RESEARCH ARTICLE

Real-Time Investigation of Tuberculosis

Transmission: Developing the Respiratory
Aerosol Sampling Chamber (RASC)
Robin Wood1,2, Carl Morrow1,2*, Clifton E. Barry, III1,3, Wayne A. Bryden4, Charles J. Call4,
Anthony J. Hickey5, Charles E. Rodes6, Thomas J. Scriba1,7, Jonathan Blackburn1,8,
Chacha Issarow1,8, Nicola Mulder1,8, Jeremy Woodward8, Atica Moosa1,9,
Vinayak Singh1,9, Valerie Mizrahi1,9, Digby F. Warner1,9
1 Institute of Infectious Disease and Molecular Medicine (IDM), Faculty of Health Sciences, University of
Cape Town, Cape Town, South Africa, 2 Desmond Tutu HIV Centre, IDM, University of Cape Town, Cape
Town, South Africa, 3 Laboratory of Clinical Infectious Diseases, National Institute of Allergy and Infectious
Diseases, US National Institutes of Health, Bethesda, Maryland, United States of America, 4 Zeteo Tech
LLC, Ellicott City, Maryland, United States of America, 5 RTI International, Research Triangle Park, North
Carolina, United States of America, 6 Aerosol Exposure Dimensions, Cary, North Carolina, United States of
America, 7 South African Tuberculosis Vaccine Initiative, Department of Paediatrics and Child Health,
University of Cape Town, Cape Town, South Africa, 8 Structural Biology Research Unit, Department of
Integrative Biomedical Sciences, Faculty of Health Sciences, University of Cape Town, Cape Town, South
Africa, 9 MRC/NHLS/UCT Molecular Mycobacteriology Research Unit & DST/NRF Centre of Excellence for
OPEN ACCESS Biomedical TB Research, Department of Pathology, Faculty of Health Sciences, University of Cape Town,
Cape Town, South Africa
Citation: Wood R, Morrow C, Barry CE, III, Bryden
WA, Call CJ, Hickey AJ, et al. (2016) Real-Time * [email protected]
Investigation of Tuberculosis Transmission:
Developing the Respiratory Aerosol Sampling
Chamber (RASC). PLoS ONE 11(1): e0146658.
doi:10.1371/journal.pone.0146658 Abstract
Editor: Pere-Joan Cardona, Fundació Institut
Knowledge of the airborne nature of respiratory disease transmission owes much to the pio-
d’Investigació en Ciències de la Salut Germans Trias
i Pujol, Universitat Autònoma de Barcelona, SPAIN neering experiments of Wells and Riley over half a century ago. However, the mechanical,
physiological, and immunopathological processes which drive the production of infectious
Received: September 2, 2015
aerosols by a diseased host remain poorly understood. Similarly, very little is known about
Accepted: December 21, 2015
the specific physiological, metabolic and morphological adaptations which enable patho-
Published: January 25, 2016 gens such as Mycobacterium tuberculosis (Mtb) to exit the infected host, survive exposure
Copyright: This is an open access article, free of all to the external environment during airborne carriage, and adopt a form that is able to enter
copyright, and may be freely reproduced, distributed, the respiratory tract of a new host, avoiding innate immune and physical defenses to estab-
transmitted, modified, built upon, or otherwise used
lish a nascent infection. As a first step towards addressing these fundamental knowledge
by anyone for any lawful purpose. The work is made
available under the Creative Commons CC0 public gaps which are central to any efforts to interrupt disease transmission, we developed and
domain dedication. characterized a small personal clean room comprising an array of sampling devices which
Data Availability Statement: All relevant data are enable isolation and representative sampling of airborne particles and organic matter from
within the paper and its Supporting Information files. tuberculosis (TB) patients. The complete unit, termed the Respiratory Aerosol Sampling
Funding: This work is funded by grants from the Bill Chamber (RASC), is instrumented to provide real-time information about the particulate out-
& Melinda Gates Foundation (OPP1116641), and the put of a single patient, and to capture samples via a suite of particulate impingers, impactors
South African Medical Research Council (MRC) with and filters. Applying the RASC in a clinical setting, we demonstrate that a combination of
funds from National Treasury under the Economic
molecular and microbiological assays, as well as imaging by fluorescence and scanning
Competitiveness and Support Package (MRC-RFA-
UFSP-01-2013/CCAMP) (RW CM TJS JB CI NM JW electron microscopy, can be applied to investigate the identity, viability, and morphology of
AM VS VM DFW); the Intramural Research Program isolated aerosolized particles. Importantly, from a preliminary panel of active TB patients,
of the NIAID, NIH (CEB); and Bill & Melinda Gates

PLOS ONE | DOI:10.1371/journal.pone.0146658 January 25, 2016 1 / 16

 Respiratory Aerosol Sampling Chamber (RASC)

Foundation Investment ID 26123 (WAB CJC AJH we observed the real-time production of large numbers of airborne particles including Mtb,
CER). as confirmed by microbiological culture and polymerase chain reaction (PCR) genotyping.
Competing Interests: WAB and CJC are employed Moreover, direct imaging of captured samples revealed the presence of multiple rod-like
by Zeteo Tech LLC. AJH is employed by RTI Mtb organisms whose physical dimensions suggested the capacity for travel deep into the
International. CER is employed by Aerosol Exposure
Dimensions. Belonging to these commercial entities
alveolar spaces of the human lung.
does not alter the authors’ adherence to PLOS ONE
policies on sharing data and materials.

