SEARCH

SEARCH BY CITATION

Keywords:

  • autotaxin;
  • enzymology;
  • inhibitors;
  • virtual screening

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. References
  9. Supporting Information

Modulation of autotaxin (ATX), the lysophospholipase D enzyme that produces lysophosphatidic acid, with small-molecule inhibitors is a promising strategy for blocking the ATX–lysophosphatidic acid signaling axis. Although discovery campaigns have been successful in identifying ATX inhibitors, many of the reported inhibitors target the catalytic cleft of ATX. A recent study provided evidence for an additional inhibitory surface in the hydrophobic binding pocket of ATX, confirming prior studies that relied on enzyme kinetics and differential inhibition of substrates varying in size. Multiple hits from previous high-throughput screening for ATX inhibitors were obtained with aromatic sulfonamide derivatives interacting with the hydrophobic pocket. Here, we describe the development of a ligand-based strategy and its application in virtual screening, which yielded novel high-potency inhibitors that target the hydrophobic pocket of ATX. Characterization of the structure–activity relationship of these new inhibitors forms the foundation of a new pharmacophore model of the hydrophobic pocket of ATX.


Abbreviations
ADMAN-LPC

analog 3-acyl-7-dimethylaminonaphthyl-1-lysophosphatidylcholine

ATX

autotaxin

FS-3

fluorogenic substrate 3

HTS

high-throughput screening

LPA

lysophosphatidic acid

LPAR

lysophosphatidic acid receptor

LPC

lysophosphatidylcholine

PDB

Protein Data Bank

pNP-TMP

p-nitrophenyl thymidine 5′-monophosphate

SAR

structure–activity relationship

SD

standard deviation

UC-DDC

University of Cincinnati Drug Discovery Center

Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. References
  9. Supporting Information

Autotaxin (ATX) inhibitors represent an emerging class of drug candidates with therapeutic potential for the treatment of disease conditions that include cancer, cardiovascular disease, chronic inflammation, and neuropathic pain [1-5]. Numerous medicinal chemistry efforts have been successful in identifying ATX inhibitors; among these are HA155, PF-8380, S32826, and the benzyl-methylphosphonic and naphthyl-methylphosphonic acid compounds RG-22 and RG-30b (Fig. 1) [6-11].

image

Figure 1. Chemical structures of known ATX inhibitors.

Download figure to PowerPoint

Despite these efforts, scant success has been reported in the preclinical development of therapeutic agents that target ATX. One reason for the slow progress in drug development is that many of the small molecules identified to date are lipid-like, which leads to unsuitable partition coefficients for drugs (log P > 5) that limit their therapeutic utility. Mechanistically, many of the ATX inhibitors reported to date have targeted the catalytic site of ATX [12], and the binding sites of those that may not work by blocking the catalytic site have not been elucidated in detail. Therefore, lipid-like inhibitors mimicking ATX's natural substrate lysophosphatidylcholine (LPC) or its product lysophosphatidic acid (LPA), or with a molecular mass of > 500 Da, all fail Lipinski's rule of five, which defines the properties for an orally bioavailable active drug [13]. Recent crystal structures of ATX with the active site inhibitor HA155 cocrystallized have provided insights into potentially inhibitory surfaces of ATX that should accelerate the rational discovery of the next generation of small-molecule drug-like ATX inhibitors [14-16].

Recently, we reported the characterization of an ATX inhibitor, 918013 (Fig. 1), that exerts its blocking action via interactions with the hydrophobic pocket of ATX, an area outside the catalytic site [17]. This competitive inhibitor showed the same pharmacological and biological effects on ATX activity as compounds that block the catalytic site. The anti-invasive and antimetastatic effect of 918013 suggested the potential utility of ATX inhibitors that interact only with the hydrophobic pocket as new tools for the elucidation of the role of ATX and LPA in vivo. Compound 918013 was only one of multiple hits identified in high-throughput screening (HTS) of ATX inhibitors containing an aromatic sulfonamide group.

Here, we report the application of a focused virtual and experimental screening approach using the University of Cincinnati Drug Discovery Center (UC-DDC) database of ~ 340 000 searchable chemical structures. In this virtual screening, we initially identified 230 in silico hits, 38 of which inhibited ATX hydrolysis of the LPC-like fluorogenic substrate 3 (FS-3) by > 50% at 10 μm, yielding an experimentally validated hit rate of 17%. Examination of the structure–activity relationships (SARs) of this series of compounds showed that small changes to the structure of these compounds at four positions can result in dramatic changes in pharmacological properties. Using this SAR examination of the aromatic sulfonamide scaffold, we developed a new and spatially more constrained pharmacophore design, and delineated the unique interactions between the aromatic sulfonamide group and residues lining the hydrophobic pocket of ATX. This search strategy identified a new ATX inhibitor, 403070, with an inhibition constant (Ki) of 8.4 nm and drug-like properties. This inhibitor provides an excellent starting point for optimization and in vivo testing, and also validates a new pharmacophore model of the hydrophobic pocket.

