis spores in a contaminated US Senate Office Laboratory biosafety manual WHO Policy on TB infection control in healthcare facilities, congregate settings and households Improved aerosol collection by combined impaction and centrifugal motion Concentrations and size distributions of airborne influenza A viruses measured indoors at a health centre, a day-care centre and on aeroplanes We thank Roland Möllby (Karolinska Institutet, Sweden) and Wayne Bryden (Zeteo Tech, USA) for suggestions and critical reading of this manuscript. We would also like to thank Sören Hartmann and Per-Erik Björk (Karolinska Institutet) and Erik Ekstedt (Akademiska Hus AB, Stockholm, Sweden) for technical assistance. Flow cytometry was performed at the Biomedicum Flow Cytometry Core facility (BFC), Department of Microbiology, Tumor and Cell Biology, Biomedicum, Karolinska Institutet. The authors declare no conflict of interest.