Introduction
In pioneering experiments over half a century ago, Wells and Riley confirmed the airborne
transmission of Mycobacterium tuberculosis (Mtb) [1,2]. Moreover, Wells proposed the concept
of an infectious quantum, a unit of infection capable of causing disease in 68.3% of susceptible
animals [1]. Key conclusions of these experiments—in particular, that there are low numbers of
infectious quanta and transmission is driven by a small subset of smear-positive patients—have
remained largely unchallenged. Subsequent data from similar animal studies appear to reinforce
these findings [3,4], although the experimental formats may have been characterized by even
lower sensitivities than the initial work of Wells and Riley owing to the need for higher levels of
diluting ventilation in modern health facilities [5]. In contrast, advances in molecular technol-
ogy have enabled the detection of large numbers of Mtb DNA copies circulating in the air of TB
treatment facilities [6,7,8]. Moreover, recent research [9] examining the exposure potentials of
healthcare workers in a South African TB clinic demonstrated the utility of personal exposure
sampling followed by PCR analysis as an effective tool to determine the risk of Mtb aerosol
exposure. Importantly, these observations implicated the close proximity of health workers to
patients—which resulted in concentrated exposure over shorter distances—as a possible reason
for the inability of fixed-location aerosol sampling to detect aerosols in the same setting. Fur-
thermore, the readily measureable breathing zone exposure described in this study strongly sup-
ports more extensive efforts to characterize the nature and sources of Mtb exposure and air
transmission routes in different environments. Such knowledge is critical to inform the most
robust mitigation recommendations for both occupational and residential settings; however it is
heavily reliant on the development of practical tools to enable a more thorough understanding
of the factors which determine Mtb particle production and transmission.
In clinical studies, there is a continued reliance on conventional bacteriology of sputum sam-
ples as a surrogate for infectiousness. Notwithstanding the practical benefits associated with a
population that is easily sampled and analyzed, the dependence on sputum obscures critical
uncertainties about the factors which ensure successful transmission of Mtb bacilli from one
individual to another. These include the number of infectious particles produced, the number of
bacilli within an infectious particle, the anatomical origin and composition of infective droplets,
and the physiological adaptations which enable airborne transmission and viability of Mtb. Crit-
ically, for a disease which is driven by the host immune response [10], there is also incomplete
knowledge about the mechanical, physiological, and immunopathological processes which
underlie the production of infectious aerosols. Addressing these questions is not a simple under-
taking, and requires the ability to observe in real time the release of Mtb particles by an infected
patient, and to integrate this information within a complex dataset describing the host, bacterial,
and environmental parameters which contribute to successful transfer of infecting bacilli to a
new recipient. In other words, developing a systems biology of TB transmission.
Towards this end, we designed a small personal clean room in which airborne Mtb could be
isolated in a clinical setting. Here, we detail the development and characterization of this

PLOS ONE | DOI:10.1371/journal.pone.0146658 January 25, 2016 2 / 16

 Respiratory Aerosol Sampling Chamber (RASC)

Respiratory Aerosol Sampling Chamber (RASC), and present initial data suggesting its poten-
tial to accommodate newly diagnosed TB cases for enumeration and morphological characteri-
zation of Mtb particles. Our ultimate goal is to utilize this information to generate models
predicting risks of TB exposure in specific environmental conditions, and to apply these mod-
els in developing practical interventions—clinical, operational, and infrastructural—to curb TB
transmission in a region characterized by an elevated force of infection [11].

Materials and Methods

Ethics Statement
Approval for the human aspects of this research was obtained from the University of Cape
Town Faculty of Health Sciences Human Research Ethics Committee (HREC/REF: 680/2013).
Participants provided written informed consent prior to acceptance into the study.

Experimental Description of RASC

The RASC (Fig 1) developed for this study utilized the basic structure of a sputum collection
booth (VividAir Cape Town, South Africa) which was modified to function as a small, personal
cleanroom with a volume of 1.4 m3, enabling the quantitation and characterization of aerosol
particles emitted by a seated study participant during normal breathing, coughing, and talking.
The RASC was positioned in a community clinic, within a well-ventilated room undergoing 12
air changes per hour (ACH) of N95-filtered cross-flow ventilation. Attending staff were
required to wear N95 respiratory masks throughout the experimental course.
During prolonged sampling procedures, study participants were able to use a small televi-
sion and entertainment system incorporated into the RASC. The RASC was hermetically
sealed, and all incoming air was HEPA filtered. The effects of HEPA-filtered air washes on the
background particle counts are shown in Fig 2A. The total particle count before the air wash
was typically around 300 particles/cm3. After air wash, and prior to patient sampling, the inte-
grated total background particle counts were generally fewer than 30 particles/cm3.
Study participants were required to wear full-body DuPont Tyvek suits throughout sampling
procedures in order to minimize background particle clutter. The sealing door, incorporated a
glass window for observation, could be opened externally as well as by subjects from inside the
device. A small electric fan maintained internal air circulation (mixing), and temperature,
humidity, and carbon dioxide levels were continuously monitored (Thermo-Hygro-CO2 moni-
tor, MIC Meter Industry Co, Taiching, Taiwan). In the absence of air sampling or purging, the
ventilation was determined as less than 0.1 ACH, and air-sampling rates during study protocols
varied between 50–200 liters per minute (l/min). The sampled air being removed from the
RASC was replaced by HEPA filtered air while the discharge after the sampling systems was
HEPA filtered and released into the room. Carbon dioxide, a natural tracer gas produced during
normal respiration, was used to quantify the volume of expired air that was sampled during
each procedure. A high-volume exhaust system enabled purging of the booth at 500 l/min via a
high-flow HEPA filter which was used prior to the patient’s exit from the RASC at the end of
each study. Surface cleaning and inbuilt ultra-violet light irradiation were used to ensure equip-
ment and surface sterilization, and to reduce cross contamination between studies.

