an aerosol exposure chamber for inactivation of inﬂuenza ... › reu › 2016 ›...

An Aerosol Exposure Chamber forInactivation of Influenza Virus Using

Ultra-Violet RadiationConnor Crickmore, David Welch, Manuela Buonanno, David Brenner

Radiological Research Accelerator Facility, Nevis Laboratories, P.O. Box 21, 136 S.Broadway, Irvington, N.Y. 10533

August 26, 2016

Abstract

In this project we aimed to construct an experimental apparatus and process to allow us to measure thesterilization capabilities of 222 nm ultra-violet light on influenza virus in an aerosol. We updated andcharacterized a previously designed piece of equipment in order to create an aerosol in a sealed exposurechamber, whilst controlling particle size, relative humidity and laminar flow rate. We then tested thisapparatus using active virus and had some success in sampling the virus transported by the aerosol aswell as sterilization of this virus with a 254 nm germicidal lamp. Further work will be completed to extendthis testing to 222 nm UV and fully characterise its sterilization capabilities. This work was completedduring a ten week REU program.

I. Introduction

I nfection control is an important area ofresearch in modern day medicine. Surgicalwound infection is still a significant

problem in modern day medicine, withbetween 200, 000 and 300, 000 cases per yearin the US, accounting for up to 8, 200 deathsand a large increase in healthcare costs [1].Furthermore, spread of airborne infectionis commonplace in areas including hospitalwards, school, offices, passenger planes andpublic transport.

The use of broadband 200− 400 nm ultra-violet

(UV) radiation for sterilization purposes is wellestablished, for example it is commonplace forsterilization of surgical equipment in hospitals.The UV light causes damage to DNA, andtherefore is effective against viruses andbacteria including drug resistant pathogenssuch as MRSA. However, current germicidallamps have major safety drawbacks in that theyare carcinogenic and cataractogenic, and socannot be used in areas with people present orduring surgery [1, 2]. Based on biophysicalprinciples and previous studies [1] and [2]we propose far-UVC light at 222 nm from aKrCl excimer lamp will not have these safetydrawbacks, yet will remain antimicrobial.

1

An Aerosol Exposure Chamber for Inactivation of Influenza Virus Using Ultra-Violet Radiation

Our proposition comes from the absorptiondata for UV light of different wavelengths.Proteins rapidly absorb UV near 200 nm,particularly the peptide bond [3, 4], meaningthat the UV is not very penetrating in proteinrich material such as human cells and tissue.Figure 1 demonstrates this.

Figure 1: Absorption against wavelength averaged over8 common proteins [1].

Because of this rapid protein absorption, UVlight near 200 nm cannot penetrate the cellwall or cytoplasm of a human cell, nor thehuman stratum corneum1 or the cornea in theeye. However, due to the smaller size of virusesand bacteria, which are typically smaller than 1µm compared to a human cell varying between10 − 25 µm micrometers, the radiation can stillreach and damage the DNA in a bacteriumor virus, therefore will still have the desiredgermicidal properties. Figure 2 illustrates thiseffect.

222 nm light in particular was chosen due toour ability to produce this light using KrClexcimer lamps [5]. Excimer lamps work byexciting a molecular complex, in this caseKrCl, which then emits specific wavelengthlight [6, 7]. The excimer lamp we will use isshown in figure 3. These lamps are relativelyinexpensive, long-lived and readily availablemeaning they have promise for widespreadusage. With these lamps we could sterilize awound throughout surgery by leaving themrunning, without causing harm to the patient;research has been done previously on thisapplication in vitro and in vivo [1, 2].

Figure 2: UV light indicated by the purple arrows incident on a human cell, MRSA and human skin. The diagramshows the UV light cannot reach the nucleus in a human cell or penetrate human skin, but can penetrate thesmaller bacteria and viruses such as MRSA.

1The layer of dead surface skin cells.

Connor Crickmore et al. 2


Figure 3: A KrCl flat excimer 4”× 4” lamp. The 222 nmlight is emitted by excited KrCl entrapped inthe microcavities. More information of thelamp can be found in [5] and [6].

Here we will focus on the application of theselamps to sterilize pathogens in the air, insuch areas as hospital wards as previouslymentioned. We hypothesise that these lampswill also have sterilizing effects on aerosols2,and therefore could be an effective means ofcombating airborne infection.

II. Experimental

Apparatus

In order to test the effectiveness of radiation onan aerosol, we needed a way of producing theaerosol with our virus in. We also needed theaerosol to be produced in a contained chamberto stop the virus escaping; we would useinfluenza A virus (A/PR/8/34 (H1N1)) as it iscommercially available and relatively safe. Wewould need a UV transparent window to allowUV into this chamber, as regular glass would

absorb the radiation. Furthermore we neededa laminar flow so that we could accurately tellthe dosage given to the aerosol particles bythe lamp, and we wished to control the relativehumidity (RH) also. Lastly, we needed a way ofmeasuring particle size and how many viruseswere deactivated.

