- Open Access
Fast quantitative determination of microbial rhamnolipids from cultivation broths by ATR-FTIR Spectroscopy
© Leitermann et al; licensee BioMed Central Ltd. 2008
- Received: 25 April 2008
- Accepted: 07 October 2008
- Published: 07 October 2008
Vibrational spectroscopic techniques are becoming increasingly important and popular because they have the potential to provide rapid and convenient solutions to routine analytical problems. Using these techniques, a variety of substances can be characterized, identified and also quantified rapidly.
The rapid ATR-FTIR (Attenuated Total Reflectance Fourier Transform Infrared Spectroscopy) in time technique has been applied, which is suitable to quantify the concentrations of microbial rhamnolipids in a typical cultivation process. While the usually applied HPLC analysis requires an extensive and time consuming multi step extraction protocol for sample preparation, the ATR-FTIR-method allows the quantification of the rhamnolipids within 20 minutes. Accuracies between 0.5 g/l – 2.1 g/l for the different analytes were determined by cross validation of the calibration set. Even better accuracies between 0.28 g/l – 0.59 g/l were found for independent test samples of an arbitrarily selected cultivation.
ATR-FTIR was found to be suitable for the rapid analysis of rhamnolipids in a biotechnological process with good reproducibility in sample determination and sufficient accuracy. An improvement in accuracy through continuous expansion and validation of the reference spectra set seems very likely.
- High Performance Liquid Chromatography
- Independent Test Sample
- Vibrational Spectroscopic Technique
Ranging from bulk chemicals like ethanol to high value proteins, biotechnological production processes are an increasingly important manufacturing route for various products . The control of these bioprocesses often can be considered as suboptimal. Usually only a few parameters, like pH, pO2 and temperature, are monitored online. All additional information required must be gained through analysis of individual samples. Typical assays often rely on enzymatic reactions or separation techniques such as high performance liquid chromatography (HPLC) . Therefore, the analysis results often will be available only with a significant time delay.
Hence, vibrational spectroscopic techniques that provide rapid and convenient solutions to routine analytical problems are being increasingly adopted. A variety of substances can be characterized, identified and also quantified rapidly in parallel from a single sample spectrum . Fourier transform infrared spectroscopy (FTIR) is a reliable and well-recognized method . For a complex analyte-matrix combination, like that which occurs in fermentations broths, the adaptation of the FTIR technique needs extensive experience and time. Nevertheless, once a method has been established, it allows for relatively fast assays of compounds, where the alternative quantitative analysis (e.g. HPLC methods) can be time-consuming. Attenuated total reflection infrared Fourier transform spectroscopy (ATR-FTIR) is a FTIR variety, which allows organic substances in aqueous solutions to be determined. ATR-FTIR involves the collection of radiation reflected from the interfacial surface between the aqueous solution and a reflection element (ATR crystal). In this crystal, evanescent waves emanate from the crystal, penetrate the aqueous solution and are absorbed by substances in this solution [6, 7]. Concerning the analysis of biosurfactants several spectroscopic methods for the characterization, identification and quantification have been reported [8–13].
Biosurfactants are microbially produced surface active compounds . Anionic glycolipid biosurfactants consisting of L-rhamnose sugars and aliphatic chains moieties are called rhamnolipids. Several variations of these rhamnolipids are known . Rhamnolipids produced by P. aeruginosa strains are often composed of one or two L-rhamnose and, additionally, one or two β-hydroxydecanoic acid moieties. They are termed rhamnolipid 1 (L-rhamnosyl-3-hydroxydecanoyl-3-hydroxydecanoate) and rhamnolipid 3 (L-rhamnosylrhamnosyl-3-hydroxydecanoyl-3-hydroxydecanoate) .
The objective of the presented work is to highlight the range of application made possible by utilizing an ATR-FTIR method for the monitoring of a biotechnological process for the production of microbial rhamnolipids. Additionally, the potential of this method for achieving structural information and identification purposes concerning rhamnolipids was investigated.
