Rapid endophytic bacterial detection by enzyme incorporated MALDI MS

Abstract

In this paper, we have attempted the use of MALDI-MS for the detection of bacteria from complex real world samples such as the root nodules of plants. We have employed lysozyme as the enzyme biosensor to enrich the bacterial molecular signatures from endophytic bacteria during the MALDI-MS analysis. We used the root nodules of Arachis hypogaea, L. as an endophytic model system to prove the enrichment of the endophytic bacterial signals for the rapid and direct identification of the endophytic bacteria. We have been able to demonstrate the functionality of lysozyme as an enzyme biosensor during the mass spectrometric analysis of complex plant tissues such as the root nodules, leading to a culture-free detection of the endophytic bacteria. We also report the direct identification of the specific dominant endophytic isolate using MALDI-MS combined with Biotyper software analysis and validation.

---

Introduction

Organisms from the microbial kingdom are always known to be the critical players for the existence of life on earth. They participate in numerous processes that drive life in the biosphere. Furthermore, they are powerful enough to even influence the global climatic change.¹ These tiny inhabitants exist widespread and they impart both beneficial and harmful effects to plants and animals. Plants harbor a vast variety of bacterial species over the rhizosphere (rhizobacteria), phyllosphere (epiphytes) and inside of the plant tissues (endophytes). Endophytic bacteria are reported to reside within the internal tissues of plants without causing damage to their hosts.² Endophytic bacterial communities are advantageous to plants³ because endophytic bacteria can contribute to plant development by producing phytohormones⁴ and siderophores,⁵ increasing resistance to pathogens⁶ and parasites,² promoting biological nitrogen fixation⁷ and antibiotic production.⁸ Thus, each individual plant is most probably a host to one or more endophytic bacteria.⁹ However, identification of the endophytic bacteria is a significant challenging task which involves laborious and time consuming processes.

The general practice of bacterial identification relies on the basic culturing of the bacteria and their responses to various physiological, serological and biochemical tests.¹⁰ However, these methods have limitations which make their identification rather complicated or misleading. Analysis of the signature molecules (such as fatty acids) of each bacterial group appears to have influence on bacterial identification systems.¹¹ However, this method also involves culture based approaches and complicated sample preparation protocols. Phylogenetic methods based on nucleotide sequence similarity and hybridization came into existence to circumvent many problems in culture based approaches. Highly conserved 16S rRNA/DNA sequences of bacterial isolates are given supreme importance in phylogenetic identifications. Moreover, the sequences of RNA polymerase (rpoB),¹² or elongation factors¹³ have also been used as other targets in bacterial identification. PCR has also been advantageous for the identification of slow growing bacteria and uncultivable bacteria. Although this technique is considered to be one of the most reliable methods for bacterial identification, discriminating the species level is not always possible with one target sequence. Moreover, this technique involves use of expensive chemicals and technical expertise. Therefore, alternative identification methods are always sought for surpassing the routine tedious protocols.

MALDI TOF-MS has turned out to be an alternative tool for bacterial identification by utilizing the proteome obtained from whole cells. The differences in the surface proteome may lead to the identification and differentiation of microorganisms, based on the peaks appearing between the mass range of 2000 and 20 000 Da.¹⁴ MALDI MS has been employed to identify bacteria,¹⁵ archaea,¹⁶ micro algae and fungal species.¹⁷ Although there is increasing interest for using MALDI-MS for the characterization and identification of microorganisms, a pure culture of the bacterial species is always essential for its identification to the genus/species level. In real world samples, however, such identification may not be practical. To overcome this difficulty of bacterial identification from the real samples, the technique of affinity mass spectrometry was developed to selectively concentrate trace amounts of bacteria from biological fluids.^18,19 This technique is more reliable if the culture is in the liquid phase. However, direct endophytic bacterial identification by MALDI MS analysis is extremely difficult because these bacteria exist inside the plant tissues. Therefore, MALDI MS that selectively targets bacterial protein/molecules is required for endophytic bacterial identification amidst plant proteomes. This type of bacterial protein enrichment could be obtained either through selectively breaking the bacterial cell wall or altering the structural integrity of the plant cell wall to release the endophytic bacteria during MALDI MS. This technique would be helpful in the identification of dominant endophytic bacteria directly from plant tissues by MALDI MS analysis because the MALDI MS signal is proportional to the bacterial numbers.^17,20