II.i. Previous DesignPrevious work had been done on designingthis experiment and so our starting point wasshown in figure 4 and a schematic of thisshown in figure 5. This design was influencedby similar designs from the Harvard School ofPublic Health [8] and [9].

Figure 4: The previously designed experiment.

Firstly we produced the aerosol by pumping airinto a high-output extended aerosol respiratorytherapy (HEART) nebulizer (Westmed, Tucson,AZ), part C in figure 5. The aerosol producedwent into a chamber which had baffles(labelled D) and a second port to allow foran RH control system input.

Figure 5: Schematic diagram of the previously designed experiment.

2A suspension of droplets in the air.



The baffles promoted droplet dryingand mixing to produce an even particledistribution [8], before the RH is measuredby E, an RH meter. Next, F was a 28 × 30 cmwindow into the exposure chamber, which wasmade of UV transparent fused silica quartz;this allowed the UV light to enter the chamberwith no virus release. The chamber itselfwas modelled to give a laminar flow past thewindow, so that the exact dosage given to theaerosol could be accurately calculated. Lastly,H and I in the schematic were the particle sizerand BioSamplersr receptively.

The Hal Technologies HAL-HPC300 particlesizer uses the refraction of laser light bythe aerosol droplets to infer their size anda vacuum pump pulls the air through theentire chamber and through two SKC Inc.BioSamplersr. The BioSamplersr directthe air through 3 nozzles aimed towardsthe glass wall of the vessel, which is filledwith 20 ml of Hank’s balanced salt solution(HBSS); this forces the HBSS to swirl andwash any viruses off the glass wall andcapture the viruses, with the laminar motionof the HBSS preventing any re-emission ofthe viruses through aerosol production inthe BioSamplersr. Two BioSamplersr wererequired as the narrow nozzles limited theflow rate to 12.5 LPM, but having twoBioSamplersr attached in parallel increasedthis flow rate analogous to current in parallelresistors in circuit theory. The machine was

connected together using Tygonr PVC tubingwith a 3/8" inner diameter, except for the tubeattaching the nebulizer, which used a tube witha 3/4" inner diameter.

II.ii. Updated Design

Figure 6 shows the apparatus after thefollowing updates had been made. Firstly,the pump requirements were reconsidered asa stronger pump was required to reach thelimiting flow rate of 12.5 LPM (26.4 SCFH)through each of the BioSamplersr; this limit isrequired to get maximum collection efficiencyin the BioSamplersr. When the BioSamplersrwere added in parallel, the limiting flowrate through both of them combined became42 SCFH, and so a new MILLIPORE vacuumpump (model number WP6111560) was usedwhich could provide a large enough pressuredrop to create this flow rate. Previously aThermo Scientific Dual-head pump (modelnumber 420-2901-00FK) was being used asboth the nebulizer and the vacuum pump;now it would just be used for the nebulizerinput. We also added GE HealthcareLife Sciences Whatman™ HEPA-VENT filters,which stopped particles, bacteria and virusesentering the system through the pumps fromthe room air. The HEPA filters stop almostall particles, with weakest filtration at around0.3 µm where at least 99.7% are still stopped.

Figure 6: Schematic diagram of the updated experiment. The additions include a desiccator, A, a water vessel to addhumidity, B and a pressure gauge, G.



The machine was tested for the first time withthe quartz windows, as previously only sheetsof acrylic had been used. We found that aircould enter our machine through small gapsaround the windows, so we inserted a 1/32"silicone rubber gasket (VIP Rubber and PlasticCompany, Inc.), which prevented the majorityof the leak. Upon testing we found that afew particles, of the order of hundreds perminute, were still leaking into the machine;however this would be negligible compared tothe hundreds of thousands that were typicallymeasured per minute when operating with thenebulizer running. Furthermore, any leakswould flow into the box due to slight negativepressure inside the box, meaning no virusshould escape, and as the lab air should bevirus free, no contamination would enter themachine.

Lastly we re-tubed the equipment, allowingthe BioSamplersr to be bypassed, so thatwhen we run the equipment steady statecould be reached without passing air throughthe BioSamplersr, and then we could beginsampling with the switch of a valve. A 3-portplastic ball valve was used to direct the floweither to the bypass or the BioSamplersr.A King Instrument Company, Inc. brassvalved 6-60 SCFH acrylic tube flow rate meter(7530 series) was placed in the bypass sothat the flow could be sufficiently lowereduntil it matched the flow rate through theBioSamplersr. Then the flow through therest of the system would be unchanged whenpassing through the bypass, meaning we couldreach the same steady state through the bypassas would be reached when running throughthe BioSamplersr.

