J . Sci. Food Agric. 1984,35,881-886

Simultaneous Determination of Proline and Betaines by High Performance Liquid Chromatography

Clive W. Ford

CSIRO Division of Tropical Crops and Pastures, 306 Carmody Road, St Lucia, Queensland, Australia 4067

(Manuscript received 9 November 1983)

A method is described for the simultaneous determination by reversed-phase h.p.1.c. of proline, glycinebetaine, trigonelline and stachydrine. The mobile phase was water containing 0 . 0 2 ~ dibutylamine phosphate at pH 3.0. The compounds were separated in a Dextropak Radial-PAK Cartridge and were detected by a differential refractometer. Total time of elution was 8 min. The method was applied to the analysis of leaves of two tropical grasses and a tropical legume which had been subjected to water stress.

Keywords: Reversed-phase h.p.1.c.; proline; quaternary ammonium compounds; glycinebetaine; trigonelline; stachydrine; water stress.

1. Introduction

Proline and glycinebetaine are secondary plant products whose concentrations in plant tissue increase appreciably, in certain taxa, in response to water deficits's2 or increased ~ a l i n i t y . ~ , ~ Hence, the association of these compounds with drought resistance and salt tolerance makes quantitative determination of these substances necessary in any study of the effects of stress on cell physiology.

The analytical methods available for glycinebetaine generally either lack sensitivity and specificity, such as the p e r i ~ d i d e , ~ phosph~tungstate~ and reineckate6 methods, or else involve preliminary purification and chromatographic which are too laborious for application to large numbers of samples. Clycinebetaine has been determined by normal phase h.p.1.c. in commercial beet sugar'0.'' and wine. 'I In the latter work amino compounds were removed prior to analysis, and the reported relative mobility of glycinebetaine disagreed with the former work. In stress physiology, proline is usually determined in a separate analysis, often on plant extracts which have been pretreated with ion exchange resin, followed by spectrophotometric determination on a benzene or toluene extraction of a heated acidic ninhydrin s ~ l u t i o n . ~ , ' ~

The present paper describes a simple procedure for the individual determination, in a single analysis, of proline, glycinebetaine, and two other ' betaines, trigonelline and stachydrine (prolinebetaine), which are known to occur in a wide variety of plant species.I3 Although the method of detection using a differential refractometer lacks sensitivity compared to spectrophotometric determinations of radiation absorbing derivatives, it is not subject to major interference from small amounts of contaminants with high absorbance. In addition, proline and glycinebetaine accumulate in species of interest to concentrations well in excess of the minimum detectable limits of the method.

2. Experimental

2.1. High performance liquid chrdmatography fh.p.1.c.) 2.1.1. Instrumentation Analyses were carried out using a Waters Associates instrument consisting of a model 6000A solvent delivery system, a Waters Intelligent Sample Processor (WISP) model 710B, a model R401

881

882 C. W. Ford

differential refractometer and a model 450 variable U.V. detector. Data reduction was performed by a Data Module and the chromatograph was operated through a model 720 System Controller.

2.1.2. Operating conditions Separations of proline and betaines were effected in a ‘Radial-PAK’ cartridge (100 x 8 mm i d . ) containing ‘Dextropak’ (10 pm particles) using a ‘Waters’ Radial Compression Module (RCM-100). The cartridge was protected with a disposable precolumn insert (RCSS Guard PAK, C18). The mobile phase was water containing 0.02M dibutylamine phosphate (1 vial Waters D-4 reagent litre-’ water, pH 3.0), which was filtered (‘Millipore’ FH 0.5 pm) and degassed by stirring under vacuum (20 mm Hg) before use. The solvent flow rate was 1.5 ml min-’, and the operating temperature was 22 +- 1°C. For optimum integration, initial peak width setting was 22 s, programmed to double after 6.5min, and noise rejection was 8OOpVs-’ drift. A variation of f 5% in the retention times was allowed for automatic peak recognition.

