Assay of calmodulin with Bordetella pertussis adenylate cyclase

ANALYTICAL

BIOCHEMISTRY

124,45-52

Assay of Cal~odulin

(1982)

with ~r~~feila

ALAN GOLDHAMMER National Institute

of

Arthritis,

~~ffssis

Adenylate

Cyclase

AND J. WOLFF

Diabetes, and Digestive and Kidney Diseases, Bethesda, Maryland 20205 Received December 22, 1981

Low levels of the calcium-de~ndent regulator protein, calm~ulin, may be measured utilizing membranes prepared from 3ordetelIa pertussis which contain an adenylate cyclase which is activated by this protein. The activation is dose dependent and tissue levels of calmodulin can be determined over a range from 2 pg to 100 ng with good reliability. We demonstrate how this bioassay may be employed to measure the levels of calmodulin in a variety of protein and cellular preparations.

The calcium-dependent regulator protein, calmodulin, is found in most eucaryotic cells and regulates the activity of a number of enzymes (l-3). Because of the postulated importance of calmodulin in cellular metabolism, a number of laboratories have sought reliable methods for measuring this protein. Several reports have appeared using a radioimmunoassay for calmodulin (4-6). A second method is to utilize the brain phosphodiesterase bioassay as outlined by Cheung and co-workers (7). Each of these methods presents problems. On the one hand, antibody preparation can be compromised by the highly conserved structure of calmodulin leading to poor antibody production, while brain phosph~iesterase is an unstable protein, and its purification is tedious. Recently, our laboratory observed that the adenylate cyclase of Bordetella pertussis is highly stimulated by calmodulin (8). Furthermore, we indicated that the organisms could be used to detect calmodulin in various protein preparations by measuring the amount of adenylate cyclase activation (9). Although the response to calmodulin is satisfactory, there are problems in using intact B. pertussis organisms: the organisms are pathogenic; activity is variable from preparation to preparation; and the enzyme activ-

ity of the organisms displays various degrees of stability with prolonged storage. Because the purified adenylate cyclase is only poorly responsive to calmodulin (lo), an alternative method was sought that would take advantage of this organism’s excellent response to the activator while circumventing the above-mentioned problems. In this communication, we report that crude membrane preparations from 3. ~ertussis spheroplasts can be utilized for a highly sensitive calmodulin bioassay. Furthermore, such preparations are stable and provide more sensitivity than either the radioimmunoassay or assays using brain phosphodiesterase. We demonstrate how this bioassay can be used to determine the calmodulin content from a variety of sources, including protein preparations, plasma membranes, and cytosols from cultured cells. EXPERIMENTAL

PROCEDURES

Chemicals and protein preparations were from the following sources: adenosine S-triphosphate, cyclic adenosine 3’:5’-monophosphate, creatine phosphokinase, cr-amylase, soybean trypsin inhibitor, acetyl phenylhydrazine, and deoxyribonuclease type I (bovine pancreas) were from Sigma Chemical 45

~3-269?/82/

Il~45-08$02.~/0

Copyright 6 1982 by Academic Press, Inc. All rights of reproduction in any form reserved.

46

GOLDHAMMER

Company. One sample of creatine kinase was purchased from Boehringer-Mannheim. [cu-~*P]ATP, 25 Ci/mmol, was from ICN and [3H]cAMP, 16 Ci/mmol, was obtained from New England Nuclear. Cholera toxin was from Calbiochem. Two-cycle tubulin was a gift from Dr. Y. C. Lee (NIADDK). Purified clathrin and bovine brain coated vesicles were from Dr. P. K. Nandi (NIAD~K). Tetr~~~~enu ~~~~~~~~ and peanut calmodulin were generously supplied by Dr. T. C. Vanaman (Duke University) and octopus calmodulin was a gift of Dr. K. Seamon (NIADDK). B. pertussis (strain 114) were grown in Stainer-Scholte medium ( 12,13). After 20 h of growth, organisms were harvested by centrifugation. Organisms not utilized immediately were stored as a paste at -20°C. Crude membranes were prepared in the following manner: 5 g of either thawed B. pertussis paste or fresh culture isolates were suspended in 100 ml of 20% sucrose, 10 mM Tris-Cl, 10 mM EDTA, pH 8.0, A smooth suspension was obtained by treating them with 5 strokes in a Dounce-type homogenizer using the loose-fitting pestle. Lysozyme (Worthington), 15 mg, was added and the suspension was kept at 25’C for 30 min. The organisms were centrifuged at 4°C at 12,OOOg for 10 min. The resulting fluffy pellet was suspended in a Dounce homogenizer in 6 ml of 20% sucrose, 20 mM MgC& together with 100 pg of deoxyribonuclease; then, 80 ml of 10 mM Tris-Cl, 5 mM MgC12, pH 8.0, was added and the spheroplasts were lysed by a few strokes of the loose-fitting pestle. The resultant membranes were collected by centrifugation at 40,OOOg for 30 min, washed once with 10 mM Tris, 5 mrvr MgC12, pH 8.0, buffer and stored at -20°C at a protein concentration between 10 and 20 mg/ml. AdenyZute cyciase assay. Adenylate cyclase activity of the B, pertussis membranes was determined by the method of Salomon et al. (14) measuring the production of [32PJ~AMP from [ CX-~*P]ATP. The reaction

