Subscriber access provided by Fudan University
Article
Medical swab analysis using desorption electrospray ionization mass spectrometry (DESI-MS) – a non-invasive approach for mucosal diagnostics Pamela Pruski, David A. MacIntyre, Holly Victoria Lewis, Paolo Inglese, Goncalo dos Santos Correia, Trevor T. Hansel, Phillip R. Bennett, Elaine Holmes, and Zoltan Takats Anal. Chem., Just Accepted Manuscript • DOI: 10.1021/acs.analchem.6b03405 • Publication Date (Web): 27 Dec 2016 Downloaded from http://pubs.acs.org on December 29, 2016
Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a free service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are accessible to all readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.
Analytical Chemistry is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.
Page 1 of 36
Analytical Chemistry
Page 1 of 27 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
1 2
Medical swab analysis using desorption electrospray ionization mass spectrometry (DESI-MS) – a non-invasive approach for mucosal diagnostics
3 4
Pamela Pruski†, David A. MacIntyre°, Holly V. Lewis°, Paolo Inglese†, Gonçalo DS Correia†, Trevor T. Hanselˠ, Phillip R. Bennett°, Elaine Holmes†, Zoltan Takats†*
5 6
†
7
Campus, London, SW7 2AZ (UK)
8
°
9
Biology, Imperial College London, London W12 0NN, (UK)
Computational and Systems Medicine, Imperial College London, South Kensington Imperial College Parturition Research Group, Institute of Reproductive and Developmental
10
ˠ
11
Imperial College London, W2 INY, (UK)
Imperial Clinical Respiratory Research Unit (ICRRU), St Mary’s Hospital, Mint Wing,
12 13
* Corresponding Author: Professor Zoltan Takats, Division of Computational and Systems
14
Medicine, Department of Surgery and Cancer, Imperial College London, South Kensington
15
Campus, Sir Alexander Fleming Building, London, SW7 2AZ. Telephone: 020 7594 2760.
16
Email:
[email protected] 17 18 19
Abstract. Medical swabs are routinely used worldwide to sample human mucosa for
20
microbiological screening with culture methods. These are usually time consuming and have
21
a narrow focus on screening for particular microorganism species. As an alternative, direct
22
mass spectrometric profiling of the mucosal metabolome provides a broader window into the
23
mucosal ecosystem. We present for the first time a least effort/least-disruption technique for
24
augmenting the information obtained from clinical swab analysis with mucosal metabolome
25
profiling using DESI-MS analysis. Ionisation of mucosal biomass occurs directly from a
26
standard rayon swab mounted on a rotating device and analysed by DESI MS using an
27
optimised protocol considering swab-inlet geometry, tip-sample angles and distances,
28
rotation speeds and reproducibility. Multivariate modelling of mass spectral fingerprints
29
obtained in this way readily discriminate between different mucosal surfaces and display the
30
ability to characterise biochemical alterations induced by pregnancy and bacterial vaginosis
31
(BV). The method was also applied directly to bacterial biomass to confirm the ability to
32
detect intact bacterial species from a swab. These results highlight the potential of direct
33
swab analysis by DESI-MS for a wide range of clinical applications including rapid mucosal
34
diagnostics for microbiology, immune responses and biochemistry.
35
ACS Paragon Plus Environment
Analytical Chemistry
Page 2 of 36
Page 2 of 27 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
36
Human mucosae are primary sites of contact and exchange with an organism’s external
37
environment. Mucosal surfaces not only act as a physical barrier for selective access
38
pathogen entry, but also as a complex ecosystem where multiple microbial species interact
39
among each other and with the host via the innate and acquired immune response
40
Through extensive chemical interactions, including excretion of antimicrobial compounds or
41
co-metabolism, host, symbiont and pathogen alike compete for niche dominance
42
However, an imbalance of the mucosal ecosystem caused by bacterial colonization can lead
43
to disease. For example, a stable environment of Lactobacillus spp. colonisation during
44
pregnancy protects against pathobiont colonisation of the reproductive tract, which is
45
strongly associated with increased risk of preterm birth and neonatal loss
46
mucosal cytokines reflect allergic inflammation, viral infection and asthma 6. Mucosal
47
analysis thus represents an important and expanding field with wide clinical applications and
48
huge diagnostic potential 7.
