Anal. Chem. 2010, 82, 1109–1116
Rapid, Absolute, and Simultaneous Quantification of Specific Pathogenic Strain and Total Bacterial Cells Using an Ultrasensitive Dual-Color Flow Cytometer Lingling Yang, Lina Wu, Shaobin Zhu, Yao Long, Wei Hang, and Xiaomei Yan* Department of Chemical Biology, College of Chemistry and Chemical Engineering, The Key Laboratory for Chemical Biology of Fujian Province, The Key Laboratory of Analytical Science, Xiamen University, Xiamen 361005, China This paper describes a rapid and sensitive strategy for the absolute and simultaneous quantification of specific pathogenic strain and total bacterial cells in a mixture. A laboratory-built compact, high-sensitivity, dual channel flow cytometer (HSDCFCM) was modified to enable dual fluorescence detection. A bacterial cell mixture comprising heat-killed pathogenic Escherichia coli E. coli O157: H7 and harmless E. coli DH5r was used as a model system. Pathogenic E. coli O157:H7 cells were selectively labeled by red fluorescent probe via antibody-antigen interaction, and all bacterial cells were stained with membrane-permeable nucleic acid dye that fluoresces green. When each individual bacterium passes through the interrogating laser beam, E. coli O157:H7 emits both red and green fluorescence, while E. coli DH5r exhibits only green fluorescence. Because the fluorescence burst generated from each individual bacterial cell was easily distinguished from the background, accurate enumeration and consequently absolute quantification were achieved for both pathogenic and total bacterial cells. By using this strategy, accurate counting of bacteria at a density above 1.0 × 105 cells/mL can be accomplished with 1 min of data acquisition time after fluorescent staining. Excellent correlation between the concentrations measured by the HSDCFCM and the conventional platecounting method were obtained for pure-cultured E. coli O157:H7 (R2 ) 0.9993) and E. coli DH5r (R2 ) 0.9998). Bacterial cell mixtures with varying proportions of E. coli O157:H7 and E. coli DH5r were measured with good ratio correspondence. We applied the established approach to detecting artificially contaminated drinking water samples; E. coli O157:H7 of 1.0 × 102 cells/mL were accurately quantified upon sample enrichment. It is believed that the proposed method will find wide applications in many fields demanding bacterial identification and quantification.
tional culture-based methods are laborious, time-consuming, and only suitable for viable and cultivable cells. In recent years, numerous culture-independent techniques have been developed for the rapid and sensitive detection of pathogenic bacteria, including DNA microarray,1-3 real-time polymerase chain reaction (RT-PCR),4-6 enzyme-linked immunosorbent assay (ELISA),7,8 surface plasmon resonance (SPR),9-11 magnetic nanoparticlebased immunoassay,12,13 various formats of biosensors,14-19 and flow cytometry.20-25 Though flow cytometry is emerging as one
Identification and quantification of infectious disease agents is important for medical diagnosis, public health, food safety, environment monitoring, and antibioterrorism. However, tradi-
(22)
* To whom correspondence should be addressed. Phone: 86-592-2184519. Fax: 86-592-2189959. E-mail:
[email protected]. 10.1021/ac902524a 2010 American Chemical Society Published on Web 12/29/2009
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10)
(11) (12) (13) (14) (15) (16) (17) (18) (19) (20) (21)
(23) (24) (25)
Cho, J. C.; Tiedje, J. M. Appl. Environ. Microbiol. 2001, 67, 3677–3682. Ahn, S.; Walt, D. R. Anal. Chem. 2005, 77, 5041–5047. Kim, B. C.; Park, J. H.; Gu, M. B. Anal. Chem. 2005, 77, 2311–2317. Malorny, B.; Lofstrom, C.; Wagner, M.; Kramer, N.; Hoorfar, J. Appl. Environ. Microbiol. 2008, 74, 1299–1304. Kramer, M.; Obermajer, N.; Bogovic Matijasic, B.; Rogelj, I.; Kmetec, V. Appl. Microbiol. Biotechnol. 2009, 84, 1137–1147. Mafu, A. A.; Pitre, M.; Sirois, S. J. Food Prot. 2009, 72, 1310–1314. Ferguson, C. M.; Booth, N. A.; Allan, E. J. Lett. Appl. Microbiol. 2000, 31, 390–394. Tamminen, M.; Joutsjoki, T.; Sjoblom, M.; Joutsen, M.; Palva, A.; Ryhanen, E. L.; Joutsjoki, V. Lett. Appl. Microbiol. 2004, 39, 439–444. Oh, B. K.; Lee, W.; Chun, B. S.; Bae, Y. M.; Lee, W. H.; Choi, J. W. Biosens. Bioelectron. 2005, 20, 1847–1850. Maalouf, R.; Fournier-Wirth, C.; Coste, J.; Chebib, H.; Saikali, Y.; Vittori, O.; Errachid, A.; Cloarec, J. P.; Martelet, C.; Jaffrezic-Renault, N. Anal. Chem. 2007, 79, 4879–4886. Dudak, F. C.; Boyaci, I. H. Biotechnol. J. 2009, 4, 1003–1011. Gu, H.; Ho, P. L.; Tsang, K. W.; Wang, L.; Xu, B. J. Am. Chem. Soc. 2003, 125, 15702–15703. Gao, J.; Li, L.; Ho, P. L.; Mak, G. C.; Gu, H.; Xu, B. Adv. Mater. 2006, 18, 3145–3148. Gehring, A. G.; Patterson, D. L.; Tu, S. I. Anal. Biochem. 