Capillary zone electrophoresis for enumeration of Lactobacillus delbrueckii subsp. bulgaricus and Streptococcus thermophilus in yogurt

Journal of Chromatography B, 877 (2009) 710–718

Contents lists available at ScienceDirect

Journal of Chromatography B journal homepage: www.elsevier.com/locate/chromb

Capillary zone electrophoresis for enumeration of Lactobacillus delbrueckii subsp. bulgaricus and Streptococcus thermophilus in yogurt Orathai Lim a , Worapot Suntornsuk a , Leena Suntornsuk b,∗ a b

Department of Microbiology, Faculty of Science, King Mongkut’s University of Technology Thonburi, Bangkok, Thailand Department of Pharmaceutical Chemistry, Faculty of Pharmacy, Mahidol University, 447 Sri-Ayudhaya Rajathevee, Bangkok 10400, Thailand

a r t i c l e

i n f o

Article history: Received 2 October 2008 Accepted 3 February 2009 Available online 11 February 2009 Keywords: Lactobacillus delbrueckii subsp. bulgaricus Streptococcus thermophilus Capillary zone electrophoresis Yogurt

a b s t r a c t Enumeration of Lactobacillus delbrueckii subsp. bulgaricus and Streptococcus thermophilus is a priority due to their importance in yogurt production. Capillary electrophoresis (CE) of both bacteria could be achieved in 7.2 min with a resolution of 3.2 in the background electrolyte (BGE) containing 4.5 mM Tris(hydroxymethyl) amminomethane (TRIS)–4.5 mM boric acid–0.1 mM ethylenediamine tetraacetate (EDTA) (TBE) buffer (pH 8.4) and 0.05% (v/v) polyethylene oxide (PEO), using a capillary of 47.5 cm (effective length) × 100 ␮m i.d., injection of 50 mbar × 3 s followed by −5 kV × 120 s, a voltage and temperature of 20 kV and 25 ◦ C, respectively. Appropriate amounts of PEO in the BGE, sample preparation (i.e. vortex) and introduction were key factors for their separation. A short hydrodynamic injection followed by applying reversed polarity voltage could compress the bacteria into narrow zones, which were detected as separated single peaks. Method linearity (r2 > 0.99), precision (%RSDs < 9.3%), recovery (%R = 91.7–106.7%) and limit of quantitation (1.0 × 106 colony forming unit per mL (CFU/mL)) were satisfactory. Results from the CE analysis of both bacteria in yogurt were not statistically different from those of the plate count method (P > 0.05). The CE method can be used as an alternative for quantitation of L. delbrueckii subsp. bulgaricus and S. thermophilus in yogurt since it was reliable, simple, cost and labor effective and rapid, allowing the analysis of 3 samples/h (comparing to 2d/sample by plate count method). © 2009 Elsevier B.V. All rights reserved.

1. Introduction Analysis of microbes (i.e. viruses, bacteria and fungi) is of great interest since they are important in many industries including food, pharmaceutical and health care industries. Conventionally, microbes are determined by the direct count method (i.e. standard plate count method) [1], which is popular for its simplicity and low cost. Although this method is still used in most microbiological laboratories, the method suffers from the long analysis time, intensive labor and the cross-contamination during the analysis, which can yield false positive results. Alternatively molecular methods (i.e. hybridization, amplification and immunoassay techniques) are available for microbial analysis [2]. However, these methods can be costly, complex and require well trained personnel. Capillary electrophoresis (CE) is now a well established method, which shows a wide range of applications including analyses of inorganic species, small organic compounds, macromolecules and biopolymers. Additionally, CE for the analysis of colloidal or nano-particles (e.g. polymers and inorganic particles, microbes, mammalian cells, cell

∗ Corresponding author. Tel.: +66 2 644 8695; fax: +66 2 644 8695. E-mail address: [email protected] (L. Suntornsuk). 1570-0232/$ – see front matter © 2009 Elsevier B.V. All rights reserved. doi:10.1016/j.jchromb.2009.02.014

subunits and other biological particles) has been recently reviewed [3–6]. Bacteria are biocolloids, which consist of a membrane composed of glycoproteins and lipids that are electrically charged by the proteolysis of amino acid residues, acidic carbohydrate groups, organophosphates and sulfates [7]. Consequently, bacteria can move under an electric field with velocities that are proportional to their electrophoretic mobilities [7]. Additionally, different bacteria have distinct electrophoretic mobilities that allow their simultaneous analysis and identification by CE [7]. However, CE analysis of bacteria can be problematic due to their various sizes, charges, particle masses, which are often unknown, their electrophoretic heterogeneity, interactions with capillary wall, particle–particle interactions, their stability and aggregation [3]. Particle adherence or particle–particle interactions can also clog the capillary. Hjerten et al. firstly demonstrated the potential of CE for viral (i.e. tobacco mosaic virus (TMV)) and bacterial (i.e. Lactobacillus casei) analyses [8]. The separation was obtained in a 100 ␮m i.d. capillary (coated with methylcellulose or linear polyacrylamide) using Tris–HCl (pH 7.5) and Tris–HOAc (pH 8.6) buffer as background electrolyte (BGE) for TMV and L. casei, respectively. Subsequent works by Ebersole and McCormick [9] and Pfetsch and Welsch [10] showed the success CE in bacterial analysis despite of the long analysis time

