Adenylate Cyclase in Rat Testis

Proc. Nat. Acad. Sci. USA Vol. 72, No. 3, pp. 1097-1101, March 1975

Development of a Mn2+-Sensitive, "Soluble" Adenylate Cyclase in Rat Testis (seminiferous tubules/cytosol/epididymal sperm/membrane-associated)

THEODOR BRAUN* AND RICHARD F. DODS Department of Physiology, Northwestern University Medical Center, Chicago, Illinois 60611

Communicated by David Shemin, January 2, 1976 A distinctive Mn2+-sensitive adenylate ABSTRACT cyclase [ATP pyrophosphate-lyase(cyclizing), EC 4.6.1.1J system insensitive to fluoride has been found in rat seminiferous tubules and epididymal sperm. The development of this distinctive adenylate cyclase in testis was studied during spermatogenesis. It was first detectable in seminiferous tubules in immature rats at about the time of the first reductive divisions and the appearance of spermatid cells. The specific activity of the enzyme increased substantially during the period of spermatogenesis when spermatids develop into mature spermatozoa, and reached maximal values in the testis of adult rats. After centrifugation of testis tissue homogenates at 105,000 X g for 60 min, the Mn2+-sensitive adenylate cyclase activity was found in the cytosol. The enzyme remains in solution after centrifugation at 300,000 X g for 5 hr or at 180,000 X g for 24 hr and passes through a 0.22 /Am Millipore filter. Electron microscopic examination showed no visible membrane fragments or vesicles in the filtered supernatant. The Mn2+-sensitive adenylate cyclase system is also present in epididymal sperm. However, in the sperm obtained from either the caput or the cauda of epididymis, the adenylate cyclase is membraneassociated and found in particulate fractions of sperm homogenates. It therefore appears that the Mn2+-sensitive adenylate cyclase is initially present in the cytoplasm either unattached or loosely bound to intracellular membranes and becomes firmly attached to sperm membranes later in development. This occurs either during the process of maturation of spermatids into sperm or duting the transport of the testicular sperm into the epididymis.

Adenosine 3': 5'-cyclic monophosphate (cAMP) generated by the adenylate cyclase [ATP pyrophosphate-lyase(cyclizing), EC 4.6.1.1] system has been implicated in the regulation of sperm metabolism and motility (1, 2). Adenylate cyclase activity has been demonstrated in all mammalian spermatozoa thus far examined, including the rat (3-7). The rat sperm adenylate cyclase exhibits special features not displayed by adenylate cyclases of a variety of rat tissues, as well as by certain testicular cyclases, which are stimulated by FSH and LH (6, 7). The adenylate cyclase systems of rat somatic cells generally are membrane-associated and found in the particulate subcellular fractions of homogenates, depend for activity on either Mg2+ or Mn2+ and are stimulated by hormones specifically and by fluoride nonspecifically (8). The gonadotropin-sensitive adenylate cyclases in rat testis have the expected general features of cyclases in other tissues. In contrast, the adenylate cyclase in epididymal sperm has been found to be Mn2+-dependent, insensitive to fluoride, and not Abbreviation: cAMP, adenosine 3':5'-cyclic monophosphate; LH, luteinizing hormone; FSH, follicle stimulating hormone. * Requests for reprints may be addressed to Dr. T. Braun. 1097

stimulated by FSH and LH (6, 7). Since mature spermatozoa have no protein synthetic ability, the adenylate cyclase in this cell type must be synthesized at a specific developmental stage in which this distinctive Mn2+-sensitive adenylate cyclase appears in the testis. The properties of the distinctive adenylate cyclase system have been studied during its development in testis and epididymal sperm. It has been found that the properties of the Mn2+-sensitive adenylate cyclase in testis and sperm homogenates are similar. However, upon centrifugation, the bulk of activity in testis is in the soluble fraction of cytoplasm whereas in sperm, the adenylate cyclase system is firmly associated with membranes. MATERIALS AND METHODS

Rats (Charles River CD®) of various age groups were used. Immature rats younger than 21 days were kept in groups of ten with their mothers. Purina chow diet and water ad lib were provided to nursing mothers, suckling and weaned (more than 21 days old) rats. They were maintained at a temperature of 22-23° and a 14-hr light and a 10-hr dark period each day.