Results

  1. Top of page
  2. Abstract
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. References
  9. Supporting Information

Identification of novel ATX inhibitors by virtual screening

We recently implemented an HTS assay to identify small-molecule inhibitors of ATX. In screening 10 000 compounds from a diversity set of the UC-DDC database, we identified 196 potential ATX inhibitors that caused 50% inhibition of the hydrolysis of FS-3 when applied at 10 μm. Six of these compounds were found to have close similarity, all sharing an aromatic sulfonamide motif that includes 918013. In order to identify new ATX inhibitors, virtual screening was performed, with application of a series of two-dimensional similarity searches. Molecular fingerprints based on molecular access system keys were calculated for the UC-DDC library of > 340 000 compounds. Three sequential rounds of database screening were performed with three different templates as queries: 918013 (5997 hits), 934313 (2360 hits), and 897131 (1754 hits). A Tanimoto threshold of > 80 was applied as a cutoff for each screening [18]. We applied an intersection inequality approach to combine the search results, maximizing the chances of identifying compounds with similar degrees of ATX inhibition. Cross-matching of the database results with the three templates yielded 230 overlapping virtual hits.

Experimental validation of virtual hit ATX inhibitors

Inhibition assays of ATX-mediated hydrolysis of FS-3 were performed with the selected 230 compounds applied at a fixed concentration of 10 μm. A compound was defined as an experimentally validated hit if it caused > 50% inhibition of the hydrolysis of the model lysophospholipase D substrate FS-3. Thirty-eight inhibitory compounds from the initial screening met this criterion (17% hit rate; Fig. S1). ATX, also known as nucleotide pyrophosphate phosphatase 2, belongs to a family of enzymes that possess phosphodiesterase activity against the nucleotide substrate p-nitrophenyl thymidine 5′-monophosphate (pNP-TMP). The size of the hit set was further decreased by applying a series of five experimental filters that included: (a) substrate selectivity for the lysophospholipase substrate FS-3 versus the nucleotide-like substrate pNP-TMP (Fig. 2A and Table S1); (b) determination of the dose–response relationship for inhibition of lysophospholipase D enzyme activity (Table 1); (c) counter-screening to eliminate molecular aggregates in the presence of 0.01% Triton X-100 (Fig. 2B); (d) inhibition of the hydrolysis of 3-acyl-7-dimethylaminonaphthyl-1-LPC (ADMAN-LPC; Table 1); and (e) cleavage of different molecular species of LPC (Table 2). A compound was designated as desirable if it had an IC50 of < 1 μm and inhibited FS-3, LPC and ADMAN-LPC hydrolysis. This stringent set of criteria was satisfied by only three compounds – 390596, 406643, and 403070 (Fig. 3) – with respective IC50 values of 141.4, 58.14 and 21.49 nm, and Ki values ranging from 8 nm to 440 nm against the FS-3 substrate. In addition, we found 12 other compounds that had potency with IC50 values of < 3 μm (Table S2).

Table 1. Summary of the in vitro pharmacological characterization of selected ATX inhibitors.
 FS-3ADMAN-LPCInvasion assay
CompoundPercentage inhibition at 10 μmIC50 (nm)Mechanism of inhibitionKi (nm)Percentage inhibition at 10 μmPercentage inhibition at 10 μm
  1. Hydrolysis of 1 μm FS-3 by 10 nm ATX was measured after 4 h of incubation in the presence and absence of 10 μm inhibitor and presented as mean percentage inhibition of ATX ± SD (n = 3). FS-3 hydrolysis was utilized to determine inhibitor potency; inhibitors at 0.03–30 μm were assayed for their effects on 10 nm ATX-mediated hydrolysis of 1 μm FS-3 after 4 h of incubation. Percentage ATX activity was plotted versus inhibitor concentration, and nonlinear regression analysis was performed in graphpad prism version 5.0a to calculate mean IC50 ± SD (n = 3). Inhibitor mechanism of action was assessed via FS-3 hydrolysis; FS-3 at 0.3–10 μm was incubated in the presence of 10 nm ATX with 0, 0.5 or two times the previously calculated IC50 concentration for each inhibitor. Reaction rate data were then simultaneously fitted via nonlinear regression in the Michaelis–Menten equations for competitive, non-competitive, uncompetitive or mixed-mode inhibition. The mechanism was selected on the basis of the highest global correlation coefficient in conjunction with interpretation of the program's calculated α-value, which reflects the change in enzyme–substrate affinity upon inhibitor binding. Ki was also calculated on the basis of the selected Michaelis–Menten nonlinear regression analysis of FS-3 hydrolysis reaction rate data, and presented as mean Ki ± SD (n = 3). Hydrolysis of 30 μm ADMAN-LPC by 30 nm ATX was assessed after 4 h of incubation and subsequent modified Bligh–Dyer extraction of lipids, which were then separated via TLC. Separated lipids were imaged under UV transillumination, and densitometric analysis was conducted in imagej for comparison of ADMAN-LPA and ADMAN-LPC, in order to calculate percentage inhibition of ATX ± SD (n = 3). A2058 human melanoma cell invasion was measured with the BD Biosciences BD BioCoat Matrigel-coated tumor invasion system. Cell invasion was measured after 16 h of incubation in the presence of 0.3 nm ATX in conjunction with 1 μm LPC 18:1 ± 10 μm inhibitor, and presented as mean percentage cell invasion ± SD (n = 3).

