(Received for publication, December 24, 1996, and in revised form, February 19, 1997)
From Searle Discovery Research, Monsanto Company, St. Louis, Missouri 63198 and the § Department of Molecular Biology and Pharmacology, Washington University School of Medicine, St. Louis, Missouri 63110
Candida albicans produces a single
myristoyl-CoA:protein N-myristoyltransferase (Nmt) that is
essential for its viability. An ADP-ribosylation factor (Arf) is
included among the few cellular protein substrates of this enzyme. An
octapeptide (GLYASKLS-NH2) derived from a N-terminal Arf
sequence was used as the starting point to identify elements critical
for recognition by the acyltransferases's peptide-binding site.
In vitro kinetic studies, employing purified Nmt and a
panel of peptides with single Ala substitutions at each position of
GLYASKLS-NH2, established that its Gly1,
Ser5, and Lys6 residues play predominant roles
in binding. LYASKLS-NH2 was found to be an
inhibitor competitive for peptide (Ki = 15.3 ± 6.4 µM) and noncompetitive for myristoyl-CoA
(Ki = 31.2 ± 0.7 µM). A survey
of 26 derivatives of this inhibitor, representing (i) a complete
alanine scan, (ii) progressive C-terminal truncations, and (iii)
manipulation of the physical-chemical properties of its residues 1, 5, and 6, confirmed the important stereochemical requirements for the
N-terminal amine, the -hydroxyl of Ser5, and the
-amino group of Lys6. Remarkably, replacement of the the
N-terminal tetrapeptide of ALYASKLS-NH2 with an
11-aminoundecanoyl group produced a competitive inhibitor,
11-aminoundecanoyl-SKLS-NH2, that was 38-fold more potent
(Ki = 0.40 ± 0.03 µM) than the
starting octapeptide. Removing the primary amine
(undecanoyl-SKLS-NH2), or replacing it with a methyl group
(dodecanoyl-SKLS-NH2), resulted in 26- and 34-fold
increases in IC50, confirming the important contribution of
the amine to recognition. Removal of LeuSer from the C terminus (11-aminoundecanoyl-SK-NH2) yielded a competitive dipeptide
inhibitor with a Ki (11.7 ± 0.4 µM) equivalent to that of the starting octapeptide,
ALYASKLS-NH2. Substitution of Ser with homoserine, cis-4-hydroxyproline, or tyrosine reduces potency by
3-70-fold, emphasizing the requirement for proper presentation of the
hydroxyl group in the dipeptide inhibitor. Substituting D-
for L-Lys decreases its inhibitory activity >100-fold,
while deletion of the
-amino group (Nle) or masking its charge
(
-N-acetyl-lysine) produces 4-7-fold attenuations.
L-His, but not its D-isomer, can fully substitute for L-Lys, producing a competitive dipeptide
inhibitor with similar potency (Ki = 11.9 ± 1.0 µM).
11-Aminoundecanoyl-S-NH2 and
11-aminoundecanoyl-S-NH2 establish that a
simple alkyl backbone can maintain an appropriate distance between
three elements critical for recognition by the fungal enzyme's
peptide-binding site: a simple
-terminal amino group, a
-hydroxyl, and an
-amino group or an imidazole. These compounds
contain one peptide bond and two chiral centers, suggesting that it may
be feasible to incorporate these elements of recognition, or
functionally equivalent mimics, into a fully de-peptidized Nmt
inhibitor.
Candida albicans is a dimorphic, asexual fungus. Ninety percent of patients with acquired immune deficiency syndrome develop C. albicans infections at some point during the course of their disease (1). The few fungicidal drugs currently available have side effects that limit their therapeutic efficacy (2). Long term suppressive or prophylactic therapy with currently effective fungistatic triazoles (3) may hasten the development of drug-resistant strains.
