Determination of Carbohydrates, Sugar Alcohols, and Glycols in Cell Cultures and Fermentation Broths Using High-Performance Anion-Exchange Chromatography with Pulsed Amperometric Detection

Analytical Biochemistry 283, 192–199 (2000) doi:10.1006/abio.2000.4653, available online at http://www.idealibrary.com on

Determination of Carbohydrates, Sugar Alcohols, and Glycols in Cell Cultures and Fermentation Broths Using High-Performance Anion-Exchange Chromatography with Pulsed Amperometric Detection Valoran P. Hanko 1 and Jeffrey S. Rohrer Dionex Corporation, 500 Mercury Drive, Sunnyvale, California 94088-3603

Received December 10, 1999

Cell cultures and fermentation broths are complex mixtures of organic and inorganic compounds. Many of these compounds are synthesized or metabolized by microorganisms, and their concentrations can impact the yields of desired products. Carbohydrates serve as carbon sources for many microorganisms, while sugar alcohols (alditols), glycols (glycerol), and alcohols (methanol and ethanol) are metabolic products. We used high-performance anion-exchange chromatography with pulsed amperometric detection (HPAE-PAD) to simultaneously analyze for carbohydrates, alditols, and glycerol in growing yeast (Saccharomyces cerevisiae) cultures and their final fermentation broths. Both cultures were grown on complex undefined media, aliquots centrifuged to remove particulates, and the supernatants diluted and directly injected for analysis. Pulsed amperometry allowed a direct detection of the carbohydrates, alditols, and glycols present in the cultures and fermentation broths with very little interference from other matrix components. The broad linear range of three to four orders of magnitude allowed samples to be analyzed without multiple dilutions. Peak area RSDs were 2–7% for 2,3-butanediol, ethanol, glycerol, erythritol, rhamnose, arabitol, sorbitol, galactitol, mannitol, arabinose, glucose, galactose, lactose, ribose, raffinose, and maltose spiked into a heat-inactivated yeast culture broth supernatant that was analyzed repetitively for 48 h. This method is useful for directly monitoring culture changes during fermentation. The carbohydrates in yeast cultures were monitored over 1 day. A yeast culture with medium consisting primarily of glucose and trace levels of trehalose and arabinose showed a

1 To whom correspondence should be addressed. Fax: (408) 7372470. E-mail: [email protected].

192

drop in sugar concentration over time and an increase in glycerol. Yeast growing on a modified culture medium consisting of multiple carbohydrates and alditols showed preference for specific carbon sources and showed the ability to regulate pathways leading to catalysis of alternative carbon sources. © 2000 Academic Press

Key Words: fermentation broth; Saccharomyces cerevisiae; yeast; carbohydrates; alditols; sugar alcohols; glycerol; ethanol; methanol; pulsed amperometric detection; anion-exchange chromatography; HPLC; HPAE; PAD; IPAD.

Fermentation broths are used in the manufacture of biotherapeutics and many other biological materials produced using recombinant genetic technology, as well as for the production of methanol and ethanol as alternative energy sources to fossil fuels. Fermentation broths are also used for the manufacture of alcoholic beverages. Carbon sources and metabolic by-products of fermentation processes can impact the yield or quality of the desired products. It is desirable to characterize these culture and fermentation broth ingredients to optimize media formulation development, manufacturing process performance with nutrient supplementation, and endpoint definition. Fermentation broths and cell cultures are complex mixtures of nutrients, waste products, cells, cell debris, and desired products. Carbohydrates (glucose, lactose, sucrose, maltose, etc.) are the major carbon sources used and are essential for cell growth and product synthesis. Alcohols (ethanol, methanol, sugar alcohols, etc.), glycols (glycerol), and organic acids (acetate, lactate, formate, etc.) are metabolic by-products that often reduce yields. Many of these compounds are nonchro0003-2697/00 $35.00 Copyright © 2000 by Academic Press All rights of reproduction in any form reserved.

