THE JOURNAL OF BKK.OGICAL CHEMISTRY Vol. 265, No. 31, Issue of November 5, pp. 18933-18943. 1990 ic; 1990 by The American Society for Biochemistry and Molecular Biology, Inc. Printed in U.S. A.

The Cyclic Diguanylic Acid Regulatory System of Cellulose Synthesis in Acetobacter xylinum CHEMICAL SYNTHESIS AND BIOLOGICAL ACTIVITY OF CYCLIC NUCLEOTIDE DIMER, TRIMER, AND PHOSPHOTHIOATE DERIVATIVES*

(Received for publication, March 6, 1990)

Peter Ross, Raphael Mayer, Haim Weinhouse, Dorit Amikam, Yassir Huggirat, and Moshe Benziman$ From the Department of Biological Chemistry, Institute of Life Sciences, The Hebrew University of Jerusalem, 91904 Jerusalem, Israel

Erik de Vroom, Alex Fidder, Paul de Paus, Leo A. J. M. Sliedregt, Gijs A. van der Mare& and Jacques H. van Boom From the Gorlaeus Laboratories, Department of Organic Chemistry, P. 0. Box 9502, 2300 RA Leiden, The Netherlands

An unusual compound, cyclic-bis(3’ + 5’) diguanylic acid (c-di-GMP or cGpGp), is involved in the regulation of cellulose synthesis in the bacterium Acetobacter xy- linum. This cyclic dinucleotide acts as an allosteric, positive effector of cellulose synthase activity in vitro (K,, = 0.31 MM) and is inactivated via degradation by a Ca’+-sensitive phosphodiesterase, PDE-A (K, = 0.26 PM). A series of 13 analogs cyclic dimer and trimer nucleotides were synthesized, employing a phospho- triester approach, and tested for the ability to mimick c-di-GMP as activators of cellulose synthase and as substrates for PDE-A.

Seven of the synthetic compounds stimulate cellulose synthase activity and all of these activators undergo the Ca’+-inhibited degradation reaction. The order of affinities for synthase activators is cGpGp = cdGpGp = cGp(S)Gp (S-diastereomer) > cIpGp > cdGpdGp > cXpGp > cIpIp > cGp(S)Gp (R-diastereomer). Three cyclic dinucleotides of negligible affinity for either enzyme are cApAp, cUpUp, and cCpCp. This same order of affinities essentially pertains to the analogs as inhibitors of PDE-A activity, but at least one cyclic dinucleotide, cXpXp, which does not bind to cellulose synthase, is also a substrate for the degradation reac- tion, demonstrating that although the two enzymes share a similar, high degree of specificity for c-di- GMP, their cyclic dinucleotide binding sites are not identical. Phosphodiester bonds of activators in which an exocyclic oxygen is replaced with an atom of sulfur (cGp(S)Gp isomers) resist the action of PDE-A, and such derivatives may be prototypes for synthetic non- hydrolyzable c-di-GMP analogs.

Cellulose biogenesis is quintessential to the plant kingdom, where the deposition of fibrils within the cell wall is an integral process to growth and development (1). Cultures of Acetobactor xylinum produce a tough pellicle of high purity cellulose, and this Gram-negative bacterium constitutes a convenient model for morphological and biochemical inves-

* The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

$ To whom correspondence should be addressed.

tigations into the nature of cellulose formation. While in plant cell walls, cellulose fibrils are an intrinsic structural element, the mat-like pellicle of cellulose fibrils produced by A. xylinum is extracellular, apparently conferring a selective advantage to this obligative aerobe under static growth con- ditions (2).

In A. xylinum, cellulose fibrils are extruded into the external mileu from an array of discrete sites arranged along the cell surface, giving rise to a single ribbon structure composed of intertwined crystalline fibrils (3). The process of glucose polymerization into polyglucan chains and the assembly of chains into rigid fibrils have been postulated to be tightly coupled processes (4). Although the underlying organizing principles have not been defined, many aspects of the polym- erizing mechanism are coming to light. Cellulose synthase, the enzyme responsible for the polymerization of glucose from UDP-glucose into the P-1,4-linked chain, has been isolated from this organism as a highly active membrane-bound cata- lytic unit (5). Comparison of various enzymatic levels in vitro (6), and of precursor concentrations in uivo (7), indicates that in the pathway from glucose, via glucose-6 phosphate, glucose- 1 phosphate, and UDP-glucose, into polyglucan chain, the unique biosynthetic step catalyzed by cellulose synthase may be rate-limiting.

The cellulose synthase from A. xylinum is subject to a complex form of regulatory control which has not been en- countered previously in a living system (5, 6, 8, 9). The basis of this regulation appears to lie in the intracellular concentra- tion of an unusual cyclic guanyl nucleotide dimer (originally isolated as an incubation product of GTP with cell extracts (8, 9)) which functions as an allosteric effector of enzyme activity. In vitro, cellulose synthase activity displays nearly absolute dependence on the presence of nanomolar concen- trations of this compound, which was conclusively identified in a recent report (10) as cyclic bis(3’ -+ 5’)-diguanylic acid (c-di-GMP or cGpGp).’ It is currently not known if c-di-GMP affects other cellular processes in addition to cellulose syn- thesis.

The pathways of synthesis and degradation of c-di-GMP are catalyzed by enzymes which occur in both soluble and

’ The abbreviations used are: c-di-GMP (or cGpGp), cyclic bis(3’ + 5’).diguanylic acid; RP-HPLC, reverse phase high performance liquid chromatography; PEI, polyethyleneimine; BAP, bacterial al- kalinephosphatase; THP, tetrahydropyranyl.

18933

by guest on Decem

ber 24, 2019http://w

ww

.jbc.org/D

ownloaded from

18934 Cyclic Dinucleotides in the Regulation of Cellulose Synthesis

membrane-associated form in cell-free extracts. The forma- tion of the nucleotide activator is attributed to diguanylate cyclase, a predominantly soluble enzyme, which condenses two molecules of GTP in a two-step reaction, via the inter- mediate linear triphosphate pppGpG. A membrane-bound phosphodiesterase, termed PDE-A, degrades the activator, cleaving a single phosphodiester bond in the cyclic structure to form the linear 5’-phosphoryl dimer pGpG (10, 11). This initial degradation product, which is devoid of stimulatory effect, is then further hydrolyzed to yield two molecules of 5’- GMP in a reaction attributed to a second phosphodiesterase, termed PDE-B. This latter phosphodiesterase has been deemed distinct from PDE-A on the basis that the ratio of PDE-A to PDE-B activity in membrane preparations is ap- proximately lO:l, whereas in soluble extracts this ratio is reversed (11). Furthermore, in contrast to the PDE-B reac- tion, PDE-A activity is inhibited at low concentrations of Ca’+ ions, suggesting that this phosphodiesterase serves as an additional locus of regulatory control, maintained by this cation.

This novel system is dually intriguing as it is the only known form of regulation of a cellulose-synthesizing enzyme currently accessible to biochemical analysis, and since cyclic dinucleotides, such as c-di-GMP, have not been recognized previously to be of natural occurrence. The chemical synthesis of c-di-GMP (cGpGp) by a modified hydroxybenzotriazole phosphotriester approach is described here in IX1 for the first time. Furthermore, several analogs of c-di-GMP, bearing modifications in either the ribose, phosphate, or guanine moieties, have been prepared with the objective of gaining more insight into the nature of the c-di-GMP regulatory system. Thus, various cyclic homodinucleotides, such as cBpBp (B = adenosine, @dine, inosine, uridine, xanthosine, or deoxyguanosine), cyclic heterodinucleotides cBpGp, a tri- nucleotide cGpGpGp, and a pair of c-di-GMP derivatives containing a chiral phosphorothioate linkage, cGp(S)Gp, were synthesized and tested for biological activity as effecters and substrates, respectively, of cellulose synthase and PDE-A.

