Ascending Chromatography of Polyphosphates GEORGE G. BERG’ Departmenf of Biology, Brookhaven National Laboratory, Upton,
b Variables involved in the chromatographic analysis of polyphosphates were surveyed in terms of their effect on R, values. The variables included selection and concentration of components of solvent, composition of atmosphere, and condition of paper. An ascending method provided constancy of experimental conditions and permitted control of R, values in terms of three properties of the solvent: choice of alcohol, amount of water, and pH. Isopropyl alcohol solvents were suitable for lower polymers only. Data on chromatographic identification of higher polymers by other solvents were obtained.
I
of individual polymers of phosphate in low concentrations is becoming a pressing problem, both in toxicology (polyphosphates are widely used as food additives) and in biochemistry, where enzyme specificity is a t stake. Filter paper chromatography, which permits separation of polyphosphates of adjacent chain length, has been used to identify polymers up to four phosphate units (n = 4) in size (6, 14) and to separate additional polymers which probably carry the series up to n = 9 (3, 8, IO, I S ) , but matching standards for identifying spots of polymers above n = 4 are not a t hand. This work examines the relation between polymer size and RI, as well as the influence of experimental variables on R, values, and demonstrates some limitations of currently accepted solvent systems. DENTIFICATION
METHOD
Reagents and Apparatus. The following phosphate polymers were used: sodium tripolyphosphate, trimetaphosphate, and tetrametaphosphate (Victor); sodium pyrophosphate and “metaphosphate” (Fisher) ; sodium tetrapolyphosphate prepared by the method of Westman (15); and five high polymers prepared by Megirian ( 7 ) (iz = 11, iz = 20, ri = 27, iz = 47, n = 65, where iz denotes average chain length). “Metaphosphate” is a com1 Present address, Department of Pharmacology, University of Rochester School of Medicine and Dentistry, Rochester 20, N. Y.
L. I . , N. Y.
mercial designation for a mixture of linear polymers (polyphosphates), One-phase alcoholic solvents were freshly prepared and electrometrically titrated (the pH values in this report are Reckman null-point readings, made with electrodes 270 and 290). Each solvent was accompanied by a matching “aqueous phase” solution, which was the solvent mixture without alcohol. The filter paper was Schleicher & Schuell No. 589, Orange Ribbon. dpparatus consisted of 1000-ml. graduated cylinders plugged with paraffin-coated one-hole stoppers. A glass rod was carried through the cork and a filter paper strip (15’/2 X 11/2 or 2 inches) was suspended from a hook a t the end of the rod. Procedure. Two pieces of filter paper (11/2 x 12 inches) mere soaked in aqueous phase solution and pressed against the inside walls of the cylinder opposite each other. The chromatographic solvent (50 ml.) was on the bottom of the cylinder; care was taken to avoid contact between the chromatographic solvent on the bottom and the aqueous phase solution on the wall. The filter paper strip of the chromatogram was then suspended in the cylinder, out of contact with walls and fluid. The cork was sealed to the cylinder by melting the paraffin around the rim 1Tith an electric soldering pencil, and the suspended paper strip was equilibrated in the sealed atmosphere for 4 to 12 hours; this was in effect equilibration with the aqueous phase. Kext the glass rod was pressed donm and the chromatogram immersed, starting development in the chromatographic solvent. Development was stopped by air-drying over a heating fan. Temperature of the room was 21” =t1” c. Two-Dimensional Chromatogram. Two-dimensional chromatograms were developed on filter paper stapled into a roll l 1 3 j r inches in circumference. The roll was suspended from a stainless steel wire hanger, which was attached in turn t o a hook, rod, and stopper as in the one-dimensional procedure. The container was a cylinder with a glass lid; the stopper was mounted in a hole in the lid. The schedule of equilibration and development in each dimension was the same as in one-dimensional chromatograms. Visualization of Spots. The method of color development (9, 15) was modified to keep s. & s. paper from charring during hydrolysis of polyphosphates. Molybdate spray was prepared from 60-62% perchloric acid, 2.5 ml 1N hydrochloric acid, 5.0 ml.