Therefore, in this paper, we have employed lysozyme as the enzyme biosensor to enrich the bacterial signals of endophytic bacteria during MALDI-MS analysis. We used the root nodules of Arachis hypogaea, L. as an endophytic model system to prove the enrichment of the endophytic bacterial signals for the rapid and direct identification of endophytic bacteria. We have demonstrated the functionality of lysozyme as an enzyme biosensor during the mass spectrometric analysis from complex plant tissues such as root nodules.

Materials and method

Chemicals

Trifluoroacetic acid (TFA) was purchased from Sigma Chemical Co. (St. Louis, MO, USA). 3,5-Dimethoxy-4-hydroxycinnamic acid (SA) was purchased from Alpha Aesar (UK). Acetonitrile (MeCN) and acetone were purchased from J.T. Baker, Phillipsburg, NJ, USA. Hen egg white Lysozyme was purchased from Sigma-Aldrich (St. Louis, MO, USA). Deionized water purified by a Milli-Q reagent system (Millipore, Milford, MA, USA) was used for all the experiments.

Collection of endophytic plant material

Root nodules are known for the presence of endophytic bacteria inside the plant tissue. Therefore, in our present study, a nodule bearing plant species, Arachis hypogaea, L. was used (Fig. S1†). Twenty day old Arachis plants were carefully uprooted from an agricultural field from Pingtung county and brought to the laboratory in an ice box. The whole plants were washed under running water using a paint brush until the sand particles were removed. Stems and roots were cut into 2 cm long sections. The tissues were thoroughly rinsed using a series of sterile water followed by sterilization using 70% ethanol (30 seconds). The plant tissues were then surface sterilized using 0.1% HgCl₂ for 3 minutes and thoroughly rinsed twice with sterile distilled water²¹ and dried using a sterile filter paper.

MALDI MS based direct endophytic bacterial analysis

The overall scheme for the identification of endophytic bacteria directly from the plant tissues by biosensor application is given in Fig. S2.† 500 μg of root and root nodules were crushed in 500 μL Tris HCl buffer (100 mM; pH 7.2) using a pre-sterilized mortar and pestle (1 inch radius) and the samples were treated with different concentrations of lysozyme, namely, 5, 50, 100, 200 and 500 μg. The samples were vortexed for 5 minutes to obtain a suspension with uniform distribution, and incubated at 37 °C for 30 minutes. The samples were then again vortexed for 10 minutes and centrifuged at 10 000 × g for 5 minutes. 4 μL of the supernatant was spotted onto the stainless steel MALDI target plate (Bruker Daltonics Inc., Germany). For validation of the spectral peak of bacterial origin in the lysozyme treated samples, the bacterial species from the crushed samples were grown on a mannitol agar medium (Fig. S3†). Then, a colony of the predominant bacterial species was picked, washed and resuspended in 100 μL sterile distilled water. 4 μL of the bacterial suspension was spotted onto the stainless steel MALDI MS target plate (Bruker Daltatonics Inc., Germany). Each spot was overlaid and mixed with 4 μL of matrix SA (0.05 M Sinapinic Acid (SA) in 3 : 1 acetonitrile : water, containing 0.1% TFA for the analysis of bacterial proteins (>2000 Da)). All the experiments were performed in triplicate.

MALDI-MS analysis

The positive ion mode was used to obtain MALDI-MS spectra using a MALDI mass spectrometer (Microflex, Bruker Daltonics, Bremen, Germany). The MALDI-MS source was equipped with a nitrogen laser (337 nm), a 1.25 m flight tube and the sample target, having the capacity to simultaneously load 96 samples. All mass spectra were acquired with following parameters set on the MALDI-MS: IS1, 19.0 kV; IS2, 16.15 kV; lens, 9.35 kV and reflector at 20.0 kV. The laser energy was adjusted to slightly above the threshold in order to obtain a good resolution and a signal-to-noise ratio of 60 Hz. 200 laser shots were maintained for obtaining each spectrum. Mass spectra from 2000 Da to 12 000 Da were acquired in the positive/linear mode with a laser energy of 63.2 μJ using SA as the matrix.