AND WOLFF

was carried out in a final volume of 60 ~1 and contained 60 mM Tris-Cl (pH 8.0), 10 mM MgC12, 50 PM CaC&, 0.16 mg/ml human serum albumin, and 1.0 mM ATP containing 0.3 &i [a-32P]ATP per assay tube. No regenerating system is required for this assay. Assays were carried out at 30°C for 10 min and initiated by substrate addition following a 1-min preincubation at the assay temperature. Reactions were stopped by addition of 100 ~1 “stopping solution” containing 1% sodium dodecyl sulfate, 20 mM ATP, and 6.25 mM CAMP (containing 1O,O~ cpm [3H]cAMP in each tube for calculation of recovery). Calmodulin was purified to homogeneity following a modification of the method of Klee (11) starting with a high-speed supernatant of bovine brain homogenate. Calmodulin standards were routinely prepared using 1 mg,/ml human serum albumin (Schwarz-Mann) as a carrier protein to prevent losses of the protein at low calmodulin concentrations. Reticulocytes were collected from Sprague-Dawley rats by cardiac puncture following induction of reticulocytosis by subcutaneous injections of 0.5 ml acetyl phenylhydrazine (70 mgfkg) for 3 consecutive days. On Day 7 after the start of the injection schedule, the reticulocyte percentage was maximal as determined by staining with methylene blue. S-49 mouse lymphoma cells were grown in Falcon bottles in Ham’s F-10 medium containing 10% horse serum, 3 mM glutamine, penicillin (100 units/ml), and streptomycin (100 ~g/ml). Y-l mouse adrenal tumor cells, the OS-3 line, and I-10 Leydig tumor cells were grown in the same medium which was further supplements with 2.5% fetal calf serum. The atmosphere was watersaturated 5% C02-95% air at 37’C. Plasma membranes were prepared by the following method. Cells were harvested by scraping in cold 0.25 M sucrose solution containing 10 mM Tris-Cl (pH 7.5) and 1 mM dithiothreitol. Low-speed pellets were pooled,

CALMODULIN

DETERMINATION

recentrifuged, and homogenized in the sucrose solution in a tight-fitting Dounce homogenizer until more than 90% of the cells had been disrupted as judged by light microscopy. Homogenates were freed of heavy particles by ~nt~fugation on a 40% sucrose cushion for 45 min at 2°C at 155,OOOg in a Beckman SW 41 rotor. The interface was collected, centrifuged in the sucrose buffer, resuspended, and stored in liquid nitrogen. Prior to the assay for calmodulin, aliquots of membrane were suspended in 150 mM NaCl, 10 mM Tris-Cl, 5 mM MgC12, pH 7.5, and washed with this buffer three times. Red cell membranes were prepared by hypotonic lysis after washing and suspending the cells in 150 mM NaCl, 10 mM Tris-Cl, and 5 mM MgCIZ (pH 7.5). Intracellular calmodulin was determined on lysates which were prepared in the following manner. Cells were collected in 10 mM Tris-Cl, 150 mM NaCl, pH 7.4, washed three times, suspended at concentrations of lo7 or lOa cells/ml, and disrupted by sonication after the addition of Triton X-100 (0.1% final concentration). In some cases, EGTA’ (1 mM final concentration) was added. A Branson sonifier with a microprobe was employed at a power setting of 3 for 10 s. Disruption of cells was confirmed by examination under a light microscope. Debris was cleared by cent~fugation 27,~g for 15 min. Some samples were incubated at 90°C for 5 min before centrifugation. Calmodulin determinations were performed on the resultant su~rnatants after aliquots had been diluted 25-fold into 10 mM Tris-Cl, 5 mM MgC12, pH 8.0. RESULTS AND DISCUSSION

Crude membranes prepared from lysed spheroplasts of B. pertussis yield an adenylate cyclase which is highly stimulated by calmodulin in a dose-dependent manner ’ Abbreviations used: EGTA, ethylene glycol bis(@amin~thyl ether) WV’-tetraacetic acid; HSA, human serum albumin; CDR, calmodulin.