O. Lim et al. / J. Chromatogr. B 877 (2009) 710–718

and poor peak shapes. The former group attempted to separate Streptococcus pyrogenes, S. agalactiae, S. pneumoniae, Staphylococcus aureus and E. faecalis in a 250 cm long, 100 ␮m i.d. capillary [9], whereas the latter group showed that the minimum concentration of Tris–borate buffer corresponds to an ionic strength of 0.2 mM should be used for the separation of various Pseudomonas species [10]. Armstrong et al. have extensively described the use of CE for analysis of various microorganisms [2,11–16]. The improved peak shapes and separation efficiency were obtained in the BGE containing poly(ethylene oxide) (PEO), which functioned as a focusing agent [11–15]. Importantly, proper preparations of bacterial samples (i.e. vortex or sonication) prior to CE injection greatly decreased the aggregation of bacteria resulting in the reduction of the number of peaks [11–13,16–17]. These techniques have been applied for testing of sterility [2], bacterial contamination in food samples [7], quantitation of bacteria and identification of the bacteria causing a urinary tract infection [12], cell viability [14] and active bacteria in dietary supplements [16]. Recently, Klodzlnska et al. demonstrated a rapid CZE method for the identification of E. coli in human urine and the analysis of pure culture of Helicobacter pylori using PEO as the focusing agent and calcium and myoinositol hexakisphosphate as specific ions to interact with the bacterial surface and to change their electrical properties and electrophoretic mobilities [18]. In this work, CE separation of Lactobacillus delbrueckii subsp. bulgaricus and Streptococcus thermophilus was demonstrated. Both bacteria are generally recognized as safe to human consumption. They are traditionally yogurt starter cultures, which benefit each other in a yogurt product. L. delbrueckii subsp. bulgaricus generally secretes proteases to digest casein in milk with the liberation of low particle mass peptides and amino acids as specific growth stimulants for S. thermophilus. S. thermophilus grows faster and produces both acids and carbon dioxide promoting L. delbrueckii subsp. bulgaricus growth. The bacteria consume lactose in milk and produce lactic acid causing distinctive and characteristic taste (i.e. sharp and acidic), which lead to (1) the coagulation of the casein (milk protein), (2) the formation of yoghurt gel and (3) the development pleasant flavors and aromas in yogurt products. The flavors and aromas come from different compounds such as non-volatile acids (e.g. lactic acid, pyruvic acid, oxalic acid and succinic acid), volatile acids (e.g. formic, acetic, propionic and butyric acids), and carbonyl compounds (e.g. acetaldehyde, acetone and diacetyl). The bacteria produce exopolysaccharide to provide the desired textural and organoleptic characteristics including mouthfeel, gel firmness and viscosity. Additionally, they serve as probiotics for promoting consumers’ health. They also improve nutritional properties of milk product by producing vitamins such as niacin and folic acid. The number and ratio of L. delbrueckii subsp. bulgaricus and S. thermophilus play a significant role on the preparation of yogurt starters, monitoring of yogurt fermentation process and quality control of yogurt products including sensory and texture properties [19]. Both bacteria have to be active in an appropriate concentration (about 108 CFU/g) and the ratio of both strains should be well balanced at ratios of 1:1 to 3:1 (L. delbrueckii subsp. bulgaricus:S. thermophilus) to retain a high quality yogurt products. Therefore, the enumeration of both bacteria is important for preparing yogurt starters, monitoring of yogurt fermentation process and quality control of yogurt products. The simultaneous separation and enumeration of these bacteria by CE have not been previously described in literatures. The optimum condition was investigated by varying the PEO concentrations, injection techniques, capillary lengths, separating voltage and temperature. Method sensitivity was significantly enhanced by the “large-volume sample stacking (LVSS)” with polarity switching [20]. Finally, the proposed method was validated and applied for the analysis of both bacteria in yogurt samples.

711

2. Experimental 2.1. Chemicals Tris(hydroxymethyl) amminomethane (TRIS), boric acid and ethylenediamine tetraacetate (EDTA) were purchased from Univar (Sydney, Australia) and PEO (Mw = 600,000) was purchased from Sigma–Aldrich (Milwaukee, WI). Yogurt samples containing L. delbrueckii subsp. bulgaricus and S. thermophilus were purchased from local stores. 2.2. Instrumentation CE measurements were performed on a 3D CE system (HewlettPackard, Waldbronn, Germany) equipped with a diode-array detector, an automatic injector, an autosampler, and a power supply. Data handling was performed by 3D CE Chemstation software (Hewlett-Packard, Waldbronn, Germany). Separations were carried out using fused-silica capillaries (Polymicro Technologies, Phoenix, AZ, USA) with an inner diameter of 100 ␮m and different effective lengths (16 and 47.5 cm). The detection wavelength at 214 nm was used as a guideline for the monitoring of both bacteria. The measured responses were from the combination of light scattering and absorbance. Separation was carried out in a positive mode (anode at the inlet and cathode at the outlet). A new capillary was washed with 1N NaOH (10 min), deionized water (15 min), 0.1N NaOH (10 min), deionized water (15 min) and the BGE (10 min). Between runs, the capillary was washed with 1N NaOH (5 min), deionized water (10 min), 0.1N NaOH (5 min), deionized water (10 min) and BGE (5 min). The between run washing was necessary to ensure the method reproducibility. 2.3. Methods 2.3.1. Bacterial isolation The bacteria were isolated from a commercial yogurt sample to ensure that the sample contained both L. delbrueckii subsp. bulgaricus and S. thermophilus as indicated on the label. Additionally, cells could be freshly prepared so that they were mostly live cells and their characteristics could be studied. The bacterial cells were isolated from the yogurt sample on deMan Rogosa Sharpe (MRS) agar (Laboratorious Britania, Argentina) containing 5% CaCO3 using the plate count method [21]. Briefly, 10 g of yogurt sample were added into 90 mL of 0.85% sterile saline solution. The diluted sample was mixed well with a vortex mixer. Ten fold serial dilutions of the sample were made. An appropriate diluted sample was then applied to MRS agar containing 5% CaCO3 . The plates were incubated at 37 ◦ C for 24–48 h. Colonies forming a clear zone around them were selected and re-streaked on MRS agar until a pure isolate was obtained. The bacterial cells were examined by Gram’s staining under microscopic examination (Nikon CoolScope, MO, USA) and by a catalase test, using 3% hydrogen peroxide pouring into their colonies, to confirm cell types. 2.3.2. Preparation of bacterial cell suspension Bacterial cell suspensions for CE analysis were prepared by cultivating each bacterium in 250-mL flasks containing 50-mL MRS broth and incubating at 37 ◦ C for 24–28 h. Bacterial cell pellets were harvested by centrifugation at 10,000 × g under 4 ◦ C for 20 min. They were washed by deionized water and collected by centrifugation. The washing procedure was repeated three times. The bacterial cells were suspended in the diluted TBE buffer (pH 8.4) containing appropriate amounts of PEO to give a cell concentration of 1.0 × 108 colony forming unit per mL (CFU/mL). Prior to CE injection, a brief vortex (1–2 min) was applied to the freshly prepared cell suspension to prevent cell aggregation. After washing,