[a-32P]ATP was purchased from International Chemical and Nuclear Corp., creatine phosphate and creatine phosphokinase (skeletal muscle, rabbit) from Calbiochem. The rats were sacrificed by decapitation and the testes were removed, decapsulated, freed of visible blood vessels and placed in ice-cold 5 mM Tris HCl buffer (pH 7.2) containing 3 mM MgCl2 and 1 mM EDTA (5 mM Tris buffer). In experiments where seminiferous tubules and interstitial cells were separated, the testes, after being excised, were placed in icecold Krebs-Ringer phosphate buffer (pH 7.2) containing 10 mg of bovine serum albumin per ml (KRP-BSA), with half of the usual calcium ion content replaced by the equivalent amount of sodium. Seminiferous tubules were isolated from testes by microdissection using procedures previously described (9). Interstitial cells were isolated by collagenase (Worthington; 1 mg/ml per 100 mg of tissue) treatment of testes for 5 min in a metabolic shaker at 60 cycles/min; cells were separated from the undigested tubules by filtration through a double layer of nylon mesh, and washed four times with KRP-BSA buffer. The cell fraction thus obtained is enriched with interstitial cells and is usually composed of 65% interstitial, 15% tubular, and 20% erythrocytes. Spermatozoa were collected in 5 mM Tris buffer from the caput, corpus, and cauda of epididymis. The epididymis was ligated in situ into these portions and separated accordingly after excision.

1098

Biochemistry: Braun and Dods

Proc. Nat. Acad. Sci. USA 72

(1975)

TABLE 1. Distribution of Mg2+- and Mn2+-sensitive adenylate cyclase activty in subceUular fractions of immature and adult rats % of total activityt Specific activity* (14-day old) Whole homogenate 600 X g X 10 min pellet 10,000 X g X 10 min pellet 105,000 X g X 60 min pellet 105,000 X g X 60 min supernatant

Mg2+ 4.1 3.0 4.5 3.1 ND

Mn2+ 5.0 6.5 9.8 10.2 ND

Mg2+ 100 21.8 11.4 1.5

Mn2+ 100 39.0 20.7 3.4

-

-

Specific activity*

% of total activityt

(110-day old) Mg2+ 2.5 2.3 2.7 1.8 NS

Mn2+

Mg2+

Mn2+

10.9

100

100 15.0 9.8 5.7 89.0

10.1

12.4

7.5 5.9 20.8

13.1 6.8 -

The values listed in the Table are from a single (immature rats) and from two separate experiments (adult rats). * pmol/mg of protein per min. t Total homogenate activity arbitrarily set as 100%. ND, not detected; NS, not significant, barely detectable amounts of cAMP formed.

Preparation of Tissue and Cell Homogenates. Seminiferous tubules and the interstitial cell fraction were washed twice in 5 mM Tris buffer before homogenization. Homogenates were prepared in the same buffer in tight-fitting glass Dounce homogenizers using three to four strokes for testicular tissue preparations and 20 strokes for epididymal sperm. Homogenates were filtered through glass wool. Subcellular fractions from whole homogenates were isolated by differential centrifugation in the Sorvall model RC2-B centrifuge at low speed (600 and 10,000 X g X 10 min) and Beckman model 12-65B ultracentrifuge at higher speeds. Adenylate cyclase activity was determined by measuring the rate of conversion of [a-32PIATP to [32P]cAMP. The assay procedure was essentially similar to that described by Bar and Hechter (10) with minor modifications. The incubation mixture contained, in a total volume of 50 Al: 40 mM TrisHCl buffer (pH 7.6); 5 mM Mn2+ or Mg2+; 0.5 mM unlabeled cAMP; 0.1% bovine serum albumin; 10 mM creatine phosphate; creatine phosphokinase 0.1 mg/ml; and 0.2 mM [a-S2PIATP. The Mn2+ was added to the incubation mixture 10-15 min before the start of the assay to prevent oxidation of Mn2+ (11). The reaction was started by adding [a-82P]ATP. The incubation was performed at 320 for 20 min. The reaction was stopped by boiling 3 min TABLE 2. Ultracentrifugation at high g forces and the Mn2+-sensitive adenylate cyclase activity from rat testis homogenates

Preparation Whole homogenate High-speed supernatantf 105,000 X g X 60 min supernatantl 180,000 X g X 60 min supernatant$ 180,000 X g X 24 hr supernatants 300,000 X g X 5 hr supernatants