CHROMATOGRAPHIC DETERMINATION OF CARBOHYDRATES

mophoric and cannot be detected by absorbance. Carbohydrates, alditols, glycols, alcohols, amines, and sulfur-containing compounds can be oxidized and, therefore, detected by amperometry. This detection method is specific for analytes that can be oxidized at a selected potential, leaving all other compounds undetected. HPAE-PAD 2 is a widely used technique for the chromatography of many carbohydrate-containing samples (1). PAD enables the direct sensitive detection of carbohydrates, glycols, and alditols at high pH following HPAE. HPAE is capable of separating complex mixtures of carbohydrates. For complex samples such as fermentation broths and cultures, the high resolving power of HPAE and the specificity of PAD allow the simultaneous determination of carbohydrates, glycols, alditols, and other alcohols such as ethanol and methanol, with little interference from other broth ingredients (2– 4). Although biosensor and flow-injection analyzer-based techniques are commonly used to analyze fermentation broths and cell cultures, these methods cannot simultaneously determine multiple compounds (5, 6). Refractive index detection has been used with HPLC, but is limited by poor sensitivity and specificity (7, 8). Postcolumn derivatization with UV-Vis detection is complicated by the additional reaction chemistry and poor sensitivity (9, 10). Using HPAE-PAD, a large number of carbohydrates can be simultaneously monitored in fermentation broths and cell cultures. This paper describes the determination of sugars, alditols, and glycols in a Saccharomyces cerevisiae (S. cerevisiae) cell culture and their final fermentation broths. These cultures used yeast extract-peptone-dextrose (YPD) broth, which is a common yeast culture medium. This culture medium is complex and contains undefined ingredients, and thus are a great challenge for most separation and detection technologies. This formulation contains inorganic and organic anionic ingredients that have been analyzed using anion-exchange chromatography with suppressed conductivity detection (11, 12). In the methods outlined in this paper, the selectivities of two anion-exchange columns (CarboPac PA1 and CarboPac MA1) are compared for the determination of carbohydrate, alcohol, and glycol ingredients in cell cultures and fermentation broths using pulsed amperometric detection. Detection limits, linearity, and precision are reported for the CarboPac MA1 column. The MA1 method was used to determine the changes in concentrations of carbohydrates, alditols, and glycols during a yeast culture incubation period. 2

Abbreviations used: HPAE-PAD, high-performance anion-exchange chromatography with pulsed amperometric detection; YPD, yeast extract-peptone-dextrose; LOD, lower limit of detection; MDL, method detection limits; LB, Luria-Bertani.

193

MATERIALS AND METHODS

Materials. All HPAE chromatography eluents and standards were prepared using 18 megaohm-cm-deionized water, free of electrochemically active impurities. The 50% (w/w) sodium hydroxide solution (low carbonate) used to make chromatography eluent was purchased from Fisher Scientific (Pittsburgh, PA) or J.T. Baker Chemical Co. (Phillipsburg, NJ). D-Arabinose, D-cellobiose, ␣-lactose, and D-xylose were purchased from Sigma Chemical Co. (St. Louis, MO). L-Arabitol, maltitol, and maltotriose were purchased from Aldrich Chemical Co. (Milwaukee, WI). Mannitol was purchased from J.T. Baker Chemical Co. Sorbitol was purchased from Eastman Chemical Co. (Rochester, NY). Sucrose was purchased from Fisher Scientific. All other carbohydrates and alditols were purchased from Pfanstiehl Laboratories, Inc. (Waukengan, IL). Ethanol, methanol, and glycerol were purchased from EM Science (Gibbstown, NJ). Bacto YPD broth, Bacto yeast extract, Bacto peptone, and LB broth were purchased from DIFCO Laboratories (Detroit, MI). Yeast (S. cerevisiae, Baker’s yeast type II) was purchased from Sigma Chemical Co. Preparation of standards. Solid standards were maintained desiccated and under vacuum prior to use. Each standard was dissolved in purified water to 10 mg/mL concentrations, correcting for mass of known water content or counterion, if specified by the manufacturer. The standards were combined and serially diluted in purified water to yield concentrations ranging from 1 mg/mL to 0.04 ␮g/mL (except methanol and ethanol). Methanol and ethanol were added at concentrations 100-times higher than those of the other standards. All solutions were maintained frozen at ⫺20°C until needed. Cell culture and fermentation broth analysis. The Bacto YPD broth used in yeast culture consisted of 2 g Bacto yeast extract, 4 g Bacto peptone, 4 g dextrose (glucose) per 10 g. The YPD broth (10 g) was dissolved in 200 mL filter-sterilized water to a concentration of 50 mg/mL. Dried yeast (1 g, S. cerevisiae, Bakers yeast type II) was dissolved in the YPD broth to initiate the yeast culture. For a study on yeast carbon source preferences, a modified multiple carbohydrate medium was prepared: 2 g of Bacto yeast extract, 4 g Bacto peptone, and 4 g of a mixture of carbohydrates (0.4 g each of glucose, sucrose, maltose, lactose, galactose, sorbitol, ribose, arabinose, rhamnose, and raffinose) were dissolved in 200 mL sterile water, and the culture was initiated with 1 g dried yeast. The aerobic yeast cultures were incubated in a 37°C shaking water bath (500 – 600 rpm) for 24 h using a vented Erlenmeyer flask. Aliquots were removed at 0, 0.5, 1, 2, 3, 4, 5, 6, 7, and 24 h and immediately placed on ice. Aliquots were centrifuged at 14,000g for 10 min