712

O. Lim et al. / J. Chromatogr. B 877 (2009) 710–718

the bacterial suspensions (2.0 × 107 CFU/mL) were also measured for their UV–vis absorption from 200 to 700 nm using an UV–visible Perkin-Elmer spectrophotometer (Lambda EZ 201, Waltham, MA). For bacterial count, after ten fold serial dilutions of cell suspensions, an appropriated diluted cell suspension was applied to Lactobacillus bulgaricus (LB) agar for L. delbrueckii subsp. bulgaricus and to Streptococcus thermophilus (ST) agar for S. thermophilus. The plates were incubated at 37 ◦ C for 24–48 h. Colonies appeared on the agar were counted and expressed as CFU. In subsequent experiments, (i.e. optimization and method validation), cells at 28 and 24 h age were used for L. delbrueckii subsp. bulgaricus and S. thermophilus, respectively. They were mostly living cells since they were at initial stage of the stationary phase (i.e. when there is no net increase or decrease in cell number) or the end of log phase. 2.3.3. Yogurt sample preparation Yogurt samples were prepared by diluting 10 g of yogurt with 90 mL diluted TBE buffer (pH 8.4) containing 0.05% (v/v) PEO. The diluted yogurt was then mixed by vortex mixer for 1–2 min before CE analysis. The samples were also examined for the bacterial count by the plate count method as described in Section 2.3.2. 2.3.4. BGE preparation The initial BGE composition was prepared as described by Armstrong et al. [16]. Appropriate amounts of TRIS, boric acid and EDTA were dissolved in deionized water to obtain a TBE buffer (pH 8.4) containing 4.5 mM TRIS, 4.5 mM boric acid and 0.1 mM EDTA. The TBE buffer was diluted with deionized water (8:1) to obtain the diluted TBE buffer. A stock polymer solution was prepared by dissolving 0.2 g of PEO in 40 mL of the diluted TBE buffer, sonicated for 2–5 h at 45–50 ◦ C (Ultrasonic bath model T28, Elma, Germany) and stirred overnight at room temperature. The running BGE was prepared by diluting the stock polymer solution with the diluted TBE buffer to desired polymer concentrations (0–0.05%, v/v). Buffers, polymer solutions and BGE were freshly prepared daily. 2.3.5. Optimization of CZE condition Several factors affecting the separation of L. delbrueckii subsp. bulgaricus and S. thermophilus were investigated. These included PEO concentration in BGE (0–0.05%, v/v), capillary effective length (16 cm vs. 47.5 cm), temperature (25 ◦ C vs. 30 ◦ C) and voltage (10 kV vs. 20 kV). Sample injections were achieved using hydrodynamic injection or by the “LVSS” modified from Huang et al. [20]. For “LVSS”, (1) the sample was hydrodynamically injected into the capillary at 50 mbar for 50 and 25 s, respectively; (2) a reversed polarity voltage (−5 kV for 100–180 s) was employed to stack the bacteria and to remove sample matrices. When the whole capillary was filled with the BGE (i.e. 90–99% of the actual current was reached), the stacking process was stopped and (3) the BGE was switched to the inlet, and the polarity of the electrodes was reversed again for normal CZE separation. 2.3.6. Method validation Validation of the method in term of linearity, precision, recovery, limits of quantitation (LOQ) were evaluated. The known concentrations of each bacterium in these experiments were also verified by the plate count method as described in Section 2.3.2. Calibration curves of both bacteria were established by triplicate injecting each bacterial suspension at five different concentrations, 1.0 × 106 , 0.5 × 107 , 1.0 × 107 , 0.5 × 108 and 1.0 × 108 CFU/mL. LOQ were defined as the standard cell concentrations that gave signal to noise ratios (S/N) of 10. Precision was determined by injection, intra-day and inter-day precision by analyzing each bacterial suspension at a concentration of 1.0 × 107 CFU/mL and percent relative standard deviations (%RSDs) of migration time and peak area