91801389 ± 0.332 ± 5.8Competitive14.6 ± 1.585 ± 11.067 ± 14.4
39059663 ± 1.0141 ± 9.4Competitive435 ± 6.533 ± 6.248 ± 13.0
40664345 ± 2.958 ± 3.7Competitive413 ± 2734 ± 18.013 ± 25.2
40307084 ± 0.221 ± 3.7Competitive8.4 ± 1.1100 ± 0.082 ± 3.5
Table 2. Inhibition of ATX-mediated hydrolysis of LPCs.
 Percentage inhibition (± SD)
CompoundLPC 14:0LPC 16:0LPC 18:0LPC 18:1
  1. Data (relative fluorescence) were recorded as a mean value of the triplicates for each sample, and reported as percentage inhibition of ATX-mediated LPC hydrolysis.

91801341.0 ± 5.155.4 ± 1.743.6 ± 2.651.8 ± 2.4
39059634.9 ± 0.654.6 ± 0.936.2 ± 2.343.9 ± 3.0
40664325.0 ± 1.829.3 ± 2.910.4 ± 3.612.4 ± 2.1
40307039.1 ± 2.742.9 ± 2.323.0 ± 3.624.5 ± 2.5
image

Figure 2. Identification and characterization of ATX inhibitors by HTS. (A) Comparison of selectivity against FS-3 (filled bars) and pNP-TMP (open bars) for the three inhibitors applied at 10 μm. Compound 966791 is a representative example of a catalytic site inhibitor that blocks the hydrolysis of both FS-3 and pNP-TMP, and is included to validate the assay. The upper bars represent inhibition of the hydrolysis of a substrate, and the lower bars represent enhancement. The inhibition was normalized to the cleavage of the substrate in the absence of the inhibitors, designated as 100%. (B) Dose–response curve of most potent aromatic sulfonamide hits from virtual screening with the FS-3 substrate in Triton X-100 counter-screening. Data points are the mean of triplicate determinations.

Download figure to PowerPoint

image

Figure 3. The structural scaffold of the aromatic sulfonamide hits and selected chemical structures used in SAR studies from the virtual screening.

Download figure to PowerPoint

The enzyme kinetics of ATX inhibition by aromatic sulfonamide inhibitors

The aromatic sulfonamide ATX inhibitors were tested in the presence of increasing concentrations of FS-3. These experiments were performed with 0.5 times and two times the IC50 of the inhibitors. The data were fitted via nonlinear regression into the Michaelis–Menten equations for competitive, noncompetitive, uncompetitive and mixed-mode inhibition to determine the best model that could be fitted to the inhibition curves (see 'Experimental procedures' and Table 1). These experiments showed that all three of our aromatic sulfonamide inhibitors inhibited ATX-mediated hydrolysis of FS-3 by a competitive mechanism of action. The Ki of 403070 is 8.43 nm, which is similar to that of our primary lead 918013 (14.64 nm, which was found from HTS [17]).

Inhibition of the ATX cleavage of natural LPC species by the aromatic sulfonamide compounds

These three compounds were tested for inhibition of the ATX-mediated hydrolysis of LPC 14:0, LPC 16:0, LPC 18:0, and LPC 18:1 (Table 2). The results of these experiments showed that every compound inhibited the cleavage of all four molecular species of LPC with similar efficacy when applied at 10 μm. Thus, these hydrophobic pocket inhibitors do not show any preference for the acyl chain of the substrate at this maximally effective concentration.

SARs for ATX inhibitors

After establishing the rank order of potency and common competitive mechanism of action of these aromatic sulfonamide compounds against lysophospholipid-like substrates, we examined their SARs (Fig. 3). The preliminary SAR analysis of this series of compounds suggested an overall trend in activity with replacement of the R1–4 groups (Fig. 3) in comparison with 918013. Replacement of the electronegative group in 918013 with a less bulky and more polar phenolic hydroxyl group in 390596 led to a five-fold decrease in potency (see Table 1 for IC50 values). The unsubstituted benzamide analog 406643 showed a 13-fold decrease in Ki as compared with 918013 (Table 1). Relocation of the fluorine to the para position and removal of the chlorine ortho to the carboxamide in 501309 reduced the potency of inhibition to 6.7 μm. Thus, changes in the R4 substituent affected efficacy (918013, 390596, 406643, 403070, and 390462). For the non-chlorine-containing 917978 and 299375, in which the two chlorines (R2 and R3) were replaced with a single bromine at R2, and the R4 substituents were either p-fluoro or p-bromo, efficacy was nearly identical. Thus, the R4 halogen group is of importance in determining the efficacy of the compound. Examination of the R4 substituent that was m-fluoro (918013) or p-bromo (403070) versus p-methyl (390462), m-hydroxyl (390596) or a hydrogen (406643) showed that the size of the R4 substituent was more important for activity than its orientation. The effect of the orientation of the R4 group is best demonstrated by comparing 918013 and 403070: the orientation of the R4 substituent is different, yet both had almost identical IC50 and Ki values (Table 1).

