cancers

Review The Structural Binding Mode of the Four Autotaxin Inhibitor Types that Differentially Affect Catalytic and Non-Catalytic Functions

Fernando Salgado-Polo and Anastassis Perrakis *

Oncode Institute and Division of Biochemistry, Netherlands Cancer Institute, Plesmanlaan 121, 1066 CX Amsterdam, The Netherlands; [email protected] * Correspondence: [email protected]



Received: 10 September 2019; Accepted: 8 October 2019; Published: 16 October 2019


Abstract: Autotaxin (ATX) is a secreted lysophospholipase D, catalysing the conversion of lysophosphatidylcholine (LPC) to bioactive lysophosphatidic acid (LPA). LPA acts through two families of G protein-coupled receptors (GPCRs) controlling key cellular responses, and it is implicated in many physiological processes and pathologies. ATX, therefore, has been established as an important drug target in the pharmaceutical industry. Structural and biochemical studies of ATX have shown that it has a bimetallic nucleophilic catalytic site, a substrate-binding (orthosteric) hydrophobic pocket that accommodates the lipid alkyl chain, and an allosteric tunnel that can accommodate various steroids and LPA. In this review, first, we revisit what is known about ATX-mediated catalysis, crucially in light of allosteric regulation. Then, we present the known ATX catalysis-independent functions, including binding to cell surface integrins and proteoglycans. Next, we analyse all crystal structures of ATX bound to inhibitors and present them based on the four inhibitor types that are established based on the binding to the orthosteric and/or the allosteric site. Finally, in light of these data we discuss how mechanistic differences might differentially modulate the activity of the ATX-LPA signalling axis, and clinical applications including cancer.

Keywords: lysophosphatidic acid; autotaxin; inhibitor; allosteric; orthosteric; lipid chaperone; signalling; GPCR

1. Introduction Lysophosphatidic acid (1- or 2-acyl-sn-glycero-3-phosphate or LPA) is a bioactive lipid found in many body fluids and involved in many physiological and pathological processes. Historically, LPA had been identified as a growth factor in serum that could induce motility in fibroblasts and cancer cells through G protein-coupled receptors (GPCRs) [1,2]. Subsequent research identified specific LPA GPCRs (LPA1–6), which have distinct expression patterns [3]. Deregulation of the LPA signalling axis has been linked to different diseases, such liver disease [4], fibrosis [5], pruritus [6], multiple sclerosis [7], inflammation, and cancer [8,9]. The LPA receptors are classified into distinct families: the endothelial cell differentiation gene (EDG) (LPA1–3) and non-EDG (LPA4–6) families. The crystallographic structures of LPA1 and LPA6 have provided the field with key mechanistic indications with respect to their ligand binding mode [10,11]. Namely, a structural comparison between the LPA1 and LPA6 substrate-binding sites has indicated contrasting LPA binding modes from the extracellular milieu or the plasma membrane, respectively. The former is consistent with a model by which LPA is carried by a lipid chaperone, such as albumin, to bind to the flexible N-terminal domain of LPA1 and deliver LPA specifically [10], whereas the latter would not require lipid-carrying molecules [12,13].

Cancers 2019, 11, 1577; doi:10.3390/cancers11101577 www.mdpi.com/journal/cancers Cancers 2019, 11, 1577 2 of 17 Cancers 2019, 11, x 2 of 17

Since LPA can can promote promote a a plethora plethora of of downstream downstream signalling signalling events, events, both both its production its production and anddegradation degradation are tightly are tightly regulated, regulated, resulting resulting in an estimated in an estimated half‐life half-life of approximately of approximately three minutes three minutesin circulation in circulation [14–16]. Such [14– 16a short]. Such‐lived a short-lived existence is existence due to its is fast due degradation to its fast degradation by three membrane by three‐ membrane-boundbound lipid phosphate lipid phosphatephosphatases phosphatases (LPPs), which (LPPs), cleave which the cleave LPA thephosphate LPA phosphate group and group release and releasesignalling signalling-inactive‐inactive monoacylglycerol monoacylglycerol [17,18]. Contrary [17,18]. Contrary to this, LPA to this, production LPA production originates originates from the fromfollowing the following two sources: two phosphatidic sources: phosphatidic acid hydrolysis acid by hydrolysis PLA1/2 and by from PLA the1/2 and enzymatic from the conversion enzymatic of conversionlysophosphatidylcholine of lysophosphatidylcholine (1‐ or 2‐acyl (1-‐sn or‐glycero 2-acyl-sn‐3-glycero-3-phosphocholine‐phosphocholine or LPC) or to LPC) LPA to LPAby bya alysophospholipase lysophospholipase D D (lysoPLD) (lysoPLD) that that has has been been established established to to be be Autotaxin Autotaxin (ATX) (ATX) (Figure (Figure 11))[ [19,20].19,20]. ATX-catalysedATX‐catalysed production production constitutes constitutes the the main main physiological physiological source source of extracellular of extracellular LPA, and LPA, therefore and ATXtherefore has been ATX widely has been studied widely as studied a target as for a drugtarget development for drug development [14,21,22]. [14,21,22].

Figure 1.1. Distinct modes for lysophosphatidic acid (LPA) binding to itsits cognatecognate GG protein-coupledprotein‐coupled receptorsreceptors (GPCRs)(GPCRs) proposed based on theirtheir crystallographiccrystallographic structures.structures. Autotaxin (ATX) isis thethe main

producerproducer ofof LPA,LPA, whichwhich cancan thenthen bindbind toto thethe extracellularlyextracellularly openopen lipid-bindinglipid‐binding pocketpocket ofof LPALPA1–31–3, potentiallypotentially assisted assisted by by lipid lipid chaperones, chaperones, or di orff usediffuse laterally laterally towards towards the membrane-open the membrane ligand-binding‐open ligand‐

sitebinding of LPA site4–6 of. LPA4–6.