MATERIALS AND METHODS’

C~m~c&-[U-14C]UDP-glucose and [CX-3*P]GTP were purchased from The Radiochemical Centre. Amersham. United Kingdom. “P- Labeled c-di-GMP was prepared’and purified as described (9). Bac- terial alkaline phosphatase Type III and all chemical reagents and nucleotides, unless otherwise specified, were purchased from Sigma. Thin-layer plates of polyethyleneimine-cellulose were from Machery Nagel (polygram Cel 300 PEI). RP-18 HPLC columns were from Merck. A detailed description of synthetic procedures for novel cyclic nucleotides appears in the accompanying Miniprint Supplement. All nucleotides were >95% pure, as judged by HPLC.

Memhne Preparations-Membrane preparations from A. xy- hum cells were prepared and stored as described previously (5).

Enzyme Assays-Cellulose synthase activity was assayed as de- scribed nreviouslv (5). The standard assav mixture contained, in a final voiume of 61 ‘ml, 50 mM Tris-HCl, pH 8.3, 20 mM MgC$, 2.5 mM CaCh. 0.1 mM EDTA. 20 UM UDPlYXlucose (50 cnm/nmol).

- , . a” . , ,

membrane preparation (0,2-0.4 mg of protein), and additional com- ponents as indicated. Incubation was for 5 min at 30 OC. Reactions were terminated, and the “C product formed was determined as described (5).

Standard PDE-A reaction mixtures contained, in a final volume of 0.1 ml, membrane preparations (l-5 pg of protein), 50 mM Tris-HCl, pH 8.3,20 mM MgCh, and either “P-labeled c-di-GMP (250,000 cpm/ reaction) or unlabeled nucleotides as indicated. Incubations were for 5 min at 30 ‘C. When employing radiolabeled substrate, reactions

* Portions of this paper (in&ding part of “Materials and Methods” and Tables S-1 and S-2) are presented in miniprint at the end of this paper. Miniprint is easily read with the aid of a standard magnifying glass. Full size photocopies are included in the microfilm edition of the Journal that is available from Waverly Press.

were terminated by application of a 20-~1 aliquot to PEI-cellulose TLC plates. Following chromatography in 1.5 M KHzPOd, the residual radiolabeled substrate was detected by autoradiography, cut out, and counted (9, 11). Enzyme activity was calculated from the percentage of substrate consumed in the course of the incubation. Unlabeled reaction mixtures were assayed by reverse phase HPLC chromatog- raphy in KHzPOd buffer as described below.

Reverse Phase 3fPX!-HPLC was performed on a Merck-Hitachi liquid chromatograph equipped with a &mm flow-through cell UV detector (252 nm), He-spargling unit, and a Spectra-Physics 4100 computing integrator. Analytical runs, employing a lo-~1 injection loop, were carried out at 1 ml/min in 0.1 M NaH2POa, pH 4.8, on a RP-18 2.50 X 4-mm Licrosphere column and a 50 X 4-mm RP-18 pre- column submerged in a 40 C water bath. Unless otherwise specified, the following gradient of 20% acetonitrile in the above buffer was used 0 min (0%); 5 min (O-5%); 10 min (5-5%); 15 min (5-12%); 37 min (12-75%); 48 min (75-O%). Preparative purifications were carried out using a RP-18 250 X lo-mm column run in 50 mM triethylam- monium bicarbonate, pH 7.1, at 6 ml/min at room temperature, employing linear gradients (O-10%) of methanol.

RESULTS

Activation of Cellulose Synthase by c-di-GMP-In crude membrane preparations from A. xylinum, the rate of incor- poration of glucose from UDP[W]glucose into alkali-insolu- ble l,4-fl-D-[14C]glucan displays a low activity level when compared with the synthetic capacity of the intact cell (7). In the presence of c-di-GMP, stimulation of synthase activity ranges from 50 to up to 200-fold (a-10). This overall fold activation effect varies among individual enzyme preparations and thus has not been calibrated as a standard value. Under the assay conditions employed here, the rate of c-di-GMP- stimulated activity in membrane preparations, expressed per mg of cells, approximates 40% of that of the intact organism.

The kinetics of activation, as a function of c-di-GMP con- centration, are relatively complex (Fig. 1). In the range of 0.3-4.0 MM, the dependence of cellulose synthase activity on effector concentration manifests a typical Michaelis-Menten relationship, from which a K,, value for c-di-GMP of 0.31 pM

can be derived from the linear portion of the plot. Below this range, synthase activity falls off rapidly with the concentra- tion of c-di-GMP, and overall kinetics of activation are suggestive of positively cooperative interactions. In order to examine this further, the stability of the activator in the assay system was examined, particularly since crude membranes contain phosphodiesterase activity which degrades c-di-GMP. In incubations carried out under parallel conditions in the presence of ‘2P-labeled c-di-GMP, partial degradation of the exogenously added activator was observed to be most promi- nent at the low initial concentrations of c-di-GMP, ranging from 2.5% (4 pM) to 50% (0.1 pM) (data not shown). This degradation can be attributed to the endogenous PDE-A activity of membrane preparations which, even when inhib- ited >99% by 2.5 mM CaCL, apparently retains sufficient activity to effectively reduce the concentration of c-di-GMP in synthase assay mixtures. The data in Fig. 1 have been corrected for this reduction in activator concentration occur- ring in the course of the assay period by employing an average value obtained from the initial and final c-di-GMP concen- tration. Thus, the deviation from typical Michaelis-Menten kinetics at low (<0.3 pM) concentrations of the activator cannot be accounted for on the basis of steady degradation. A more detailed kinetic analysis of c-di-GMP degradation under these particular assay conditions has not yet been performed. Notably, however, the kinetics of c-di-GMP acti- vation of highly purified preparations of cellulose synthase (l2), which is essentially devoid of PDE-A activity, do not

by guest on Decem

ber 24, 2019http://w

ww

.jbc.org/D

ownloaded from

500

E” 300 z .- E

27 75 E 200

,a

z

100

0

Cyclic Dinucleotides in the Regulation of Cellulose Synthesis 18935

0.0 0.5 1 .o 1.5

vi[c-di-GMP] (pmolelminimg prot/nM)

FIG. 1. Kinetics of c-di-GMP activation of cellulose syn- thase. Cellulose synt.hase activity in membrane preparations (200 pg of protein) was assayed in standard reaction mixtures in the absence and presence of various concentrations (0.1-4.0 PM) of c-di-GMP as described under “Materials and Methods.” The level of enzyme activ- ity measured above basal level (8 pmol/min/mg protein) was used in the Eadie-Hofstee plot displayed. The data have been corrected for partial breakdown of c-di-GMP which occurs in the course of the assay. Residual c-di-GMP was determined in parallel incubation mixtures, containing “P-labeled c-di-GMP and unlabeled UDP-glu- case, following termination of the reaction by addition of 10% tri- chloroacetic acid, centrifugation, and then thin-layer chromato- graphic analysis of the supernatant as described under “Materials and Methods” for standard PDE-A reaction assays. The values for c- di-GMP used to plot the curve were taken from the average of the initial and residual concentrations.

display such a sigmoidal pattern.” Analysis of the rate of production of alkali insoluble product

as a function of UDP-glucose concentration (Fig. 2) in either the absence or presence of c-di-GMP yields a plot that is typical of Michaelis-Menten enzyme kinetics with respect to substrate. The apparent K,,, value (125 PM) calculated for UDP-glucose is independent of effector concentration, as reported previously for the activation produced by GTP and (diguanylate cyclase-containing) crude cell extract (5, 13). Rather, c-di-GMP stimulation of the enzyme is a result of an increase in the V,,, parameter of synthase activity.