Molecular docking of 403070 into the hydrophobic pocket of ATX

To further elucidate the mechanism of ATX inhibition by 403070, molecular docking studies were performed with the ATX crystal structure of Protein Data Bank (PDB) 2XRG [15]. The previous kinetic studies suggested that 918013 is a competitive inhibitor of FS-3 without inhibiting pNP-TMP hydrolysis [17]. This pattern leads to the hypothesis that 403070, like 918013 and its analogs, occupies the hydrophobic pocket of ATX without interfering with the pNP-TMP placement at the catalytic site. Indeed, when 403070 was docked into this ATX structure (Fig. 4A), it occupied the ATX hydrophobic pocket without protruding into the catalytic cleft. On the basis of this model, 918013 and 403070, which are different only at the R1 and R4 positions (Fig. 3), docked in a similar space in the hydrophobic pocket and engaged in the same interactions with this inhibitory surface (Fig. 4B,C). These results are consistent with the competitive binding mechanism of 403070 against FS-3 and its lack of inhibition of pNP-TMP cleavage. Furthermore, the conservation of binding interactions between 918013 and 403070 provides a plausible explanation for the similarities in their potency and efficacy (Table 1).

image

Figure 4. Molecular modeling of the interaction of ATX with its inhibitors. (A, B) Docked positions of active hits for 918013 [green (A)] and 403070 [yellow (B)], shown overlayed on HA155 (orange) in the hydrophobic pocket (blue). (C) Overlay of 918013 (yellow) with 403070 (green) in the hydrophobic pocket (blue) shows the striking similarities between the positions and binding interactions of the two inhibitors. (D) Key π–π interactions between 403070 (yellow) and ATX involve the hydrophobic residues Tyr307, Phe274, and Phe275. (E) Overlay of 38 virtual hit compounds individually docked to ATX shows very tight placement in the hydrophobic pocket. (F) Simultaneous docking of pNP-TMP and 403070 into the ATX structure shows no interference in binding. This type of docking simulation provides a plausible explanation for the lack of inhibition and enhancement of pNP-TMP hydrolysis by 403070.

Download figure to PowerPoint

Our docking simulations predict that the aromatic sulfonamide group undergoes aromatic interactions with the aromatic residues on the interface between the hydrophobic pocket and nucleotide-binding region of ATX (Fig. 4D). This ring portion of the aromatic sulfonamide series of compounds can engage residues Phe211, Tyr307, and Phe275 in a π–π charge–charge interaction. The R4 halogen on 918013 and 403070 occupies space deep in the hydrophobic pocket lined by Phe274, Ala305, Ser170, and Ala218. From these docked positions, the aromatic sulfonamide scaffold shows interactions that are analogous to that of LPC or the cocrystallized inhibitor HA155. Although some interacting residues are shared between our aromatic sulfonamide analogs and other previously identified inhibitors, the aromatic sulfonamide compounds do not protrude into the catalytic site. For example, the catalytic site residues His360, His475 and His315 are too far away to interact with the aromatic sulfonamides in their docked position.

Because the F275A ATX mutant retains FS-3 hydrolytic activity [17], we tested the inhibitory potency of 403070 on this mutant. We hypothesized that 403070 would no longer inhibit this mutant, owing to the loss of its interaction with Phe275 in the hydrophobic pocket. Both 918013 and 403070 showed a dramatic IC50 shift from the nanomolar range to > 3 μm when the F275A ATX mutant was tested. These results provide important experimental proof for the interaction of aromatic sulfonamides with the inhibitory surface of the hydrophobic pocket in which Phe275 plays a crucial role.

Effect of inhibitors on cell invasion

To evaluate the effect of the new aromatic sulfonamide inhibitors in a cell-based system, we used ATX-dependent invasion of A2058 human melanoma cells. The compounds were initially screened at 10 μm. Dose–response curves were then generated for compounds that showed > 50% inhibition of A2058 cell invasion. Both 918013 and 403070 caused a concentration-dependent decrease in A2058 invasion measured after a 16-h incubation with similar potencies (Fig. 6).

image

Figure 5. (A, B) ATX hydrophobic pocket residues in complex with HA155 [orange (A)] and 403070 [yellow (B)]. (C) Traditional structure-based pharmacophore hypothesis with targeting of the catalytic site of ATX. (D) Revised ATX pharmacophore hypothesis with targeting of the hydrophobic pocket. Three-dimensional spatial arrangements and distances between the centroids are represented by solid black lines. The pharmacophore features are shown as purple (hydrogen bond donor), red (hydrogen bond acceptor), cyan (hydrophobic group), and green (aromatic ring). The numbers represent distances between these functional groups.

Download figure to PowerPoint

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. References
  9. Supporting Information

Here, we report the identification and pharmacological characterization of a novel series of ATX enzyme inhibitors with a common aromatic sulfonamide group at their core. This was accomplished by starting with an in silico search strategy, and then performing biochemical and enzyme kinetic determinations and cell-based assays. This approach led to the identification of three novel ATX inhibitors with nanomolar inhibitory activity. The presented results further explore targeting of the hydrophobic pocket of ATX as a means to competitively inhibit LPA production.