Characterization and interpretation of the absorbance spectrum of rhamnolipid 3
Basic investigations for the applicability of ATR-FTIR
Rhamnolipids are surface active compounds, and, therefore, absorbance effects at the interfacial surface of the ATR crystal to liquid phase must be expected. With respect to the small volume of the sample, 20 μl, even evaporation may affect the absorbance intensity of the spectra and influence the quantification of the analytes. Hence, time dependant spectra of the samples were recorded for comparison. Four spectra of the same sample were recorded in time intervals of 2 minutes. Neither intensity changes of single bands nor an overall intensity change of the absorbance spectrum were observed. Therefore evaporation and interfacial absorbance effects were not relevant for all practical purposes.
Quantification of rhamnolipids by ATR-FTIR
In a second step, the evaluation of the prediction error of the established calibration models versus an independent test samples set from an arbitrarily selected fermentation process was investigated by test set validations. The results are displayed in Figure 3. At a rank of 5, for rhamnolipid 1, a determination coefficient R2 of 87.88 and a RMSEP of 0.278 g/l were calculated. When considering a rank of 5, the optimized correlation for rhamnolipid 3 responds with a R2 of 99.43 and a RMSEP of 0.441 g/l. For the total rhamnolipid content a R2 of 99.31 and a RMSEP of 0.586, at a rank of 5, were found. Similar to the cross validated models, the best results were found by restricting of a single frequency area from 900 – 1200 cm-1. Also, the similar mathematical pre-treatments, as described above, were responding the best results.
The spectrum of pure rhamnolipid 3, as shown in Figure 1, indicates characteristic absorption bands at 2921 and 2855 cm-1 derived from symmetric C-H and a C=O stretching band at 1730 cm-1. Gartshore et al.  suggested the correlation of a single peak area when analyzing artificially prepared samples of three biosurfactants, lecithin, cholesterol laurate, and spiculi sporic acid. The peak utilized in this study was between 1720 – 1770 cm-1. Unlike these spectroscopic experiments, the calibrations derived from the cultivation samples showed the best correlations for the quantitative determination of rhamnolipids when the fingerprint regions of the spectra were utilized. This could be explained by the complexity of the analyte-matrix of real cultivation samples. Besides rhamnolipids, a great variety of additional substances, like plant oil, fatty acids, buffer salts or ester compounds were present, which influence the absorption intensities in the spectrum area of the three characteristic bands upon 1700 cm-1. Hence, the unapparent information comprised in the fingerprint area of the spectra was found to be most suitable for selective, quantitative determinations of rhamnolipids.
Overall, the ATR-FTIR assays showed a good reproducibility. Even suspected interfacial or evaporation effects could not be observed, as can be seen from the consecutive spectra shown in Figures 2 and 3. The differentiation of rhamnolipid 1, rhamnolipid 3 and the total rhamnolipid was possible.
By cross validation, predictive errors between 0.496 g/l to 2.06 g/l were determined, which are adequate for a fast in time analysis. Similar or even better results, with predictive errors from 0.278 g/l to 0.586 g/l, were obtained by test set validation of independent cultivation samples. Accounting for the highest rhamnolipid 1 value in Figure 3, resulted from a possible outlier, the derived correlation error seems to be even lower. Also, it has been shown that reliable correlations could be found, despite different fermentation conditions with respect to the substrates. An improvement of the accuracy and stability of the correlations by expansion of the calibrations data sets with samples from additional cultivations is expected. In addition, the major advantage of an ATR-FTIR compared to a HPLC quantification of rhamnolipids is the time needed. Whereas HPLC data is available after approximately 24 h, caused by multiple extractions, evaporation and derivatisation steps, ATR-FTIR provides process values within 20 minutes.
