Sage Journals: Discover world-class research

Abstract

The microbial threat to human health is growing due to the dramatic increase in the number of multidrug-resistant organisms. The decline in effective antibiotics available to treat these growing threats has provided greater urgency to the search for new antibiotics. Clearly, new approaches must be developed against novel targets to control these resistant infectious organisms. The screening of low molecular weight compounds against new protein targets provides an opportunity to identify novel inhibitors as starting points for the development of new antibiotics. Custom fragment libraries have been assembled and screened against 3 representative forms of a key enzyme in an essential microbial biosynthetic pathway. Although each of these aspartate semialdehyde dehydrogenases (ASADHs) catalyzes the same reaction and each shares identical active site functional groups, subtle differences in enzyme structures have led to different binding selectivity among the initial hits from these fragment libraries. Amino acid analogues have been identified that show selectivity for either the gram-negative or gram-positive bacterial enzyme forms. A series of benzophenone analogues selectively inhibit the gram-negative ASADH, whereas some haloacids and substituted aromatic acids have been found to inhibit only the fungal form of ASADH. Each of these low molecular weight compounds possesses high ligand binding efficiency for their target enzyme forms. These results support the goal of designing lead compounds that will selectively target ASADHs from different microbial species.

Keywords

fragment library screening kinetic studies aspartate semialdehyde dehydrogenase antibiotic development ligand binding efficiency

Introduction

We are facing a growing threat from infectious organisms that are becoming resistant to even the most recently developed antibiotics that target essential steps in cell wall assembly and protein biosynthesis. Plasmid-encoded resistance to erythromycin and tetracycline was discovered in Enterococcus faecalis in the early 1970s,¹ and high transfer rates of multidrug-resistant plasmids have been observed between different bacterial species. Enterococci are now the leading cause of antibiotic-resistant, hospital-acquired infections. Encounters of methicillin-resistant strains of Staphylococcus aureus (MRSA) increased from less than 20% in 1999 to greater than 50% in 2003, and community-associated MRSA strains are now the predominant isolates found in US hospitals.² Many different bacterial genus now exhibit multidrug resistance, including Staphylococcus, Enterococcus, Gonococcus, Streptococcus, Salmonella, and Mycobacterium. To combat this threat, we need a broader approach to antimicrobial development that identifies novel targets with different modes of action, thereby leading to new classes of drugs that can bypass the developing resistance mechanisms.

The examination of new protein targets to identify potent lead compounds has traditionally been conducted by screening compound libraries of drug-like molecules. Although the overall success rate in identifying compounds that bind to a given target with high affinity is generally quite low, the increasing use of robotics to rapidly screen large, million-compound libraries continues to yield useful initial hits that can be further developed. An alternative approach, the screening of small compound libraries containing low molecular weight compounds (herein called “fragments”) with relatively few functional groups and limited drug-like properties, has some advantages over the traditional screening approach. The size of these fragment libraries is several orders of magnitude smaller than the large drug libraries, requiring less time and resources to screen. Because these fragments are smaller than typical drug-like molecules, the binding affinities are expected to be much weaker than drug library hits, and the specificity for the target protein is likely to be much lower. However, the success rates can be much higher for fragment libraries since different criteria are used to identify promising initial hits. Unlike screening with drug libraries, the criterion used to identify successful hits in fragment library screening is not high binding affinity but high ligand efficiency. Ligand efficiency (LE) is a measure of binding affinity that is adjusted for the smaller size of these fragments and is defined³ as

LE = - Δ G / no . of heavy (i . e ., nonhydrogen) atoms .

Most successful drug candidates have nanomolar affinities for their targets but also possess ligand efficiencies of 0.3 kcal/mol/heavy atom or greater. The goal for fragment library screening is to identify compounds that bind to the protein target with at least modest affinities but with initial LE values of 0.3 or greater. As the size and complexity of these initial hits are increased through structural elaborations and fragment coupling, maintaining high LE is used as the selection criterion to guide the development of lead compounds.⁴

The aspartate biosynthetic pathway is essential for microorganism survival, but this entire pathway is absent in mammals, thus requiring the acquisition of the products of this pathway from dietary sources. Aspartate β-semialdehyde dehydrogenase (ASADH) catalyzes the second step in the aspartate pathway, a reductive dephosphorylation that converts β-aspartyl phosphate to aspartate β-semialdehyde ( Fig. 1 ). The product of this reaction occupies the first branch point in this pathway, and inhibition of this enzyme will interrupt the production of 4 essential amino acids required for protein synthesis.⁵ In addition, this pathway produces several essential metabolites that play crucial roles in key microbial processes, including cell wall biosynthesis,⁶ protective dormancy,⁷ quorum sensing signaling,^8,9 and virulence factor production.¹⁰ Thus, interruption of the flux through this pathway will adversely affect the survival of microorganisms that depend on the products of the aspartate pathway. Earlier studies have shown that deletion of the asd gene (encoding ASADH) is lethal to these organisms,¹¹ and different studies to identify essential genes have each included the asd gene in the minimal set required for organism survival.^12-14

Fig. 1.

Physiological reaction catalyzed by aspartate β-semialdehyde dehydrogenase (ASADH).

