Analysis of pathogenic bacteria using exogenous volatile organic compound metabolites and optical sensor detection

Department of Applied Sciences, Northumbr upon Tyne, NE1 8ST, UK. E-mail: John.Dean@ 3519; Tel: +44 (0)191 227 3047 Department of Microbiology, Freeman Hosp † Electronic supplementary information ( of the optimisation processes for dete 2-nitrophenol detection, aniline detec 2-nitrophenol and aniline. In addition, ex 3-amino-N-phenylpropanamide, TFA and as analytical data for 3-amino-N-phenylpr three Tables are provided. See DOI: 10.10 Cite this: RSC Adv., 2015, 5, 15494


Introduction
The detection of pathogenic bacteria in clinical and food samples can oen be a time-consuming and laborious process. The rapid identication of bacteria is essential for effective patient treatment in clinical settings and for determining the source of contamination in food samples. Bacteria have been shown to liberate a wide range of volatile organic compounds (VOCs). 1,2 Several analytical methods which have focussed on the detection of VOCs liberated by bacteria have been developed. Detection of bacterial VOCs using headspace solid phase microextraction (SPME) coupled with gas chromatography-mass spectrometry (GC-MS) has shown potential as a diagnostic tool for bacterial detection. 3 Other analytical methods used for bacterial VOC analysis include multi-capillary column-ion mobility spectrometry 4 and secondary electrospray ionizationmass spectrometry. 5 However, clinical and food laboratories are unlikely to possess such instruments for VOC analysis; in addition, they are expensive to purchase, require trained laboratory staff to operate them and regular maintenance. Alternative VOC detection methods that do not require instrumentation would be highly desirable. Visual VOC detection methods have recently been developed. For example, a disposable colorimetric sensing array to identify 10 strains of bacteria has been developed. 6 Dyes incorporated into the sensor changed colour upon exposure to VOCs generated by bacteria. The use of enzyme substrates as a potential tool for bacterial detection, via exogenous VOC evolution, has previously been described. 7 E. coli was detected via the generation of the exogenous 2-nitrophenol using the enzyme substrate 2-nitrophenyl-b-D-galactoside; ion mobility spectrometry was used for VOC detection. Subsequently xerogels have been used to trap and colorimetrically detect VOCs, specically 4-nitrophenol released by enzymatic hydrolysis of 4nitrophenyl-b-D-glucuronide 8 and p-dimethylaminocinnamaldehyde for the detection of E. coli (a known indole producer). 9 In the case of the latter example, p-dimethylaminocinnamaldehyde reacted with indole to generate a green colour within the xerogel.
Agarose is a polysaccharide with repeating units of D-galactose and 3,6-anhydro-L-galactose and upon gelation it forms a porous semi-solid matrix. Agarose gels are commonly used for separating molecules during electrophoresis; however, here it is used as a potential matrix for VOC trapping. Agarose gel is inexpensive and its preparation is simple. In addition, it is possible to modify the agarose gel by the addition of specic and selective colorimetric reagents that will generate colour in the presence of target compounds. The aim of this paper is to develop a novel, simple optical method for bacterial identication via the detection of exogenous VOCs. Following bacterial enzyme activity, the exogenous VOCs are trapped in modied agarose gel and optically detected by either the naked eye or colorimetric analysis; the latter allows quantication of the exogenous VOC. The proof of concept is demonstrated using two enzyme substrates: the commercially available 2-nitrophenyl-b-D-glucoside which liberates 2-nitrophenol in the presence of bacteria with b-glucosidase activity and a synthesised substrate, 3-amino-N-phenylpropanamide, TFA salt which liberates aniline in the presence of bacteria with b-alanine aminopeptidase activity. The 2-nitrophenol liberated from 2-nitrophenyl-b-Dglucoside is yellow coloured in alkaline conditions whereas the aniline released from a substrate targeting b-alanine aminopeptidase activity can be optically detected by its reaction with 1,2-naphthoquinone-4-sulfonic acid, sodium salt (NQS) giving a red colour. NQS has been used for the colorimetric detection of amines in solution 10 and in the vapour phase. 11 b-Glucosidase substrates are widely used in chromogenic media; b-glucosidase activity has been demonstrated in several species, including the Gram-positive organism Enterococcus faecium and the Gram-negative species Klebsiella pneumoniae. 12 Pseudomonas aeruginosa is a known b-alanine aminopeptidase producer and enzyme substrates targeting b-alanine aminopeptidase activity have been synthesised and successfully used in chromogenic media for the detection of Ps. aeruginosa. 12,13 Pathogenic bacteria known to produce b-glucosidase and b-alanine aminopeptidase activity were selected to demonstrate this proof of concept, illustrating the use of exogenous VOC detection, via modied agarose gels, as a novel method of bacterial identication based on optical sensor detection. This novel approach for identifying or detecting bacteria, via the generation of exogenous VOCs, allows for label-free, noninvasive and continuous monitoring of a sample.