Spectral analysis

Statistical analyses of the MALDI MS spectra of the lysozyme treated and untreated plant tissues were carried out using ClinPro Tools 2.0 (Bruker Daltonics). The spectra of different classes were loaded into the software for offline analysis. Baseline correction was achieved using a top hat algorithm with a 10% minimal baseline width. Normalization was done for all the spectra to their own TIC (total ion count) by default programming. Thus, for each spectrum the TIC was determined. The TIC was calculated based on the total sum of intensities from all data points in the spectrum. Subsequently, all data point intensities of this spectrum were divided by the obtained TIC value, bringing all intensities into the range of [0,1]. To identify the differentially displayed peaks, statistical comparison of the peak intensities of each class of spectrum was obtained through ANOVA (analysis of variance). The peaks showing 0.05 P values according to the Kruskal–Wallis test and Anderson–Darling test were considered to statistically present the significant differences. The reproducibility of the mass spectral peaks was demonstrated by the stack gel view.

MALDI Biotyper 1.0 software (Bruker Daltonics) with a mass spectral library consisting of about 967 strains of bacteria was used for the identification of unknown bacterial MS spectrum. The identification was done by using the default settings. The identification was carried out in the automatic mode without any user intervention. A total number of 100 peaks were picked up from each bacterial spectrum by the software. The peak lists generated were used for matches against the reference library, by directly using the integrated pattern matching algorithm of the software.

Results and discussion

Fig. S1† shows the morphology of the plant and the root nodules, which are the sources from where the bacterial endophytic association was obtained. The morphology of the Arachis (Fig. S1b†) plant above the ground level is shown in the image. After uprooting the plants from the agricultural fields, we can clearly see the distinct structures known as root nodules (Fig. S1b†) occurring throughout the tap root system of the Arachis plants. These root nodules are the sites where the endophytic bacteria are actually housed. The present study involved the direct identification of these endophytic bacteria within the root nodules.

In the present study, lysozyme was used as the enzyme biosensor for selectively enriching the bacterial peaks during MALDI-MS analysis. In addition, the bacterial peaks enriched by the lysozyme treatments were compared with the MALDI MS spectrum obtained from the pure dominant colony of bacterium, isolated from the root nodule (Fig. S3†). Fig. 1 shows the MALDI MS spectrum of the crushed root nodule (Fig. 1a), the root nodules treated with different concentrations of lysozyme (Fig. 1b–f) and the bacteria culture isolated from the root nodule of the Arachis plant (Fig. 1g). The crushed root nodules of Arachis which had not been treated with lysozyme showed less number of MALDI MS peaks. Further, in these samples, we could obtain MALDI MS peaks up to m/z 7494.070. The lowest concentration of lysozyme treatment (5 μg) exhibited few bacterial peaks (indicated by star) such as 9358.993, 10 941.509 and 13 279.655 along with the plant peaks (indicated by a spherical symbol) (Fig. 1b). On increasing the concentration of lysozyme (50, 100, 200 and 500 μg) for enriching the bacterial peaks from the crushed samples, an increase in the bacterial peak numbers and their intensity was observed (Fig. 1c–f). Approximately 9 and 13 bacterial mass spectral peaks were observed, when the crushed nodule was treated with a higher concentration of lysozyme such as 200 and 500 μg, respectively (Fig. 1e, f and S4†). To date, MALDI MS has been successfully employed for the bacterial identification of axenic cultures isolated by the traditional culture-based approach.¹⁵ In few other reports, nanoparticle (NP) based sensors have been used for the enrichment of bacterial signals in MALDI MS applications.^22,17 The NP based sensors work because of their affinity towards intact bacterial membranes and aid in better ionization of the bacterial proteins on laser irradiation during MALDI MS.^23,24 However, these NPs will be ineffective for bacterial detection in a complicated system such as plant roots/nodules, where the bacteria lie encased within the plant tissues.²⁴ Moreover, the penetration of NP based sensors into the intra- and intercellular spaces of plant cells, where the bacteria are housed, is highly limited. Whereas, lysozyme is a unique functional biosensor, which selectively cleaves the glycoside bonds between the N-acetyl glucosamine and N-acetyl muramic acid of bacterial cell walls,^25,26 thus presumably enriches the endophytic bacterial signals. Lysozyme can easily penetrate into the intercellular spaces and can selectively act on the endophytic bacteria and release the bacterial surface signature molecules, which can be mobilized out of the plant tissues by simple pretreatments such as vortexing and centrifugation and then detected using MALDI MS.