1998, 258, 293– 298. Wong, Y. Y.; Ng, S. P.; Ng, M. H.; Si, S. H.; Yao, S. Z.; Fung, Y. S. Biosens. Bioelectron. 2002, 17, 676–684. Susmel, S.; Guilbault, G. G.; O’Sullivan, C. K. Biosens. Bioelectron. 2003, 18, 881–889. Wang, C.; Irudayaraj, J. Small 2008, 4, 2204–2208. Liao, W. C.; Ho, J. A. Anal. Chem. 2009, 81, 2470–2476. Lu, Q.; Lin, H.; Ge, S.; Luo, S.; Cai, Q.; Grimes, C. A. Anal. Chem. 2009, 81, 5846–5850. Ferrari, B. C.; Oregaard, G.; Sorensen, S. J. Microb. Ecol. 2004, 48, 239– 245. Moragues, M.; Comas-Riu, J.; Vives-Rego, J. Folia Microbiol. (Praha) 2004, 49, 587–590. Busam, S.; McNabb, M.; Wackwitz, A.; Senevirathna, W.; Beggah, S.; Meer, J. R.; Wells, M.; Breuer, U.; Harms, H. Anal. Chem. 2007, 79, 9107–9114. Hibi, K.; Mitsubayashi, K.; Fukuda, H.; Ushio, H.; Hayashi, T.; Ren, H.; Endo, H. Biosens. Bioelectron. 2007, 22, 1916–1919. Hahn, M. A.; Keng, P. C.; Krauss, T. D. Anal. Chem. 2008, 80, 864–872. Qin, D.; He, X.; Wang, K.; Tan, W. Biosens. Bioelectron. 2008, 24, 626– 631.
Analytical Chemistry, Vol. 82, No. 3, February 1, 2010
1109
of the best choices for microbe quantification by providing rapid, quantitative, and multiparameter measurement at the single-cell level, its applications to sensitive detection of bacterial cells are frequently hindered by bacteria’s small sizes and consequently the low contents of specific cellular constituents.26,27 To improve the sensitivity of bacteria detection by flow cytometry, various fluorescent-bioconjugated nanoparticles, including fluorescent latex particles, dye-doped silica nanoparticles, and quantum dots,24,28-30 have been adopted as probes to enable amplification of the analytical signal. Although fluorescent nanoparticles have superiority over conventional fluorophores in terms of fluorescence intensity and photostability, they usually generate significantly higher false positives.25,31 On the other hand, efforts have been made to develop compact, disposable, and ultrasensitive flow cytometers for bacteria measurement. Employing microchip technology, flow cytometers were miniaturized on microfabricated fluidic devices. Detection, sorting, and recovery of Escherichia coli (E. coli) cells have been demonstrated on the chip.32-34 Bacterial cells were quantified in river water (containing 106-107 cells/ mL) using a commercially available on-chip flow cytometer via a standard curve generated by fluorescent beads of known particle concentrations.35 With an improved microfluidic device coupled with charge-coupled device (CCD) camera detection, bacterial cells were determined at a density on the order of 104-106/mL without sample preconcentration.36 For this system, bacteria was quantified by calculating the cell count at each flow volume, and 10 min of analysis time was required for recording a movie to obtain reliable results. By costaining with SYTO 9 (membrane-permeable) and propidium iodide, differentiation of live and dead bacterial cells was demonstrated on a microfluidic chamber when a color CCD camera was used for detection.37 Combined with bioconjugated luminescent nanoparticles, a simple flow cytometer using silica microcapillary (inner diameter ) 50 µm) as flow cell was reported for bacterial detection.29 However, hardly any of the above techniques have been applied to multiparameter detection for a bacterial mixture with different species. Because total direct counts of bacterial abundance are central in assessing the biomass and bacteriological quality of water in ecological and industrial applications,38 it is desirable not only to rapidly identify specific species or strains in complex matrices but also (26) Vives-Rego, J.; Lebaron, P.; Nebe-von Caron, G. FEMS Microbiol. Rev. 2000, 24, 429–448. (27) Czechowska, K.; Johnson, D. R.; van der Meer, J. R. Curr. Opin. Microbiol. 2008, 11, 205–212. (28) Su, X. L.; Li, Y. Anal. Chem. 2004, 76, 4806–4810. (29) Zhao, X.; Hilliard, L. R.; Mechery, S. J.; Wang, Y.; Bagwe, R. P.; Jin, S.; Tan, W. Proc. Natl. Acad. Sci. U. S. A. 2004, 101, 15027–15032. (30) Edgar, R.; McKinstry, M.; Hwang, J.; Oppenheim, A. B.; Fekete, R. A.; Giulian, G.; Merril, C.; Nagashima, K.; Adhya, S. Proc. Natl. Acad. Sci. U. S. A. 2006, 103, 4841–4845. (31) Ferrari, B. C.; Bergquist, P. L. Cytometry A 2007, 71, 265–271. (32) McClain, M. A.; Culbertson, C. T.; Jacobson, S. C.; Ramsey, J. M. Anal. Chem. 2001, 73, 5334–5338. (33) Fu, A. Y.; Chou, H. P.; Spence, C.; Arnold, F. H.; Quake, S. R. Anal. Chem. 2002, 74, 2451–2457. (34) Dittrich, P. S.; Schwille, P. Anal. Chem. 2003, 75, 5767–5774. (35) Sakamoto, C.; Yamaguchi, N.; Nasu, M. Appl. Environ. Microbiol. 2005, 71, 1117–1121. (36) Sakamoto, C.; Yamaguchi, N.; Yamada, M.; Nagase, H.; Seki, M.; Nasu, M. J. Microbiol. Methods 2007, 68, 643–647. (37) Inatomi, K. I.; Izuo, S. I.; Lee, S. S. Lett. Appl. Microbiol. 2006, 43, 296– 300.
