Translocation of rhoA associated with Ca2+ sensitization of smooth muscle.

We determined the relationship between the localization of rhoA and Ca2+ sensitization of force in smooth muscle. In alpha-toxin-permeabilized rabbit portal vein at pCa 6.5, the particulate hydrophobic fraction of rhoA (10 +/- 1.6% of the total) was significantly increased by phenylephrine to 18 +/- 5.5% at 5 min, by AlF4- to 26 +/- 8.4% at 20 min, and dose-dependently up to 62 +/- 9.5% by guanosine 5'-O-(3-thiotriphosphate) (GTPgammaS; 0.3-50 microM). Translocation of rhoA was selective (Rac1 and Cdc42 were not translocated) and was quantitatively correlated (up to approximately 50%; r = 0.91, p < 0.05) with Ca2+ sensitization; high GTPgammaS concentrations (>/=10 microM) further increased translocation without increasing force. The initial recruitment of rhoA to the membrane paralleled the time course of contraction, but sensitization could be reversed without a decrease in particulate rhoA. High [Ca2+] (pCa 4.5) also increased particulate rhoA to 31 +/- 5.8%. Membrane-associated rhoA in unstimulated portal vein was a good substrate for in vitro ADP-ribosylation, whereas the large amount translocated by GTPgammaS was not. We conclude that 1) translocation of rhoA plays a causal role in Ca2+ sensitization, and 2) membrane-bound rhoA can exist in two or more states.

We determined the relationship between the localization of rhoA and Ca 2؉ sensitization of force in smooth muscle. In ␣-toxin-permeabilized rabbit portal vein at pCa 6.5, the particulate hydrophobic fraction of rhoA (10 ؎ 1.6% of the total) was significantly increased by phenylephrine to 18 ؎ 5.5% at 5 min, by AlF 4 ؊ to 26 ؎ 8.4% at 20 min, and dose-dependently up to 62 ؎ 9.5% by guanosine 5-O-(3-thiotriphosphate) (GTP␥S; 0.3-50 M). Translocation of rhoA was selective (Rac1 and Cdc42 were not translocated) and was quantitatively correlated (up to ϳ50%; r ‫؍‬ 0.91, p < 0.05) with Ca 2؉ sensitization; high GTP␥S concentrations ( 10 M) further increased translocation without increasing force. The initial recruitment of rhoA to the membrane paralleled the time course of contraction, but sensitization could be reversed without a decrease in particulate rhoA. High [Ca 2؉ ] (pCa 4.5) also increased particulate rhoA to 31 ؎ 5.8%. Membrane-associated rhoA in unstimulated portal vein was a good substrate for in vitro ADP-ribosylation, whereas the large amount translocated by GTP␥S was not. We conclude that 1) translocation of rhoA plays a causal role in Ca 2؉ sensitization, and 2) membrane-bound rhoA can exist in two or more states.
Phosphorylation of the regulatory light chain of myosin (MLC 20 ) 1 by a calcium/calmodulin-dependent protein kinase is the primary determinant of force developed by smooth muscle. However, this phosphorylation can also be increased ("Ca 2ϩ sensitization") at constant [Ca 2ϩ ] by a G-protein-coupled mechanism (1)(2)(3)(4) that inhibits the trimeric phosphatase (SMPP-1M) (5-7) that dephosphorylates MLC 20 . The Ca 2ϩ -sensitizing effect of recombinant p21 rhoA and the inhibition of agonist-induced Ca 2ϩ sensitization by selective ADP-ribosylation of p21 rhoA (either recombinant or endogenous) have implicated this monomeric G-protein in Ca 2ϩ sensitization (8 -11). Because bacterially expressed p21 rhoA that lacks the prenylated C terminus required for membrane association (12) did not show significant Ca 2ϩ -sensitizing activity and even active (geranylgeranylated) p21 rhoA failed to Ca 2ϩ -sensitize preparations heavily permeabilized with Triton X-100 (11), we suggested that association of p21 rhoA with the plasma membrane may be required for its Ca 2ϩ -sensitizing effect (11). Recruitment of cytosolic proteins to the membrane is an important component of several other signaling systems, such as the Raf-Ras pathway (13) and protein kinase C cascades (conventional and novel) (14). The purpose of this study was to determine whether p21 rhoA signaling of Ca 2ϩ sensitization also involves its translocation to the cell membrane in vivo. We now show that GTP␥S-induced translocation of p21 rhoA is quantitatively and kinetically associated with Ca 2ϩ sensitization of smooth muscle and provide evidence of more than one conformational state of membrane-associated p21 rhoA .