The aspartate biosynthetic pathway presents a number of attractive and as yet untested targets for novel drug intervention. The identification of selective inhibitors of these critical microbial enzymes can lead to the development of new classes of antimicrobials that can be highly effective against the growing threat from multidrug-resistant infectious organisms. Structures have been determined of ASADHs isolated from a variety of microbial species, and these structures are organized into related groups based on the source of the enzyme. All of the ASADHs studied to date possess the same set of substrate binding and active site catalytic groups^15,16 despite sequence homologies ranging from more than 90% to less than 10%. The enzymes from gram-negative bacteria are closely related and share many common structural features.^17,18 Domain and secondary structural comparisons of the archael¹⁹ and fungal²⁰ forms of ASADH show that they are most closely related to each other and have the greatest differences with the gram-negative enzyme forms. The gram-positive bacterial ASADHs possess structural elements that are situated between those of the fungal and the gram-negative enzyme families.²¹ The combined application of kinetic and mutagenic studies, along with structural characterization of key catalytic intermediates,^21,22 has contributed to a detailed understanding of the mechanism of action of ASADH. Because of its crucial position at an early branch point in this essential microbial pathway, identifying new inhibitors against ASADH can be a valuable strategy for the development of novel antibacterial and antifungal drugs. In this study, ASADHs from the gram-positive Streptococcus pneumoniae (spASADH) and the gram-negative Vibrio cholerae (vcASADH) bacterium, as well as from a yeast species, Candida albicans (caASADH), were each screened against small fragment libraries to find initial hits that inhibit ASADH activity with high ligand efficiencies and, in some cases, with unexpected selectivities.

Materials And Methods

Materials

ASADHs from S. pneumonia, V. cholerae, and C. albicans were cloned, expressed, and purified as previously described^21,23,24 and then concentrated and stored in 50 mM HEPES (pH 7) with 1 mM EDTA and dithiothreitol (DTT) at −20°C. Aspartate β-semialdehyde (ASA) was synthesized as previously described²⁵ and stored in 4 M HCl at −20°C due to the instability of this compound under basic conditions. Working solutions of ASA were neutralized prior to addition to the assay mixtures.

Fragment libraries

Two fragment libraries were assembled by including a range of different low molecular weight compounds to probe binding to the target enzymes. The water-soluble fragment library (SFL) is composed of 384 compounds equally divided among 4 different compound classes:

Amino acids and derivatives—natural and unnatural amino acids; N-derivatized and amino acid esters; halo- and phospho-amino acids; dipeptides

Metabolites and analogues—mono-, di-, and tricarboxylic acids; halo acids; phosphorylated metabolites

Carbohydrates and bases—sugars, sugar acids, and amino sugars; nucleobases

Water-soluble organics and aromatics—5- and 6-membered heterocyclic derivatives; amino alcohols; halo acids and amides; nitro- and hydroxybenzyl derivatives

The average molecular weight of these compounds is ~155 Da, with 90% of the compounds between 90 and 250 Da. An organic fragment library (OFL) of 384 compounds was also assembled from 4 different classes of compounds:

Benzene derivatives—halo, cyano, and hydroxybenzenes; benzaldehydes and benzoic acids

Five-membered heterocycles—oxygen-, nitrogen-, and sulfur-containing heterocycles; halo, amino, carboxy, and methyl derivatives

Six-membered heterocycles—nitrogen- and oxygen- containing heterocycles; halo, amino, nitro, hydroxy, and carboxy derivatives

Fused/multiple ring systems—naphthalenes and indoles; quinines and benzyls; diphenyl and dibenzyl derivatives; cycloalkanes

The average molecular weight of the OFL members is 160 Da, with 90% of the compounds having molecular weights between 90 and 240 Da. Stock solutions of each compound were prepared in water (200 mM) or DMSO (400 mM), with some pH adjustment as needed to solubilize acids or bases in the SFL. For screening purposes, fragment cocktails were assembled by mixing 1 compound from each class, resulting in 96 fragment cocktails containing 4 compounds for each library.

Kinetic assay

ASADH catalyzes the reductive dephosphorylation of an acyl phosphate to generate an aldehyde product ( Fig. 1 ).

Due to the instability of aspartyl phosphate, this reaction is most easily followed in the reverse direction by monitoring the production of NADPH at 340 nm. Initial velocity kinetics for inhibition of ASADH were carried out at 25°C with the reaction mixture composed of 120 mM CHES (pH 8.6), 200 mM KCl, and the substrates ASA and NADP added at twice their respective K_m values. To identify ASADH inhibitors from the fragment libraries, this reaction was examined in the presence of different 4-compound fragment library cocktails.

Because the organic fragment library members are dissolved in 100% DMSO, it was necessary to identify the optimal enzyme assay conditions in the presence of this organic solvent. Fortunately, the ASADHs are quite stable and active in aqueous DMSO mixtures, retaining nearly 70% of their native activity in DMSO concentrations as high as 50%. Typical assays were run in 20% DMSO in the same reaction mixtures as was used to screen the SFL.

