February 17, 1998
95 (4) 1568-1573
Insights into structure-function relations of many proteins opens the possibility of engineering peptides to selectively interfere with a protein’s activity. To facilitate the use of peptides as probes of cellular processes, we have developed caged peptides whose influence on specific proteins can be suddenly and uniformly changed by near-UV light. Two peptides are described which, on photolysis of a caging moiety, block the action of calcium-calmodulin or myosin light chain kinase (MLCK). The efficacy of theses peptides is demonstrated _in vitro_ and _in vivo_ by determining their effect before and after photolysis on activities of isolated enzymes and cellular functions known to depend on calcium-calmodulin and MLCK. These caged peptides each were injected into motile, polarized eosinophils, and when exposed to light promptly blocked cell locomotion in a similar manner. The results indicate that the action of calcium-calmodulin and MLCK, and by inference myosin II, are required for the ameboid locomotion of these cells. This methodology provides a powerful means for assessing the role of these and other proteins in a wide range of spatio-temporally complex functions in intact living cells.
Get alerts for new articles, or get an alert when an article is cited.
Several methods have been used in the past to probe the role of a protein in cell function, each with advantages as well as limitations. Organic compounds are available that can modulate the activity of proteins, but interpretation of effects often is complicated by their relatively slow onset and low selectivity for a specific protein. Impressive progress toward single protein specificity has been made with antisense (1) and homologous recombination (2) methods, which disrupt the expression, and thus the function of a specific protein. Interpretation of effects, or lack thereof, on a cellular function may be complicated, however, by compensatory pathways enhanced by the absence of the targeted protein. Peptides that bind to proteins with high affinity and high selectivity provide a means to rapidly and potently inhibit the activity of selected proteins, but peptides must be microinjected into cells, and microinjection itself can at least transiently alter cell function. It thus would be desirable to have a way to make a peptide that is initially inactive or “caged” because of a strategically placed photolabile moiety (3, 4). Such a peptide could be injected into a cell, and time allowed for it to distribute evenly and for normal cell function to be verified. The peptide’s biological activity then could be unmasked by light-directed removal of the photolabile group. Each cell would be its own control, thereby diminishing effects of cell to cell variability, and active peptides could be produced rapidly (within milliseconds) and with good spatial resolution.
We describe here the preparation and use of photoactivatable caged peptides targeted against calmodulin and myosin light chain kinase (MLCK). Because these two proteins are known to be essential in the control of smooth muscle contractility, the efficacy of the caged peptides was verified _in vitro_ by using smooth muscle MLCK (5) and in smooth muscle cells of toad stomach (6). There is also evidence that calmodulin and MLCK regulate cell motility in fibroblasts (7), but there is little information in other cells such as leukocytes where cell shape changes and intracellular Ca 2+ levels fluctuate rapidly, and cells move considerably faster (8). The caged peptides therefore were used to determine the extent to which individual eosinophil cells rely on calmodulin, MLCK, and by inference myosin II to control their rapid ameboid cell motility.
Caged peptides were synthesized by automated solid-phase methods using 9-fluorenylmethoxycarbonyl (Fmoc) amino acids. l-Tyrosine protected on its α-amino and α-carboxyl groups was coupled to 2-bromo-2′nitrophenylacetic acid methyl ester through its phenolic group (9). Methyl α-carboxyl caged tyrosine (MW = 374.1, _m_/_z_ 375.1) then was derivatized with an Fmoc moiety and incorporated into peptides during automated synthesis (Fig. 1_A_). Cleavage from the solid support and deprotection of side chains (90% trifluoroacetic acid) was followed by HPLC purification of the peptide then by hydrolytic removal of the methyl ester from the cage moiety using 10% aqueous K 2 CO 3. α-Carboxyl 2′-nitrobenzyl tyrosine containing peptides were repurified by HPLC and lyophilized. Peptide compositions were confirmed by amino acid analysis and MS: 5cgY-RS-20: MS, MW = 2434.2, found 2434.5. 5Y-RS-20 (obtained by photolysis of 5cgY-RS-20): MS, MW = 2256.2, found 2256.2. 5,18cgY-RS-20: MS, MW = 2663.2, found 2663.2. 9cgY-LSM1: MS, MW = 1862.3, found 1861.7. LSM1 (obtained by photolysis of 9cgY-LSM1) MS, MW = 1682.3, found 1682.
(_A_) Amino acid sequences of calmodulin inhibitory peptide RS-20 and variants containing glutamate (E), tyrosine (Y), and caged tyrosine (cgY). Peptide and caged peptide structures were confirmed by amino acid composition and mass spectrometry (see _Materials and Methods_). (_B_) Effects of RS-20 peptides on Ca 2+/calmodulin-dependent MLCK activity _in vitro_. 5cgY-RS-20 before (•) and after (○) photoconversion; the photoreleased peptide was indistinguishable from authentic 5Y-RS-20 in inhibitory potency, chromatographic behavior, and mass spectra. There was no detectable inhibition by 5E-RS-20 (▵), whereas 5,18cgY-RS-20 inhibited only weakly both before (▴) and after photolysis (▪). Data points represent means of triplicate measurements with SEM values less than 10% of the mean. (_C_) Structure and photolysis reaction of cgY.