47

BY BIOASSAY

-11

-10

-9 -8 log (w~~lin)

-7

-6

M

FIG. 1. The dose-response curve for the activation of B. pertussis membrane adenylate cyclase by calmodulin. The assay used 9 pg of membrane protein to measure varying amounts of caimodulin which were added in lopl aliquots (see Experimental Procedures for assay conditions).

(Fig. 1). The range of stimulation is extremely broad stretching from 0.01 nM to the saturation point at about 0.3 PM calmodulin when 10 gg of membrane protein is used for the assay. Similar curves were obtained with membranes prepared from three separate B. pertussis cultures indicating that such preparations are fairly uniform. Furthermore, we have prepared membranes from fresh cultures as well as frozen cell paste with little difference in activities. The time course for the reaction is linear for 30 min and no lag phase can be detected at I min when the reaction is initiated with substrate, demonstrating that the calmodulin activation of the adenylate cyclase had already taken place. The membrane preparations that were used in this study had activities of 35-50 nmol CAMP produced/ min/mg membrane protein in the presence of 1.67 X 10m9 M calmodulin final. Unlike the phosphodiesterase reaction, the B. pertussis adenylate cyclase calmodulinactivated reaction displays a more complex calcium ion specificity. Activation of the enzyme by Ca*+ is complex (manuscript in preparation). This problem is easily overcome by keeping the Ca2’ ion at a constant and saturating level (50 ,uM).

48

GOLDHAMMER TABLE

1

SE~SIT~VI~OF~ pettscssis MEMBRANEADENY~TE CVCLASE IN DETERMINING Low LEVELS OF CALMODULIN Added caimodulin (P#z)

Adenylate cyclase activity fpmol cAMP/min)

0 1.67 16.7

5.8 It 0.8 12.2 e 1.6 40.1 -t 4.1

Note. Assays were run as outlined under Experimental Procedures using 9 kg of membrane protein as the cyciase source. Calmodulin standards were made up in 0. t % HSA, such that the final concentration in the assay was 0.0167%. The basal value also contained an equivalent amount of HSA. Each result is an average of nine determinations with the indicated standard deviation.

The membrane preparations are extremely stable. They may be stored at -20°C in con~ntrat~ form (>lO mgfml membrane protein) for as long as 6 months without significant loss of activity. In addition, aliquots may be diluted to working concentration (- 1 mg/ml protein) lyophilized and stored at - 10°C until use. Reconstitution requires only the addition of water. Two samples were stored at room temperature for as long as 3 weeks without a loss in calmodulin responsiveness, though the baseline activity in the absence of calmodulin was elevated (data not shown). This effect was not investigated further and routine storage should be at freezer temperatures though brief exposure of the lyophilized powder to higher tem~ratures should have little effect on the enzyme. Since the calmodulin effect can be observed at very low activator concentrations where there is a problem of protein loss to vessel walls, the use of a carrier protein is obligatory (11). Since many protein preparations are contaminated with calmodulin, the sensitivity of the assay may be compromised (9). Three proteins were found suitable as carrier proteins for the assay: human serum albumin, soybean trypsin inhibitor, and n-amylase from Bacifius subtilis. Each

AND

WOLFF

of these preparations gave low blank values when employed at concentrations of 0.16 mg/ml. Several samples of bovine serum albumin were tested, but these proved to be contaminated. Because of the variability in different lots, each should be tested for contamination. The sensitivity of the response (Fig. 1) should permit the determination of very small amounts of calmodulin. This part of the titration curve was checked for reliability by repeated determination at two very low concentrations of calmodulin. As can be seen in Table 1, 1.67 pg of calmodulin can be determined in an unknown sample. This provides more sensitivity than does the radioimmunoassay (4). It may be possible to determine smaller amounts of calmodulin by TABLE