MATERIALS AND METHODS
Isometric Tension Measurement-Small strips (200 m wide and 3 mm long) of rabbit portal vein and ileum longitudinal smooth muscle were dissected, and isometric tension was measured as published (15)(16)(17).
Separation of Particulate and Cytosolic Fractions-A minimum of 10 small (200 m wide and 3 mm long) strips of rabbit portal vein or ileum longitudinal smooth muscle were used to provide sufficient protein for reliable separation of cytosolic and particulate fractions. Stimulated and control strips were homogenized in ice-cold homogenization buffer (10 mM Tris-HCl, pH 7.5, 5 mM MgCl 2 , 2 mM EDTA, 250 mM sucrose, 1 mM dithiothreitol, 1 mM 4-(2-aminoethyl)bezenesulfonyl fluoride, 20 g/ml leupeptin, and 20 g/ml aprotinin) and centrifuged at 100,000 ϫ g for 30 min at 4°C (Optima™ TLX ultracentrifuge, TLA 120.1 rotor, Beckman Instruments), and the supernatant was collected as the cytosolic fraction. Pellets were resuspended, and membrane proteins were extracted by incubation for 30 min in homogenization buffer containing 1% Triton X-100 and 1% sodium cholate or only 2% Triton X-114. The latter buffer was used to avoid the increase in the cloudy point of Triton X-114 by a second detergent. The extract was centrifuged at 800 ϫ g for 10 min. The supernatant was collected and is referred to as the particulate fraction, and the pellet was collected and is referred to as the detergent-insoluble particulate fraction. Cytosolic, particulate, and detergent-insoluble particulate fraction proteins were separated by SDSpolyacrylamide gel electrophoresis. Only the cytosolic and particulate p21 rhoA proteins are shown in most of the figures, as no immunoblotdetectable p21 rhoA was found in the detergent-insoluble particulate fraction. The absence of p21 rhoA in the detergent-insoluble particulate fraction verified the completion of the extraction of membrane p21 rhoA proteins and completion of homogenization. Prompt termination of the reaction in homogenization buffer was verified by the absence of translocation of p21 rhoA when control strips were homogenized in GTP␥S (50 M)-containing homogenization buffer.
Phase Separation by Triton X-114 -Precondensed Triton X-114 stock solution was added to tissue homogenates or cytosolic fractions to a final concentration of ϳ2%, and proteins were extracted by incubation for 30 min on ice with occasional mixing (18). The mixture was centrifuged at 10,000 ϫ g for 10 min at 4°C; the pellet was solubilized in sample buffer; and proteins were separated by SDS-polyacrylamide gel electrophoresis to determine cellular proteins insoluble in nonionic detergent. The supernatant was collected in a fresh tube and warmed to 37°C in a water bath until the solution became cloudy (for ϳ5 min). Phase separation was achieved by centrifuging the solution in a tabletop centrifuge for 10 min at 800 ϫ g at room temperature. The upper aqueous phase contains soluble proteins, and the lower, detergentenriched phase contains proteins bearing hydrophobic domains.
Western Blots-After transfer to polyvinylidene difluoride membrane, the membranes were blocked with 5% nonfat dry milk in phosphate-buffered saline containing 0.05% Tween 20 for 1 h and then incubated with primary antibody for 3 h and secondary antibody for 1 h at room temperature. Blots were detected with enhanced chemiluminescence (ECL, Amersham Corp.) and quantitated by densitometry using a Bio-Rad GS-670 imaging densitometer. Optimal primary antibody concentration was determined by antibody titration (1:100, 1:500, 1:1000, and 1:5000) using a Mini-protein II multiscreen apparatus (Bio-Rad). Preliminary experiments established that the amount of protein loaded was within the range of linearity of the assays. The percent of particulate p21 rhoA was calculated according to particulate p21 rhoA /(particulate ϩ cytosolic p21 rhoA ).