Library screening

Earlier studies had shown that the active sites are highly conserved throughout the whole ASADH family,²¹ and the optimum reaction pH for the Escherichia coli enzyme (ecASADH) is between 8 and 9. Therefore, the pH of each fragment cocktail was adjusted to 8~9 before screening by titrating with NaOH or HCl. To improve the screening efficiency, the reactions were monitored in a 96-well plate by using a SPECTRAmax® 340PC Microplate Spectrophotometer (Molecular Devices, Sunnyvale, CA). For each row of the plate, 160 µL of the stock buffer and substrate solution (final substrate concentrations: 1 mM ASA, 1.5 mM NADP, 20 mM phosphate) was added to each well with an 8-channel multitip pipettor, followed by 20 µL of each fragment cocktail solution. The reaction was initiated by addition of 20 µL of enzyme solution to each well in the row. To minimize the variation between the reactions in different rows, the initial velocity of the reaction was compared to an internal standard by assuming the reaction with the highest initial velocity rate in each row has no inhibitor present and then comparing those values to controls run in the absence of inhibitor. Good inhibitors of ASADH were defined as those compounds that resulted in less than 10% of the initial rate compared to the control reactions. Fragments from the cocktails containing these inhibitors were then screened individually by using the same protocol but now varying the concentration of each inhibitor. Inhibition constants (K_i values) were determined by a Dixon analysis²⁶ in which the inhibitor is assumed to compete with one of the substrates for binding to the enzyme. To eliminate false-positive readings, stock solutions of each individual inhibitor were scanned from 300 to 500 nm, and the concentration of any strongly absorbing inhibitors was decreased to allow the enzyme assay to remain in a linear absorbance range.

Results and Discussion

Fragment library screening

Earlier studies had identified several substrate analogue inhibitors against the target enzyme, ASADH obtained from E. coli, including S-methyl-L-cysteine sulfoxide, an active site-directed inactivator, and aspartyl β-difluorophosphonate, a reversible competitive inhibitor.^27-29 To introduce diversity and novelty into the list of ASADH inhibitors, a custom fragment library of water-soluble compounds (SFL) was assembled (as described in the Materials and Methods section) to test this experimental approach to inhibitor development against different members of the ASADH enzyme family. This SFL was screened kinetically against representative gram-negative (V. cholerae), gram-positive (S. pneumoniae), and fungal (C. albicans) forms of ASADH in 96 four-compound cocktails, containing 1 member from each compound class at 20-mM final concentrations. These ASADHs have very similar overall structures, reasonably high-sequence homologies, and an identical constellation of active site functional groups.^15,16 However, surprisingly, quite different inhibition patterns were observed against these enzymes that use an identical mechanism to catalyze the same chemical reaction with the same substrates. Only 4 of the 96 cocktails in this initial screen contained a compound that showed good inhibition against all 3 enzyme forms ( Fig. 2A ), whereas an additional 11 cocktails were found to contain an inhibitor against 2 of the 3 enzymes. Unexpectedly, 28 cocktails contained a compound that is inhibitory against only one of the ASADHs, with most of them showing selectivity for the bacterial ASADHs, and 53 cocktails (a total of 212 compounds) showed little or no inhibition against any ASADH even at these very high concentrations and were eliminated from further consideration.

Fig. 2.

Fragment library screening against Vibrio cholerae (vcASADH), Streptococcus pneumoniae (spASADH), and Candida albicans (caASADH). (A) Water-soluble fragment library (SFL); (B) organic-soluble fragment library (OFL). Each well contains a 4-compound cocktail, one from each structural class for the respective libraries. Cocktails that showed good inhibition against all 3 enzyme forms are shown in red, cocktails with inhibition against 2 enzyme forms are in orange, and those causing good inhibition against only 1 of the 3 enzymes are shown in yellow. Cocktails giving little or no inhibition are shown in white.

A second 384-compound, DMSO-soluble OFL was used to probe the aromatic binding selectivity between these different ASADHs. ASADH retains more than half of its catalytic activity in 20% DMSO, which is sufficient to maintain the solubility of the compounds in this OFL and expands the capability to measure inhibition by these water-insoluble compounds. Again, significant differences were observed in the inhibition patterns between the representative ASADHs. In contrast to the results obtained from the SFL screening, in this case 24 different cocktails from this OFL showed good inhibition against all 3 forms of ASADHs ( Fig. 2B ). However, given the highly colored nature of a number of these aromatic compounds, it is likely that strong absorbance is causing interference with the enzyme assay in a number of these cocktails. An additional 13 cocktails showed inhibition against 2 of the 3 enzymes, and 12 more inhibited only one of the representative target enzymes ( Fig. 2B ). A total of 188 of the 384 compounds in this library were immediately eliminated as potential inhibitors since the cocktails containing these compounds showed little or no inhibition against any of these representative ASADHs.

Inhibitor identification

The individual compounds present in these inhibitory cocktails from each fragment library were examined separately to identify the inhibitory component and also in various combinations if the individual compounds from each of these cocktails did not cause appreciable inhibition. As expected, in a number of cases, the individual compounds strongly absorbed at 340 nm, especially those compounds from the OFL, thus causing interference with the assay and leading to false-positive readings during cocktail screening. These compounds were each rescreened at the highest concentrations that still allowed accurate rate measurements to determine if any of the highly absorbent compounds were able to function as enzyme inhibitors. Inhibition constants (K_i values) were measured for each of the inhibitory compounds, first against the enzyme form identified from the initial screening studies and then against the other enzyme forms to measure their relative selectivities. From these kinetic studies, a total of 8 compounds were identified that inhibited spASADH, 13 compounds that inhibited vcASADH, and only 5 that inhibited caASADH with measureable affinities ( Table 1 ). Seven of these compounds were found to be inhibitors for 2 different enzyme forms, primarily for the bacterial ASADHs, whereas, surprisingly, only one of the 18 initial hits, 4-hydroxybenzophenone, gave measureable inhibition for all 3 representative ASADHs. However, even with this less selective inhibitor, there was significant discrimination in relative affinities between the different enzyme forms.