Bioaerosol Characterization
The RASC was equipped with a suite of continuously operating particle sensors to characterize
aerosols with a temporal response of a few seconds. Central to the real-time measurement of
aerosol particle concentration and particle size distribution was an Aerodynamic Particle Sizer

PLOS ONE | DOI:10.1371/journal.pone.0146658 January 25, 2016 3 / 16

 Respiratory Aerosol Sampling Chamber (RASC)

Fig 1. The Respiratory Aerosol Sampling Chamber (RASC). (A) Photograph of the RASC (with the door open) on site in a community TB clinic (1)
aerodynamic particle sizer (2) Filter samplers (3) Andersen impactor (4) Mixing fan (5) CO2, temperature and RH (6) PM10 impactor (7) Chair for participant.
(B) Block diagram depicting the fluidic and electronic configuration of the RASC. Thick connecting lines indicate airflow and aerosol paths; thin lines indicate
electronic connections. All air leaving the RASC is HEPA filtered.
doi:10.1371/journal.pone.0146658.g001

PLOS ONE | DOI:10.1371/journal.pone.0146658 January 25, 2016 4 / 16

 Respiratory Aerosol Sampling Chamber (RASC)

Fig 2. The in-built APS characterizes the particle size distribution spectrum within the RASC. (A) Typical background particle spectrum before and
after air wash. Note the 10-fold decrease in particle counts across all size ranges following the air wash. Total count from a typical five seconds of sampling.
(B) Artificial dry release of fluorescent polystyrene latex (PSL) microspheres. The APS instrument groups all particles with an aerodynamic diameter less
than 0.523 μm in the number bin on the far left of each chart. Note that the 1μm release (b) contained approximately ten times more particles in the release.
Total count from ten seconds of sampling at the peak of particle concentration. (Inset) corresponding SEM images of the released particles.
doi:10.1371/journal.pone.0146658.g002

(APS Model 3321, TSI, Shoreview, MN USA) and a Laser Scattering Particle Counter (Model
3016, Lighthouse Worldwide Solutions, Medford, Oregon USA). These two instruments worked
in concert to characterize the aerosol particle environment in the RASC. The APS provided
high-resolution aerodynamic (as opposed to physical) particle size distributions in the particle
size range 0.7 μm to 20 μm. We assumed, based upon known deposition sites in the distal alveoli
of the lungs, that the modal size of Mtb-infected sputum droplets emitted by patients would be
centered within this range. Therefore, the measure of aerodynamic sizing was expected to
mimic the way aerosols are sized and transported through the human respiratory system. The
laser scattering particle counter provided lower resolution measurement of physical particle size
but was particularly useful for the study of submicron-sized particles with a minimum size of

PLOS ONE | DOI:10.1371/journal.pone.0146658 January 25, 2016 5 / 16

 Respiratory Aerosol Sampling Chamber (RASC)

0.3 μm. In general, the information obtained from the APS was considered more likely to be rel-
evant to the predicted behavior of particles in the lungs; however, the correlation between mea-
surements on the two instruments (APS and laser scattering particle counter) provided rich
data on the bio-aerosols generated by the patient during occupation of the RASC.

Particle Sampling System

Our longer term goal is to utilize the RASC to define the morphological and molecular charac-
teristics that might be key to transport of particles in the respiratory system and, in turn, deter-
mine the infectiveness of the pathogen. To this end, three main types of sample collectors were
installed in the RASC, specifically: liquid impingers, physical impactors, and filters.
The liquid impingers (BioSampler, SKC, Eighty Four, PA) were set to sample at 12.5 l/min
into liquid medium (PBS). Liquid impaction was included in the design of the RASC in order to
enable the future use of FACS-based sorting of aerosolized Mtb bacilli from patients for down-
stream genotypic, phenotypic and infectivity characterizations, while solid impaction should
additionally provide samples for whole-genome sequencing and mass spectrometry imaging to
characterize the composition of airborne material collected on the impaction plates [12].
Two types of physical impactors were employed. A Dekati three-stage impactor (PM10
Impactor, Dekati, Kangasala, Finland) was set to sample at 30 l/min. The Dekati impactor was
selected since it collects particles in three size ranges: for particles that have a density of 1 g.cm-
3
, the first stage collects particles >14.1 μm, the second stage collects particles in the 14.1 μm to
3.5 μm range, and the final stage collects particles in the 3.5 μm to 1.4 μm range. Therefore, this
impactor was considered especially useful for imaging respired aerosol matter. A Six-stage Via-
ble Andersen Cascade Impactor (Model 10830-EPD, Thermo Scientific, USA) was also
employed since this device measures culturable particle counts as a function of particle size:
each of the six stages captures smaller particles [13]. The airflow rate of the sampler was 28.3 l/
min and the impaction plates comprised a thick coating of agar growth media to minimize
bounce and enable microbial growth. Each of the impactor samplers was backed with a 0.4 μm
filter to collect any particles that bypassed the impaction stages.
We also used two types of filter media to capture samples: micro-porous polycarbonate
membranes and gel filters. The polycarbonate membrane filters (Sterlitech Corporation, WA
USA) were nominally rated at 0.4 μm and provided high collection efficiencies when used in 47
mm open-faced filter holders, with flow rates through the filters usually around 28 l/min. The
gel filters (Model 12602-37-ALK, Sartorius, Goettingen, Germany) were 37 mm in diameter,
and rated to capture >99.99% of both bacterial and viral particles. These filters were selected
since the gel-filtering medium maintains moisture so that collected organisms can maintain
significant viability. Furthermore, the gel medium is soluble in water enabling high levels of
organism recovery from the filters. The filters, which sample at up to 20 l/min, were used for
shorter periods of time during the experiments to reduce dehydration and increase the likeli-
hood of capturing viable organisms.