194

HANKO AND ROHRER

and the supernatant was diluted 100-fold in water for analysis. Chromatography system. The chromatography system used for this work was a Dionex (Sunnyvale, CA) DX-500 system. The system consisted of an ED40 electrochemical detector with a gold working electrode run in the integrated amperometry mode, a GP40 gradient pump (standard bore, PEEK) with degas option, a LC30 chromatography oven, and an AS3500 autosampler (Thermo Separation Products, Fremont, CA) with a stainless-steel injection needle and 100-␮L injection loop. PeakNet chromatography software (Dionex) was used for system control and data analysis. The two anion-exchange columns used in this work were the CarboPac PA1 (Dionex, 4 ⫻ 250 mm) with guard (4 ⫻ 50 mm), and the CarboPac MA1 (Dionex, 4 ⫻ 250 mm) with guard (4 ⫻ 50 mm). A BorateTrap column (Dionex) was placed between the pump and the injection valve to remove any trace borate ions that may be present in the eluent. Borate ions form complexes with some carbohydrates and alditols, causing peak tailing (13). The columns were installed in the LC30, which was maintained at 30°C. The ED40 waveform was ⫹0.05 V for 0.00 to 0.40 s, then ⫹0.75 V from 0.41 to 0.60 s, followed by ⫺0.15 V from 0.61 to 1.00 s (end of cycle). Integration occurred between 0.20 and 0.40 s. Chromatography. New water sources were prescreened for eluent and sample-diluent suitability by injecting 10-␮L volumes and measuring background peaks. Any repairs or routine maintenance to water purification systems constituted a reason to requalify the water source, as some filtration devices contain nonconductive surfactants that are electrochemically active (e.g., glycerol) and contribute to large background responses (⬎30 nC). All eluents were kept blanketed under 34 –55 kPa (5– 8 psi) of helium at all times to reduce carbonate buildup and biological contamination. One hundred millimolar NaOH and 250 mM NaOH eluents were prepared for CarboPac PA1 chromatography by dilution of 50% NaOH. The CarboPac PA1 column was eluted with 16 mM sodium hydroxide for 16 min. From 60 to 70 min, the column was washed with 250 mM sodium hydroxide. From 70 to 90 min, the column was reequilibrated to the initial eluent conditions in preparation for the next injection. The eluent was pumped at a flow rate of 1.0 mL/min. Hydroxide eluent (480 mM) for CarboPac MA1 chromatography was made by diluting 50 mL 50% sodium hydroxide with 1950 mL water. The flow rate for these separations was 0.4 mL/min and the run times were 60 –70 min. Analyte quantification and method testing. Analytes were identified in samples by comparison to retention times of standards. These analytes were mea-

FIG. 1. Common carbohydrate, alditol, alcohol, and glycol standards found in cell cultures and fermentation broths, separated on the CarboPac MA1 analytical and guard column with pulsed amperometry (gold electrode). The eluent was 480 mM NaOH and the flow rate was 0.4 mL/min. All standards were at 10 ␮g/mL concentration, except methanol (73 mg/mL) and ethanol (3 mg/mL); 10-␮L injection. Peak identities: 1, butanediol (2,3-); 2, ethanol; 3, methanol; 4, glycerol; 5, erythritol; 6, rhamnose; 7, arabitol; 8, sorbitol; 9, galactitol; 10, mannitol; 11, arabinose; 12, glucose; 13, galactose; 14, lactose; 15, ribose; 16, sucrose; 17, raffinose; 18, maltose.

sured by comparison to calibration curves prepared with known quantities of standards. Calibration curves, prepared for each analyte, were constructed by plotting peak area vs amount injected. Upper limits of linearity were estimated for 16 model compounds by progressively plotting the peak areas for higher amounts injected of each analyte until the correlation coefficient (r 2 ) dropped in value below 0.998. The upper limit of linearity was defined in this paper as the highest amount injected where the r 2 ⱖ 0.998. The estimated lower limit of detection (LOD) for 10-␮L injections of 18 representative compounds was calculated from the minimum concentration or injected mass required to produce a peak height signal-to-noise ratio of 3. The noise level was measured using the peak-topeak noise algorithm in the PeakNet software for a 1-min interval of a blank injection corresponding to the same retention time as the analyte of interest. Peak area and retention time precision were determined for replicate injections of 16 analyte standards commonly found in cell cultures and fermentation broths (see Table 2 for list). These carbohydrate, alcohol, alditol, and glycol standards were added (10 mg/L) to a 100fold dilution of a heat-treated (100°C, 10 min) yeast fermentation broth supernatant and then analyzed over 48 h (10 ␮L per injection, 42 injections) on the CarboPac MA1 column. Fermentation broth samples for the reproducibility experiment were heat-treated to inactivate enzymes and eliminate time-dependent changes in analyte concentrations resulting from continued metabolism.

