ARCHIVES
OF BIOCHEMISTRY
AND
BIOPHYSICS
Vol. 280, No. 2, August 1, pp. 397-404,199O
Potent Peptide Inhibitors of Smooth Muscle Myosin Light Chain Kinase: Mapping of the Pseudosubstrate and Calmodulin Binding Domains Carolyn
J. Foster,*gl
Shona A. Johnston,*
Brooks
Sunday,?
Departments of *Cardiovascular Pharmacology and tMedicina1 86 Orange Street, Bloomfield, New Jersey 07003
Received December
Chemistry,
and Federico Schering-Plough
C. A. Gaeta? Research,
15,1989, and in revised form April 2,199O
Smooth muscle myosin light chain kinase (MLCK) is activated by calcium-calmodulin and, in turn, phosphorylates and activates the smooth muscle actomyosin ATPase, resulting in muscle contraction. The amino acid sequence of the regulatory domain of MLCK is known, and it contains a region that binds calmodulin and also bears a strong homology to the phosphorylation site in the substrate. Thus, it has been called the “pseudosubstrate.” It has been proposed that calmodulin activates MLCK by binding to and reversing the autoinhibitory function of the pseudosubstrate. Synthetic peptides based on this sequence inhibit MLCK both by binding to calmodulin and by competing with the substrate at the active site. In the work reported here, we have synthesized a large number of peptides from the regulatory region of MLCK (MLCK 480-516). The region was systematically analyzed by dividing it into fragments of two to six amino acids, each containing one or more basic residues, in order to map in detail the calmodulin binding site and the autoinhibitory region. It was observed that both calmodulin binding and autoinhibition are mediated by several different fragments of the regulatory sequence. Two nonoverlapping peptides, MLCK 480-493 and MLCK 494-504, are similar in potency in inhibiting the enzyme (ICBO’s of 2 and 6 PM, respectively). Larger fragments, combining multiple inhibitory regions, are more potent inhibitors. For example, MLCK 480-504 is extremely potent, with an IC& of 13 nM. The calmodulin binding site and active site directed inhibitory regions overlap, but are not identical. Residues 505-512 are important only for calmodulin binding. B 1990 Academic Press, Inc.
Smooth muscle contraction is triggered by an increase in intracellular calcium. Ca-calmodulin complexes form 1 To whom correspondence
should be addressed.
0003.9861/90 $3.00 Copyright CC 1990 by Academic Press, All rights of reproduction in any form
Inc. reserved.
in the smooth muscle cells and bind to MLCK,’ resulting in phosphorylation of myosin and activation of the actomyosin ATPase (1). Structural studies on the myosin light chain kinase molecule have shown that the enzyme consists of functional domains including a catalytic domain and a regulatory domain that binds calmodulin (2, 3). Gentle proteolytic hydrolysis of MLCK produces, first, an inactive enzyme of approximately half the size of the original molecule (4-6) that cannot be stimulated by calmodulin. On further hydrolysis, a fragment 30004000 Da smaller, with calmodulin-independent enzyme activity, is generated (4-6). These observations suggested that there was an autoinhibitory domain in the enzyme that could be removed by proteolysis. When the calmodulin binding domain of MLCK was sequenced, both by amino acid sequencing techniques (7) and by nucleotide sequencing of a cDNA clone encoding a portion of the carboxy terminal of the enzyme (8), it was observed to contain a sequence with strong homology to the recognition sequences in the substrate, myosin light chain. Kemp and collaborators (9) synthesized peptides based on this pseudosubstrate sequence and showed that these peptides were able to inhibit MLCK both by binding calmodulin and by a direct effect on the enzyme. Certain of the peptides were shown to be competitive with the natural substrate, myosin light chain, as well as the synthetic 13 amino acid substrate, KKRPQRATSNVFS-NH,. A model was proposed for regulation of MLCK in which, in the absence of calcium and calmodulin, the pseudosubstrate sequence functioned as an internal inhibitor of the enzyme by binding to the active site. In the presence of Ca-calmodulin, the inhibi-
‘Abbreviations used: MLCK, ethylene glycol bis(P-aminoethyl tert-butoxycarbonyl.
myosin light chain kinase; EGTA, ether) NJ’-tetraacetic acid; t-BOC;
397
398
FOSTER
tion was reversed (9). The model was directly confirmed by sequencing of proteolyzed MLCK (10). The work reported here confirms and expands these studies by using a large number of synthetic peptides to investigate in detail which amino acid clusters are involved in calmodulin binding and substrate antagonism. Basic amino acids are often essential for recognition of substrates by protein kinases. Kemp has previously shown the importance of clusters of basic amino acids to recognition of substrates by MLCK (11,lZ). For MLCK, the substrate recognition site is extremely specific, consisting of the following sequence, KKRPQRATSNVFA (MLC ll-23), where Ser” is the phosphorylation site (11). All four of the basic amino acids, the spacing between the arginines, and the spacing between the basic residues and the phosphorylatable serine are important to the affinity of binding of substrates and specificity of phosphorylation (11,12). Hydrophobic amino acids carboxy-terminal to Serlg contribute to the V,,, of phosphorylation but not to binding (13). Peptides in which the basic residues are retained, but the serine is absent or replaced with an amino acid that is not phosphorylatable, bind to the enzyme and act as inhibitors (14, 15), as does peptide MLC (11-19) itself due to its extremely (13). low vmax of phosphorylation The pseudosubstrate region of MLCK centers on the sequence RRKWQKTG (MLCK 493-501), which has substantial homology to the light chain recognition sequence. The peptide ARRKWQKTG (MLCK 492-501) functions as an active site directed inhibitor (9). Aminoterminal and carboxy-terminal extended segments of the MLCK sequence including these residues are both inhibitors and potent binders of calmodulin (9,16). The current study systematically explores each of the regions of the MLCK pseudosubstrate/calmodulin binding sequence containing one or more lysine or arginine residues for their importance in enzyme inhibition. Evidence is shown that separate segments of the enzyme sequence, in particular MLCK (480-493) (Peptide 18), which does not contain any of the basic residues of the pseudosubstrate, are able to inhibit the enzyme independently, apparently by competing with substrate-enzyme interaction. Peptides consisting of combinations of these segments are more potent inhibitors. Calmodulin binding resides in a segment that overlaps the substrate competitive regions. Of interest is that at least one portion of the sequence (MLCK 505-512) is important to calmodulin binding while playing a minor role in substrate antagonism. This sequence is located at the carboxy1 end of the binding site, and this is consistent with the fact that calmodulin binding is lost by mild proteolysis while the autoinhibitory function is retained. These results have been reported in preliminary form (17). MATERIALS
AND
METHODS
Protein preparations. Calmodulin-dependent MLCK of greater than 95% homogeneity prepared
chicken gizzard by the method of
ET AL. Ngai et al. (18) was generously provided by Dr. Paul Trotta, ScheringPlough Research. Calmodulin-independent MLCK was prepared by modifications of the digestion procedure of Tanaka et al. (19) and the isolation procedure of Walsh et al. (2) as described by Foster et al. (4). This preparation of calmodulin-independent MLCK is stable for over 2 years at -70°C. Enzyme a.ssuys. MLCK was assayed using the synthetic peptide substrate KKRPQRATSNVFS-NH,, a 13 amino acid fragment with a sequence corresponding to residues 11-23 of gizzard light chain except for a carboxy-terminal serine-NH, (11). Kemp has demonstrated similar results with this synthetic substrate and light chain (9), and studies in this laboratory with several classes of MLCK inhibitors have shown that IC,, values do not change significantly when the synthetic peptide is substituted for light chain. The standard assay was pH performed in a total volume of 100 ~1 containing 20 mM Tris-HCI, 7.2, 10 mM MgCI,, 100 pM [Y-~‘P]ATP with a specific activity of 300and 0.02 Fg en1000 cpm/pmol, 50 pM KKRPQRATSNVFS-NH2, zyme. Calmodulin was added at the concentrations indicated in the text and tables, and assays contained 1 mM EGTA or 1 mM EGTA plus approximately 200 pM excess free calcium when calcium-dependent activity was measured. Calcium-independent activity was negligible in the calcium-dependent gizzard enzyme. Assays were initiated by the addition of ATP and stopped after 30 min by the addition of 200 FM HCl. Phosphate incorporated into the basic substrate was measured by spotting an aliquot on phosphocellulose filter paper, washing, and counting the papers. Assays were performed by hand or by using the Biomek Automated Pipetting Station (Beckman, Inc.). Many of the peptides studied contain serine or threonine residues. In order to demonstrate that these peptides were not phosphorylated by MLCK, a model peptide was synthesized. This peptide, RRKWQKTSHAVRAIGRLSS-NH,, contained Thr’“, Se?“, and Se?*, in addition to Se?“‘, added as another possible phosphorylation site. This peptide was not phosphorylated under the standard assay conditions. Since a large number of peptides were studied, determination of K, values for each was impractical. Therefore, I& values, determined under constant conditions, were used to compare inhibitor potencies. For determination of I& values, inhibitors were tested in duplicate at four concentrations, including concentrations above and below the I&,, and values were determined graphically. K, values were determined when possible for key inhibitors. Peptide synthesis. KKRPQRATSNVFS-NH2 and all inhibitor peptides were prepared by stepwise solid-phase synthesis on an Applied Biosystems Model 430A peptide synthesizer. The t-BOC protection methodology was used, and final cleavage was with HF/anisole. Peptides were purified by reverse-phase HPLC using a Rainin Dynamax C8 column and a Water’s gradient HPLC system. Peptides were eluted with a gradient of 0.1% trifluoroacetic acid in acetonitrile. Peptides were a single peak by HPLC and were analyzed by FAB mass spectrometry to confirm structure. All peptides described in this work were carboxy-terminal amides. Reagents. [y-“‘P]ATP was obtained from ICN, calmodulin was purchased from Boehringer-Mannheim, and all other chemicals were obtained from Sigma.