1110
Analytical Chemistry, Vol. 82, No. 3, February 1, 2010
to reveal total bacterial counts. Therefore, technical advances in instrumentation and methodology are needed. Very recently, a compact, high-sensitivity, dual channel flow cytometer (HSDCFCM) was developed in our laboratory for the individual analysis of nanosized particles and biomolecules.39 On the basis of the simultaneous detection of fluorescence and light scattering signals from individual nanoparticles, reliable and accurate enumeration of nanoparticles was demonstrated with a volumetric system. In the present work, the HSDCFCM was modified to comprise red and green fluorescence detection channels for the rapid, absolute, and simultaneous quantification of specific pathogenic strain and total bacterial cells. A bacterial cell mixture composed of heat-killed pathogenic E. coli O157:H7 and harmless E. coli DH5R was used as a model system. Pathogenic E. coli O157:H7 cells were selectively labeled with biotinylated anti-E. coli O157:H7 monoclonal antibodies and Alexa Fluor 647-R-phycoerythrin streptavidin conjugates that fluoresce red. The green fluorophore SYTO 9, a membrane-permeable nucleic acid stain, was used to stain all the bacterial cells including E. coli O157:H7 and E. coli DH5R. Double-stained E. coli O157: H7 can be specifically identified and enumerated using two-color fluorescence coincidence detection, and E. coli DH5R can be quantified by green fluorescence measurement. The correlation between the concentrations measured by the HSDCFCM and the plate-counting was examined. Bacterial cell mixtures with varying proportions of E. coli O157:H7 and E. coli DH5R were measured with good ratio correspondence. Additionally, the applicability of the established method was demonstrated with the artificially contaminated drinking water samples spiked with low abundance of E. coli O157:H7 (102 to 5.0 × 103 cfu/mL). EXPERIMENTAL SECTION Reagents and Chemicals. Alexa Fluor 647-R-phycoerythrin streptavidin conjugate and SYTO 9 nucleic acid stain were purchased from Molecular Probes (Eugene, OR). Anti-E. coli O157:H7 monoclonal antibody (mAb, purchased from Meridian Life Science, Cincinnati, OH) was biotinylated following the instruction of a biotin-XX microscale protein labeling kit (Molecular Probes).40 E. coli O157:H7 was kindly provided by the Xiamen Center for Disease Control and Protection. E. coli DH5R was a laboratory-kept reagent. All other chemicals for buffer preparation were obtained from Sigma (St. Louis, MO). Distilled, deionized water supplied by a Milli-Q RG unit (Millipore, Bedford, MA) was used in the preparation of buffer solutions. Instrumentation. The compact and high-sensitivity dualchannel flow cytometer described previously39 was modified to enable the two-color fluorescence detection of bacteria. Figure 1 shows the schematic diagram of the optical path for the modified HSDCFCM. Briefly, a 488 nm continuous-wave laser (Newport Corp., Irvine, CA; maximum output of 10 mW) was used to replace the 532 nm laser as the excitation source. Laser excitation power (measured after mirror reflection) of 1.0 mW was used in the present study. The 0.7 mm laser output beam was focused to a ∼11 µm diameter spot (1/e2) by an achromatic-doublet lens onto (38) Lisle, J. T.; Hamilton, M. A.; Willse, A. R.; McFeters, G. A. Appl. Environ. Microbiol. 2004, 70, 5343–5348. (39) Yang, L.; Zhu, S.; Hang, W.; Wu, L.; Yan, X. Anal. Chem. 2009, 81, 2555– 2563. (40) http://probes.invitrogen.com/media/pis/mp30010.pdf.