Bacteriological, Molecular and Imaging Analyses

The M. smegmatis::gfp reporter mutant expressing green fluorescent protein (GFP) was con-
structed by introduction of the pMSP12GFP plasmid [14] into M. smegmatis mc2155. For
determination of CFU in controlled aerosol release experiments, bacilli were propagated in liq-
uid 7H9 medium (Difco) supplemented with 0.05% Tween-80, 0.2% glycerol and albumin/
NaCl/ glucose (ADC) complex and/or isolated on solid 7H10 agar (Difco) supplemented with
0.2% glycerol and oleic acid/albumin/dextrose/catalase (OADC) complex. Owing to the selec-
tivity provided by the aph cassette on pMSP12GFP, all media contained kanamycin (Sigma) at

PLOS ONE | DOI:10.1371/journal.pone.0146658 January 25, 2016 6 / 16

 Respiratory Aerosol Sampling Chamber (RASC)

a final concentration of 20 μg/mL. For the pilot TB studies, the Andersen impactor contained
solid 7H10 agar (Difco) plates supplemented with 0.2% glycerol and oleic acid/albumin/dex-
trose/catalase (OADC) complex.
PCR genotyping of the Mtb RD9 region was performed by amplifying genomic DNA from
putative Mtb CFU using the primer set RD9/qRTF (5’-tgagtggcgatggtcaacac-3’)
and RD9/qRTR (5’-gatggcgttcggaaagaaac-3’).
For analysis by scanning electron microscopy (SEM), samples were air-dried on aluminium
foil and imaged without coating using a Zeiss/Leo 1450 Scanning Electron Microscope
equipped with a lanthanum hexaboride (LaB6) cathode. Images were captured in secondary
electron mode at 10 kV.

Results
Aerosol characterization of the RASC
In advance of patient testing in the RASC, it was important to characterize the behavior of parti-
cles in the chamber and demonstrate that accurate results could be obtained from both artificial
and live aerosol releases. For artificial release, fluorescent polystyrene latex (PSL) microspheres
(Sigma-Aldrich Corp, St. Louis, MO and Polysciences Inc., Warrington, PA) were “dry” aerosol-
ized into the chamber. The resulting size distribution graphs (Fig 2B) confirmed accurate cali-
bration of the APS system for the three bead sizes: 0.5 μm, 1.0 μm, and 2.0 μm. Quantitative
SEM methodology (Fig 2B) indicates that the actual bead sizes were 0.47 μm, 0.91 μm and
1.74 μm. After PSL beads were introduced into the RASC, they were observed by the APS within
6 seconds and were well mixed within 20 seconds of release. Analysis of the APS size-bins asso-
ciated with the 2 μm beads indicated that the count of PSL beads relative to background clutter
prior to release was approximately 50. This ratio was defined as the “signal to clutter ratio” and,
for the 1 μm beads, the ratio was calculated as approximately 250.
For live aerosol release, we utilized a fluorescent reporter mutant of M. smegmatis mc2155,
since the non-pathogenic mycobacterial model organism is of equivalent size and morphology
to M. tuberculosis [15]. Suspensions of the exponentially growing M. smegmatis::gfp reporter
strain were aerosolized into the RASC at several different concentrations to test the capacity of
the system to characterize bioaerosols of mycobacteria accurately. Two different protocols
were used to aerosolize M. smegmatis::gfp: an adapted micro paint sprayer that was fed by high
pressure HEPA-filtered air to create a “wet” aerosol with heavy particle clustering; and an ultra-
sonic, vibrating orifice nebulizer (Model DPD-1b, BioProcess Diagnostics, Albuquerque, NM)
that fed aerosol through a heated tube to create a “dry” aerosol. The aerosol was considered
“dry” if its aerodynamic size distribution was unchanged over time scales of 10–20 s after exit-
ing the drying tube. The two different release protocols had differential effects on the particle
size distribution (Fig 3). Wet release skewed the particle size distribution to larger particles
with a secondary maximum at a cluster size near 2 μm. In contrast, the dry release resulted in a
particle size distribution with a single maximum near 0.8 μm, which was indicative of much
smaller amounts of agglomeration and clustering.
We also used chalk dust (Visolite Tracer Compound PN 100–0164, BHA Group, Inc., Kan-
sas City, MO) to characterize the loss of particles over time owing to surface deposition as a
function of particle size (data not shown). Following release of chalk dust into the RASC, the
peak was observed at 1 μm, but there were significant numbers of particles up to 10 μm in size.
In 6 minutes following dust aerosolization, both total particle counts and the 1 μm particle
counts decreased at a rate of approximately 4% per minute. Slightly less than 10% of this loss
could be attributed to the APS sampling, which permanently removed particles from the cham-
ber at a rate of 5 litres per minute. The 5 μm particle loss rate was 9% per minute.

PLOS ONE | DOI:10.1371/journal.pone.0146658 January 25, 2016 7 / 16

 Respiratory Aerosol Sampling Chamber (RASC)

Fig 3. The effect of aerosol hydration on particle size distribution. The graph shows APS measurements of “wet” and “dry” releases of M. smegmatis::
gfp. (see text for details).
doi:10.1371/journal.pone.0146658.g003

Capture of viable organisms: the Andersen Cascade Impactor

In order to confirm the ability to capture live mycobacteria in the RASC, a Six-Stage Viable
Andersen Cascade Impactor was tested in a series of 3 replicates of 3 wet M. smegmatis::gfp test
releases (Fig 4). These experiments incorporated a range of cell concentrations—from ~0.2
CFU/L to ~22 CFU/L of RASC air—to ensure the capacity of the RASC to capture the variable
numbers of Mtb bacilli expected to be produced by TB patients. The Andersen impactor com-
prised six stages, with different aerodynamic particle sizes captured on the solid 7H10 agar
plate contained in each stage. The first plate (A1) collected the largest particles (>8 μm) and
the last plate (A6) collected the smallest particles (< 1 μm). Each of the other plates had a col-
lection size range between the two extremes. At the highest concentration range, the colonies
growing under the jets of the Andersen device were too numerous to count individually with
accuracy. In the bottom row, for the extremely dilute sample, a mean of 15 colonies were cap-
tured in the impactor giving a capture efficiency of 49.3% of expected CFU. As the protocol
evolves, and more is known about the patient output, the sampling times for the impactor can
be adjusted to be less sensitive (shorter sampling times) or more sensitive (longer sampling
times). Based on preliminary data, we expect that an active TB patient will generally
produce < 1 CFU/L, which lies between the last two rows in Fig 4A. After each release, we
swabbed a 15x15 cm square with a sterile felt filter moistened with sterile PBS solution on three
walls of the chamber (just above the release point, on the wall opposite the release point and on
a side wall) and we did not grow any colonies from any of the releases indicating that there was
practically no adhesion of viable aerosol to walls of the chamber.