Table 1.

Aspartate β-Semialdehyde Dehydrogenase Inhibitors Identified from Fragment Library Screening

	K_i Values (mM)
Inhibitors	Chemical Formula	spASADH	vcASADH	caASADH	Selectivity^a sp:vc:ca
L-cystine	(⁻OOC-CH(NH₃ ⁺)CH₂S-)₂	0.036 ± 0.004	0.00023 ± 0.00006	>20^b	160:1:>87000
L-cysteine ethyl ester	HS-CH₂CH(NH₃ ⁺)COOC₂H₅	0.85 ± 0.22	0.00043 ± 0.00011	>20^b	2000:1:>46000
L-cysteine sulfinate	⁻O₂S-CH₂CH(NH₃ ⁺)COO^–	1.3 ± 0.4	0.25 ± 0.02	>20^b	5:1:>80
L-homocystine	(⁻OOC-CH(NH₃ ⁺)(CH₂)₂-S-)₂	>200^b	0.47 ± 0.10	>20^b	>400:1:>40
S-carbamoyl-L-cysteine	NH₂C(O)SCH₂CH(NH₃ ⁺)COO^–	>200^b	0.13 ± 0.02	>20^b	>1500:1:>150
D-glucosamic acid	HOCH₂(CHOH)₃CH(NH₂)COO^–	>200^b	3.2 ± 0.6	>20^b	>60:1:>6
Propionate	CH₃CH₂COO^–	0.46 ± 0.14	0.57 ± 0.07	>20^b	1:1.2:>40
D-2,3-diaminopropionate	CH₂(NH₂)CH(NH₂)COO^–	0.27 ± 0.05^c	0.47 ± 0.08^c	>10^b	1:1.7:>40
3-Bromopyruvate	CH₂Br-CH₂COO^–	>200^b	>200^b	0.34 ± 0.07	>500:>500:1
Maleimide	C₄H₂NH(O)₂	>200^b	>200^b	0.14 ± 0.04	>1400:>1400:1
N-iodosuccinimide	C₄H₄NI(O)₂	0.006 ± 0.002	0.016 ± 0.004	>10^b	1:2.8:>1600
Pyridoxal-5-phosphate	C₅HN(OH)(CHO)(CH₃)(CH₂OPO₃ ⁼)	0.14 ± 0.04	>200^b	>20^b	1:>1400:>140
4-Hydroxybenzophenone	C₆H₄OHC(O)C₆H₅	2.4 ± 0.42	0.33 ± 0.05	10 ± 2	8:1:30
5-Chloro-2-nitrobenzaldehyde	C₆H₃Cl(NO₂)CHO	>10^b	0.53 ± 0.17	>10^b	>20:1:>20
2,6-Dibromoquinone chlorimide	C₆H₂Br₂(O)(NCl)	>10^b	0.99 ± 0.11	>10^b	>10:1:>10
2-Chloro-3′,4′-dihydroxyacetophenone	C₆H₂Cl(OH)₂C(O)CH₃	>10^b	1.2 ± 0.2	>10^b	>8:1:>8
(Bromomethyl)cyclohexane	CH₂Br-C₆H₁₁	>20^b	>20^b	0.25 ± 0.08	>80:>80:1
3,4-Dihydroxyphenylacetate	C₆H₃(OH)₂CH₂COO^–	>20^b	>20^b	0.63 ± 0.14	>30:>30:1

Ratio of K_i values against representative gram-positive, gram-negative, and fungal ASADHs.

No inhibition was observed; the K_i is estimated to be at least 10 times greater than the highest concentration examined.

Adjusted for inhibition by only the D-isomer.

Bold font indicates compounds from the sets of new inhibitors that were selected for further examination and development. ASADH, aspartate semialdehyde dehydrogenase; vc, Vibrio cholerae; sp, Streptococcus pneumoniae; ca, Candida albicans.

Overall, 10 inhibitory compounds were identified from the aqueous SFL (a 2.6% hit rate) and 8 from the DMSO-soluble OFL (a 2.1% hit rate). In most cases, the observed inhibition constants were generally quite modest, typically in the low millimolar range, as would be expected for these lower molecular weight fragments. However, several compounds were found to inhibit one of the ASADHs with K_i values in the low micromolar range ( Table 1 ) and with extremely high ligand efficiencies. The basis for this high affinity and target selectively was then examined.

Selectivity of ASADH inhibitors

On the basis of our initial screening studies, significant differences were observed in the affinity of fragment library hits for each of these representative enzymes despite the presence of a highly conserved active site throughout the ASADH family. The affinity differences between the gram-negative and gram-positive bacterial enzymes suggest that some of the spASADH inhibitors are accessing a binding pocket or a combination of binding groups that are either not present or are less accessible in vcASADH. The complete lack of overlap between the initial hits for the yeast form of ASADH indicates that there must be unique interactions in caASADH that are not shared by the bacterial forms of this enzyme and can potentially be exploited for the development of antifungal agents against this enzyme target.

