ANALYTICAL

BIOCHEMISTRY

194,388-393

(19%)

The Concentration of Red Blood Cell UDPGlucose UDPGalactose Determined by High-Performance Liquid Chromatography

and

Michael J. Palmieri, Gerard T. Berry, Deborah A. Player, Shirley Rogers, and Stanton Segal Division of Biochemical Developmentand Molecular Diseases,The Children’sHospital of Philadelphia, 34th Street and Civic Center Boulevard, Philadelphia, Pennsylvania 19104; and Department of Pediatrics, University of Pennsylvania School of Medicine, Philadelphia, Pennsylvania 19104

Received

October

26,199O

We have developed a sensitive method that employs high-performance liquid chromatography to separate and quantitate uridine diphosphogalactose (UDPGal) and uridine diphosphoglucose (UDPGlu) in human red blood cells. The trichloracetic acid extracts of red blood cells were chromatographed using a Dionex CarboPac anion-exchange resin and a 20-40% potassium phosphate buffer, pH 4.6, in a gradient-elution program. UDPGal and UDPGlu were detected spectrophotometritally at 264 nm. Recoveries of UDPGal and UDPGlu ranged from 96 to 106%. Under these conditions, there was exceptionally good reproducibility in stored specimens, and the method was sensitive in the low picamole range. The mean values and standard deviations in adults were 2.9 f 0.4 and 7.8 f 0.8 rmol/lOO g Hgb for UDPGal and UDPGlu, respectively. The values in children were 4.5 f 1.2 and 10.2 A 1.7 Hmol/lOO g Hgb for UDPGal and UDPGlu, respectively. Values determined by the HPLC method are in excellent agreement with those obtained by enzyme analysis. Q 1~91 Academic Preee.

Inc.

Uridine diphosphoglucose (UDPGlu)l and uridine diphosphogalactose (UDPGal) play pivotal roles as precursors in the synthesis of complex carbohydrates and glycolipids as well as lactose. A major concern from a clinical point of view, however, has been their role in the intermediatry metabolism of galactose. This pathway involves the reaction of galactose l-phosphate with 1 Abbreviations used: UDPGal, uridine diphosphogalactose; UDPGlu, uridine diphosphoglucose; ADP, adenosine diphosphate; IMP, inosine monophosphate, CMP, cytidine monophosphate; UMP, u&line monophosphate; Hgb, hemoglobin; TCA, trichloroacetic acid.

UDPGlu to form UDPGal via galactose-l-phosphate uridyltransferase with the subsequent conversion of UDPGal to UDPGlu mediated by UDPGal-4-epimerase. The defective activity of both of the involved enzymes resulting in clinical disorders of galactosemia has provided an impetus to the development of methods for assessment of tissue concentrations of these compounds. Enzymatic methods for the determination of UDPGlu by reaction with UDPGlu dehydrogenase (1) coupled with the conversion of UDPGal to UDPGlu and assay of the latter with the dehydrogenase is well established. Keppler and Decker (2) and Berman et al. (3) employed the addition of glucose l-phosphate and reagent galactose-l-phosphate uridyltransferase to convert UDPGal to UDPGlu in liver. Maxwell (1) utilized reagent UDPGal-4-epimerase to convert UDPGal to UDPGlu. Fujimura et al. (4) adopted this method to measure the UDPGal level in whole blood spots. Employing a modification of the latter method, Ng et al. (5) have reported the levels of both UDPGlu and UDPGal in normal human red cells, liver, and cultured fibroblasts and demonstrated that there is a deficiency of UDPGal in tissues of galactosemic infants. Considerable differences in the concentration of the nucleotide sugars in normal red blood cells have appeared which are highlighted by the findings of Simmonds et al. (6) and Kirkman (7) whose values appear to be a fifth of those reported by Ng et al. (5) and one twenty-fifth those of Fujimura et al. (4). This broad range of reported normal red blood cell values and the importance placed by Ng et al. (5) on the low red blood cell UDPGal levels in uridyltransferase-deficient galactosemics with regard to the etiology of complications of the disorder led us to devise an HPLC method for separation and quantitation of UDPGlu and UDPGal with-

388 All

Copyright 0 1991 rights of reproduction

oLm3-2697/91 $3.00 by Academic Press, Inc. in any form reserved.