Fig. 1 MALDI MS analysis of endophytic bacteria from the root nodule of the plant Arachis hypogaea. Spectra obtained from (a) crushed root nodule; (b) crushed root nodule with 5 μg of lysozyme; (c) crushed root nodule with 50 μg of lysozyme; (d) crushed root nodule with 100 μg of lysozyme; (e) crushed root nodule with 200 μg of lysozyme; (f) crushed root nodule with 500 μg of lysozyme; (g) bacterial isolate from the root nodule (symbol ● denotes peaks of plant origin; ★ denotes peaks of bacterial origin).

Fig. 2 shows the statistical reproducibility of the mass spectral peaks of the nodule (without lysozyme treatment), which we consider as the untreated control (Fig. 2a) versus the nodule with lysozyme treatment (at highest concentration) (Fig. 2b). The protein peaks observed in the spectra with m/z differences ±10 were considered reproducible. Thus, in all the plant spectra without lysozyme treatment, we observed that the higher intensity peaks appeared below m/z of 8000 (Fig. 1a and 2a). The dominant bacterial isolates obtained from the Arachis nodule were used for the comparison and identification of bacterial peaks from the biosensor treated nodule samples (Fig. 1g and 2c). MALDI MS analysis of bacteria lead to the wide spread appearance of m/z peaks up to 13 280, these peaks were very useful for identifying the bacterial peaks directly from the real sample, which in this case is the root nodule. Lysozyme treated root nodules exhibited peaks from plant origin as well as enriched bacterial mass spectral peaks (Fig. 2b). Thus, the clear distinction of bacterial peaks enrichment was observed on treatment with the lysozyme biosensor, which aided the identification of the predominant bacterium.

Fig. 2 Mass spectral reproducibility of (a) crushed root nodule; (b) crushed root nodule treated with 500 μg of lysozyme; (c) bacterial isolate isolated from the root nodule of Arachis hypogaea plant represented by the gel view of the Clin pro software.

Table 1 accounts the differentially displayed peaks for three groups of samples with their statistical significance. A total number of 39 peaks were picked up, based on the default programming, from the three groups of samples. Out of the peaks picked, 21 peaks were detected with a differential expression in all three groups of samples by their significant level. Eight peaks from plants were also found in the MALDI MS spectra of an axenic bacterial culture isolated from root nodules (m/z 5733.54; 6361.00; 3995.56; 3265.07; 4071.03; 5632.22; 5946.51and 3980.33) with extremely low intensities (negligible level at bacteria), thus signifying their plant origin. Twelve bacterial peaks (m/z 9800.83; 8266.79; 8932.59; 7371.33; 9359.11; 12 668.05; 3822.48; 6800.28; 13 279.68; 7854.45; 10 284.81 and 7859.45) were expressed in very low levels in the root nodule without the lysozyme treatment. The lysozyme treatment significantly (PTTA < 0.000001) increased the MALDI MS peaks in the root nodules, which again confirms the sensing effect of lysozyme. We also extend our strategy to the roots, which might contain very few bacterial entities compared with that of nodule. Fig. 3 shows the MALDI MS spectra for the control (Fig. 3a), biosensor (lysozyme) treated root tissues (Fig. 3b–f) and axenic culture of root nodule bacteria (Fig. 3g). The control and biosensor treatment did not result in the enrichment of the bacterial peaks, which could be attributed to the extremely low quantities of endophytic bacteria.