CHROMATOGRAPHIC DETERMINATION OF CARBOHYDRATES TABLE 1

Retention Times for Carbohydrates and Alcohols on the CarboPac MA1 Analyte

Retention time (min)

2,3-Butanediol Ethanol Methanol Glycerol Erythritol Rhamnose Fucose Arabitol Galactosamine Glucosamine Sorbitol Trehalose Galactitol Ribitol Mannitol 2-Deoxy-D-Glucose Mannose Arabinose Glucose Xylose Galactose Maltitol Lactose Fructose Ribose Cellobiose Sucrose Raffinose Maltose Maltotriose

6.6 7.3 7.7 8.7 10.7 13.6 13.8 14.7 14.7 15.3 16.1 17.0 17.6 17.8 19.4 20.2 21.6 21.6 24.0 24.6 27.0 27.7 28.9 29.0 31.5 43.2 44.9 51.8 59.4 ⬎60.0

195

resolved on the PA1. 2,3-Butanediol coeluted with ethanol, and glycerol coeluted with methanol. Erythritol/ methanol, galactitol/ribitol/sorbitol, and galactosamine/ arabinose were also not well resolved. Except for the galactitol/ribitol pair, all the aforementioned compounds were resolved on the MA1. Some pairs of sugars are better resolved on the PA1. For example, the rhamnose/ fucose, arabitol/galactosamine, mannose/arabinose, and lactose/fructose pairs that were difficult to resolve on the MA1 column were well resolved on the PA1. When stronger eluents (ⱖ100 mM NaOH) are used with the PA1 column it elutes larger carbohydrates faster than the MA1 column. For example, maltotriose, which does not elute within 60 min on the MA1 column, elutes at 43.5 min using 100 mM NaOH eluent with the PA1. When larger carbohydrates (disaccharides and trisaccharides) are of interest, the PA1 column should be used. The CarboPac MA1 and PA1 columns have different strengths for determining the carbohydrates in cell cultures and fermentation broths. The appropriate column choice will be dictated by the analytes present, the analytes of interest, analyte concentrations, and the desired analysis time. Because the MA1 was able to separate a wide variety of alcohols, sugar alcohols, and carbohydrates, we chose it for our studies. Method detection limits. The method detection limits (MDL) for a 10-␮L injection of representative fermentation broth and cell culture constituents using the MA1 column are shown in Table 2. Detection limits were generally estimated to be about 1 ng, or about 100

RESULTS AND DISCUSSION

Chromatography. Figure 1 shows the separation of alcohols (2,3-butanediol, ethanol, methanol), a glycol (glycerol), alditols (erythritol, arabitol, sorbitol, galactitol, mannitol), and carbohydrates (rhamnose, arabinose, glucose, galactose, lactose, sucrose, raffinose, maltose) commonly found in cell cultures and fermentation broths using a CarboPac MA1 column set. The retention times of these and additional carbohydrates, alcohols, and alditols that may be present in unusual cell cultures and fermentation broths are listed in Table 1. Generally, alcohols eluted first, followed by glycols, alditols, monosaccharides, disaccharides, and trisaccharides. Rhamnose and fucose coeluted, as did the arabitol/galactosamine, galactitol/ribitol, mannose/ arabinose, and lactose/fructose pairs. The trisaccharide maltotriose did not elute within 60 min. Figure 2 shows the analysis of common fermentation broth and cell culture alcohols, glycols, alditols, and carbohydrates using the CarboPac PA1 column set. The elution order of the PA1 was similar to that of the MA1. Many of the alcohols, glycols, and alditols that were well resolved on the MA1 column were not as well

FIG. 2. Common carbohydrate, alditol, alcohol, and glycol standards found in cell cultures and fermentation broths, separated on the CarboPac PA1 analytical and guard column with pulsed amperometry (gold electrode). The eluent was 16 mM NaOH and the flow rate was 1.0 mL/min. All standards were at 10 ␮g/mL concentration, except methanol (69 mg/mL) and ethanol (4 mg/mL); 10-␮L injection. Peak identities: 1, butanediol (2,3-)/ ethanol; 2, methanol/ glycerol; 3, erythritol; 4, arabitol; 5, galactitol; 6, sorbitol; 7, mannitol; 8, rhamnose; 9, arabinose; 10, galactose; 11, glucose; 12, sucrose; 13, lactose; 14, raffinose.