CHROMATOGRAPHIC

ANALYSIS

out resorting to coupled enzyme manipulations. sults form the basis of this report. MATERIALS

AND

OF

UDPGALACTOSE

AND

-GLUCOSE

IN

RED

CELLS

389

Our re-

METHODS

Materials All solvents and buffers were HPLC analytical reagent quality chemicals and were purchased from EM Science (Cherry Hill, NJ). The water used in all preparations was greater than 18 Mohm in purity (Marcor Co., Harleysville, PA). Uridine diphosphogalactose, uridine diphosphoglucose, the oxidized form of nicotinamide adenine phosphate (NAD), and the reduced form (NADH) were purchased from the Sigma Chemical Co. (St. Louis, MO). Uridine-5’-diphosphoglucose dehydrogenase was a highly purified preparation purchased from the Sigma Chemical Co. UDPGal and UDPGlu contained cations, water of hydration, and solvent residues with a calculated molecular weight of 670 (theoretical, 566). Standard solutions were prepared taking the formula weight into consideration.

Methods Blood was obtained from 12 fasting adult volunteers ranging in age from 27 to 63 years, and from 16 fasting children from 1.5 to 12 years. The children were already having venipuncture performed prior to the scheduled day of surgery. The acquisition of blood for these determinations was approved by the Children’s Hospital IRB and consent was obtained from the parents. Two to 8 ml of blood was drawn into heparinized tubes which were immediately placed in an ice-water mixture. An aliquot of blood was used for the determination of hemoglobin (Hgb) and hematocrit by automated Coulter counter analysis. Subsequently, packed red blood cells were prepared by centrifugation for 5 min at 9OOgat 5°C. Hemolysates were made by mixing 1 vol of packed red blood cells with 2 vol of ice-cold water, and the hemoglobin was determined calorimetrically using the Sigma diagnostic kit (No. 625-A). For analysis of UDPGal and UDPGlu, 1 vol of packed red blood cells was added to 2 vol of ice-cold trichloroacetic acid (TCA). After centrifugation at 30,OOOgfor 20 min at 5”C, the supernate was collected and the TCA was extracted with 8 X 3 vol of water-saturated ether. Residual ether was removed under a stream of nitrogen, and the final pH of the aqueous solution was 5.0. Further partitioning with ether failed to increase the pH above 5.0. High-performance liquid chromatography was carried out on a Beckman Model 126 solvent delivery system equipped with a Model 166 uv detector and a Dionex CarboPac PA-l anion-exchange column (4.9 X 250 mm). Eluants containing potassium phosphate were filtered through a 0.22~pm nylon filter prior to use. Samples were injected with a Beckman Model 506 auto-

pm01 FIG. 1.

Separation by HPLC of various concentrations of UDPGal and UDPGlu on a Dionex CarboPac anion-exchange resin. Elution was performed with a 20-40% concave gradient formed from water and potassium phosphate buffer, pH 4.5, as described in the text. These are actual tracings of the elution patterns generated with known standards having actual elution times of 23.5 and 24.5 min for UDPGal and UDPGlu, respectively.

sampler via a Rheodyne injector valve equipped with a 20-~1 sample loop. Chromatographic data were plotted and integrated using a Hewlett-Packard 3390A integrator. Separation of UDPGal and UDPGlu was achieved using a 20-40% concave gradient (Beckman preprogrammed Curve No. 4) formed from water (eluant A) and 0.5 M potassium phosphate buffer, pH 4.5 (eluant B). A flow rate of 0.6 ml/min was maintained over a period of 30 min. The column was initially equilibrated with a 20% mixture of water and potassium phosphate buffer pH 4.5. Nucleotides were detected at a wavelength of 254 nm. UDPGlu in TCA filtrates was also quantitated by a modification of the method of Keppler and Decker (2) originally described by Kirkman and Maxwell (8). The stoichiometric formation of NADH generated from the enzymatic conversion of UDPGlu to uridine diphosphoglucuronic acid by UDPGlu dehydrogenase was measured fluorometrically. Results are expressed in two ways: (a) micromoles UDPGal or UDPGlu per 100 g Hgb, the latter being measured directly in the red blood cell lysate; and (b) micromoles UDPGal or UDPGlu per liter of red blood cells, where the red blood cell volume was calculated from the Coulter counter indices. RESULTS

Separation

of UDPGal

and UDPGlu

Figure 1 shows the elution profile of standard UDPGal and UDPGlu. There is a well-defined separation between the two compounds, the retention times being 1 min apart. When multiple injections (120 pmol) of the same standards were made, the integrated areas under each peak were reproducible. The coefficient of variation of each nucleotide based on 66 injections of 20 ~1 each was 4.2%. This indicated that although baseline

390

PALMIERI

ET AL.