ATX isis the the only only member member of the of ectonucleotidethe ectonucleotide pyrophosphatase pyrophosphatase/phosphodiesterase/phosphodiesterase family (ENPP)family that(ENPP) presents that presents lysoPLD lysoPLD activity (EC activity 3.1.4.39) (EC over3.1.4.39) lysophospholipids over lysophospholipids [22]. ATX [22]. is first ATX translated is first astranslated a pre-proenyme as a pre‐ thatproenyme undergoes that twoundergoes proteolytic two processing proteolytic steps, processing resulting steps, in a resulting mature, glycosylated in a mature, andglycosylated secreted formand [23secreted]. The determinationform [23]. The of thedetermination structure of ATXof the by X-raystructure crystallography of ATX by enabled X‐ray thecrystallography determination enabled of its domain the determination organization, of i.e.,its thedomain two N-terminalorganization, somatomedin i.e., the two B N (SMB)-like‐terminal domainssomatomedin are followed B (SMB) by‐like a central domains catalytic are followed phosphodiesterase by a central (PDE) catalytic domain, phosphodiesterase which is adjacent (PDE) to an inactivedomain, nuclease-likewhich is adjacent domain. to Substratean inactive hydrolysis nuclease‐ requireslike domain. a bimetallic Substrate active hydrolysis site containing requires two a Znbimetallic2+ ions active and a site threonine containing nucleophile, two Zn2+ ions which and act a threonine in an associative nucleophile, two-step which in-line act in displacementan associative catalytictwo‐step mechanism in‐line displacement [24]. catalytic mechanism [24]. Structural studiesstudies have have also also established established that that ATX ATX has a has unique a unique tripartite tripartite binding binding site. The site. catalytic The bimetalliccatalytic bimetallic site is next site to is a next hydrophilic to a hydrophilic shallow grooveshallow that groove accommodates that accommodates the hydrophilic the hydrophilic glycerol moietyglycerol of moiety lipid substrates. of lipid substrates. This groove This is connectedgroove is connected by a T-junction by a T to‐junction a hydrophobic to a hydrophobic pocket where pocket acyl chainswhere canacyl bind, chains and can a tunnel bind, (oftenand a calledtunnel the (often “hydrophobic called the channel”)“hydrophobic that leads channel”) to the that other leads side to of the PDEother domain side of [ 25the]. PDE It is noteworthy domain [25]. that It theis noteworthy tunnel (or channel) that the is tunnel only partially (or channel) hydrophobic is only inpartially nature andhydrophobic has hydrophilic in nature patches, and has unlike hydrophilic the pocket patches, (Figure unlike2). The the tunnel pocket binds (Figure steroid 2). The molecules tunnel binds [ 26], assteroid well asmolecules the LPA [26], product as well [27, 28as], the which LPA results product in a[27,28], modulation which of results catalytic in a e ffimodulationciency. Thus, of catalytic we refer toefficiency. the tunnel Thus, as the we allostericrefer to the site, tunnel while as we the refer allosteric to the site, substrate-binding, while we refer to hydrophilic the substrate groove‐binding, and thehydrophilic hydrophobic groove pocket, and the as the hydrophobic orthosteric pocket, site (Figure as the3). orthosteric site (Figure 3).

Cancers 2019, 11, 1577 3 of 17 CancersCancers 2019 2019, ,11 11, ,x x 33 ofof 1717

FigureFigure 2. 2.2. HydrophobicityHydrophobicity comparisoncomparison betweenbetween between thethe pocketpocket and and the the tunnel. tunnel.tunnel. SideSide Side sectionssections sections ofof ATXATX (Protein(Protein(Protein Data Data Bank Bank (PDB): (PDB):(PDB): 5dlw) 5dlw)5dlw) showing showingshowing the thethe modes modesmodes of of bindingof bindingbinding of LPA ofof LPALPA and andand tauroursodeoxycholic tauroursodeoxycholictauroursodeoxycholic acid acid(TUDCA)acid (TUDCA)(TUDCA) in the inin orthosteric thethe orthostericorthosteric and allosteric andand allostericallosteric sites, respectively. sites,sites, respectively.respectively. The partly TheThe hydrophobic partlypartly hydrophobichydrophobic tunnel presents tunneltunnel presentshydrophilicpresents hydrophilichydrophilic patches that patchespatches accommodate thatthat accommodateaccommodate the polar moieties thethe polarpolar of moieties TUDCA.moieties of Proteinof TUDCA.TUDCA. surface ProteinProtein was colored surfacesurface from waswas coloredorangecolored tofrom from turquois orange orange using to to turquois turquois ChimeraX using using (version ChimeraX ChimeraX 0.91). (version (version 0.91). 0.91).

FigureFigure 3. 3.3. Key Key residues residuesresidues involved involved inin in binding bindingbinding the thethe four fourfour classes classes classes of ofof ATX ATXATX inhibitors. inhibitors. inhibitors. Cartoon CartoonCartoon depiction depictiondepiction of the ofof theATXthe ATXATX tripartite tripartitetripartite site andsitesite and theand crucial thethe crucialcrucial interacting interactinginteracting residues residuesresidues for inhibitor forfor inhibitorinhibitor binding. binding.binding. The dashed TheThe dasheddashed lines depict lineslines depictthedepict binding the the binding binding site for site site lysophosphatidylcholine for for lysophosphatidylcholine lysophosphatidylcholine (LPC) or (LPC) (LPC) LPA or ator LPA theLPA orthosteric at at the the orthosteric orthosteric site, as wellsite, site, as as thewell well binding as as the the bindingsitebinding for steroidssite site for for steroids atsteroids the tunnel. at at the the tunnel. tunnel.