The recently solved ATX crystal structure revealed that there are multiple molecular surfaces at which small molecules may exert inhibitory actions, including: (a) the catalytic site; (b) the hydrophobic pocket; and (c) the hydrophobic channel [14, 16, 17]. Although previously described ATX inhibitors have evolved from lipid-like to nonlipid-like, the sites of interaction for many of them have not been thoroughly described and experimentally characterized. More recently, we described two new nonlipid classes of compound identified by an HTS campaign as ATX inhibitors. One class of compounds, designated as dual inhibitors (966791; Fig. 1), block the hydrolysis of both lysophospholipid-like and nucleotide-like substrates, and bind near the catalytic site. In contrast to many published ATX inhibitors, another class, designated as single inhibitors, inhibit ATX via competition with lysophospholipid-like substrates but without inhibiting the phosphodiesterase-mediated hydrolysis of the nucleotide substrate pNP-TMP. Our present data on the aromatic sulfonamide series of compounds shows an increase in pNP-TMP hydrolytic activity, suggesting that these compounds not only fail to block the catalytic site shared by LPC and pNP-TMP, but also, possibly via allosteric regulation of substrate binding and/or catalysis, enhance the phosphodiesterase activity. Our modeling studies are consistent with the finding that aromatic sulfonamide single inhibitors can bind to ATX simultaneously with pNP-TMP and enhance its hydrolysis (Fig. 4F). The comparable IC50 values of hydrophobic pocket and catalytic site inhibitors strongly suggest that inhibition of the hydrophobic pocket of ATX is sufficient to disrupt the ATX–LPA signaling axis, as shown by our A2058 invasion assay. We have also demonstrated that the aromatic sulfonamide compounds inhibit ATX via a competitive mechanism of action against the LPC-like FS-3 substrate.

In an effort to identify additional hydrophobic inhibitors, 918013 was used as a starting point for SAR studies. We began by performing a two-dimensional similarity search of the UC-DDC database, using an aromatic sulfonamide scaffold. The search resulted in 5997 hits based on homology with three different templates. Although the availability of the ATX crystal structures provides the opportunity for structure-based design, we devised a virtual screening approach to filter and rank the aromatic sulfonamide derivatives for experimental validation. Docking of analogs of 918013 showed that these all interacted in the same region of ATX (Fig. 4E). To reduce the set of 5997 hits, we applied an intersection inequality approach. This strategy reduced the set to 230 compounds, which were screened for inhibition of FS-3 hydrolysis, leading to 16 compounds with an IC50 of < 3 μm. Compound 403070 was the most potent ATX inhibitor identified in this study, with a Ki of 8.43 nm against FS-3. In addition, 390596 and 406643 also met our filtered screening criteria.

We explored the SAR of this aromatic sulfonamide scaffold by investigating whether small changes at four sites of this scaffold affect biological activity (Fig. 3). Although the selected inhibitors show structural similarity with 918013, they also have some unique features. From the SAR analysis, we determined the tolerance of the substituents at four positions and the effect of these substitutions on ATX activity. The molecular docking results are in agreement with the experimental SAR observations. The combined approach allowed identification of functional groups conserved between our most active compounds.

The ATX crystal structure allowed us to investigate the binding mode of the aromatic sulfonamide derivatives. Docking of 403070 into ATX results in an orientation similar to that of our lead compound, 918013 (Fig. 4C). The primary interactions of the aromatic sulfonamide compounds with ATX are with R3 and R4 aromatic rings, which undergo a π–π interaction with the aromatic amino acids lining the hydrophobic pocket (Figs 4D and 5B). In contrast, changes to the R1 portion of ring 1 seem to extend from the hydrophobic pocket of ATX. This R1 position appears to be a potential site for additional chemical modifications in future lead optimization studies. Thus, the present results provide insights into the structural requirements for the activity of hydrophobic pocket inhibitors and will guide the design of further derivatives.

image

Figure 6. Effect of ATX inhibitor hits on cancer cell invasion in vitro. A2058 human melanoma cells were applied to Matrigel-coated BD BioCoat chambers. ATX (0.1 nm) plus 1 μm LPC 18:1 with or without 10 μm inhibitor was applied, and invasion was quantified 16 h later. For dose–response curves, the inhibitors were applied at increasing concentrations ranging from 0.01 μm to 10 μm. The numbers shown are the means from three experiments ± SDs.

Download figure to PowerPoint

The availability of the ATX crystal structure has provided understanding of the molecular surfaces that are important for regulation of lysophospholipase and phoshodiesterase enzymatic activity. Although previous drug discovery efforts have focused on targeting the catalytic site to develop competitive ATX inhibitors (Fig. 5A), our studies continue to exploit blocking of the hydrophobic pocket with small molecules to provide a strong inhibitory effect on the catalytic activity of ATX (Figs 4 and 5) [15, 16, 19]. Current competitive ATX inhibitors (e.g. HA155; Fig. 6A) are designed to block ATX enzymatic activity by binding to the catalytic site, thus blocking both LPC and pNP-TMP binding/hydrolysis. Blocking of the hydrophobic pocket, as achieved with our aromatic sulfonamide derivatives (Fig. 4), offers new opportunities for structure-based design of small-molecule competitive ATX inhibitors, and supports the design of a more focused and structurally restricted pharmacophore model that no longer covers the entire LPC-binding site (Fig. 5C), but instead is limited to the hydrophobic binding pocket (Fig. 5D).