Recovery In a series of four studies, addition of 4.5 and 9.0 nmol of UDPGal and UDPGlu, respectively, to 3 ml of a 10% TCA-red blood cell suspension resulted in a recovery of nucleotides ranging from 96 to 106%. Replication

I 20

30

Time FIG. 2. Typical elution profile mixing 1 vol of packed red blood minutes.

of a protein-free filtrate prepared by cells with 2 vol of 10% TCA. Time in

Preparation

of Red Blood Cell Filtrates

Two methods for preparing red blood cell filtrates for enzymatic analysis have been utilized by various investigators. These include (a) perchloric acid precipitation of protein (2), and (b) hemolysis of packed red blood cells followed by heating in a boiling water bath to precipitate all protein (5). We attempted to separate UDPGal and UDPGlu by HPLC in samples prepared by both methods. Unfortunately, both methods proved to be inadequate. Perchloric acid-treated samples neutralized with KOH permitted no separation of UDPGal and UDPGlu. Apparently, enough ions remained in solution to grossly distort the chromatographic elution pattern. On the other hand, separation of UDPGal and UDPGlu was achieved in samples prepared by hemolysis followed by boiling for 10 min. Values of UDPGal and UDPGlu obtained on five samples were comparable to those determined with the TCA-precipitation method. However, after several analyses the anion-exchange column failed to give adequate resolution of the compounds. Perhaps the presence in the filtrate of some

Sensitivity

TABLE Sensitivity

Various dilutions of a TCA filtrate from a normal adult containing 4.4 and 9.5 ~mol/lOO g Hgb of UDPGal and UDPGlu, respectively, were made in water and subsequently analyzed by HPLC. We found that sample dilutions up to fivefold could be quantitated with a high degree of accuracy (Table 1). The ratio of UDPGal to UDPGlu remained constant in this dilution range. At higher dilutions (lo-fold), UDPGal and UDPGlu could not be quantitated accurately, and there was a sharp decrease in the ratio of UDPGlu to UDPGal.

on Storage

Table 2 shows the results of analyses on two different red blood cell filtrates. Filtrates 1 and 2 were analyzed on five different occasions, with little or no variation in UDPGal and UDPGlu. During this span of 32 days, the same specimen was frozen at -40°C and subsequently thawed before each analysis. Two other filtrates were treated in the same manner and similar results were obtained. It is apparent from these studies that (a) there is little variation in the concentrations of UDPGal and UDPGlu, and (b) the filtrates are very stable over a prolonged period of time. Several (at least five) freeze-thaw cycles had no effect on the stability of the nucleotides. Optimum

separation between UDPGal and UDPGlu was not achieved, the integrator was able to discern the quantities of each compound. Figure 1 also demonstrates the linearity of concentration with peak areas. Injections of 30 to 240 pmol of each compound revealed a constancy of separation and a linear relationship of the areas of each nucleotide sugar with concentration. Concentrations as low as 12 pmol of each were accurately determined by increasing the sensitivity of the integrator. Figure 2 shows the typical elution pattern of 20 ~1 of protein-free filtrate prepared by the addition of 1 vol of packed red blood cells to 2 vol of 10% TCA. The separation of UDPGal and UDPGlu is clearly well established, and retention times are identical to those obtained in the standards. Retention times for UDPGal and UDPGlu on a typical CarboPac anion-exchange resin usually were around 23.5 and 24.5 min, respectively. The other peaks labeled in the figure were identified by comparison with the retention times of standard nucleotide sugars.

and Stability

of HPLC

Determination

1

of UDPGlu

and UDPGal

Dilution

UDPGlu (pmol)

UDPGal bmol)

Ratio

0 2.5 5.0 10.0

206.9 87.1 44.4 13.0

92.1 37.1 18.0 3.3

2.36 2.40 2.50 1.57

Note. A TCA filtrate of red cells from a normal adult was diluted with water as indicated. The absorbance unit full scale was decreased from 0.025 to 0.005 to accommodate the dilution.