RESULTS
Active Site Directed Peptide Inhibitors Table I shows the sequence of MLCK in the region of the calmodulin binding site (residues 480-516) as it was deduced from the sequence of the cloned DNA (8). It has been divided into sections, each containing one or more basic amino acids. These sections have been systematically examined for their importance to enzyme inhibition using the peptides listed in Table I. Assays were per-
POTENT
PEPTIDE
INHIBITORS
OF MYOSIN
LIGHT
CHAIN
KINASE
399
TABLE I Active Site Directed
Peptide Inhibitors
I’eptide number
of Calmodulin-Independent
Residue number
Sequence AKK-LSKD-RM-KKYMA-RRKWQ-KTGHAV-RAIGRLSSMAMI'
A 480 1 2 3 4 r, 6 7 8 9 10 11 12 13 14 15 16 17 18
A
A
A
A
485
490
495
500
Myosin Light Chain Kinase
A 505
A 510
480-516
Ic,,, (PM) Or % inhibition Not made
A 515
KTGHAV-R* KTGHAV-F TGHAV-R RAIGRLSS RLSSMAMI AV-RAIGRLSS KWQ-KTGHAV KWQ-KTGHAV-RAIGRL RRKWQ-KTGHAV RRKWQ-KTGHAV-RAIGRLSS RRKWQ-KTGHAV-FAIAAL KKYMA-RRKWQ-KTGHAV KKYMA-RRKWQ RM-KKYMA-RRKWQ LSKD-RM-KKYMA-RRKWQ AKK-LSKD-RM-KKYMA-RRKWQ AKK-LSKD-RM-KKYMA-RRKWQ-KTGHAV AKK-LSKD-RM-KKYMA
499-505 500-505 505-512 509-516 503-512 496-504 496-510 494-504 494-512 489-504 489-498 487-498 483-498 480-498 480-504 480-493
” Sequence of the calmodulin binding site from myosin light chain kinase from Guerriero rt al. (8). ’ All peptides are carhoxy-terminal amides. (A) Intervals of five amino acid residues. I& values are the average of two or more determinations.
formed using calmodulin-independent MLCK in the absence of calmodulin so that direct enzyme inhibition could be studied without the complicating influence of calmodulin binding. The first peptide synthesized was RRKWQKTGHAV (Peptide 9, Table I) because it showed a high degree of homology to the peptide, KKRPQRATSNVFS-NH2, an excellent substrate for MLCK. Peptide 9 contained all
the characteristics of the myosin light chain recognition site, that is, three basic amino acids separated from the fourth by two amino acids, with glutamine in the second position as it is in the substrate. This peptide will be referred to as the “core peptide.” It was indeed an inhibitor of MLCK, with an IC& of 2 ~.LMunder standard assay conditions. To characterize the mode of inhibition, kinetics studies were performed. At 100 yM ATP, the peptide was apparently competitive with the peptide substrate (Fig. 1A). Secondary plots of apparent K, vs inhibitor concentration were linear, and the apparent K,, an average of three separate determinations, was 2 PM. The equality of IC,,, values and Ki values is a coincidence due to a favorable choice of assay conditions, however, an excellent correlation between IC,, values and Kl’s has been observed for MLCK in this laboratory and others (Ref. (15) and C. Foster, unpublished observations), and IC,,, values will be reported for most peptides. A similar study was performed varying ATP, and a nonlinear double-reciprocal plot was observed at
strongly inhibitory concentrations of the peptide (Fig. 1B). The explanation for this phenomenon was not known, and it was not possible to conclude whether the inhibition was competitive or noncompetitive from these studies. Since the core peptide was an inhibitor as predicted, with a K, of the same order of magnitude as the K,,, for the substrate (myosin light chain, Km = 8.6 PM; Ref. (ll)), the effect of the addition of amino acids in both the carboxy and amino directions was explored. Peptides in the region 499-512 (Peptides l-6) were poor inhibitors of the enzyme and had little effect on inhibitory potency when covalently bound to the core peptide as demonstrated by comparing the IC,, of the core peptide (2.0 PM) with that of Peptide 10 (2.4 PM). Replacing Arg”“” and Arg50g with Phe and Ala, respectively (Peptide II), had no significant effect on inhibitory potency. By contrast, amino acids to the amino-terminal side had a significant influence on inhibitory potency. The arginine pair at position 494-495 was critical to inhibitory activity, since the ICso of the core peptide was at least 150 times lower than that of the peptide missing these two residues (Peptide 7). Adding residues 489-493 (KKYMA) to the core peptide improved the inhibition fivefold (Peptide 12). The peptide KKYMARRKWQ (Peptide 13) was an equal inhibitor to the core peptide, and amino-terminal additions to this peptide further increased the inhibitor potency. RM (487-488) had an ap-
FOSTER
-40
-20
0
20 11
40
[PEPTIDE]
FIG. 1. Kinetics studies with chain kinase as described under tion with the peptide substrate, HIM. For competition with ATP, ( ?) pM. Assays were performed
60
80
100
120
ET AL.
-100
0
(mhl)-1
100 11
IAlP]
200
300
400
(mM)-1
the peptide MLCK 494-504 (Peptide 9). Assays were performed with calmodulin-independent myosin light Materials and Methods. Velocity was measured as nanomoles phosphate incorporated in 30 min. For competiATP was held constant at 100 WM (A). Inhibitor concentrations were 0 (n), 1 (II), 2 (B), 3 (c), 5 (E), and 10 (+) the peptide substrate was held constant at 100 pM (B). Inhibitor concentrations were 0 (O), 1 (+), 3 (P), and 5 in duplicate.
proximately threefold effect (Peptide 14), while LSKD (483-486) increased inhibitor potency approximately an additional sixfold (Peptide 15). AKK (480-482) had no significant effect (Peptide 16), suggesting that this is the amino-terminal end of the pseudosubstrate region. The two halves of the core peptide were investigated individually. KTGHAV (499-504) increased potency 5 fold when added to the carboxy terminal of KKYMARRWQ (489-498) (Peptide 12) and g-fold when added to the longer peptide, AKKLSKDRMKKYMARRKWQ (480-498) (Peptide l?), although the peptide KTGHAVR (Peptide 1) was not an inhibitor on its own. RRKWQ (494-498) had a 67-fold effect when added to the carboxy terminal of AKKLSKDRMKKYMA (480-493) (Peptide 16). Thus, both halves of the core peptide contribute to inhibitory potency, and RRKWQK, or perhaps the two arginines alone (see above), has the greatest individual effect. Combination of all five of the regions shown to contribute to inhibition results in an extremely potent inhibitor with an I&, of 13 nM, AKKLSKDRMKKYMARRKWQKTGHAV (Peptide 17), which was almost lo-fold more potent than any other peptide tested. To prove the point that more than one region of the calmodulin binding sequence can act to inhibit MLCK, Peptide 18 was tested. This peptide does not overlap with the core peptide, and it has little homology with the substrate sequence. Nevertheless, it is an excellent inhibitor with an IC& of 6 PM. Similar to the core peptide, this peptide is apparently competitive with the peptide substrate, with an apparent Kc of 2 PM (data not shown).