Gel preparation for the detection of 2-nitrophenol
Agarose gel (1%) was prepared by dissolving 0.5 g agarose in distilled water and microwaving until boiling (approximately 1 min). The solution was allowed to cool and then modied by addition of 0.2 mol L À1 NaOH; the nal total volume was 50 mL. The nal NaOH concentration was 2 mmol L À1 . A sub-sample aliquot of the NaOH modied-agarose gel (150 mL) was added to a cuvette within a vial and allowed to set.

Gel preparation for the detection of aniline
A 0.5% solution of NQS was prepared by dissolving 0.025 g in 5 mL of distilled water. This reagent was prepared fresh daily. Agarose gel (1%) was prepared as described in Section 2.3. To 1% agarose gel, 0.25 mL of 0.5% NQS was added to give a total volume of 50 mL. A sub-sample aliquot of the NQS modied agarose gel (150 mL) was added to a cuvette within a separate vial and allowed to set.
For each gel type, a 1 mL total volume of sample in BHI media was added aseptically to a 1.5 mL Eppendorf tube within a sterilised vial. The BHI media was made up according to manufacturer's guidelines. Sterilisation of media was achieved by autoclaving at 121 C for 11 minutes. Vials and cuvettes were sterilised by exposure to UV radiation for 90 min in a UV cabinet. The vial with sample was sealed and placed overnight in a 37 C incubator. The absorbance of the gel in the cuvette was measured at the experimentally determined wavelengths of 415 nm for 2-nitrophenol and 470 nm for the coloured complex formed by aniline/NQS, compared against a reagent blank aer 18 h incubation.

Optimisation of parameters
Aniline and 2-nitrophenol stock solutions were prepared in N-methyl-2-pyrrolidinone. The optimum percentage agarose in gels was determined using 50 mg mL À1 of 2-nitrophenol (or 359 mmol L À1 ) and 20 mg mL À1 (or 215 mmol L À1 ) of aniline in 1 mL of BHI media. The optimum volume of gel and NaOH concentration in gel were determined with 50 mg mL À1 2-nitrophenol in 1 mL BHI media following overnight incubation at 37 C. [Note: as the yellow colour is generated by the ionic form i.e. 2-nitrophenolate (pK a of 2-nitrophenol is 7.23) it was necessary to increase the pH (with NaOH) to generate more of the anion; this resulted in a higher molar extinction coefficient in the presence of NaOH and a more intense colour.] Similarly, optimum NQS and NaOH concentrations in gel were determined with 20 mg mL À1 aniline in 1 mL BHI media. Absorption spectra of gels exposed to 2-nitrophenol and aniline from 300 nm to 700 nm were produced and maximum absorption wavelengths for both VOCs were determined. Calibration curves for both VOCs were produced using a VOC concentration range in BHI media and measuring absorbance aer 18 h incubation at 37 C. Absorbance readings below 0.03 were omitted. An absorption spectrum of NQS measured against agarose gel with no reagents added was also produced. A Pharmacia Biotech Ultraspec 2000 UV/visible spectrophotometer was used for measuring the absorbance of samples and blanks.
2.6 Synthesis of 3-amino-N-phenylpropanamide, TFA salt 3-Amino-N-phenylpropanamide, TFA salt was synthesised using t boc-b-alanine-OH as a starting material. To a stirred solution of t boc-b-alanine-OH (0.50 g, 2.6 mmol) and N-hydroxysuccinimide (0.30 g, 2.6 mmol) in dry dichloromethane (DCM) (10 mL) at room temperature with a CaCl 2 trap was added dicyclohexylcarbodiimide (DCCI) (0.55 g, 2.6 mmol). The solution was stirred at room temperature for 1.5 h. To dry DCM (10 mL) was added aniline (0.24 g, 2.6 mmol) and to this was added the contents of the other ask slowly over 1-2 min. The mixture was allowed to stir overnight at room temperature aer which the precipitate was ltered and solvent evaporated on a rotary evaporator. The intermediate, t boc-3-amino-N-phenylpropanamide (0.49 g, 71% yield), was puried by column chromatography over silica gel using hexane : ethyl acetate (2 : 1, v/v) as the eluent. A portion (0.35 g) of the product was deprotected with triuoroacetic acid (TFA) (used as solvent and reagent) yielding the TFA salt of 3amino-N-phenylpropanamide (0.31 g, 84% yield). For structure verication and comparison with literature, 3-amino-N-phenylpropanamide, TFA salt was dissolved in ice-cold water (2 mL), made alkaline with 2 mmol L À1 sodium hydroxide and extracted with chloroform (5 Â 2 mL). The combined organic phases were dried with sodium sulphate and the solvent evaporated. Structure verication was achieved using 1 H-NMR spectrum (400 MHz) and 13 C-NMR spectrum (101 MHz) recorded on a Jeol ECS400 instrument. Low resolution mass spectrum (LRMS) was recorded via direct injection of a dilute methanolic solution into a Thermo Finnigan LCQ Advantage MS detector using electrospray ionisation (ESI). Melting point was determined using Stuart SMP1 melting point apparatus. Analytical data for 3-amino-N-phenylpropanamide, TFA salt and the scheme for synthesis are given in the ESI. †