Phosphodiesterase A Activity in Membrane Fraction-The hydrolysis of c-di-GMP to pGpG is defined as PDE-A activity. Employing an assay based on measuring the rate of conversion of “‘P-labeled c-di-GMP to its degradation products, resolved by thin-layer ion-exchange chromatography (9, ll), it has been possible to analyze the initial rate kinetics of the PDE- A reaction in membrane preparations (Fig. 3). The PDE-A activity conforms to typical Michaelis-Menten saturation ki- netics with respect to the concentration of c-di-GMP (appar- ent K,,, = 0.25 FM). Under the conditions employed here (20 mM MgC12), the V,,,,, parameter of the reaction is reduced by 53% in the presence of 100 PM CaCl,, whereas the apparent affinity for substrate appears to increase slightly. Much higher concentrations (2.5 mM) of this cation are required to effec- tively inhibit the reaction by up to 99% (data not shown).

Synthesis of Cyclic Deoxyribonucleotide Analogs-The

” R. Mayer and M. Benziman, manuscript in preparation.

d[UDP-glucose] (nmotelmitdmg prot/mM)

FIG. 2. Effect of c-di-GMP on the kinetics of cellulose syn- thase activity with respect to UDP-glucose. Cellulose synthase activity was measured in membrane preparations (400 pg of protein), as described under “Materials and Methods,” in both the absence (inset) and presence of either 0.5 FM (0) or 10 ~.LM (0) c-di-GMP, at varying concentrations of UDP[W]glucose (0.02-l mM; 400,000 cpm/ reaction).

j/

0 2 4 6 8 10 I2 14 16

v/[c-di-GMPI (nmole/min/mg proUa)

FIG. 3. Kinetics of PDE-A activity and effect of CaC12. Mem- brane preparations (1 rg of protein) were assayed for PDE-A activity, as described under “Materials and Methods,” in the presence of varying concentrations of “‘P-labeled c-di-GMP (0.1-3.0 PM; 500,000 cpm/reaction) in both the absence (solid squares) and presence of 100 pM (open squares) CaCL. Enzyme activity as a fun&on of substrate concentration is displayed in Eadie-Hofstee form.

chemical synthesis of cyclic deoxyribonucleotides 8 (Scheme 2, Fig. 4) was achieved in two distinct stages by a modified hydroxybenzotriazole phosphotriester approach (14). In the first stage, as outlined in Scheme I (Fig. 4), a properly pro- tected linear building block, enclosing 2 or 3 deoxyribo-

by guest on Decem

ber 24, 2019http://w

ww

.jbc.org/D

ownloaded from

18936 Cyclic Dinucleotides in the Regulation of Cellulose Synthesis

z-xx:0 b:X :S

Scheme 1

I

1. M*

2 Pdlplphl,l‘/ PIPhI]/ BuNkI

B- A& - N6-Ehzoykdanin-9-yl CAII - N4-Anisoykytoain-l-y1 GDPA - N2-LXphenylac@-9uenin-9-yl INPSE - 06~[2-(4-nitrophenylsulfanyl)elhyl~~inosin~9-yl U - uddin-1 -yl X(NPE)2 - 02,06.bis[2-(4-nilrophenyl~elhyl~-xanlh0sin~9-yl

FIG. 4. Scheme for the synthesis of cyclic deoxyrihonucleotides. See text for details.

nucleotide units (i.e. 6 or 6) was synthesized employing the bifunctional phosphorylating reagent 2a in a two-step phos- phorylation process (15). For the preparation of the phospho- rothioate-containing dimer 5 (X = S; B = GDP*), phosphoro- thioylating reagent 2b, applied previously for the synthesis of 3’,5’-cyclic ribonucleotide phosphorothioates (16) and the introduction of a phosphorothioate linkage in 2,5-A deriva- tives (1’7), was used for coupling 1 with 4. In the second stage (see Scheme 2, Fig. 4), dimer 5 or trimer 6 was converted into a building block suitable for cyclization (i.e. 7). For the temporary protection of the 3’-terminal phosphodiester, the ally1 (18) group was introduced by coupling 5 or 6 with ally1 alcohol. After removal of the 4,4’-dimethoxytrityl group (19) and the ally1 function (18), an intramolecular condensation was effected with 2,4,6-triisopropylbenzenesulfonyl-3-nitro- 1,2,4-triazole (TPSNT) (20), leading to the fully protected cyclic di- or trimer 8.

Thus, nucleoside 1 was phosphorylated with 2a in dioxane in the presence of pyridine. After 5 min, TLC analysis indi- cated the complete formation of 3 (X = 0). A slight excess of 4, having its 5’- and 3’-hydroxyl function unprotected, was added. Work-up of the reaction mixture after 45 min, followed by short column chromatography (21), afforded dimer 5 (X = 0) in an average yield of 75%.

A similar procedure was followed for the synthesis of dimer 5 (X = 5’) with the exception that phosphorylating agent 2b was used for the condensation and N-methylimidazole was added in the second phosphorylation step. After work-up, the diastereomers were separated by short column chromatogra- phy and isolated in a total yield of 69%.

Trimer 6 (X = 0, B = GDP*; Rz = 0-THP) was isolated in

81% yield after coupling 5 (X = 0; B = GDP*; R’ = O-THP) with 4 (B = GDP*; R’ = 0-THP), followed by work-up and purification by short column chromatography.

Next, compounds 5 or 6 were co-evaporated with dioxane and phosphorylated with 2a in the presence of a slight excess of pyridine. After 5 min, when TLC analysis indicated com- plete conversion of starting material into a compound with zero mobility, ally1 alcohol was added. After 30 min, work-up and purification afforded pure 7 in an average yield of 79%. The dimethoxytrityl group at the 5’-position was removed with toluene-p-sulfonic acid and the ally1 function at the 3’- terminal phosphotriester was deblocked within 5 min with palladium tetrakistriphenylphosphine/triphenylphosphine/ n-butylamine in tetrahydrofuran. The reaction mixture was co-evaporated and diluted with anhydrous pyridine and TPSNT was added. Cyclization reactions were normally com- plete within 2 h. Usual work-up and purification by short column chromatography furnished cyclic deoxyribonucleotides 8 in an average overall yield of 62%.

The fully protected compounds 8 were deblocked by treat- ment with oximate ions (22), followed by ammonolysis at 50 YZ and in case R’ = 0-THP by acid hydrolysis (23). Crude deprotected cyclic nucleotides 8 thus obtained were purified by anion-exchange chromatography and converted into their sodium salts. The homogeneity and identity of the deprotected cyclic deoxyribonucleotides 6 was acertained by fast protein liquid chromatography analysis, ‘H and “P NMR spectros- copy, mass spectroscopy, and UV absorption (Table I). Minor contaminations from some synthetic preparations (cXpXp, cIpGp, cGpGpGp, and cGp(S)Gp:Rp- and Sp-diastereomers) were removed by preparative HPLC. All of the compounds

by guest on Decem

ber 24, 2019http://w

ww

.jbc.org/D

ownloaded from

Cyclic Dinucleotides in the Regulation of Cellulose Synthesis 18937

TABLE I Relevant data of deprotected cyclic deoxyribonucleotides

Compound alp NMR” ‘H NMRb 6 rnlll

~GPGP

cdGpGp

-0.62

-0.11 -0.20

cdGpdGp

cGp(S)Gp: W

6%)