CHROMATOGRAPHIC

ANALYSIS OF UDPGALACTOSE

391

AND -GLUCOSE IN RED CELLS

TABLE 2 Reproducibility Sample

of Red Cell Uridine Nucleotide Sugars Analyzed by HPLC Sample 2*

1”

Area’ Day

Area’

of

analysis

Day

UDPGlu

UDPGal

Ratio

analysis

UDPGlu

UDPGal

Ratio

1 1 1 13 32

535,666 536,548 547,113 520,244 545,788

215,468 188,933 216,395 196,207 193,199

2.49 2.84 2.53 2.65 2.82

537,072 +10,750

202,040 +12,946

2.67 +0.16

1 3 3 10 22

422,962 419,838 411,351 427,472 409,858

207,462 208,317 209,239 217,031 199,972

2.04 2.02 1.96 1.97 2.05

Mean

418,296 &7,546

208,404 +6,065

2.01 kO.04

SD

of

n Filtrates after Day 1 of analysis were kept at -40°C. Subject: 6 years * Part of recovery experiments: 4.5 nmol UDPGal and 9.0 nmol UDPGlu old. ’ Arbitrary units.

nonprecipitable, low-molecular-weight peptides bound to the column interfered with subsequent chromatographic analyses. Column regeneration using acid-base recycling, although restoring the resolving capabilities of the resin, shortened the life span considerably. TCA proved to be the method of choice for both the preparation of the sample and the chromatographic separation of UDPGal and UDPGlu. Samples prepared by

the TCA method were found (a) to give good resolution of UDPGal and UDPGlu; (b) to be stable over prolonged periods of time at -4O”C, even when subjected to repeated freeze-thaw cycles; (c) to have excellent chromatographic reproducibility; and (d) to maintain the integrity of the anion-exchange resin over hundreds of analyses. The influence of blood handling was also examined. Probably the single, most important factor responsible for poor chromatographic resolution of UDPGal and UDPGlu was the quantity of compound present in the peak just prior to elution of UDPGal. An analysis of standards indicated this to be inosine monophosphate (IMP) (Fig. 2). Large quantities of IMP produced peak tailing which, in turn, caused baseline shifting in the region of UDPGal. Thus, integration of the UDPGal peak was severely affected. Since formation of IMP is a consequence of the deamination of AMP by adenylate deaminase, special precautions were taken to minimize this phenomenon. These included (a) placing the blood in an ice-water mixture until the TCA filtrate was prepared, usually within 60 min after drawing, and (b) eliminating the washing of the red blood cells with physiological saline. The results of two studies demonstrated that washing with 3 X 3 vol of 0.085% NaCl over a period of 20 min dramatically increased IMP production. Immediate freezing of packed red blood cells on dry ice also proved to be inadequate for obtaining subsequent separation and quantitation of UDPGal. Analysis

old. were added

to 3 ml of TCAhed

blood

cell mixture.

Subject:

47 years

of six samples prepared from packed red blood cells within 60 min and also after freezing of the cells on dry ice, revealed a marked increase in IMP. This impaired the quantitation of UDPGal such that the values were increased from 25-lOO%, while the UDPGlu values remained the same. In one frozen sample, quantitation of UDPGal or UDPGlu could not be achieved.

Concentration of UDPGal and UDPGlu Human Red Blood Cells

in Normal

The levels in red blood cells of 12 adults and 16 children ranging in age from 1.5 to 8 years are shown in Fig. 3A. The mean values and standard deviations in adults were 2.9 + 0.4 and 7.8 + 0.8 pmol/lOOg Hgb for UDPGal and UDPGlu, respectively. The values in children were 4.5 f 1.2 and 10.2 + 1.7 for UDPGal and UDPGlu, respectively. The difference between the UDPGal and UDPGlu values in adults and children is significantly different with a P < 0.001. The ratio of UDPGal to UDPGlu is 1:2.7 + 0.24 in adults, while in children it is 1:2.3 f 0.40. These values are significantly different with P < 0.01. To compare our results with those reported by several other investigators (6,7), we have also expressed our data as micromoles per liter of red blood cells (Fig. 3B).