Adenylate cyclase activity (pmol of cAMP/ mg of protein per min) 11.5 d4 0.5* 18.5 A 1.3* 18.3 4 1.0 19.8 4 1.5 21.7 4 0.6 22.2 4= 0.6

* Mean ± SEM from six separate preparations, each of which was assayed in duplicate; rats 57- to 120-day old were used. t Supernatant obtained either at 105,000 or 180,000 X g X 60 min. t Separate experiment; rats 115-day old were used. Each value was the mean of duplicate determination.

after addition of 50 ,ul of unlabeled 10 mM ATP, ADP, and cAMP and 20 mM EDTA. The addition of EDTA before boiling prevents nonenzymatic formation of cAMP in the presence of Mn2+ (4). The [32P]cAMP formed was separated from other radioactive substances by chromatography on polyethylenimine-impregnated thin-layer cellulose sheets developed in 0.3 M LiCl (10), or more recently on silica gel thin-layer sheets developed in an organic solvent system, chloroform 40: methanol 20: water 3 (12). Separation of cAMP by either of these two methods gave identical results (ref. 12, and unpublished observations). ATP levels during incubation were maintained by the regenerating system at 80-90%0 of the initial value. The degree of ATP breakdown during incubation was determined by measuring the nucleotide composition of the incubated sample by chromatography on polyethylenimine-impregnated thin-layer cellulose sheets developed in 1 M LiCl (13). Protein was determined after solubilization by boiling for 15 min in 1 N NaOH by the Lowry et al. (14) method with bovine serum albumin as standard. The results were expressed in pmol of cAMP formed/mg of protein per min.

Negative Staining. High-speed supernatant fraction was negatively stained on carbon-coated copper grids with 1% phosphotungstic acid (pH 6.8). After blotting and drying for 30 min, samples were examined on a Hitachi HU 12 electron microscope at an accelerating voltage of 75 kV. RESULTS Separation of the Mg2+- and Mn2+-sensitive adenylate cyclase activity in rat testis homogenates by differential centrifugation In rat testis homogenates cAMP is formed in the presence of Mg2+ or Mn2+. In whole homogenate from 14-day-old rats, Mg2+ and Mn2+ are about equally effective in stimulating adenylate cyclase activity. Following centrifugation of testes homogenate from 14-day-old rats, the Mg2+- and Mn2+sensitive enzyme activity was found in the particulate fractions which sedimented at 600 and 10,000 X g for 10 min and 105,000 X g for 60 min. Adenylate cyclase activity was not recovered in the cytosol fraction from testes homogenate of 14-day-old rats. The recovery of the particulate adenylate cyclase upon differential centrifugation is variable in different experiments and was found to amount to 25-60% of the total activity originally present in the whole homogenate. This is


(1975)

"Soluble" Adenylate Cyclase in Rat Testis

due to the lability of the adenylate cyclase activity in homogenates from immature rats and the decline of its activity with time when allowed to stand (even at ice temperature). In whole homogenates from 120-day-old rats the enzyme activity is considerably greater with Mn2+ than with equiniolar concentrations of Mg2+ (Table 1). Centrifugation of testes homogenate of adult sexually-mature rats results in the sedimentation of the Mg2+- and Mn"-sensitive adenylate cyclase activity in the particulate fractions. However, in adult rats, in contrast to the immature rats, the bulk (89%/) of the enzyme activity originally present in the whole homogenate is recovered in the cytosol (Table 1). The cyclase activity in the cytosol is significantly activated by Mn'+, but not by Mg'+. The adenylate cyclase activity in the cytosol is extremely stable upon high-speed centrifugation (Table 2).

Localization of the Mn2+-sensitive "soluble" adenylate cyclase activity in the seminiferous tubules of adult rats The activity of the Mn'+-sensitive adenylate cyclase in whole testis, isolated interstitial cells, and seminiferous tubules was compared. The bulk of enzyme activity present in the whole tissue was recovered in the seminiferous tubules isolated by microdissection (Table 4) Adenylate cyclase activity was also observed in the supernatant fraction isolated from the homogenate of interstitial cells; however, the enzyme activity in this tissue fraction was only 10% of the activity in the isolated seminiferous tubules. The occurrence of this amount of the .