~GPGPGP

~IPGP

-0.24

56.89 -0.21 55.35

-0.54 -0.12

-1.03 -1.15

CXPGP

~CPGP

-0.74 -0.83

-0.92 -1.08

CIPIP -0.88

CXPXP

CCPCP

-0.71

-1.21

~APAP -1.18

CUPUP -0.85

7.84 (5; H-8) 5.87 (d; 1.8 Hz; H-l’) 8.09 (s: H-8) 8.02 (sf H-8j 6.02 (t; 5.s Hz; H-l’) 5.93 (s; H-l’) 7.94 (s; H-8) 6.21 (t; 5.0 Hz; H-l’)

7.91 (s; broad; H-8) 5.90 (s; broad; H-l’) 7.93 (s; broad; H-8) 5.87 (s; broad; H-l’) 8.01 (s: H-8) 5.94 (d; 7.8 tiz; H-l’) 7.99 (s; H-8) 7.97 (s; H-8) 7.73 (s; H-2) 5.80 (m; H-l’) 8.04 (s: H-8) 8.00 (s; H-8) 5.89 (m; H-l’) 7.84 (s; H-8) 7.75 (d; 7.5 Hz; H-6) 6.01 (d; 7.5 Hz; H-5) 5.88 (m; H-l’) 8.01 (s; H-8) 7.77 (s; H-2) 5.87 (s; H-l’) 7.88 (s; H-8) 5.82 (s; H-l’) 8.02 (d; 7.6 Hz; H-6) 5.96 (d; 7.6 Hz; H-5) 5.82 (s; H-l’) 8.24 (s; H-8) 7.87 (s; H-2) 5.99 (s; H-l’) 7.97 (d; 8.1 Hz; H-6) 5.82 (d; 8.1 Hz; H-5) 5.80 (s; H-l’)

252

245

236

252

250

253

251

250

261

249

250

271

259

262

’ ‘iP NMR in DPO. Chemical shifts are in parts/million relative to 85% HzPO+

b ‘H NMR in D,O. Tetramethylammonium chloride was used as internal reference. Chemical shifts are in parts/million relative to tetramethylsilane.

tested below were judged to be of 298% purity as judged by HPLC analysis.

Cellulose Synthuse Effector Specificity-Seven compounds, cdGpGp, cdGpdGp, cIpGp, cIpIp, cXpGp, and cGp(S)Gp (both R and S isomers), from the group of 13 cyclic deoxyri- bonucleotides prepared above, proved to be effective activa- tors of cellulose synthase activity (Fig. 5). Although in each case enzyme activation displays saturation kinetics with re- spect to the cyclic nucleotide concentration, most of the activators differ with regard to K, value (estimated from the saturation curve) and, in some cases, do not activate the enzyme to the maximal possible level. An extreme example is presented by the diastereomeric pair of monophosphothioate derivatives: whereas the S-isomer (K,, = 0.35 /IM) appears to be roughly comparable in efficacy with the natural (c-di- GMP) activator, the affinity displayed toward the R-isomer (K, = 10 pM) is more than an order of magnitude lower and, in this case, the level of enzyme activity at the plateau concentration approaches only 70% that of the potential maximal level of activity (i.e. at saturating concentrations of c-di-GMP).

None of the other cyclic nucleotides tested, cApAp, cXpXp, cCpGp, cCpCp, cUpUp, and cGMP, affect synthase activity

when present in standard assay mixtures at concentrations ranging from 1 to 100 PM, with the exception of the cyclic trimer, cGpGpGp, which produces a marginal (3-4-fold), but consistently observed, activation at concentrations above 50 PM. A number of noncyclic nucleotides related to the c-di- GMP structure, pGpG, GpG and 5’-GMP, also proved to be without effect when tested in this concentration range. The impotency of all of these compounds is due to a low affinity for the enzyme, as evidenced by the results of competition experiments wherein the nucleotides were tested for the abil- ity to interfere with c-di-GMP stimulation of synthase activ- ity. None of the compounds, when included at a concentration of 100 PM in standard assay mixtures containing 1 PM c-di- GMP affected the level of enzyme activity, when compared with mixtures containing 1 PM c-di-GMP alone (data not shown). Based on these results, the affinity of cellulose syn- thase for any of these structures is at least two orders of magnitude lower than that for the c-di-GMP activator. As with c-di-GMP, all of the activators appear to be relatively more effective at higher concentrations, i.e. display sigmoidal saturation curves.

Stimulation of cellulose synthase activity by c-di-GMP in crude membrane preparations is greatly enhanced in the presence of CaC12, which is a potent inhibitor of PDE-A activity (10, ll), and this salt (2.5 mM) is included in standard synthase reaction mixtures. When CaClz was excluded from the reaction mixtures, the stimulation of activity produced by either the natural or synthetic activators was dramatically reduced (data not shown). To judge this effect more closely, aliquots from activator-containing standard reaction mixtures were tested for stimulatory activity following a preincubation period under conditions parallel to that of the synthase assay (Table II). In the absence of CaCl*, each of the activators was destroyed in the course of the standard incubation period (not shown). However, although in general CaC12 preserves the activators, the analogs cIpIp and cGp(S)Gp:R, appear to be less protected, suggesting that their relative affinities might actually be 2-3-fold higher than those estimated from the saturation curves (Fig. 5). The results of more direct experi- ments (see below) confirm that the Ca*+ requirement is at- tributable to inhibition of cyclic dinucleotide-degrading phos- phodiesterase (PDE-A) activity occurring in the membrane preparation.

Stability of Cyclic Dinucleotides and Related Compounds in the Presence of Membrane Preparations-The stability of the synthetic cyclic nucleotides, as well as of several other related phosphodiester-containing guanyl nucleotides, toward expo- sure to the phosphodiesterase activity of membranes was assessed employing reverse phase HPLC analysis of standard PDE-A reaction mixtures containing 10 PM (40 times the K,,, for c-di-GMP) nucleotide test substrate (Table III). Of all of the compounds tested, eight (cdGpGp, cdGpdGp, cIpGp, cIpIp, cXpGp, cXpXp, cGp(S)Gp:S,, and cGp(S)Gp:R,) are degraded at a pace comparable with that of c-di-GMP, judging by either the rate of consumption of substrate or appearance of product(s). Both the linear dimer pGpG and the cyclic trimer cGpGpGp are also degraded, although at slower rates. Presumably, as is true for c-di-GMP hydrolysis which yields predominantly the linear dimer pGpG under these conditions, the cyclic nucleotide degradation products observed here re- sult from the initial hydrolysis of a single phosphodiester bond in the molecule. However, aside from the case of the monophosphothioate derivatives (see below), the structures of the cyclic nucleotide degradation products have not been characterized, although their relatively high retention time values indicate that the major products detected in each case

by guest on Decem

ber 24, 2019http://w

ww

.jbc.org/D

ownloaded from

18938 Cyclic Dinucleotides in the Regulation of Cellulose Synthesis

FIG. 5. Activation of cellulose synthase by synthetic analogs. Cel- lulose synthase activity was assayed as described under “Materials and Meth- ods” in the presence of varying concen- trations of the synthetic analogs, as in- dicated (sold .s/mpes). Enzyme activity is expressed as the percent of maximal activity, i.e. at saturating concentration of c-di-GMP (open S@XS). The rmrnkrs in pare&zesk refer to the Ka value (the concentration required for half-maximal stimulatory effect) relative to that for c- di-GMP (relative Ka = l), as estimated from the saturation curves. The data have not been corrected for degradation and so the actual Ka for c-di-GMP here is slightly (~25%) higher than that ex- trapolated in Fig. 1. In at least some cases (cIpIp and cGp(S)Gp:R& the rela- tive affinity may actually be 2-3.fold higher, due to partial enzymatic degra- dation in the assay system (see text for details).