were calculated. Injection precision was determined from eight injections of each bacterial suspension. Intra-day precision was performed by triplicate analyzing each bacterial suspension on the same day and each suspension was injected four times. Inter-day precision was performed by analyzing each bacterial suspension on five different days and each suspension was injected in triplicate. Recovery was performed by analyzing the bacteria in 3 yogurt samples (brands A, B and C) using the developed CE method in comparison with the plate count method. The plate count method is considered as a conventional method and commonly used for bacterial enumeration in food samples, thus it can be employed as a standard method and cell number obtained by this method is reliable and considered as 100% [1,21]. Percent recovery (%R) was calculated from (the number of bacteria obtained from the CE method/the number of bacteria obtained from the plate count method) × 100. 2.3.7. Statistical analysis Numbers of L. delbrueckii subsp. bulgaricus and S. thermophilus in yogurt samples obtained from the proposed CZE and the standard plate count methods were statistically analyzed by simple paired t-test (SPSS 13.0, Chicago, IL). The means were compared and were considered significantly different at the confidence level of 95% (P ≤ 0.05). 3. Results and discussion 3.1. Characterization and growth of L. delbrueckii subsp. bulgaricus and S. thermophilus isolated from the yogurt samples Isolation of bacteria from the yogurt samples showed two bacterial strains with different characteristics. One was rod and the other was cocci. Both bacteria were Gram positive, had similar morphology as those reported in literatures and gave negative results for the catalase test (Table 1 and Fig. 1). Base on their morphology and catalase test results and according to Bergey’s manual [22], these bacteria were L. delbrueckii subsp. bulgaricus and S. thermophilus. The number of peaks and peak shape are two main considering features for CE analysis of bacteria. In addition to CE conditions (e.g. BGE concentration and pH, capillary dimension, injection techniques, temperature and separating voltage), bacterial age, growth condition and sample pre-treatment highly influence their separation. In the present study, the bacterial concentrations and incubation times were investigated to standardize the analysis procedure. Plots of colony forming unit per mL (CFU/mL) for L. delbrueckii subsp. bulgaricus and S. thermophilus against time were established (Fig. 1). Bacterial cells were then taken from the stationary phase (i.e. when the number of viable cells was constant) after incubation times of 28 and 24 h for L. delbrueckii subsp. bulgaricus and S. thermophilus, respectively, to avoid the physiological changes of cells in the growth phase. 3.2. Bacterial sample preparation Sample preparation of bacteria for CE analysis is important since bacteria have high tendency to aggregate in an aqueous matrix, especially under acidic condition [2]. This self-aggregation can be weak association, strong electrostatic or even covalent bond, which cannot be easily reversed. The aggregation results in cluster formations that can complicate the CE analysis of bacteria [13–14]. Moreover, the charge-to-size ratio of bacteria can vary due to cluster formations, which eventually affect electrophoretic mobilities of bacterial cells and yield spurious peaks in electropherograms. Additionally, bacterial cells can settle and aggregate easily due to

O. Lim et al. / J. Chromatogr. B 877 (2009) 710–718

713

Table 1 Characteristics of yogurt starter cultures. Bacteria

L. delbrueckii subsp. bulgaricus S. thermophilus

Characteristics Shape

Size (␮m)

Grama

Surface chargea

Catalase test

Rod Cocci in pair to long chain

Diameter: 0.5–0.8; length: 2–9 Diameter: 0.7–0.9

+ +

− −

No bubble formation No bubble formation

+: Positive, −: Negative.

the gravity force. This also causes varied electrophoretic mobilities. Most publications, therefore, recommended the use of low energy (vortex) or high energy (sonication) processes to disperse the clusters of cells, which enabled the observation of a single peak with more symmetric peak shape during the electrophoresis of bacteria [11–13,16–17]. Initially, sonication (3 min) was applied for the sample pre-treatment of L. delbrueckii subsp. bulgaricus and S. thermophilus. However, sonication did not benefit the CE analysis of both microorganisms (electropherograms not shown). The sonicated samples even gave more numerous peaks than samples without sonication. In addition, these peaks were of low sensitivity with noisy signals due to cell lyses, which were shown by a microscope. Both L. delbrueckii subsp. bulgaricus and S. thermophilus might have weak association strengths so the cells were damaged and broken during sonication, which could cause poor precision for quantitative analysis. Thus, the cell suspensions were briefly vortexed (1–2 min) to minimize cell aggregation and to avoid cell disruption. Under this mild condition, intact cells were obtained (as observed under a microscope), which would enhance the separation reproducibility. 3.3. Optimization of CZE condition BGE compositions and several instrumental parameters (e.g. injection technique, capillary length, separating voltage and temperature) played important roles for the separation of L. delbrueckii

subsp. bulgaricus and S. thermophilus. Initially, the CE optimization (Sections 3.3.1–3.3.3) was separately performed for each bacterium to avoid confusion due to spurious peaks that are normally obtained from CE analysis of bacteria. Finally, the optimization was carried out in a mixture of both bacteria (Section 3.3.4). 3.3.1. Effect of PEO concentration in BGE The initial condition for the CE separation of L. delbrueckii subsp. bulgaricus and S. thermophilus was studied using the diluted TBE buffer containing various amounts of PEO (0, 0.0125, 0.025 and 0.05%, v/v) using a capillary with an effective length of 16 cm, 100 ␮m i.d., a separating voltage of 10 kV, temperature of 25 ◦ C and hydrodynamic injection of 34 mbar for 10 s. Under this condition, both bacteria were swept out from the capillary with one major peak and spurious small peaks within 3.60 min (electropherograms not shown). The migration times of the major peaks increased from 1.70 to 2.57 min for L. delbrueckii subsp. bulgaricus and from 1.66 to 3.60 min for S. thermophilus when the amount of PEO was increased from 0 to 0.05% (v/v). In the absence of PEO, separation of the bacteria could not be achieved because the electroosmotic flow (EOF) was too fast and the bacteria co-migrated with the EOF as broad peaks. Increasing amounts of PEO decreased the EOF and the mobilities of the bacteria resulting in the longer migration times. PEO is a polymer that acts as a focusing agent and helps diminishing or reducing the EOF and wall adsorption. Adding PEO into the BGE decreased the EOF velocity, thus enabled the