Preparation of bacterial samples
All bacteria were sub-cultured overnight at 37 C on BHI agar one day prior to the preparation of samples. Aer overnight incubation, bacteria were inoculated in sterile BHI broth and incubated at 37 C. Samples were prepared by measuring the absorbance of the incubated bacterial suspension at OD 600nm . At an absorbance reading of 0.132 (equivalent to 0.5 McFarland units/1.5 Â 10 8 CFU mL À1 ) an aliquot of 10 mL of bacterial suspension (1.5 Â 10 6 CFU) was added to a 1.5 mL Eppendorf containing 0.99 mL sterile BHI broth with enzyme substrate.
Samples were placed in a sterile vial alongside a cuvette containing agarose gel with added reagents and incubated overnight ( Fig. 1(a)). Media with enzyme substrate was prepared by dissolving 10 mg of enzyme substrate in N-methyl-2-pyrrolidinone then transferring aliquots aseptically to BHI broth.

Application of optimised methods to bacterial samples
E. faecium, S. agalactiae, E. coli and K. pneumoniae were tested with 2-nitrophenyl-b-D-glucoside. Ps. aeruginosa, Ps. uorescens and E. coli were evaluated with 3-amino-N-phenylpropanamide, TFA salt. Parameters investigated with E. faecium and 2-nitrophenyl-b-D-glucoside and Ps. aeruginosa with 3-amino-N-phenylpropanamide, TFA salt were substrate concentration and sensitivity in terms of initial inoculum used. All samples were tested aer overnight incubation at 37 C, aer which the absorbance of the gel in the cuvette was measured at 415 nm with 2-nitrophenyl-b-D-glucoside and at 470 nm with 3-amino-Nphenylpropanamide, TFA salt, both against reagent blanks. Uninoculated broth with added substrate and bacteria inoculated in broth without substrate were tested to ensure no colour developed due to spontaneous breakdown of the substrate or interfering VOCs. All bacterial samples, as well as uninoculated BHI samples with substrate and uninoculated BHI samples without substrate, were tested in triplicate.

Results and discussion
Modied agarose gel has been identied as a suitable matrix for the trapping of exogenous VOCs. It was found that colour development was homogenous throughout the gel, allowing accurate absorbance measurements to be recorded. The reagents required to generate colour were easily added to agarose gels and their optimisation for increased colour generation was straightforward (see ESI †).