Studies on Inhibition
of MLCK
by Peptide 480-504 (17)
Peptide 480-504 (Peptide 17, I& = 13 nM) is a covalent combination of Peptide 9 (I&, = 2 PM) and Peptide 18 (IC& = 6 PM), both of which are apparently competitive with the peptide substrate. The two smaller peptides were tested in combination, to see whether the significant increase in potency observed vs calmodulin-independent MLCK with the combined peptide resulted simply from simultaneous occupation of two inhibitory sites, or whether the covalent connection of the peptides contributed to the inhibition. As can be seen in Fig. 2, a 1:l mixture of the two peptides gave an inhibitor curve that was intermediate between the two individual curves, not shifted to the nanomolar range as is observed with Peptide 17, demonstrating that covalent connection of the peptides is necessary for the improved inhibitory potency. Kinetic studies were performed on Peptide 17. Competition with the peptide substrate yielded nonlinear double-reciprocal plots under the usual assay conditions (not shown). However, in the standard assay, enzyme concentrations are in the low nanomolar range, similar to the peptide concentration. Therefore, kinetics studies were repeated at l/10 the usual enzyme concentration. At these low activity levels, experimental errors were higher, but double-reciprocal plots were linear and showed that this peptide was apparently noncompetitive with the peptide substrate (Fig. 3A). When the peptide substrate was held constant, and ATP was varied, double-reciprocal plots indicated that this peptide was apparently competitive with ATP (Fig. 3B). The explanation for these results is not clear at this time; however, a
POTENT
PEPTIDE
INHIBITORS
100
60
g i z z z
6o
40
20
0 -t.I
r
1 PEPTIDE
10 CONC.
100 (FM)
FIG. 2. Inhibition of myosin light chain kinase by MLCK 494-504 (Peptide 9) and MLCK 480-493 (Peptide 18) alone and in combination. Calmodulin-independent myosin light chain kinase was tested in the absence of inhibitory peptides or in the presence of peptide concentrations as shown (Peptide 9,O; Peptide 18, *). For the combination of peptides (m), Peptides 9 and 18 were mixed at equal concentrations so that the total peptide concentration is shown on the r-axis, and the concentration of each peptide is half the total. Inhibition by the combination is approximately the average of the inhihitions by the two peptides. The IC,, values in this experiment were 2 ELM for Peptide 9, 4.5 @M for Peptide 18, and 2.9 fiM for the combination. The IC,, for MLCK 480-504 (Peptide 17), in which these two peptides are covalently linked, is 13 nhf.
secondary plot of apparent K,,, vs inhibitor concentration was nonlinear, suggesting that the kinetics are complex (data not shown). Similar kinetic results were observed by Ikebe et al. (20). Peptide Inhibitors
Involved
in Calmodulin
Binding
To determine the portions of the calmodulin binding region that are important to calmodulin antagonism, assays were performed using both calmodulin-dependent MLCK and calmodulin-independent MLCK. Earlier studies have shown that active site directed MLCK inhibitors have similar IC& values against both forms of the enzyme. This is true both for antagonists of the peptide substrate (21) and for antagonists of ATP (C. Foster, unpublished observations). However, if an active site directed inhibitor also binds to calmodulin, its ability to inhibit the calmodulin-independent enzyme will be reduced in the presence of calmodulin to the extent that the free inhibitor molecule is removed from the enzyme by binding to calmodulin. In assays for calmodulin-dependent MLCK, the results with inhibitors that interact both with calmodulin and the enzyme are complex. Whether the IC,,, of a calmodulin binding peptide against calmodulin-dependent MLCK is higher or lower
OF MYOSIN
LIGHT
CHAIN
KINASE
401
than that for inhibition of calmodulin-independent MLCK will depend on the relative affinities for the enzyme and calmodulin, as well as the concentrations of each of the components of the reaction mixture. In general, however, a change in the ICSOvalue for a peptide in the presence of calmodulin implies binding to calmodulin. Table II shows a selection of peptides used to determine the regions that are critical to calmodulin antagonism. The 1C5,,for inhibition of calmodulin-independent enzyme and calmodulin-dependent enzyme are shown for comparison. The parent peptide is divided arbitrarily into three regions, each of which contributes to calmodulin binding. The core peptide (Peptide 9) shows essentially no difference in IC& with the two enzymes, suggesting relatively poor binding to calmodulin, at least in the micromolar affinity range. Means (16) has measured the binding of calmodulin and ARRKWQKTG, a peptide similar to the core peptide, and found a binding constant of 700 PM, significantly above the inhibitory concentration (48 PM) and consistent with the idea that similar IC,,, values using the two enzyme preparations indicates poor binding. Larger peptides from the calmodulin binding site have binding affinities in the nanomolar range (16). The importance of the two arginine residues Arg4”4Arg4’” was investigated by testing Peptide 7. Although the ability to inhibit the calmodulin-independent enzyme was low, there was a similar low potency of inhibition against the dependent enzyme, consistent with poor calmodulin binding, even at 300 WM peptide, significantly in excess of the 200 nM calmodulin concentration. When additions were made to the core peptide in the carboxy-terminal direction, however, a significant improvement in calmodulin binding was observed. Peptides 10 and 20 were dramatically better inhibitors of the calmodulin-independent enzyme, and this improved inhibitory activity seemed to require basic amino acids, since the Peptide 11 with Arg5”5 replaced by phenylalanine and Arg50g replaced by alanine had a lo-fold higher ICSOvalue. On the amino-terminal side, Peptide 18 was a poor calmodulin binder on its own, but in combinat.ion with RRKWQ or the core peptide (Peptides 16 and 17), the resulting peptides showed a significant shift in their ability to inhibit calmodulin-dependent MLCK. In these cases, the IC&‘s were actually higher in the presence of calmodulin, and dependent on the calmodulin concentration, although they were of much smaller magnitude than the shifts for Peptides 10 and 20. These results are in general agreement with actual binding measurements (16). These studies led to the question of whether the mechanism of alteration of ICSo values was purely due to calmodulin binding by the peptides. Results with Peptides 10 and 20 suggest that the mechanism is solely by binding to and antagonism of calmodulin, since the I&, val-
402
FOSTER
ET AL.
C
0 80
-60
-40
-20
0
l/(Kemp)
20
40
60
80
100
120
-100
(mM)-1
0
100
200
11 ATP
(mM)-1
300
400
FIG. 3. Kinetics studies with MLCK 480-504 (Peptide 17). Assays were performed with calmodulin-independent myosin light chain kinase as described under Materials and Methods. Velocity was measured as nanomoles phosphate incorporated in 30 min. For competition with the peptide substrate, ATP was held constant at 100 FM, and enzyme concentration was lo-fold lower than ordinarily used (A). Inhibitor concentrations were 0 (O), 10 (+), and (B) 30 nM. For competition with ATP, the peptide substrate was held constant at 100 PM (B). Inhibitor concentrations were 0 (O), 15 (+), 20 (W), and (c‘) 25 nM.
ues for the calmodulin-dependent enzyme are 16 and 14 nM, respectively, concentrations where there would be no inhibition at all of calmodulin-independent MLCK. Peptide 17 was tested further, since it was the most potent inhibitor in the absence of calmodulin, and there-
TABLE
fore the I&, values in the presence of calmodulin were in a range where there would be a direct effect on calmodulin-independent MLCK. An experiment was performed using calmodulin-independent MLCK in the presence of calmodulin and calcium. Since calmodulin
II
Peptide Inhibitors of Myosin Light Chain Kinase from the Calmodulin Binding Site I&, Peptide number
Residue number
Sequence AKK-LSKD-RM-KKYMA-RRKWQ-KTGHAV-RAIGRLSSMAMIn
A A 480 485 9 7 19
12 10 20 11 8
21 17 18 15 16
A 490
A 495
A 500
CaM-Independent MLCK
480~516
Not made
494-504 496-504
17%at 300 GM
.
or % mhibition
CaM-Dependent MLCK
A A A 505 510 515
RRKWQKTGHAV KWQKTGHAV RRKAAKTGHAV KKYMA-RRKWQKTGHAV RRKWQKTGHAV-RAIGRLSS RRKWQKTGHAV-RAIGRL RRKWQKTGHAV-FAIAAL KWQKTGHAV-RAIGRL KTGHAV-RAIGRL AKKLSKDRMKKYMA-RRKWQKTGHAV AKKLSKDRMKKYMA LSKDRMKKYMA-RRKWQ AKKLSKDRMKKYMA-RRKWQ
’ Sequence of the calmodulin * 20 nM CaM. ’ 25 nM CaM. d 10 nM CaM.