Fig. 1. (a) Microscopic examination and (b) growth curves of the bacteria isolated from yogurt samples.

714

O. Lim et al. / J. Chromatogr. B 877 (2009) 710–718

Fig. 2. Effects of injection techniques on the CZE of bacterial suspension of L. delbrueckii subsp. bulgaricus and S. thermophilus (1.0 × 108 CFU/mL). Conditions: TBE buffer (pH 8.4) containing 0.025% (v/v) PEO; capillary, 24.5 cm full length (8.5 cm to detector), 100 ␮m ID; injection, (a) hydrodynamic injection (34 mbar × 10 s) and (b) “LVSS” with polarity switching, 50 mbar × 50 s, and 25 s followed by −5 kV × 120 s; voltage, 10 kV; temperature, 25 ◦ C; detection by UV absorbance at 214 nm.

differentiation of the EOF mobility and electrophoretic mobilities of bacteria and increased the separation efficiency. The mechanism of PEO on the CE separation is not fully understood, until recently when a charge-coupled device (CCD) camera coupled with laser

induced fluorescence has been employed to obtain the moving pictures of the electrophoresis process. The proposed mechanisms included (1) the field-induced aggregation, (2) the hairy particle and (3) the shape-induced differential mobility models [15,23].

Fig. 3. Effects of temperature on the CZE of bacterial suspension of L. delbrueckii subsp. bulgaricus and S. thermophilus (1.0 × 108 CFU/mL). Conditions: TBE buffer (pH 8.4) containing 0.025% (v/v) PEO; capillary, 56 cm full length (8.5 cm to detector), 100 ␮m ID; injection, “LVSS” with polarity switching, 50 mbar × 3 s (twice) followed by −5 kV × 120 s; voltage, 20 kV; temperature, (a) 25 ◦ C and (b) 30 ◦ C; detection by UV absorbance at 214 nm.

O. Lim et al. / J. Chromatogr. B 877 (2009) 710–718

715

Fig. 4. Effects of amounts of PEO on the CZE of a mixture of L. delbrueckii subsp. bulgaricus and S. thermophilus (1.0 × 108 CFU/mL). Conditions: TBE buffer (pH 8.4) containing (a and b) 0.025%, (c) 0.0375% (v/v) and (d) 0.05% (v/v) PEO; capillary, 56 cm full length (8.5 cm to detector), 100 ␮m ID; injection, “LVSS” with polarity switching, (a) 50 mbar × 3 s (twice) followed by −5 kV × 120 s and (b–d) 50 mbar × 3 s followed by −5 kV × 120 s; voltage, 20 kV; temperature, 25 ◦ C; detection by UV absorbance at 214 nm.

In these experiments, 0.025% (v/v) PEO was selected to allow sufficient migration time differences (ca. 0.5 min) between the two bacteria. Additionally, the major peaks of both bacteria were more separated from other small peaks in the diluted TBE buffer containing 0.025% (v/v) PEO. 3.3.2. Injection technique Sample loading can play significant roles on separation efficiency and sensitivity. Two injection techniques, hydrodynamic injection (34 mbar × 10 s) and “LVSS” were attempted for the separation of L. delbrueckii subsp. bulgaricus and S. thermophilus. Results revealed that hydrodynamic injection provided one major peak with numerous small peaks whereas “LVSS” gave eight and five major peaks for L. delbrueckii subsp. bulgaricus and S. thermophilus, respectively (Fig. 2). The stacking time of 100 s gave signals of low sensitivity with baseline drift as well as several small peaks, whereas 150 and 180 s overfilled the capillary. The optimized stacking time was at 120 s, which greatly reduced the number of peaks with improved sensitivity.

The hydrodynamic injection in “LVSS” was divided into two steps since we experienced the precision problem when one long injection time was made. In this work, the “LVSS” could be achieved despite of little/no differences on the field strength between the sample zone and the BGE. The field strength of the sample zone and the BGE was similar and we reasoned that “the stacking” of the bacteria could occur by a different mechanism. L. delbrueckii subsp. bulgaricus and S. thermophilus are Gram-positive bacteria, which generally consist of a thick peptidoglycan layer without a lipopolysaccharide layer. These bacteria possess negative charges at pH > 5.0 (the investigated pH was 8.4), thus they moved against the EOF direction toward the anode at the inlet. When a negative voltage was applied at the inlet, immediately after loading the bacteria into the capillary, the bacteria started to move toward the anode at the outlet. At this step, the sample was compacted into a smaller zone and then was separated by CZE using positive polarity voltage. This process served as a pre-concentration procedure, which helped improving the sensitivity and reducing the number of peaks and baseline noise in CE separation.

716

O. Lim et al. / J. Chromatogr. B 877 (2009) 710–718

Table 2 Linearity, limits of quantitation data (n = 3).