Mn2+-sensitive adenylate cyclase activity in the interstitial cell fraction can be accounted for by the 'presence of cells originating from the seminiferous tubules (usually about 15% of total cells in this fraction). Development of the "soluble" Mn2+-sensitive adenylate cyclase in rat testis during spermatogenesis The Mn2+-dependent adenylate cyclase activity was undetectable in the 105,000 X g X 10 min supernatant from

TABLE 3. Millipore filtration and the Mn2+-sensitive adenylate cyclase activity in the supernatant fraction from adult rat testes homogenate Adenylate cyclase activity Supernatant (180,000 X g X 60 min)

(pmol of cAMP/ mg of protein per min)

Unfiltered Filtratet

23. 9 0.1* 25.3 4 1.5*

* Mean

SEM of triplicate determinations.

t Filtrate passed through 0.22 pm Millipore filter.

0

0 N

0

-W

~~0E .:0

0 c

E 30

0

co

.6p*

0l.

'I"

C

&-

a

a

o

20-

0 0'

E

IS-E 10 0

0

75 E 0.

Studies testing the solubility of the Mn'2+-sensitive adenylate cyclase

The Mn2+-sensitive adenylate cyclase activity in the cytosol fraction isolated from testis homogenates was not sedimented by further centrifugation either at 180,000 X g for 60 mm or 105,000 X g for 24 hr or -at 300,000 X g for 5 hr (Table 2). After filtration of the high-speed supernatant through 0.22 Am millipore filter, the adenylate cyclase activity was entirely in the filtrate (Table 3). Moreover, the high-speed supernatant fraction filtered through millipore filter did not contain membrane fragments or vesicles when examined by electron microscopy of negatively stained preparations.

01

c

o

1099

I ^ 0

I

.

.

.

I

.

.

.

0 10 20 30 40 50 60 70 age in days

80

a ' ' ' 90 100110120 I

u

.

FIG. 1. Development of MnO+-sensitive adenylate cyclase in rat testis. Enzyme activity was determined in the cytosol isolated from testes homogenates. Stages of morphological development in the seminiferous tubules are shown.

homogenates of rat 14-days old (Fig. 1). At this developmental stage the tubular epithelium is composed primarily of Sertoli cells, spermatogonia, and primary spermatocytes. Adenylate cyclase activity in high-speed supernatant was first detected in rats 22- to-26-days old, at the developmental stage when the first spermatid cells appear and the rapid rise of its activity between 26 and 35 days of age correlates with the appearance of a considerable number of spermatids. The Mn +-sensitive enzyme activity reaches its maximal plateau values in adult, sexually mature rats. Distribution of the Mn'+-sensitive adenylate cyclase activity in subeellular fractions of epididymal sperm In contrast to the Mn2t-sensitive adenylate cyclase activity in

the seminiferous tubules, the adenylate cyclase in the epididymal sperm obtained either from the caput or cauda was associated mainly with the particulate fractions which sedimented at 10,000 X g for 10 min and 105,000 X g for 60 min (Table

5). TABLE 4. Distribution of Mn2+-sensitive adenylate cyclase activity between isolated seminiferous tubules and interstitial cells from testis of adult rats

Preparation

Whole testis tissue Isolated seminiferous tubules Interstitial cell fraction

Whole Supernatant* homogenate pmol of cAMP/mg of protein per min b 0.3

26.9 4± 1.5

6.3 ± 0.1

24.5 ± 1.5

2.3 ± 0. 1

2.7 4 1.4

6.3

Each value is the mean ± SEM of duplicate (whole homogenate) or triplicate determinations (supernatant). * Filtrate obtained by Millipore filtration of 20,000 X g X 10 min supernatant fraction from testes homogenate; rats 90-days old were used.

1100

Biochemistry: Braun and Dods


TABLE 5. Relative distribution of adenylate cyclase activity in subcellular fractions of rat spermatozoa from caput and cauda of epididymia8

Fraction

Whole homogenate 600 X g X 10 min pellet 10,000 X g X min pellet 105,000 X g X 60 min pellet 105,000 X g X 60 min supernatant

Epididymal sperm from Cauda Caput (% of total homogenate activity) 100 3.5 22.1 17.6 4.3

100 7.4 42.0 47.0 3.2

The values listed in the table are from three separate (caput) and a single experiment (cauda). Adenylate cyclase assays performed in the presence of 5 mM Mn2+ as the divalent cation.