( fiM)
2.0
2.2b
110
30%at 300pM* 28%at 100 pM*
489-504 494-512 494-510
0.4 2.4 5.0
0.9’ 0.016* o.014d
1.0
0.17'
496-510 499-510
287
480-504 480-493 483-498 480-498
binding site from myosin light chain kinase from Guerriero
16% at 300 0.013 6.0
GM
0.1 0.09
et al. (8). (A) Intervals
0.17’ 150d 0.21’, 0.105b 13.5’ 0.4d, 4.0h Oxd,
1.5*
of five amino acid residues.
POTENT TABLE
PEPTIDE
INHIBITORS
III
Comparison of the ICso Values for Peptide 17 in the Presence and Absence of Calmodulin 1% (nM) Enzyme type CaM-independent CaM-dependent
0 CaM 13 NT
25 nM
CaM
25 21
200 nM CaM 75 105
has no direct effect on the activity of this enzyme, any alterations in the I&, value of the peptide must be due to binding of the peptide to calmodulin, thereby preventing the peptide from inhibiting the enzyme. Table III shows the results of this experiment and compares it to a similar experiment using calmodulin-dependent enzyme. The ICsOvalues are similar for both enzyme preparations at the same calmodulin concentrations, making it unnecessary to invoke any mechanism other than binding of the peptide to calmodulin to explain the effects. In addition, the shift in IC,,, values in the presence of calmodulin occurred only when calcium was present, indicating that binding of the peptide to calmodulin required calcium. Note that in these experiments the peptide concentration is less than or approximately equal to the calmodulin concentration, and since the affinity of MLCK for Ca’+-calmodulin is high (& = 1 nM, (22)), adequate calmodulin is probably present to activate the enzyme, despite binding of the peptide. DISCUSSION The amino acid sequence of chicken gizzard myosin light chain kinase contains a region with homology to the substrate. Synthetic peptides mimicking this region of MLCK inhibit the catalytic activity of the enzyme directly and also bind to and antagonize calmodulin. By using results with both calmodulin-dependent and calmodulin-independent smooth muscle myosin light chain kinase and a series of synthetic peptides based on the enzyme sequence, regions of the MLCK sequence critical to direct inhibition and calmodulin binding have been mapped. Twenty-two amino acids, MLCK (483504) contribute to both types of inhibition, and five separate peptide sequences of 4-6 amino acids have been identified that contribute individually to direct inhibition, as well as being involved in calmodulin binding. These are LSKD (483-486), RM (487-488), KKYMA (489-493), RRKWQ (494-498), and KTGHAV (499504). Residues MLCK (505-512), by contrast, are critical to calmodulin binding but are unimportant to direct inhibition of the enzyme. Previous studies on gizzard myosin light chain kinase (4-6) have demonstrated that calmodulin binding can be
OF MYOSIN
LIGHT
CHAIN
403
KINASE
destroyed by proteolysis while autoinhibition is retained. A study by peptide sequencing (10) has shown that Arg505 is the carboxy-terminal amino acid of an inactive fragment of MLCK that has lost the ability to bind calmodulin. Thus, it seems that loss of MLCK (505-512), a peptide shown in these studies to be critical only to calmodulin binding and not to active site inhibition, is sufficient to eliminate effective calmodulin binding in proteolyzed enzyme, even though segments of the sequence that remain after proteolysis bind calmodulin when they are tested as independent peptides. While results obtained with isolated peptides must be interpreted with caution since conformation may not be retained, it is possible that these peptide regions do indeed contribute to calmodulin binding in the nonproteolyzed enzyme. In another report (6) an inactive fragment was sequenced and found to terminate at Arg4”. In our studies, MLCK (480-493) was shown to be an active site directed inhibitor with little calmodulin binding activity. Thus, a proteolyzed enzyme terminating at residue 493 would be expected to have lost calmodulin binding activity while retaining autoinhibition, consistent with the conclusions of Ikebe et al. (6). This is of interest, since MLCK (480-493) does not include the putative pseudosubstrate sequence (9). However, while Ikebe concludes that the inhibitory sequence includes only residues 474-490, the inhibitory potency of peptides including residues 493504 (note especially Peptide 17) strongly suggests the additional role of these amino acids in autoinhibition. The carboxy terminal of an active, calmodulin-independent form of the enzyme has been identified as Lys4”’ (6). This is consistent with the results shown here, since all of the predicted inhibitory peptides are missing from this fragment. The results in this study expand on earlier studies by providing a detailed description of the amino acids that are important for calmodulin binding and autoinhibition. The complex overlapping of the pseudosubstrate and calmodulin binding site and the synergistic participation of multiple, linearly arranged regions of the enzyme in regulation show how elegantly and economically the enzyme myosin light chain kinase has evolved. ACKNOWLEDGMENTS The authors thank Dr. Michael Czarniecki and Dr. Barry Pitts for their helpful suggestions and Dr. David Hartshorne for his excellent advice on the research.
REFERENCES 1. Kamm, K. E., and Stull, J. T. (1985) Annu. cd.
Reu. Pharmacol.
TOG-
25,593-620.
2. Walsh, M. P., Dahrowska, R., Hinkins, S., and Hartshorne, (1982)Biochemistry21,1919-1925. 3. Foyt, H. L., Guerriero, V., and Means, A. R. (1985) J. Bid. 260,1765-7774.
D. .J. Chem.
404
FOSTER
ET AL.
11. Kemp, B. E., Pearson, R. B., and House, C. (1983) Proc. Nat. Acad. Sci. USA 80,7471-7475.
13. Pearson, R. B., Misconi, L. Y., and Kemp, B. E. (1986) J. Biol. Chem. 261,25-27. 14. Pearson, R. B., Floyd, D. M., Hunt, J. T., Lee, V. G., and Kemp, B. E. (1988) Arch. Biochem. Biophys. 260,37-44. 15. Hunt, J. T., Floyd, D. M., Lee, V. G., Little, D. K., and Moreland, S. (1989) Biochem. J. 257,73-78. 16. Means, A. R. (1988) Recent Prog. Horm. Res. 44,223-262. 17. Foster, C. J., and Gaeta, F. C. A. (1988) Biophys. J. 53,182a. 18. Ngai, P. J., Carruthers, C. A., and Walsh, M. P. (1984) Biochem. J. 2 l&863-870. 19. Tanaka, T., Naka, M., and Hidaka, H. (1980) Biochem. Biophys. Res. Commun. 92,313-318. 20. Ikebe, M., Kemp, B. E., and Hartshorne, D. J. (1988) Biophys. J. 53,182a. 21. Gaeta, F. C. A., Lehman De Gaeta, L., Or, Y. S., Foster, C. J., and Czarniecki, M. (1990) J. Med. Chem. 33,964-972.
12. Kemp, B. E., and Pearson, R. B. (1985) J. Biol. Chem. 260,33553359.
22. Walsh, M. P., Hinkins, S., Flink, (1982) Biochemistry 21,6890-6896.
4. Foster, C. J., Van Fleet, M., and Marshak, A. M. (1986) Arch. Biothem. Biophys. 25 1,616-623. 5. Ikebe, M., Stepinska, M., Kemp, B., Means, A. R., and Hartshorne, D. J. (1987) J. Biol. Chem. 262, 13,828-13,834. 6. Ikebe, M., Maruta, S., and Reardon, S. (1989) J. Biol. Chem. 264, 6967-6971. 7. Lukas, T. J., Burgess, W. H., Prendergast, F. G., Lau, W., and Watterson, D. M. (1986) Biochemistry 25,1458-1464. 8. Guerriero, V., Russo, M., Olson, N., Putkey, A. R. (1986) Biochemistry 25,8372-8381.