Several observations indicate that myristoyl-CoA:protein
N-myristoyltransferase (Nmt)1
may be a good target for the development of a new class of fungicidal agents. C. albicans contains a single NMT gene
(4). Metabolic labeling studies indicate that this Nmt covalently
attaches [3H]myristate (C14:0) to the N-terminal Gly
residues of fewer than 10 cellular proteins during exponential growth
on rich media (5). These Nmt substrates include an ADP-ribosylation
factor (Arf) (5, 6) and Cga, a protein of unknown function that can
complement the growth arrest and mating defects found in strains of
Saccharomyces cerevisiae containing a null allele of its G
protein subunit gene, GPA1 (7). Genetic studies have
shown that Nmt is essential for viability. A strain of C. albicans was constructed in which one copy of its NMT
gene was deleted. A Gly447
Asp mutation was
introduced into the remaining NMT allele (designated nmt447D). This amino acid substitution produces a marked
reduction in the enzyme's catalytic efficiency at 24 and 37°C,
reflecting, in part, a reduction in its affinity for myristoyl-CoA (8). Unlike isogenic NMT/NMT or NMT/nmt
strains,
nmt
/nmt447D cells require myristate for their growth in
rich media at 24 and 37°C. Removing myristate results in cell death
(8). This lethality correlates with levels of cellular protein
N-myristoylation. Arf is completely
N-myristoylated in the NMT/nmt
strain, whether it is grown in the presence or absence of myristate at 24 or 37°C. When nmt
/nmt447D cells are grown at 24°C in media
supplemented with myristate, most (>75%) of cellular Arf is
N-myristoylated. However, 2 or 4 h after withdrawal of
myristate, the level of N-myristoylated Arf falls to <50%
of total cellular Arf (9). Attenutation of Nmt activity also
compromises the ability of these organisms to survive in vivo.
The NMT/
nmt strain produces 100% lethality within 7 days after intravenous administration into a group of immunosuppressed
mice. However, when an identical number of nmt
/nmt447G
cells are infused into immunosuppressed animals, no death is
observed even after 21 days (8).
Protein N-myristoylation occurs co-translationally and appears to be irreversible (reviewed in Ref. 10). Nmt has a preferred order reaction mechanism (11-13). The apoenzyme first binds myristoyl-CoA to form a myristoyl-CoA·Nmt binary complex which is competent for peptide binding. Following generation of a myristoyl-CoA·Nmt·peptide ternary complex and catalytic transfer of myristate to the glycyl amine, the CoA and myristoylpeptide products are released. In vitro kinetic analyses using a large panel of myristoyl-CoA analogs and purified human and fungal Nmts indicate that the requirements for molecular recognition at the enzymes' acyl-CoA-binding site have been highly conserved throughout the course of eukaryotic evolution (e.g. Refs. 4 and 14). However, differences in peptide substrate specificities have been noted among orthologous Nmts in vitro (4, 9, 12, 15, 16). These differences can be exploited to develop species-specific enzyme inhibitors.
To design such inhibitors of C. albicans Nmt, the subsite specificity of the enzyme for peptide ligands needs to be defined. To begin this process, we have systematically replaced each amino acid residue in a known, high affinity, Arf-derived octapeptide substrate with alanine. Having identified the importance of residues 1, 5, and 6 in molecular recognition, we then synthesized an additional series of peptides and de-peptidized analogs containing various substituents at these critical positions.
Protected amino acids were obtained from Applied Biosystems, Inc. or Bachem California. tert-Butyloxycarbonyl (t-Boc)-protected 11-aminoundecanoic acid was purchased from Omni Biochem. t-Boc-4-aminobutanoic, t-Boc-5-aminopentanoic, t-Boc-6-aminohexanoic, and t-Boc-8-aminooctanoic acids were synthesized using the method of Tarbell et al. (17).
With the exception of position 1 derivatives of ALYASKLS-NH2 (see below), all peptides were produced using t-Boc-protected amino acids and an Applied Biosystems Model 470A automated synthesizer. Peptides were cleaved from the solid support resin and de-protected by treatment with anhydrous HF:anisole (10:1) for 60 min at 5°C. After evaporation of HF, the peptide resin was extracted with ether and then 10% acetic acid. The acid washings were combined, diluted with water, and lyophilized. Peptides were subsequently purified by reverse phase HPLC and characterized by HPLC, amino acid analysis, and fast atom bombardment mass spectrometry. All peptides used in this study were >95% pure.
Peptides incorporating substitutions in position 1 of ALYASKLS-NH2 were prepared by the standard orthogonal solid-phase peptide synthesis protocol (18) using 9-fluorenylmethoxycarbonyl-protected amino acids. After de-protection with piperdine, washing and drying of the resin, the peptides were cleaved using 95% trifluoroacetic acid containing trace thiol for 2 h. The peptide product was collected by filtration, diluted, and lyophilized, and the product purified by HPLC. Characterization and purity of the position 1 peptides were similar to that of peptides produced by automated t-boc synthesis.