49
Medical cotton swabs are the standard mucosal sampling device and have been long used,
50
for microbiological culture, drug testing and genetic screening. Conventionally, swabs of
51
mucosal surfaces are used for culture-based characterisation of microbial colonization.
52
Commercial platforms using Matrix Assisted Laser Desorption Ionisation Time of Flight
53
(MALDITOF) MS based analysis of bacterial proteins (e.g. Bruker MALDI Biotyper and
54
bioMerieux VITEK MS) facilitate strain-level identification of bacteria from isolated, purified
55
cultures
56
addition of matrix to assist lysis and ionisation. Culture-independent approaches are also
57
increasingly used for characterisation of microbial composition in clinical samples. These
58
methods largely rely upon next-generation sequencing of hypervariable regions of the 16S
59
rRNA subunit of bacteria using region-specific primers and high-throughput DNA sequencing
60
10,11
61
host interactions. Integration of LC MS-derived metabonomic data with microbiomic data has
62
recently been used to evaluate host-microbiota interactions in mucosal samples in collected
63
from women with bacterial vaginosis, a common vaginal condition characterised by reduction
64
in commensal Lactobacillus species, elevated pH and a malodorous discharge12,13.
65
Moreover, BV is accompanied by increased diversity of anaerobic bacteria in the vaginal
66
microbiome including Atopobium, Gardnerella, Prevotella, Megasphaera, Dialister and
67
several other bacteria. These studies showed that these BV-associated bacteria (BVAB)
68
increase vaginal pH by producing short chained fatty acids (SCFA, e.g. butyrate, succinate
69
and propionate) and polyamines (e.g. putrescine and cadaverine) and are also capable of
70
utilizing lactic acid as an energy source. Despite the quality of the information provided by
8,9
4,5
1,2
.
3
.
. Similarly, nasal
). However, these methods are limited to cultivatable bacterial and require the
. However none of the aforementioned approaches have the ability to assess bacterial-
ACS Paragon Plus Environment
Page 3 of 36
Analytical Chemistry
Page 3 of 27 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
71
LC-MS and 16S rRNA, it has limited application for clinical point of care (POC)-diagnostics
72
due to extensive and time-consuming sample preparation and extraction protocols.
73
For routine clinical application, faster and simpler methodologies are desirable to enable real
74
time metabolic profiling that could assist in the generation of objective, effective and
75
immediate biochemical data to inform clinicians and direct personalized treatment strategies.
76
As an alternative to LC-MS analysis, ambient ionization MS techniques permit rapid and cost
77
effective metabolic profiling of unmodified and complex biological materials, including
78
bacteria, without the need for sample preparation
79
applications
80
all on the same principle, which is to generate gas phase ions directly from the samples,
81
without the need of chromatographic separation or sample preparation resulting in a
82
straightforward handling and rapid analysis of intact biofluid. Paper spray (PS) MS has been
83
already introduced as one of a main potential ambient ionization technique for POC testing,
84
which allows the quantification of analytes in dried samples spots of blood, plasma, urine
85
and oral fluid. Moreover, PS MS has been shown to readily discriminate between sixteen
86
different cultures bacteria species
87
wooden tip electrospray, coated blade spray and touch spray were which introduce the
88
concept of simultaneous sample collection and ionization directly from the surface of the
89
sampling device
90
glycerophospholipids of cultured strep throat causing bacterial species using a metallic
91
modified medical swab device
92
application of touch spray MS as a semi-quantitative MS method for direct oral fluid drug
93
testing
94
samples for POC diagnostic using touch spray MS has yet to be demonstrated. In a different
95
study, Bregy et al. used secondary electrospray ionisation (SESI) MS on saliva to identify 18
96
potential markers from a single periodontitis patient with confirmed infection where the
97
obtained metabolic profile was compared with 120 in vitro detected compounds from four
98
cultures periodontal pathogens
99
volatile and semi-volatile compounds e.g. fatty acids it is therefore more suitable for direct
21
16,17
14,15
, enabling a wide range of POC
. To date, more than 80 ambient ionization techniques are known, that work
19,20
18
. Other examples of direct sampling analysis include
. For microbiological analysis, touch spray MS was used to profile 17
. Furthermore, the same group has extended the potential
. However, to date, the detection of bacterial associated metabolites in real clinical
22
. But since SESI-MS covers preferably the detection of
100
headspace analysis including exhaled breath samples.