470 ammonium molybdate, 12.5 ml.
(freshly prepared) Water to make 100 ml.
Sprayed chromatograms were suspended in a water-saturated atmosphere a t about 85” C., and hydrolysis was completed in 15 minutes or more. The end point was indicated by charring (yellow color) of a sprayed test strip, which was placed in the oven 5 minutes ahead of the chromatograms. After hydrolysis, maximum visualization of color was obtained by spraying with a strong reducing agent freshly prepared from stock (stock, 10 grams of stannous chloride in 25 ml. of concentrated hydrochloric acid; spray, 1 part of stock in 100 parts of 1N sulfuric acid). Proper handling resulted in a permanent pattern of blue spots on white background. EXPERIMENTAL VARIABLES
Variables involved in chromatographic separation of polyphosphates included the composition of the atmosphere, the brand and hydration of paper, and the composition of the solvent. These variables were surveyed in terms of R, values of phosphate polymers, primarily orthophosphate, pyrophosphate, and tetrametaphosphate. Composition of Atmosphere and Hydration of Paper. When paper was enclosed prior to development near a blotter soaked in the aqueous phase solution, it became hydrated and lengthened visibly in a matter of hours. This did not take place in an atmosphere of the alcoholic solvent alone (the paper lengthened only after immersion). Hydration of paper before development lowered Rl values (Table I, 1 ) . The presence of aqueous atmosphere during development increased R/ values. The combined effect of equilibration and development in aqueous atmosphere was tn.ofold: Chromatograms were “faster” (Table I, 2) ; and R , values n-ere more stable, as shown by the increased sensitivity of Rl to changes in pH when the aqueous atmosphere was m-ithheld (Figure 1). Schleicher 8: Schuell No. 589 Orange Ribbon paper was “slower” than washed Khatnian KO.4 Paper used by Ebel (4)(Table I, 3). p H and Water Content of Solvent. The response of R, values to changes of p H depended on the water content VOL. 30, NO. 2, FEBRUARY 1958
213
~
of the solvent; conversely, the RI response to changes in water content depended on pH. Eighteen solvents were tested, covering the p H range from 1.2 to 3.0 and water content range from 15 t o 35 parts. Figure 1 summarizes the effect of water content a t three pH's, and the effect of p H a t three dilutions, in a total of 13 solvent formulas. Figure 1 also presents the response of Rr t o pH and water content of solvent in the chromatographic method of Ebel (4). Ebel's pH values are based on measurement made in this laboratory (Ebel did not report pH) ; his chromatograms were on washed Thatman No. 4 paper, not equilibrated, and in an atmosphere of solvent (solvent: sum of volumes of isopropyl alcohol and water, 100 ml.; trichloroacetic acid, 5 grams; titrated with 22' BB. ammonia). In acid solvents, R/ values were higher, the higher the water content and the lower the pH, and R/ values for polymers approached the R, of orthophosphate as the upper limit. An exception to this rule was found in the region of pH 1.4 and 20 to 25 parts of water; there the spread of spots was somewhat increased, favoring chromatographic resolution. The formula of pH 1.4 and 20 parts of water was consequently adopted as the standard acid solvent for low polymers. Acids and Alkalies. Formic acid, trichloroacetic acid, and ammonia were tested singly and in combination, to determine their buffering power in solvents and their effect on R, values. I n acid solvents, the buffering action of formic acid (up to 22%) was not satisfactory; pH dropped by 0.4 unit or more during development. Trichloroacetic acid buffered well: 0.6 gram added to 100 ml. of solvent containing formic acid and ammonia stabilized pH within &0.05 unit. I n alkaline solvents, ammonia was a satisfactory buffer only in concentrations of 15 parts or more (of a 26' BB solution) per 100. [Ebel's (4) alkaline solvent containing 1 part of 22' B6. ammonia dropped from pH 11.5 to 10.9 during development.] R, values dropped when either formic acid (in acid solvents) or ammonia (in alkaline solvents) was added in place of water, while alcohol concentration and pH of solvent were kept constant. The apparent retarding effect of formic acid was due solely to the withdrawal of water; when formic acid was added to a solvent in which pH and alcohol-water ratio were kept constant (and consequently the alcohol concentration of solvent was lowered), R/ values showed little change. Addition of trichloroacetic acid did not change the volume of solvent, and 214
e
ANALYTICAL CHEMISTRY
Table I.