2.2. Materials All compounds were dried under vacuum over phosphorus pentoxide at room temperature before use, and were obtained from the following commercial sources: proline and trigonelline hydrochloride (Sigma Chemical Co.), anhydrous glycinebetaine (ICN Pharmaceuticals), stachyd- rine hydrochloride (Carl Roth Laboratories),

2.3. Sample Preparation Aqueous solutions of the above compounds were prepared in concentrations ranging from 1 to 20 mg ml-’. Aliquots, in triplicate, were applied to small columns (70 x 6 mm) containing Bio-Rad AG 50X-X8 (H’) cation exchange resin (1.5 ml). After washing the resin with water (10 ml), the nitrogenous compounds were eluted with 2111 NH40H (6ml) at ca 1 ml min-’. The eluates were air dried, redissolved in water (2 ml), and clarified through a Swinnex syringe filter (Millipore HA 0.45 km and prefilter). The injection volume was 10 or 20 pl with a run time of 11 min. Non-deionised aliquots of each compound (in triplicate) were used for external calibration.

2.4. Extraction of plant material Freeze-dried, ground tissue (1-2 g, < 1 mm particle size) was sequentially extracted in a Soxhlet apparatus with diethyl ether (4 h) and then 95% aqueous ethanol (16 h). The ethanol solution was adjusted to 100 ml. Aliquots (25 ml) were evaporated to dryness, redissolved in water (1 ml) and then treated with cation exchange resin as described in section 2.3. The air-dried ammonia eluate was redissolved in water (1 ml), and part of the solution was clarified into vial inserts (0.2 ml capacity) before automatic injection into the liquid chromatograph.

3. Results and discussion

The possibility of separating proline and glycinebetaine directly by h.p.l.c., using water as the mobile phase, was first indicated by results obtained during a study of the analysis of sugars on Dextropak.12 Later it was found that trigonelline and stachydrine were also well resolved, and eluted after glycinebetaine. When pure water was the eluent, all four compounds showed considerable tailing. Attempts to improve the peak shapes by decreasing the polarity of the mobile phase or using paired-ion chromatography were unsuccessful. The best results were obtained when di-butylamine phosphate (Waters D4 reagent) was added to the mobile phase. A typical chromatogram showing the excellent separation of proline and the three betaines is shown in Figure 1. Glycinebetaine had the same retention volume as stachyose in this system. To obviate the interference of this and any other neutral or acidic plant constituents, tissue extracts were passed through cation exchange resin, and the basic compounds were subsequently eluted with ammonia.

Valine was found to co-elute with proline, while phenylalanine did not resolve from stachydrine. Eight other common protein amino acids did not interfere at all. As concentrations of free amino

Determination of proline and betaines by h.p.1.c. 883

I n P 5

C 0 v)

+ 0 W W 0 +

Figure 1. High performance liquid chromato- gram of a mixture of (1) proline; (2) glycinebe- taine; (3) trigonelline; and (4) stachydrine.

I

0 4 8 12

Time (rnin)

acids, other than proline, are typically low in stressed or unstressed leaf the errors introduced by their presence should be minimal relative to the magnitude of stress-induced increase in the levels of the other compounds.

Precision of the method was generally satisfactory (Table 1). Coefficients of variation were highest at the lowest concentration of each compound but in most cases, with the exception of stachydrine where concentrations of only 1 4 m g m l - ' were used for calibration, were less than 2%. The response factors for proline and glycinebetaine were not significantly different. However, the factor

Table 1. Detector response factors" for proline, glycinebetaine, trigonelline and stachydrine

Response factorb S.d

Proline 0.1531 0.0060 Glycinebetaine 0.1495 0,0027 Trigonelline 0.1292 0.0023 Stachydrine 0.2083 0.0065

Calibration concentrations were mg mi-'; differen- tial refractometer attenuation was x8; injection volume was 10 pl.

Mean of five different concentrations (analysed in triplicate) of each compound, calculated as the free base. Factor expressed as concentration x area-'.

58

884 C. W. Ford

Table 2. Recoveries" of proline and glycinebetaine from cation exchange resin with 2M ammonia

Proline (mg ml-') Glycinebetaine (mg mi-') Recovery Recovery

Original Found ("/.I S.d. Original Found (%) S.d

5.15 4.70 91.3 5.2 5.68 5.43 95.6 2.5 10.30 9.82 95.3 1.6 11.37 10.83 95.7 1.1 15.45 14.71 95.2 0.5 17.05 16.25 95.3 2.0 20.60 18.86 91.6 0.3 22.74 21.19 93.2 0.2 25.76 24.00 92.9 0.1 28.42 26.78 94.2 1 .0 Mean recovery 93.3 1.7 Mean recovery 94.8 1 .o

a Mean of triplicate analysis

Table 3. Recovery of proline, and glycinebetaine from synthetic mixtures

Concentration (mg ml-') Recovery

Mixture Original Found" (%I S.d.