101
Another widely more established ionization technique for direct analysis of clinical sample is
102
desorption electrospray ionization (DESI) MS
103
technique to MALDI MS or secondary ionization mass spectrometry (SIMS), also suitable for
104
the rapid analysis of samples under ambient conditions and on arbitrary sample surfaces
105
In DESI, a pneumatically-assisted electrospray of charged aqueous droplets is directed onto
106
a sample, where it then forms a liquid film that gently dissolves molecules from the sample
23
. DESI is a comparably soft ionization
ACS Paragon Plus Environment
24
.
Analytical Chemistry
Page 4 of 36
Page 4 of 27 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
107
and desorbs secondary charged micro droplets. A subsequent ESI-like process occurs,
108
leading to the formation of gaseous ions of analytes, which are directed via the usual
109
atmospheric pressure to the vacuum ion optics interface. DESI has also proved applicable
110
for the ionization of a variety of compounds including lipids, peptides, proteins or drug
111
molecules
112
analysis but also for discriminating purified bacterial samples at species and subspecies
113
level from a glass slide or directly from the agar plate using DESI or nano-DESI MS following
114
statistical PCA analysis
115
proposed but not yet been demonstrated 29.
116
In this study, we describe an analytical protocol for direct analysis of standard medical
117
swabs, by DESI-MS (Figure 1). No modified swabs or alterations to the default clinical
118
sample collection protocols for microbial analysis are required. We illustrate the potential of
119
the method with analysis of human mucosa samples, including vaginal mucosa samples
120
acquired in a clinical setting. We aimed to extend the DESI-MS application to direct analysis
121
of standard medical swabs for rapid assessment of mucosal chemistry perturbations and the
122
development of a novel, non-invasive POC-diagnostic technique.
123
Experimental Section
124
Reagents and materials. All chemicals used were analytical reagent grade. HPLC grade
125
methanol and water for DESI-MS analysis were purchased from Sigma Aldrich (St Louis,
126
MO). Human plasma was purchased from Sigma Aldrich (St Louis, MO).Transwab® Amies
127
medical rayon swabs provided by MWE medical wire (Wiltshire, UK) were used for mucosal
128
sampling.
129
Sample collection. Written, informed consent was provided from all patients prior to sample
130
collection. 20 volunteer subjects contributed nasal fluid according to NHS REC ethical
131
protocol RESPI-SAM (15/LO/0444). Oral mucosa samples were taken from the inner cheek
132
cavity of the mouth from 15 volunteers. For the vaginal mucosa study samples were
133
obtained from 23 non-pregnant healthy volunteers (13/LO/0126) and from 43 pregnant
134
healthy women according the ethical protocol VMet2-Vaginal Microbiome and Metabonome
135
in Pregnancy (14/LO/0328) at Queen Charlotte’s and Chelsea Hospital, Imperial College
136
Healthcare NHS Trust (London, UK). Exclusion criteria included HIV positive women and
137
women who had had sexual intercourse or vaginal bleeding in the preceding 48 h prior to
138
sampling. Vaginal secretions were collected by a trained clinician by placing the cotton swab
139
in the vaginal cavity for 5 s; transferred to a sterile tube without buffer or storage medium
140
solution and stored at -80 °C in the freezer. Matched vaginal swabs were also sent for
141
clinical microbial assessment and diagnosis of BV was performed via the Hay/Ison criteria
25
. DESI MS provides high specifity, selectivity and sensitivity for direct tissue
26-28
. Direct analysis of gastrointestinal mucosa by DESI has been
ACS Paragon Plus Environment
Page 5 of 36
Analytical Chemistry
Page 5 of 27 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
142
and assessment of gram staining30. Gestational age at sampling varied from 12 to 40 weeks
143
of gestation for pregnant women.
144
To examine inter-swab variability, quality control swabs were prepared by first precipitating
145
human plasma using a 4 to 1 ratio of ice-cold acetonitrile and incubating for 1h at -80°C. The
146
precipitate was recovered following removal of the supernatant after centrifugation (1000 x g
147
for 10 seconds). The precipitate was dried for approximately 2h at room temperature until
148
reaching a gel-like consistency. The pellet was then distributed uniformly on the swab
149
surface rotating the swab on aluminium foil so that each swab contained approximately 50
150
mg of plasma precipitate (see Supplementary Information, Figure S1).