Table 1 Statistical analysis of MALDI MS peaks from plant tissue treated with biosensors and authentic bacterial strains obtained from Arachis root nodule^a

Index	Mass	D ave^b	PTTA^c	PWKW^d	PAD^e	Average nodule bacteria	Average biosensor treatment	Average nodule	Std dev. 1	Std dev. 2	Std dev. 3
a Std dev. – Standard deviation of the peak area/intensity average.b Difference between the maximal and the minimal average peak area/intensity of all classes.c PTTA is the P-value of t-test ANOVA (<0.01 significant).d PWKW is the P-value of Wilcoxon test or Kruskal–Wallis test (<0.01 significant).e PAD is the P-value of Anderson–Darling test (1-normal distribution).
1	5733.54	74.07	<0.000001	0.000499	0.182	6.17	50.71	80.24	4.03	18.63	5.76
2	9800.83	174.53	<0.000001	0.000702	0.151	213.86	137.99	39.33	19.43	57.01	8.36
3	8266.79	334.19	<0.000001	0.00106	0.185	422.42	291.98	88.24	43.06	116.67	21.41
4	8932.59	149.12	<0.000001	0.000655	0.206	188.03	119.29	38.91	19.48	42.45	9.92
5	7371.33	163.11	<0.000001	0.000935	0.182	259.89	190.79	96.78	10.31	46.41	13.93
6	9359.11	822.72	<0.000001	0.000499	0.327	900.98	422.58	78.26	97.54	145.62	16.28
7	6361	504.19	<0.000001	0.000499	0.143	13.74	279.04	517.93	1.18	134.39	37.74
8	12 668.05	242.8	<0.000001	0.00106	0.0908	297.21	204.16	54.41	30.41	68.35	7.04
9	3996.81	44.07	<0.000001	0.000499	0.132	1.19	20.85	45.27	0.62	10.77	3.8
10	5632.22	47.04	<0.000001	0.00252	0.0908	10.87	45.21	57.91	6.74	18.18	4.25
11	5946.51	584.7	<0.000001	0.000499	0.183	20.51	331.53	605.21	5.79	168.23	55.39
12	3980.33	163.49	<0.000001	0.000499	0.177	5.59	80.85	169.08	2.56	41.38	16.5
13	3995.56	132.28	<0.000001	0.000499	0.177	4.08	65.61	136.36	1.11	32.32	14.57
14	3822.48	48.59	<0.000001	0.000499	0.327	9.85	33.43	58.44	4.54	12.06	7.1
15	6800.28	98.98	<0.000001	0.000499	0.177	26.6	80.91	125.58	3.08	29.58	13.02
16	13 279.68	748.25	<0.000001	0.000499	0.191	787.42	373.97	39.17	130.82	188.4	8.13
17	7854.45	366.53	<0.000001	0.00127	0.0286	395.25	152.34	28.72	61.67	131.21	6.75
18	10 284.81	347.32	<0.000001	0.000935	0.191	437.9	276.71	90.57	65.15	118.38	15.3
19	7859.45	208.22	<0.000001	0.000655	0.143	237.74	110.67	29.52	38.3	59.22	8.96
20	3265.07	65.08	<0.000001	0.000655	0.183	4.53	38.4	69.61	2.2	21.3	8.75
21	4071.03	76.13	<0.000001	0.000499	0.268	21.06	58.58	97.19	3.94	19.47	11.75
22	5943.72	360.96	2.16 × 10^−6	0.000499	0.154	22.58	207.31	383.54	8.31	98.55	57.13
23	7825.35	240.68	4.07 × 10^−6	0.00106	0.177	265.86	157	25.18	54.41	84.43	4.18
24	10 941.99	670.67	2.29 × 10^−5	0.00106	0.387	846.66	567.05	176	218.43	220.38	25.14
25	6604.22	113.63	0.000035	0.000499	0.177	145.61	79.62	31.98	37.82	29.23	2.01
26	6398.54	147.86	4.96 × 10^−5	0.000499	0.183	7.09	76.95	154.95	3.43	37	42.92
27	5421.91	74.7	4.96 × 10^−5	0.0023	0.327	100.36	78.45	25.66	34.87	25.59	5.83
28	7317.14	46.5	5.55 × 10^−5	0.00123	0.678	55.62	89.74	102.12	8.71	12.76	11.99
29	8344.63	113.5	5.88 × 10^−5	0.000499	0.143	169.18	92.87	55.68	43.53	22.29	5.73
30	5260.89	50.59	6.99 × 10^−5	0.00127	0.886	30.41	60.97	81	13.2	16.64	9.09
31	3702.16	49.24	0.000171	0.00449	0.237	64.56	44.37	15.33	20.33	18.41	6.69
32	5148.63	34.37	0.000358	0.00559	0.597	61.69	39.93	27.33	12.11	14.55	6.16
33	3082.46	20.32	0.00241	0.0179	0.0908	38.13	37.8	17.81	12.83	19.73	2.78
34	7491.32	140.83	0.00267	0.00988	0.327	333.73	320.78	192.9	68.46	73.88	47.86
35	4875.59	42.34	0.00314	0.115	0.0313	88.24	79	45.9	33.72	29.14	4.72
36	3064.82	9.76	0.0144	0.0122	0.00308	9.41	19.17	13.32	2.72	10.9	2.22
37	4981.97	16.04	0.335	0.18	0.0908	65.04	79.62	81.08	21.36	17.75	10.42
38	5001.64	13.07	0.541	0.525	0.478	106.87	93.8	97.01	20.55	27.18	11.76
39	2750.3	0.49	0.993	0.874	0.886	25.06	25.54	25.14	11.12	6.09	9.07