MLCK was assayed as described (5) with the following modifications. Various concentrations of peptides were added to the assay mixture containing 0.5 μg/ml smooth muscle MLCK, 0.2 mg/ml smooth muscle regulatory light chain, 100 nM calmodulin, 0.1 mM CaCl 2, 100 mM KCl, 1 mM MgCl 2 and 30 mM Tris at pH 7.5, 25°C. Phosphorylation was initiated by addition of 50 μM 32 P-γ-ATP and continued for 5 min. Constitutively active catalytic domain of MLCK was prepared by trypsinization (5) and assayed in the above mixture containing 1 mM EGTA and no added Ca 2+. The selectivity of LSM1 peptide for MLCK over other protein kinases was examined as follows. At a concentration of 100 μM, LSM1 inhibited protein kinase A (PKA), protein kinase C (PKC), and calmodulin-dependent protein kinase II (CaMPKII) activity by 38%, 49%, and 47%, respectively. At a concentration of 37 nM, where LSM1 inhibited MLCK activity by 50%, LSM1 had no significant inhibitory effect on PKA, PKC, or CaMPKII.
LSM1 also inhibited less than 15% of Rho kinase activity (10) at concentrations up to 100 μM. Rho kinase was expressed in COS-7 cells, immunoprecipitated, then assayed for activity by using smooth muscle light chains as a substrate.
Calmodulin binding was assessed by adding 100 nmol aliquots of peptides to a 5-ml calmodulin agarose column (Sigma) equilibrated in 32 μM free Ca 2+ at 4°C. Peptides that bound were eluted from the column in 1 mM EGTA. All peptides then were identified by UV spectra and HPLC.
Newt eosinophil cells were prepared as described (8, 11). Briefly, cells were settled onto coverslips from 1:5 newt blood in amphibian culture medium, washed and counted in a chamber on an inverted microscope. Cells were injected with 1 mM caged peptide in a vehicle of 20 mM Hepes, pH 7.2, containing 0.1 mM fluorescein dextran (3 kDa) to estimate injection volume and peptide concentration (see below).
Cells were activated with newt serum and motility recorded at 1-sec intervals by a Dage video charge-coupled device camera and stored on video disks. Each cell served as its own control by recording for 5 min before UV exposure. Effects of UV exposure then were determined in the subsequent 5-min period by examining three categories of motility parameters: (_i_) whole-cell locomotion, shape, and polarity, (_ii_) lamellipod formation, expansion, ruffling, withdrawal, and distribution around the cell periphery, and (_iii_) granule motion patterns. Four types of control cells were examined. Cells injected with vehicle only or with 5,18cgY-RS-20 (an inactive variant of RS-20) were subjected to the same UV exposure protocol as the experimental cells. In addition, the behavior of uninjected cells and cells injected with caged peptide was recorded without being exposed to UV light. This method permitted double-blind analysis of responses by two scorers, each positive response to the active peptides and each negative response in control cells requiring consensus. Video records were viewed and scored a total of four times, the final scoring session occurring several weeks after the experiment. Cells remained active on the microscope for up to 3 hr after injection, and the peptides were effective throughout, indicating stability of caged peptides in the cytosol over this period.
Photorelease of peptides was accomplished with the near-UV output of an argon ion laser delivered in a series of 10 pulses each of 5-msec duration spaced 20 msec apart. The extent of photoconversion of caged peptide was estimated to be 45–55% by comparison with caged fluorescein. This estimate is consistent with the finding that enough caged peptide remained after a single exposure to obtain a response to a second exposure (in cells that recovered from the first). Depending on the injection volume, the cyotosolic caged peptide concentrations were estimated to be 20–100 μM, and final free peptide 10–50 μM. The extent and persistence of peptide effects on cell locomotion and granule flow was correlated with the volume of caged peptide introduced into each cell assessed by fluorescein dextran fluorescence.
Exposure of newt eosinophils to high levels of near-UV light resulted in some rundown of motility. This rundown usually could be distinguished from the effects of peptides in that it was nonspecific (e.g., ruffling also was lost) and cells never recovered. To minimize these effects of near-UV light, each series of experiments began with exposure of at least a dozen uninjected cells to determine the maximum dose of laser light that had no detectable effect on any of the motility parameters evaluated. In addition, at least one uninjected cell from each coverslip was exposed to near-UV light and recorded to rule out the possibility that UV sensitivity varied among different batches of cells.
A method was developed to prepare caged peptides for the purpose of inhibiting the activity of the calcium-calmodulin complex, as this complex is believed to play a key role in mediating many of the effects of calcium in cells. Our efforts were focused on RS-20 (Fig. 1_A_), a 20-aa calcium-calmodulin binding domain of smooth muscle MLCK (12). In an abbreviated structure-activity analysis of RS-20, we found that substitution of hydrophobic residues W-5, L-18, or both with E greatly reduced the affinity of RS-20 for calmodulin as assessed by an _in vitro_ MLCK assay (Fig. 1_B_). This finding confirmed that hydrophobic side chains are preferred in positions 5 and 18 for high affinity calmodulin binding (13) and showed that negatively charged side chains at these sites interfered with formation of the complex.
Several candidate caged forms of RS-20 were prepared by replacing W-5, L-18, or both with a caged tyrosine that featured a negatively charged photolabile group linked to tyrosine’s phenolic function (Fig. 1_C_). Analysis of these peptides by a quantitative calmodulin-dependent MLCK assay (5) identified only one that displayed properties appropriate for a caged peptide. The peptide, designated 5cgY-RS-20 (Fig. 1_A_), was largely inactive before photolysis and then showed smooth and complete photoconversion to 5Y-RS-20, resulting in an increase in apparent calmod