ADP-ribosylation of Cytosolic and Particulate p21 rhoA -The detergent concentration and the volumes of cytosolic and particulate fractions were adjusted to identical values, and the following reagents were added: 200 M GTP, 10 mM dithiothreitol, 2 mM thymidine, and 1 g/ml Clostridium botulinum exoenzyme C3. After initiation of ADP-ribosylation by addition of [ 32 P]NAD (final concentration of 50 Ci/ml), the mixture (total volume of 100 l) was incubated for 30 min at 30°C. The reaction was stopped by trichloroacetic acid (24%, 250 l) and deoxycholate (2%, 6 l), and the final volume was adjusted to 1 ml with water. After centrifugation (5000 ϫ g, 10 min), the supernatant was carefully removed, and the pellet was resuspended in 20 l of 2 ϫ sample buffer. 10 l of 1 M Tris base was added to neutralize the pH. Samples were heated at 85°C for 5 min, and the proteins were separated by SDSpolyacrylamide gel electrophoresis. ADP-ribosylation of p21 rhoA in ␤-escin-permeabilized strips was carried out as described previously (11).
As shown in Fig. 1A, the translocation of p21 rhoA , up to 50%, was quantitatively correlated (r ϭ 0.91, p Ͻ 0.05) with Ca 2ϩ sensitization of force induced by agonist (PE plus GTP), AlF 4 Ϫ , and various concentrations of GTP␥S. Higher concentrations (Ն10 M) of GTP␥S caused further translocation of p21 rhoA without further increase in force, indicating a "ceiling effect." To ascertain whether the observed translocation is ␣-adrenergic receptor-specific, we also determined the effect of a muscarinic agonist on the localization of p21 rhoA in permeabilized ileum smooth muscles. Surprisingly, a high percentage (61 Ϯ 6.8%, n ϭ 14) of p21 rhoA was located in the particulate fraction of unstimulated ileum. This high basal level of particulate p21 rhoA was not due to Ca 2ϩ (submaximal, pCa 6.5) because 59 Ϯ 1.4% (n ϭ 2) of p21 rhoA was particulate even at cytoplasmic [Ca 2ϩ ] Ͻ pCa 8 (no free Ca 2ϩ added, 10 mM EGTA present). In contrast, inclusion of the muscarinic antagonist atropine (10 M) during and following dissection decreased particulate p21 rhoA to 28 Ϯ 11.6% (n ϭ 6, p Ͻ 0.05). This significant decrease in particulate p21 rhoA by a highly specific muscarinic antagonist indicates that acetylcholine released from nerve endings in the richly innervated ileum causes translocation of p21 rhoA to the particulate fraction and that such translocation is not limited to the action of ␣-adrenergic agonists.
Selective Translocation of p21 rhoA , but Not Rac1 and Cdc42, by GTP␥S-To determine the specificity of translocation of p21 rhoA by GTP␥S, we also determined the distribution of two other Rho family proteins, Rac1 and Cdc42, under conditions identical to those used for determining the partitioning of p21 rhoA . GTP␥S (50 M, 20 min) (Fig. 2) had no significant effect on the amount of either Rac1 or Cdc42 present in the particulate fraction: 46 Ϯ 2.9% (n ϭ 5) of Rac1 and 14 Ϯ 4.1% (n ϭ 5) of Cdc42 in pCa 6.5 solution and 47 Ϯ 4.2% (n ϭ 6, p Ͼ 0.005) of Rac1 and 19 Ϯ 4.8% (n ϭ 6, p Ͼ 0.05) of Cdc42 after stimulation with GTP␥S. Unlike p21 rhoA and Cdc42, which were not detected in the detergent-insoluble particulate fraction, 23 Ϯ 7.0% (n ϭ 5) of Rac1 was in the detergent-insoluble particulate fraction, and this fraction was also not changed by GTP␥S (25 Ϯ 5.7%, n ϭ 6, p Ͼ 0.05).