The biggest change we made was creating theRH control system. To do this we connected aWilkersonr X06 desiccant air dryer to providedry air, and a 2 litre conical flask filled withwater which we bubbled air through to givemoist air. We attached to each of these anothervalved flow rate meter so that we could controlrelative amounts of them, and therefore the

RH, but we HEPA filtered them to ensure noaerosol got into the main chamber, only vapour.However, we found that controlling the RHcreated pressure build up in the box, whichmeant the thin quartz would not withstand thepressure and therefore would not be suitablefor the window material. Through re-tubingthe RH control system, using two tubes tobubble the air through the conical flask inorder to allow more airflow, and adding inextra HEPA filters in parallel to again allowmore airflow, the maximum pressure in thechamber was reduced to 0.04 atm below thelab pressure when running. We would nowuse a UV transparent plastic film (TOPASr8007X10) with a metal supporting frame, seefigure 7, instead of a quartz window to ensurethe chamber seal would not be compromisedduring machine operation.

Figure 7: The supporting frame for the film windows.The vertical bar design ensures all aerosol willstill be irradiated, the bars just reduce the UVexposure time, which we can factor into ourcalculations. The airflow is in the directionindicated (perpendicular to the steel bars), andwill be laminar past the window.

The window dimensions, accounting for thetwo steel beams and any parts of the windowcut off by the rim of the chamber mount,



became 26×25.6×6.3 cm, meaning the totalvolume of air which would be irradiated bythe beam was 4.2 litres. By using the overallflow rate for two BioSamplersr in parallelas 42 SCFH and assuming the flow is trulylaminar, the total exposure time will be 12.7 s.This exposure time can be lowered by coveringup part of the lamp, so that only part of thisvolume of air is irradiated. The exposure timecan be increased by slowing down the overallflow rate through the chamber, for exampleslowing the flow to 24 SCFH would result in aexposure time of 22.2 s, however this will causea reduction in the BioSamplerr samplingefficiency, so tests would be conducted to findthe lowest flow rate where normal machineoperation was still possible.

III. Preliminary

Testing and Discussion

Before conducting biological testing with ourapparatus we first wished to fully characterizethe machine, to understand the effect of eachvariable on our aerosol. Research conducted in[10] and [11] suggested that the size of aerosoldroplets expelled by humans are roughlydistributed around 1 µm, with many less than1 µm, and a large variation between subjects.Figures 8 and 9 show this. Therefore we wouldaim to be in this range when producing ouraerosol, in order to replicate conditions in anenvironment with infection spreading betweenhumans via the air.

During each experiment we could control theRH inside the box by varying the wet anddry inlet flow rates, we could control the flowrate into the nebulizer, as well as the volumeand composition of its contents, and we couldcontrol the overall flow rate through the systemand the time we sampled particles for with theparticle sizer. The RH and temperature of thelab were not controlled, but the RH in the box

could still be controlled by allowing more airthrough one inlet to counteract any changes inthe lab RH. We were interested in measuringthe particle size and the amount of virus in theBioSamplersr, comparing tests with the lampon and lamp off to see if the virus numberswere reduced. For our preliminary testing wewould not use any lamp or virus, but wouldpurely characterise the effect of our controlledvariables on the particle size.

Figure 8: Data from [10] for one subject of the study,showing the aerosol particle distribution fordifferent respiratory actions.

Figure 9: Data from [11] showing particle size againstthe concentration of droplets from a cough.

Various tests were conducted by using differentcombinations of the controlled variables andobserving the particle sizes that resulted; thisdata is presented in figures 10 to 15.



Figure 10: Data for a test with 75 ml of tap water in a nebulizer labelled D. The flow rate through the RH systemwas set to 100% dry air and the flow rate into the nebulizer was set at 10 LPM. The particle distributionproduced was comparable to the literature [10, 11].

Figure 11: This test kept all variables constant except for increasing the volume of the liquid, in this case water, in thenebulizer. The data shows that the liquid volume had virtually no effect on the particle distribution. Thetemperature was 28.1°C, the RH in the box was 37%, the nebulizer flow rate was 8 LPM and the overallflow rate was 42 SCFH.

Figure 12: This test kept all variables constant except for increasing the RH in the chamber by increasing the wet airflow rate. The data shows RH has a significant effect on the particle distribution. The temperature was28.1°C, the nebulizer was filled with 150 ml of water, the nebulizer flow rate was 8 LPM and the overallflow rate was 42 SCFH.



Figure 13: This test kept all variables constant except for changing the nebulizer flow rate. The data shows this rate hasa significant effect on the particle distribution. The temperature was 28.3°C, the nebulizer was filled with75 ml of water and the overall flow rate was 42 SCFH. The RH in the box was 50% at 10 LPM and 53% at12 LPM; the increase is because less air is pulled through the RH control system and also due to increasedevaporation from the larger droplets. Therefore some of the particle size increase seen here comes from theincrease in RH between the tests.

Figure 14: This test kept all variables constant except for changing the nebulizer liquid from water to 100% HBSS. Thedata shows the liquid composition has a very large effect on particle sizes. The temperature was 28.4°C, theRH in the box was 43%, the nebulizer was filled with 75 ml of liquid, the nebulizer flow rate was 10 LPMand the overall flow rate was 42 SCFH.