TABLE II

Stability of actiuators in the presence oj nzembrunes Standard cellulose synthase reaction mixtures (UDP-glucose omit-

ted) containing 10 pM cyclic dinucleotide as indicated were incubated under standard conditions (5 min at 30 “C) and the reaction termi- nated by heating at 100 C for 3 min, followed by centrifugation. The resultant supernatants were tested for the ability to stimulate cellu- lose synthase activity in the standard assay system as in Fig. 5. Controls were carried out by substituting heat-inactivated (3 min at 100 -C) membrane preparations in the preincubation mixture.

Activator Supematant

vohne added

d

Stimulation of cellulose synthase activity

Control Preincubated

cpm above basal leuel

CGPGP

cdGpGp

cdGpdGp

CIPIP

~IPGP

CXPGP

cGp(S)Gp: SP

4

5 18,000 10 25,200

5 17,500 10 26,000

10 9,000 20 14,400

20 7,600 40 11,200

IO 14,400 20 20,600

20 6,200 40 12,000

5 18,000 10 25,200

20 2,570 40 4,530

17,100 23,900

18,000 24,000

8,700 14,000

3,300 6,400

13,500 19,600

7,000 11,200

16,200 22,690

1,150 2,550

are probably the corresponding linear dimers or trimer. As- suming that the same reaction pathway applies in every degradation reaction, the number of products formed is con- sistent with a mechanism in which either of the two phospho- diester bonds in the cyclic dinucleotide is susceptible to cleav- age, yielding the corresponding 5’-phosphorylated open di- mer. Thus homodinucleotides (cGpGp, cXpXp, cdGpdGp, cIpIp) yield a single product, whereas heterodinucleotides (cdGpGp, cXpGp, cIpGp) give rise to two products, in yields which presumably reflect preferences related to the rotational orientations in which binding of the cyclic structure to the enzyme might occur.

log [cyclic dinucleotide] (nM)

None of the other cyclic dinucleotides tested, cApAp, cCpGp, cCpCp, and cUpUp, nor the cyclic monomer cGMP and linear dimer GpG, were judged to be labile to the PDE-A reaction conditions, even when the amount of membrane preparation employed in the degradation assay was increased to an amount sufficient to completely hydrolyze the equiva- lent of twice the quantity of c-di-GMP. In the case of the four cyclic dinucleotides, no marker compounds of possible hydro- lytic products were available, and the possibility was consid- ered that the apparent stability of these compounds might actually be due to a poor chromatographic resolution of prod- ucts. However, attempts to improve the separation by modi- fying the conditions (i.e. methanol gradient) of elution in the HPLC analysis did not reveal UV-absorbing peaks in the reaction mixtures in addition to that of the original material. The observed stability of most of these compounds to PDE- A activity, namely, cApAp, cCpCp, cUpUp, and cGMP, is compatible with their low affinity for the enzyme, as deter- mined in inhibition assays (see below).

All of the cyclic dinucleotide degradations, as tested in Table II, are completely inhibited in the presence of 2.5 mM CaCIZ, implying that a common Ca2+-sensitive enzyme (i.e. PDE-A) executes each of these reactions. However, the results obtained from the preincubation assays mentioned above (Table III), in which the substrates were subjected to a 40- fold greater amount of membranes, indicated that the extent of this inhibition is disproportionate with regard to substrate. Indeed, when assayed at sub-saturating (100 FM) concentra- tions of CaCl*, the hydrolysis of some analogs (cIpIp, cGp( S)Gp:&, cXpXp) proceeds relatively unimpeded (15- 25% inhibition) relative to the 50-60% retarding effect ob- served on the hydrolysis rate of c-di-GMP and the other substrates (Table III). Under these conditions, the degrada- tion rate of cGpGpGp was too slow to accurately estimate the inhibitory effect, although the hydrolysis of this compound is also probably impeded by Ca2+, since its stimulatory effect on synthase activity in membrane preparations was judged to be Ca’+-dependent. The same applies to the pGpG-degrading reaction in membrane preparations; in this case, employing radiolabeled substrate (10,ll) and the PEI-TLC assay, a 20% inhibition was observed at 2.5 mM CaCIZ (not shown).

Analysis of Monophosphothioate Degradation Products- The degradation of the monophosphothioate-containing de- rivatives (the Rp- and &,-diastereomers) was studied in detail

by guest on Decem

ber 24, 2019http://w

ww

.jbc.org/D

ownloaded from

Cyclic Dinucleotides in the Regulation of Cellulose Synthesis

TABLE III HPLC analysis of PDE-A reaction mixtures containing cyclic dinwleotides and related compounds

Standard PDE-A reaction mixtures containing membranes (20 pg of protein) and 10 pM nucleotide were incubated for 5 min at 30 “C and the reaction terminated by heating at 100 ‘C for 3 min, followed by centrifugation. The supernatants were analyzed by RP-HPLC as described under “Materials and Methods.” Gradient solvent C: 0 min (0%); 5 min (O-5%); 10 min (5-5%); 15 min (5-12%); 37 min (12-75%); 37-48 min (75-O%). Reaction rat,es were estimated from the percent of total absorbance (i.e. substrate and product) appearing as product. The relative rate of degradation is the ratio of the observed rate to that of c-di-GMP in parallel incubations. Under these reaction conditions, typically 25% of the c-di-GMP substrate was converted to pGpG and <2% to 5’-GMP (280 s). Retention times recorded for other guanyl nucleotides were guanosine, 1015 s; 3’-GMP, 467 s; 2’-GMP, 896 s. In cases where two major products were detected, the percent of the total absorbance present in each is presented in parenthesis. In some cases, additional products occurred, but these accounted for less than 2% of the total absorbance. The effect of CaClz on the degradation rate was tested in parallel incubation mixtures containing the addition of 100 PM CaCl*.

18939

Compound Retention Relative rate of Inhibition in presence time Product(s) retention time degradation Of 100 PM Cd&

9%

CGPGP 1169 1059 1.00 60 cdGpGp 1136 1111 (12%), 1312 (5%) 0.70 55 cdGpdGp 1109 1328 0.80 50 CIPGP 1026 1060 (13%), 1207 (8%) 0.80 50 CIPIP 901 1214 0.75 15 CXPGP 849 1060 (13%), 1170 (7%) 0.80 55 CXPXP 616 1493 0.65 25 CCPGP 410 None detected co.05 CAPAP 1273 None detected <0.05” CUPUP 374 None detected CO.05” CCPCP 310 None detected <0.05” cGp(S)Gp:

RP 1451 1472 1.50 15 SP 1706 1454 0.70 55

CGPGPGP 1478 1464 0.05” cGMP 1025 None detected <0.05” PGPG 1059 280 0.10 GPG 1616 None detected <0.05”

’ In these reaction mixtures both the amount of membranes and the duration of incubation were doubled.

in order to ascertain the relative stability of the modified phosphodiester bond towards PDE-A activity. Assuming that PDE-A action forms the corresponding 5’-phosphorylated linear dimer products from each of these, two products would be expected from each diastereomer if the hydrolysis of either phosphodiester bond in the molecule proceeds at equal rates. However, in the case of either diastereomer, only a single product was detected upon HPLC analysis of PDE-A reaction mixtures allowed to proceed until complete consumption of the substrate (Fig. 6, a, b, k and g). The formation of a unique hydrolysis product (as indicated by co-chromatography of the reaction mixtures which yielded two distinct peaks) from each diastereomer suggests that each yields the corresponding 5’- phosphorylated linear dimer in which the chiral phospho- thioate group is contained in internal phosphodiester linkage, e.g. pG(S)pG. Further characterization supports this struc- tural assignment. Incubation of the reaction mixtures with BAP yielded clearly distinguishable products (Fig. 6, c and h), confirming that the original degradation products of the S,- and R,-diastereomers possess a mono-esterified phosphate group and a modified internal structure, since neither BAP product co-chromatographs with GpG (relative time = 1616 s compared with 1673 and 1935 s for the PDE-A/BAP S,, and Rp products, respectively). Furthermore (Fig. 6, d, e, i, and j), each of the original degradation products is susceptible to periodate oxidation (P-elimination reaction), which conforms with the free 3’-terminal ribose structure in which the BAP- labile group occupies the 5’-position.