Comparison

with the Enzymatic

Method

Table 3 shows a comparison of UDPGlu values determined by the HPLC and enzymatic methods on TCA filtrates from five normal adults. Whether the results are expressed as micromoles/lOO g Hgb or micromoles per liter of red blood cells, there is excellent agreement between both methods. Enzyme values range from 96 to 118% of those obtained from the HPLC analysis.

392

PALMIERI

ET

AL.

A UDPQIU

14.0-

. 13.0-

B UDPQIU

12.0-

UDPGal

4&O11.0.

I lO.O-

40.0 -

f

C 3 so-

36.0.

B g aoI

g 30.0.

0” 7.0? a g BO-

::g m

: b

2&O-

4.0-

i

l

r

10.0

’

2.0-

0’

-A-

Adult8

Chlldren

Adults

Levels of UDPGal and UDPGlu in normal g Hgb (A) and Fmol/liter red blood cells (B).

Children

We describe a sensitive method for the quantitation of UDPGal and UDPGlu in normal red cells which, with suitable changes, could be applied to other tissue. The advantage of the HPLC method is that it is a direct chemical assay of UDPGal and UDPGlu. There is no need to measure the production of NADH generated in the one-step reaction of UDPGlu with UDPGlu dehydrogenase or the two-step procedure used in the analyTABLE Cell UDPGlucose and Enzymatic

pmol/lOO 1 2 3 4 5

8.0 8.0 7.4 7.9 7.3

i . i

t

+

:

Enzyme g Hgb 7.7 9.1 a.7 8.2 7.5

3 Values Methods

Determined

HPLC rmol/liter 26.0 25.9 25.5 27.9 25.2

1 Adults

Children

Adults

Chlldrm

adults and children as determined by the HPLC method. The horizontal line represents the mean of each group.

DISCUSSION

HPLC

I

6.0

to-

Sample

:

i5 5 16.0 !

B ;

3.0-

HPLC

l

:

d 20.0 1 L

of Red

+

0

6.0 -

Comparison

: ;

J

P

FIG. 3. pmol/lOO

-L: . .

by

Enzyme cells 25.1 29.5 29.8 29.0 25.7

Note. Both analyses were performed on the same TCA filtrate of five normal adult subjects. The enzyme analysis was performed using UDPGlucose dehydrogenase with fluorometric measurement of NADH resulting from the reaction described under Materials and Methods.

Values

are expressed

as

sis of UDPGal with UDPGlu dehydrogenase and UDPGal-4-epimerase. The conditions for separation on the Dionex anionexchanger column must be rigidly adhered to since sample preparation, flow rates, and column regeneration are critical. Ions in the cell filtrates prepared by the perchloric acid-extraction method, residual protein in boiled extracts and impurities in the TCA or ether can dramatically alter the separation of UDPGal and UDPGlu. Cells immediately frozen on dry ice and stored at -4O”C, those left at room temperature for 24 hr, or those washed with physiological saline generate IMP which, at high concentrations, interferes with the elution of UDPGal. The wide range of reported results for red cell UDPGal and UDPGlu is shown in Table 4. These values, varying widely, have been reported as milligram percent whole blood (4), micromoles per liter red blood cells (6,7), and micromoles per 100 g hemoglobin (5) according to the investigator. The only previous report of an analysis of UDPGlu by HPLC in the red blood cells of children was that of Simmonds et al. (6) whose value of 36 -t 8 rmoles/liter cells was similar to our value of 31.6 + 5.2. They did not report any data for UDPGal. Our results for both of these nucleotide sugars determined per liter of packed cells agree closely with those of Kirkman et al. (7) determined enzymatically. As noted in Table 3, our values for UDPGlu measured by the HPLC

CHROMATOGRAPHIC

ANALYSIS

OF

UDPGALACTOSE TABLE

Normal Red Cell UDPGlu Investigator

Fujimura et al. (4) Ng et al. (5) Kirkman (7) Simmonds et al. (6) Palmieri et al.

’ Number of analyses. ’ Mean + SD. ’ Converted from values d Adults and children. = Adults. ’ Children.

Method

(N)”

in whole

26.0 36 25.9 34.3

blood

as ma%. I

IN

RED

393

CELLS

and UDPGal Concentrations UDPGal

(pmol/liter + 3.2 +8 + 2.5 + 5.2

mean.