InIn this this review, review, we wewe will willwill first firstfirst review reviewreview the thethe catalytic catalyticcatalytic mechanism mechanism of of ATX, ATX, especially especiallyespecially in in light light of of the the allostericallosteric modulation modulation we wewe have havehave recently recentlyrecently described, described,described, discuss discussdiscuss the thethe non-catalytic nonnon‐‐catalyticcatalytic functions functionsfunctions of ATX, ofof ATX,ATX, and and howand howthesehow thesethese are involved areare involvedinvolved in the inin ATX-LPA thethe ATXATX‐‐LPA signallingLPA signallingsignalling axis. axis.axis. Then, Then,Then, we will wewe presentwillwill presentpresent the four thethe four familiesfour familiesfamilies of ATX ofof ATXinhibitorsATX inhibitorsinhibitors from afromfrom structural aa structuralstructural biology biologybiology perspective, perspective,perspective, as they areasas they classifiedthey areare classified dependingclassified dependingdepending on their occupancy onon theirtheir occupancyofoccupancy the orthosteric of of the the orthosteric andorthosteric/or the allostericand/or and/or the the site. allosteric allosteric Finally, site. site. we Finally, will Finally, discuss we we will will how discuss discuss the diff how erenthow the the types different different of inhibitors types types ofof inhibitors inhibitors might might interfere interfere with with catalytic catalytic and and non non‐‐catalyticcatalytic functions functions to to differentially differentially affect affect the the ATX ATX‐‐ LPALPA signalling signalling axis axis in in vivo. vivo.

Cancers 2019, 11, 1577 4 of 17 might interfere with catalytic and non-catalytic functions to differentially affect the ATX-LPA signalling axis in vivo.

2. Autotaxin Catalytic and Non-Catalytic Functions ATX has long been established as the lysoPLD that converts LPC into LPA, but the exact catalytic mechanism remained a subject of study and debate. The first complete characterization of ATX catalysis showed that LPC binding was slow and rate limiting and offered clear evidence for a model where catalysis first results in choline release, which is followed by the slow release of nitrobenzoxadiazole (NBD)-LPA (tens of seconds) [29]. More recently, we have observed an approximately 10 min lag phase in time-course measurements of ATX activity, which could be alleviated by the addition of external LPA [28]. We proceeded to show that LPA binding increases the turnover rate (kcat) of LPC hydrolysis and promotes its own production. Specifically, our results established that binding of LPA takes place at the low-affinity (~1 µM) allosteric tunnel site. This binding is consistent with earlier results [27] that attributed residual electron density in crystallographic structures of mouse autotaxin to LPA (16:0, 18:0, 18:1, 18:3 and 22:6). A recent study indicated that active ATX can bind to the surface of secreted exosomes and carry LPA [30]. Such an event could hypothetically lead to LPA-bound ATX at the cell surface, after which ATX would release LPA and activate LPARs. Indeed, the authors indicated that it was by this mechanism that ATX yielded activation of LPA1 and LPA3 in the employed in vitro experimental setup [30]. However, molecular dynamics simulations and kinetic modelling [28] argue that the presence of LPA in the ATX allosteric tunnel is an independent event, which does not represent an exit pathway of produced LPA by LPC hydrolysis in the adjacent orthosteric site. The source of the LPC substrate poses another exciting question, that is, its rate of production, lifetime in circulation, and local concentrations in specific organs, which have received rather limited attention compared to the importance of the question. Even though the tunnel does not play a critical role in recruiting LPC substrates from BSA or detergent micelles [27], introduction of different tunnel-occluding loop insertions, based on those present in ENPP1 and ENPP3, resulted in mutants with much impaired cell motility-stimulating activity. Although these observations appear contradictory, the loop insertions could have induced structural changes destabilizing the orthosteric site and resulting in catalytically inactive ATX without just occluding substrate trafficking through the tunnel. Taken together, the evidence is consistent with the role of ATX as an LPA carrier, transporting LPA to distal locations from those where LPC can be taken. LPC is present in the blood at high concentrations (100–200 µM[19]); most of it is bound to serum albumin and cannot be “extracted” by ATX because of their similar affinities (approximately 1 µM) [31], while a small fraction (about 1 µM) is free and can be recognized by ATX [20]. Besides its catalysis-dependent functions related to LPA production, ATX also mediates diverse cell signalling events through binding to integrins and heparan sulfate proteoglycans (HSPGs). On the one hand, the more abundant ATXβ isoform binds to integrins, which in turn promotes cell proliferation [32] and directional cell motility [33]. This interaction takes place through the ATX SMB domains, which have high structural similarity with the SMB domain of the cell adhesion factor vitronectin. Specifically, ATXβ can interact with αvβ3 or αIIbβ3 integrins, both of which cause platelet activation upon interaction with ATX [25,34,35]. Accordingly, integrin binding enables uptake and intracellular sequestration of ATX, which redistributes to the front of the migrating cells, however, blockade of integrin binding did not abolish cell migration completely. Interestingly, cell stimulation with only the ATX SMB domains promotes directional cell migration independently of lysoPLD activity [33]. On the other hand, the less abundant ATXα isoform (but not ATXβ) binds to HSPGs through a polybasic insertion [36]. This interaction results in an increase of ATXα catalytic turnover, which may elicit a membrane-anchored burst of LPA production and downstream signalling. Even though this ATXα-specific function lacks further characterization, it has been recently reported that ATXβ can interact in an SMB-independent manner with the HSPG syndecan-4, which affects Cancers 2019, 11, 1577 5 of 17 cell proliferation and cellular metastatic potential [32]. Thus, accumulating evidence suggests several mechanisms for cell receptor binding in the ATX structure, highlighting the relevance of plasma membrane recruitment.