Nucleotide Inhibition of the PDE-A Reaction-To comple- ment the degradation studies described above, and to assess the overall specificity of the PDE-A reaction, the initial degradation rate of 32P-labeled c-di-GMP (1 PM) was com-

pared with reaction rates in the presence of equimolar and excess concentrations of various nucleotides (Table IV). With the exceptions of GMP, GDP, and ATP, all of the compounds tested were observed to inhibit the degradation of c-di-GMP, although to widely varying extents. In general, the most potent inhibitors are those cyclic dinucleotides (cdGpGp, cdGpdGp, cIpGp, cIpIp, cXpGp, cXpXp, and phosphothioate diastereomers) rated as good substrates of the degradation reaction. Interestingly, a gross correlation exists between the potency of these compounds as cellulose synthase activators and as PDE-A inhibitors; in general, the more effective acti- vators are also more effective inhibitors. Albeit of slightly lower potency, several other guanyl nucleotides, cGpGpGp, pGpG, cCpGp, pppGpG, GpG, and GTP, are also inhibitors of the phosphodiesterase reaction. Although the former two compounds might possibly be substrates for the enzyme, the latter were judged to be >95% stable under the standard reaction conditions, by either HPLC analysis (cCpGp and GpG) or by PEI-TLC (pppGpG and GTP), employing radio- labeled material as substrate (10) (not shown). The other cyclic nucleotides tested, including cGMP, cCpCp, cUpUp, and cApAp, all of which are also >95% stable under the assay condition, produce detectable effects on the reaction rate only when present at 25-fold molar excess to substrate.

DISCUSSION

Although the effect of c-di-GMP on cellulose synthase activity is clearly dramatic, the r&on d’etre of this unique control mechanism is still open to speculation. As an allosteric effector, c-di-GMP is unusual in that its stimulatory effect on the cellulose synthase reaction is manifested as an increase in the V,,, parameter of enzyme activity, rather than the

by guest on Decem

ber 24, 2019http://w

ww

.jbc.org/D

ownloaded from

18940

a

b

n d

L- I

e

Cyclic Dinucleotides in the Regulation of Cellulose Synthesis

f

t-J-- TABLE IV

Nuckotide itzhibition oj the PDE-A reaction Standard PDE-A reaction mixtures containing membranes (2 pg

of protein) and 1 fiM “P-labeled c-di-GMP were assayed as described in Fig. 3, in the absence and presence of either 1, 5, and 25 KM unlabeled nucleotide as indicated. In the absence of added nucleotide, typically 20-30% of the labeled substrate was degraded in the course of the assay. For comparison of effects, K, values were calculated for each individual concentration tested based on the assumption that the inhibitor is competitive with c-di-GMP (apparent Km = 0.25 PM), and the average of these for each compound is presented as a relative value to that of (unlabeled) substrate (relative K, = 1).

h

I . . . . . . 3 -a . ,

0 IO m 30 40 minutes

FIG. 6. Analysis of monophosphothioate degradation prod- ucts. Standard PDE-A reaction mixtures contained 50 pg of protein membranes and 10 pM of either cGp(S)Gp:$, or cGp(S)Gp:Rp, in a final volume of 0.25 ml. Following incubation at 30 ‘C for 20 min, the mixtures were centrifuged and the resultant supernatant heated at 100 ‘C for 4 min. Zero time controls were carried out by substituting heat-inactivated membrane preparation in the reaction mixture. For phosphatase treatment, 0.05 ml of the PDE-A reaction mixtures were incubated at 30 ‘C for 10 min in the presence of 0.01 unit of BAP. Prior to NaIOd treatment, the reaction products were purified by HPLC: 0.1 ml of each PDE-A reaction mixture was run as described below and the product peaks collected in 1 ml, rotor-vaporized to dryness, and resuspended in 0.1 ml of HzO. To each of these was added, in the dark, 5 ~1 of 25 mM NaI04 and allowed to stand at 22 ‘C for 30 min. Following this, 15 ~1 of 1.3 M lysine, pH 8.75, was added and the mixtures incubated for an additional 40 min at 40 ‘C. The reactions were then exposed to room light and 2 ~1 of 4% glycerol

Inhibition ohserved at

~GPGP cdGpGp cdGpdGp ~IPGP CIPIP ~XPGP CXPXP ~CPGP ccpcp ~APAP CUPIJP cGp(S)Gp:

& &

~GPGPGP cGMP PGPG GPG GDG (2'.5')

45 65 45 25 20 45 1.5

5 <5 <5 -3

25 65 10

10 10

80 >95 1.0 95 z-95 0.3 80 1.0 75 95 1.6 40 4.6 75 95 1.1 40 75 5.1 20 15.6

<5 10 180.0 <5 10 180.0 <5 10 180.0

50 >95

40 80 <5 10 30 65 30 60

30

3.2 0.3 6.1

180.0 9.1

10.0 46.7

180.0 180.0

k-380.0 b-380.0 b-380.0

80.0 9.3

b380.0

G;A APA 3’.GMP 5’.GMP GDP GTP

10 10

<5 <5 <5

<5 c5 20

10 35 60 <5

result of enhanced affinity for the UDP-glucose substrate. This type of regulation is usually associated with the physio- logical situation in which the intracellular concentration of substrate is well above the saturating level, although in A. xylinum cells that of UDP-glucose probably does not exceed more than twice the synthase apparent K,,, value (125 PM). Although this observation does not exclude the hypothesis that the rat,e of c-di-GMP turnover maintains synthase activ- ity in step with the metabolic state of the cell, the cyclic nucleotide might be involved, as well, in an organizational role with regard to the mechanism of the synthase reaction. However, since the current assay method does not distinguish between the various stages of polymerization, such as polyglu- can chain initiation, elongation, or termination, it has not been ascertained if the effect of c-di-GMP on the synthase reaction is to accelerate a specific step in this process. The recent finding that cellulose synthase in both highly (500- fold) purified (12) and immobilized (24) form retains its sensitivity to c-di-GMP suggests that the activator binding site is integral to the structure of the enzyme. The partially

added. The controls shown were carried out by addition of Hz0 instead of NaIO.,. In additional controls performed, neither the phos- phatase nor the NaIOd treatment were observed to affect the intact analogs (prior to exposure to PDE-A). HPLC analysis was performed as described in Table III. Chomatograms a-e refer to S,,-diastereomer and j-j to Rp-diastereomer: a and j, zero times; b and g, PDE-A reaction mixtures; c and h, BAP treatment; d and i, isolated (NaI04 control) PDE-A products; and e and j. NaIOJLysine treatment.

by guest on Decem

ber 24, 2019http://w

ww

.jbc.org/D

ownloaded from

Cyclic Dinucleotides in the Regulation of Cellulose Synthesis 18941

sigmoidal kinetics observed under the conditions employed here warrant further examination. That cooperative interac- tions are not an inseparable regulatory property of the syn- thase has been demonstrated for c-di-GMP activation in highly purified enzyme preparations devoid of PDE-A activ- ity.” For a more definitive analysis of affinities and mecha- nisms, the sensitivity of the crude and purified forms of the synthase should be compared with regard to c-di-GMP and its analogs.