Relaxation of Phenylephrine-induced Ca 2ϩ Sensitization Is Unaccompanied by the Return of p21 rhoA from the Particulate to the Cytosolic Fraction-To determine whether the return of p21 rhoA from the particulate to the cytosolic fraction is required for the reversal of Ca 2ϩ sensitization, the distribution of p21 rhoA after "washout" of the agonist was determined. After increasing the steady-state pCa 6.5-induced contraction by PE plus GTP from 7 Ϯ 2.2% (n ϭ 3) to 32 Ϯ 1.2% (n ϭ 3, p Ͻ 0.001), the muscles were transferred into relaxing solution (no added Ca 2ϩ , PE, or GTP and containing 1 mM EGTA) and exchanged three times for a total of 25 min. At this time, the pCa 6.5induced contraction was not significantly different from before exposure to PE plus GTP (9 Ϯ 0.6%, n ϭ 3), indicating that the muscles were no longer Ca 2ϩ -sensitized, but 21 Ϯ 4.6% (n ϭ 9) of p21 rhoA still remained in the particulate fraction; this was not significantly different from that found in the presence of PE (18 Ϯ 5.5%, n ϭ 9, 5 min, p Ͼ 0.05). Even when strips were washed in 10 mM EGTA-containing solution for 60 min, the translocated p21 rhoA remained in the particulate fraction (data not shown).
Time Courses of GTP␥S-induced Contraction and Translocation of p21 rhoA and G ␣q/11 -The time courses of GTP␥S-induced potentiation of force and translocation of p21 rhoA were determined to evaluate whether they were kinetically consistent with the potential role of p21 rhoA as a mediator of agonistinduced Ca 2ϩ sensitization. As shown in Fig. 3 (A and B), within 1 min following addition of GTP␥S to permeabilized portal vein smooth muscle at pCa 6.5, force reached 21 Ϯ 4.2% (n ϭ 10) of the maximal GTP␥S-induced contraction; this was accompanied by translocation of p21 rhoA to the particulate fraction, increasing from the control value of 10 Ϯ 1.6% (n ϭ 23) to 32 Ϯ 9.7% (n ϭ 6, p Ͻ 0.0001). Thus, within the time resolution of this study, the kinetics of translocation of p21 rhoA were consistent with its role in GTP␥S-induced Ca 2ϩ sensitization of force. However, the later time course of GTP␥S-induced p21 rhoA translocation was slower than that of force development: contraction peaked at 5 min, at which time ϳ51 Ϯ 4% (n ϭ 6) of p21 rhoA was in the particulate fraction, whereas p21 rhoA continued to translocate, reaching its peak of 62 Ϯ 9.5% (n ϭ 4) at 20 min, consistent with the ceiling effect in the translocationforce relationship (Fig. 1A).
We also determined the time course of translocation of G ␣q/11 , the heterotrimeric G-protein implicated in the activation of phospholipase C, a major contributor to pharmacomechanical coupling in smooth muscle (reviewed in Ref. 4). Under control conditions (pCa 6.5), 86 Ϯ 1.7% (n ϭ 25) of the total G ␣q/11 was in the particulate fraction, and this was reduced by GTP␥S to 60 Ϯ 12% (n ϭ 6, p Ͻ 0.01) at 1 min and 70 Ϯ 8.8% (n ϭ 6, p Ͻ 0.05) at 5 min (Fig. 3, A and B). In contrast to p21 rhoA , the translocation of G ␣q/11 was transient: by 60 min, the previously translocated protein had returned to the particulate fraction (Fig. 3, A and B).