Some of these new inhibitors, such as propionate and 2,3-diaminopropionate, do not show appreciable selectivity between the bacterial forms of ASADH; however, a number of other compounds show considerable selectivity between the gram-negative and gram-positive ASADHs. S-carbamoyl-L-cysteine has more than a 1500-fold preference for vcASADH, and L-homocystine has more than a 400-fold higher affinity, whereas L-cysteine and L-cysteine ethyl ester are each more than 40,000-fold more potent against the gram-positive spASADH enzyme form ( Table 1 ). The particularly strong potency of these sulfur-containing substrate analogues against the bacterial ASADHs prompted further inquiry into the mode of inhibition for this group of compounds. Preincubation of vcASADH and spASADH with each of these compounds prior to kinetic analysis did not reveal a time-dependent component to their inhibition, suggesting a reversible, noncovalent mechanism of inhibition. However, subsequent structural studies with vcASADH in the presence of S-carbamoyl-L-cysteine provided conclusive evidence to the contrary. This enzyme-inhibitor complex structure shows clearly connected density between the thiolate group of the active site cysteine and the sulfur atom of the inhibitor, indicative of inactivation through covalent modification of the enzyme (Pavlovsky, Potente, and Viola, unpublished results). Although this selective reactivity may bode well for the development of related compounds as potent ASADH inhibitors, it is unlikely that even protected thiol-containing drugs would survive to reach this target enzyme without encountering other reactive sites in mammals. Because of this concern, we elected not to pursue the development of this class of potential inhibitors as initial lead compound candidates. However, the remaining substrate analogues in the second and third groups of compounds in Table 1 present some intriguing possibilities for inhibitor design and development.

Inhibitor development

Several interesting compounds from these sets of new inhibitors ( Table 1 , bold entries) were selected for further examination and development. 2,3-Diaminopropionate has a K_i in the low millimolar range for both spASADH and vcASADH, with a very high ligand efficiency of >0.5, and already shows selectivity (>40:1) for bacterial versus fungal ASADHs ( Table 1 ). Modeling and docking studies suggested that longer homologues of aminocarboxylates could fit into this active site and potentially make some favorable new interactions. This hypothesis was tested with some mono- and dicarboxylic acids containing either 1 or 2 amine functional groups. The 4- and 5-carbon diaminocarboxylate homologues were found to be good inhibitors of both bacterial enzyme forms with high LE values ( Table 2 ). Although the corresponding 4-carbon diaminodicarboxylate (2,3-diaminosuccinate) does not inhibit either enzyme form, the 5-carbon aminocarboxylate (5-aminocaproate) is a good inhibitor of both bacterial enzymes. The 5-carbon aminodicarboxylate (glutamate) is also a reasonably good inhibitor of vcASADH, but inhibition is only observed with the D-isomer, whereas the L-isomer shows no inhibition at concentrations up to 10 mM ( Table 2 ). D-glutamate was also found to be a micromolar inhibitor of spASADH with a 15-fold selectivity for this gram-positive enzyme over the gram- negative enzyme form. Subsequent structural studies support this chiral discrimination by showing that only the D-isomer from the racemic DL-2,3-diaminopropionate is bound to spASADH (Pavlovsky, Potente, and Viola, unpublished results). Accordingly, the measured K_i value for this mixture was divided by 2 to reflect binding by only one of the isomers.

Table 2.

Amino Acid Inhibitors of Bacterial ASADHs

		vcASADH		spASADH
Inhibitor	Chemical Formula	K_i (mM)	LE	K_i (mM)	LE	Selectivity vc:sp
DL-2,3-diaminopropionate	CH₂(NH₂)CH(NH₂)COO^–	0.47 ± 0.08^a	0.65	0.27 ± 0.05^a	0.70	1.7:1
L-2,4-diaminobutyrate	CH₂(NH₂)CH₂CH(NH₂)COO^–	0.14 ± 0.04	0.66	0.81 ± 0.11	0.53	1:6
L-2,5-diaminopentanoate	CH₂(NH₂)(CH₂)₂CH(NH₂)COO^–	0.59 ± 0.09	0.49	0.75 ± 0.08	0.48	1:1.3
5-Aminocaproate	H₂N(CH₂)₄COO^–	0.56 ± 0.08	0.56	0.41 ± 0.06	0.58	1.4:1
D-glutamate	⁻OOC(CH₂)₂CH(NH₃ ⁺)COO^–	1.3 ± 0.4	0.40	0.087 ± 0.014	0.56	15:1
Trans-3-hexenedioate	⁻OOC-CH₂CH=CHCH₂COO^–	5.2 ± 1.3	0.31	0.24 ± 0.05	0.49	22:1
3-Nitropropionate	⁻O₂N(CH₂)₂COO^–	0.24 ± 0.06	0.62	0.39 ± 0.03	0.58	1:1.6
2-Amino-4-phosphonobutyrate	⁼O₃P-(CH₂)₂CH(NH₃ ⁺)COO^–	1.56 ± 0.28	0.35	NI^b	—	1:>60
2-Amino-3-phosphonopropionate	⁼O₃P-CH₂CH(NH₃ ⁺)COO^–	1.53 ± 0.26	0.39	NI	—	1:>60
D-aspartate^c	⁻OOC-CH₂CH(NH₃ ⁺)COO^–	0.49 ± 0.09	0.50	NI	—	1:>200
D-2-aminoadipate	⁻OOC-(CH₂)₃CH(NH₃ ⁺)COO^–	0.23 ± 0.05	0.45	NI	—	1:>400
DL-2-aminobutyrate	CH₃CH₂CH(NH₃ ⁺)COO^–	NI	—	0.44 ± 0.05^d	0.66	>200:1
L-glutamate	⁻OOC(CH₂)₂CH(NH₃ ⁺)COO^–	NI	—	0.31 ± 0.07	0.48	>300:1

Bold font indicates the parent compound from the fragment library screen. ASADH, aspartate semialdehyde dehydrogenase; LE, ligand efficiency; NI, no inhibition; vc, Vibrio cholerae; sp, Streptococcus pneumoniae.