DISCUSSION In rat testis, multiple adenylate cyclase systems exist in distinctive specific cell types (6, 7). An adenylate cyclase sensitive to luteinizing hormone (LH) is present in the interstitial Leydig cells and a separate adenylate cyclase sensitive to follicle stimulating hormone (FSH) is present in the tubular Sertoli (15, 16), and perhaps spermatogonial cells (6, 7). These hormone-sensitive adenylate cyclases are membrane-associated (17, 18), and in tissue homogenates their divalent cation requirement can be satisfiied with both Mg2+ and Mn2+. Accordingly, the presence of Mn2+-sensitive adenylate cyclase activity in particulate fractions of testis homogenates was expected. Our findings reveal, however, a distinctive adenylate cyclase in the cytosol, which, in contrast to the adenylate cyclase activity in particulate fraction, has an obligatory requirement for Mn2+, and is Mg2+-, fluoride-, FSH- and LH-insensitive, and the stimulatory effect of Mn2+ is potentiated by Ca2. (refs. 6 and 7, and unpublished observations)t. The Mn2+-sensitive "soluble" adenylate cyclase is localized in the seminiferous tubules. It was not detected in the developing seminiferous tubules before the appearance of spermatid cells and a rapid rise of its activity appears to correlate with the increase in the number of these cells. In the epididymal sperm, the adenylate cyclase has properties similar to the Mn2+-sensitive adenylate cyclase in the seminiferous tubules (6, 7). However, the adenylate cyclase in the epididymal sperm was found to be firmly membrane-bound. During the past 15 years, it has been widely recognized that adenylate cyclases are complex, multi-component systems, t The adenylate cyclase in the cytosol exhibits some properties similar to those seen in guanylate cyclases [GTP pyrophosphatelyase(cyclizing) EC 4.6.1.2] from a variety of mammalian tissues (10), in that it is Mn2+-dependent, fluoride-insensitive and the stimulatory effect of Mn2+ is potentiated by Ca2+. However, we have found that these two cyclases are separate entities. The adenylate cyclase activity can be filtered through a 0.22 m1A Millipore filter, whereas the bulk of guanylate cyclase activity is retained. Furthermore, the two cyclases differ with regard to nucleotide specificity. The guanylate cyclase activity is inhibited by ATP, whereas the adenylate cyclase activity is unaffected by GTP. (These findings will be published in detail in a forthcoming paper.)

(1972)

associated primarily with the cell plasma membrane in various mammalian cell types and in some few cases with intracellular membranous structures (8). Our present findings reveal, however, that the distinctive Mn2+-sensitive adenylate cyclase in the seminiferous tubules is entirely in the cytosol. These findings suggest that the germ cell adenylate cyclase in the testis is present in the cytoplasm, unattached or loosely associated with intracellular membranes. Upon transfer of sperm from testis to epididymis, the adenylate cyclase in sperm, whether obtained from the caput, corpus, or cauda of the epididymis, is now found to be firmly membrane-associated. It is found associated almost entirely with "mitochondriar' and "microsomal" particulate fractions following differential centrifugation of sperm homogenates. The attachment of the Mn2+-sensitive adenylate cyclase to sperm membranes occurs in testis either in the maturing spermatids or during transport of testis sperm into the epididymis. Our present findings suggest that in the germ cell, the adenylate cyclase system and the entity required for its inclusion into the membrane either do not develop at the same time or are brought into association after being synthesized at a certain developmental stage. Both the nature of the entity and the mechanism of this phenomenon are unknown and merit further investigation. The biogenesis of the molecular components of the adenylate cyclase system and their assembly in the membranes is unknown. The phenomena associated with the formation of the germ cell adenylate cyclase and its association with membrane(s) during further development might be a useful developmental model for gaining some insight into these questions. We do not know whether the "soluble", Mn2+-sensitive adenylate cyclase in the seminiferous tubules represents merely a developmental stage in the formation of the germ cell cyclase system or whether it has a yet unexplored role in the synthesis of specific structures and in the transformation of the spermatid cells into spermatozoa. These questions merit further investigation. Since the tail of the sperm is formed during the process of conversion of spermatids into mature sperm, it is of interest to recall a previous observation by Yokota and Gots (19), i.e., "cAMP is absolutely necessary for flagella formation and hence motility, in cAMP-deficient mutants of Escherichia coli and Salmonella typhimurium". We are indebted to Ms. Elizabeth Krohn for excellent technical assistance, to Dr. S. Sepsenwol for performing the negative staining, and to Dr. 0. Hechter for his continued interest in this work. Supported by the Rockefeller Foundation Grant RF71074. 1. Garbers, D. L., Lust, W. D., First, N. L. & Lardy, H. A. (1971) "Effect of phosphodiesterase inhibitors and cyclic nucleotides on sperm respiration -and motility," Biochemistry 10, 1825-4831. 2. Tash, J. & Mann, T. (1973) "Relation of cyclic AMP to sperm motility," J. Reprod. Fert. 35, 591. 3. Casilas, E. 'R. & Hoskins, D. D. (1971) "Adenyl cyclase activity and cyclic 3',5'-AMP content of ejaculated monkey spermatozoa," Arch. Biochem. Biophys. 147, 148-155. 4. Gray, J. P. (1971) "Cyclic AMP and cyclic GMP in gametes," Ph.D. Dissertation, Vanderbilt University Medical School Library, Nashville, Tenn. 5. Morton, B., 'Harrigan-Lum, J., Albagli, L. & Jooss, T. (1973) "The activation of motility in quiescent hamster sperm from the epididymis by calcium and cyclic nucleotides," Biochem. 'Biophys. Res. Commun. 56, 372-379.