Particulate p21 rhoA Is Hydrophobic-The hydrophobic domain of cytosolic, geranylgeranylated p21 rhoA is masked by bound Rho-GDI, and activated p21 rhoA is thought to bind to the cell membrane through the unmasked hydrophobic geranylgeranyl group exposed by the release of Rho-GDI (22). Because the particulate fraction obtained through centrifugation may contain both hydrophobic (membrane) and nonhydrophobic (e.g. cytoskeletal) components, we determined by phase separation with Triton X-114 whether the p21 rhoA translocated to the particulate fraction by GTP␥S in vivo was hydrophobic, as indicated by partitioning into Triton X-114. Indeed, GTP␥S (50 M, 30 min) increased the fraction of p21 rhoA partitioned into the detergent phase when whole homogenates were treated with Triton X-114. p21 rhoA in the Triton X-114 phase increased from the control value of 22 Ϯ 6.6% (n ϭ 11) to 81 Ϯ 1.8% (n ϭ 6, p Ͻ 0.0001), indicating that most of the particulate p21 rhoA was associated with hydrophobic (presumably membrane) components.
To further evaluate whether cytosolic p21 rhoA (complexed with Rho-GDI) is hydrophilic, whereas particulate p21 rhoA is hydrophobic, the whole homogenate was first separated into cytosolic and particulate fractions, which were subsequently phase-separated with Triton X-114 (see "Materials and Methods"). As shown in Fig. 4, only 5 Ϯ 1.8% (n ϭ 9) of the cytosolic p21 rhoA partitioned into the Triton X-114 phase. In contrast, most of the particulate p21 rhoA partitioned into the Triton X-114 phase, with only 7 Ϯ 2.5% (n ϭ 8) partitioning into the aqueous phase. Again, there was a dramatic increase from 11 Ϯ 8.6% (n ϭ 3) to 63 Ϯ 13.5% (n ϭ 3, p Ͻ 0.05) in the amount of p21 rhoA in the Triton X-114-treated particulate fraction of GTP␥S (50 M, 20 min)-stimulated muscles. The small quantities of cytosolic p21 rhoA partitioning into the Triton X-114 phase and particulate p21 rhoA partitioning into the aqueous phase may result from "carryover" during the experimental procedure.
Translocated p21 rhoA Is Not a Good Substrate for C3-catalyzed ADP-ribosylation-The cytosolic and particulate fractions of control and GTP␥S-stimulated tissues were incubated with C3 and [ 32 P]NAD (see "Materials and Methods") to determine if the translocated p21 rhoA is a good substrate for C3catalyzed ADP-ribosylation. In unstimulated strips, the largely cytosolic p21 rhoA (Figs. 1, 2, and 5) was only minimally ADPribosylated; a very faint band was detected in the autoradiograph (Fig. 5, upper left panel), whereas the very small amount of particulate p21 rhoA was highly ADP-ribosylated (lower left panel). This is consistent with previous results showing that cytosolic p21 rhoA is complexed with Rho-GDI and that the complex is a poor substrate for C3-catalyzed ADP-ribosylation (23,24), but becomes a better substrate after its dissociation from Rho-GDI. In contrast, the large amount of particulate p21 rhoA in GTP␥S-stimulated tissues (Figs. 1-3 and 5) was only minimally ADP-ribosylated (Fig. 5, right panels). This lower level of ADP-ribosylation of particulate p21 rhoA in GTP␥S-stimulated strips was not due to the loss of a membrane component to the cytosol because ADP-ribosylation was significantly less even in total tissue homogenates containing both cytosolic and particulate fractions of GTP␥S-treated tissue compared with controls (data not shown), and cytosolic p21 rhoA in GTP␥S-stimulated tissue was still not a good substrate for ADP-ribosylation (Fig.  5, right panels). This was also not the result of reassociation with Rho-GDI because the latter was not detectable in any of the particulate fractions. These results suggest that membrane-associated p21 rhoA exists in at least two states: a resting state that is a good substrate and an activated (and/or inactivated; see "Discussion") state that is a poor substrate for C3catalyzed ADP-ribosylation.