Effects of Equilibration, Atmosphere, and Paper
R/ Pyrophosphate
Orthophosphate 1. Aqueous atmospherea Equilibrated paper (S. & S.)* Dry paper (S. & S.) 2. Equilibrated paperC(S. & S.)
Aqueous atmosphere Alcohol atmosphere 3. Alcohol atmospherec S. & S. paper Washed Whatman No. 4 paper
Tetrametaphosphate
0.67 0.71
0.40 0.61
0.13 0.50
0.60 0.28
0.50 0.19
0.56 0.38
0.28 0 . 3 6 (4 j
0.19 0.31 ( 4 )
0.38 0.41 ( 4 )
a Solvent, Isopropyl alcohol, 65; water, 30; 89% formic acid, 5; 26" BB. ammonia, 0.2; titrated with trichloroacetic acid to pH 1.9. b Schleicher & Schuell No. 589 Orange Ribbon. c Solvent. Isopropyl alcohol, 40; isobutyl alcohol, 20; water, 39; 22" BB. ammonia, 1; pH 11.5.
Rf
30 PARTS WATER
.8 r
pH 1.4
L
KL .2
-
0
0 .8
r
25 WRTS WATER
-
pH 165
.8r
-
0
20 PARTS WATER
pH 1.95
I
\
.2h
-
0
"[
25 PARTS WATER x3txro*
I.4
p H l 2 13
ORTHO
.6
GRAHAM'S SALT
I
40
25 20 PARTS WATER
36
30
15
Figure 1 . Effect of pH and water content of solvent on Rf values of polyphosphates Top three graphs in each column represent standard acid chromatograms equilibrated and developed in aqueous atmosphere (solvent: sum of volumes of isopropyl alcohol and water, 9 5 ; 89% formic acid, 5; 26' Be' ammonia, 0.1 j titrated with TCA to indicated pHAot+om graph in each column presents results of Ebel (4). Each curve represents one polymer. Graham's salt in bottom graph was that ( 5 ) , in top graph was ii = 65 (7).
6 = 22
consequently had no effect on volume ratios. Formic acid, as well as trichloroacetic acid, suppressed spreading of spots in acid solvents (4). This action required relatively high concentrations of acid-e.g., 5 parts of 89% formic acid in 100 parts of acid solvent-and appeared unrelated t o either buffering power or pH reading (the latter was regulated with small increments of ammonia). Alcohols and Mixtures of Alcohols. Five alcohols were tested alone and in mixtures: methyl, isopropyl, n-butyl, isobutyl, and tert-amyl. I n the following series, the mixtures are listed in order of decreasing R f values: methanol-tert-amyl alcohol 3 : 1, methanol-n-butyl alcohol 5:1, same 4:1, same 3: 1, isopropyl alcohol, and isopropyl alcohol-isobutyl alcohol 2: 1. Smaller and more branched molecules gave solvents with higher Rf values and faster rates of advance of the front. Configuration outweighed molecular size as shown by relative ranking of nbutyl and tert-amyl alcohol. Effect of alchol on stability of solvent was checked by measuring drift of pH in acid solvents made up with formic acid but without trichloroacetic acid. Branched alcohols gave more stable solvents: tert-amyl alcohol was superior to n-butyl alcohol, and isopropyl alcohol to mixtures of methanol and n-butyl alcohol. I n well buffered solvents (with trichloroacetic acid) deterioration also took place but was revealed only by a drop of Rt values. Solvents deteriorated faster during a clromatographic run than in storage. Separation of Higher Polymers. T o learn whether rules controlling Izf values of lower polymers applied also t o higher polymers, solvents nere sought t h a t would move high polymers off the starting line. As a first step, acid solvents of the isopropyl alcohol series were "speeded up" by addition of water and decrease of pH, and in some cases by development on dry paper. Three high polymers \yere tested: metaphosphate (Fisher), and the ;f, = 11 and 6 = 65 preparations ( 7 ) . The results (Table 11) indicated that information about the size of high polymers could be derived from RI of spot and composition of solvent. It was possible to show that the ii = 11 preparation contained higher polymers than metaphosphate, as it lagged behind in solvents that moved both preparations off the starting line (Table 11). The 6 = 65 spot remained on the starting line in a solvent (not shown) which moved metaphosphate to an Rf above 0.4, but even the ii = 65 preparation could in turn be moved off the starting line and to an R f greater than 0.2 (Figure 1 and Table 11) by
Table II.