Determination of Peptide Kinetic ParametersC. albicans Nmt was expressed in Escherichia coli strain JM101 and purified to apparent homogeneity using procedures described in Ref. 19. A coupled in vitro Nmt assay was utilized to determine peptide kinetic parameters (12). In a typical reaction, myristoyl-CoA was generated using [3H]myristate, CoA, and Pseudomonas acyl-CoA synthetase (12). The final reaction mixture (110 µl) contained variable amounts of peptide, 0.23 or 0.3 µM [3H]myristoyl-CoA, and purified C. albicans Nmt (10-100 ng). After a 10-min incubation at 24°C, the [3H]myristoylpeptide product was purified by reverse phase HPLC (12) and quantitated using an in-line flow scintillation detector (Radiomatic A250, Packard). Km and Vmax were calculated using nonlinear regression analysis of the initial velocities with the program k·cat (version 1.55, Biometallics, Princeton, NJ). All experiments were performed in triplicate and assays were repeated on at least two separate occasions.
Determination of IC50 ValuesAssays (110 µl) contained variable amounts of inhibitor, 0.11 nmol of purified [3H]myristoyl-CoA (1 µCi, 9.09 Ci/mmol), 2.2 nmol of GNAASARR-NH2, and 7-12 ng of purified C. albicans Nmt in reaction buffer (buffer = 200 mM HEPES, pH 7.4, 2 mM dithiothreitol, 0.2 mM EGTA). Incubations were allowed to proceed for 10 min at 24°C. Assays were performed in duplicate and repeated at least once.
Determination of Ki ValuesKi values for inhibitors were determined using methods described earlier (16). Briefly, competition against peptide was evaluated by varying the concentrations of two previously characterized peptide substrates, GNAASARR-NH2 or GARASVLS-NH2 (19), between 5 and 80 and 5 and 40 µM, respectively, and fixing the concentration of purified[3H]myristoyl-CoA at 1 µM. Competition against myristoyl-CoA was evaluated by varying the concentration of the acyl-CoA between 0.165 and 2.64 µM and fixing the concentration of GNAASARR-NH2 at 20 µM. Reactions were begun by adding the purified fungal Nmt to a final concentration of 1.2-2.2 nM. The acylpeptide product was quantitated as described above. Data are reported as apparent inhibition constants and were calculated by nonlinear regression analysis of double-reciprocal (Lineweaver-Burk) plots using k·cat. Both competitive and noncompetitive models were tested.
Arf proteins are produced in a
wide variety of species and many are known to be substrates for Nmts
in vivo (e.g. Refs. 20-24). An octapeptide
representing the N terminus of C. albicans Arf (GLTISKLF-NH2) is a substrate for purified C. albicans Nmt in vitro (Km = 0.6 µM; Vmax = 48,000 pmol/min/mg
enzyme). GLYASKLF-NH2 representing the N terminus of
S. cerevisiae Arf2p is also accommodated by C. albicans Nmt (Km = 0.4 µM). Previous studies had shown that a derivative of
GLYASKLF-NH2 containing a Phe8 Ser
substitution (GLYASKL-NH2) is a high affinity
substrate for S. cerevisiae Nmt1p (Km = 0.07 µM; Ref. 12). Subsequent replacement of its
Gly1 with Ala (LYASKLS-NH2; Ref.
25) yielded the first known high affinity competitive peptide inhibitor
of an Nmt (Ki = 5 µM with purified
S. cerevisiae Nmt1p). Based on these observations, we chose
GLYASKLS-NH2 to begin our identification of functional groups required by the binary C. albicans
Nmt·myristoyl-CoA complex for recognition of its peptide ligands.