ATR-FTIR was found to be appropriate for the rapid analysis of rhamnolipids in a biotechnological process with sufficient accuracy and good reproducibility. Further improvements to the quantification accuracy can be expected by expansion of the calibration data set with samples from additional cultivations. By opening a "spectroscopic eye" into bioprocesses for the production of microbial rhamnolipids, in time ATR-FTIR monitoring makes rapid process characterization and process control much easier. Future objectives for the modifications of the utilized ATR-FTIR technique towards an online in situ measurement system by using integrated ATR probes seemsto be very promising.
Pseudomonas aeruginosa DSM 7108 was obtained from the German Collection of Microorganisms and Cell Cultures. The strain is stored as glycerol stock culture at -80°C.
For seed cultures, a lysogeny broth (LB) medium containing 5 g/l NaCl was used . For the bioreactor cultivations, a Ca-free mineral salt medium was used. The medium is based on a 0.1 M sodium phosphate buffer pH 6.5 supplemented with 0.5 g/l MgSO4 × 7 H2O, 1 g/l KCl and 15 g/l NaNO3 (modified, according to ). The phosphate buffer was separately autoclaved from the other salts. A sterile filtered trace elements solution consisted of 2 g/l sodium citrate × 2 H2O, 0.28 g/l FeCl3 × 6 H2O, 1.4 g/l ZnSO4 × 7 H2O, 1.2 g/l CoCl2 × 6H2O, 1.2 g/l CuSO4 × 5 H2O and 0.8 g/l MnSO4 × H2O was added to the media as described below. All chemicals were of analytical grade. As a carbon source, three types of sunflower oils (high oleic sunflower oils HOS 90+, HOS 80+, conventional sunflower oil), linseed oil, rapeseed oil, corn oil soy bean oil (all by courtesy of Dr. B Schlüter, Unternehmensberatung und Dienstleistung im vor- und nachgelagerten Bereich der Landwirtschaft, Bornheim, Germany) and fish oil (by courtesy of Dr. Krumbholz, KD-Pharma Bexbach GmbH, Bexbach, Germany) of technical/food grade were used.
As for the bioreactor cultures, 25 ml LB medium was inoculated with 100 μl of a glycerol stock culture in a 100 ml shaking flask. This culture was incubated for 12 h at 30°C and 140 rpm in an incubation shaker (HT Infors Multitron II), until an optical density at 580 nm (OD580) of 2.5 – 3.0 was reached (Amersham Biosciences Ultrospec 1100pro). With 2 ml of this seed culture, the main cultures were inoculated. The cultivations were performed in a 500 ml parallel bioreactor system (Sixfors, IFORS GmbH Einsbach, Germany) with 214 ml (200 ml media + 14 ml plant oil) initial volume. The temperature was controlled and maintained at 30°C. During the process, the pH was controlled and maintained at 6.5. An additional 14 ml substrate was added after approximately 40 h of process time. One ml per liter of the trace element solution was added at distinct process times (t = 0; 20; 40; 70; 120 h). The reactors were equipped with additional 4 blade Rushton stirrers with blade geometry of 1:1 for mechanical foam destruction. For chemical defoaming Contraspum™ A4050 (Zschimmer&Schwarz GmbH & Co KG Chemische Fabriken, Lahnstein, Germany) was used. The processes were controlled and recorded by a personal computer system with the process software Iris (INFORS GmbH Einsbach, Germany).
A ttenuated T otal R eflectance (ATR) Fourier transform infrared spectra were collected with a bench-top spectrometer (Tensor 27, Bruker Optics GmbH, Ettlingen, Germany) equipped with a liquid nitrogen cooled, linear LN-MCT-Photovoltaic detector (Kolmar Technologies Inc., Newburyport, MA 01950, USA) and a BioATR II cell (Harrick Scientific Products Inc., Pleasantville, NY 10570, USA). The Bio-ATR II cell was purged by a continuous flow of dried air to minimize water vapour in the IR-beam path. The ATR cell was temperature stabilized at 25°C by a refrigerated circulator bath (HAAKE DC30-K20, Thermo Haake GmbH, Karlsruhe, Germany) to avoid water signals in the IR-sample spectrum resulting from temperature shifts between reference and sample measurement. Before the measurement, the crystal of the ATR cell was washed several times with 20 μl H2O. Cleanliness of the ATR-crystal was controlled by the measurement of a baseline of the water covered crystal before and after washing.