Linear regression LOQ (CFU/mL)

L. delbrueckii subsp. bulgaricus

S. thermophilus

y = 853.6x − 4786.3 (r = 0.9942) 1.0 × 106 (2.4)

y = 678.4x − 3723.4 (r2 = 0.9905) 1.0 × 106 (1.0)

2

x = log CFU/mL, number in parentheses represents %RSDs.

3.3.3. Capillary length, voltage and temperature Although the “LVSS” injection greatly enhanced the CE separation of L. delbrueckii subsp. bulgaricus and S. thermophilus, a number of peaks (up to 8, in case of L. delbrueckii subsp. bulgaricus) were still observed. Our ultimate goal was to obtain a single peak for each microbe, therefore further CE optimization was carried out by reducing the hydrodynamic injection time to 3 s and by varying the capillary effective length (i.e. 16 cm vs. 47.5 cm) and temperature (i.e. 25 and 30 ◦ C). The shorter capillary and lower voltage was not sufficient for the baseline separation of both bacteria and numerous peaks were obtained. The use of a longer capillary caused an increase of migration times, therefore a higher voltage (i.e. 20 kV) was employed for the capillary with an effective length of 47.5 cm. Reducing the hydrodynamic injection time, increasing the capillary effective length by a factor of about 3 and doubling the voltage, significantly reduced the number of peaks down to a single peak for both bacteria with an expense of longer migration times (Fig. 3a). In order to keep the minimum migration time, increasing of the capillary temperature was attempted. Effects of temperature on the separation of L. delbrueckii subsp. bulgaricus and S. thermophilus were investigated at 25 and 30 ◦ C (Fig. 3). At 25 ◦ C, the bacteria were separated as a single peak with migration times of 4.20 and 4.99 min for L. delbrueckii subsp. bulgaricus and S. thermophilus, respectively (Fig. 3a). However, at a higher temperature (30 ◦ C) the system was unstable because of the Joule heating, which caused bubble formation and might disrupt the analysis. Major peaks at the migration times of 2.67 and 3.07 min for L. delbrueckii subsp. bulgaricus and S. thermophilus, respectively, and other small peaks were observed at 30 ◦ C (Fig. 3b). Increasing of temperature enhanced the EOF due to the decrease of the BGE viscosity, which resulted in the shorter migration times and might cause overlapped peaks when a mixture of the bacteria was analyzed. From the results, a lower temperature (25 ◦ C) was preferred since each bacterium could be separated as a single peak with high separation efficiency. 3.3.4. Final optimization The CE condition in Fig. 3a was efficient for the analysis of L. delbrueckii subsp. bulgaricus and S. thermophilus, when they existed separately. However, baseline separation could not be achieved when a mixture of the bacteria was analyzed (Fig. 4a). There were too many bacteria (2 × 108 CFU/mL) in the capillary since now there were present as a mixture, which caused overlapped peaks. Diluting the sample did not improve the separation due to the poor sensitivity. Thus, the “LVSS” was further modified by reducing the sample loading of 50 mbar for 3 s (once) followed by applying a negative polarity voltage (−5 kV for 120 s). This injection was no longer “LVSS”, however the bacteria could be detected as two major peaks with good sensitivity (Fig. 4b). The bacteria could be still compressed into sharp zones by the focusing effect of the bacteria themselves by the three mechanisms described by Armstrong et al. [15]. Instead of a group of individual cell, the bacteria then migrated as a single charged entity, which moved toward the cathode at a velocity governed by its size-to-charge ratios. The stacking process (i.e. reversed polarity voltage) due to the highly negative charges on the cell surface also assisted the complete focusing of the bacteria. Additionally, the amount of PEO (i.e. 0.025, 0.0375 and 0.05%, v/v) in the BGE was re-investigated to ensure the complete

separation of both bacteria (Fig. 4b–d). At 0.05% (v/v) PEO, baseline separation of the bacteria was obtained with a resolution (Rs ) of 3.2 and migration times of 6.8 and 7.2 min for L. delbrueckii subsp. bulgaricus and S. thermophilus, respectively (Fig. 4d). The effective mobility of each bacterium was −0.25 × 10−4 and −0.40 × 10−4 cm2 /V s, respectively. The negative mobility indicated that they migrated against the EOF since both bacteria possess negative charges at their surface and migrated toward the anode. However, they could eventually migrate to the detector at the cathode by the effect of the EOF. Other small peaks were not from cell lyses because normal cells were observed under a microscope. These peaks might come from the BGE and yogurt matrices (Fig. 5a and b). The final optimized condition for the separation of a mixture of L. delbrueckii subsp. bulgaricus and S. thermophilus was in the BGE containing diluted TBE (pH 8.4) and 0.05% (v/v) PEO, using the temperature of 25 ◦ C, voltage of 20 kV and a capillary of 100 ␮m (i.d.) × 47.5 cm (effective length) and hydrodynamic injection of 50 mbar × 3 s followed by −5 kV × 120 s (Fig. 4d). 3.4. Method validation The CE condition in Fig. 4d was selected as the optimum condition to ensure the separation of L. delbrueckii subsp. bulgaricus and S. thermophilus in yogurt samples. The condition was evaluated for the method linearity, LOQ, precision and recovery. Linearity data of both bacteria in a range of 1.0 × 106 , 0.5 × 107 , 1.0 × 107 , 0.5 × 108 Table 3 Precision data. No.