Proc. Nat. Acad. Sci. USA 72 (1975) 6. Braun, T. (1974) "Evidence for multiple, cell specific and distinctive adenylate cyclases in rat testis," in Current Topis in Molecular Endocrinology, eds. Dufau, M. & Means, A. R. (Raven Press, New York), Vol. 1, 243-264. 7. Braun, T. (1975) "Stimulation of sperm adenylate cyclase activity by divalent cations: Potentiation by calcium," Advances in Cyclic Nucleotide Research, eds. Greengard, P. & Robison, G. A. (Raven Press, New York), Vol. 5, in press. 8. Robison, G. A., Butcher, R. W. & Sutherland, E. W. (1972) Cyclic AMP (Academic Press, New York). 9. Braun, T. & Sepsenwol, S. (1974) "Stimulation of "IC, cyclic AMP accumulation by FSH and LH in testis from

mature and immature rats," Endocrinology 94, 1028-1033.

10. Bar, H. P. & Hechter, 0. (1969) "Adenyl cyclase assay in fat cell ghosts," Anal. Biochem. 29, 476-489. 11. Hardman, J. G. & Sutherland, E. W. (1969) "Guanyl cyclase, an enzyme catalyzing the formation of guanosine 3',5'-monophosphate from guanosine triphosphate," J. Biol. Chem. 244, 6363-6370. 12. Flouret, G. & Hechter, 0. (1974) "Adsorption chromatography of cyclic nucleotides on silica gel and alumina thinlayer sheets," Anal. Biochem. 58, 276-285.

"Soluble" Adenylate Cyclase in Rat Testis

1101

13. Hechter, O., Bar, H. P., Matsuba, M. & Soiffer, D. (1969) "ACTH-sensitive adenyl cyclase in bovine adrenal cortex membrane fractions," Life Sci. 8, 935-942. 14. Lowry, O. H., Rosebrough, N. J., Farr, A. L. & Randall, R. J. (1951) "Protein measurement with the folin phenol reagent," J. Biol. Chem. 193, 265-275. 15. Dorrington, H. J. & Fritz, I. B. (1974) "The effect of gonadotropins on cyclic AMP production by isolated seminiferous tubule and interstitial cell preparations," Endocrinology 94, 395-403. 16. Means, A. R. & Huckins, C. (1974) "The Sertoli cell is the primary target for FSH in the testis," Endocrinology 94, Suppl. A-107. 17. Means, A. R. (1973) "Specific interactions of 3H-FSH with rat testis binding sites," Advan. Exp. Med. Biol. 36, 431448. 18. Catt, K. J., Tsuruhara, T. & Dufau, M. L. (1972) "Gonadotrophin binding sites of the rat testis," Biochim. Biophys. Acta 279, 194-201. 19. Yokota, T. & Gots, J. S. (1970) "Requirement of adenosine 3',5'-cyclic phosphate for flagella formation in Escherichia coli and Salmonella typhimurium," J. Bacteriol. 103, 513516.