J. A., and Means,
9. Kemp, B. E., Pearson, R. B., Guerriero, V., Bagchi, Means, A. R. (1987) J. Biol. Chem. 262,2542-2548.
I. C., and
10. Pearson, R. B., Wettenhall, R. E. H., Means, A. R., Hartshorne, D. J., and Kemp, B. E. (1988) Science 241,970-973.
I. L., and Hartshorne,
D. J.
OF BIOCHEMISTRY
AND
BIOPHYSICS
Vol. 280, No. 2, August 1, pp. 397-404,199O
Potent Peptide Inhibitors of Smooth Muscle Myosin Light Chain Kinase: Mapping of the Pseudosubstrate and Calmodulin Binding Domains Carolyn
J. Foster,*gl
Shona A. Johnston,*
Brooks
Sunday,?
Departments of *Cardiovascular Pharmacology and tMedicina1 86 Orange Street, Bloomfield, New Jersey 07003
Received December
Chemistry,
and Federico Schering-Plough
C. A. Gaeta? Research,
15,1989, and in revised form April 2,199O
Smooth muscle myosin light chain kinase (MLCK) is activated by calcium-calmodulin and, in turn, phosphorylates and activates the smooth muscle actomyosin ATPase, resulting in muscle contraction. The amino acid sequence of the regulatory domain of MLCK is known, and it contains a region that binds calmodulin and also bears a strong homology to the phosphorylation site in the substrate. Thus, it has been called the “pseudosubstrate.” It has been proposed that calmodulin activates MLCK by binding to and reversing the autoinhibitory function of the pseudosubstrate. Synthetic peptides based on this sequence inhibit MLCK both by binding to calmodulin and by competing with the substrate at the active site. In the work reported here, we have synthesized a large number of peptides from the regulatory region of MLCK (MLCK 480-516). The region was systematically analyzed by dividing it into fragments of two to six amino acids, each containing one or more basic residues, in order to map in detail the calmodulin binding site and the autoinhibitory region. It was observed that both calmodulin binding and autoinhibition are mediated by several different fragments of the regulatory sequence. Two nonoverlapping peptides, MLCK 480-493 and MLCK 494-504, are similar in potency in inhibiting the enzyme (ICBO’s of 2 and 6 PM, respectively). Larger fragments, combining multiple inhibitory regions, are more potent inhibitors. For example, MLCK 480-504 is extremely potent, with an IC& of 13 nM. The calmodulin binding site and active site directed inhibitory regions overlap, but are not identical. Residues 505-512 are important only for calmodulin binding. B 1990 Academic Press, Inc.
Smooth muscle contraction is triggered by an increase in intracellular calcium. Ca-calmodulin complexes form 1 To whom correspondence
should be addressed.
0003.9861/90 $3.00 Copyright CC 1990 by Academic Press, All rights of reproduction in any form
Inc. reserved.
in the smooth muscle cells and bind to MLCK,’ resulting in phosphorylation of myosin and activation of the actomyosin ATPase (1). Structural studies on the myosin light chain kinase molecule have shown that the enzyme consists of functional domains including a catalytic domain and a regulatory domain that binds calmodulin (2, 3). Gentle proteolytic hydrolysis of MLCK produces, first, an inactive enzyme of approximately half the size of the original molecule (4-6) that cannot be stimulated by calmodulin. On further hydrolysis, a fragment 30004000 Da smaller, with calmodulin-independent enzyme activity, is generated (4-6). These observations suggested that there was an autoinhibitory domain in the enzyme that could be removed by proteolysis. When the calmodulin binding domain of MLCK was sequenced, both by amino acid sequencing techniques (7) and by nucleotide sequencing of a cDNA clone encoding a portion of the carboxy terminal of the enzyme (8), it was observed to contain a sequence with strong homology to the recognition sequences in the substrate, myosin light chain. Kemp and collaborators (9) synthesized peptides based on this pseudosubstrate sequence and showed that these peptides were able to inhibit MLCK both by binding calmodulin and by a direct effect on the enzyme. Certain of the peptides were shown to be competitive with the natural substrate, myosin light chain, as well as the synthetic 13 amino acid substrate, KKRPQRATSNVFS-NH,. A model was proposed for regulation of MLCK in which, in the absence of calcium and calmodulin, the pseudosubstrate sequence functioned as an internal inhibitor of the enzyme by binding to the active site. In the presence of Ca-calmodulin, the inhibi-
‘Abbreviations used: MLCK, ethylene glycol bis(P-aminoethyl tert-butoxycarbonyl.
myosin light chain kinase; EGTA, ether) NJ’-tetraacetic acid; t-BOC;
397
398
FOSTER
tion was reversed (9). The model was directly confirmed by sequencing of proteolyzed MLCK (10). The work reported here confirms and expands these studies by using a large number of synthetic peptides to investigate in detail which amino acid clusters are involved in calmodulin binding and substrate antagonism. Basic amino acids are often essential for recognition of substrates by protein kinases. Kemp has previously shown the importance of clusters of basic amino acids to recognition of substrates by MLCK (11,lZ). For MLCK, the substrate recognition site is extremely specific, consisting of the following sequence, KKRPQRATSNVFA (MLC ll-23), where Ser” is the phosphorylation site (11). All four of the basic amino acids, the spacing between the arginines, and the spacing between the basic residues and the phosphorylatable serine are important to the affinity of binding of substrates and specificity of phosphorylation (11,12). Hydrophobic amino acids carboxy-terminal to Serlg contribute to the V,,, of phosphorylation but not to binding (13). Peptides in which the basic residues are retained, but the serine is absent or replaced with an amino acid that is not phosphorylatable, bind to the enzyme and act as inhibitors (14, 15), as does peptide MLC (11-19) itself due to its extremely (13). low vmax of phosphorylation The pseudosubstrate region of MLCK centers on the sequence RRKWQKTG (MLCK 493-501), which has substantial homology to the light chain recognition sequence. The peptide ARRKWQKTG (MLCK 492-501) functions as an active site directed inhibitor (9). Aminoterminal and carboxy-terminal extended segments of the MLCK sequence including these residues are both inhibitors and potent binders of calmodulin (9,16). The current study systematically explores each of the regions of the MLCK pseudosubstrate/calmodulin binding sequence containing one or more lysine or arginine residues for their importance in enzyme inhibition. Evidence is shown that separate segments of the enzyme sequence, in particular MLCK (480-493) (Peptide 18), which does not contain any of the basic residues of the pseudosubstrate, are able to inhibit the enzyme independently, apparently by competing with substrate-enzyme interaction. Peptides consisting of combinations of these segments are more potent inhibitors. Calmodulin binding resides in a segment that overlaps the substrate competitive regions. Of interest is that at least one portion of the sequence (MLCK 505-512) is important to calmodulin binding while playing a minor role in substrate antagonism. This sequence is located at the carboxy1 end of the binding site, and this is consistent with the fact that calmodulin binding is lost by mild proteolysis while the autoinhibitory function is retained. These results have been reported in preliminary form (17). MATERIALS
AND
METHODS
Protein preparations. Calmodulin-dependent MLCK of greater than 95% homogeneity prepared
chicken gizzard by the method of
ET AL. Ngai et al. (18) was generously provided by Dr. Paul Trotta, ScheringPlough Research. Calmodulin-independent MLCK was prepared by modifications of the digestion procedure of Tanaka et al. (19) and the isolation procedure of Walsh et al. (2) as described by Foster et al. (4). This preparation of calmodulin-independent MLCK is stable for over 2 years at -70°C. Enzyme a.ssuys. MLCK was assayed using the synthetic peptide substrate KKRPQRATSNVFS-NH,, a 13 amino acid fragment with a sequence corresponding to residues 11-23 of gizzard light chain except for a carboxy-terminal serine-NH, (11). Kemp has demonstrated similar results with this synthetic substrate and light chain (9), and studies in this laboratory with several classes of MLCK inhibitors have shown that IC,, values do not change significantly when the synthetic peptide is substituted for light chain. The standard assay was pH performed in a total volume of 100 ~1 containing 20 mM Tris-HCI, 7.2, 10 mM MgCI,, 100 pM [Y-~‘P]ATP with a specific activity of 300and 0.02 Fg en1000 cpm/pmol, 50 pM KKRPQRATSNVFS-NH2, zyme. Calmodulin was added at the concentrations indicated in the text and tables, and assays contained 1 mM EGTA or 1 mM EGTA plus approximately 200 pM excess free calcium when calcium-dependent activity was measured. Calcium-independent activity was negligible in the calcium-dependent gizzard enzyme. Assays were initiated by the addition of ATP and stopped after 30 min by the addition of 200 FM HCl. Phosphate incorporated into the basic substrate was measured by spotting an aliquot on phosphocellulose filter paper, washing, and counting the papers. Assays were performed by hand or by using the Biomek Automated Pipetting Station (Beckman, Inc.). Many of the peptides studied contain serine or threonine residues. In order to demonstrate that these peptides were not phosphorylated by MLCK, a model peptide was synthesized. This peptide, RRKWQKTSHAVRAIGRLSS-NH,, contained Thr’“, Se?“, and Se?*, in addition to Se?“‘, added as another possible phosphorylation site. This peptide was not phosphorylated under the standard assay conditions. Since a large number of peptides were studied, determination of K, values for each was impractical. Therefore, I& values, determined under constant conditions, were used to compare inhibitor potencies. For determination of I& values, inhibitors were tested in duplicate at four concentrations, including concentrations above and below the I&,, and values were determined graphically. K, values were determined when possible for key inhibitors. Peptide synthesis. KKRPQRATSNVFS-NH2 and all inhibitor peptides were prepared by stepwise solid-phase synthesis on an Applied Biosystems Model 430A peptide synthesizer. The t-BOC protection methodology was used, and final cleavage was with HF/anisole. Peptides were purified by reverse-phase HPLC using a Rainin Dynamax C8 column and a Water’s gradient HPLC system. Peptides were eluted with a gradient of 0.1% trifluoroacetic acid in acetonitrile. Peptides were a single peak by HPLC and were analyzed by FAB mass spectrometry to confirm structure. All peptides described in this work were carboxy-terminal amides. Reagents. [y-“‘P]ATP was obtained from ICN, calmodulin was purchased from Boehringer-Mannheim, and all other chemicals were obtained from Sigma.