CancersAnchoring 2019, 11, x on the cell surface is of critical importance in light of the crystallographic structures5 of 17 of LPA1 and LPA6, which suggested novel mechanisms with respect to their ligand binding modes[10,11].[ 10The,11 ]LPA. The1 substrate LPA1 substrate-binding‐binding site is located site is located inside insidea central a central globular globular cavity cavitycapped capped by an byextracellular an extracellular N‐terminal N-terminal helical helical lid. The lid. large The (also large named (also named“baggy”) “baggy”) ligand‐binding ligand-binding pocket remains pocket remainsclosed at closed the membrane at the membrane side, indicating side, indicating it is solely it is solelyaccessible accessible from fromthe extracellular the extracellular space. space. This Thisenables enables not only not only LPA LPA binding, binding, but butalso also potentially potentially phosphorylated phosphorylated endocannabinoids endocannabinoids binding binding and andactivation activation of LPA of LPA1. Such1. Such an anextracellular extracellular lipid lipid-binding‐binding site site capped capped by by flexible flexible helices helices is consistentconsistent withwith aa modelmodel byby whichwhich LPALPA is is carried carried by by a a lipid lipid chaperone, chaperone, suchsuch asas albumin,albumin, whichwhich bindsbinds toto thethe flexibleflexible N-terminal N‐terminal domain domain of of LPA LPA1 and1 and delivers delivers LPA LPA specifically specifically [10]. [10]. Conversely, Conversely, the zebrafish the zebrafish LPA6 structureLPA6 structure [11] showed [11] showed a much smallera much ligand-binding smaller ligand site,‐binding that is site, not openthat is towards not open the extracellulartowards the milieu,extracellular restricting milieu, the restricting receptor’s the specificity receptor’s solely specificity to LPA [ 11solely]. The to LPA LPA6 structure[11]. The isLPA consistent6 structure with is aconsistent binding mode with a by binding lateral mode diffusion by lateral of the diffusion LPA in the of the plasma LPA in membrane, the plasma which membrane, would which not require would lipid-carryingnot require lipid molecules‐carrying to molecules take place to [12 take,13]. place [12,13].

3.3. TheThe AutotaxinAutotaxin InhibitorInhibitor FamilyFamily Initially,Initially, classical classical ATXATX inhibitors inhibitors relied relied on on lipid lipid analogues analogues targetingtargeting ATXATX based based on on similarities similarities withwith sphingosine-1-phosphate sphingosine‐1‐phosphate (S1P) (S1P) or LPA.or LPA. Activity-based Activity‐based lead discovery lead discovery campaigns, campaigns, using artificial using substratesartificial substrates as activity as reporters,activity reporters, subsequently subsequently made importantmade important contributions, contributions, many many of which of which are reviewedare reviewed by Castagna by Castagna et al. et al. and and Nikolaou Nikolaou et alet al [37 [37,38].,38]. The The structural structural characterization characterization of of rat rat and and mousemouse ATXATX structures structures in in 2011 2011 [ 25[25,27],,27], enabled enabled a a remarkable remarkable potential potential for for selective selective inhibitor inhibitor designdesign byby focusingfocusing onon thethe three-dimensionalthree‐dimensional architecturearchitecture ofof thethe ATXATX active active site. site. WhileWhile attentionattention initiallyinitially focusedfocused on on the the lipid-binding lipid‐binding pocket pocket and and the catalyticthe catalytic site, current site, current emphasis emphasis is on the is so-called on the tripartiteso‐called sitetripartite that we site introduced that we introduced earlier (Figure earlier3). (Figure 3). TheThe tripartite tripartite ATX ATX binding binding site representssite represents a remarkable a remarkable potential potential for selective for designselective of inhibitors.design of Overinhibitors. time, Over ATX time, inhibitors ATX inhibitors of a distinct of a chemicaldistinct chemical nature werenature designed, were designed, including including lipid-based lipid‐ inhibitorsbased inhibitors [39], DNA [39], aptamers DNA aptamers [40], as [40], well as smallwell as molecules. small molecules. The last The group, last thegroup, focus the of focus this review, of this canreview, be in can turn be classified in turn classified in four distinct in four types distinct (I, II,types III and(I, II, IV) III depending and IV) depending on their modeon their of bindingmode of tobinding the ATX to tripartitethe ATX tripartite site (Figure site4)[ (Figure41]. 4) [41].

FigureFigure 4.4. ClassificationClassification of of the the four four reported reported ATX ATX inhibitor inhibitor types types based based on their on their binding binding modes. modes. Left: Leftschematic: schematic view of view the binding of the bindingmodes; right modes;: crystallographicright: crystallographic structures of structures ATX bound of to ATX compounds bound tofrom compounds each type; from from each top type;to bottomfrom: topPF‐8380 to bottom (5l0k),: PF-8380PAT‐352 (5l0k), (4zg9), PAT-352 TUDCA (4zg9), (5dlw), TUDCA and FP‐ (5dlw),Cpd 17 and(5m0m). FP-Cpd 17 (5m0m).

In this review we focus on compounds for which there is a crystal structure available. We present, by inhibitor type, the common interactions with ATX (Table 1), known application in in vivo models (Table 2), and discuss their properties in light of their experimentally determined binding pose. It should be noted that due to the unclear naming of some inhibitors as “compound #” in the literature, we will refer to those as “FL‐Cpd‐#”, where FL stands for the first and last names of the publication’s first author.

Cancers 2019, 11, 1577 6 of 17

In this review we focus on compounds for which there is a crystal structure available. We present, by inhibitor type, the common interactions with ATX (Table1), known application in in vivo models (Table2), and discuss their properties in light of their experimentally determined binding pose. It should be noted that due to the unclear naming of some inhibitors as “compound #” in the literature, we will refer to those as “FL-Cpd-#”, where FL stands for the first and last names of the publication’s first author.