196

HANKO AND ROHRER TABLE 2

Estimated Detection Limits and Linear Range Using the CarboPac MA1 with Pulsed Amperometry Detection limits a

Linear range

Analyte

ng

ng/mL b

␮g/mL

r2

2,3-Butanediol Ethanol Methanol Glycerol Glycerol Erythritol Rhamnose Arabitol Sorbitol Galactitol Mannitol Arabinose Glucose Galactose Lactose Ribose Sucrose Raffinose Maltose

1 300 7000 0.4 — 0.2 1 0.5 0.8 0.7 0.7 1 0.9 1 2 1 4 5 9

100 30000 700000 40 — 20 100 50 80 70 70 100 90 100 200 100 400 500 900

0.1–1000 ND ND 0.04–200 0.04–1000 0.02–1000 0.1–1000 0.05–1000 0.08–1000 0.07–1000 0.07–1000 0.1–1000 0.09–1000 0.1–1000 0.2–1000 0.1–1000 0.4–1000 0.5–1000 0.9–1000

0.9995 ND c ND 0.9980 0.9998 d 0.9965 0.9987 0.9986 0.9978 0.9971 0.9987 0.9993 0.9991 0.9991 0.9998 0.9992 0.9987 0.9987 0.9991

a

Lower limit of detection is based on 3⫻ baseline noise. 10-␮L injections. c ND, not determined. d Second degree polynomial regression. b

ng/mL for a 10-␮L injection. Detection limits tended to increase with longer retention times because of peak broadening. Later eluting peaks such as lactose, sucrose, raffinose, and maltose had MDLs higher than 1 ng per injection. Ethanol had 570 times less response than glucose, and methanol had 3600 times less response than glucose. Consequently, the MDL for these alcohols were higher: 300 ng for ethanol and 7000 ng for methanol per injection. The lower response (and higher detection limits) for alcohols may make this method unsatisfactory for some alcohol applications. Alternatively, alcohols can be determined with greater sensitivity using PAD with a platinum electrode (2). Linearity. Table 2 shows that the MA1 method was linear for 2,3-butanediol, rhamnose, arabitol, sorbitol, mannitol, arabinose, glucose, galactose, lactose, ribose, sucrose, raffinose, and maltose over the range of 0.1 to 1000 ␮g/mL (1 to 10,000 ng) for a 10-␮L injection. The r 2 values were 0.998 to 0.999 for these 13 analytes. Glycerol, erythritol, and galactitol were not linear over this concentration range (r 2 ⫽ 0.987, 0.997, 0.997, respectively). Glycerol was linear over the range of 0.04 –200 ␮g/mL; erythritol over the range of 0.04 –100 ␮g/mL; and galactitol over the range of 0.07–100 ␮g/ mL. The best fits for the glycerol, erythritol, and galactitol results were second-order polynomial regressions. Linearity was demonstrated for these 16 analytes over

at least three orders of magnitude, and for most analytes, over four orders of magnitude. We found that these broad linear ranges helped reduce the need to dilute samples and repeat the analysis when component concentrations varied greatly. Precision and ruggedness. Peak area and retention time RSDs were determined for replicate injections of a heat-treated yeast fermentation broth supernatant (diluted 100-fold) that was spiked with a standard mix of 16 carbohydrates and alditols (Table 3). This sample was analyzed repetitively for 48 h (42 injections) using the MA1 method. Peak area RSDs ranged from 2 to 7%, and retention time RSDs ranged from 0.2 to 0.4%. There was no negative or positive trending in the peak area or retention times over the course of this experiment. This suggested that the sample was not fouling either the column or the working electrode. We found that precision was affected by concentration (i.e., RSD values increased as concentrations approached the MDL). Long-term stability studies over several months of continuous electrode use have shown decreases in peak area response for carbohydrate standards (14) using the triple-pulse waveform described under Materials and Methods. An alternative quadruple-pulse waveform was recommended by Rocklin et al. (14) as a replacement for the triple-pulse to improve the precision, ruggedness, and long-term reproducibility of the PAD method. The quadruple-pulse waveform was developed and published after this study was conducted. We did investigate the feasibility of using the newer quadruple-pulse waveform on carbohydrate, alditol, and glycol standards in water using the same chromatographic conditions as those described for the CarTABLE 3