-GLUCOSE

4

UDPGlu

Enzyme (14) Enzyme(14) Enzyme (8) d HPLC (15) HPLC (12)e HPLC (16)’

renorted

AND

UDPGlu

UDPGal (pmol/lOO

red blood cells)’ 190 (45-298)

38.1 + 7.1

8.5 k 3.6 9.7 -c 1.5 15.1 + 4.2

g Hgb) * 25 (6-39)’ 11.2 ‘_ 2.5”’ -

7.8 + 0.8 10.2 k 1.7

2.9 + 0.4 4.5 + 1.2

and range.

method were also confirmed by us with the enzymatic procedure. When calculated on the basis of hemoglobin, they differ by a factor of three- to fourfold from those reported by Ng et al. (5), who used the enzymatic procedure, and many-fold more from the recalculated values of Fujimura et al. (4). The values reported by Ng et al. differ by fivefold from those of Kirkman (7) and from our own enzymatically determined values. The average values for red cell UDPGal and UDPGlu of children are significantly higher than those of adults, although there is overlap in the range for both (Fig. 3). Factors determining the steady-state levels are, indeed, complicated. The rates of formation of UDPGlu via UDPGal pyrophosphorylase from glucose l-phosphate and UTP, and of UDPGal and UDPGlu from galactose l-phosphate via galactose-l-phosphate uridyltransferase, are determinants. Obviously, an important factor is UDPgalactose-4-epimerase which catalyzes the interconversion of UDPGlu and UDPGal (1). The enzyme has an equilibrium value of 3 for UDPGluIUDPGal in yeast and calf livers, and 3.5 in bovine mammary gland (9). We have found that the average ratio of UDPGlu/ UDPGal in adult red cells is 2.7 which differs significantly from the 2.3 seen in children. The lower ratio in children is due to a relatively greater increase in UDPGal in the cells of younger subjects. Whether diet plays a role in the levels of UDPGal and UDPGlu is unknown. Although the subjects were in the fasting state, prior diet may be a factor in that children might be expected to have a greater milk and galactose intake thus supplying an additional source of UDPGal. Bergren et al. (10) have shown that epimerase activity is high in human newborn red cells, but that activity of older children’s cells does not differ from that of adult cells. The levels of UDPGal and UDPGlu achieve great importance in relation to the observation of Ng et al. (5) that UDPGal concentration of cells from uridyltransferase-deficient galactosemics is significantly lower

than normal. This has generated the hypothesis that long-term complications and poor outcome of galactosemit patients well-treated by dietary galactose restriction is due to a deficiency of an important precursor of glycoproteins and galactolipids. As treatment, the administration of uridine to boost UDPGal levels has been recommended (5). Since these important considerations depend on the observation of low UDPGal in red cells, the findings of Ng et al. (5) should be verified by the HPLC method, especially since there is such a great discrepancy in the normal values reported by them from those found by Kirkman (7), by Simmonds et al. (6), and by us. ACKNOWLEDGMENT This work was supported by a grant from the National Diabetes, Digestive, and Kidney Diseases, DK 42785.

Institute

of

REFERENCES 1. Maxwell,

E. S. (1957)

J. Biol.

Chem.

229,

139-151.

2. Keppler, D., Rudigier, J., and Decker, K. (1970) Anal. Biochem. 38,105-114. 3. Berman, W., Rogers, S., Bautista, J., and Segal, S. (1978) Met&olism 27,1721-1731. 4. Fujimura, Exp. Med.

Y., Kawahura, 14 1,263-268.

M.,

and Naruse,

H. (1983)

Tokoku

J.

5. Ng, W. G., Xu, Y. K., Kaufman, F. R., and Donnell, G. N. (1989) J. Inherited Metab. Dis. 12, 257-266. 6. Simmonds, H. A., Fairbanks, L. D., Morris, G. S., Webster, D. R., and Harley, E. H. (1988) Clin. Chim. Actu 171,197-210. 7. Kirkman, H. N. (1990) J. Pediatr. 117,838-839. 8. Kirkman, H. N., and Maxwell, E. S. (1960) J. Lab. Clin. Med. 56, 161-166. 9. Tsai, C. M., Holmberg, U., and Eboner, K. E. (1970) Arch. Biothem. Biophys. 136,233-244. 10. Bergren, W. R., Ng, W. G., and Donnell, G. N. (1973) B&him. Biophys. Actu 315,464-472.