Application of optimised methods to bacterial samples
The structures of both enzyme substrates are shown in Fig. 2. 3-Amino-N-phenylpropanamide has previously been synthesised. 14,15 A comparison of the NMR data is given in the ESI. † Bacterial samples were prepared with a nal 2-nitrophenyl-b-Dglucoside concentration of 200 mg mL À1 (664 mmol L À1 ). There was an increase in absorbance and the development of a yellow colour aer 18 h incubation with both E. faecium and K. pneumoniae (known producers of b-glucosidase). The generation of a yellow colour indicated that 2-nitrophenol had been trapped in the modied agarose gel. Both S. agalactiae and E. coli showed no activity with 2-nitrophenyl-b-D-glucoside and no yellow colour was generated (Table 1). It was noted that all species were able to grow in the presence of enzyme substrate. 2-Nitrophenol generating bacteria could be visually differentiated from those species negative for 2-nitrophenol production ( Fig. 1(b)).
Bacterial samples were tested with a nal 3-amino-N-phenylpropanamide, TFA salt concentration of 100 mg mL À1 (or 359 mmol L À1 ). Ps. aeruginosa hydrolysed 3-amino-N-phenylpropanamide, TFA salt and this was detected by an increase in absorbance of the gel and the generation of an orange colour within the gel. No hydrolysis occurred with Ps. uorescens and E. coli and no orange colour developed with these species ( Table  2). All species grew in the presence of 3-amino-N-phenylpropanamide, TFA salt. Aniline generating bacteria, and therefore b-alanine aminopeptidase producing species, could be differentiated visually from species that did not liberate aniline ( Fig. 1(b)). No substrate hydrolysis occurred in uninoculated BHI controls with either enzyme substrate. In addition, there was no colour development in bacterial samples in BHI without the addition of enzyme substrates.
A 2-nitrophenyl-b-D-glucoside concentration range of 50-200 mg mL À1 was tested with E. faecium (Table 3); the optimum substrate concentration was determined to be 200 mg mL À1 (or 664 mmol L À1 ). An increase in absorbance readings and a yellow colour were detected from a 2-nitrophenyl-b-D-glucoside concentration of 100 mg mL À1 (or 332 mmol L À1 ); 2-nitrophenol generation continued to increase as the substrate concentration increased up to 200 mg mL À1 . A 3-amino-N-phenylpropanamide, TFA salt concentration range of 10-200 mg mL À1 was tested with Ps. aeruginosa (Table 3). Aniline was detected from a 3-amino-Nphenylpropanamide, TFA salt concentration of 20 mg mL À1 (or 72 mmol L À1 ); the optimum 3-amino-N-phenylpropanamide, TFA salt concentration was 100 mg mL À1 (or 359 mmol L À1 ). At a 3-amino-N-phenylpropanamide, TFA salt concentration of 200 mg mL À1 (or 719 mmol L À1 ), aniline generation decreased from that detected at a substrate concentration of 100 mg mL À1 . An initial inoculum of 1-1.5 Â 10 4 CFU mL À1 was required for detectable levels of colour to develop with both enzyme substrates aer 18 h incubation ( Table 4).
The developed method allowed for the discrimination of b-glucosidase producing bacteria, as well as b-alanine aminopeptidase producing species. Bacterial VOC analysis has previously suffered from a lack of specicity; in addition, experimental variables such as choice of growth medium have    been shown to alter the VOCs generated. 3 Use of enzyme substrates increases the specicity of bacterial VOC proles as VOCs released from substrates would act as markers for a particular species. 3-Amino-N-phenylpropanamide, TFA salt demonstrated potential as a suitable substrate for the detection of Ps. aeruginosa. The development of an orange colour indicated aniline liberation and its subsequent coupling with NQS, and thereby signifying b-alanine aminopeptidase activity. A 3-amino-Nphenylpropanamide, TFA salt concentration of between 20-100 mg mL À1 was sufficient for colour generation; this was comparable to b-alanine aminopeptidase enzyme substrate concentration used in chromogenic media. 13 However, the absorbance of gels decreased at 3-amino-N-phenylpropanamide, TFA salt concentrations above 100 mg mL À1 , possibly due to the increased amount of solvent at higher substrate concentrations as 3-amino-N-phenylpropanamide, TFA salt was less soluble in N-methyl-2-pyrrolidinone than 2-nitrophenyl-b-D-glucoside.