Figure 15: This test kept all variables constant except for changing the total flow rate from 42 to 26 SCFH. The datashows the change in particle distribution, however this is caused by a change in the RH. The same amountof air comes from the nebulizer at both total flow rates and therefore less air is pulled through the RH controlsystem so it is less effective. The temperature was 28.3°C, the RH in the box was 53% for 42 SCFH and67% for 26 SCFH. The nebulizer was filled with 75 ml of water and its flow rate was 12 LPM.



Through further testing with a similarapproach, we also found that changing thenebulizer unit for a brand new unit had anegligible effect on the particle distributioncompared to other factors. Also DI water wastested, however almost no particles above our0.3 µm lower bound were produced. Thismade sense as the increase in the salinityfrom water to HBSS increased the particlesizes, so reducing salinity from water to DIwater reduced the particle sizes until they werebelow the precision of our particle sizer. Wehypothesise this change in particle size withliquid is driven by surface tension effects.

HBSS in particular was investigated as itwould be used as our collection liquid in theBioSamplersr. Also, unpublished data fromthe laboratory suggested both HBSS and waterhave minimal UV attenuation and so would besuitable for use as the aerosol liquid.

One thing that was apparent from this testing,especially the total flow rate test in figure 15,is that all of these variables are closely linked.If any of the flow rates on the nebulizer orRH control or output change, then they allchange due to the combined effect of the RH onparticle size and the fact that the nebulizer flowrate and total flow rate out are set, and the flowrate through the RH is the difference betweenthese. However, we could simply set up theconditions we want, then keep everything fixedand just sample the output of viruses withoutand then with the lamp, which would tell us if

the lamp is an effective method for sterilization.

IV. Biological Testing

and Discussion

IV.i. Set Up

In order to conduct testing with influenzavirus in the machine, we moved the equipmentinto a LABCONCOr Purifier Biological SafetyCabinet (Class II, Type A2), and directed thetubing so that the RH inputs and the pumpswere outside the cabinet, so that virus free labair would be drawn in. We also directed theoutput of the vacuum pump into two 1800 mlconical flasks filled with water mixed with10% bleach, in order to create a bleach tap tokill any viruses that were not stopped by theBioSamplersr or the HEPA filters, meaningthey would not be released into the lab air.

Next, we considered how to sterilize the boxonce an experiment had been conducted. Wehypothesized that using aerosol produced byputting 10% bleach in the nebulizer could work.To test this we ran this set up for 15 minuteswith food dye in the BioSamplersr. A colourchange was observed as shown in figure 16,which confirmed that the bleach was beingspread throughout the system.

Figure 16: The left-hand image shows the BioSamplersr filled with diluted yellow and green food dye. The right-handimage shows the colour change due to bleach which was collected during a 15 minute run of the nebulizerfilled with 10% bleach. The colour change showed the bleach penetrated the entire system.



Afterwards, we would wash the nebulizer andBioSamplersr with 10% bleach and then withDI water, and then we would run the machinewith just DI water in the nebulizer to flush outany remaining bleach aerosol. This shouldfully sterilize the internals of the machine,and to sterilize the outside the equipment wasleft in the biosaftey cabinet overnight with agermicidal UV light running in order to killany surrounding virus. Furthermore, the viruswould die once it had dried out, and so anyleft in or out of the machine would becomeinactive once left overnight to dry.

At this point, we decided that we would onlyuse one BioSamplersr for our initial biologicaltesting, as we realised by investigating thepower output of the excimer lamp thatwe would likely want to have a longerexposure time than the 12.7 seconds from twoBioSamplersr at maximum flow rate. Wecould just slow down the overall flow ratewith two BioSamplersr, but this would likelycause issues with the capture efficiency. Usingonly one BioSamplerr would mean that theoverall flow would be lower, capped at the12.5 LPM limit of the BioSamplerr, yet asthe limiting flow for this single BioSamplerrwas reached the efficiency would still beoptimized. However, due to decreasing thetotal flow rate whilst the nebulizer flow rateremained fixed, the flow rate through theRH control would be lower, making it lesseffective, but still giving us RH control overa decent range. Furthermore, having onlyone BioSamplerr cup to sample from wouldsimplify experimental operation. The optionto reintroduce the second BioSamplerr forshorter exposure times would now becomea simple adjustment we could make tothe machine, without changing any othercharacteristics except for the total flow rate(and therefore the effectiveness of the RHcontrol). For the one BioSamplerr set upoperating at the 12.5 LPM total flow rate limit,the exposure time would be 20.2 s, which againcould be increased by using the valved flow

rate meter to reduce the overall flow, withthe effects of this on the sampling efficiencyneeding to be investigated.

Lastly, as mentioned in the preliminary testingsection, changing the nebulizer liquid has alarge effect on particle size distribution. Sothat we could use could use sterile controlledsubstances, rather than using tap water, yetstill get a human-like particle distribution,we investigated different mixtures of HBSSand DI water to give us the desired particledistribution. We found that 0.25% HBSS in DIwater gave a similar distribution to tap water;this is shown in figure 17, and it gives a similardistribution to figure 10.