Table
I shows the kinetic effects of replacing residues in
GLYASKLS-NH2 with Ala. Substitution of Ala at position 1 (LYASKLS-NH2) represents addition of a methyl
group to the -carbon of Gly. The result is to transform a substrate
to an inhibitor (IC50 = 29 ± 4 µM; see
Table I). Double-reciprocal plots established that the inhibition was
competitive versus peptide (Ki = 15.3 ± 6.4 µM) and noncompetitive versus
myristoyl-CoA (Ki = 31.2 ± 0.7 µM) (Fig. 1, A and
B). LYASKLS-NH2 does not serve as a
substrate: experiments employing a wide range of Nmt,
[3H]myristoyl-CoA, and peptide concentrations failed to
yield detectable amounts of product
([3H]myristoyl-LYASKLS; see legend to Table
I).
|
Substitution of an Ala for the Leu2 in GLYASKLS-NH2 replaces an isobutyl group with a methyl. There was a modest 3-fold reduction in Km which was offset by a 4-fold drop in Vmax. Substitution of an Ala for its Tyr3 replaces a p-hydroxybenzyl with a methyl and had insignificant effects on these kinetic parameters (Table I).
The importance of the -methyl group of Ala4 was explored
by placing Gly at this position (GLYSKLS-NH2).
Removal of the methyl group results in a 15-fold increase in
Km and a 4-fold augmentation of
Vmax (Table I).
Replacement of Ser5 with Ala can be viewed as a substitution of a hydrogen for a hydroxyl. The effect of this substitution is much more dramatic than any of the alterations at positions 1-4: Km increases by at least 10,000-fold and Vmax decreases 6-fold, resulting in a catalytic efficiency (Vmax/Km) which is 66,000-fold lower than that of the parental octapeptide, GLYASKLS-NH2 (Table I). The aminoalkyl side chain in the adjacent Lys6 also appears to play an important role in recognition: replacement of this Lys with Ala results in a 236-fold increase in Km (Table I).
As in the case of Leu2, replacement of Leu7 with Ala produces only minor (2-3-fold) effects on Km and Vmax (Table I). Finally, unlike the dependence on the hydroxyl of Ser at position 5, replacement of Ser with Ala at position 8 produces no significant change in Km and a less than 2-fold alteration in Vmax.
The results of the scanning alanine mutagenesis suggest that C. albicans Nmt recognizes the Arf substrate based in large part on the nature of functional groups present at residues 1, 5, and 6. This hypothesis was supported by an alanine scan of ALYASKLS-NH2 (Table I). Substitution of Leu2 increases inhibitory potency by 7-fold while placing Ala at positions 3 and 8 produces no significant change in IC50. Introducing Ala at positions 4 and 7 results in ~10-fold reductions in inhibitory activity. However, replacement of Ser5 and Lys6 changes the IC50 from 29 ± 4 to 1520 ± 50 µM and 2680 ± 50 µM, respectively (50- and 90-fold increases; Table I).
C-terminal truncations of the parental substrate (GLYASKLS-NH2) and inhibitor (ALYASKLS-NH2) also revealed the important contribution of Lys6. Comparison of GLYASKL-NH2, GLYASK-NH2, and GLYAS-NH2 (Table I) established that (i) deletion of Ser8 has no significant effect on kinetic parameters; (ii) deletion of Leu7-Ser8 produces a modest 7-fold increase in Km and a 5-fold augmentation of Vmax, and (iii) removal of Lys6-Leu7-Ser8 results in barely detectable amounts of product, even when enzyme and peptide concentrations are increased 10-100-fold over that used for assaying the other peptides. Similarly, loss of Leu7-Ser8 from ALYASKLS-NH2 only produces a 10-fold reduction in IC50 (Table I).
Further Characterization of Structure-Activity Relationships at Positions 1, 5, and 6 of ALYASKLS-NH2We reasoned that if positions 1, 5, and 6 of ALYASKLS-NH2 provide essential elements for recognition by Nmt, then it might be possible to de-peptidize this parental inhibitor by removing the nonessential residues and replacing them with hydrocarbon linkers. However, before attempting such an exercise, we further defined structure-activity relationships at these three positions to obtain additional information about the spectrum of functional groups that might be incorporated into de-peptidized inhibitors.