Figure 1. Schematic diagram of the optical path for the laboratorybuilt high-sensitivity dual channel flow cytometer (HSDCFCM): P, halfwave plate; S, polarizing beam splitter; M, mirror; L, achromaticdoublet lens; C, flow cell; A1, A2, and A3, aspheric lenses; Dic-F, dichroic filter; LP, long-pass filter; EF, edge filter; BP-1 and BP-2, band-pass filter; APD-1 and APD-2, single-photon counting avalanche photodiode.
the hydrodynamically focused sample stream inside the 250 µm square quartz flow channel (NSG Precision Cells, Farmingdale, NY). The emitted light from the sample stream was collected by an aspheric lens and then directed by a dichroic beam splitter (FF560-Di02, Semrock Inc., Rochester, NY) into two light paths for red and green fluorescence detection, respectively. The transmitted fluorescence passed through a long-pass filter (565ALP, Omega Optical Inc. Brattleboro, VT) and a bandpass filter (FF01-670/30, Semrock) and then was focused on APD-1 (SPCM-AQR-12, EG&G Canada, Vaudreuil, Canada) for red fluorescence detection. The reflected light was spectrally filtered by a Raman edge filter (LP03-488RS, Semrock) and a bandpass filter (FF01-520/35, Semrock) and then focused on the APD-2 (SPCM-AQR-14, EG&G Canada) for green fluorescence detection. The output signals from APD-1 and APD-2 were simultaneously counted by a National Instruments DAQ card (PCI-6713, Austin, TX) which was controlled with a program written in LabVIEW 8.0 (National Instruments Inc.). Data processing and the criteria used for peak identification were the same as those described previously.39 Sample solutions were delivered pneumatically via a precise pressure regulator, and the ultrapure water (Millipore) served as the sheath fluid via gravity feeding. The sample and sheath volumetric flow rate, measured by decrement and increment methods,39 respectively, were controlled at 55 nL/min and 15 µL/min, respectively, in the present study unless otherwise stated. For each bacterial sample, 60 s of data acquisition time was used. Dual Fluorescence Staining of Bacterial Cells. Since E. coli O157:H7 are highly pathogenic, experiments directly utilizing these pathogens were conducted in a Biosafety Level 2 laboratory. The pure cultures of E. coli O157:H7 or E. coli DH5R were grown overnight in Luria-Bertani (LB) broth (10 g of tryptone, 5 g of yeast extract, and 10 g of NaCl/L) at 37 °C in baffled flasks with rotary aeration for about 16 h until stationary phase was reached. The concentrations of the E. coli O157:H7 and E. coli DH5R stock
solutions were determined to be 4 × 109 and 3.8 × 109 cfu/mL, respectively, by the conventional plate-counting method. The harvested E. coli O157:H7 and E. coli DH5R cells were heatkilled and stored at 4 °C. The E. coli O157:H7 stock sample was diluted in a 10-fold series to concentration range of 106-108 cfu/mL with dilution buffer (1% bovine serum albumin (BSA) in phosphate-buffered saline (PBS)), and blocking was allowed to take place for 30 min. Biotinylated anti-E. coli O157:H7 monoclonal antibody (40 µL, 0.4 µg/mL) and the diluted E. coli O157:H7 bacterial sample (20 µL, 106-108 cfu/mL) were mixed and incubated at 37 °C for 30 min. Alexa Fluor 647-Rphycoerythrin streptavidin conjugate (40 µL, 0.4 µg/mL) was then added and incubated at 37 °C for 20 min under dark condition. In the end, SYTO 9 stain (100 µL, 40 nM) was added and 5 min of room temperature incubation was allowed prior to the analysis on the HSDCFCM. The final concentrations of the E. coli O157:H7 samples ranged from 1.0 × 105 to 4.0 × 107 cfu/mL. Note that no washing was performed upon each incubation step. E. coli DH5R samples (final concentrations ranging from 3.8 × 105 to 3.8 × 107 cfu/mL) and bacterial cell mixtures of E. coli O157:H7 and E. coli DH5R were stained with the same procedure. Preparation of Bacteria Spiked Drinking Water Samples. Bottled natural mineral water was purchased from a local supermarket and filtered through a 0.2 µm membrane. Pathogenic bacterial spiked drinking water samples were prepared by adding 10, 20, and 50 µL of 4.0 × 106 cfu/mL E. coli O157:H7 into 40 mL of the filtered natural mineral water contained in a 50 mL tube. The final concentrations of the bacteria spiked water samples were 1.0 × 103, 2.0 × 103, and 5.0 × 103 cfu/mL, respectively. These