Effect of Ca 2ϩ and of the Phosphatase Inhibitor Tautomycin on the Localization of p21 rhoA -To determine whether Ca 2ϩ alone can induce translocation of p21 rhoA , ␣-toxin-permeabilized rabbit portal vein strips were incubated in Ca 2ϩ -free solution (no Ca 2ϩ added and containing 10 mM EGTA) for 15 min, homogenized, and fractionated. The particulate fraction under this Ca 2ϩ -free condition contained 9 Ϯ 3.0% (n ϭ 6) of the total p21 rhoA , which is not significantly different from strips incubated in pCa 6.5 solution: 10 Ϯ 1.6% (n ϭ 23, p Ͼ 0.05). However, increasing Ca 2ϩ to very high levels (pCa 4.5, 15 min) increased the p21 rhoA content of the particulate fraction to 31 Ϯ 5.8% (n ϭ 10, p Ͻ 0.01). This translocation of p21 rhoA induced by pCa 4.5 was not due to the release of norepinephrine from nerve endings because inclusion of the ␣-adrenergic blocker FIG. 4. Triton X-114 partitioning of p21 rhoA in the cytosolic and particulate fractions of control and GTP␥S-stimulated rabbit portal vein. GTP␥S (50 M, 20 min)-treated and control (in pCa 6.5 solution) portal vein strips were homogenized and fractionated by centrifugation (see "Materials and Methods"), after which the cytosolic and particulate fractions were further phase-separated with Triton X-114 and blotted with anti-p21 rhoA antibody. C, cytosol; P, particulate; Aq, aqueous phase; Tx, Triton X-114 phase.
FIG. 5. Reduced availability of particulate p21 rhoA for ADPribosylation in GTP␥S-stimulated, ␣-toxin-permeabilized rabbit portal vein. GTP␥S (50 M, 20 min)-treated and control (in pCa 6.5 solution) portal vein strips were homogenized and fractionated by centrifugation (see "Materials and Methods") with the volumes and concentration of the detergent in the two fractions adjusted to the same level. Samples were incubated with [ 32 P]NAD (50 Ci/ml), C3 (1 g/ml, 30 min, 30°C), and other reagents (see "Materials and Methods") to ADP-ribosylate p21 rhoA . The ADP-ribosylation level was determined by autoradiography (AutoRad). The particulate p21 rhoA in the control strips was a good substrate for C3-catalyzed ADP-ribosylation, whereas the particulate p21 rhoA in the GTP␥S-stimulated strips was a poor substrate, although markedly increased (see Western blots). Results are representative of three independent experiments. C, cytosol; P, particulate.
prazosin (10 M) in the solutions during permeabilization and thereafter had no effect on the pCa 4.5-induced translocation of p21 rhoA to the particulate fraction (39 Ϯ 6.9%, n ϭ 8, p Ͼ 0.05). To exclude the possibility that the translocation of p21 rhoA by pCa 4.5 was due to trapping of cytosolic p21 rhoA in the cytoskeletal components of the contracted tissue, cytosolic and particulate fractions of pCa 4.5-stimulated tissues were phase-separated by Triton X-114. Similar to the GTP␥S-stimulated tissues, 89 Ϯ 8.6% (n ϭ 3) of p21 rhoA in the particulate fraction (29.0 Ϯ 13.9% (n ϭ 3) of the total p21 rhoA ) was partitioned into the Triton X-114 phase, suggesting that high [Ca 2ϩ ] alone can translocate p21 rhoA to the membrane.
To determine whether p21 rhoA translocated by high [Ca 2ϩ ] returns to the cytosol after removal of Ca 2ϩ , portal vein strips stimulated with pCa 4.5 for 15 min were washed five times in Ca 2ϩ -free solution (no Ca 2ϩ added and containing 10 mM EGTA) for a total of 60 min, and the distribution of p21 rhoA was determined. Even after this extensive wash, the same amount (38 Ϯ 2%, n ϭ 3) of translocated p21 rhoA remained in the particulate fraction.