Figure 17: Data for a test with 20 ml of DI water with0.25% HBSS added. The particle distributionproduced was comparable to data given infigure 10, and therefore with the literaturedata in figures 8 and 9.



We would use a DI water HBSS mix in therange close to 0.25% as starting point forthe tests run with the virus, but this was infact one of the most important variables wevaried during the tests that we ran, which arediscussed in the following.

IV.ii. Testing and Results

A series of five tests were ran with active virus,each testing a number of conditions. Duringthese tests the particle sizer took continuousone minute samples, and provisionally theBioSamplerr would sample each conditionfor 15 minutes whilst filled with 20 ml of HBSSto capture the virus. By passing the air throughthe bypass for 5 minutes at the start of eachrun, steady state would be reached; the two3-port vales would then be switched so that theflow went through the BioSamplerr for the 15minutes of sampling time.

Once the sample was complete, we centrifugedthe contents of the BioSamplerr and then took250 µl from the bottom 1 ml of the centrifugedsolution and deposited this onto a plated layerof Madin-Darby Canine Kidney Epithelial cells(MDCK). These were incubated at 37°C for45 minutes and then washed with HBSS andput into fresh Gibcor Dulbecco’s ModifiedEagle Medium (DMEM) to be left overnight. Astandard immunocytochemistry protocol wasthen conducted, including fixing the cells andadding a primary antibody (Anti-Influenza AVirus Nuceloprotien antibody [C43]) to identifythe infected cells. A secondary antibody (Goatanti-Mouse Alexa Fluorr 555) was then added,which would attach to the primary antibodyand can fluoresce, thus allowing identificationof the infected cells when observed under themicroscope at ×10 magnification.

For the first test, 100 ml of DI water was used inthe nebulizer with 0.5 ml of HBSS added. Thenebulizer flow rate was set at 10 LPM, the totalflow rate was the maximum 26 SCFH and the

RH control valves were both fully opened. Weran a control test by sampling for 15 minuteswith this solution and then we added the virusinto the nebulizer and sampled again for 15minutes. The virus we used was separatedinto vials of 108 fluorescent forming units permillilitre (FFU/ml) in 1 ml of DMEM. For thistest we diluted a 1 ml vial in 20 ml of DI water,and then added 1 ml of this diluted mixture toour nebulizer fluid, effectively adding 1 ml ofDI water with 5% DMEM and 5 × 106 FFU/mlof virus. After sampling the solutions in theBioSamplerr and nebulizer and completingthe immunocytochemistry on the cells, thevirus was clearly absent in the BioSamplerrfor the control test and present in the nebulizersolution for the virus test, figures 18 and 19.

Figure 18: The results of the cell analysis for the controltest BioSamplerr sample. The top imageshows the cell layer under a bright field.The lower image shows the cell layer underfluorescence, where any cells infected with thevirus fluoresce. This means no contaminatingvirus was present for the control.



Figure 19: These were the results of the cell analysisfor the nebulizer contents in the virustest. Unlike the BioSamplerr sample, the250 µl were taken straight from the nebulizerwithout centrifuging. The top image showsthe cell layer under a bright field. The lowerimage shows the cell layer under fluorescence,with a large number of infected fluorescingcells showing the nebulizer contained activeviruses.

No virus was detected in the BioSamplerrfor the virus run, giving similar images to thecontrol test in figure 18. This meant viruswas not being successfully transported andsampled by the system, which also meant thenext test we conducted with virus at a slowerflow rate gave a virus free result. The RHwas 55.5% for the control test and 57% forthe virus test, at a temperature of 29.9°C forboth. For the control test the average particledistribution throughout the test was 40.5%,31.0% and 28.4% in the 0.3 − 0.5 µm, 0.5 − 0.7µm and 0.7+ µm ranges respectively. For the

virus test the average particle size increasedslightly to 37.1%, 31.0% and 31.9% in the sameranges, due the addition of the diluted solutionof the virus in DMEM increasing the particlesize and also the RH.

In an attempt to get virus detection, we madethe following adjustments in test 2. Firstly, welowered the volume of the DI water, HBSS mixin the nebulizer from 100 ml to 20 ml; thiswould have little effect of the aerosol producedas shown in our preliminary testing but wouldallow us to have a high concentration of viruswithout having to use as much as we wouldfor 100 ml. We also used 0.25% HBSS inthis mix, rather than the 0.5% used in test1, in order to achieve the smaller human-likeparticles. Testing on the 20 ml of water, 0.05 mlHBSS mix indicated 11 LPM should be used forthe nebulizer flow rate to get a good particledistribution. The RH was found to increasewhen running with less liquid in the nebulizer,perhaps due to a larger area of plastic for theliquid to spread over, thus aiding evaporation;therefore only the desiccator valve was beopened in the RH control system for test 2.Lastly, we added the full 1 ml of 108 FFU/mlvirus in DMEM to the nebulizer, straight fromthe vial, in order to have more virus to startwith.