RESULTS
Active Site Directed Peptide Inhibitors Table I shows the sequence of MLCK in the region of the calmodulin binding site (residues 480-516) as it was deduced from the sequence of the cloned DNA (8). It has been divided into sections, each containing one or more basic amino acids. These sections have been systematically examined for their importance to enzyme inhibition using the peptides listed in Table I. Assays were per-
POTENT
PEPTIDE
INHIBITORS
OF MYOSIN
LIGHT
CHAIN
KINASE
399
TABLE I Active Site Directed
Peptide Inhibitors
I’eptide number
of Calmodulin-Independent
Residue number
Sequence AKK-LSKD-RM-KKYMA-RRKWQ-KTGHAV-RAIGRLSSMAMI'
A 480 1 2 3 4 r, 6 7 8 9 10 11 12 13 14 15 16 17 18
A
A
A
A
485
490
495
500
Myosin Light Chain Kinase
A 505
A 510
480-516
Ic,,, (PM) Or % inhibition Not made
A 515
KTGHAV-R* KTGHAV-F TGHAV-R RAIGRLSS RLSSMAMI AV-RAIGRLSS KWQ-KTGHAV KWQ-KTGHAV-RAIGRL RRKWQ-KTGHAV RRKWQ-KTGHAV-RAIGRLSS RRKWQ-KTGHAV-FAIAAL KKYMA-RRKWQ-KTGHAV KKYMA-RRKWQ RM-KKYMA-RRKWQ LSKD-RM-KKYMA-RRKWQ AKK-LSKD-RM-KKYMA-RRKWQ AKK-LSKD-RM-KKYMA-RRKWQ-KTGHAV AKK-LSKD-RM-KKYMA
499-505 500-505 505-512 509-516 503-512 496-504 496-510 494-504 494-512 489-504 489-498 487-498 483-498 480-498 480-504 480-493
” Sequence of the calmodulin binding site from myosin light chain kinase from Guerriero rt al. (8). ’ All peptides are carhoxy-terminal amides. (A) Intervals of five amino acid residues. I& values are the average of two or more determinations.
formed using calmodulin-independent MLCK in the absence of calmodulin so that direct enzyme inhibition could be studied without the complicating influence of calmodulin binding. The first peptide synthesized was RRKWQKTGHAV (Peptide 9, Table I) because it showed a high degree of homology to the peptide, KKRPQRATSNVFS-NH2, an excellent substrate for MLCK. Peptide 9 contained all
the characteristics of the myosin light chain recognition site, that is, three basic amino acids separated from the fourth by two amino acids, with glutamine in the second position as it is in the substrate. This peptide will be referred to as the “core peptide.” It was indeed an inhibitor of MLCK, with an IC& of 2 ~.LMunder standard assay conditions. To characterize the mode of inhibition, kinetics studies were performed. At 100 yM ATP, the peptide was apparently competitive with the peptide substrate (Fig. 1A). Secondary plots of apparent K, vs inhibitor concentration were linear, and the apparent K,, an average of three separate determinations, was 2 PM. The equality of IC,,, values and Ki values is a coincidence due to a favorable choice of assay conditions, however, an excellent correlation between IC,, values and Kl’s has been observed for MLCK in this laboratory and others (Ref. (15) and C. Foster, unpublished observations), and IC,,, values will be reported for most peptides. A similar study was performed varying ATP, and a nonlinear double-reciprocal plot was observed at
strongly inhibitory concentrations of the peptide (Fig. 1B). The explanation for this phenomenon was not known, and it was not possible to conclude whether the inhibition was competitive or noncompetitive from these studies. Since the core peptide was an inhibitor as predicted, with a K, of the same order of magnitude as the K,,, for the substrate (myosin light chain, Km = 8.6 PM; Ref. (ll)), the effect of the addition of amino acids in both the carboxy and amino directions was explored. Peptides in the region 499-512 (Peptides l-6) were poor inhibitors of the enzyme and had little effect on inhibitory potency when covalently bound to the core peptide as demonstrated by comparing the IC,, of the core peptide (2.0 PM) with that of Peptide 10 (2.4 PM). Replacing Arg”“” and Arg50g with Phe and Ala, respectively (Peptide II), had no significant effect on inhibitory potency. By contrast, amino acids to the amino-terminal side had a significant influence on inhibitory potency. The arginine pair at position 494-495 was critical to inhibitory activity, since the ICso of the core peptide was at least 150 times lower than that of the peptide missing these two residues (Peptide 7). Adding residues 489-493 (KKYMA) to the core peptide improved the inhibition fivefold (Peptide 12). The peptide KKYMARRKWQ (Peptide 13) was an equal inhibitor to the core peptide, and amino-terminal additions to this peptide further increased the inhibitor potency. RM (487-488) had an ap-
FOSTER
-40
-20
0
20 11
40
[PEPTIDE]
FIG. 1. Kinetics studies with chain kinase as described under tion with the peptide substrate, HIM. For competition with ATP, ( ?) pM. Assays were performed
60
80
100
120
ET AL.
-100
0
(mhl)-1
100 11
IAlP]
200
300
400
(mM)-1
the peptide MLCK 494-504 (Peptide 9). Assays were performed with calmodulin-independent myosin light Materials and Methods. Velocity was measured as nanomoles phosphate incorporated in 30 min. For competiATP was held constant at 100 WM (A). Inhibitor concentrations were 0 (n), 1 (II), 2 (B), 3 (c), 5 (E), and 10 (+) the peptide substrate was held constant at 100 pM (B). Inhibitor concentrations were 0 (O), 1 (+), 3 (P), and 5 in duplicate.
proximately threefold effect (Peptide 14), while LSKD (483-486) increased inhibitor potency approximately an additional sixfold (Peptide 15). AKK (480-482) had no significant effect (Peptide 16), suggesting that this is the amino-terminal end of the pseudosubstrate region. The two halves of the core peptide were investigated individually. KTGHAV (499-504) increased potency 5 fold when added to the carboxy terminal of KKYMARRWQ (489-498) (Peptide 12) and g-fold when added to the longer peptide, AKKLSKDRMKKYMARRKWQ (480-498) (Peptide l?), although the peptide KTGHAVR (Peptide 1) was not an inhibitor on its own. RRKWQ (494-498) had a 67-fold effect when added to the carboxy terminal of AKKLSKDRMKKYMA (480-493) (Peptide 16). Thus, both halves of the core peptide contribute to inhibitory potency, and RRKWQK, or perhaps the two arginines alone (see above), has the greatest individual effect. Combination of all five of the regions shown to contribute to inhibition results in an extremely potent inhibitor with an I&, of 13 nM, AKKLSKDRMKKYMARRKWQKTGHAV (Peptide 17), which was almost lo-fold more potent than any other peptide tested. To prove the point that more than one region of the calmodulin binding sequence can act to inhibit MLCK, Peptide 18 was tested. This peptide does not overlap with the core peptide, and it has little homology with the substrate sequence. Nevertheless, it is an excellent inhibitor with an IC& of 6 PM. Similar to the core peptide, this peptide is apparently competitive with the peptide substrate, with an apparent Kc of 2 PM (data not shown).