---

Fig. 3 MALDI MS analysis of endophytic bacteria from the root of Arachis hypogaea plant. Spectra obtained from (a) crushed root; (b) crushed root with 50 μg of lysozyme; (c) crushed root with 100 μg of lysozyme; (d) crushed root with 200 μg of lysozyme; (e) crushed root with 500 μg of lysozyme; (f) bacterial isolate from the root nodule.

From the extended library of the Biotyper1 software of Bruker Daltonics, the spectra obtained from the predominant colony of the root nodule was compared with the spectra obtained from the axenic cultures of the different strains of nitrogen fixing bacteria. The bacteria showed the highest mass homology for Bradyrhizobium sp. USDA 3187 (Fig. 4), followed by a local Rhizobium sp and another Rhizobium sp USDA 3138 (Table S1†). Moreover, to identify the sensor assisted peak enrichment for sensing the predominant endophytic bacteria, the MALDI MS spectra obtained from the root nodule treated with lysozyme (500 μg) were subjected to the Biotyper 1 identification tool. The spectra exhibited the highest similarity again with Bradyrhizobium sp. USDA 3187 (Fig. 5), however the score value were comparatively lower than the MALDI MS spectra of the axenic culture (Table S2†), thus the identity of the lysozyme treated nodules more affirmatively enriched the MALDI MS peaks of bacterial origin (Fig. 1 and Table 1), thus easing the identification of the predominant bacterium in the real sample, like the root nodule. The Biotyper software is already proven to be very effective for diversity and ecological studies and is applicable for the analysis of large populations of isolates, allowing the effective differentiation of strains, species and genera.^27,28 The use of this software combined with the lysozyme incorporated approach for the selective release of bacterial proteins has led to the successful detection and identification of the endophytic bacteria in the root nodules of the peanut plant.

Fig. 4 Identification of root nodule bacteria isolated from Arachis hypogaea plant from the MALDI MS library constructed using the Biotyper 1 software. Spectrum above the X-axis signify the unknown endophytic bacteria. Spectrum below the X-axis denotes the most similar spectrum of the library (Bradyrhizobium sp). Mass spectral peak of the test spectra having similar mass spectral peak from the spectral library is highlighted in Green (mismatched peaks shown in red).

Fig. 5 Identification of nodule bacteria directly from Arachis hypogaea nodules after lysozyme treatment using the MALDI MS library constructed using Biotyper 1 software. Spectrum above the X-axis signify the unknown endophytic bacteria. Spectrum below the X-axis denotes the most similar spectrum of the library (Bradyrhizobium sp). Mass spectral peak of the test spectra with similar mass spectral peak from the spectral library is highlighted in Green (mismatched peaks shown in red), unidentified peaks are shown in yellow.