Considering the unusual structure of c-di-GMP, the high degree of specificity displayed by its molecular targets is not unexpected. In general, modifications in the c-di-GMP struc- ture reduce the affinity of the molecule for both cellulose synthase and PDE-A, as judged by the ability to interact as activators, substrates, or inhibitors. In this respect, the sig- nificance of the cyclic structure is evident in the lower affinity of pGpG, relative to that of c-di-GMP, by a factor of lo-fold for PDE-A and by at least 3 orders of magnitude for the synthase. The intact guanine rings, with the exception of the C-2 amino group, also appear to be particularly relevant to the biologically active cyclic dinucleotide structure, which is essentially devoid of activity in compounds containing alter- native bases such as cApAp, cCpCp, cUpUp, and cCpGp. At the C-2 position, which has been studied in the most detail, the modifications have both partial and absolute effects on activity; substitution of the amine with carbonyl either re- duces (cXpGp) or destroys (cXpXp) affinity for the synthase, whereas PDE-A also recognizes both molecules less effectively as substrates; replacement of the amine with hydrogen (cIpGp, cIpIp) results in compounds which are functional as both activators and substrates, but of reduced affinity. In addition, the importance of the C-6 carbonyl is demonstrated in the preference of PDE-A for cXpXp as a substrate relative to the C-6 amine-bearing cApAp, which is neither a substrate nor effective inhibitor of this enzyme. The phosphoribosyl ring “backbone” also bears on the interaction of the molecule with its targets, as exemplified by the altered affinities af- forded by substitution at the sugar hydroxyl (cdGpdGp) and at the exocyclic oxygen (cGp(S)Gp:R,) of the phosphodiester bond. The vitalness of the dimeric structure is evident in the negligible affinity of both the lower homologue, cGMP, and the marginal affinity of the higher homologue, cGpGpGp, for either enzyme.

Given the 2-fold rotational symmetry of c-di-GMP, its protein binding sites might reflect a similar symmetrical disposition, as has been suggested from studies of cyclic dinucleotide inhibition of bacterial RNA polymerase (25). Comparison of the relative affinities of cdGpGp uersw cd- GpdGp, cIpGp uersus cIpIp, and cXpGp uersus cXpXp, re- veals that a given substitution is always more deleterious to the active structure when carried out in both GMP residues, suggesting that both participate in the recognition process. This might be taken as evidence that the cyclic dinucleotide binding occurs in either one of the two 180” rotational orien- tations and that both binding modes contribute to the ob- served potency of each analog. Such a situation would explain the apparent selectivity of PDE-A, which produces two prod- ucts in varying yield, when presented with a heterodinucleo- tide (cdGpGp, cXpGp, cIpGp). The occurrence of alternate binding modes might explain the partial activity of the phos- phothioate analog cGp(S)Gp:R, toward cellulose synthase; maximal activation of the enzyme may be prohibited even at apparent saturating concentrations of the activator due to competition between an “active” and an “inactive” binding orientation. Although this argument may also account for the potent effect on cellulose synthase activity of cXpGp relative

to cXpXp, it might only be valid when the substitution is a purine, since cCpGp is of negligible affinity.

The foregoing analysis does not take into account the effect that substitutions can have on the overall conformation of the molecule. Cyclic dinucleotides can adopt a variety of asymmetrical conformations in solution (26) and crystalline state (27). The expectation that protein-nucleotide interac- tion might involve specific conformational properties which are not evident in the structural formula of the analog limits the conclusions regarding the relationships of structure to function and of alternate binding orientations.

The use of membrane preparations as a source of both synthase and PDE-A activities has facilitated the comparison of their c-di-GMP-binding sites, which would appear to be closely related, since under these conditions their respective apparent affinities (& = 0.31 pM; apparent K, = 0.25 pM)

are similar in value. However, although each of these enzymes are highly specific toward the c-di-GMP structure, PDE-A appears to be of wider specificity than cellulose synthase. Their overlapping specificity is evident when considering that all of the cyclic dinucleotide activators of cellulose synthase serve as substrates in the PDE-A reaction, and a number of other cyclic dinucleotides (i.e. cApAp, cCpCp, and cUpUp) show low or no affinity for either enzyme. However, at least one compound, cXpXp, which lacks affinity for the synthase, is readily hydrolyzed by PDE-A activity. As noted, highly purified cellulose synthase (13) is essentially devoid of PDE- A activity,” confirming that each is a distinct enzyme.

The PDE-A reaction is subject to inhibition by a variety of cyclic and linear guanyl nucleotides (cCpGp, pGpG, GpG), which are demonstrably lacking in affinity for the synthase. Thus, either the c-di-GMP-binding site of the phosphodies- terase accommodates structures which are not recognized at all by the regulatory site of the synthase, as is apparently the case for cXpXp, or the observed inhibition is not of a com- petitive nature. In this connection, it is interesting to consider the inhibitory effects on PDE-A activity of GTP andpppGpG, the phosphorylated structures of which sufficiently digress from that of the substrates and other inhibitors to suggest the involvment of an additional, possibly regulatory, site on the enzyme which responds to these compounds, particularly when considering that other, more closely related, guanosine derivatives such as cGMP, GDP, and GMP show little affin- ity. The physiological basis for this inhibition may lie in the fact that both GTP and pppGpG are substrates in the acti- vator-forming diguanylate cyclase reaction. A more detailed analysis of the nature of guanyl nucleotide inhibition would facilitate the comparison of the overall degree of relatedness between the c-di-GMP-binding sites.

Throughout this discussion it has been assumed that the degradation of cyclic dinucleotides in the presence of mem- brane preparations is due to the action of a single highly specific enzyme, that is, PDE-A. This assumption is supported by the findings that all of the cyclic dinucleotide degradation reactions occur at similar rates and, without exception, all of the apparent substrates are potent inhibitors of PDE-A activ- ity, as assayed employing c-di-GMP as substrate. Further- more, all of the cyclic dinucleotide degradation reactions are inhibited at low CaCl, concentrations.

The inhibitory effect of Ca’+ on the rate of hydrolysis of c- di-GMP appears to result from a reduction in the turnover number (V,,,) of PDE-A. If the mechanism of inhibition were allosteric in nature, as via a specific Ca’+-binding site which inactivates the enzyme when occupied, then the extent of the inhibition would be expected to be independent of substrate. However, the inhibitory effect of Ca’+ varies among the cyclic

by guest on Decem

ber 24, 2019http://w

ww

.jbc.org/D

ownloaded from

Cyclic Dinucleotides in the Regulation of Cellulose Synthesis

dinucleotide substrates by up to a factor of 4-fold, suggesting that this divalent cation impedes catalysis through direct interaction with the substrate, possibly by competition with M$+ ions, which are required for PDE-A activity (9, 11).

7. Aloni, Y., and Benziman, M. (1982) in Ce&bse and Other Natural Polymer Swtems: Biogenesis, Strwzture and Dewadation (R. M.“Brown,“Jr., ed) pp. !l41-361, Plenum Publishing Corp., New York

8. Ross, P., Aloni, Y., Weinhouse, H., Michaeli, D., Weinberger- Ohana, P., Mayer, R., and Benziman, M. (1985) FEBS Mt. 186,191-196

The structure-function analysis described here should be useful in the synthesis of additional analogs and new molec- ular tools with which to study the various components of the c-di-GMP regulatory system. For example, the finding that the phosphothioate linkage is relatively resistant toward PDE-A activity raises the prospect of obtaining a c-di-GMP analog which is a highly potent activator of cellulose synthase, but which is stable to enzymatic degradation. One structure which promises to fulfill these requirements is a diphospho- thioate analog of c-di-GMP in which both phosphodiester bonds contain a substituted exocyclic oxygen in the S-config- uration. This study has also led to the identification of func- tional groups which are relatively amenable to chemical mod- ifications with regard to the preservation of biological activity, such as the exocyclic phosphodiester oxygen, the C-2 amino, and the ribosyl hydroxyl; this information should facilitate the design of affinity labels and columns, based on the c-di- GMP structure.