Both a 15 minute and a 30 minute samplewere taken, with the hope that the longersample would allow for more detection. Viruswas sampled and successfully detected inboth samples. The additional sampling timedid have the desired effect of increasing theamount of virus collected. However, wehad not accounted for the effect of the virusmedium on the particle size, and so for thesetests we had an average particle distributionof 2.5%, 3.1% and 94.4% in the 0.3 − 0.5 µm,0.5 − 0.7 µm and 0.7+ µm ranges respectively.Therefore we hypothesized that the virusdetection was caused by very large aerosoldroplets, which were not representative ofa human-like aerosol. For these tests thetemperature was 29.6°C and the RH was



between 53% and 57% throughout the tests.The results from test 2 are shown in figure 20.

Figure 20: These were the results of the cell analysis fortest 2. The top image shows the cell layerunder a bright field (for the 15 minute test,the 30 minute test had a similar appearance).The middle image shows the cell layer fromthe 15 minute test under fluorescence, whichreveals a few infected cells. The lower imagefrom the 30 minute test has more fluorescingcells due to the longer sample time meaningmore viruses were collected and detected.

As in test 1, the nebulizer liquid was sampled.However no fluorescence was seen becausethe high concentration of virus was enoughto disrupt the cell layer, meaning the cells diedbefore fixing and so were lost in the washingprocesses. This was not an issue however,as we could infer from the disrupted layerthat there was virus present, and we coulddilute down the nebulizer liquid and apply thediluted sample to the cells to obtain fluorescingcells.

For the third virus test we repeated the30 minute run from test two, except wecentrifuged the virus vial and removed themedium, and then re-suspended the virus in1 ml of DI water. We then attempted thissame test at various lower total flow rates. Forthe 26 SCFH test we had a more human-likeparticle distribution of 43.4%, 24.8% and 31.8%in the 0.3 − 0.5 µm, 0.5 − 0.7 µm and 0.7+ µmranges respectively; the RH was 52% andthe temperature was 29°C. However at thislower particle size distribution we found nofluorescing cells at any flow rate, meaningthe virus was again not being transported andsampled at a detectable level. The sample fromthe nebulizer contents was similar to the resultof test 2, confirming virus was active in thenebulizer, therefore we concluded the smallerdroplets simply didn’t transport enough virus.This made sense as the volume of the dropletgoes as the radius cubed, so reducing theparticle radius rapidly reduced the volume ofliquid transported, and therefore amount ofvirus transported.

From testing the 20 ml DI water, 0.05 ml HBSSmix without the virus, we found the particledistribution to be almost identical to that intest 3, which indicated that adding the virusre-suspended in DI water would not changethe particle size, only the DMEM would alterthe particle size distribution. Therefore for test4 we tested different concentrations of HBSSwithout the virus, knowing adding the virusin DI water would not alter the particle size.We also tested 20 ml DI water with 0.05 ml



HBSS and 1 ml DMEM, and knowing the virusitself doesn’t have an effect on particle size, wecould recreate the particle distribution of theaerosol in test 2 that successfully transportedvirus through. We found that the distributionwas 17.8%, 13.5% and 68.6% in the 0.3− 1.0 µm,1.0 − 2.0 µm and 2.0+ µm ranges respectively(note the change in size ranges to the previous).We decided we would add 0.5 ml of HBSS tothe 20 ml of DI water, which would give aparticle distribution of 41.1%, 29.1% and 29.9%in the 0.3 − 1.0 µm, 1.0 − 2.0 µm and 2.0+ µmranges respectively. This would be a goodcompromise between having a human-likedistribution from [10] and [11] and having thelarger particles from test 2 which carried thevirus through the system.

For test 4 we also re-suspended each vial in0.5 ml of water and then added 2 vials worthof virus for one 30 minute run, and all 4 vialsfor a second 30 minute run. Despite havingup to 4 times the virus in the larger particles,no virus was detected in the cells, meaningwe still did not have large enough droplets totransport detectable amounts of virus.

Lastly for test 5 we characterised exactly whatthe particle distribution was for the test 2aerosol with the full 1 ml of DMEM. We thencompared this to the same mixture with 1 mlof HBSS instead of DMEM, shown in figure21. The distributions are similar for the twomixtures, and so we decided to run the 30minute sample from test 2 again, and then runit with the 1 ml of DMEM replaced with 1 mlof HBSS, to see if somehow the DMEM washelping the virus transport. We also ran the1 ml DMEM mixture with a Hygeaire 254 nmgermicidal lamp running for the duration ofthe 30 minute sample, to see if any virusreduction would be detectable. The lamp wasplaced 7.5" from the exposure chamber, whichgave a power of 0.95 mW at the centre of thechamber, including the attenuation from thewindow. Over the 20.2 s transit time this wouldgive an average dose of 19.2 mJ to the virus,which we expect to sterilize it.

Figure 21: The upper table shows the particledistribution of the 20 ml DI water with0.05 ml HBSS and 1 ml DMEM at 54%RH and a nebulizer flow rate of 11 LPM.The lower table shows the distribution of the20 ml DI water with 1.05 ml HBSS at 49%RH and a nebulizer flow rate of 11 LPM.