Conclusions

We have reported the application of a lysozyme biosensor for the MALDI MS peak enrichment of endophytic bacteria directly from the root nodules of the Arachis hypogaea plant. This enzyme biosensor selectively acts on the bacterial cell wall and alters the bacterial cell wall structure without agitating plant cells. Thus, the MALDI MS signal of the endophytic bacterial proteins had been enriched. The enriched bacterial signals were validated using the dominant bacterial isolates obtained from the root nodule. The root nodule bacteria had been identified as the close relative of Bradyrhizobium sp USDA 3187. The enrichment of the bacterial MALDI MS signals from the real samples (root nodules) treated with lysozyme biosensors plays a significant role for the rapid analysis of endophytic bacteria from plant tissues without complex cultivation and additional confirmation steps.

Acknowledgements

We thank the Ministry of Science and Technology of Taiwan for the financial support.

References

1. E. DeLong, Report from American academy of microbiology, ed. Ann Reid, 2011, February, p. 21.
2. J. Hallmann, W. F. Mahaffee, J. W. Kloepper and A. Quadthallmann, Can. J. Microbiol., 1997, 43, 895.
3. A. Mengoni, S. Mocali, G. Surico and S. Tegli, Microbiol. Res., 2003, 158, 363.
4. Y. Feng, D. Shen and W. Song, J. Appl. Microbiol., 2006, 100, 938.
5. G. I. Burd, D. G. Dixon and B. R. Glick, Appl. Environ. Microbiol., 1998, 64, 3663.
6. B. Reiter, U. Pfeifer, H. Schwab and A. Sessitsch, Appl. Environ. Microbiol., 2002, 68, 2261.
7. J. I. Baldani, V. L. D. Baldani, L. Seldin and J. Döbereiner, Int. J. Syst. Bacteriol., 1986, 36, 86.
8. G. Strobel and B. Daisy, Microbiol. Mol. Biol. Rev., 2003, 67, 491.
9. G. Strobel, B. Daisy, U. Castillo and J. Harper, J. Nat. Prod., 2004, 67, 257.
10. V. Sintchenko, J. R. Iredell and G. L. Gilbert, Nat. Rev. Microbiol., 2007, 5, 464.
11. Automated Microbial Identification and Quantitation: Technologies for the 2000s, ed. W. P. Olson, Interpharm Press, Buffalo Grove, IL, USA, 1996.
12. M. Drancourt and D. Raoult, J. Clin. Microbiol., 2002, 40(4), 1333.
13. B. Schneider, K. S. Gibb and E. Seemuller, Microbiology, 1997, 143(10), 3381.
14. M. Welker, Proteomics, 2011, 11, 3143.
15. E. Carbonnelle, C. Mesquita, E. Bille, N. Day, B. Dauphin and J.-L. Beretti, Clin. Biochem., 2011, 44, 104.
16. B. Dridi, D. Raoult and M. Drancourt, APMIS, 2012, 120, 85.
17. M. Manikandan, H. F. Wu and N. Hasan, Biosens. Bioelectron., 2012, 35(1), 493.
18. J. Bundy and C. Fenselau, Anal. Chem., 2001, 73, 751.
19. Y.-S. Lin, P.-J. Tsai, M.-F. Weng and Y.-C. Chen, Anal. Chem., 2005, 77(6), 1753.
20. S. L. Gantt, N. B. Valentine, A. J. Saenz, M. T. Kingsley and K. L. Wahl, J. Am. Soc. Mass Spectrom., 1999, 10, 1131.
21. S. Gagne, C. Richard, H. Roussean and H. Antoun, Can. J. Microbiol., 1987, 33, 996.
22. J. Gopal, H. F. Wu and Y. H. Lee, Anal. Chem., 2010, 82, 9617.
23. K. Shrivas, K. Agrawal and H. F. Wu, Analyst, 2011, 136, 2852.
24. F. Ahmad, M. Siddiqui, O. Babalolaa and H. F. Wu, Biosens. Bioelectron., 2012, 35(1), 235.
25. J. A. Rupley, Biochim. Biophys. Acta, 1964, 83, 245.
26. E. Holler, J. A. Rupley and G. P. Hess, Biochemistry, 1975, 14, 2377.
27. L. Ferreira, F. Sanchez-Juanes, P. Garcıa-Fraile, R. Rivas and P. F. Mateos, PLoS One, 2001, 6(5), e20233.
28. D. Hahn, B. Mirza, C. Benagli, G. Vogel and M. Tonolla, Appl. Microbiol., 2011, 34, 63.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c4ra05604c

---