1. Proline 4.09 Glycinebetaine 6.27

2. Proline 8.17 Glycinebetaine 12.54

3. Proline 12.26

4. Proline 12.60

Trigonelline 10.74

Glycinebetaine 18.81

Glycinebetaine 12.20

Stachydrine 7.98

"Mean of triplicate analysis.

4.24 104 6.25 100 8.55 105

12.73 102 12.84 105 19.47 104 12.97 103 11.39 93 11.13 104 7.49 94

2 2 2 1 3 1 1 2 3 3

for trigonelline was lower, and that for stachydrine higher than for the former compounds. These differences may be caused by differential column adsorption effects, as the recoveries of the compounds from the cation exchange resin were generally high. Data are given for proline and glycinebetaine where recoveries of 93.3 and 94.8% respectively were obtained (Table 2).

The accuracy of the method was tested by analysing prepared mixtures of proline and glycinebetaine (Table 3). Good recoveries with low standard deviations were obtained. When the method was applied to plant material which had been previously analysed by a different procedure,* good agreement was obtained for the proline and glycinebetaine values in the two grasses (Table 4). It was found, however, that glycinebetaine, which had been previously reported in the tropical

Table 4. Determination of proline and betaines in leaves of two tropical grasses, Panicum marimurn (green panic) and Cenchrus ciliaris (Buffel grass), and a tropical legume Macroprilium

afropurpureum (Siratro) (mg g-' dry tissue)

Water Species status' Proline Glycinebetaine Trigonelline

P. marimurn W 0.1 (tr)b 5.2 (4.8) 0 D 9.9 (11.2) 15.6 (15.1) 0

C. ciliaris W 0.1 (tr) 9.3 (10.2) 0 D 4.7 (4.4) 20.1 (21.8) 0

D 0 (0) 0 (27.0) 3.9 Siratro W 0 (0) 0 (19.0) 2.7

Tr = trace; 0 =not detected. W, well watered; D, water stressed. Figures in parentheses are values obtained previously by thin-layer electrophoresis or an

amino acid analyser.'

Determination of proline and betaines by h.p.1.c. 885

pasture legume siratro,'was not present in that plant. Trigonelline was found but at a concentration much lower than that reported for glycinebetaine. The previous analytical method employed thin-layer electrophoresis which did not resolve glycinebetaine and trigonelline efficiently, which may account for the misidentification. Quantitation had been obtained by visual estimation after spraying the chromatograms with Dragendorff's reagent3 followed by sulphuric acid. Using these conditions it was found, in the present work, that trigonelline gave five to six times the colour intensity of the same amount of glycinebetaine, which may thus account for about 90% of the reported glycinebetaine values previously obtained. The identity of trigonelline in siratro was confirmed by isolating the relevant peak from h.p.1.c. and obtaining a U.V. spectrum identical with authentic material.

Good recoveries were obtained from proline and glycinebetaine which had been added to ethanol extracts of the two grasses (Table 5) . Thus, the method offers a simple, direct identification and quantitative analysis for proline and the three betaines described. In this work, most attention has been paid to proline and glycinebetaine as these are the compounds most likely to accumulate tohigh concentrations in cells subjected to stress situations, and hence are of primary interest. However, it was important to know that trigonelline and stachydrine could also be determined by the method. Small amounts of trigonelline, in the presence of high proline concentrations in water-stressed legumes, have been most conveniently measured by simultaneous monitoring with a U.V. detector, using the high correlation of peak height with trigonelline level^.'^

Table 5. Recovery of proline and glycinebetaine added to ethanol extracts of leaves of Panicum maximum (green panic) and Cenchrus ciliaris (buffel erass)

Proline (mg) Glycinebetaine (mg) Water Recovery Recovery status" Original Added Found (%) Original Added Found (a)

P. maximum W 0.13 4.09 4.37 104 6.35 6.27 12.41 98 D 11.84 4.09 16.06 101 19.42 6.27 27.00 105

W 0.10 4.09 4.44 106 6.37 6.27 11.92 94 D 12.55 4.09 16.92 102 19.38 6.27 25.25 97

C. ciliaris W 0.15 4.09 4.25 100 4.01 6.27 10.55 103 D 1.54 4.09 5.64 100 8.57 6.27 15.63 105

W 0.47 4.09 4.66 102 5.03 6.27 10.13 90 D 2.16 4.09 6.80 109 9.18 6.27 15.84 103

Mean ( f s .d . ) recovery 103+3 Mean (fs.d.) recovey 9 9 f 5

a w, well-watered; D, water-stressed.'