Imaging bacilli captured after aerosolization

To study the morphology of particles (including bacteria) isolated in the RASC, samples were
captured onto metal plates in a three-stage Dekati impactor. The utility of this system was vali-
dated following release of M. smegmatis::gfp: a representative image (Fig 4B, left panel) presents
a 20 μm × 20 μm view of a metal foil inserted on the final Dekati impactor plate which collects
particles in the 2.5 μm to 1 μm range. A single M. smegmatis::gfp bacillus is visible near the cen-
ter of the image. The identity of the M. smegmatis test strain was confirmed in the right-hand

PLOS ONE | DOI:10.1371/journal.pone.0146658 January 25, 2016 8 / 16

 Respiratory Aerosol Sampling Chamber (RASC)

Fig 4. Isolation and visualization of viable mycobacteria in the RASC. (A) M. smegmatis::gfp growth on solid 7H10 agar plates from the Six-Stage Viable
Andersen Cascade Impactor after wet release of 200 μl diluted culture into the RASC (30 000, 3000 and 300 colony forming units—CFU). The columns
indicate the particle sizes captured on each plate across the 6 stages of the impactor, and the rows indicate the estimated total number of CFU passing
through the impactor. Each release was repeated three times and the mean and SD for each plate are presented below the typical growth pattern distribution
seen in the particle release. In all the releases the sampling was run for 5 minutes at 28 l/min resulting in the potential total capture of 3000, 300 and 30 CFU
respectively. (B) SEM (left) and fluorescent microscopy (right) of M. smegmatis::gfp isolated on a PM10 impactor following experimental release.
doi:10.1371/journal.pone.0146658.g004

PLOS ONE | DOI:10.1371/journal.pone.0146658 January 25, 2016 9 / 16

 Respiratory Aerosol Sampling Chamber (RASC)

panel, containing a 20 μm × 20 μm fluorescent photomicrograph of the same sample, in which

the bacillus was brightly fluorescent. It was noticeable, though, that this particular organism
appeared to exhibit a slightly less curved morphology than that observed in the SEM image,
perhaps as a consequence of the fluorescence signal.

Demonstrating the utility of the RASC in a TB clinic

In order to demonstrate key capabilities of the RASC, a standard protocol was developed that was
approved by the University of Cape Town Faculty of Health Sciences Human Research Ethics
Committee. Participants provided written informed consent prior to acceptance into the study.
Adult patients with a recent diagnosis of TB were recruited immediately prior to initiation
of therapy. After entering the RASC, the sealed door was closed; the surgical mask removed,
and a cough sputum sample provided. Following occupation of the RASC, carbon dioxide lev-
els increased consistently, reaching levels of approximately 4,000 parts per million (ppm)
above ambient levels owing to exhaled breath. Given that the concentration of CO2 in exhaled
air is close to 40,000 ppm, the target concentration of 4,000 ppm above ambient atmospheric
level (typically ~400 ppm) indicated that approximately 10% of the air in RASC comprised
expired air at that point. The typical time required for CO2 levels to reach the target concentra-
tion was approximately 25 to 30 min. During this period, only the APS was set to sample from
the chamber, at a rate of 5 l/min. Once the CO2 level reached the 4,000 ppm target, sampling
was initiated at approximately 50 l/min: the first 10 minutes were utilized for collection of bac-
teriology samples (Andersen impactor and gel filter), after which a further 10 minutes was
allowed for imaging and molecular samples (PM10 impactor and polycarbonate filter). This 20
minute sampling period was expected to maintain CO2 levels at approximately the target con-
centration. Finally, high flow sampling was initiated at 300 l/min for PCR samples, a 10 minute
process in which the CO2 concentration was expected to return to background ambient levels.
Details of the flows out of the RASC are given in Table 1. The total sample therefore repre-
sented the expired air over the full study period of approximately 60 minutes. The mathemati-
cal relationship between CO2 production, resultant CO2 levels, and effect of sampling rates, is
described in Appendix 1 (S1 File).
During the entire course of the 60 minute protocol, aerodynamic particle size was deter-
mined at 10 second intervals along with continuous monitoring of CO2, humidity and temper-
ature levels. Fig 5 details RASC CO2 concentrations and particle size counts (in the 1–2.5 μm
size range) for a single patient. It is notable that changes in particle count parallel the CO2

Table 1. Sampling rates of the capture devices and the periods in which they are used.