Tautomycin, a potent inhibitor of protein phosphatases 1 and 2A, (25) causes substantial MLC 20 phosphorylation and smooth muscle contraction even in the absence of Ca 2ϩ (26) by inhibiting the catalytic subunit of SMPP-1M. We also wished to localize p21 rhoA in tautomycin-stimulated tissues to determine whether phosphorylation of a protein that can be dephosphorylated by phosphatase 1 or 2A is involved in regulating translocation of p21 rhoA . No significant translocation of p21 rhoA was detected in tautomycin-stimulated tissue at pCa 6.5 (data not shown), although it caused 86 Ϯ 6.3% (n ϭ 3) of the maximal Ca 2ϩ -induced contraction. DISCUSSION The relationships between the extent and time course of translocation of p21 rhoA to the particulate fraction (Figs. 1 and 3) and enhancement of force at constant [Ca 2ϩ ] are consistent with a causal role of p21 rhoA recruitment to the membrane in Ca 2ϩ sensitization. This conclusion is also supported by the hydrophobicity of particulate p21 rhoA (partitioning into Triton X-114; this study), the abolition of the Ca 2ϩ -sensitizing effect of recombinant p21 rhoA by extensive permeabilization of smooth muscle with detergent, and the inactivity of nonprenylated p21 rhoA (11). Similarly, GTP␥S-induced activation of NADPH oxidase in neutrophils involves the dissociation of p21 rho from Rho-GDI and its translocation to the membrane by a mechanism that requires a heat-and trypsin-labile membrane component (27). The translocation was limited to p21 rhoA in our study: the localization of two other Rho family proteins, Rac1 and Cdc42, was not affected by GTP␥S under conditions causing extensive translocation of p21 rhoA .
The ␣-adrenergic agonist phenylephrine, GTP␥S, and AlF 4 Ϫ induced translocation of p21 rhoA , whereas the muscarinic antagonist atropine inhibited the translocating effect of endogenous acetylcholine. The significant reduction by atropine of the large amount of particulate p21 rhoA in "unstimulated" ileum smooth muscle and the translocation of p21 rhoA by high [Ca 2ϩ ] indicate that preparatory conditions alone can affect its localization. Thus, caution is required in interpreting results of fractionation obtained in the presence of high [Ca 2ϩ ] and/or locally stored and released transmitters.
Because agonists acting on heptameric serpentine receptors coupled to trimeric G-proteins induced Ca 2ϩ sensitization and because AlF 4 Ϫ was not reported to interact with Ras family proteins (21), we had previously thought that the Ca 2ϩ sensitization by AlF 4 Ϫ was mediated by a trimeric G-protein (4). Subsequently, we (11) and others (9) found that ADP-ribosylation of p21 rhoA inhibited AlF 4 Ϫ -induced Ca 2ϩ sensitization, and we now show (Fig. 1) that AlF 4 Ϫ , like agonists and GTP␥S, also translocates p21 rhoA to the membrane. Furthermore, it has now been shown that although AlF 4 Ϫ cannot interact with another monomeric G-protein (elongation factor G) in solution, such an interaction can occur when elongation factor G is associated with ribosomes (28). On the other hand, other reports show p21 rhoA to be a downstream effector of G ␣12 -and G ␣13 -induced stress fiber formation (29) and of G ␣q -initiated hypertrophic signaling to cardiomyocytes (30). Neither these G ␣ subunits nor agonist-induced Ca 2ϩ sensitization (11) is sensitive to pertussis toxin, suggesting that they could act as upstream initiators (activated by AlF 4 Ϫ ) of a p21 rhoA -mediated cascade. At present, we consider it equally likely that a complex of p21 rhoA with a cytosolic protein can directly interact with AlF 4 Ϫ . A larger fraction of cytoplasmic p21 rhoA (Ͼ60% of the total) could be translocated by GTP␥S (50 M) to the membrane than required for maximal Ca 2ϩ sensitization of force. This ceiling effect presumably reflects the fact that the extent (ϳ50%) (2) of inhibition of SMPP-1M attainable by G-protein-coupled mechanisms is achieved by activating only ϳ50% of endogenous p21 rhoA (Fig. 1A). The p21 rhoA translocated to smooth muscle membranes did not return to the cytosol during 60 min of incubation without the agonist, although Ca 2ϩ sensitization of force by this time was already reversed. This is consistent with the finding that activated p21 rhoA associated with the particulate fraction is transformed, with time, into an inactivated membrane-bound form, and/or its downstream effector(s) is down-regulated (31). According to a recent report, rho translocation in cultured fibroblasts is rapidly reversible (32). The difference between these and our findings may be tissue-dependent or indicate an inhibitory effect of permeabilization (this study) on the reversibility of translocation.