Six of the biologically active cyclic dinucleotides contain naturally occurring moieties, namely deoxyribose, inosine, and xanthine. When supplied with the corresponding nucleo- tide triphosphates, i.e. dGTP and ITP, crude diguanylate cyclase-containing cell extracts did not produce cellulose syn- thase activators in detectable yield (28). In preliminary ex- periments,4 employing highly purified preparations of digua- nylate cyclase and an assay based on reverse phase HPLC analysis, the low yield of products formed from dGTP or ITP in either the absence or co-presense of GTP is compatible with the notion that GTP is the preferred substrate of the enzyme. A more rigorous examination of the diguanylate cyclase reaction is currently underway, since the specificity of this enzyme is critical in determining the cyclic dinucleotide composition of the cell.

REFERENCES

1. Haigler, C. H. (1985) in Cellulose C!zern&ry and Its App~icutions, pp. 30-83, Ellis Horwood Ltd., Chichester, United Kingdom

2. Valla, S., and Kjosbakken, J. (1982) J. Gen. Micro&ol. 128, 1461-I408

3. Brown, R. M., Jr. (1973) in S&ucture cm& &o&em&ry o/Nu&rui Biological Systems (Walk, C. M., ed) pp. 51-123, Philip Morris, New York

4. Benziman, M., Haigler, C. H., Brown, R. M., Jr., White, A. R., and Coouer, K. M. (19801 Proc. N&l. Acad. Sci. U. S. A. 77. 6678-6682

5. Aloni, Y., Delmer, D. P., and Benziman, M. (1982) proc. N&l. Acad. Sci. U. S. A. 79,6448-6452

6. Ross, P., Mayer, R., and Benziman, M. (1990) in Biosyrzthesis and Biodegradation of Cellulose and Cellulose h4aterials (Haig- ler, C., ed) pp. 219-243, Marcel Dekker, Inc. New York

’ P. Ross and M. Benziman, manuscript in preparation.

9.

10.

11.

12.

13.

14.

15.

16.

17.

18.

19.

20.

21.

22. 23.

24.

25.

26.

27.

28.

29.

30.

31.

32.

Ross, P., Aloni, Y., Weinhouse, H., Michaeli, D., Weinberger- Ohana, P., Mayer, R., and Benziman, M. (1986) Carbohydr. Res. 149, 101-117

Ross, P., Weinhouse, H,, Aloni, Y., Michaeii, D., Weinberger- Ohana. P.. Mayer. R.. Braun. S.. de Vroom. E.. van der Marel. G. A., ban’Boo&, b. II., and Be&man, M. (1987) Nczture 325; 279-281

Mayer, R. (1987) Characterization of thy Phosphodiesterases Act- ing on c-di-GMP’ in Acetobucter xylinum. MSc. thesis, The Hebrew University of Jerusalem, Jerusalem, Israel

Mayer, R., Ross, P., Weinhouse, H., Amikam, D., Volman, G., Ohana, P., and Benziman, M. (1989) Fifth CeH Wall Meeting (Brett, E. T., and Reid, J. S. G., eds) Edinburgh, United Kingdom

Aloni, Y., Cohen, R., Benziman, M., and Delmer, D. (1983) J. Biol. Ckm. 258,4419-4423

van der Marel, G. A., Van Boeckel, C. A. A., Wille, G., and van Boom, J. H. (1981) Tetrahedron L&t. 22,3887-3890

de Vroom, E., Fidder, A., Marugg, J. E., van der Marel, G. A., and van Boom, J. H. (1986) Nucleic Acid Res. 14,5885-5900

de Vroom, E., van der Marel, G. A., and van Boom, J. H. (1987) Red. Frau. Chim. Pays-Bas 106,577-580

de Vroom, E., Fidder, A., Saris, C. P., van der Marel, G. A., and van Boom, J. H. (1987) Nucleic Acid Res. 15,9933-9943

Hayakawa, Y., Uchiyama, M., Kato, H., and Noyori, R. (1985) Tetruhedron L&t. 26,6505-6508

Stawinsky, J., Hozumi, T., Narang, S. A., Bahl, C. P., and Wu, R. (1977) NucZeic Acid Res. 4,353-371

de Rooij, J. F. M., Willie-Hazeleger, G., van Deursen, P. H., Serdin. J.. and van Boom. J. H. (1979) Reel. Truu. Chim. Paw- Bas 9&i, 5,37-548 ‘

Hunt, B. J., and Rigby, W. (1967) Chem. Znd. (tindon) 186% 1869

Reese, C. B., and Zard, L. (1981) Nucleic Acids Res. 9,4611-4626 Griffin, B. E., Jarman, M., and Reese, C. B. (1968) Tetrahedron

24,639-662 Thelen, M. P., and D. P. Delmer (1986) Plant Physiol. 81, 913-

918 Hsu, C.-J., and Dennis, D. (1982) Nucleic Acids Res. 10, 5637-

5647 Blommers, M. J. J., Haasnoot, C. A. G., Walters, J. A. L. I., van

der Marel, G. A., van Boom, J. H., and Hilbers, C. W. (1988) Biochemistry 27,8361-8369

Frederick, C. A., Coil, M., van der Mare& G. A., van Boom, J. H., and Wang, A. H.-J. (13%) Biochemishy 2?, 8350-8361

Benziman, M., Aloni, Y., and Delmer, D. P. (1983) J. At&. Polym. Sci. 37,131-143

-.

van Boom. J. H.. and Wreesmann. C. T. J. (1984) in Okonuck- otide S&w&; A Practical Appioach (Gait, M. .J., ed) pp. 153- 183, IRL Press, Oxford, United Kingdom

Marugg, J. E., Tromp, M., Jhurani, P., Hoyng, C. F., van der Marel, G. A., and van Boom, J. H. (1984) !Z’etrahedron 40,73- 78

Baraniak, J., Kinas, R. W., Lesiak, K., and Stec, W. (1979) J. Chem. Sot. Chem. Commun. 940-941

Himmelsbach, F., Schulz, B. S., Trichtinger, T., Charubala, R., and Pfleiderer, W. (1984) Tetrahedron 40, 59-72

by guest on Decem

ber 24, 2019http://w

ww

.jbc.org/D

ownloaded from

Cyclic Dinuckotides in the Regulation of CelluQse Synthesis 18943

by guest on Decem

ber 24, 2019http://w

ww

.jbc.org/D

ownloaded from

Fidder, P de Paus and L A SliedregtP Ross, R Mayer, H Weinhouse, D Amikam, Y Huggirat, M Benziman, E de Vroom, A

trimer, and phosphothioate derivatives.xylinum. Chemical synthesis and biological activity of cyclic nucleotide dimer,

The cyclic diguanylic acid regulatory system of cellulose synthesis in Acetobacter

1990, 265:18933-18943.J. Biol. Chem. 

  http://www.jbc.org/content/265/31/18933Access the most updated version of this article at

 Alerts:

  When a correction for this article is posted• 

When this article is cited• 

to choose from all of JBC's e-mail alertsClick here

  http://www.jbc.org/content/265/31/18933.full.html#ref-list-1

This article cites 0 references, 0 of which can be accessed free at

by guest on Decem

ber 24, 2019http://w

ww

.jbc.org/D

ownloaded from