Adjusted for inhibition by only the D-isomer.

No inhibition observed at 10 mM.

No inhibition observed for the corresponding L-isomer.

Adjusted for inhibition by only the L-isomer.

The 4-carbon dicarboxylic acid succinate does not inhibit either enzyme form, but a nitro analogue of succinate (3-nitropropionate) is a strong inhibitor of both enzymes with high LE values. A conformationally constrained, 6-carbon dicarboxylic acid (trans-3-hexenedioic acid) was also examined and shows >20-fold preference for binding to the gram-positive enzyme form and diminished LE to the gram-negative enzyme. Although the K_i values for these compounds vary by a factor of about 60 from the most to least potent, each of these inhibitors still shows good ligand efficiency values ( Table 2 ).

Although these particular aminocarboxylate inhibitors show only modest selectivities between these bacterial enzymes, several highly selective inhibitors were identified for both the gram-negative and the gram-positive forms of ASADH. Aminophosphonates have low millimolar inhibition for only vcASADH, and the D-isomers of aspartate and 2-aminoadipate are also selective inhibitors of only vcASADH ( Table 2 ). By contrast, the L-isomers of glutamate and 2-aminobutyrate are selective inhibitors of only the gram-positive spASADH ( Table 2 ). The K_i value for the racemic DL-2-aminobutyrate was divided by 2 to reflect inhibitor by only one of the isomers, and subsequent measurements with L-2-aminobutyrate (K_i = 0.49 ± 0.10 mM) confirmed that the inhibition of spASADH is caused by binding of the L-isomer. These results validate our modeling studies, suggesting the need for a 4- to 6-carbon dicarboxylate core structure and a preference for at least 1 amino group as the minimal pharmacophore ( Fig. 3 ).

Fig. 3.

Minimal inhibitory amino acid pharmacophore (n = 0-3, m = 1-2).

The L-stereoisomers of these amino acids show preferential inhibition of the gram-positive ASADH, whereas the D-isomers are selective inhibitors of the gram-negative enzyme form. This chiral selection was unexpected because the substrate for this enzyme family (ASA) requires L-stereochemistry at the α-carbon. The basis for this stereoselectivity between these structurally and functionally related enzymes must be established by structural studies of the respective enzyme-inhibitor complexes.

4-Hydroxybenzophenone shows some selectivity (8:1) for gram-negative versus gram-positive bacterial ASADHs and only very weak inhibition of the fungal enzyme ( Table 1 ). To examine the structural basis for this selectivity, a series of benzophenone analogues were evaluated as potential inhibitors of vcASADH. The parent compound (benzophenone) shows no inhibition of this enzyme even when tested at the 20-mM concentration. Addition of a single amine or a carboxyl group to the benzophenone ring at either the 2- or 4-position yields modest inhibitors ( Table 3 ), in contrast to the low millimolar inhibition observed with the 4-hydroxy derivative. Increasing the number of hydroxyl groups that decorate the ring structure from 1 to 4 produces a further 10-fold improvement in affinity while maintaining the LE. Replacing the central carbonyl group of benzophenone with an imine leads to reasonably potent inhibition even for the parent compound, suggesting the need for a hydrogen-bond donor at this position. But derivatizing this imine with either a hydroxyl group (oxime) or an amine (hydrazone) to introduce additional hydrogen-bond donor groups actually causes diminished potency and much lower LE values ( Table 3 ). The hydroxybenzophenones and the imine derivative each possess good ligand efficiency values and suggest core structural requirements that can be further elaborated to produce more potent and selective inhibitors of gram- negative ASADHs.

Table 3.

Benzophenone Inhibitors of vcASADH

Inhibitor^a	X	R	K_i (mM)	LE
4-hydroxybenzophenone	O	OH	0.33 ± 0.05	0.32
Benzophenone	O	—	NI^b	—
4-Aminobenzophenone	O	NH₂	>10^c	<0.18
4-Benzoyl benzoate	O	COO^–	1.1 ± 0.3	0.24
2-Benzoyl benzoate	O	COO^–	2.5 ± 0.8	0.21
2,2′,4,4′-Tetrahydroxybenzophenone	O	(OH)₄	0.041 ± 0.007	0.33
Benzophenone imine	NH	—	0.37 ± 0.08	0.34
Benzophenone hydrazone	N-NH₂	—	26 ± 4	0.15
Benzophenone oxime	N-OH	—	29 ± 6	0.14

Bold font indicates the parent compound from the fragment library screen. ASADH, aspartate semialdehyde dehydrogenase; LE, ligand efficiency; vc, Vibrio cholerae.

Parent compound: .

No inhibition observed at 20 mM

Strongly absorbs at 340 nm above 0.1 mM.