Polymer Metaphosphate E = 11 Metaphosphate E = 65
Comparison of
R,
Values of High Polymers
RP 0.03
Solvent Water, 30, pH 1.9 (dry paper) As above Water, 25, pH 1.3 Water, 30, pH 1.4
Solvent Water, 34, pH 1.9 (dry paper) As above Water, 30, pH 1.3 Water, 35, pH 1.4
0.01 0.01 0
Rfa
0 . Ogb
0.026 0.48* 0 . 23b
Measured to last spot given by each preparation; Rj of slovest component. Streaking, with R f values measured to end of streak. Solvents as in Figure 1; (dry paper) denotes that equilibration was omitted.
a
b
use of faster solvents. Thus, the rules for controlling Rl values by hydration and pH of solvent held good for high polymers. Standard Procedure and Lower Polymers. A standard procedure, utilizing an aqueous phase atmosphere, was calibrated for linear and ring polymers up to Fi = 4, and yielded reproducible Rfvalues (Table 111). The acid solvent resolved the six lower polymers (four linear and two ring compounds) from each other and from the larger polymers (Graham's salt), failing only to give separation of tetrapolyphosphate from trimetaphosphate (Table 111). The alkaline solvent (Table 111) resolved the same polymers, but failed to separate tripolyphosphate from tetrapolyphosphate. The alkaline solvent placed ring polymers ahead of all linear polymers, regardless of chain length. Of the low polymers, only trimetaphosphate was visibly unstable, yielding a second spot of orthophosphate in both solvents. The method yielded spots with less than 0 . 2 ~of phosphorus as orthophosphate. Larger loads were carried as sharp spots: in an acid chromatogram, orthophosphate loads of 0.4, 3, and 5 y of phosphorus yielded spots measuring, respectively, 13, 21, and 29 mm. in diameter. In two-dimensional chromatograme the alkaline solvent was used before the acid solvent (which contained tri-
Table 111.
chloroacetic acid). As R, values in each dimension remained unchanged, all six lower polymers were separated from each other and from the high polymers. Where experimental design required the initial run to be in acid solvent, the following formula was useful: methanol, 60; n-butyl alcohol, 15; 89% formic acid, 25; ammonia, 0.05 (Rl values: orthophosphate, 0.63; pyrophosphate, 0.40; tripolyphosphate, 0.18; trimetaphosphate, 0.13; tetrametaphosphate, 0.05). Debelopment was about twice as rapid as in the standard acid solvent, but the chromatographic pattern was inferior. DISCUSSION
Condition of Paper. Existing chromatographic methods for polyphosphates used paper equilibrated with the atmosphere of solvent. These were compared with paper equilibrated with a mater-saturated atmosphere. Observations of stretching of paper indicated that only paper that had been equilibrated in proximity t o a watersoaked blotter was fully hydrated a t the beginning of the run; all strips regardless of equilibration showed the full hydration stretch a t the end of the run. The previously published methods appeared then to employ a paper with properties that varied in the course of each run in a manner resembling
Standard Chromatographic Solvents for Lower Polymers
Acid Solvent Substance Analyzed Ammonium orthophosphate Tetrasodium pyrophosphate Sodium tripolyphosphate Sodium tetrapolyphosphate Sodium trimetaphosphate Sodium tetrametaphosphate Graham's salt
R f rt S. D. 0.70 f 0 073 0 55 f 0 097 0 41 f 0 043 0 29 f 0 045 0 24 f 0 033 0 08 f 0.025 0 1 0
Total of runs 14 15 15 3 7 10
6
Alkaline Solvent Total Rj f S. D. of runs 0 44 f 0 048 9 0 22 f 0 040 11 0 17 f 0 028 4 0.18 0 81 f 0.028 0 68 i 0 073 Of0
1 4 9
7
Acid solvent. Isopropyl alcohol, 7 5 ; water, 20; 89% formic acid, 5; 26' BB. ammonia, 0.1; titrated with trichloroacetic acid t o pH 1.4. Rate of ascent of front in acid solvent. 81/2 inches in 12 hours, lll/zinches in 24 hours, 13 inches in 34 hours (20-2' c.). Alkaline solvent, Methanol, 50; n-butyl alcohol, 10; water, 22; 26" BB. ammonia, 18; titrated with 25y0 formic acid t o pH 11.4. Rate of ascent of front in alkaline solvent. 14 inches in 14l/~hours (20-2' C . ) . -