Limitations of the current method include use of a small sample size, testing with a limited range of species and only one strain of each species used. Applying the method to more species is required. As a result of the size of vials used (30 mL), the headspace volume in the vial was large. Use of a smaller vial would decrease the headspace volume, could lead to an improvement in the linear range for quantication purposes, as well as improving method sensitivity and hence increasing the amount of colour developed. An initial inoculum of 1-1.5 Â 10 4 CFU mL À1 was required for colour generation aer 18 h incubation with both enzyme substrates. However, using a xerogel, p-nitrophenol was detected aer 16 h incubation with an initial E. coli inoculum of 10 5 CFU mL À1 . 8 The initial inoculum required to produce a detectable amount of colour could be reduced by decreasing the headspace volume. Following the implementation of a new experimental set-up with a smaller sample headspace volume, a study investigating the minimum time required for sufficient exogenous VOC liberation and detection would be useful. Pure bacterial cultures inoculated in a liquid culture medium were used to demonstrate the proof of concept; however, this method could potentially be applied to the identication of pathogenic bacteria in "real" clinical and food samples.
Producers of b-glucosidase and/or b-alanyl aminopeptidase can be readily detected by inclusion of chromogenic or uorogenic enzyme substrates in the medium that contains the bacterial cells without the need to detect released VOCs in the headspace, and such methods are well established. However, our approach has several possible advantages. For example, inoculation of foodstuffs or other test samples into enrichment broths may lead to opacity or discolouration that could confound the interpretation of whether a chromogenic substrate has been hydrolysed (or lead to the quenching of released uorogens, if uorogenic substrates are used). Furthermore, development of this technology could lead to the design of a sensor that is divided into compartments, each of which contains specic reagents for trapping and visualizing distinct VOCs. This could potentially allow generation of a biochemical prole showing the presence (or absence) of multiple bacterial enzymes within an enrichment broth allowing the detection of bacterial pathogens with high specicity.
In addition, it is also possible to compare the theoretical trapping efficiency of 2-nitrophenol in agarose. This would allow this approach to be directly compared with other trapping media (e.g. xerogels). 8 Unfortunately some uncertainty exists in the scientic literature on the values for the molar extinction coefficient and Henry's constant for 2-nitrophenol. For example, the molar extinction coefficient (molar absorptivity) for 2-nitrophenol has been reported to have values that range from 2150-21 300 mol À1 L cm À1 . 16 In addition, the Henry's constant for 2-nitrophenol also have some variability in their numerical values (e.g. 4.13 Â 10 À5 and 1.28 Â 10 À5 atm m 3 mol À1 ). 8,17 Similar variability is also evident for aniline. On that basis it is not possible to determine an accurate trapping efficiency for either VOC.

Conclusions
This work highlights the potential of designing enzyme substrates to liberate exogenous VOCs for bacterial identication, as well as the suitability of agarose gel as a matrix for VOC trapping. In addition, the ability to tailor reagents present in the gel for the colorimetric detection of specic exogenous VOCs is demonstrated and reagents added to gel can be optimised for maximum colour generation. This innovative method displays potential for the development as a novel, simple and low-cost optical VOC sensor for the detection of bacteria in clinical and food samples. Table 4 Sensitivity of modified agarose gels method in terms of initial inoculum used Initial inoculum (CFU mL À1 BHI) 2-Nitrophenol concentration b (mg mL À1 ) Aniline concentration c (mg mL À1 ) 1-1.5 Â 10 6 31.5 8.25 1-1.5 Â 10 5 27.8 6.48 1-1.5 Â 10 4 16.4 5.03 1-1.5 Â 10 3 ND a ND 1-1.5 Â 10 2 ND ND 1-1.5 Â 10 1 ND ND a ND, not detected. b E. faecium with 200 mg mL À1 2-nitrophenyl-b-Dglucoside, absorbance measured at 415 nm. c Ps. aeruginosa with 3-amino-N-phenylpropanamide, TFA salt, absorbance measured at 470 nm.