A reference spectrum of water was measured. Then the cell was washed twice with 20 μl sample and the actual sample measurement was performed. Both for reference and sample measurement, 64 interferograms were averaged resulting in an overall scanning time of about 1 minute. Spectra were recorded in a range from 850 – 4000 cm-1 at a resolution of 4 cm-1 and with an aperture of 6 mm. Data storage, spectra processing, substance comparison and the quantitative analysis of the spectra were done with the software OPUS 5.5 (Bruker Optics GmbH, Ettlingen, Germany).
The recorded FT-IR spectra together with the results from HPLC reference analysis were analyzed using partial least-squares (PLS) regression. PLS calibration development, including cross- and test-set validation were performed with the QUANT 2 module of the OPUS 5.5 software, according to the multivariate calibration techniques described by Conzen . Multiple mathematical pre-treatments were iteratively applied and for evaluation of the fits of the correlations to the reference HPLC data, the determination coefficients R2 were calculated. Additionally, the predictive abilities of the correlation methods were proofed by calculation of two types of prediction errors. For validation within the whole calibration set, cross validations with one leave out sample were applied, and the root mean square errors of cross validation (RMSECV) were determined.
The evaluation of the predictive quality of an established calibration model applied on an independent test samples set from an arbitrarily selected fermentation process was done by test set validations and calculation of the root mean square error of prediction (RMSEP). All errors were calculated for different numbers of PLS factors. During the establishment of the calibration models, spectra were checked for outliners by visual inspection of principal component score plots. The final correlations were selected according to the lowest prediction errors and PLS ranks versus highest R2 values were used to obtain confidential and stable calibration models.
Reference rhamnolipid determination by HPLC
The ATR-FTIR analysis was calibrated and validated using HPLC as reference method. A rhamnolipid 1 standard was produced for the calibration of the HPLC device by enzymatic hydrolysis of rhamnolipid 3  and purified by chromatography. At the stationary phase, fine silica gel was used, and at the mobile phase, a chloroform-methanol 75:25 was utilized. The rhamnolipid 3 standard with a purity grade of 97.3% was purchased from Hoechst AG (Frankfurt, Germany).
Applied linear blending parameters for the solvents A & B for the HPLC rhamnolipid analysis.
solvent A in %
solvent B in %
We thank the FNR e.V. (Fachagentur Nachwachsende Rohstoffe e.V.) for funding, the industrial project partners Bruker Optics GmbH, Sartorius AG, Ecover Belgium and Henkel AG for technical support, and our academic partners Dr. Rosenau (University of Düsseldorf), Prof. Dr. Chmiel and PD Dr. Janke (UPT, Gesellschaft für umweltkompatible Prozesstechnik mbH i.L., Saarbrücken) and the group of Dr. Behrensmeier (Forschungszentrum Karlsruhe) for lively discussions.