VOL. 30, NO. 2, FEBRUARY 1958
215
“double front” development (11, p. 82). The data (Figure 1) indicated that the spread between “slower” and “faster” spots was most sensitive t o changes in solvent composition in chromatograms on nonhydrated paper, as nould be expected from this interpretation. Equilibration of the paper with the aqueous phase atmosphere was consequently adopted as standard procedure. Acids and Ammonia. R j values proved independent of the concentration of acid as long as p H and the alcohol-n ater ratio were kept constant. While various organic acids have been used in acid solvents for polyphosphate chromatography (4,Is), it now appears that changes in R j values w.ith choice of acid could be entirely accounted for by differences in p H of solvent. The highly volatile formic acid \?-asselected here to suppress streaking in acid solvents (11, p. 82), while the pH level was regulated by small increments of trichloroacetic acid and ammonia. The trichloroacetic acid-ammonia mixture, first introduced was shown t o have special by Ebel (i), buffering powers. There alkaline solvents were concerned, similar considerations applied to the selection of strong rather than weak ammonia mixtures, and to buffering with traces of formic acid. Standard Method. The standard method was derived from the method of Ebel (4). There were two significant modifications in acid solvents: (1) the equilibration step, which gave a fully hydrated strip prior t o development ( I f ? ) , and ( 2 ) the electrometric titration of the solvent to a constant end point (16). In alkaline solvents there was also a n improvement in the stability of p H during development. The method mas calibrated on pure low polymers and has since been applied to analysis of unknowns ( 2 ) . The level of reproducibility shown in Table I11 n-as reached in spite of rudimentary temperature regulation (drafts from air-conditioning fans alone were sufficient t o slon doivn the rate of advance of the front); wide changes !\-ere also allowed in such other parameters as the amount of sodium ion in a spot, duration of equilibration, duration of runs, and direction of cut (lengthwise or crosswise) when making paper strips. By contrast, Ebel ( 4 ) found that Rj values were not reproducible in his procedure unless he redistilled his alcohols; others, using Ehel’s method, found it necessary to resort t o Ro values t o correct for the variability of Rj values (8, 10). Control of R, Values. The “pH” values given here represented Beckman readings in alcohol and could not be evtrapolated to hydrogen ion concentration values; they did, how216
ANALYTICAL CHEMISTRY
ever, make it possible t o obtain solutions with reproducible chroniatographic properties (and, by inference, with reproducible hydrogen ion concentrations), and under proper circumstances could be taken to show the direction of change of actual pH. These circumstances prevailed in the pH/Rj curves (Figure l), where the alcohol concentration was constant and the acid-base content nearly constant. In choosing between acid and alkaline solvents it was found that both of Martin’s sets of conditions 111, p. 68) applied to polyphosphates. Alkaline solvents were preferable in resolving meta (ring) from linear polymers; in these solvents ring polymers were readily moved out from among other polymers and on to much higher R, values than the rest (4, 14, and Table 111). On the other hand, resolution of linear polymers from each other was more readily accomplished in acid solvents. R, values of linear polymers in acid solvents increased with drop in size of polymer, in pH, in