Medical Swab Analysis Using Desorption ... - ACS Publications

Dec 27, 2016 - Computational and Systems Medicine, Imperial College London, South ... minimal-disruption technique for augmenting the information...
3 downloads 0 Views 3MB Size
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