- Antranikian G: Angewandte Mikrobiologie. 1st edition. Berlin, Heidelberg, New York: Springer-Verlag; 2006.Google Scholar
- Mazarevica G, Diewok J, Baena JR, Rosenberg E, Lendl B: On-line fermentation monitoring by mid-infrared spectroscopy. Appl Spectrosc 2004, 58: 804-810. 10.1366/0003702041389229View ArticleGoogle Scholar
- Gillie JK, Hochlowski J, Arbuckle-Keil GA: Infrared spectroscopy. Anal Chem 2000, 72: 71-80. 10.1021/a1000006wView ArticleGoogle Scholar
- Workman J, Veltkamp DJ: Process analytical chemistry. Anal Chem 1999, 71: 121-180. 10.1021/a1990007sView ArticleGoogle Scholar
- Tanaka T, Nagao S, Ogawa H: Attenuated Total Reflection Fourier Transform Infrared (ATR-FTIR) Spectroscopy of Functional Groups of Humic Acid Dissolving in Aqueous Solution. Anal Sci 2001, 17: i1081-i1084.Google Scholar
- Günzler H, Gremlich H-U: IR-Spektroskopie. 1st edition. Wiley-VCH; 2003.View ArticleGoogle Scholar
- Otto A: Excitation of Nonradiative Surface Plasma Waves in Silver by Method of Frustrated Total Reflection. Zeitschrift für Physik 1968, 216: 398. 10.1007/BF01391532View ArticleGoogle Scholar
- Gartshore J, Lim YC, Cooper DG: Quantitative analysis of biosurfactants using Fourier Transform Infrared (FT-IR) spectroscopy. Biotechnol Lett 2000, 22: 169-172. 10.1023/A:1005670031432View ArticleGoogle Scholar
- Lébron-Paler A, Pemberton JE, Becker BA, Otto WH, Larive CK, Maier RM: Determination of the acid dissociation constant of the biosurfactant monorhamnolipid in aqueous solution by potentiometric and spectroscopic methods. Anal Chem 2006, 78: 7649-7658. 10.1021/ac0608826View ArticleGoogle Scholar
- Makkar RS, Cameotra SS: Structural characterization of a biosurfactant produced by Bacillus subtilis at 45°C. JSD 1999, 2: 367-372.Google Scholar
- Monteiro SA, Sassaki GL, de Souza LM, Meira JA, de Araujo JM, Mitchell DA, Ramos LP, Krieger N: Molecular and structural characterization of the biosurfactant produced by Pseudomonas aeruginosa DAUPE 614. Chem Phys Lipids 2007, 147: 1-13. 10.1016/j.chemphyslip.2007.02.001View ArticleGoogle Scholar
- Tuleva BK, Ivanov GB, Christova NE: Biosurfactant Production by a New Pseudomonas putida Strain. Z Naturforsch [C] 2002,57(3-4):356-360.View ArticleGoogle Scholar
- Heyd M, Kohnert A, Tan T, Nusser M, Kirschhöfer F, Brenner-Weiss G, Franzreb M, Berensmeier S: Development and trends of biosurfactant analysis and purification using rhamnolipids as an example. Anal Bioanal Chem 2008.Google Scholar
- Nitschke M, Costa SGVAO, Contiero J: Rhamnolipid surfactants: An update on the general aspects of these remarkable biomolecules. Biotechnol Prog 2005, 21: 1593-1600. 10.1021/bp050239pView ArticleGoogle Scholar
- Lang S, Trowitzsch-Kienast W: Rhamnoselipide. In Biotenside. 1st edition. Stuttgart, Leipzig, Wiesbaden: B. G. Teubner; 2002:66-81.View ArticleGoogle Scholar
- Trummler K, Effenberger F, Syldatk C: An integrated microbial/enzymatic process for production of rhamnolipids and L-(+)-rhamnose from rapeseed oil with Pseudomonas sp DSM 2874. Eur J Lipid Sci Technol 2003, 105: 563-571. 10.1002/ejlt.200300816View ArticleGoogle Scholar
- Bertani G: Studies on Lysogenesis.1. The Mode of Phage Liberation by Lysogenic Escherichia-Coli. Journal of Bacteriology 1951, 62: 293-300.Google Scholar
- Giani C, Wullbrandt D, Rothert R, Meiwes J: Pseudomonas aeruginosa and its use in a process for the biotechnological preparation of L-Rhamnose. US Patent 1997. Pat. No.: 5,658,793.Google Scholar
- Conzen J-P: Multivariate Kalibration – Ein praktischer Leitfaden zur Methodenentwicklung in der quantitativen Analytik. Ettlingen. 4th edition. 2005.Google Scholar
- Schenk T, Schuphan I, Schmidt B: High-performance Liquid Chromatographic determination of the rhamnolipids produced by Pseudomonas aeruginosa . J Chromatogr A 1995, 693: 7-13. 10.1016/0021-9673(94)01127-ZView ArticleGoogle Scholar
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.