alcohol-water ratio, and in niolecular size of alcohol. Even this simple system, however, involved five variables, and any formulation of rules implied keeping three of the five variables a t some arbitrarily constant level. Thus, a logarithmic relationship betn een Rj and polymer size, reported for acid solvents (8, I O ) , was also found in the standard isopropyl alcohol solvents used here; but all these solvents representtd a special case in which pH, alcoholnater ratio, and choice of alcohol were set to keep a11 but one or t n o spots below R j 0.5. It remained to h scen whether a logarithmic or sigmoid spacing would be found in the more general case (faster solvents), where spots n-odd be clustered near the front (R, = 1) as well as near the starting line. To devise acid solvents to fit polymers, a relatively narrow range of pH d u e s (1.2 t o 4) and alcohol-water ratios (60:10 to 7 5 : 2 5 ) was imposed by requirements of stability of polyphosphates and tightness of spots ( 1 , 4). For that range the generalization could he made that, for a given alcohol, there was a group of polymers nhich could be moved hetween starting line and front by changes in p H and alcoholmater ratio; polymers lower than the group would remain near the front, and those higher than the group near the starting line. I n the case of isopropyl alcohol solvent studied here, orthophosphate remained near the front; low polymers from 6 = 2 up could be manipulated by changes in p H and alcohol-water ratio along the curves plotted in Figure 1; and Graham’s salt polymers remained near the start-
ing line unless the solvent was pushed beyond its ability t o hold tight spots (Table 11). The resolution of higher polymers nould depend on a similar calibration of solvents based on alcohols faster than isopropyl alcohol. and requiring less extreme alcohol-water ratios and p H values. The Rj-pH curves (Figure 1) appear to be a more sensitive tool than Rf values alone in helping to identify these polymers, for LVhich matching pure conipouncls are not available.
ACKNOWLEDGMENT
The author thanks Jack Gross for guidance in problenis of chroniatography, and Robert Gosselin for contributing the polymers used in this research. The hospitality of llontrose J. Moses and the Department, of Biology, Brookhaven Sational Lahoratory, is gratefully acknon ledged.
LITERATURE CITED
(1) Bandurski, R. S., .iselrod, B., J . Biol. Chem. 193,405 (1951). (2) Berg, G. G., J . Histochetu. Cytochem. 4 , 424 (1956). . 26, 1383 (3) Crowther, J., ~ A L CHEII. (1984). (4) Ebel, J. P., Bull. S O C . chini. Francr 20, 991 (1953). (5) Ebel, J. P., Colas, J., Busch, S . , Ibid., 22, 1087 (1955). (6) Ebel, J. P., Volmar, T.! (‘otiipt. lend. 233, 415 (1951). (7) Gosselin, R. E., Ilegirian, R., J . Pharmucol. Erpti. l’herap. 115, 402 (1955). (8) Grunze, H., Thilo, E., ,.lAe Papier-
chromatographie der kondensierten Phosphate,” 2nd ed., Akademie-T’erlag, Berlin, 1955. (9) Hanes, C. S., Isherwood, F. A., ,$-atwe 164, 1107 (1849). (10) Karl-Kroripa, E., .IsAI.. CHEX 28, 1091 (1956). (11) Lederer, E., Lederer, JI.,,‘Chromatography,” Elsevier, Sen- l-ork, 1953. (12) Wade, H. E., JIorgan, I). JI,, Biochem. J. 60, 264 (1955;. (13) Westman, d. E. R., Crovither, J., J . .4/u. Ceram. Soc. 37, 420 (1954). (14) Westman, A. E. R., Scott, -1,E., S a l i u e 168, 740 (1951 I. (15) Krstnian, A. E. R., Scott, -1.E., Petllej-, J. T., (‘heiiii’cit,y in Con. 4 , 35 (1952).
RECEIVED for rrview Jul>-5 , 1956. *kccepted October 4, 1957. Research carried out at. Brookhaven Sational Laboratory under the auspices of the L-. S. Atomic Energv Commission, with partial aid of Grant‘So. G-3322 (C),U. S.Public Health Service, and completed at the Atomic Energy Project, University of Rochester, with partial aid of Grant No. -Ir10S9 (31and G), E.S. Public Health Service.