L. delbrueckii subsp. bulgaricus

S. thermophilus

tm (min)

Peak area

tm (min)

Peak area

Injection precision (n = 8) 1 5.1 2 4.9 3 4.9 4 4.9 5 5.0 6 4.8 7 4.9 8 4.9

1157 1137 1203 1246 1099 1235 1335 1235

5.4 5.6 5.6 5.6 5.9 5.3 5.7 5.9

1237 1033 1046 1266 1076 1126 1033 1127

Average %RSD

1202 6.2

5.6 3.9

1118 8.1

Intra-day precision (n = 3) 1 4.9 2 4.9 3 4.9

1154 1035 1252

5.5 5.7 5.6

1248 1239 1253

Average %RSD

1148 9.3

5.6 1.8

1247 0.6

Inter-day precision (n = 5) 1 4.9 2 4.6 3 4.9 4 4.9 5 5.0

1146 1099 1124 1139 1288

5.4 5.5 5.6 5.4 5.8

1107 1045 1103 1089 999

Average %RSD

1184 7.6

5.6 3.6

1064 5.3

4.9 1.9

4.9 0.0

4.9 1.2

Precision data is represented as %RSD and was performed from the concentration of 1.0 × 107 CFU/mL for both bacteria.

O. Lim et al. / J. Chromatogr. B 877 (2009) 710–718

717

Fig. 5. Typical electropherograms of (a) BGE, (b) filtered yogurt through a 0.2 ␮m membrane filter, (c) L. delbrueckii subsp. bulgaricus and S. thermophilus in un-spiked yogurt brand C and (d) L. delbrueckii subsp. bulgaricus and S. thermophilus in spiked yogurt brand C. Conditions: see Fig. 4d.

and 1.0 × 108 CFU/mL was shown in Table 2. Results revealed good linearity for both bacteria with r2 > 0.99. LOQ was 1.0 × 106 CFU/mL with %RSD of less than 2.4% (Table 2). Injection precision shows %RSDs of less than 3.9% and 8.1% for migration time and peak area, respectively (Table 3). Intra- and inter-day precision gave %RSDs of less than 3.6% and 9.3% for migration time and peak area, respectively (Table 3). Washing the capillary between run was important for the method precision. Percent recoveries, determined from the comparison the results from the proposed method with the plate count method, were in a range of 94.8–98.6% for L. delbrueckii subsp. bulgaricus and 94.3–102.1% for S. thermophilus (Table 4). These recoveries were in acceptable ranges (where the cell num-

bers obtained from plate count method was considered as 100% recovery). 3.5. Applications The proposed CE method was applied for the enumeration of L. delbrueckii subsp. bulgaricus and S. thermophilus in three brands of yogurt. For analysis of yogurt samples, the yogurt samples contained live cells because they were well preserved at 4 ◦ C and were used for the analysis before their expired date. Therefore, the electrophoresis results represented mostly living cells in the freshly prepared samples (Table 4).

Table 4 Assay of L. delbrueckii subsp. bulgaricus and S. thermophilus in yogurt samples (n = 3)* . Brand

L. delbrueckii subsp. bulgaricus CE method (%RSD)

S. thermophilus

Plate count method (%RSD)

A

8

1.3 × 10 1.3 × 108 1.1 × 108

8

1.3 × 10 1.4 × 108 1.2 × 108

Average %RSD

1.2 × 108 9.3

B

%R

CE method (%RSD)

Plate count method (%RSD)

%R

100.0 92.9 91.7

8

1.5 × 10 1.6 × 108 1.7 × 108

8

1.5 × 10 1.6 × 108 1.6 × 108

100.0 100.0 106.3

1.3 × 108 7.7

94.8 4.7

1.6 × 108 6.2

1.6 × 108 3.7

102.1 3.5

1.8 × 109 1.4 × 109 1.6 × 109

1.9 × 109 1.5 × 109 1.6 × 109

94.7 93.3 100.0

2.4 × 109 1.9 × 109 2.3 × 109

2.5 × 109 2.0 × 109 2.5 × 109

96.0 95.0 92.0

1.6 × 109 12.5

1.7 × 109 12.5

96.0 3.7

2.2 × 109 12.0

2.3 × 109 12.4

94.3 2.2

C

4.4 × 108 4.6 × 108 4.6 × 108

4.5 × 108 4.6 × 108 4.7 × 108

97.8 100.0 97.9

6.3 × 108 5.5 × 108 5.9 × 108

6.2 × 108 5.5 × 108 5.9 × 108

101.6 100.0 100.0

Average %RSD

4.5 × 108 2.6

4.6 × 108 2.2

98.6 1.3

5.9 × 108 6.8

5.9 × 108 6.0

100.5 0.9

Average %RSD

*

Assay data is presented in CFU/g of each bacterium in yogurt samples.