The results for the repeat of test 2 withoutthe lamp are shown in figure 22 and the testwith the UV lamp in figure 23. The RH wasbetween 50% and 57%, the temperature wasbetween 27.5°C and 27.9°C and the averageparticle distribution over these two tests was25.5%, 18.0% and 56.5% in the 0.3 − 1.0 µm,1.0 − 2.0 µm and 2.0+ µm ranges respectively.

Figure 22: Cells fluorescing from the test using the 20 mlDI water with 0.05 ml HBSS and 1 mlDMEM mix with 108 FFU/ml of virus.



Figure 23: The cells from the UV lamp test for 20 ml DIwater with 0.05 ml HBSS and 1 ml DMEMmix with 108 FFU/ml of virus. Two differentareas of the cell layer are shown. Most ofthe virus was inactivated except for a fewconcentrated areas shown by the fluorescence.

Much more virus is seen in figure 22 incomparison to the lower image in figure 20as the cells were left for a longer time periodbefore fixing. The test with the lamp showsa large reduction in the florescence, meaningthe virus was successfully deactivated by thegermicidal lamp.

The results of the test replacing the DMEMwith HBSS are shown in figure 24. Thevirus is found to have not transported intothe BioSamplerr even though the HBSS andDMEM size distributions were similar, with adistribution for the HBSS case of 40.7%, 24.6%and 34.7% in the 0.3 − 1.0 µm, 1.0 − 2.0 µmand 2.0+ µm ranges respectively. The RH was56.6% and the temperature was 28.3°C.

Figure 24: Only one cell can be seen fluorescing forthe test replacing DMEM with HBSS,even through the particle distributions weresimilar.

The contents of the nebulizer were checkedfor the HBSS test and there was the expectedamount of active virus in the nebulizer,therefore the washing and re-suspendingprocess could not have been deactivatingthe virus, and the virus can survive in theHBSS alone without the DMEM. Therefore wemust conclude that changing the DMEM toHBSS is preventing our virus from travellingthrough the machine or from being sampledor both of these. We considered a numberof explanations for this. Firstly, the slightreduction in particle size distribution withHBSS could be preventing transportation; thisseems unlikely as this would cause a reductionat best rather than a full eradication of the virus.However, we believe the DMEM evaporatesmuch faster than the HBSS, meaning we couldhave started with much larger droplets whenusing the DMEM mix, which had becomemuch smaller by the time they reach theparticle sizer, and therefore may have beenmuch more concentrated in virus. It is possiblethat a combination of these two factors couldaccount for the reduction seen, otherwise wemust assume that somehow the virus will onlytravel through our system in DMEM and notHBSS.

The only tests we have done as of yet with



DMEM are with 1 ml (except test 1 which usedmuch less virus), so next it will be worth testingusing less virus medium with the same amountof virus and seeing if this gets virus through.This would help us tell if the reduction we seeis caused by a particle size effect or due to thecontents of the aerosol liquid. Furthermore,if transmission is successful with less virusmedium it would be desirable as it wouldgive us smaller more human-like particles, yetstill get the virus through to the BioSamplerr.However, the 222 nm UV transmittance of anaerosol containing DMEM must be checked,as the high protein content may prevent thelower wavelength UV reaching the virus, eventhough the 254 nm lamp was clearly able to.The small droplet size in our aerosol may meanany attenuation of the 222 nm light due to theDMEM can be ignored.

In addition to more DMEM tests, we shouldnow test using more than 1.05 ml of HBSS,to recreate exactly the particle sizes measuredwith the DMEM mixture, to ensure the slightreduction in particle size we had for this test isnot the main factor in the virus reduction. Theevaporation rates of HBSS and DMEM shouldalso be accurately measured and comparedto judge the importance of any evaporationeffects. Lastly, an alternative medium shouldtested, such as phosphate-buffered saline (PBS),which may lead to more insight.

V. Conclusions

Overall, I feel great progress has been madewith this project during the 10 weeks ofmy REU program. We have successfullyset up and run our aerosol machine andcompleted a good numbers of virus testswith it, some of which had successful virusdetection. Furthermore, we have had clearindication from the germicidal UV lamp testthat we can measure if the virus is beingdeactivated by the irradiation source.

However, there are still a number of hurdles toovercome before getting conclusive results onthe effectiveness of 222 nm UV on sterilizing anaerosol. Firstly the issue of not sampling anyvirus through the machine at lower particlesizes must be addressed. This could be dueto needing DMEM to transport the virussuccessfully, which we aim to investigatein the tests outlined previous. Thereforemoving forwards our focus will be on fullyunderstanding the role of DMEM in the virustransport. If in fact the DMEM and HBSS turnout to be equally as effective at transportingvirus then the solutions to this problem couldbe a combination of producing more aerosol,using larger particles, sampling for longer andusing more virus. The latter three options areundesirable for practical reasons and becauseusing larger particles would not representthe data that has been measured for humanaerosol production. Therefore we wouldaim to produce a larger number of aerosoldroplets that are still in the human-like range,with a change of nebulizer or using multiplenebulizers being potential options to achievethis.