151
DESI-MS swab analysis. The in-house built DESI source mounted on the mass
152
spectrometer consists of an electronic spray emitter and an automatic rotatable swab holder
153
that are separately secured onto a 3D manual moving stage. The sample holder consists of
154
a rotor for controlled rotation of the medical swab followed by a plastic tube to stabilize the
155
swab and prevent sideway tripping during rotation. All experiments were performed on a
156
LTQ-Orbitrap Discovery mass spectrometer (Thermo Scientific, Bremen, Germany) and
157
Xevo G2-S Q Tof (Waters, Manchester, UK) mass spectrometer. The DESI source operating
158
parameters and instrumental settings are listed in Table 1. All samples were randomised
159
prior to analysis, to avoid bias in the statistical analysis due to analytical batch effects. For
160
DESI-MS analysis the medical swabs were fixed into the swab holder and positioned
161
orthogonally in front of the MS inlet capillary with a swab-capillary distance of less than 2
162
mm. The DESI sprayer was directed to the centre of the medical swab with a tip-sample
163
distance between 1.5-2 mm and an altitude difference between the tip and the inlet capillary
164
of 2 mm. The entire surface of the medical swabs was analysed by DESI through clockwise
165
rotation of the swab toward the MS capillary. For desorption of the sample material a mixed
166
methanol:water solution (95:5, v:v) spray solvent was used for DESI measurement. The
167
mucosal membrane was absorbed from the swab tip by gently desorbing the molecules with
168
charged droplets of organic solvent and subsequently transferred to the mass spectrometer
169
for analysis. Full scan mass spectra (m/z 50-1000, R = 30.000 (FWHM)) were recorded in
170
the negative and positive ion mode.
171
For microbial culture samples, a loop scraping of biomass was diluted into 50 µL of methanol
172
before a swab was rapidly dipped into the solution and DESI-MS analysis was performed on
173
the swab surface.
174
Mass spectral data pre-processing. Raw mass spectral data were converted from .raw
175
files into .mzML format via the ProteoWizard msConverterGUI (Vanderbilt University,
ACS Paragon Plus Environment
Analytical Chemistry
Page 6 of 36
Page 6 of 27 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
31
176
Nashville, TN, USA)
177
package
32
178
0.01 ∆m/z and an averaged mass spectrum obtained by averaging all the scans in a
179
measurement (in average 30 scans), thus obtaining a matrix of mass spectral profiles. For
180
peak detection, Savitzky-Golay smoothing
181
window increasing linearly with the m/z value was applied before detection of local maximum
182
by analysis of second and first derivatives. The peaks detected in each samples were
183
matched sequentially (window of 0.01 m/z). Peaks only detected in less than 10 samples
184
(50% of samples in the smallest of the mucosal types, nasal mucosa) were discarded,
185
resulting in a final peak list of 926 peaks in negative mode and 2418 peaks in positive mode.
186
Two other versions of this peak matrix were generated by applying TIC and probabilistic
187
quotient normalization (see Supplementary Information, Figure S2).
188
Multivariate statistical analysis. Data was analysed in the same manner in parallel using
189
the non-normalized, TIC and PQN
190
and effect of normalization on model quality and discriminatory variables. Principal
191
component analysis (PCA) was used for unsupervised analysis of the data set and
192
visualisation of general clustering trends. For the group discrimination analyses (nasal vs
193
oral vs vaginal mucosa and pregnant vs non-pregnant), a random forest classifier
194
used (number of trees; n = 400). Feature importance was calculated by checking the out-of-
195
bag error after permuting the data. Model fitting and feature importance evaluation were
196
performed using the ‘randomForest’ R package
197
random forest models was validated through leave one out cross validation (CV). This
198
procedure involved leaving out one sample from the sample set and calculation of a new
199
model using the remaining data set. Statistical measures including sensitivity (TPR),
200
specificity (TNR), positive predictive value and negative predictive value. No outliers were
201
removed for any of the analyses described. The distribution of the identified features was
202
visualised by violin plots, which feature a kernel density estimation of the underlying
203
distribution. P-values for each peak were calculated by using the Kruskal Wallis rank sum
204
test and were assigned to be statistical significant with p