samples were then centrifuged at 8000 rpm for 30 min. Supernatants were carefully removed with sterile micropipets without disturbing the bacteria pellet. The cell pellet was transferred to a 1.5 mL Eppendorf tube followed by three washes with 500 µL sterile PBS each to remove residual bacterial cells attached to the surface of the 50 mL tube. The washing solutions were all collected to the same 1.5 mL Eppendorf tube. Then, the bacteria suspensions were concentrated to 20 µL by centrifugation at 10000 rpm for 5 min. Finally, the bacteria spiked water samples were double-stained to a final volume of 200 µL. That is, the 40 mL spiked natural mineral water was 200-fold concentrated before being analyzed. Sterile PBS buffer was used as negative control. Three parallel samples were prepared for each concentration. Similarly, a great bulk of 240 mL filtered natural mineral water spiked with 1.0 × 102 cfu/mL of E. coli O157:H7 was 1200-fold concentrated to a final concentration of 1.2 × 105 cfu/mL and then analyzed on the HSDCFCM. RESULTS AND DISCUSSION Detection Principle. This paper aims to develop a rapid and sensitive strategy for the absolute and simultaneous quantification of pathogenic and total bacterial cells in a mixture. The detection principle is illustrated in Scheme 1. In the first step of cell staining, E. coli O157:H7 cells were specifically labeled with biotinylated monoclonal antibodies and further conjugated with Alexa Fluor 647-R-phycoerythrin streptavidin conjugates, while E. coli DH5R cells were not recognized by the antibody. Then all the bacterial cells, including both E. coli O157:H7 and E. coli DH5R, were Analytical Chemistry, Vol. 82, No. 3, February 1, 2010
1111
Scheme 1. Schematic Representation of the Detection Principle for the Simultaneous Enumeration of Pathogenic Bacteria and Total Bacterial Cells in a Mixture
stained with nucleic acid dye SYTO 9. Alexa Fluor 647-Rphycoerythrin streptavidin tandem conjugate has excitation and emission maxima at ∼496, 565/667 nm. It has been widely used to detect biotinylated probes by flow cytometry in multicolor applications using a single excitation wavelength of 488 nm.41 SYTO 9 is a membrane-permeable nucleic acid stain with excitation/emission maxima of 485/498 nm, and a large fluorescence enhancement can be obtained upon binding with nucleic acids in both live and dead cells.42 Upon double fluorescent staining, the bacterial sample was analyzed by the HSDCFCM. The HSDCFCM uses a 488 nm laser to excite Alexa Fluor 647-R-phycoerythrin streptavidin conjugate and SYTO 9 concurrently, and the emitted red and green fluorescence signals are detected simultaneously by the APD-1 and APD-2 detectors. For a bacterial cell mixture, E. coli O157:H7 exhibits signals derived from both Alexa Fluor 647-R-phycoerythrin (red fluorescence) and SYTO 9 (green fluorescence), while E. coli DH5R only shows green fluorescence. Consequently, E. coli O157:H7 cells can be quantified by counting the correlated peaks on the two APD detectors, and the total bacterial cells can be quantified by counting the green fluorescence peaks detected by the APD-2. E. coli O157:H7 Enumeration by Two-Color Fluorescence Coincidence Detection. Before attempting flow cytometric analysis of double-stained pathogenic bacteria, E. coli O157:H7 cells were stained with biotinylated anti-E. coli O157:H7 monoclonal antibodies and Alexa Fluor 647-R-phycoerythrin streptavidin conjugates and tested on the HSDCFCM. Representative data of single-stained E. coli O157:H7 are shown in Figure 2. Negative control experiment of PBS buffer spiked with the same amount of staining reagents demonstrated that there was no appreciable fluorescence burst peaks observed on both fluorescence channels (Figure 2a,b). Because no washing steps was carried out upon fluorescent staining, the existence of many free forms of fluorescent probes (Alexa Fluor 647-R-phycoerythrin streptavidin conjugate) in the solution yielded some level of background signals on both channels (APD-1 channel, 244 ± 34 counts/bin; APD-2 channel, 41 ± 7 counts/bin). Nevertheless, the red fluorescence bursts generated from each individual E. coli O157:H7 cells were (41) http://probes.invitrogen.com/media/pis/mp00888.pdf. (42) http://probes.invitrogen.com/media/pis/mp07572.pdf.