Options for replacing the N-terminal amino acid were explored by
examining the effects of (i) adding different carbon substituents; (ii) removing, substituting for, or masking the nitrogen acceptor; and
(iii) altering the distance between this nitrogen and the Ser-Lys
dipeptide. The results are presented in Table II. A
comparison of L- and D-Ala revealed that
substitution of the D-isomer abolishes inhibitory activity
(IC50 >1000 versus 29 ± 4 µM). This finding suggests that either an
-amino or
methyl group, or both, in the R-configuration prevents
binding to the binary myristoyl-CoA·Nmt complex. Extension of the
side chain at position 1 by substituting the methyl with either a
propargyl or a propylguanidino group yielded compounds
(S-propargylglycine-LYASKLS-NH2 and
N-L-arginyl-LYASKS-NH2) that have
similar inhibitory activities (IC50 = 51 ± 4 and
54 ± 6 µM, respectively) as
L-alanyl-LYASKLS-NH2. These results indicate that the peptide recognition site of the myristoyl-CoA·Nmt complex is
able to accommodate marked variations in the polarity and steric bulk,
but not in the stereochemistry, of the
carbon side chain of residue
1.
|
When the primary amine of GLYASKLS-NH2 is replaced with a hydroxyl (N-hydroxyacetyl-LYASKLS-NH2), capped with an N-methyl group (N-sarcosyl-LYASKLS-NH2), or its charge neutralized by acetylation (N-acetylglycyl-LYASKLS-NH2), there are no remarkable effects on inhibitory activity (IC50 changes <3-fold; Table II). When the amine nitrogen is removed entirely (N-acetyl-LYASKLS-NH2), the IC50 increases 8-fold (225 ± 2 µM). The hydroxyl, N-methyl, and N-acetylglycyl substituents all have heteroatom H-bond donor and acceptor groups that may mimic the amine present in GLYASKSL-NH2, but which is absent from N-acetyl-LYASKLS-NH2.
The distance between the terminal H-bond donor or acceptor group and
the interior Ser-Lys dipeptide is sensed by the enzyme. N-Acetyl-LYASKLS-NH2 can be viewed as an analog
lacking a terminal amine but having a heteroatom more proximal to the
Ser-Lys dipeptide than ALYASKLS-NH2. In addition, altering
the distance by substituting Ala with N--alanyl
(IC50 = 108 ± 1 µM) or
N-methylcarbamoyl (IC50 = 82 ± 3 µM) also compromises inhibitory activity (3-4-fold
compared with their ALYASKLS-NH2 parent).
Because of the potential to build product-like or transition state-like
binding modes into a de-peptidized derivative, we also added two groups
that extend from the terminal nitrogen: myristate (yielding
the product N-myristoylglycyl-LYASKLS-NH2) and
an N-2,2-difluoro-3-ketohexadecanoyl group (yielding a
possible tetrahedral transition-state mimic; cf. Ref. 26).
These compounds are only marginally better inhibitors (IC50 = 5 ± 1 and 17 ± 0.3 µM, respectively) than
ALYASKLS-NH2 (29 ± 4 µM) (Table
II).
Changing the stereochemistry of Ser5 or O-methylation markedly attenuates the inhibitory activity of ALYASKLS-NH2 (IC50 increases to 920 ± 110 and 522 ± 30 µM; Table III). Neither the isosteric hydroxyl in Thr nor the OH group of Asp effectively substitutes for the side chain hydroxyl in L-Ser (Table III).
|
Substituting D-Lys at position 6 of ALYASKLS-NH2 is deleterious (8-fold increase in IC50 to 220 ± 30 µM) but is not as damaging as changing the stereochemistry at Ser5 (30-fold increase; Table III). Arg6 is inferior to Lys6 (4-fold increase in IC50) but is slightly better tolerated than the shortened aminoalkyl side chain of Orn (8-fold increase). Comparison of Nle6 (10-fold increase in IC50) and Lys6 emphasizes the importance of a primary amine at this position (Table III).
De-peptidization of ALYASKLS-NH2The alanine scan of ALYASKLS-NH2 established that amino acids 2 or 3 could be substituted with modest or no effects on inhibitory potency. As a prelude to de-peptidization, both Leu2 and Tyr3 of ALYASKLS-NH2 were replaced by Ala (AASKLS-NH2). This "simplified" compound was 7-fold more potent an inhibitor (IC50 = 4 ± 1 µM) than its parent (Table IV).
|
Our initial approach for de-peptidization was to replace the N-terminal two residues of ALYASKLS-NH2 with an aminoalkyl moiety of comparable length. Remarkably, when aminopentanoyl was used to replace the 6 backbone atoms of Ala1-Leu2, the resulting compound, 5-aminopentanoyl-YASKLS-NH2, was 180-fold more potent than the starting octapeptide (IC50 = 0.16 ± 0.01 versus 29 ± 4 µM; Table IV). Further kinetic analysis confirmed that it was a competitive inhibitor for peptide (Ki = 0.133 ± 0.019 µM). Experiments employing a large (2,000-fold) excess of Nmt over that used in standard assays, a 10-fold increase in incubation time, and up to 1000 µM 5-aminopentanoyl-YASKLS-NH2 failed to yield detectable amounts of [3H]myristoyl-peptide analog, thereby establishing that this compound does not serve as a Nmt substrate.