Table 1. Common interactions needed in distinct types of Autotaxin (ATX) inhibitors.

Residues Establishing Ligand Contacts (Rat ATX) Type Active Site-Hydrophilic Groove Hydrophobic Pocket Allosteric Tunnel Common Thr209, Asp311, His474 - - I Frequent His315, His359 Leu213, Phe273, Phe274 *, Trp275 - Common - Leu213, Phe274 *, Trp276 - II Frequent - Phe273, Tyr306 - Common - - Lys248, Phe249, Trp254, Trp260 III Frequent - - Phe274 * Common - Leu213, Phe273, Trp275, Tyr306 Phe249, Trp260 IV Frequent - Phe274 *, Phe210 His251, Trp254, Phe274 * * Phe274 sidechain has two predominant conformers at the pocket-tunnel boundary depending on the interacting ligand.

Table 2. ATX inhibitors employed in in vivo models. Further details discussed in [21,42,43].

Type Inhibitor Disease LPA Inhibition References Inflammation I SBJ-Cpd 1 >50% [7] Multiple sclerosis Glioblastoma >90% [44] I PF-8380 Liver fibrosis >90% [4] I GK442 Pulmonary fibrosis [45] I BMP22 Melanoma metastasis 50% [46] Pulmonary fibrosis [47] II GWJ-A-23 50% inflammation [48] III PharmAkea -Cpd A-E Metabolic disorder 25–35% [49] Liver fibrosis III PAT-505, PAT-048 40–80% [50] Skin fibrosis Pulmonary fibrosis 84–95% [51] IV GLPG1690 Clinical trials in IPF patients 84–95% [52] Breast cancer >60% [53] ? ONO-8430506 Thyroid cancer >70% [54]

3.1. Type I Inhibitors Type I compounds occupy the orthosteric site, mimicking the LPC substrate mode of binding. As such, they include competitive inhibitors with a long and flexible structure that occupies the catalytic site, the shallow groove, and the hydrophobic pocket. Several analogues based on LPA have naturally been put forward as ATX inhibitors such as: cyclic phosphatidic acid or cPA (half maximal inhibitory concentration (IC50) = 0.14 µM, bis-pNPP) [55], thiophosphate (IC50 = 0.6 µM, bis-pNPP) [56], α-bromomethylene phosphonates like BrP-LPA (IC50 = 0.7–1.6 µM, LPC) [57], the synthetic analogue for S1P FTY720-P (inhibitor constant (Ki) = 0.2 µM, Bis-pNPP) [58], and the most potent lipid-based inhibitor S32826 (IC50 = 5.6 nM, LPC) [59]. The latter showed poor in vivo stability, which prevented it from further use in animal models. The first small molecule type I inhibitors were thiazolidinediones discovered in a screen using the artificial substrate CPF4 [14,25]. Optimization led to the identification of HA-130 (IC50 = 28 nM, LPC), a boronic acid-based inhibitor that could attack the nucleophilic oxygen at the catalytic threonine, and it was able to hamper LPA production both in vivo and in vitro [14]. Moreover, a positional isomer of HA-130, HA-155 (IC50 = 5.7 nM, LPC) (Table3), was co-crystallized with ATX, which revealed interactions with Thr209, Asp311, His359 and His474 at the active site, as well as Leu213 and Cancers 2019, 11, x 7 of 17 Cancers 2019, 11, x 7 of 17 structureCancers 2019‐activity, 11, x improvement studies based on the HA‐155 crystal structure. These consisted7 of in17 Cancerstestingstructure 2019 the‐activity, 11different, x improvement linkers added studies to the basedthiazolidine on the‐2,4 HA‐dione‐155 corecrystal [60]. structure. These consisted7 of in17 Cancerstestingstructure 2019 the‐activity, 11different, x improvement linkers added studies to the basedthiazolidine on the‐2,4 HA‐dione‐155 crystalcore [60]. structure. These consisted7 of in17 CancersTable 3.2019 Type, 11, I 1577 ATX inhibitors based on their kinetic and crystallographic analysis. * denotes that TG‐ 7 of 17 structureCancerstesting 2019 the‐activity, 11different, x improvement linkers added studies to the basedthiazolidine on the‐2,4 HA‐dione‐155 crystalcore [60]. structure. These consisted7 of in17 mTMPTable 3. was Type used I ATX as the inhibitors substrate based for reportingon their kinetic the IC and50. crystallographic analysis. * denotes that TG‐ CancerstestingstructureCancers 20192019 the‐,activity, 11different11,, xx improvement linkers added studies to the basedthiazolidine on the‐2,4 HA‐dione‐155 corecrystal [60]. structure. These consisted77 ofof 17in17 structuretestingPhe274mTMPTable the‐activity 3. differentwas in Type the used Iimprovement hydrophobicATX linkersas the inhibitors substrate added pocket basedstudies tofor the reportingon (Figure basedthiazolidinetheir kinetic 3theon, Table ICthe and50‐2,4. 1HA). crystallographic‐dione This‐155 series crystalcore [60]. was structure.analysis. subjected * denotesThese to structure-activity thatconsisted TG‐ in structuretestingstructureimprovementTablemTMP the‐‐activityactivity 3. different was Type used Iimprovement improvementATX studies linkersas theinhibitors substrate based added basedstudiesstudies on tofor the the onreporting HA-155 basedbasedtheirthiazolidine kinetic theonon crystal ICthethe and50‐2,4 . HAHA structure. crystallographic‐dione‐‐155155 crystalcorecrystal These [60]. structure.analysis.structure. consisted * denotes TheseThese in testing thatconsistedconsisted TG the‐ di ffininerent Type I Inhibitor PDB ID Activity (IC50) Reference mTMP was used as the substrate for reporting the IC50. testingtestinglinkersTable thethe 3.differentdifferent added Type I toATX linkerslinkers the inhibitors thiazolidine-2,4-dione addedadded based toto thethe on thiazolidinetheirthiazolidine kinetic core [and‐60‐2,42,4]. crystallographic‐‐dionedione corecore [60].[60]. analysis. * denotes that TG‐ Type I Inhibitor PDB ID Activity (IC50) Reference TablemTMP 3. was Type used I ATX as the inhibitors substrate based for reportingon their kinetic the IC and50. crystallographic analysis. * denotes that TG‐