1112
Analytical Chemistry, Vol. 82, No. 3, February 1, 2010
easily distinguished from the background (Figure 2c). Meanwhile, several small bursts with peak height around 100 counts/bin were observed on the green fluorescence APD-2 channel (Figure 2d) with good time correlation with red fluorescence peaks in the APD-1 channel. As peaks of similar intensity were observed on the APD-2 green fluorescence channel for unstained bacterial cells (data not shown), these peaks can be mainly ascribed to the leakage of light-scattering signals despite the edge filter and bandpass filter placed in the optical path. Because rod-shaped E. coli bacterial cells can pass through the interrogating laser beam at different angles, the peak intensity observed on the APD-2 channel (due to scattering) does not necessarily correlate with the peak intensity observed on the APD-1 red fluorescence channel. The bivariate dot-plot of the red fluorescence burst area versus the green fluorescence burst area is shown in Figure 2e. The green fluorescence burst area of single-stained E. coli O157: H7 was calculated by integrating the photon counts detected by APD-2 within the corresponding burst duration of the peaks identified on the red fluorescence channel. Figure 2e shows that E. coli O157:H7 cells emitted bright red fluorescence signals on the APD-1 channel. For single-stained E. coli O157:H7 enumeration by red fluorescence measurement, multiple data sets (n ) 3) were collected consecutively for bacterial samples prepared at each concentration. Excellent correlation (R2 ) 0.9995) was obtained between the concentrations (1.0 × 105 to 4.0 × 107 cfu/mL) measured by the HSDCFCM and the plate-counting (Figure 2f). The average detection efficiency was 95 ± 11%, and the variation was mainly from the lower concentration samples. As shown in Figure 3, for double-stained PBS negative control and E. coli O157:H7, the addition of nucleic acid stain SYTO 9 did not cause obvious increase of the background signal on both fluorescence channels (APD-1 channel, 249 ± 34 counts/bin; APD-2 channel, 50 ± 7 counts/bin) as compared to the singlestained samples. When E. coli O157:H7 cells were double-stained, the peak height of fluorescence bursts detected on the green fluorescence APD-2 channel (Figure 3d) was much higher than that without SYTO 9 staining (Figure 2d). As shown in Figure 3c,d, for double-stained E. coli O157:H7 at concentration of 4.0 × 106 cfu/mL, 12 fluorescence peaks were detected within a 3 s time interval, and the red and green fluorescence bursts
Figure 2. Flow cytometric analysis of single-stained E. coli O157:H7 on the laboratory-built HSDCFCM. (a and c) Red fluorescence burst trace from APD-1 channel for PBS negative control and single-stained E. coli O157:H7 cells, respectively; (b and d) green fluorescence burst trace from APD-2 channel for the above two samples, respectively; (e) bivariate dot-plot of red fluorescence burst area versus green fluorescence burst area for E. coli O157:H7 cells; (f) relationship between HSDCFCM counting and plate-counting of E. coli O157:H7 cells. Error bars indicate standard deviation (n ) 3). For c-e, concentration of E. coli O157:H7 was 4.0 × 106 cfu/mL. For a-d, the data were binned into 100 µs intervals.
Figure 3. Flow cytometric analysis of double-stained E. coli O157:H7 on the laboratory-built HSDCFCM. (a and c) Red fluorescence burst trace from APD-1 channel for PBS negative control and double-stained E. coli O157:H7 cells, respectively; (b and d) green fluorescence burst trace from APD-2 channel for the above two samples, respectively; (e) bivariate dot-plot of red fluorescence burst area versus green fluorescence burst area for E. coli O157:H7 cells; (f) relationship between HSDCFCM counting and plate-counting of E. coli O157:H7 cells. Error bars indicate standard deviation (n ) 3). For c-e, concentration of E. coli O157:H7 was 4.0 × 106 cfu/mL. For a-d, the data were binned into 100 µs intervals.
generated from the same bacterial cell correlated quite well in the time frame. On the basis of the mean of the E. coli O157: H7 fluorescence burst height and the standard deviation of the background signal, the signal-to-noise ratios (S/N) were calculated to be 15 and 16 for E. coli O157:H7 detection on the red and green fluorescence channels, respectively. As shown in the dot-plot of the red fluorescence burst area versus the green fluorescence burst area (Figure 3e), double-stained E. coli O157:H7 cells emitted bright fluorescence signals on both
channels. For double-stained E. coli O157:H7 enumeration by twocolor fluorescence coincidence detection, the concentrations (1.0 × 105 to 4.0 × 107 cfu/mL) of E. coli O157:H7 determined by the HSDCFCM and the plate-counting approach correlate quite well (R2 ) 0.9993). The average detection efficiency of HSDCFCM is 93 ± 8%. Here we should note that for our laboratorybuilt HSDCFCM, the sample flow rate of 30-100 nL/min is most suitable for accurate analysis by enabling stable flow and small probe volume, thus protecting the APD detectors from Analytical Chemistry, Vol. 82, No. 3, February 1, 2010
1113
Figure 4. Flow cytometric analysis of double-stained E. coli DH5R on the laboratory-built HSDCFCM. (a) Red fluorescence burst trace from APD-1 channel; (b) green fluorescence burst trace from APD-2 channel; (c) dot-plot of red fluorescence burst area versus green fluorescence burst area; (d) relationship between HSDCFCM counting and plate-counting of E. coli DH5R cells. Error bars indicate standard deviation (n ) 3). For a-c, concentration of E. coli DH5R was 3.8 × 106 cfu/mL. For a and b, the data were binned into 100 µs intervals.