The fungal form of ASADH has a similar overall structure to that of the bacterial enzymes²⁰ but shows a very different fragment library screening profile in which each of the inhibitors identified against caASADH was found to be noninhibitory toward the bacterial enzymes ( Table 1 ). Several compounds from the OFL show low millimolar inhibition of this fungal enzyme and no detectable inhibition of the bacterial enzyme forms. Both (bromomethyl)cyclohexane and 3,4-dihydroxyphenylacetate from this library have LE values of >0.3 for caASADH. The parent ring structure (phenylacetate) is still an effective inhibitor with an improved LE, but the introduction of a functional group at the 4-position leads to a decrease in binding potency ( Table 4 ). However, retaining these functional groups at the 3-position maintains the binding affinity for caASADH. Extension of the carboxyl side chain through the introduction of either an α-hydroxy group (phenyllactate) or an α-amino group (phenylglycine) causes complete abolition of binding.

Table 4.

Organic Acid Inhibitors of caASADH

Inhibitor	Chemical Formula	K_i (mM)	LE
3,4-Dihydroxyphenylacetate	(HO)₂-C₆H₃CH₂COO^–	0.63 ± 0.14	0.37
Phenylacetate	C₆H₅CH₂COO^–	0.83 ± 0.28	0.42
4-Aminophenylacetate	H₂N-C₆H₄CH₂COO^–	2.2 ± 0.2	0.33
4-Nitrophenylacetate	O₂N-C₆H₄CH₂COO^–	3.6 ± 0.7	0.26
4-Hydroxyphenylacetate	HO-C₆H₄CH₂COO^–	NI^a	—
3-Aminophenylacetate	H₂N-C₆H₄CH₂COO^–	0.61 ± 0.13	0.40
3-Hydroxyphenylacetate	HO-C₆H₄CH₂COO^–	0.49 ± 0.15	0.41
Phenyllactate	C₆H₅CH₂C(OH)COO^–	NI^a	—
Phenylpropionate	C₆H₅(CH₂)₂COO^–	0.63 ± 0.08	0.40
Phenylglycine	C₆H₅CH(NH₃ ⁺)COO^–	NI^a	—
3-Bromopyruvate	CH₂Br-C(O)COO^–	0.34 ± 0.07	0.68
3-Fluoropyruvate	CH₂F-C(O)COO^–	0.14 ± 0.02	0.76
3-Iodopyruvate	CH₂I-C(O)COO^–	0.18 ± 0.02	0.73
Chloroacetate	CH₂Cl-COO^–	4.6 ± 0.6	0.64
Bromoacetate	CH₂Br-COO^–	3.2 ± 0.6	0.68
Chlorodifluoroacetate	CF₂Cl-COO^–	0.59 ± 0.05	0.63
3-Bromopropionate	CH₂Br-CH₂COO^–	0.64 ± 0.05	0.73
3-Chloropropionate	CH₂Cl-CH₂COO^–	0.24 ± 0.01	0.83
4-Bromobutyrate	CH₂Br-(CH₂)₂COO^–	0.16 ± 0.01	0.74
5-Bromovalerate	CH₂Br-(CH₂)₃COO^–	0.11 ± 0.03	0.68
6-Bromohexanoate	CH₂Br-(CH₂)₄COO^–	0.21 ± 0.02	0.56
Bromosuccinate	-OOC-CH₂Br-CH₂COO^–	0.55 ± 0.07	0.50

Bold font indicates the parent compound from the fragment library screen. ASADH, aspartate semialdehyde dehydrogenase; LE, ligand efficiency; ca, Candida albicans.

No inhibition observed at 10 mM.

3-Bromopyruvate from the SFL is a sub-millimolar inhibitor of caASADH, and its small size leads to a very high ligand efficiency value ( Table 4 ). Based on this initial hit from screening studies, a set of structurally related compounds was examined. For a series of structural analogues in which the chain length and the nature of the halogen substituent were varied, the K_i values remained in the low millimolar range, and the LE values each remained consistently high (>0.5) ( Table 4 ).

As the nature of the interactions between these low molecular weight inhibitors and the different forms of ASADH is elucidated, elaboration and functionalization of these different core structures will produce compounds with more drug-like properties. In addition, coupling of fragments that bind at adjacent sites on the target enzyme will increase the complexity of these inhibitors and is expected to lead to the development of lead compounds with increased potency and enhanced selectivity.

Footnotes

Acknowledgements

The authors thank DeMarco Camper and Buenafe Arachea for providing purified samples of the V. cholerae and C. albicans forms of ASADH, as well as Drs. Paul Erhardt and Rahul Khupse (University of Toledo, Center for Drug Design and Development) for helpful discussions on the synthesis of new enzyme inhibitors.

This work was supported by a grant from the National Institutes of Health (AI077720).

References

Courvalin

Carlier

Chabbert

: Plasmid linked tetracycline and erythromycin resistance in group D Streptococcus . Ann Inst Pasteur (Paris) 1972;123:755-759.

Maree

Daum

Boyle-Vavra

Matayoshi

Miller

: Community-associated methicillin-resistant Staphylococcus aureus isolates causing healthcare-associated infections. Emerg Infect Dis 2007;13:236-242.

Hopkins

Groom

Alex

: Ligand efficiency: a useful metric for lead selection. Drug Discov Today 2004;9:430-431.