Type I Inhibitor PDB50 ID Activity (IC50) Reference TablemTMPTableTable 3.3. was TypeType 3. used TypeII ATXATX as I the inhibitorsinhibitors ATX substrate inhibitors basedbased for basedonreportingonO theirtheir on kinetickinetic theirthe IC kinetic andand. crystallographiccrystallographic and crystallographic analysis.analysis. analysis. ** denotesdenotes * thatthat denotes TGTG‐‐ that 50 mTMP was used as the substrateType for I reportingInhibitor the PDBIC50. ID Activity (IC ) Reference mTMPTG-mTMP was used was as the used substrate as the substrate for reportingON for reporting the IC50. the IC50. O S 50 HO Type I Inhibitor PDB ID Activity (IC ) Reference O B ON HO 2xrg 5.7 nM [25,60] O Type SI Inhibitor PDB ID Activity (IC50) Reference HO OH O B ON HO F 2xrg 5.7 nM 50 [25,60] O TypeType ISI InhibitorInhibitor PDBPDB IDID ActivityActivity (IC(IC50)) ReferenceReference HO OH Type I Inhibitor PDB ID Activity (IC50) Reference ON B O HO O S F 2xrg 5.7 nM [25,60] HO OH HA‐155 O B ON HO O S F 2xrg 5.7 nM [25,60] HO OH Cl O O HA‐155 O N B O HO O S F 2xrg 5.7 nM [25,60] HO OH O Cl HAN‐155 O N B ON O S F O HO S 2xrg 5.7 nM [25,60] HO O ON HO OH NO Cl HA‐N155 H 510k 1.7 nM [61,62] BCl O O HO B O O 2xrg 2xrg5.7 nM 5.7 nM[25,60] [25,60] HO F 2xrg 5.7 nM [25,60] OH O O N OH NO Cl Cl HAN‐155 O H 510k 1.7 nM [61,62] F F O O O N O Cl HAN‐155 O N Cl PF8380 H 510k 1.7 nM [61,62] HA-155 O O N O HA‐155 NO Cl Cl HA‐N155 O H 510k 1.7 nM [61,62] Cl PF8380 OH O O O N O Cl Cl N O N Cl O OH 510k 1.7 nM [61,62] Cl PF8380N OH O O O N O O N ON Cl O N N H 510k 510k1.7 nM 1.7 nM[61,62] [61,62] PF8380 NO O Cl N H O 5m0e ‐ [62] Cl O O N OH O N N O Cl O NH 510k 1.7 nM [61,62] Cl O NPF8380 H 510k 1.7 nM [61,62] Cl Cl OH NO 5m0e ‐ [62] O N H O PF8380 O O O NPF8380 O Cl Cl N OH N 5m0e ‐ [62] FP‐Cpd 3 H O O NPF8380 O PF8380 NO Cl Cl N OH H 5m0e ‐ [62] FP‐Cpd 3Cl Cl O H3C O O N O Cl OH N Cl Cl N N OH H 5m0e5m0e ‐ -[62] [62] FP‐Cpd 3Cl Cl O

H3CO O N O N N ON 5m0e ‐ [62] Cl N N S H FP‐Cpd 3 O Cl Cl O O N O O N 3wav 180 nM * [63] Cl H3C FP-CpdN N 3 N 5m0e ‐ [62] Cl S NH 5m0e ‐ [62] N FP‐Cpd 3Cl Cl H O H3C O O 3wav 180 nM * [63] N N N S FP‐Cpd 3Cl Cl

H3C N O 3wav 180 nM * [63] N FP‐NCpdS 3 MKFP‐‐Cpd 310Cl Cl H3C 3wav3wav 180 nM * 180 nM *[63] [63] N N O N S HO Cl Cl OH MK‐Cpd 10Cl Cl BH3C O H3C N 3wav 180 nM * [63] N N S Cl Cl N HO OH MK‐Cpd 10 B N N O 3wav 180 nM * [63] MK-CpdN N S 10 S Cl Cl HO MK‐Cpd 10 OH N O 3wav 180 nM * [63] B N N S 3wav 180 nM * [63] N Cl Cl 3waw 580 nM * [63] HO OH MK‐Cpd 10 B N NO O S N Cl Cl 3waw 580 nM * [63] HO OH MK‐Cpd 10O B N N N S Cl Cl HO MK‐Cpd 10 3waw3waw 580 nM 580* nM *[63] [63] OHMK‐Cpd 10O B N N N S 2BoA Cl Cl HO 3waw 580 nM * [63] HO OH B OH N N O B N S Cl Cl OH 2BoA Cl Cl 3waw 580 nM * [63] O N N 2BoAN S Cl Cl B N HO OH 2BoA 3waw 580 nM * [63] N N O N N S Cl Cl B 2BoA S 3waw 580 nM * [63] HO N 3waw 580 nM * [63] OH S N N O N Cl Cl 3wax 13 nM * [63] B 2BoA HO OH N N O S N O Cl Cl B 3wax3wax 13 nM * 13 nM *[63] [63] O HO OH 2BoA N N N S Cl Cl B 2BoA 3wax 13 nM * [63] HO OH 2BoAN O N N S 3BoA Cl Cl B 3wax 13 nM * [63] HO OH OH 3BoAN N O N S Cl Cl B 3BoA ClCl ClCl B 3wax 13 nM * [63] HO O HO N N N S NN HO 3BoA Cl Cl 3wax 13 nM * [63] B N O NN N S 3BoAN SS HO OH Cl Cl 3wax3wax 1313 nMnM ** [63][63] O B N 3way3way 22 nM * 22 nM *[63] [63] N NN S N Cl Cl OH 3BoA Cancers 2019, 11, x HO O 8 of 17 OO 3way 22 nM * [63] B N N N S HO 3BoA Cl Cl OH B 4BoAN O 3way 22 nM * [63] 3BoAN N S HO OH CH3 3BoA4BoA Cl Cl B N N O 3way 22 nM * [63] N S ON N Cl Cl HO OH 4BoA Cl Cl O B N 3way 22 nM * [63] N N S 5l0b 5l0b520 nM 520 nM[61] [61] HO NNN HO OH 4BoAH B N O 3way 22 nM * [63] B N N S S OH 4BoA OH SBJ-Cpd 1 3way 22 nM * [63] N O 3way 22 nM * [63] SBJ4BoA‐Cpd N 1 O O 4BoA O O H 4BoA N N 4BoA NH