2

CALMODULIN CONTENT OF SEVERAL PROTUN, MEMBRANE, ANDCELLPREPARATIONS Protein soilrce

Calmodulin content ng/mg source protein

Creatine phosphokinase(Sigma) Creatine phosphokinase (~hrin~r-Mannheim) Bovine brain tubulin (2 cycle) Bovine brain clathrin Bovine brain coated vesicles Human red blood cell membranes Human red blood cell membranes (EGTA treated) I-10 Membranes Y-l Membranes

220 f (n = 12) 147 rt (n = 4) 4195 85 122 40 40 177 270 MoI~ul~/~ll

Rabbit red blood Human red blood Rat red blood S-49 Mouse lymphoma Y-I Rat adrenal tumor OS-3 Rat adrenal tumor

0.042 f 0.042 f O.llOf 1.54 f 3.12 1.04

x lo-’

0.004 (n 0.003 (n 0.006 (n 0.30 fn

= = = =

4) 3) 4) 7)

Nore. Protein sampleswere either dissolvedor dialyzed in 10 mht T&-Cl, 5 rnr+i MgCl, pH 8.0. All samples,except red blood cell sumples,were heated at 90°C for 5 min and centrifuged. Supernatants were diluted 25-fold into the abovebuffer prior to assay.One sample of human red blood cell membranes was washed three times against the abovebuffer containlag 1 msr EGTA. Results are the average of at least two determinations each in triplicate with the indicated standard deviations.

CALMODULIN

DETERMINATION

lowering the concentration of ATP in the assay medium, increasing the specific radioactivity of the substrate. Care in the preparation of samples for calmodulin assay is, perhaps, the most critical aspect of the method. For example, the heating procedure may result in artifactually low values because the denatured proteins may entrap calmodulin. When determinations of calmodulin in red blood cell lysates were attempted, levels of calmodulin were 20-30s below those of unheated controls. Attempts to solve this problem by sonication of the aggregates gave inconsistent results hence the values reported for red blood cells (Table 2) were on unheated samples. Similar controls on the cultured cells showed no such behavior, as heated and unheated samples agreed within 5%. An important question deals with the specificity of the assay for calmodulin. While it is difficult to rule out the presence of an as yet undiscovered protein from some other tissue, we had previously shown that troponin C exhibited about 0.001 of the cyclasestimulating activity of calmodulin (9). At this concentration it is difficult to rule out contamination by trace amounts of calmodulin. We have also tested the potency of frog parvalbumin on the activation of the membrane adenylate cyclase. It exhibited
BY BIOASSAY

49

Although the amount of calmodulin necessary for half-maximal activation in each case was the same as the required amount of bovine brain calmodulin (-2 X lo-’ M from Fig. 1 ), the maximal activations varied. Calmodulin from T. pyriformis yielded 56% of the maximum activation attained with bovine brain calmodulin, while peanut calmodulin yielded 75% of the brain calmodulin maximum. Octopus calmodulin was also active but we had insufficient material for a complete activation curve. The Tetrahymenu calmodulin displays the same bioreactivity toward phosphodiesterase as brain CDR (16), while peanut calmodulin is 85% as active (17). We are currently pursuing further structure-function relationships in the calmodulin family. Applications. Several protein and plasma membrane preparations were checked for calmodulin content using the B. pertussis membrane adenylate cyclase assay. These data are shown in Table 2. Contamination of protein preparations has been previously pointed out and the present method yields similar results to those obtained with intact organisms and brain phosphodiesterase (9). In addition, membrane preparations contain considerable calmodulin. One lot of creatine phosphokinase was subjected to multiple determinations both within the same and different assays with results agreeing within 5%, indicating that this method is reliable. It is of interest that repeated treatment of erythrocyte membranes with EGTA does not change the calmodulin content of the sample. This implies that this amount of calmodulin may be held via a calcium-insensitive mechanism. Similar data were obtained for other plasma membrane preparations (not shown). Crude brain membranes washed with an EGTA buffer had calmodulin levels of 240 ng/mg protein (18). Although tubulin preparations have a high calmodulin content, the calmodulin concentration can be reduced 20-fold by phosphocellulose chromatography which removes microtubule associated proteins (MAPS)

50

GOLDHAMMER

FIG. 2. Calmodulin content of rat red blood cells following induction of reticulocytosis. Cells were lysed as outlined under Experimental Procedures but were not boiled prior to assay. Reticulocytes were measured following staining with methylene blue.