There were only minimal (3-fold) increases in IC50 when the chain length of the aminoalkyl group was adjusted by adding or subtracting one methylene (4-aminobutanoyl- and 6-aminohexanoyl-YASKLS-NH2, respectively; Table IV). This indicates that (i) the fungal enzyme is able to both measure and tolerate one atom variation in the length of the flexible alkyl chain, and (ii) a 6-atom linker is the optimal length for replacement of the N-terminal dipeptide.
As noted above, deletion of Ser8 in ALYASKLS-NH2 produces no effect on its inhibitory activity while deletion of Leu7-Ser8 results in 10-fold reduction in potency (cf. Table I). Virtually identical results were noted when these residues were deleted from 5-aminopentanoyl-YASKLS-NH2 (Table IV). Further C-terminal truncation, i.e. removing Lys from 5-aminopentanoyl-YASK-NH2, produced an inactive tripeptide (IC50 >1000 µM) with only two of the three postulated essential elements for recognition.
When the -methyl group of Ala4 was removed in
ALYASKLS-NH2 and GLYASKLS-NH2 by Gly
substitution, the IC50 increased 10-fold and the
Km rose 15-fold, respectively. Therefore, before proceeding with further de-peptidization, we explored the importance of
Ala in the minimal fully active, 5-aminopentanoyl-containing compound
(5-aminopentanoyl-YASKL-NH2). Sixteen derivatives were examined (Table IV). The methyl group of Ala "remains" important: replacement with Gly results in a 76-fold reduction in inhibitory activity (IC50 = 32 ± 4 versus 0.42 µM). The stereochemistry of this methyl is critical since
substitution of L-Ala with D-Ala produces a
>2000-fold reduction in potency. Moving the methyl group from the
-carbon to the amide nitrogen (alanine
sarcosine) is poorly
tolerated (17-fold reduction). Appending a hydroxyl to the methyl group
(Ser) produces a 90-fold reduction in potency. Amino acids with
branched, moderately bulky side chains (Val, Ile, and Leu) are
equivalent to or slightly better than Ala (IC50 = 0.17-0.88 µM). Substitutents with greater bulk,
phenylglycine, t-butylalanine, phenylalanine,
cyclohexylalanine, and norleucine, are deleterious
(IC50 values increase 6-145-fold relative to
5-aminopentanoyl-YASKL-NH2). Norvaline and propargylglycine
contain slightly smaller unbranched side chains that are tolerated by
Nmt (IC50 = 0.34 ± 0.17 and 1.25 ± 0.07 µM, respectively). The geminal di-substituted
aminoisobutyrate and cyclopropylglycine residues both have a pronounced
negative impact on inhibition (IC50 = 45 and 100 µM), further confirming that substitutions at the
-carbon corresponding to the R-configuration have a
pronounced negative effect on inhibition.
Although Nmt is sensitive to the physical-chemical properties of residues occupying the place of Ala in these 5-aminopentanoyl-containing analogs, we found that this Ala and its adjacent Tyr could be replaced with an alkyl moiety of comparable length, producing an inhibitor, 11-aminoundecanoyl-SKLS-NH2, that was still 60-fold more potent (IC50 = 0.49 ± 0.04 µM) than the starting ALYASKLS-NH2 octapeptide. Like ALYA-NH2 and 5-aminopentanoyl-YA-NH2, 11-aminoundecanoyl--NH2 exhibits a competitive pattern of inhibition versus peptide (Ki = 0.40 ± 0.03 µM).
Structure-Activity Studies of 11-Aminoundecanoyl-SKLS-NH2This synthetic
organic-peptide hybrid retains the three critical elements of
recognition defined from the original alanine scan and truncations of
ALYASKLS-NH2: an N-terminal primary amine, a
Ser5 hydroxyl, and an -amino group at position 6. Removing the primary amine from
11-aminoundecanoyl-SKLS-NH2
(undecanoyl-SKLS-NH2), or replacing it with a methyl group
(dodecanoyl-SKLS-NH2), results in substantial reductions in
inhibitory potency (26- and 34-fold, respectively, as defined by
IC50; Table IV), thereby confirming the importance of this
amine for recognition by Nmt.