N N 5l0e 2.5 nM [61]

O

SBJ‐Cpd 2

F O F

NN F N O HN

H N 5ohi 2.2 nM [64]

F F O F BI‐2545 * TG‐mTMP used as subtrate

Subsequently, numerous type I ATX small molecule inhibitors have appeared in both academic and patent literature. An example of these are the thiazolone derivatives similar to the HA series produced by Kawaguchi and collaborators [63]. A library of 81,600 compounds was screened for inhibiting the hydrolysis of the fluorescence probe TG‐mTMP. This led to the identification of KM‐ Cpd 10 (180 nM, TG‐mTMP), 2BoA (580 nM, TG‐mTMP), 3Boa (13 nM, TG‐mTMP), and 4BoA (22 nM, TG‐mTMP), which were co‐crystallized with ATX (Table 3). These structures revealed interactions with Arg284, as well as hydrophobic contacts with Trp260 and Phe274 (Figure 3, Table 1). Moreover, these compounds were able to decrease LPA levels both in vitro and in vivo [63]. The most potent type I inhibitor to date is PF‐8380 (IC50 = 1.7 nM, LPC) (Table 3), reported by Pfizer [61,62]. In general, it has been widely used because of its high potency and its favourable pharmacokinetic properties, which has allowed in vivo evaluation of ATX inhibition. PF‐8380 is a benzoxazolone that exhibits the general chemotype of lipophilic tail, core spacer, and acidic head group. This general motif contains a benoxazolone as the acidic head group, a functionalized piperazine as the spacer, and the dichlorocarbamate moiety as the lipophilic portion [61,62]. In the active site, the acidic head group makes essential interactions with one of the Zn2+ ions, and the lipophilic tail is accommodated within the hydrophobic pocket. Moreover, it is in close proximity to Thr209 and has hydrophobic interactions with Leu213, Phe273, and Phe274, as well as a hydrogen bond with the Trp275 amino group (Figure 3). The addition of a hydroxyethyl group to PF‐8380 resulted in FP‐Cpd‐3, which had an increased solubility and was co‐crystallisation with ATX (Table 3) [65]. ATX inhibition with PF‐8380 was shown to attenuate bleomycin‐induced pulmonary fibrosis, owing to decreased LPA levels in plasma and bronchioalveolar lavage fluid, together with a decrease in inflammation and collagen deposition (Table 2). However, effectiveness of treatment with this compound varies in the literature [42,66]. Recently, treatment of mice on a high‐fat diet with PF‐8380 reduced plasma LPA levels, resulting in lower diet‐induced cardiac dysfunction and inflammatory response [67]. The compound SBJ‐Cpd‐1 (IC50 = 520 nM, LPC) [7] (Table 3) was obtained from an aminopyrimidine series that was further improved by the addition of the benzoxazolone moiety present in PF‐8380. This was crucial for synthesizing the far more active SBJ‐Cpd‐2 (IC50 = 2.5 nM, LPC), which reaches the active site Zn2+ ions in a similar manner to that of PF‐8380. Additionally, it also contacts Asp171, Asp311, His315 and His474 at the active site, and Leu213, Phe273, Phe274 and Trp275 at the pocket (Figure 3, Table 1). This compound was further tested in rats by a single oral dose (10 mg kg‐1), where plasma LPA levels decreased 80% upon 12 h treatment (Table 2). Lastly,

Cancers 2019, 11, x 7 of 17 structure‐activity improvement studies based on the HA‐155 crystal structure. These consisted in Cancers 2019, 11, x 8 of 17 testingCancers the 2019different, 11, 1577 linkers added to the thiazolidine‐2,4‐dione core [60]. 8 of 17 Cancers 2019, 11, x 8 of 17

Table 3. Type I ATX inhibitorsCH3 based on their kinetic and crystallographic analysis. * denotes that TG‐

ON N mTMP was used as theCH3 substrate for reporting theTable IC50. 3. Cont. 5l0b 520 nM [61] ON NNN H NN 5l0b 520 nM [61] H SBJ‐CpdTypeType 1 I Inhibitor I Inhibitor PDB PDBID IDActivity Activity (IC50) (ICReference50) Reference

SBJ‐Cpd 1 O

O H O N N O NH O H N N NH N N N 5l0e 2.5 nM [61] O S HO N N O 5l0e 5l0e2.5 nM 2.5 nM[61] [61] B O HO 2xrg 5.7 nM [25,60] OH O SBJ‐Cpd 2 F SBJ-Cpd 2 SBJ‐Cpd 2 F HA‐155O F

NN F F O F Cl N O O HN NN F H O O N N N 5ohi 5ohi2.2 nM 2.2 nM[64] [64] HN O O N F F H N Cl O N F 510k5ohi 1.72.2 nMnM [61,62][64] H O F F O BI-2545 F BI‐2545 PF8380BI‐2545 * TG‐mTMP* TG-mTMP used used as subtrate as subtrate.