(19). This may imply that calmodulin binds to these protein species. The calmodulin content of cells was determined as outlined in Table 2. In order to achieve consistent results, an efficient method had to be developed for lysing the cell and then eliminating proteins which might subsequently interfere with the adenylate cyclase assay. This was accomplished by sonication in the presence of Triton X100 as described under Experimental Procedures. More than 90% of the cells were lysed by this technique as judged by microscopy. EGTA (1 mM) was routinely employed to disrupt calmodulin-protein interactions but was not necessary for cell disruption when deleted. Samples were then heat-treated to denature the bulk of cellular proteins, taking advantage of the heat stability of calmodulin. All samples were diluted prior to assay, such that the Ca’+ concentration was in excess. Attempts were made to see whether the present method permitted the detection of changes in calmodulin concentration as a function of maturation. The rat reticulocyte system was examined since the changes in the levels of hormone receptors and adenyl-

AND

WOLFF

ate cyclase activity of the reticulocyte as it matures to an erythrocyte have been well documented (20). Figure 2 shows the results of calmodulin determinations in rat red blood cells following induction of reticulocytosis by injection of acetyl phenylhydrazine. Reticul~ytes have about four times as much calmodulin as erythrocytes. Furthermore, the level decays approximately in parallel with the rate of maturation of the reticulocyte. The mechanism of this loss of cellular calmodulin is unknown. Another example of the application of this calmodulin assay system is presented in Table 3. The Y- 1 cell line, a steroid-secreting adrenal tumor line, was treated with or without ACTH or cholera toxin to measure the effect of these agents on the level of calmodulin activity. Cells incubated with hormone and cholera toxin showed greater intracellular levels of calm~ulin than control cells. This coincides with the ability of these ligands to initiate steroidogenesis (21,22). The OS-3 cell line is a mutant of the Y-l line that lacks functional receptors to ACTH, TABLE

3

THEEFFECTOF HORMONEAND~HOLERAGEN TREATMENT ON CALMODULIN LEVELS IN Y-l CELW Treatment None Synacthen (ACTH l-24), 2.85 X lo-’ M Synacthen, 5.7 X lo-* M Cholera toxin, 100 rig/ml

Calmodulin (molecules/cell X IO-‘) 3.01 5.84 3.96 7.01

No&. Cells were grown as outlined under Experimental Procedures in lOO-mm dishes until confluent. Medium containing the indicated amount of hormone or toxin was added and cells were incubated for 24 h. Cells were harvested by scraping, washed three times in 10 mM Tris-Cl, 150 mrvr NaCl, pH 7.4, and suspended at approximately 10’ cells/ml. Calmodulin was assayed in disrupted preparations as outlined. The values are from a single experiment; three separated experiments gave similar results, with calmodulin levels agreeing within 10%.

CALMODULIN

DETERMINATION

although the steroidogenic pathway is intact (23); these cells have demonstrably less calmodulin than Y-l cells (Table 3). We do not presently know if there is a functional link between the receptor-cyclase system and calmodulin. Calmodulin has, however, been implicated as a cofactor in the steroidogenic pathway (24). Applications similar to these have been made recently using the radioimmunoand/ or the phosphodiesterase assays. Means and his colleagues (25) were able to demonstrate by radioimmunoassay that viral transformation of cells increased the levels of calmodulin twofold. This was ascribed to increased synthesis and not degradation. Levels of calmodulin in these cell lines, 3T3 and normal rat kidney, ranged from 5.5 X IO6 to 3.22 X 10’ molecules of calmodulin/cell(25). Our results for cultured cells in the absence of hormone fall in this range (Table 3). Two reports employing the phosph~iesterase assays describe the subcellular distribution of calmodulin in T. pyriformis (26) and the levels of calmodulin in dispersed human parathyroid cells (27). Despite the fact that various methods for determining calm~ulin levels may not give comparable results (25), it is clear that changes in cellular calmodulin content can be correlated with alterations in the cell by transformation (25,28), or hormone or toxin treatment (Table 3). The B. pert~ssi~ adenylate cyclase assay compares favorably with either of these alternative methods because it is more rapid than the radioimmunoassay and has a degree of amplification much greater than the phosphodiesterase assay. Moreover, the need for preparation of either a calm~ulin-deficient enzyme or antibody is eliminated. One caveat must be stated which is applicable to any of the assays which might be employed. The method of sample preparation is of utmost importance. All calm~ulin present in cells is not released by chelator treatment (this study, 18,26). Heat treatments can unmask this additional calmodulin by denatur-

51

BY BIOASSAY

ing a great many of the calmodulin-binding proteins, but may cause entrapment of calmodulin in the resultant precipitate. It is advisable to run a number of coincidental dilutions and unheated controls for comparison. REFERENCES Wang, J. H., and Waisman, D. M. (1979) Catrr. Top. Cell.