Sampling period Sampling Device Flow out of booth Total outflow

Bacteriology sampling Andersen Impactor 28 l/min
Gelatine Filter 20 l/min
Aerodynamic Particle Sizer 5 l/min 53 l/min
Imaging Sampling PM10 Impactor 30 l/min
Polycarbonate Filter 20 l/min
Aerodynamic Particle Sizer 5 l/min 55 l/min
PCR sampling Felt Filter Approx. 300 l/min
Aerodynamic Particle Sizer 5 l/min Approx. 300 l/min

The air being removed from the sampling space is replaced through large HEPA filters. The laser scattering particle counter and CO2, temperature and
humidity monitor run at low flow rates and the exhaust air is returned to the sampling space. During the patient sampling the liquid impingers were not
used.

doi:10.1371/journal.pone.0146658.t001

PLOS ONE | DOI:10.1371/journal.pone.0146658 January 25, 2016 10 / 16

 Respiratory Aerosol Sampling Chamber (RASC)

Fig 5. Particle production as a function of respiration in a clinical TB patient. CO2 concentration (solid line and left ordinate) and particle counts (dots
and right ordinate) in the 1–2.5 μm size range for a TB patient.
doi:10.1371/journal.pone.0146658.g005

concentrations within the RASC: as the patient exhaled and coughed over the first non-sam-
pling 30 min period, the CO2 and particle count built up; subsequently, the counts plateaued,
ultimately reaching an approximate steady state as the samplers evacuated the sample space;
and, finally, the counts dropped rapidly as the chamber was exhausted with a high-volume
sampler. From a preliminary analysis of data collected, the temperature increased by 5.9
degrees during the whole sampling period (24.0 to 29.9°C, n = 19) and the humidity increased
by 10.9% from the start of the sampling to the bacterial and imaging sampling and then
reduced by 11.9% during the PCR sampling (52.7% through 63.6% to 51.7%, n = 19).
It is evident that the aerosol particle data (such as those shown in Fig 5) will become most
important when correlations between particle counts and disease status can be drawn by com-
parisons across multiple patients. The preliminary patient samples are too few to enable any
definitive inferences; nevertheless, these early results enabled a key observation from the very
first patient tested in the RASC: a single putative Mtb bacillus is visible in an SEM image
obtained under high magnification of a 10 μm by 10 μm region of the lower plate of the Dekati
sampler (Fig 6). In the full image, multiple similar rod-like organisms were clearly visible (data
not shown). The presence of Mtb in the chamber was confirmed by culture of a colony isolated
on plate 5 of the Andersen impactor, and preliminary genotyping through PCR amplification
of the RD9 locus which differentiates Mtb and M. canettii from other mycobacteria within the
Mycobacterium tuberculosis Complex [16].
A number of other particles were visible on the Dekati plate (Fig 6), and these likely arose
from the respired air from the patient, especially since the particulate background in the cham-
ber was quite low. Most of these particles were characterized by a surrounding “halo”, an

PLOS ONE | DOI:10.1371/journal.pone.0146658 January 25, 2016 11 / 16

 Respiratory Aerosol Sampling Chamber (RASC)

Fig 6. Isolation of Mtb from a TB patient. SEM image of patient sample impacted on the lower plate of the PM10 impactor. The dimensions and
morphology of the rod-shaped structure (denoted by *) are consistent with the presence of Mtb bacilli in the untreated TB patient. There is also evidence of
multiple “splats” of unknown identity (one example is denoted by **) which might comprise organic matter derived from patient lung or respiratory tract. Note
the “halo” structures (dark shadows) surrounding each particle.
doi:10.1371/journal.pone.0146658.g006

observation which suggested the intriguing possibility that this feature was caused by dried liq-
uid that might contain complementary biomarkers of Mtb infection. Again, however, more
samples—from bothTB patients and control, non-TB active individuals—will need to be pro-
cessed before a definitive correlation can be dawn between halo formation (and composition)
and TB disease status.

Discussion
Mtb survival during the airborne phase of its life cycle is an understudied area of research. In
contrast to prior studies which have focused on culture of aerosols created immediately follow-
ing periods of forced coughing [17,18], the RASC enables study of more aged airborne organ-
isms that have been selected or adapted for environmental survival.

PLOS ONE | DOI:10.1371/journal.pone.0146658 January 25, 2016 12 / 16

 Respiratory Aerosol Sampling Chamber (RASC)

The RASC was developed to enable quantitation and characterization of the bio-aerosols
emitted during normal breathing and coughing of TB patients in future clinical studies, and to
incorporate the resulting information into a refined model of TB transmission. Critically, this
long-term goal requires that the performance levels of the sampling sub-systems within the
RASC be calibrated to ensure reproducibly robust bio-aerosol detection, quantification, and
isolation. The results presented here suggest that the RASC is performing as intended and can
be usefully applied in collecting and characterizing patient-generated aerosols, and for assess-
ing changes in the nature and content of respired aerosols over time. The air mixing in the
RASC is designed to replicate air currents that occur in normal indoor environments, and
which decrease particle settling compared to still air. Information that can be acquired in the
RASC therefore includes the number, size distribution (physical and aerodynamic), and mor-
phology of aerosolized particles.
Infection of a new human host by Mtb depends on the generation of a source aerosol popu-
lation that is defined both by the number of bacilli and the phenotypic and genomic character-
istics of the organisms that are emitted from the infected individual into the environment. In
addition, it is important to consider the evolution of the emitted particles as a function of parti-
cle size and morphological structure since these parameters alter how they travel and are trans-
formed in the environment [19]. Particularly important in this context are the processes by
which the aerosol particles dry (with environmentally variable kinetics) into derivative
(smaller) particles of size and morphology that have been shown to promote travel deeper into
alveolar spaces in the lungs of the recipient [20].
We did demonstrate the novel ability to image individual airborne rod shaped organisms.
However, we are currently unable to definitely identify these as Mycobacteria or other contami-
nating organisms. The general morphology of individual mycobacteria is a rod shape, approxi-
mately 0.2–0.5μm in cross-sectional diameter and 1–4μm in length [21]. However, bacillary
dimensions are known to vary according to whether observed in sputum and bronchoalveolar
lavage (BAL) fluid or in TB cavities: organisms found in sputum and BAL are significantly lon-
ger, whereas the lengths of bacilli harvested in cavities are more reminiscent of stationary
phase bacilli in vitro [22]. Under certain growth conditions, colonies can also appear in groups
as elongated “cord” structures [23]. It seems plausible, therefore, that some proportion of the
emitted aerosol produced by the infected donor may retain the elongated features of the micro-
organisms in the airways depending on the composition of the airway fluid in which they are
ejected. This is also likely to be a function of the region of the lungs of origin, as it is known
that the periphery is largely composed of surfactant whereas the ciliated airways possess mucus
containing glycolipids and glycoproteins. The presence of inflammatory cells, immune media-
tors, and general debris associated with infection almost certainly contributes to a complex sce-
nario which cannot easily be predicted and requires direct experimental evidence.
Once in air, drying of the aerosol will proceed at a rate dictated by ambient temperature and
humidity (Appendix 2, S2 File). The impact of exposure for individuals breathing nearby in
either residential or occupational settings will therefore vary in both space and time, and must
be heavily influenced by particle morphology at the time of inhalation. With regard to the
influence of other key single particle features such as shape and agglomerate density, there are
known phenotypes that might promote a rod shaped presentation (referred to as “cording” in
Koch’s original description of Mtb [24]) that would further enhance the tendency for periph-
eral lung deposition [25].
Our preliminary results suggest that RASC sampling has the potential to give qualitative
information about particles and organisms which have been exhaled and remained airborne. A
standardized protocol, together with the ability to monitor the proportion of sampled air as a
function of total exhaled air, should enable quantitative assessments of particle and organism