Peak Area and Retention Time Precision (RSD) over 48 h of Repetitive 10-␮L Injections (n ⫽ 42) of a Heat-Inactivated Yeast Fermentation Broth Supernatant Spiked with Standards (10 ␮g/mL) Analyte

Peak area (%)

Retention time (%)

2,3-Butanediol Ethanol Glycerol Erythritol Rhamnose Arabitol Sorbitol Galactitol Mannitol Arabinose Glucose Galactose Lactose Ribose Raffinose Maltose

2.4 2.7 2.0 3.4 1.8 3.0 2.7 2.7 2.7 3.1 3.3 3.5 3.6 3.1 4.8 6.8

0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.3 0.3 0.3 0.2 0.3 0.3 0.3 0.4 0.4

197

CHROMATOGRAPHIC DETERMINATION OF CARBOHYDRATES

FIG. 3. Saccharomyces cerevisiae culture and fermentation broth supernatant diluted 100-fold (10-␮L injection) and analyzed using the method described in Fig. 1: (A) 0 h of incubation (initial cell culture) and (B) 24 h of incubation (final fermentation broth). Peak identities: 1, unknown; 2, unknown; 3, ethanol; 4, glycerol; 5, unknown; 6, erythritol; 7, unknown; 8, rhamnose; 9, arabitol; 10, trehalose; 11, unknown; 12, arabinose; 13, glucose; 14, unknown; 15, cellobiose; 16, unknown.

boPac MA1 method over 56 h (48 injections). The peak area RSDs ranged from 3 to 5%, and the retention time RSDs ranged from 0.2 to 0.8%. These results suggest that the quadruple-pulse waveform will produce results equivalent to those found when the triple-potential waveform was used for the analysis of cell cultures and fermentation broths. The advantage of the quadruple-potential waveform is the greater long-term (⬎48 h) reproducibility of peak area responses. Analysis of yeast cell cultures and fermentation broths. Yeast was grown in Bacto YPD broth at 37°C for up to 24 h. Aliquots of cell culture were removed at

specified time points and centrifuged to remove cell debris, and the supernatant was diluted in water and directly injected for analysis using the MA1 method. This procedure was also performed on the final fermentation broth after 24 h of incubation. Alternatively, samples could be filtered instead of centrifuged to remove particulates. Figure 3 shows the separation of cell culture ingredients at the beginning (Fig. 3A) and after 24 h (Fig. 3B) of incubation in the fermentation broth. Glucose (dextrose) was prominent at the beginning of the culture. During the first 3 h, glucose levels decreased, and after 3 h glucose was not detected (Table 4). Trace levels of glycerol, erythritol, rhamnose, trehalose, arabinose, and cellobiose were detected at the beginning of the culture. Glycerol increased over the first 3 h and remained constant up to 7 h. Glycerol is known to be produced by yeast for the regeneration of nicotinamide adenine dinucleotide (NAD⫹) from its reduced form (NADH), which is involved in providing intracellular metabolic energy (15). The glycerol formed during glycolysis leaves the cell by passive diffusion. Erythritol and rhamnose concentrations did not change, as expected for S. cerevisiae (16). Cellobiose concentrations decreased by 50%, which was unexpected because this strain of yeast is not known to assimilate this carbohydrate (16). Trehalose and arabinose were depleted between 7 and 24 h. The loss in arabinose after 7 h was also unexpected because S. cerevisiae does not utilize pentoses (17). Trehalose, along with glycogen, is considered as a type of energy reserve for yeast, and consequently its loss under starvation conditions is expected (18). Ethanol was found at a relatively high level at the beginning of the culture incubation and remained constant up to 7 h. The analysis of raw ingredients used in the cell culture indicated that the initially high level of ethanol was introduced from both the YPD medium (40%) and the yeast sample used to inoculate the culture (60%). Between 7 and 24 h ethanol concentration decreased, presumably

TABLE 4

Determination of Carbohydrates, Alcohols, Alditols, and Glycols in a Yeast Culture and Fermentation Broth (24 h) Broth concentration (mg/mL) Elapsed time (h)