Studies on Inhibition
of MLCK
by Peptide 480-504 (17)
Peptide 480-504 (Peptide 17, I& = 13 nM) is a covalent combination of Peptide 9 (I&, = 2 PM) and Peptide 18 (IC& = 6 PM), both of which are apparently competitive with the peptide substrate. The two smaller peptides were tested in combination, to see whether the significant increase in potency observed vs calmodulin-independent MLCK with the combined peptide resulted simply from simultaneous occupation of two inhibitory sites, or whether the covalent connection of the peptides contributed to the inhibition. As can be seen in Fig. 2, a 1:l mixture of the two peptides gave an inhibitor curve that was intermediate between the two individual curves, not shifted to the nanomolar range as is observed with Peptide 17, demonstrating that covalent connection of the peptides is necessary for the improved inhibitory potency. Kinetic studies were performed on Peptide 17. Competition with the peptide substrate yielded nonlinear double-reciprocal plots under the usual assay conditions (not shown). However, in the standard assay, enzyme concentrations are in the low nanomolar range, similar to the peptide concentration. Therefore, kinetics studies were repeated at l/10 the usual enzyme concentration. At these low activity levels, experimental errors were higher, but double-reciprocal plots were linear and showed that this peptide was apparently noncompetitive with the peptide substrate (Fig. 3A). When the peptide substrate was held constant, and ATP was varied, double-reciprocal plots indicated that this peptide was apparently competitive with ATP (Fig. 3B). The explanation for these results is not clear at this time; however, a
POTENT
PEPTIDE
INHIBITORS
100
60
g i z z z
6o
40
20
0 -t.I
r
1 PEPTIDE
10 CONC.
100 (FM)
FIG. 2. Inhibition of myosin light chain kinase by MLCK 494-504 (Peptide 9) and MLCK 480-493 (Peptide 18) alone and in combination. Calmodulin-independent myosin light chain kinase was tested in the absence of inhibitory peptides or in the presence of peptide concentrations as shown (Peptide 9,O; Peptide 18, *). For the combination of peptides (m), Peptides 9 and 18 were mixed at equal concentrations so that the total peptide concentration is shown on the r-axis, and the concentration of each peptide is half the total. Inhibition by the combination is approximately the average of the inhihitions by the two peptides. The IC,, values in this experiment were 2 ELM for Peptide 9, 4.5 @M for Peptide 18, and 2.9 fiM for the combination. The IC,, for MLCK 480-504 (Peptide 17), in which these two peptides are covalently linked, is 13 nhf.
secondary plot of apparent K,,, vs inhibitor concentration was nonlinear, suggesting that the kinetics are complex (data not shown). Similar kinetic results were observed by Ikebe et al. (20). Peptide Inhibitors
Involved
in Calmodulin
Binding
To determine the portions of the calmodulin binding region that are important to calmodulin antagonism, assays were performed using both calmodulin-dependent MLCK and calmodulin-independent MLCK. Earlier studies have shown that active site directed MLCK inhibitors have similar IC& values against both forms of the enzyme. This is true both for antagonists of the peptide substrate (21) and for antagonists of ATP (C. Foster, unpublished observations). However, if an active site directed inhibitor also binds to calmodulin, its ability to inhibit the calmodulin-independent enzyme will be reduced in the presence of calmodulin to the extent that the free inhibitor molecule is removed from the enzyme by binding to calmodulin. In assays for calmodulin-dependent MLCK, the results with inhibitors that interact both with calmodulin and the enzyme are complex. Whether the IC,,, of a calmodulin binding peptide against calmodulin-dependent MLCK is higher or lower
OF MYOSIN
LIGHT
CHAIN
KINASE
401
than that for inhibition of calmodulin-independent MLCK will depend on the relative affinities for the enzyme and calmodulin, as well as the concentrations of each of the components of the reaction mixture. In general, however, a change in the ICSOvalue for a peptide in the presence of calmodulin implies binding to calmodulin. Table II shows a selection of peptides used to determine the regions that are critical to calmodulin antagonism. The 1C5,,for inhibition of calmodulin-independent enzyme and calmodulin-dependent enzyme are shown for comparison. The parent peptide is divided arbitrarily into three regions, each of which contributes to calmodulin binding. The core peptide (Peptide 9) shows essentially no difference in IC& with the two enzymes, suggesting relatively poor binding to calmodulin, at least in the micromolar affinity range. Means (16) has measured the binding of calmodulin and ARRKWQKTG, a peptide similar to the core peptide, and found a binding constant of 700 PM, significantly above the inhibitory concentration (48 PM) and consistent with the idea that similar IC,,, values using the two enzyme preparations indicates poor binding. Larger peptides from the calmodulin binding site have binding affinities in the nanomolar range (16). The importance of the two arginine residues Arg4”4Arg4’” was investigated by testing Peptide 7. Although the ability to inhibit the calmodulin-independent enzyme was low, there was a similar low potency of inhibition against the dependent enzyme, consistent with poor calmodulin binding, even at 300 WM peptide, significantly in excess of the 200 nM calmodulin concentration. When additions were made to the core peptide in the carboxy-terminal direction, however, a significant improvement in calmodulin binding was observed. Peptides 10 and 20 were dramatically better inhibitors of the calmodulin-independent enzyme, and this improved inhibitory activity seemed to require basic amino acids, since the Peptide 11 with Arg5”5 replaced by phenylalanine and Arg50g replaced by alanine had a lo-fold higher ICSOvalue. On the amino-terminal side, Peptide 18 was a poor calmodulin binder on its own, but in combinat.ion with RRKWQ or the core peptide (Peptides 16 and 17), the resulting peptides showed a significant shift in their ability to inhibit calmodulin-dependent MLCK. In these cases, the IC&‘s were actually higher in the presence of calmodulin, and dependent on the calmodulin concentration, although they were of much smaller magnitude than the shifts for Peptides 10 and 20. These results are in general agreement with actual binding measurements (16). These studies led to the question of whether the mechanism of alteration of ICSo values was purely due to calmodulin binding by the peptides. Results with Peptides 10 and 20 suggest that the mechanism is solely by binding to and antagonism of calmodulin, since the I&, val-
402
FOSTER
ET AL.
C
0 80
-60
-40
-20
0
l/(Kemp)
20
40
60
80
100
120
-100
(mM)-1
0
100
200
11 ATP
(mM)-1
300
400
FIG. 3. Kinetics studies with MLCK 480-504 (Peptide 17). Assays were performed with calmodulin-independent myosin light chain kinase as described under Materials and Methods. Velocity was measured as nanomoles phosphate incorporated in 30 min. For competition with the peptide substrate, ATP was held constant at 100 FM, and enzyme concentration was lo-fold lower than ordinarily used (A). Inhibitor concentrations were 0 (O), 10 (+), and (B) 30 nM. For competition with ATP, the peptide substrate was held constant at 100 PM (B). Inhibitor concentrations were 0 (O), 15 (+), 20 (W), and (c‘) 25 nM.
ues for the calmodulin-dependent enzyme are 16 and 14 nM, respectively, concentrations where there would be no inhibition at all of calmodulin-independent MLCK. Peptide 17 was tested further, since it was the most potent inhibitor in the absence of calmodulin, and there-
TABLE
fore the I&, values in the presence of calmodulin were in a range where there would be a direct effect on calmodulin-independent MLCK. An experiment was performed using calmodulin-independent MLCK in the presence of calmodulin and calcium. Since calmodulin
II
Peptide Inhibitors of Myosin Light Chain Kinase from the Calmodulin Binding Site I&, Peptide number
Residue number
Sequence AKK-LSKD-RM-KKYMA-RRKWQ-KTGHAV-RAIGRLSSMAMIn
A A 480 485 9 7 19
12 10 20 11 8
21 17 18 15 16
A 490
A 495
A 500
CaM-Independent MLCK
480~516
Not made
494-504 496-504
17%at 300 GM
.
or % mhibition
CaM-Dependent MLCK
A A A 505 510 515
RRKWQKTGHAV KWQKTGHAV RRKAAKTGHAV KKYMA-RRKWQKTGHAV RRKWQKTGHAV-RAIGRLSS RRKWQKTGHAV-RAIGRL RRKWQKTGHAV-FAIAAL KWQKTGHAV-RAIGRL KTGHAV-RAIGRL AKKLSKDRMKKYMA-RRKWQKTGHAV AKKLSKDRMKKYMA LSKDRMKKYMA-RRKWQ AKKLSKDRMKKYMA-RRKWQ
’ Sequence of the calmodulin * 20 nM CaM. ’ 25 nM CaM. d 10 nM CaM.
