Unfair competition governs the interaction of pCPI-17 with myosin phosphatase (PP1-MYPT1)

The small phosphoprotein pCPI-17 inhibits myosin light-chain phosphatase (MLCP). Current models postulate that during muscle relaxation, phosphatases other than MLCP dephosphorylate and inactivate pCPI-17 to restore MLCP activity. We show here that such hypotheses are insufficient to account for the observed rapidity of pCPI-17 inactivation in mammalian smooth muscles. Instead, MLCP itself is the critical enzyme for pCPI-17 dephosphorylation. We call the mutual sequestration mechanism through which pCPI-17 and MLCP interact inhibition by unfair competition: MLCP protects pCPI-17 from other phosphatases, while pCPI-17 blocks other substrates from MLCP’s active site. MLCP dephosphorylates pCPI-17 at a slow rate that is, nonetheless, both sufficient and necessary to explain the speed of pCPI-17 dephosphorylation and the consequent MLCP activation during muscle relaxation. DOI: http://dx.doi.org/10.7554/eLife.24665.001

where is the initial dephosphorylation rate, is the inhibitor concentration, and we have approximated ≈ because = 0.1 nM ≪ , so little phosphosubstrate was bound in all cases. These equations were analytically solved to determine as a function of .
The best-fit value of the OA was determined by nonlinear regression of ( ) to the pMRLC and pMyBP dephosphorylation data. Analogous experiments to determine the calyculin A cly . The analyses gave OA = 20 ± 2 nM ( = 7) and 0.13 nM < cly < 0.22 nM ( = 4). ( cly could only be limited to a range because we only had a lower bound on pC-ERMAD .) The pCPI-17/MLCP was then determined by nonlinear regression of ( ) to the data for each inhibitor using the calculated . Combining the OA ( = 0.59 ± 0.06; = 8) and calyculin A ( ≤ 0.42 ± 0.04; = 3) estimates using Bayes' theorem gave = 0.48 ± 0.03 nM. This is an upper bound since the analysis assumes that both OA and calyculin A are purely competitive inhibitors. If inhibition is mixed, then the actual is lower.

Figure 5: In vitro time-course of MLCP inhibition by pCPI-17
The analysis was complicated by two factors: (1) the small amount of contaminating phosphatase activity that remained even in the presence of a thio(p)CPI-17 concentration (50 nM) that completely inhibited MLCP, and (2) the slow degradation of MLCP that was noticeable as a slow decrease in the rate of pC-ERMAD dephosphorylation in the absence of pCPI-17. We accounted for these effects by including a background pC-ERMAD dephosphorylation rate, , and an MLCP degradation rate, in the analysis. The time-dependent quasi-steady state equations in the presence of pCPI-17 are where pS denotes pC-ERMAD, = 6 M is the reaction volume, and 4 ( ) is the amount of phosphate released by time . Solving the equations with these values gave the theoretical prediction for the time-course in the presence of pCPI-17 (red line, Figure 5). These experiments were conducted with 890 nM 32 P-labeled pCPI-17 and extract total protein concentrations of = 0.093 mg/ml (1:278 dilution) or = 0.047 mg/ml (1:556 dilution). Because these concentrations were so low, the MLCP concentration was < 1 nM in both cases, so protection of pCPI-17 by MLCP was negligible and essentially all dephosphorylation was by other, more efficient, phosphatases (i.e., PPU). It is evident that most of the phosphatase activity was inhibited by OA with subnanomolar , suggesting that this came from PP2A-, PP4-, and/or PP6-containing enzymes (84,86). The inhibition curves were biphasic, thereby revealing the presence of a second component contributing a small amount of pCPI-17-dephosphorylating activity that was much less sensitive to OA inhibition. Therefore, the data were

Experiments with mouse uterus extracts
where and denote the unbound PPU and PPU concentrations and is the dephosphorylation rate. Nonlinear regression of three experiments at each dilution was used to determine and the total specific activities of PPU ( ∕ ∕ ) and PPU ( ∕ ∕ ) in the extract (Tables 1 and 2). This analysis showed that PPU and PPU contribute ∼ 85% and ∼ 15% of the total dephosphorylating activity, respectively. was too small to be determined from the data, but the PPU IC 50 provided the upper bound ≤ IC 50 ∼ 0.43 nM.
The large value ∼ 450 nM precludes PPU being a PP2A-, PP4-, or PP6-type phosphatase. In addition, its activity is an order-of-magnitude greater than that predicted for MLCP. Therefore, we conclude that PPU is not MLCP. It might be a mixture of PP1 and other phosphatases even more resistant to OA leading to the large apparent inhibition constant.  To more closely mimic cellular protein concentrations, reactions were performed using highly concentrated uterus extract; the total protein concentration was = 17.5 mg/ml. (While this was still ∼ 10× lower than the estimated physiological concentration, it was adequate to test for sequestration of pCPI-17 by MLCP.) The pCPI-17 dephosphorylation reactions were performed at 0 • C to limit adequately the fraction of substrate that was dephosphorylated. Based on the physiological MLCP concentration [∼ 1 M (3, 82)], the MLCP concentration in the reactions was estimated to be ∼ 80 nM.