Saxty

Woodheard

Berdini

Davies

Verdonk

Wyatt

: Identification of inhibitors of protein kinase B using fragment-based lead discovery. J Med Chem 2007;50:2293-2296.

Viola

: The central enzymes of the aspartate family of amino acid biosynthesis. Acc Chem Res 2001;34:339-349.

Van Heijenoort

: Recent advances in the formation of the bacterial peptidoglycan monomer unit. Nat Prod Rep 2001;18:503-519.

Ragkousi

Eichenberger

van Ooij

Setlow

: Identification of a new gene essential for germination of Bacillus subtilis spores with Ca²⁺-dipicolinate. J Bacteriol 2003;185:2315-2329.

Lyon

Novick

: Peptide signaling in Staphylococcus aureus and other gram-positive bacteria. Peptides 2004;25:1389-1403.

Chen

Schauder

Potier

Van Dorsselaer

Pelczer

Bassler

: Structural identification of a bacterial quorum-sensing signal containing boron. Nature 2002;415:545-549.

10.

Niu

Clemmer

Bonomo

Rather

: Isolation and characterization of an autoinducer synthase from Acinetobacter baumannii . J Bacteriol 2008;190:3386-3392.

11.

Galan

Nakayama

Curtiss

: Cloning and characterization of the asd gene of Salmonella typhimurium: use in stable maintenance of recombinant plasmids in Salmonella vaccine strains . Gene 1990;94:29-35.

12.

Kobayashi

Ehrlich

Albertini

Amati

Andersen

Arnaud

: Essential Bacillus subtilis genes. Proc Natl Acad Sci USA 2003;100:4678-4683.

13.

Salama

Shepard

Falkow

: Global transposon mutagenesis and essential gene analysis of Helicobacter pylori . J Bacteriol 2004;186:7926-7935.

14.

Akerley

Rubin

Novick

Amaya

Judson

Mekalanos

: A genome-scale analysis for identification of genes required for growth or survival of Haemophilus influenzae . Proc Natl Acad Sci USA 2002;99:966-971.

15.

Blanco

Moore

Faehnle

Coe

Viola

: The role of substrate-binding groups in the mechanism of aspartate-β-semialdehyde dehydrogenase. Acta Cryst 2004;D60:1388-1395.

16.

Blanco

Moore

Faehnle

Viola

: Critical catalytic functional groups in the mechanism of aspartate-β-semialdehyde dehydrogenase. Acta Cryst 2004;D60:1808-1815.

17.

Hadfield

Kryger

Ouyang

Petsko

Ringe

Viola

: Structure of aspartate-β-semialdehyde dehydrogenase from Escherichia coli, a key enzyme in the aspartate family of amino acid biosynthesis. J Mol Biol 1999;289:991-1002.

18.

Blanco

Moore

Kalabeeswaran

Viola

: A structural basis for the mechanism of aspartate-β-semialdehyde dehydrogenase from Vibrio cholerae . Protein Sci 2003;12:27-33.

19.

Faehnle

Ohren

Viola

: A new branch in the family: structure of aspartate-β-semialdehyde dehydrogenase from Methanococcus jannaschii . J Mol Biol 2005;353:1055-1068.

20.

Arachea

Liu

Pavlovsky

Viola

: Expansion of the aspartate-β-semialdehyde dehydrogenase family: the first structure of a fungal ortholog. Acta Cryst 2010;D66:205-212.

21.

Faehnle

Liu

Le Coq

Viola

: An examination of key intermediates in the catalytic cycle of aspartate semialdehyde dehydrogenase from a gram-positive bacteria. J Biol Chem 2006;281:31031-31040.

22.

Blanco

Moore

Viola

: Capture of an intermediate in the catalytic cycle of L-aspartate-β-semialdehyde dehydrogenase. Proc Natl Acad Sci USA 2003;100:12613-12617.

23.

Moore

Bocik

Viola

: Expression and purification of aspartate β-semialdehyde dehydrogenase from infectious microorganisms. Protein Expr Purif 2002;25:189-194.

24.

Camper

Viola

: Fully automated multi-step protocols for protein purification. Anal Biochem 2009;393:176-181.

25.

Black

Wright

: Aspartate semialdehyde dehydrogenase and aspartate semialdehyde. J Biol Chem 1955;213:39-50.

26.

Dixon

: The determination of enzyme inhibitor constants. Biochem J 1953;55:170-171.

27.

Hadfield

Shammas

Kryger

Ringe

Petsko

Ouyang

: Active site analysis of the potential antimicrobial target aspartate semialdehyde dehydrogenase. Biochemistry 2001;40:14475-14483.

28.

Cox

Hadfield

Mayo-Martin

: Difluoromethylene analogues of aspartyl phosphate: the first synthetic inhibitors of aspartate semialdehyde dehydrogenase. Chem Commun 2001;18:1710-1711.

29.

Han

Moore

Viola

: A facile synthesis of a difluoromethylene analog of β-aspartyl phosphate as an inhibitor of L-aspartate-β-semialdehyde dehydrogenase. Synlett 2003:845-846.

Identification of Selective Enzyme Inhibitors by Fragment Library Screening

Abstract

Keywords

Introduction

Materials And Methods

Materials

Fragment libraries

Kinetic assay

Library screening

Results and Discussion

Fragment library screening

Inhibitor identification

Selectivity of ASADH inhibitors

Inhibitor development

Footnotes

Acknowledgements

References