high background noise. Under these flow conditions, theoretically 3-10 bacteria (depending on the flow rate) can be detected within 1 min of data acquisition time for a bacterial sample of 1.0 × 105 cells/mL. Since the flow rate is quite stable, HSDCFCM can accurately quantify bacteria at concentrations of above 105 cfu/mL within 1 min of analysis time. For bacteria samples at abundance lower than 1.0 × 105 cfu/mL, prolonged acquisition time or preconcentration of bacteria samples needs to be applied so that an adequate number of cells per sample can be analyzed to ensure an acceptable level of statistical confidence. E. coli DH5r Enumeration by Green Fluorescence Measurement. To assess the accuracy of bacteria quantification with nucleic acid staining, double-stained E. coli DH5R were analyzed on the HSDCFCM with bacteria concentrations ranging from 3.8 × 105 to 3.8 × 107 cfu/mL. As shown in Figure 4a,b, fluorescence signal generated from each individual E. coli DH5R cells can only be detected on the green fluorescence channel (APD-2), suggesting that the nonspecific interaction of anti-E. coli O157:H7 monoclonal antibody to E. coli DH5R was negligible. Compared to the small bursts detected in Figure 2d (originated from leakage of light scattering signals of bacterial cells), the peak intensity of SYTO 9 stained E. coli DH5R was much stronger. Therefore, accurate enumeration of bacterial cells on the APD-2 channel will not be affected by the coexisting nonbiological particles or debris unless their sizes are much bigger than the bacteria. Moreover, because the APD-2 green fluorescence channel was designed for total bacteria counting, the leakage of light-scattering signals from 1114
Analytical Chemistry, Vol. 82, No. 3, February 1, 2010
bacterial cells can be considered as a good asset to some extent. The dot-plot of the red fluorescence burst area versus the green fluorescence burst area is shown in Figure 4c. The red fluorescence burst area of E. coli DH5R was calculated by integrating the background photon counts detected by APD-1 within the corresponding burst duration of the peaks identified on the green fluorescence channel. Figure 4d shows that, for E. coli DH5R, excellent correlation (R2 ) 0.9998) was obtained between the concentrations measured by the HSDCFCM and the platecounting. Identification and Quantification of Pathogenic E. coli O157:H7 in Bacterial Cell Mixtures. The capability of the laboratory-built HSDCFCM for the absolute and simultaneous quantification of specific pathogenic strain and total bacterial cells was assessed by analyzing mixtures of E. coli O157:H7 and E. coli DH5R cells. The percentages of E. coli O157:H7/total bacterial cells were 51/100, 41/100, 26/100, and 10/100 with the total concentration of 3.9 × 106 cfu/mL and 1/100 with the total concentration of 2 × 107 cfu/mL. Figure 5 shows two-color flow cytometric analysis of a mixture with E. coli O157:H7/total bacterial cells percentage of 51/100. As we expected, E. coli O157: H7 yielded signals derived from Alexa Fluor 647-R-phycoerythrin (red fluorescence) and SYTO 9 (green fluorescence), while E. coli DH5R only exhibited green fluorescence from SYTO 9. In panel b, the peaks marked by number are signals generated by E. coli O157:H7 cells, which have corresponding peaks on the red fluorescence channels. The peaks shown on the green
Figure 5. Flow cytometric analysis of a double-stained mixture with the percentage of E. coli O157:H7/total bacterial cells of 51/100 at a total concentration of 3.9 × 106 cfu/mL. (a) Red fluorescence burst trace from APD-1 channel; (b) green fluorescence burst trace from APD-2 channel; (c) dot-plot of red fluorescence burst area versus green fluorescence burst area. In panel b, the peaks marked by number are the signal from E. coli O157:H7 cells, which have corresponding peaks on the red fluorescence channels. For a and b, the data were binned into 100 µs intervals. In panel c, a discriminant line was drawn to facilitate an easy discrimination between E. coli O157:H7 and E. coli DH5R on the red fluorescence channel. Table 1. Comparison of the Theoretical and Detected Ratios of Esherichia coli O157:H7/Total Bacterial Cells for Bacterial Cell Mixtures Measured on the HSDCFCM E. coli O157:H7/total bacterial cells (theoretical)
E. coli O157:H7/total bacterial cells (experimental)a
RSD (n ) 3, %)
51/100 41/100 26/100 10/100 1/100b
52/100 41/100 26/100 12/100 0.99/100
2 4 4 4 9
a Average ratio of E. coli O157:H7/total bacterial cells measured from three 60 s runs. b The total concentration of the mixture was 2 × 107 cfu/mL.