PLOS ONE | DOI:10.1371/journal.pone.0146658 January 25, 2016 13 / 16

 Respiratory Aerosol Sampling Chamber (RASC)

production rates from infected individuals. Importantly, the potential to acquire quantitative
data suggests the additional opportunity for inter-personal variation to be assessed and for
intra-personal variability to be determined in repeated investigations, including after initiation
of treatment.
Transmission within TB endemic populations can be modeled [26] from the measured vol-
ume of air rebreathed from others which is determined by socio-environmental factors [27,28],
the proportion of rebreathed air derived from prevalent TB infected individuals in that popula-
tion, and the surviving pathogen concentration of exhaled air from an infected individual [5].
Quantitative assessments of pathogen load in exhaled air may allow more accurate modeling of
those individuals and factors which are driving the elevated rates of TB disease which charac-
terize high-burden settings such as our own [29].

Supporting Information
S1 File. Appendix 1 describing the mathematical relationship between CO2 production,
resultant CO2 levels, and effect of sampling rates in the RASC.
(DOCX)
S2 File. Appendix 2 describing the drying of the aerosol at a rate dictated by ambient tem-
perature and humidity.
(DOCX)

Acknowledgments
Academic and technical support was provided from The Tuberculosis Aerobiology Advisory
Group (TAAG) funded by the Bill and Melinda Gates Foundation.

Author Contributions
Conceived and designed the experiments: RW CM CEB WAB CJC AJH CER TJS JB JW AM
VS VM DFW. Performed the experiments: RW CM WAB CJC AM VS. Analyzed the data: RW
CM WAB CJC CI NM JW AM VS. Contributed reagents/materials/analysis tools: RW CB
WAB CJC AJH CER CI NM JW VM DFW. Wrote the paper: RW CM CEB WAB CJC AJH
CER TJS JB CI NM JW AM VS VM DFW.

References
1. Riley RL, Mills CC, Nyka W, Weinstock N, Storey PB, Sultan LU, et al. (1959) Aerial dissemination of
pulmonary tuberculosis. A two-year study of contagion in a tuberculosis ward. Am J Hygiene 70: 2.
2. Riley RL, Mills CC, O'Grady F, Sultan LU, Wittstadt F, Shivpuri DN (1962) Infectiousness of air from a
tuberculosis ward. Ultraviolet irradiation of infected air: comparative infectiousness of different patients.
Am Rev Respir Dis 85: 511–525. PMID: 14492300
3. Escombe AR, Moore DAJ, Gilman RH, Pan W, Navincopa M, Ticona E, et al. (2008) The infectiousness
of tuberculosis patients coinfected with HIV. PLoS Med 5: e188. doi: 10.1371/journal.pmed.0050188
PMID: 18798687
4. Dharmadhikari AS, Basaraba RJ, Van Der Walt ML, Weyer K, Mphahlele M, Venter K, et al. (2011) Nat-
ural infection of guinea pigs exposed to patients with highly drug-resistant tuberculosis. Tuberculosis
(Edinb) 91: 329–338.
5. Issarow CM, Mulder N, Wood R (2015) Modelling the risk of airborne infectious disease using exhaled
air. J Theor Biol 372: 100–106. doi: 10.1016/j.jtbi.2015.02.010 PMID: 25702940
6. Mastorides SM, Oehler RL, Greene JN, Sinnott JT 4th, Kranik M, Sandin RL (1999) The detection of air-
borne Mycobacterium tuberculosis using micropore membrane air sampling and polymerase chain
reaction. Chest 115: 19–25. PMID: 9925058
7. Wan GH, Lu SC, Tsai YH (2004) Polymerase chain reaction used for the detection of airborne Myco-
bacterium tuberculosis in health care settings. Am J Infect Control 32: 17–22. PMID: 14755230

PLOS ONE | DOI:10.1371/journal.pone.0146658 January 25, 2016 14 / 16

 Respiratory Aerosol Sampling Chamber (RASC)