In summary, we have identified aromatic sulfonamide derivatives as a new class of ATX inhibitors that are potent and competitive ATX inhibitors. Hydrophobic pocket inhibitors represent new chemical tools that may be useful in the further development of ATX inhibitors. Because of their nanomolar potency and nonlipid properties, these hydrophobic pocket inhibitors are important lead compounds that may be useful as chemical tools for the further development of novel, selective ATX inhibitors. Specific in silico, biochemical and pharmacological assays targeting the hydrophobic pocket and the hydrophobic channel will need to be developed to guide the identification of novel types of ATX inhibitor.

Experimental procedures

  1. Top of page
  2. Abstract
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. References
  9. Supporting Information

Materials

The screening compound library was from the UC-DDC screening collection. The UC-DDC numbering nomenclature was used throughout this study for compound identification. The compounds were dissolved in dimethylsulfoxide at a 10 mm stock concentration and stored at −80 °C. For some studies, additional compounds were purchased from the Chembridge (San Diego, CA, USA) screening collection (http://www.hit2lead.com). The compounds were > 90% pure as certified by Hit2lead.com. FS-3 and pNP-TMP were from Echelon Biosciences (Salt Lake City, UT, USA) and Sigma-Aldrich (St Louis, MO, USA), respectively.

Expression, purification and activity of ATX

The expression, purification and enzymatic assay conditions of human ATX and its F275A mutant have been described previously [7, 17].

Virtual screening for ATX inhibitors

Similarity searching

A structural similarity search was conducted on the UC-DDC database with molecular operating environment [20] software. Specifically, the database was preprocessed with the module ‘wash’, which prepared the ligands with respect to protonation and tautomerization states, and this was followed by the generation of molecular access key fingerprints for each compound. Similarity searching was carried out independently for the three highly potent aromatic sulfonamide inhibitors from our HTS: 918013, 897131, and 934313. For the similarity search tool, the Tanimoto coefficient was used as the similarity metric, set at a minimum of 0.8. In this step, the initial virtual subset of ~ 340 000 compounds was reduced by eliminating compounds that were distant from 918013, 390596 and 406643 on the basis of their two-dimensional structural fingerprints. We analyzed the filtered databases of compounds obtained with the three parallel searches. The three filtered databases were merged to yield a library of 230 candidates for further analysis.

Molecular docking analysis

The ATX structure was obtained by downloading a crystal structure of ATX complexed with HA155 from the PDB (entry code 2XRG) into the molecular operating environment software. All water molecules, all sugars, iodoacetamide and all nonpolar hydrogen atoms were removed, but the zinc heteroatom records were retained. The search space was set as centered around the cocrystallized HA155. Before docking, structures of compounds and ATX were preoptimized with the molecular operating environment software and saved in PDB format. The ATX macromolecule and all of the compounds were imported into autodock tools [21]. Automated docking was performed with autodock vina [22].

Experimental validation of in silico hits

ATX enzymatic activity was determined by measuring the hydrolysis of LPC species, FS-3, ADMAN-LPC, or pNP-TMP. In brief, compounds were initially tested at 10 μm. The enzyme reactions were conducted in buffer consisting of 10 mm Tris/HCl (pH 7.2), 10 mm MgCl2, and 0.1% BSA. These assays contained 4 nm recombinant ATX and 1 μm FS-3. Compounds were added to the mixture to a final volume of 20 μL. Compounds that showed > 50% inhibition of ATX in the primary assay were subsequently titrated in a 10-point dose–response reaction against the FS-3 substrate. These reactions were performed with 4 nm recombinant ATX with the indicated drug concentrations in reaction buffer and 1 μm FS-3 substrate in a total volume of 20 μL in 384-well plates for 120 min at room temperature. Fluorescence was measured with a FlexStation 3 plate reader (Molecular Devices, Sunnyvale, CA, USA).

Compounds were also assayed for inhibition of ATX phosphodiesterase activity with pNP-TMP as substrate. The assay contained final concentrations of 4 nm ATX, 1 mm pNP-TMP and 10 μm compound in the assay buffer described above. Absorbance was monitored at 485 nm to measure enzyme activity for 2 h.

ADMAN-LPC at a final concentration of 30 μm was resuspended in assay buffer containing 2 mg·mL−1 BSA with or without 10 μm inhibitor and 30 nm ATX. The reaction was incubated at 37 °C for 4 h. Lipids were extracted with a modification of the Bligh and Dyer protocol by adding 3.5 volumes of citrate phosphate buffer (61.4 mm citric acid and 74.8 mm sodium phosphate, pH 4.0), before extraction with chloroform and methanol (2 : 1, v/v). Lipids were dried and resuspended in 30 μL of chlorform/methanol (1 : 1), and separated on silica gel 60 TLC plates (Merck) with CHCl3/MeOH/NH4OH (60 : 35 : 8, v/v/v). Fluorescent LPC and LPA species were visualized by UV transillumination, and quantified by densitometric analysis with NIH imagej software [23] to quantify the inhibition of LPA production by ATX.