Regul.

15, 47-108.

Klee, C. B., Crouch, T. H., and Richman, P. G. (1980) Annu. Rev. Biochem. 49, 489-515. Cheung, W. Y. (1980) Science 207, 19-27. Chafouleas, J. G., Dedman, J. R., Munjaal, R. P., and Means, A. R. ( 1979) J. Biol. Chem. 254, 10,262-10,267. Wallace, R. W.,

and Cheung, W. Y. (1979). J Bioi. Chem. 254,6564-6571. Van Eldik, L. J., and Watterson, D. M. (1981) J. Biol. Chem. 256,4205-4210. Wallace, R. W., Lynch, T. J., Tallant, E. A., and Cheung, W. Y. (1978) Arch. B&hem. Biophys. 187, 328-334.

Wolff, J., Cook, G. H., Goldhammer, A. R., and Berkowitz, S. A. (1980) Proc. Nat. Acad. Sci. USA 77, 3841-3844. 9. Goldhammer, A. R., Wolff, J., Cook, G. H., Berkowitz, S. A., Klee, C. B., Manclark, C. R., and Hewlett, E. L. (1981) Eur. J. Biochem. 115, 605-609. 10. Hewlett, E. L., and Wolff, J. (1976) J. Bucteriol. 8.

127, 890-898.

11. Klee, C. B. (1977) Biochemistry 16, 1017-1024. 12. Hewlett, E. L., Urban, M. A., Manclark, C. R., and Wolff, J. (1976) Proc. Naf. Aead. Sci. USA 73, 19261930. 13. Hewlett, E. L., Wolff, J., and Ma&ark, C. R. (1978) Advan. Cycl. Nucleotide Res. 9,621-628. 14. Salomon, Y., Londos, C., and Rodbell, M. (1974) Anal. Biochem. 58, 541-548. 15. Vanaman, T. C. (1980) in Calcium and Cell Function (Cheung, W. Y., ed.), pp. 41-58, Academic Press, New York. 16. Jamieson, G. A., Jr., Vanaman, T. C., and Blum, J. J. (1979) Proc. Nat. Acad. Sei. USA 76,64716475. 17. Cormier, M. J., Anderson, J. M., Carbonneau, H. P., and McCann, R. 0. (1980) in Calcium and Cell Function (Cheung, W. Y., ed.), pp. 201218, Academic Press, New York. 18. Piascik, M. T., Wisler, P. L., Johnson, C. L., and Potter, J. D. (1980) J. Biof. Gem. 255,41764181.

52

GOLDHAMMER

19. Berkowitz, S. A., and Wolff, J. (1981) J. Biol. Chem 256, 11,216-l 1,223. 20. Bilezikian, J. P., Spiegel, A. M., Brown, E. M., and Aurbach, G. D. (1977) Mol. Pharmacol. 13, 775-785. 21. Buonassisi, V., Sato, G., and Cohen, A. I. (1962) Proc. Nat. Acad. Sci. USA 48, 1184- 1190. 22. Wolff, J., Temple, R., and Cook, G. H. (1973) Proc. Nat. Acad. Sci. USA, 70, 2741-2744. 23. Schimmer, B. P. (1969) .I. Cell Physiol. 74, 115122. 24. Hall, P. F., Osawa, S., and Thomasson, C. L. (1981) J. Cell Biol. 90, 402-407.

AND WOLFF 25. Chafouleas, J. G., Pardue, R. L., Brinkley, B. R., Dedman, J. R., and Means, A. R. ( 1981) Proc. Nat. Acad. Sci. USA 78,996-1000. 26. Nagao, S., Banno, Y., Nozawa, Y., Sobue, K., Yamazaki, R., and Kakiuchi, S. (198 1) J. Biochem. (Tokyo) 90,897~899. 27. Brown, E. M., Dawson-Hughes, B. F., Wilson, R. E., and Adragna, N. (1981) J. Gin. Endocrirwl. Metab. 53, 1064-1071. 28. MacManus, J. P., Braceland, B. M., Rixon, R. H., Whitfield, J. F., and Morris, H. P. (1981) FEES Lett. 133, 99-102.