Ethanol

Glycerol

Erythritol

Rhamnose

Arabitol

Trehalose

Arabinose

Glucose

Cellobiose

0 0.5 1 2 3 4 5 6 7 24

181 167 170 174 166 175 170 173 169 14

0.4 1.1 1.4 2.0 2.3 2.4 2.3 2.3 2.3 2.3

0.006 0.004 0.004 0.006 0.005 0.006 0.007 0.006 0.006 0.008

0.028 0.024 0.028 0.038 0.039 0.039 0.041 0.029 0.036 0.085

0.005 0.000 0.000 0.000 0.000 0.000 0.000 0.005 0.004 0.000

0.97 0.93 0.90 0.88 0.82 0.81 0.75 0.73 0.69 0.00

0.030 0.034 0.032 0.030 0.016 0.000 0.000 0.000 0.000 0.000

28.0 22.5 17.5 7.8 0.5 0.0 0.1 0.0 0.0 0.0

0.14 0.17 0.12 0.15 0.12 0.12 0.13 0.13 0.10 0.07

198

HANKO AND ROHRER

FIG. 4. S. cerevisiae culture grown on medium consisting of multiple carbohydrates and alditols and analyzed using the method described in Fig. 1.

due to evaporative losses. The increase in ethanol was not observed because the culture was conducted under aerobic conditions, favoring a respiratory metabolism (citric acid cycle) over anaerobic (alcoholic) fermentative metabolism. The relative insensitivity of this PAD method for alcohols, such as methanol and ethanol, is an advantage when determining alditols, glycols, and carbohydrates in cell cultures that produce large concentrations of alcohols. The lower response for alcohols reduces the interference for the determination of carbohydrates and glycols in these types of broths. When the yeast culture medium was modified to contain 10 different carbohydrates and alditols, at the same total carbohydrate concentration as the standard Bacto YPD broth, it was apparent that yeast preferred to use certain carbohydrates over others, and that some carbohydrates or alditols could not be used as a carbon source during the 24-h incubation period. Figure 4 shows the concentrations of broth components over 24 h. Glucose and raffinose were metabolized within the first hour. The rapid metabolism of glucose is nearly universal among the different strains of S. cerevisiae; however, the metabolism of the trisaccharide raffinose varies among different strains (19). The ability to use raffinose as a carbon source is dependent on the strain’s ability to produce the periplasmic enzymes invertase and melibiase (20). Sucrose could not be measured, even at 0.15 h (9 min) of incubation, which was the earliest time point possible due to the time required to dissolve the yeast inoculum. Sucrose was probably quickly digested by the periplasmic enzyme invertase, which is present in large amounts in the dried yeast. Invertase cleaves sucrose into its monosaccharides, glucose and fructose. Glucose was measured at levels higher than expected at the first time point, which supports this hypothesis. The lower rate in the decrease of raffinose compared to sucrose probably reflects a lower substrate binding affinity or catalytic activity of invertase for raffinose. After 1 h, the yeast began to consume maltose and galactose. The

delayed metabolism of these carbohydrates has been described as the result of an inducible transport system (21). In the case of galactose, delayed metabolism may also be the result of three inducible intracellular enzymes involved in its catabolism: galactokinase, hexose-1-phosphate uridyltransferase, and UDP-glucose4-epimerase (22). Rhamnose, sorbitol, arabinose, lactose, and ribose were either unchanged or decreased slightly over 24 h. The lack of assimilation of these compounds as carbon sources is consistent with the literature (16). Glycerol increased for the first hour and then leveled off. The capability of detecting carbohydrates, alditols, and glycols with high sensitivity and without prederivatization is a major advantage of PAD over other detection methods. Broth and culture solutions are simply centrifuged or filtered to remove particulates, diluted, and then injected. The rapid and simultaneous analysis of many different analytes renders this technique ideal for broth optimization during process development and assessment of the condition of the yeast culture for harvesting products. It can also be used for taxonomic identification of microorganisms or characterization of their variants (16, 23). The ability to monitor a culture for a broad range of substances, and their changes during fermentation, gives this method an advantage over other methods (e.g., flow-injection analysis) that can only determine a few analytes in one injection. The high sensitivity of PAD permits lower detection limits and less sample waste. PAD is also insensitive to some chromophoric compounds in cell cultures and fermentation broths that might interfere with absorbance detection. This method was also used to investigate Escherichia coli grown on Luria-Bertani (LB) broth (results not presented). Many identified and unidentified peaks increased or decreased in peak area during the 24-h incubation period. Therefore, this method also exhibited the potential to monitor metabolically related substrates and products in prokaryotic cell cultures, which could be useful for optimizing product yields. CONCLUSION

The results presented in this paper show that both yeast and bacterial cell cultures and fermentation broths could be analyzed for carbohydrate, alcohol, alditol, and glycol composition using HPAE-PAD. The application of this technique to both eukaryotic and prokaryotic cell culture samples demonstrated the feasibility of simultaneously determining a large number of carbohydrate and alcohol compounds in cell cultures and their final fermentation broths. Time course studies of these cultures revealed that this method could measure time-dependent changes in the concentrations of these ingredients. This method can be used to