Furthermore, no tests have yet been performedwith the 222 nm excimer lamp. We willneed to figure how to fit the lamp so thatwe can irradiate the aerosol. We will alsoneed to fit dosimeters either behind or insidethe chamber to measure how much dose thelamp is producing. Moreover, testing on thecurrent lamp suggests the beam producedmay not provide sufficient intensity over theentire chamber window, so we may need toinvestigate reducing the size of the exposurechamber. A plan we are considering is to insertimplants inside the chamber to reduce theinternal volume of the chamber, and thereforereduce the area we would need to irradiate.This lack of lamp intensity is the reason wehave attempted to investigate slowing the totalflow rate down, as by slowing the flow ratedown the exposure time is increased, meaninga higher dose can be given from a lower



intensity. As our flow rate tests have not yetbeen successful, we will need to redo theseafter solving the virus sampling yield issue.This will tell us the highest exposure timefor which we can still sample virus effectively,which will then give the minimum intensityrequired to provide a certain dose.

Ultimately we should be able to use thismachine to record data on how 222 nm UVeffects the virus count for a variety of differenthumidities, particle distributions and lampintensities, which will potentially aid infectioncontrol, both airborne and surgical.

VI. Acknowledgements

I would like to thank Professor David Brennerand all the staff at RARAF for supporting methroughout my placement. In particular I wishto thank my supervisor Dr. David Welch whohas worked closely with me throughout thisproject and provided invaluable advice, andalso Dr. Manuela Buonanno who gave up lotsof time to conduct all of the virus tests forme. I would also like to thank the people atColumbia University and Nevis labs who havemade the REU program possible, especiallyProfessor John Parsons and Amy Garwoodfor their warm hospitality. I have thoroughlyenjoyed my time here in New York at RARAF.

References

[1] Buonanno M, Randers-Pehrson G,Bigelow AW, Trivedi S, Lowy FD, SpotnitzHM, Hammer SM, Brenner DJ. 207-nmUV Light - A Promising Tool for SafeLow-Cost Reduction of Surgical SiteInfections. I: In Vitro Studies. PLoS ONE,DOI: 101371/journal.pone.0076968. 2013PMCID: PMC3797730.

[2] Buonanno M, Stanislauskas M, PonnaiyaB, Bigelow AW, Randers-Pehrson G,Xu Y, Shuryak I, Smilenov L, OwensDM, Brenner DJ. 207-nm UV Light-APromising Tool for Safe Low-CostReduction of Surgical Site Infections.II: In-Vivo Safety Studies. PLoS ONE,DOI: 10.1371/journal.pone.0138418. 2016PMCID: PMC4898708

[3] Goldfarb AR, Saidel LJ. Ultravioletabsorption spectra of proteins. Science.1951;114:156-7.

[4] Setlow J. The molecular basis of biologicaleffects of ultraviolet radiation andphotoreactivation. In: M, E, A, H, editors.Current topics in radiation research.Amsterdam: North Holland PublishingCompany; 1966. p. 195-248.

[5] Eden G, Park S-J. Sheetlike microplasmashave many applications. Laser FocusWorld 2010; 46: 33-7.

[6] Sosnin EA, Oppenlander T, TarasenkoVF. Applications of capacitive and barrierdischarge excilamps in photoscience. JPhotochem Photobiol C: Photochem Rev2006; 7: 145-63.

[7] Volkova GA, Kirillova NN, PavlovskayaEN, Yakovleva AV. Vacuum-ultravioletlamps with a barrier discharge in inertgases. J Appl Spectrosc 1984; 41: 1194-7.

[8] Lai KM, Burge HA, First MW. Size andUV germicidal irradiation susceptibilityof Serratia marcescens when aerosolizedfrom different suspending media. ApplEnviron Microbiol. 2004;70:2021-7 PMCID:PMC383042.

[9] McDevitt JJ, Lai KM, Rudnick SN,Houseman EA, First MW and Milton DK.Characterization of UVC Light Sensitivityof Vaccinia Virus. Appl Environ Microbiol.2007;73:5760-6 PMCID: PMC2074914

[10] Papineni RS, Rosenthal FS. The sizedistribution of droplets in the exhaled



breath of healthy human subjects. Journalof aerosol medicine : the official journalof the International Society for Aerosolsin Medicine 1997; 10: 105-16

[11] Morawska L, Johnson GR, Ristovski ZD,

Hargreaves M, Mengersen K, Corbett S,Chao CYH, Li Y, Katoshevski D. Sizedistribution and sites of origin of dropletsexpelled from the human respiratory tractduring expiratory activities. Journal ofAerosol Science 2009; 40: 256-69


an aerosol exposure chamber for inactivation of inﬂuenza ... › reu › 2016 ›...

Documents