Substitution of the N-terminal tetrapeptide of ALYASKLS-NH2 with an 11-aminoundecanoyl group removes three backbone amide bonds. The contribution of these bonds to recognition was evaluated by preparing two compounds, one with an amide bond analogous to that linking Ala1-Leu2 (glycyl-8-aminooctanoyl-SKLS-NH2), the other with a bond analogous to that linking Leu2-Tyr3 (5-aminopentanoyl-5-aminopentanoyl-SKLS-NH2). In both cases, introduction of the amide bonds diminished inhibitory potency relative to 11-aminoundecanoyl-SKLS-NH2 (23- and 5-fold respectively; Table IV). These results further emphasize the surprising nature of the finding that this peptide N-myristoyltransferase seems to "prefer" competitive peptidomimetic inhibitors that lack peptide bonds.
11-Aminoundecanoyl-SK-NH2, A Dipeptide Inhibitor that Retains Critical Elements of RecognitionAs with ALYASKLS-NH2 and 5-aminopentanoyl-YASKLS-NH2, deletion of the C-terminal Ser from 11-aminoundecanoyl-SKLS-NH2 produces only a minimal (<3-fold) effect on its inhibitory activity while deletion of its C-terminal Leu-Ser dipeptide (11-aminoundecanoyl-SK-NH2) results in a more substantial (30-fold) reduction in potency.
11-Aminoundecanoyl-SK-NH2 represents a dipeptide inhibitor (IC50 = 14.5 ± 1.6 µM) that is competitive for peptide (Ki = 11.7 ± 0.4 µM; Fig. 1C). The Ki of this compound is equivalent to the Ki of the starting octapeptide inhibitor, ALYASKLS-NH2 (15.3 ± 6.4 µM).
The contributions of the remaining amino acids in
11-aminoundecanoyl-SK-NH2 to recognition were
explored. Presentation of the hydroxyl group is important: extension of
this OH in the context of homoserine, cis-4-hydroxyproline,
or tyrosine worsens inhibitory potency by 3-70-fold (Table IV). The
stereochemistry of the Lys side chain is also critical: substituting
D-Lys abolishes inhibitory activity (Table IV). Deletion of
the -amino group (Nle) or masking its charge
(
-N-acetyl-lysine) produces more modest attenuation (4-7-fold). In the context of this dipeptide aminoalkyl inhibitor, L-His can substitute for L-Lys, producing an
inhibitor competitive for peptide with similar potency
(Ki = 11.9 ± 1.0 versus 11.7 ± 0.4 µM). Again stereochemistry is important:
D-His is inactive (IC50 > 1000 µM). Masking either of the imidazole nitrogens
(1-methyl-histidine or 3-methyl-hisitidine) also eliminates inhibitory
activity (IC50 > 1000 µM).
p-Aminophenylalanine contains a terminal basic amine: the
length of the side chain is similar to Lys, although the bulk is
greater and the pKa of the amine is lower (~10
versus ~5). It cannot substitute for Lys (IC50 > 1000 µM; Table IV).
Our findings indicate that an
11-aminoundecanoyl backbone can serve to maintain an appropriate
distance between three elements critical for recognition by the
peptide-binding site in the fungal Nmt·myristoyl-CoA binary complex.
These elements include a simple -terminal amino group, a
-hydroxyl, and an
-amino group or an imidazole. Each of these
competitive inhibitors has one peptide bond and two chiral centers.
They are equipotent with the starting peptide inhibitor,
ALYASKLS-NH2, which was derived from a known Nmt substrate
(Arf) and contained 7 peptide bonds and 8 chiral centers.
11-Aminoundecanoyl-S-NH2 and
11-aminoundecanoyl-S-NH2 demonstrate that
this fungal Nmt can be inhibited by simplified organic-peptide hybrids
and suggests that it may be possible to incorporate these elements of
recognition, or functionally equivalent mimics, in a fully
de-peptidized organic molecule.
We are grateful to Mike Jennings for amino acid analysis and Jim Doom for mass spectrometry of peptides and peptidomimetics.