Inhibition of ATX-mediated hydrolysis of LPCs assesed with the Amplex Red choline release assay

ATX inhibition by the selected compounds was assessed with the Amplex Red choline release assay. Triplicate wells were loaded with 60 μL of reaction cocktail in ATX assay buffer (50 mm Tris/HCl, 5 mm CaCl2, 30 μm BSA, pH 7.4), resulting in overall concentrations, per well, of 100 μm lipid substrate, 0 or 10 nm ATX, and 0 or 10 μm inhibitor. Fluorescence was read initially and after 2 h of incubation at 37 °C with excitation and emission wavelengths of 560 nm and 590 nm, respectively. Relative fluorescence was recorded as a mean value of the triplicates for each sample, and reported as percentage inhibition of ATX-mediated LPC hydrolysis.

Mechanism of inhibition

ATX activity assays were used to determine the inhibitory mechanism based on nonlinear regression analysis as described previously [24, 25]. Briefly, inhibitor concentrations of 0, 0.5 and two times the experimentally calculated IC50 value were used in conjunction with a range of FS-3 concentrations of 0.3–20 μm to determine the initial reaction rate. Background-corrected fluorescence at excitation and emission wavelengths of 485 nm and 528 nm was plotted as a function of time. A carboxyfluorescein standard curve was then used to transform the data, as carboxyfluorescein is analogous to the fluorescent product of ATX-mediated FS-3 hydrolysis, and can be correlated as such in a 1 : 1 ratio. The reaction rate was then plotted versus substrate concentration, and simultaneously fitted via nonlinear regression into the Michaelis–Menten equations for competitive, noncompetitive, uncompetitive and mixed-mode inhibition (see below) with graphpad prism v5 (GraphPad Software, San Diego, CA, USA). The mechanism of inhibition was assigned on the basis of the model providing the highest global nonlinear fit (R2) value while simultaneously accounting for graphpad's interpretation rules for the calculated α-value. According to graphpad, the α-value is a measure of affinity, such that, when α = 1, enzyme–substrate binding is not affected, and a noncompetitive mechanism is assigned. A very large α-value indicates inhibitor interference with enzyme–substrate binding, denoting a competitive model. When the α-value is very small, the inhibitor enhances the binding of enzyme and substrate, and thus an uncompetitive model is indicated. Finally, an intermediate α-value is indicative of mixed-mode inhibition. After the inhibitory mechanism had been determined, the Ki (inhibitor affinity for the enzyme) was calculated on the basis of the consequent regression analysis. In our case, the best fit was obtained when the formula for competitive inhibition was used:

  • display math

Cell invasion assay

Cell invasion was measured in 24-well invasion chambers (BD Biosciences, San Jose, CA, USA) with the Matrigel-coated film insert (8-μm pore size), as described previously [7]. In brief, A2058 cells (5 × 104), which were resuspended in serum-free DMEM supplemented with 0.1% BSA, were added to the top compartment of the invasion chamber. The various compounds were preincubated in serum-free DMEM/0.1% BSA with recombinant ATX for 30 min at 37 °C, and this was followed by the addition of 1 μm LPC 18:1 to the bottom chamber. The invasion chambers were incubated at 37 °C in 5% CO2 for 16 h. Then, the filter inserts were removed from the wells and transferred to a new 24-well plate containing 4 μg·mL−1 of calcein-AM (Molecular Probes, Invitrogen) in Hank's balanced salt solution. After a 1-h incubation, the fluorescence of invaded cells was measured with a FLEXStation 3 plate reader at excitation and emission wavelengths of 485 nm and 530 nm, respectively.

Statistical data analysis

The data were plotted with prism version 4 or 5 (GraphPad Software, San Diego, CA, USA), and presented as mean ± standard deviation (SD). For compounds that behaved as competitive inhibitors, the Ki was calculated by fitting the data at various FS-3 concentrations and comparing them with the value obtained via the Cheng–Prussoff equation.

Acknowledgements

  1. Top of page
  2. Abstract
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. References
  9. Supporting Information

This study was supported by the National Institutes of Health National Cancer Institute (Grant CA092160), the American Cancer Society (Grant 122059-PF-12-107-01-CDD), and the Van Vleet Endowment.