( fiM)
2.0
2.2b
110
30%at 300pM* 28%at 100 pM*
489-504 494-512 494-510
0.4 2.4 5.0
0.9’ 0.016* o.014d
1.0
0.17'
496-510 499-510
287
480-504 480-493 483-498 480-498
binding site from myosin light chain kinase from Guerriero
16% at 300 0.013 6.0
GM
0.1 0.09
et al. (8). (A) Intervals
0.17’ 150d 0.21’, 0.105b 13.5’ 0.4d, 4.0h Oxd,
1.5*
of five amino acid residues.
POTENT TABLE
PEPTIDE
INHIBITORS
III
Comparison of the ICso Values for Peptide 17 in the Presence and Absence of Calmodulin 1% (nM) Enzyme type CaM-independent CaM-dependent
0 CaM 13 NT
25 nM
CaM
25 21
200 nM CaM 75 105
has no direct effect on the activity of this enzyme, any alterations in the I&, value of the peptide must be due to binding of the peptide to calmodulin, thereby preventing the peptide from inhibiting the enzyme. Table III shows the results of this experiment and compares it to a similar experiment using calmodulin-dependent enzyme. The ICsOvalues are similar for both enzyme preparations at the same calmodulin concentrations, making it unnecessary to invoke any mechanism other than binding of the peptide to calmodulin to explain the effects. In addition, the shift in IC,,, values in the presence of calmodulin occurred only when calcium was present, indicating that binding of the peptide to calmodulin required calcium. Note that in these experiments the peptide concentration is less than or approximately equal to the calmodulin concentration, and since the affinity of MLCK for Ca’+-calmodulin is high (& = 1 nM, (22)), adequate calmodulin is probably present to activate the enzyme, despite binding of the peptide. DISCUSSION The amino acid sequence of chicken gizzard myosin light chain kinase contains a region with homology to the substrate. Synthetic peptides mimicking this region of MLCK inhibit the catalytic activity of the enzyme directly and also bind to and antagonize calmodulin. By using results with both calmodulin-dependent and calmodulin-independent smooth muscle myosin light chain kinase and a series of synthetic peptides based on the enzyme sequence, regions of the MLCK sequence critical to direct inhibition and calmodulin binding have been mapped. Twenty-two amino acids, MLCK (483504) contribute to both types of inhibition, and five separate peptide sequences of 4-6 amino acids have been identified that contribute individually to direct inhibition, as well as being involved in calmodulin binding. These are LSKD (483-486), RM (487-488), KKYMA (489-493), RRKWQ (494-498), and KTGHAV (499504). Residues MLCK (505-512), by contrast, are critical to calmodulin binding but are unimportant to direct inhibition of the enzyme. Previous studies on gizzard myosin light chain kinase (4-6) have demonstrated that calmodulin binding can be
OF MYOSIN
LIGHT
CHAIN
403
KINASE
destroyed by proteolysis while autoinhibition is retained. A study by peptide sequencing (10) has shown that Arg505 is the carboxy-terminal amino acid of an inactive fragment of MLCK that has lost the ability to bind calmodulin. Thus, it seems that loss of MLCK (505-512), a peptide shown in these studies to be critical only to calmodulin binding and not to active site inhibition, is sufficient to eliminate effective calmodulin binding in proteolyzed enzyme, even though segments of the sequence that remain after proteolysis bind calmodulin when they are tested as independent peptides. While results obtained with isolated peptides must be interpreted with caution since conformation may not be retained, it is possible that these peptide regions do indeed contribute to calmodulin binding in the nonproteolyzed enzyme. In another report (6) an inactive fragment was sequenced and found to terminate at Arg4”. In our studies, MLCK (480-493) was shown to be an active site directed inhibitor with little calmodulin binding activity. Thus, a proteolyzed enzyme terminating at residue 493 would be expected to have lost calmodulin binding activity while retaining autoinhibition, consistent with the conclusions of Ikebe et al. (6). This is of interest, since MLCK (480-493) does not include the putative pseudosubstrate sequence (9). However, while Ikebe concludes that the inhibitory sequence includes only residues 474-490, the inhibitory potency of peptides including residues 493504 (note especially Peptide 17) strongly suggests the additional role of these amino acids in autoinhibition. The carboxy terminal of an active, calmodulin-independent form of the enzyme has been identified as Lys4”’ (6). This is consistent with the results shown here, since all of the predicted inhibitory peptides are missing from this fragment. The results in this study expand on earlier studies by providing a detailed description of the amino acids that are important for calmodulin binding and autoinhibition. The complex overlapping of the pseudosubstrate and calmodulin binding site and the synergistic participation of multiple, linearly arranged regions of the enzyme in regulation show how elegantly and economically the enzyme myosin light chain kinase has evolved. ACKNOWLEDGMENTS The authors thank Dr. Michael Czarniecki and Dr. Barry Pitts for their helpful suggestions and Dr. David Hartshorne for his excellent advice on the research.
REFERENCES 1. Kamm, K. E., and Stull, J. T. (1985) Annu. cd.
Reu. Pharmacol.
TOG-
25,593-620.
2. Walsh, M. P., Dahrowska, R., Hinkins, S., and Hartshorne, (1982)Biochemistry21,1919-1925. 3. Foyt, H. L., Guerriero, V., and Means, A. R. (1985) J. Bid. 260,1765-7774.
D. .J. Chem.
404
FOSTER
ET AL.
11. Kemp, B. E., Pearson, R. B., and House, C. (1983) Proc. Nat. Acad. Sci. USA 80,7471-7475.
13. Pearson, R. B., Misconi, L. Y., and Kemp, B. E. (1986) J. Biol. Chem. 261,25-27. 14. Pearson, R. B., Floyd, D. M., Hunt, J. T., Lee, V. G., and Kemp, B. E. (1988) Arch. Biochem. Biophys. 260,37-44. 15. Hunt, J. T., Floyd, D. M., Lee, V. G., Little, D. K., and Moreland, S. (1989) Biochem. J. 257,73-78. 16. Means, A. R. (1988) Recent Prog. Horm. Res. 44,223-262. 17. Foster, C. J., and Gaeta, F. C. A. (1988) Biophys. J. 53,182a. 18. Ngai, P. J., Carruthers, C. A., and Walsh, M. P. (1984) Biochem. J. 2 l&863-870. 19. Tanaka, T., Naka, M., and Hidaka, H. (1980) Biochem. Biophys. Res. Commun. 92,313-318. 20. Ikebe, M., Kemp, B. E., and Hartshorne, D. J. (1988) Biophys. J. 53,182a. 21. Gaeta, F. C. A., Lehman De Gaeta, L., Or, Y. S., Foster, C. J., and Czarniecki, M. (1990) J. Med. Chem. 33,964-972.
12. Kemp, B. E., and Pearson, R. B. (1985) J. Biol. Chem. 260,33553359.
22. Walsh, M. P., Hinkins, S., Flink, (1982) Biochemistry 21,6890-6896.
4. Foster, C. J., Van Fleet, M., and Marshak, A. M. (1986) Arch. Biothem. Biophys. 25 1,616-623. 5. Ikebe, M., Stepinska, M., Kemp, B., Means, A. R., and Hartshorne, D. J. (1987) J. Biol. Chem. 262, 13,828-13,834. 6. Ikebe, M., Maruta, S., and Reardon, S. (1989) J. Biol. Chem. 264, 6967-6971. 7. Lukas, T. J., Burgess, W. H., Prendergast, F. G., Lau, W., and Watterson, D. M. (1986) Biochemistry 25,1458-1464. 8. Guerriero, V., Russo, M., Olson, N., Putkey, A. R. (1986) Biochemistry 25,8372-8381.
J. A., and Means,
9. Kemp, B. E., Pearson, R. B., Guerriero, V., Bagchi, Means, A. R. (1987) J. Biol. Chem. 262,2542-2548.
I. C., and
10. Pearson, R. B., Wettenhall, R. E. H., Means, A. R., Hartshorne, D. J., and Kemp, B. E. (1988) Science 241,970-973.
I. L., and Hartshorne,
D. J.