CHROMATOGRAPHIC DETERMINATION OF CARBOHYDRATES

determine the preference for specific carbohydrates as carbon sources, and the generation of metabolic byproducts, for a given strain of S. cerevisiae, and other types of metabolically active microorganisms. REFERENCES 1. Lee, Y. C. (1996) J. Chromatogr. A 720, 137–149. 2. Robinett, R. S. R., and Herber, W. K. (1994) J. Chromatogr. A 671, 315–322. 3. Herber, W. K., and Robinett, R. S. R. (1994) J. Chromatogr. A 676, 287–295. 4. Marko-Varga, G., Buttler, T., Gorton, L., Olsson, L., Durand, G., and Barcelo, D. (1994) J. Chromatogr. A 665, 317–332. 5. Schugerl, K., Brandes, L., Wu, X., Bode, J., Ree, J. I., Brandt, J., and Hitzmann, B. (1993) Anal. Chim. Acta 279, 3–16. 6. Rank, M., Gram, J., and Danielsson, B. (1993) Anal. Chim. Acta 281, 521–526. 7. Buttler, T., Gordon, L., and Marko-Varga, G. (1993) Anal. Chim. Acta 279, 27–37. 8. Van de Merbel, N. C., Lingeman, H., Brinkman, U. A. T., Kolhorn, A., and de Rijke, L. C. (1993) Anal. Chim. Acta 279, 39 –50. 9. Van de Merbel, N. C., Kool, I. M., Lingeman, H., Brinkman, U. A. Th., Kolhorn, A., and de Rijke, L. C. (1992) Chromatographia 33, 525–532. 10. Marko-Varga, G., Dominguez, E., Hahn-Hagerdal, B., Gorton, L., Irth, H., De Jong, G. J., Frei, R. W., and Brinkman, U. A. Th. (1990) J. Chromatogr. 523, 173–188. 11. Dionex Corporation (1998) Application Note 123, Determination of organic and inorganic anions in fermentation broths, Dionex Corporation, Sunnyvale, CA. 12. Hanko, V. P., and Rohrer, J. S. (1999) Gen. Eng. News 19, 19, 51.

199

13. Weitzhandler, M., Slingsby, R., Jagodzinski, J., Pohl, C., Narayanan, L., and Avdalovic, N. (1996) Anal. Biochem. 241, 135–136. 14. Rocklin, R. D., Clarke, A. P., and Weitzhandler, M. (1998) Anal. Chem. 70, 1496 –1501. 15. Gancedo, C., and Serrano, R. (1989) in The Yeasts, 2nd ed. (Rose, A. H., and Harrison, J. S., Eds.), Vol. 3, pp. 227–229, Academic Press, New York. 16. Yarrow, D. (1984) in The Yeasts, a Taxonomic Study (Kreger-van Rij, N. J. W., Ed.), p. 384, Elsevier Science, New York. 17. Gancedo, C., and Serrano, R. (1989) in The Yeasts, 2nd ed. (Rose, A. H., and Harrison, J. S., Eds.), Vol. 3, p. 214, Academic Press, New York. 18. Gancedo, C., and Serrano, R. (1989) in The Yeasts, 2nd ed. (Rose, A. H., and Harrison, J. S., Eds.), Vol. 3, pp. 249 –250, Academic Press, New York. 19. Barnett, J. A., Payne, R. W., and Yarrow, D. (1983) Yeasts, Characteristics and Identification, pp. 467– 469, Cambridge Univ. Press, New York. 20. Gancedo, C., and Serrano, R. (1989) in The Yeasts, 2nd ed. (Rose, A. H., and Harrison, J. S., Eds.), Vol. 3, pp. 211–212, Academic Press, New York. 21. Cartwright, C. P., Rose, A. H., Calderbank, J., and Keenan, M. H. J. (1989) in The Yeasts, 2nd ed. (Rose, A. H., and Harrison, J. S., Eds.), Vol. 3, pp. 5–56, Academic Press, New York. 22. Gancedo, C., and Serrano, R. (1989) in The Yeasts, 2nd ed. (Rose, A. H., and Harrison, J. S., Eds.), Vol. 3, pp. 213–214, Academic Press, New York. 23. de van Broock, M. R., Sierra, M., and de Figueroa, L. Intergeneric fusion of yeast protoplasts. (1980) in Current Developments in Yeast Research; Advances in Biotechnology (Stewart, G. G., and Russell, I., Eds.) pp. 171–176, Pergamon, New York.