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. References
  9. Supporting Information
  • 1
    Houben AJ & Moolenaar WH (2011) Autotaxin and LPA receptor signaling in cancer. Cancer Metastasis Rev 30, 557565.
  • 2
    Nakanaga K, Hama K & Aoki J (2010) Autotaxin – an LPA producing enzyme with diverse functions. J Biochem 148, 1324.
  • 3
    Okudaira S, Yukiura H & Aoki J (2010) Biological roles of lysophosphatidic acid signaling through its production by autotaxin. Biochimie 92, 698706.
  • 4
    Liu S, Murph M, Panupinthu N & Mills GB (2009) ATX–LPA receptor axis in inflammation and cancer. Cell Cycle 8, 36953701.
  • 5
    van Meeteren LA & Moolenaar WH (2007) Regulation and biological activities of the autotaxin–LPA axis. Prog Lipid Res 46, 145160.
  • 6
    St-Coeur PD, Ferguson D, Morin P Jr & Touaibia M (2013) PF-8380 and closely related analogs: synthesis and structure–activity relationship towards autotaxin inhibition and glioma cell viability. Arch Pharm (Weinheim) 346, 9197.
  • 7
    Gupte R, Patil R, Liu J, Wang Y, Lee C, Fujiwara Y, Fells J, Bolen AL, Emmons-Thompson K, Yates CR, et al. (2011) Benzyl and naphthalene methylphosphonic acid inhibitors of autotaxin with anti-invasive and anti-metastatic activity. Chem Med Chem 6, 922935.
  • 8
    Gierse J, Thorarensen A, Beltey K, Bradshaw-Pierce E, Cortes-Burgos L, Hall T, Johnston A, Murphy M, Nemirovskiy O, Ogawa S et al. (2010) A novel autotaxin inhibitor reduces lysophosphatidic acid levels in plasma and the site of inflammation. J Pharmacol Exp Ther 334, 310317.
  • 9
    Albers HM, van Meeteren LA, Egan DA, van Tilburg EW, Moolenaar WH & Ovaa H (2010) Discovery and optimization of boronic acid based inhibitors of autotaxin. J Med Chem 53, 49584967.
  • 10
    Ferry G, Moulharat N, Pradere JP, Desos P, Try A, Genton A, Giganti A, Beucher-Gaudin M, Lonchampt M, Bertrand M et al. (2008) S32826, a nanomolar inhibitor of autotaxin: discovery, synthesis and applications as a pharmacological tool. J Pharmacol Exp Ther 327, 809819.
  • 11
    Kawaguchi M, Okabe T, Okudaira S, Nishimasu H, Ishitani R, Kojima H, Nureki O, Aoki J & Nagano T (2013) Screening and X-ray crystal structure-based optimization of autotaxin (ENPP2) inhibitors, using a newly developed fluorescence probe. ACS Chem Biol 8, 17131721.
  • 12
    Albers HM & Ovaa H (2012) Chemical evolution of autotaxin inhibitors. Chem Rev 112, 25932603.
  • 13
    Lipinski CA, Lombardo F, Dominy BW & Feeney PJ (2001) Experimental and computational approaches to estimate solubility and permeability in drug discovery and development settings. Adv Drug Deliv Rev 46, 326.
  • 14
    Nishimasu H, Ishitani R, Aoki J & Nureki O (2012) A 3D view of autotaxin. Trends Pharmacol Sci 33, 138145.
  • 15
    Nishimasu H, Okudaira S, Hama K, Mihara E, Dohmae N, Inoue A, Ishitani R, Takagi J, Aoki J & Nureki O (2011) Crystal structure of autotaxin and insight into GPCR activation by lipid mediators. Nat Struct Mol Biol 18, 205212.
  • 16
    Hausmann J, Kamtekar S, Christodoulou E, Day JE, Wu T, Fulkerson Z, Albers HM, van Meeteren LA, Houben AJ, van Zeijl L et al. (2011) Structural basis of substrate discrimination and integrin binding by autotaxin. Nat Struct Mol Biol 18, 198204.
  • 17
    Fells JI, Lee S-C, Fujiwara Y, Norman DD, Lim KG, Tsukahara R, Liu J, Patil R, Miller DD, Kirby RJ et al. (2013) Hits of a high-throughput screen identify the hydrophobic pocket of autotaxin/lysophospholipase D as an inhibitory surface. Mol Pharm 84, 415424.
  • 18
    Willett P, Barnard JM & Downs GM (1998) Chemical similarity searching. J Chem Inf Comput Sci 38, 983996.
  • 19
    Koyama M, Nishimasu H, Ishitani R & Nureki O (2012) Molecular dynamics simulation of autotaxin: roles of the nuclease-like domain and the glycan modification. J Phys Chem B 116, 1179811808.
  • 20
    Chemical Computing Group. Molecular Operating Environment (MOE) Software. Chemical Computing Group, Montreal.
  • 21
    Morris GM, Huey R, Lindstrom W, Sanner MF, Belew RK, Goodsell DS & Olson AJ (2009) AutoDock4 and AutoDockTools4: automated docking with selective receptor flexibility. J Comput Chem 30, 27852791.
  • 22
    Trott O & Olson AJ (2010) AutoDock Vina: improving the speed and accuracy of docking with a new scoring function, efficient optimization, and multithreading. J Comput Chem 31, 455461.
  • 23
    Schneider CA, Rasband WS & Eliceiri KW (2012) NIH Image to ImageJ: 25 years of image analysis. Nat Methods 9, 671675.
  • 24
    Hoeglund AB, Bostic HE, Howard AL, Wanjala IW, Best MD, Baker DL & Parrill AL (2010) Optimization of a pipemidic acid autotaxin inhibitor. J Med Chem 53, 10561066.
  • 25
    North EJ, Howard AL, Wanjala IW, Pham TC, Baker DL & Parrill AL (2010) Pharmacophore development and application toward the identification of novel, small-molecule autotaxin inhibitors. J Med Chem 53, 30953105.

Supporting Information

  1. Top of page
  2. Abstract
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. References
  9. Supporting Information
FilenameFormatSizeDescription
febs12674-sup-0001-FigS1-TableS1-S2.zipapplication/ZIP275K

Fig. S1. Structures of the 38 active ATX inhibitors identified in this study.

Table S1. Per cent inhibition of FS-3 or pNP-TMP hydrolysis of ATX by a 10 μM concentration of the in silico hits.

Table S2. IC50 values against the FS-3 substrate for the most potent ATX inhibitors identified in this study.

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.