A further observation of interest was that a new Dextropak cartridge purchased in September 1982 did not resolve proline and glycinebetaine. However, after about 4 months in use for sugar analysis, proline and glycinebetaine were completely resolved on the column. It would appear that the separation of these compounds required interaction with active sites on the column packing, which were protected on a new column, and the poor peak shapes obtained with pure water as the mobile phase would support this idea. With use, some of the previously capped silanol groups on Dextropak may become uncovered, providing sites for interaction with certain compounds. Another Dextropak column which had been used for over a year for analysing sugars in fermentation solutions gave the expected separations of proline and betaines.

Acknowledgements The author thanks Mrs L. Howse for skillful technical assistance, and Mr K. Dobson for providing a used Dextropak column.

886 C. W. Ford

References 1. Hanson, A. D.; Nelsen, C. E. Betaine accumulation and [“C] formate metabolism in water-stressed barley leaves. Plant

Physiol.‘ 1978, 62, 305-12. 2. Ford, C. W.; Wilson. J. R. Changes in levels of solutes during osmotic adjustment to water stress in leaves of four tropical

pasture species. Aust. 1. Plant Physiol. 1981,8,77-91. 3. Storey, R.; Wyn Jones, R. G. Quaternary ammonium compounds in plants in relation to salt resistance. Phytochemistry

4. Wyn Jones, R. G.; Storey, R. Salt stress and comparative physiology in the Gramineae. 11. Glycinebetaine and proline accumulation in two salt- and water-stressed barley cultivars. Aust. J . Plant Physiol. 1978, 5 , 801-16.

5. Davies, W. L.; Dowden, H. C. Betaine content and nitrogen distribution of beet molasses and other beet by-products. J. SOC. Chem. Ind. 1936, 55,175-9T.

6. Walker, H. G. Jr.; Erlandsen, R. Rapid method for determination of betaine. Analyt. Chem. 1951, 23, 13W11. 7. . Burham, J.; Coughlan, S. J.; Storey, R.; Wyn Jones, R. G. Estimation of quaternary ammonium and tertiary sulphonium

compounds by thin-layer eIectrophoresis and scanning reflectance densitometry.^J. Chromatogr. 1981, 210, 5 5 W . 8. Hitz, W. D.; Hanson, A. D. Determination of glycine betaine by pyrolysis-gas chromatography in cereals and grasses.

Phytochembtry (Oxf..) 1980, 19, 2371-4. 9. Gorham, J.; McDonnell, E.; Wyn Jones, R. G. Determination of betaines as ultraviolet-absorbing esters. AnaZyt. Chim.

Acta 1982, 138,27743. 10. Steinle, G.; Fischer, E. Hochdruckflussigkeitschromatographische Methode zur Bestimmungvon Betain in Zuckerfabriks-

produkten. Zuckerindustrze 1978, 103, 129-31. 11. Vialle, J , ; Kolosky, M.; Rocca, I. L. Determination of betaine in sugar and wine by liquid chromatography. J. Chrornatogr.

1981,204,429-35. 12. Singh, T. N.; Paleg, L. G.; Aspinall, D. Stress metabolism. I. Nitrogen metabolism and growth in the barley plant during

water stress. A u t . 1. Biol. Sci. 1973, 26, 45-56. 13. Wyn Jones, R. G.; Storey, R. In The Physiology and Biochemistry of Drought Resistance in Plants (Paleg, L. G.; Aspinall,

D., Eds), Academic Press, Sydney, 1981, pp. 171-204. 14. Ford, C. W.; Howse, L. Analysis of fermentable sugars in tropical grasses by high performance liquid chromatography.

Trop. Agron. Tech. Mem. Trop. Crops and Past. CSIRO Aust. 1982, No. 31. 15. Ford, C. W. Accumulation of low molecular weight solutes in water-stressed tropical legumes. Phytochembtry (Oxf.) 1984,

OX^.) 1977, 16, 447-53.

23.1007-1015.