Determining [PPU ] from OA inhibition of unsequestered pCPI-17 dephosphorylation
The pCPI-17 concentration of 1.35 M (black data, Figure 8) was in great excess over MLCP; so, in this case, only a negligible fraction was sequestered and almost all dephosphorylation was due to PPU. Since PPU accounts for most of the PPU activity and < 0.5 nM (Figure 7), OA bound PPU almost stoichiometrically over the experimental range. Therefore, the IC 50 was close to half the total PPU concentration, . For best accuracy, was determined by nonlinear regression

Dephosphorylation of pCPI-17 at low concentration is catalyzed by MLCP
When the pCPI-17 concentration was 10 nM (purple data, Figure 8), MLCP was in significant excess and the unfair competition model predicted that it would bind essentially all pCPI-17. Consequently, because the (0.48 nM) is much smaller than the (20 nM) for OA-inhibition, the model predicted that, as observed,an OA concentration much greater than the would be required for inhibition. Moreover, the dephosphorylation rate in the absence of OA was ∼ 0.008 nM/sec, in excellent agreement with the predicted MLCP dephosphorylation rate of (0 • C) × [pCPI-17] ∼ 0.007 nM/sec.

Figure 8 -figure supplement 1: Identifying the phosphatase activities against pC-ERMAD in uterus extracts
A priori, we allowed for the possibility that pC-ERMAD can be dephosphorylated by MLCP and other PP1-containing enzymes, PPU , and PPU by modeling the pC-ERMAD dephosphorylation rate as where denotes the unbound pC-ERMAD concentration, ≡ ∕ denotes the effective first-order activity of phosphatase 'P' against pC-ERMAD, and the s are given in where , and , are the unbound concentration and Michaelis constant of substrate/inhibitor of phosphatase P. Σ , the sum of all the terms arising from the binding of the phosphatase to the competing substrates and inhibitors, measures the strength of the competition.
We determined Σ ℎ , the competition parameter at physiological total protein concentration, by measuring the ability of uterus extracts that had been diluted to different total protein concentrations, , to dephosphorylate a substrate that was known to be primarily dephosphorylated by phosphatase P. We modeled the variation of and Σ with as The first equation is precise, since total concentrations vary linearly with . While the second equation is an approximation that ignores concentration-dependent changes in the ratio between the unbound and total S , concentrations, we found that it well-fit either the entire data range (panel A) or the high-concentration region (i.e., ≥ 0.01 ℎ ; panel B), which was adequate for determining Σ ℎ .
Since pC-ERMAD is primarily dephosphorylated by PP1-type enzymes (see Figure 8 -supplement figure 1), it was a good substrate for determining Σ ℎ . Since ∕ was small, it was ignored in Eq. (1a). Nonlinear regression using Eqs. (1) and (2) with the data with ≥ 0.01 ℎ with an adjustment for the additional change in phosphatase activity as → 0 gave a best-fit of Σ = 15 ± 2 (panel B). We assumed that this was representative of competition for MLCP and used this value as Σ ℎ .
Σ ( ) and Σ ( ) were ∼ 2 at the extract concentrations used in the experiments of Figure 8, but this did not affect the the computation of [PPU ], or the predicted dephosphorylation rate at low pCPI-17 concentration. In contrast, these estimates do play a significant role in calculating the time-courses in Figure 9, as described in the next section.
Experiments with rabbit artery smooth muscle strips   (2) activation of PKC (and the resultant increase in CPI-17 phosphorylation) was sufficient to induce both MRLC phosphorylation and vasocontraction. These facts motivated the working hypothesis that the observed decrease in MRLC phosphorylation was due primarily to increased MLCP activity. In this case, Eq.
(3) implies that (vd)∕ (vc) = 2.4, and we used this value in the simulations described below. We also performed simulations that either used the previous experimental result (vd)∕ (vc) = 3 (30) or that ascribed some of the reduction in MRLC phosphorylation during vasodilation to decreased MLCK activity (thereby implying a smaller MLCP activation). These controls showed that the computed half-lives for pCPI-17 dephosphorylation and MLCP activation were not highly sensitive to the value used for