fluorescence channel were defined as total bacteria. The dot-plot of red fluorescence burst area versus green fluorescence burst area (Figure 5c) shows that the bacterial mixture was distinctly separated into two populations of E. coli O157:H7 and E. coli DH5R based on the red fluorescence intensity of Alexa Fluor 647-Rphycoerythrin. A discriminant line was drawn to facilitate an easy discrimination between E. coli O157:H7 and E. coli DH5R on the red fluorescence channel (Figure 5c). Table 1 summarizes the theoretical and experiment detected ratios of E. coli O157:H7 versus total bacterial cells for bacterial cell mixtures analyzed on the HSDCFCM. Multiple data sets (n ) 3) were consecutively collected for each sample. Evidently, a bacterial cell mixture of large ratio difference can be quantified with good accuracy and precision. To further demonstrate the specificity of the proposed approach for bacterial mixture analysis, two other types of bacteria (E. coli ER2738 and Micrococcus lysodeikticus (M. lysodeikticus))
and their mixtures with E. coli O157:H7 were double-stained and examined on the HSDCFCM. The results indicate that the nonspecific interaction of anti-E. coli O157:H7 monoclonal antibody with E. coli ER2738 and M. lysodeikticus was also negligible (Supporting Information, Figure S1 and Figure S2). Clearly, it is feasible to simultaneously enumerate E. coli O157:H7 and total bacteria in the mixtures with great accuracy using the HSDCFCM detection combined with dual fluorescence staining. Application of E. coli O157:H7 Quantification in Drinking Water Samples. To test the feasibility of applying the aforeestablished system to bacteria quantification in real samples, artificially contaminated drinking water was prepared by spiking different concentrations of E. coli O157:H7 in natural mineral water. The spiked E. coli O157:H7 samples were preconcentrated by centrifugation and then analyzed on the HSDCFCM after dual fluorescence staining. The sample flow rate was controlled at 70 nL/min for this experiment. As shown in Table 2, using a 40 mL dose of water sample, bacterial concentration as low as 1.0 × 103 cfu/mL can be reliably detected using preconcentration steps. If 240 mL of sample volume is available, the detection limit of the original bacterial concentration can be further reduced to 1.0 × 102 cfu/mL after 1200-fold preconcentration. Upon preconcentration, the detection efficiencies of E. coli O157:H7 in the spiked water samples varied from 81 to 88% with an average of 85 ± 3%. This value is comparable to the aforereported 93% of detection efficiency for E. coli O157:H7 without preconcentration. These results clearly demonstrate that the HSDCFCM has the ability to efficiently detect and quantify bacterial cells in drinking water samples. Analytical Chemistry, Vol. 82, No. 3, February 1, 2010
1115
Table 2. Enumeration of Escherchia coli O157:H7 in Spiked Water Sample on the HSDCFCM ICa (cfu/mL) 1.0 × 10 1.0 × 103 2.0 × 103 5.0 × 103 2
concentration fold
CCb (cfu/mL)
event ratec (cells/min)
MCc (cfu/mL)
MC/CC (%)
1200 200 200 200
1.2 × 10 2.0 × 105 4.0 × 105 1.0 × 106
7 ± 1.7 12 ± 0.6 24 ± 1.0 57 ± 1.0
(1.0 ± 0.14) × 105 (1.8 ± 0.08) × 105 (3.4 ± 0.38) × 105 (8.1 ± 0.22) × 105
83 88 86 81 av ± SD: 85 ± 3
5
a Initial concentration (IC) of E. coli O157:H7 in spiked water sample. b Calculated concentration (CC) of E. coli O157:H7 in spiked water sample after preconcentration. c Measured concentration (MC) of E. coli O157:H7 on the HSDCFCM based on dual-channel coincidence detection.
CONCLUSION By integrating a laboratory-built HSDCFCM with antigen and nucleic acid double fluorescence staining, we have developed a sensitive approach for the rapid, absolute, and simultaneous quantification of specific pathogenic strain and total bacteria cells in mixtures. Since fluorescence burst generated from each individual bacterial cell was well-resolved from the background, the HSDCFCM provides direct and absolute enumeration of bacteria. With an acquisition time of 1 min/sample, bacteria concentration of 1.0 × 105 cfu/mL and above can be accurately quantified. With the implementation of a preconcentration step, bacteria with abundance as low as 1.0 × 102 cfu/mL were reliably enumerated. For a bacterial cell mixture, the HSDCFCM can specifically identify the pathogenic bacteria and simultaneously quantify both pathogenic and total bacteria cells accurately. The usefulness of this approach was demonstrated by the accurate detection of E. coli O157:H7 bacterial cells in spiked natural mineral waters. Applicability of the proposed approach in environmental and food sample analysis is under investigation. By using selective antibodies to other pathogens, the HSDCFCM holds great potential for the rapid detection of
1116
Analytical Chemistry, Vol. 82, No. 3, February 1, 2010
a wide variety of pathogenic bacteria in biomedical and biotechnological areas. ACKNOWLEDGMENT This work was supported by the National Natural Science Foundation of China (Grants 20645001, 20675070, 20975087, and 90913015), Department of Science and Technology of Fujian Province (Grant 2005NZ1013), the Program for New Century Excellent Talents in University (Grant NCET-07-0729), and the Scientific Research Foundation for the Returned Overseas Chinese Scholars, State Education Ministry (SRF for ROCS, SEM), for which we are most grateful. SUPPORTING INFORMATION AVAILABLE Figures showing two-color flow cytometric analyses. This material is available free of charge via the Internet at http://pubs.acs.org. Received for review December 8, 2009. AC902524A
November
4,
2009.
Accepted