The large cytosolic fraction of p21 rhoA present in unstimulated smooth muscle was not ADP-ribosylated, whereas ADPribosylation (Fig. 5) of the small amount of membrane-associated p21 rhoA was extensive and sufficient to inhibit the Ca 2ϩsensitizing effect of the ␣-adrenergic agonist PE and of GTP␥S. In other types of smooth muscle, ADP-ribosylation of endogenous p21 rhoA only slowed GTP␥S-induced Ca 2ϩ sensitization without reducing its amplitude (rabbit mesenteric artery). It also inhibited Ca 2ϩ sensitization by carbachol (ileum smooth muscle) (9,11) and the increase in MLC 20 phosphorylation induced by GTP␥S in cultured smooth muscle cells (33).
The lack of correlation between p21 rhoA detected by Western blotting and ADP-ribosylation, respectively, in several tissues (Ref. 24 and this study) raises the possibility that the variable and often incomplete (agonist-and tissue-dependent) inhibition of Ca 2ϩ sensitization by in situ ADP-ribosylation may reflect slow recruitment of p21 rhoA from a cytosolic pool protected from ADP-ribosylation by Rho-GDI. Alternatively, other p21 rhoA -independent pathways of Ca 2ϩ sensitization may account for incomplete inhibition of Ca 2ϩ sensitization following ADP-ribosylation of endogenous p21 rhoA .
The variable availability of particulate p21 rhoA for ADPribosylation by C3 (Fig. 5) suggests that membrane-associated p21 rhoA can exist in more than one (conformational) state. One state is membrane-associated under resting conditions, available for ADP-ribosylation and presumably ready, "with the trigger cocked," to be activated by an agonist, GTP␥S, or AlF 4 Ϫ .
This would account for the more rapid rate of Ca 2ϩ sensitization achieved by agonists or GTP␥S than by exogenous p21 rhoA (11). Another state (or states) is much less available for ADPribosylation, as indicated by the large quantity of p21 rhoA that was translocated by GTP␥S to the membrane, but not ADPribosylated (Fig. 5). This may represent a combination of transitional states being activated through membrane association, as well as a third membrane-bound but inactivated state of p21 rhoA (31). However, we cannot completely rule out the possibility that a very small fraction of activated p21 rhoA that was not translocated is involved in Ca 2ϩ sensitization. Tautomycin, a potent inhibitor of protein phosphatases 1 and 2, did not translocate p21 rhoA , although it had a large "Ca 2ϩsensitizing" effect on contraction (see also Ref. 26). Therefore, it is unlikely that phosphorylation of a site susceptible to dephosphorylation by these phosphatases plays a role in the translocation of p21 rhoA . The possibility of a role of tyrosine phosphorylation in this process is currently under study.
Ca 2ϩ -sensitizing mechanisms, whether mediated by protein kinase C (34,35,41) or other effectors, appear to converge to inhibit dephosphorylation of a cytosolic SMPP-1M substrate, MLC 20 (2). Smooth muscle myosin light chain phosphatase is not known to be associated with the plasma membrane. Therefore, p21 rhoA -coupled Ca 2ϩ sensitization is thought to involve the participation of additional downstream messengers/mechanisms, such as inhibitory phosphorylation of the regulatory subunit of SMPP-1M by a p21 rhoA -regulated and/or SMPP-1Massociated kinase (36 -40) or an atypical protein kinase C (or a related kinase) (41), and/or activation of a phosphatase inhibitor (1). SMPP-1M can also be inhibited directly by arachidonic acid (42). It is likely that multiple mechanisms, including some mediated by p21 rhoA , are involved in Ca 2ϩ sensitization.