718

O. Lim et al. / J. Chromatogr. B 877 (2009) 710–718

A major problem encountered when the proposed method was applied for the analysis of L. delbrueckii subsp. bulgaricus and S. thermophilus in yogurt was the shift of the migration time of the bacteria in the samples comparing to that of the standard suspension. This shift was due to changes in ionic strength caused by sample matrices. Moreover, the shift was varied among brands of yogurts since each brand might contain different yogurt compositions (e.g. lactose, lactic acid, minerals, vitamins, emulsifiers/stabilizers, etc.) and are usually not disclosed to consumers. For confirmation, the bacterial peaks were verified by spiking the standard bacterial suspension (5.0 × 105 CFU/mL) into the yogurt samples. The spiked samples gave larger peak areas and heights than the un-spiked samples and typical electropherograms of the un-spiked and spiked samples are shown in Fig. 5c and d, respectively. The migration time shift of bacterial peaks did not affect the bacterial enumeration in yogurt by CE. Results from the enumeration of L. delbrueckii subsp. bulgaricus and S. thermophilus in yogurt samples by the developed method did not significantly differ (P > 0.05) from those obtained from the conventional plate count method (Table 4). 4. Conclusion This work firstly illustrated the potential of CE for the analysis of L. delbrueckii subsp. bulgaricus and S. thermophilus in yogurt samples. Previously, only products from the fermentation process were determined, for example, volatile and non-volatile products of milk fermentation process [24], furosine in dairy products [25], citrate and inorganic phosphate in milk [26] and diacetyl in yogurt [27]. The BGE containing PEO was essential for the separation of both bacteria, other key factors such as the use of low energy process (i.e. vortex) for the sample preparation and the injection technique (i.e. 50 mbar × 3 s followed by −5 kV × 120 s) for sample loading were also important. This injection was not true “LVSS”, but could focus the bacteria into sharp zones, which were eventually detected as separated peaks. The focusing effect occurred from unique mechanisms due to the highly negative charges on the cell surfaces. The proposed CE method can serve as a novel tool in yogurt science and technology for the quality control of yogurt production because of its high efficiency, simplicity and speed.

Acknowledgement The authors are thankful to the Faculty of Science, King Mongkut’s University of Technology Thonburi for the financial support in this work. References [1] The United State Pharmacopeia (USP 28) Asian ed., Convention Inc., Washington, DC, 2004. [2] M.A. Rodriguez, A.W. Lantz, D.W. Armstrong, Anal. Chem. 78 (2006) 4759. [3] M.A. Rodriguez, D.W. Armstrong, J. Chromatogr. B 800 (2004) 7. [4] V. Garcia-Canas, A. Cifuentes, Electrophoresis 28 (2007) 4013. [5] B. Buszewski, E. Klodzinska, H. Dahm, H. Roucki, J. Szeliga, M. Jackowski, Biomed. Chromatogr. 21 (2007) 116. [6] M.J. Desai, D.W. Armstrong, Microbiol. Mol. Biol. Rev. 67 (2003) 38. [7] B. Palenzuela, B.M. Simonet, R.M. Garcia, A. Rios, M. Valcarcel, Anal. Chem. 76 (2004) 3012. [8] S. Hjerten, K. Elenbring, F. Kilar, J.L. Llao, A.J. Chen, C.J. Siebert, M.D. Zhu, J. Chromatogr. 403 (1987) 47. [9] R.C. Ebersole, R.M. McCormick, Biotechnology (NY) 11 (1993) 1278. [10] A. Pfetsch, T. Welsh, Fresenius J. Anal. Chem. 359 (1997) 198. [11] D.W. Armstrong, G. Schulte, J.M. Scneiderheinze, D.J. Westenberg, Anal. Chem. 71 (1999) 5465. [12] D.W. Armstrong, J.M. Scneiderheinze, Anal. Chem. 72 (2000) 4474. [13] J.M. Scneiderheinze, D.W. Armstrong, G. Schulte, D.J. Westenberg, FEMS Microbiol. Lett. 189 (2000) 39. [14] D.W. Armstrong, L. He, Anal. Chem. 73 (2001) 4551. [15] D.W. Armstrong, M. Girod, L. He, M.A. Rodriguez, W. Wei, J. Zheng, E.S. Yeung, Anal. Chem. 74 (2002) 5523. [16] D.W. Armstrong, J.M. Scneiderheinze, J.P. Kullman, L. He, FEMS Microbiol. Lett. 194 (2001) 33. [17] L.J. Yu, S.F.Y. Li, Chromatographia 62 (2005) 401. [18] E. Klodzinska, H. Dahm, H. Rozycki, J. Szeliga, M. Jackowski, B. Buszewski, J. Sep. Sci. 29 (2006) 1180. [19] A.Y. Tamine, R.K. Robinson, Yogurt Science and Technology, second ed., Woodhead Publishing Ltd., Cambridge, 1999. [20] H.Y. Huang, C.W. Chiu, S.L. Sue, C.F. Chen, J. Chromatogr. A 995 (2003) 29. [21] K.M.J. Swanson, F.F. Busta, E.H. Peterson, M.G. Johnson, in: C. Vanderzant, D.F. Splittstoesser (Eds.), Compendium of Methods for the Microbiological Examination of Foods, third ed., The American Public Health Association, Washington, DC, 1992, p. 75. [22] P.H.A. Sneath, N.S. Mair, M.E. Sharpe, J. Holt, Bergey’s Manual of Systematic Bacteriology, vol. 2, Williams and Wilkins, Baltimore, 1986. [23] B. Khusid, A. Acrivos, Phys. Rev. 54 (1996) 5428. [24] M. Ligor, R. Jarmalaviciene, M. Szumski, A. Maruska, B. Buszewski, J. Sep. Sci. 31 (2008) 2707. [25] B. Vallejo-Cordoba, M.A. Mazorra-Manzano, A.F. Gonzalez-Cordova, J. Agric. Food Chem. 52 (2004) 5787. [26] J.M. Izco, M. Tormo, A. Harris, P.S. Tong, R. Jimenez-Flores, J. Dairy Sci. 86 (2003) 86. [27] E.J. Guerra Hernandez, R.G. Estepa, I.R. Rivas, Food Chem. 53 (1995) 315.