8. Chen P-S, Li C-S (2005) Quantification of airborne Mycobacterium tuberculosis in health care setting
using real-time qPCR coupled to an air-sampling filter method. Aerosol Science and Technology 39:
371–376.
9. Matuka O, Singh TS, Bryce E, Yassi A, Kgasha O, Zunqu M, et al. (2015) Pilot study to detect airborne
Mycobacterium tuberculosis exposure in a South African public healthcare facility outpatient clinic. J
Hosp Infect 89: 192–196. doi: 10.1016/j.jhin.2014.11.013 PMID: 25623206
10. Ernst JD (2012) The immunological life cycle of tuberculosis. Nat Rev Immunol 12: 581–591. doi: 10.
1038/nri3259 PMID: 22790178
11. Middelkoop K, Bekker LG, Liang H, Aquino LD, Sebastian E, Myer L, et al. (2011) Force of tuberculosis
infection among adolescents in a high HIV and TB prevalence community: a cross-sectional observa-
tion study. BMC Infect Dis 11: 156. doi: 10.1186/1471-2334-11-156 PMID: 21631918
12. van Belkum A, Welker M, Erhard M, Chatellier S (2012) Biomedical mass spectrometry in today's and
tomorrow's clinical microbiology laboratories. J Clin Microbiol 50: 1513–1517. doi: 10.1128/JCM.
00420-12 PMID: 22357505
13. Andersen AA (1958) New sampler for the collection, sizing, and enumeration of viable airborne parti-
cles. J Bacteriol 76: 471–484. PMID: 13598704
14. Chan K, Knaak T, Satkamp L, Humbert O, Falkow S, Ramakrishnan L (2002) Complex pattern of Myco-
bacterium marinum gene expression during long-term granulomatous infection. Proc Natl Acad Sci U S
A 99: 3920–3925. PMID: 11891270
15. Gordon RE, Smith MM (1953) Rapidly growing, acid fast bacteria. I. Species' descriptions of Mycobac-
terium phlei Lehmann and Neumann and Mycobacterium smegmatis (Trevisan) Lehmann and Neu-
mann. J Bacteriol 66: 41–48. PMID: 13069464
16. Warren RM, Gey van Pittius NC, Barnard M, Hesseling A, Engelke E, de Kock M, et al. (2006) Differen-
tiation of Mycobacterium tuberculosis complex by PCR amplification of genomic regions of difference.
Int J Tuberc Lung Dis 10: 818–822. PMID: 16850559
17. Fennelly KP, Martyny JW, Fulton KE, Orme IM, Cave DM, Heifets LB (2004) Cough-generated Aero-
sols of Mycobacterium tuberculosis. American Journal of Respiratory and Critical Care Medicine
169:604–609. doi: 10.1164/rccm.200308-1101OC PMID: 14656754
18. Fennelly KP, Jones-López KC, Ayakaka I, Kim S, Menyha H, Kirenga B, et al. (2012) Variability of Infec-
tious Aerosols Produced during Coughing by Patients with Pulmonary Tuberculosis. American Journal
of Respiratory and Critical Care Medicine 186:450–457. doi: 10.1164/rccm.201203-0444OC PMID:
22798319
19. Crowder TM, Rosati JA, Schroeter JD, Hickey AJ, Martonen TB (2002) Fundamental effects of particle
morphology on lung delivery: predictions of Stokes' law and the particular relevance to dry powder
inhaler formulation and development. Pharm Res 19: 239–245. PMID: 11934228
20. Swift D, Asgharian B, Kimbell JS (2007) Use of mathematical aerosol deposition models in predict-
ing the distribution of inhaled therapeutic aerosols. In: Hickey AJ, editor. Inhalation Aerosols, Physi-
cal and Biology Basis for Therapy, Second Edition. Second ed. New York: Informa Healthcare. pp.
55–82.
21. Thanky NR, Young DB, Robertson BD (2007) Unusual features of the cell cycle in mycobacteria: polar-
restricted growth and the snapping-model of cell division. Tuberculosis (Edinb) 87: 231–236.
22. Eum SY, Kong JH, Hong MS, Lee YJ, Kim JH, Hwang SH, et al. (2010) Neutrophils are the predominant
infected phagocytic cells in the airways of patients with active pulmonary TB. Chest 137: 122–128. doi:
10.1378/chest.09-0903 PMID: 19749004
23. Caceres N, Vilaplana C, Prats C, Marzo E, Llopis I, Valls J, et al. (2013) Evolution and role of corded
cell aggregation in Mycobacterium tuberculosis cultures. Tuberculosis (Edinb) 93: 690–698.
24. Koch R (1882) The Aetiology of Tuberculosis, Max Pinner Dr. and Mrs., ed. New York: National Tuber-
culosis Association.
25. Glickman MS, Cox JS, Jacobs WR Jr. (2000) A novel mycolic acid cyclopropane synthetase is required
for cording, persistence, and virulence of Mycobacterium tuberculosis. Mol Cell 5: 717–727. PMID:
10882107
26. Andrews JR, Morrow C, Walensky RP, Wood R (2014) Integrating social contact and environmental
data in evaluating tuberculosis transmission in a South African township. J Infect Dis 210: 597–603.
doi: 10.1093/infdis/jiu138 PMID: 24610874
27. Wood R, Morrow C, Ginsberg S, Piccoli E, Kalil D, Sassi A, et al. (2014) Quantification of shared air: a
social and environmental determinant of airborne disease transmission. PLoS One 9: e106622. doi:
10.1371/journal.pone.0106622 PMID: 25181526

PLOS ONE | DOI:10.1371/journal.pone.0146658 January 25, 2016 15 / 16

 Respiratory Aerosol Sampling Chamber (RASC)

28. Richardson ET, Morrow CD, Kalil DB, Ginsberg S, Bekker LG, Wood R (2014) Shared air: a renewed
focus on ventilation for the prevention of tuberculosis transmission. PLoS One 9: e96334. doi: 10.1371/
journal.pone.0096334 PMID: 24804707
29. Andrews JR, Hatherill M, Mahomed H, Hanekom WA, Campo M, Hawn TR, et al. (2015) The dynamics
of QuantiFERON-TB gold in-tube conversion and reversion in a cohort of South African adolescents.
Am J Respir Crit Care Med 191: 584–591. doi: 10.1164/rccm.201409-1704OC PMID: 25562578

PLOS ONE | DOI:10.1371/journal.pone.0146658 January 25, 2016 16 / 16