Subsequently,Cl numerous type* ITG ATXOH ‐mTMP small used molecule as subtrate inhibitors have appeared in both academic and O Subsequently, numerous typeN I ATX small molecule inhibitors have appeared in both academic patent literature. An example of these are theO thiazolone derivatives similar to the HA series produced by O N and patentSubsequently, literature.Cl numerous An example type Iof ATX these small N are moleculethe5m0e thiazolone ‐ inhibitors derivatives have appeared similar[62] into boththe HA academic series Kawaguchi and collaborators [63]. A libraryH of 81,600 compounds was screened for inhibiting the hydrolysis O andproduced ofpatent the fluorescenceby literature. Kawaguchi An probe andexample TG-mTMP. collaborators of these This are[63]. led the toA thethiazolonelibrary identification of 81,600derivatives of compounds KM-Cpd similar 10 wasto (180 the screened nM, HA TG-mTMP series for ), inhibitingproduced2BoA the (580by Kawaguchihydrolysis nM, TG-mTMP), ofandFP the‐ Cpdcollaborators fluorescence 3Boa 3 (13 nM, [63].probe TG-mTMP), A TG library‐mTMP. of and 81,600 This 4BoA led compounds (22 to nM,the identification TG-mTMP), was screened whichof KM for‐ were Cpdinhibiting 10 (180 the nM, hydrolysis TG‐mTMP), of the 2BoA fluorescence (580 nM, probe TG‐mTMP), TG‐mTMP. 3Boa This (13 lednM, to TG the‐mTMP), identification and 4BoA of KM (22‐ co-crystallized with ATX (TableCl 3).Cl These structures revealed interactions with Arg284, as well as CpdnM, 10TG (180‐mTMP), nM, TGwhich‐mTMP),H3C were 2BoA co‐crystallized (580 nM, TG with‐mTMP), ATX 3Boa (Table (13 nM,3). TheseTG‐mTMP), structures and 4BoArevealed (22 hydrophobic contactsN with Trp260 and Phe274 (Figure3, Table1). Moreover, these compounds were able interactionsnM, TG‐mTMP), with Arg284,which aswere Nwell co as‐ crystallizedhydrophobic with contacts ATX with (Table Trp260 3). andThese Phe274 structures (Figure revealed 3, Table to decrease LPA levels both inS vitro and in vivo [63]. 3wav 180 nM * [63] interactions1). Moreover, with these Arg284, compounds as wellN were as hydrophobic able to decrease contacts LPA withlevels Trp260 both in and vitro Phe274 and in (Figure vivo [63]. 3, Table The most potent type I inhibitor to date is PF-8380 (IC50 = 1.7 nM, LPC) (Table3), reported by 1). Moreover,PfizerThe most [61 ,these 62potent]. Incompounds type general, I inhibitor it wereO has beenableto date to widely decrease is PF‐ used8380 LPA because(IC levels50 = 1.7 ofboth nM, its in high LPC) vitro potency (Tableand in 3),vivo and reported its[63]. favourable by PfizerThe [61,62]. most Inpotent general, type itI inhibitorhas been towidely date isused PF‐ 8380because (IC50 of = its1.7 highnM, LPC)potency (Table and 3), its reported favourable by pharmacokinetic properties,MK‐Cpd which 10 has allowed in vivo evaluation of ATX inhibition. PF-8380 is pharmacokineticPfizera benzoxazolone[61,62]. In properties,general, that exhibitsit haswhich been the has generalwidely allowed used chemotype in vivobecause evaluation of of lipophilic its high of ATX tail, potency coreinhibition. spacer,and its PF and favourable‐8380 acidic is a head HO OH pharmacokineticbenzoxazolone that properties, exhibitsB thewhich general has allowed chemotype in vivo of lipophilic evaluation tail, of coreATX spacer,inhibition. and PFacidic‐8380 head is a group. This general motif containsCl a benoxazoloneCl as the acidic head group, a functionalized piperazine benzoxazolonegroup.as theThis spacer, general that and exhibits motif the dichlorocarbamate N contains the general a benoxazolone chemotype moiety of asas lipophilic the the lipophilic acidic tail, head portioncore group,spacer, [61, 62 aand ].functionalized In acidic the active head site, N piperazine as the spacer, and the dichlorocarbamateS moiety as the lipophilic portion [61,62]. In the group.the This acidic general head group motif makes contains essential a benoxazolone interactions3waw withas the one 580acidic of nM the *head Zn 2+ group,ions,[63] and a thefunctionalized lipophilic tail is 2+ piperazineactiveaccommodated site, asthe the acidic spacer, within head and thegroup the hydrophobicN dichlorocarbamate makes essential pocket. interactions Moreover, moiety as it thewith is inlipophilic one close of proximity theportion Zn [61,62]. toions, Thr209 and In andthe has O 2+ lipophilicactivehydrophobic site, tail the is acidicaccommodated interactions head group with within Leu213, makes the Phe273, essentialhydrophobic and interactions Phe274, pocket. as Moreover,with well asone a hydrogen of it isthe in Znclose bond ions, proximity with and the the Trp275to Thr209 and has hydrophobic interactions with Leu213, Phe273, and Phe274, as well as a hydrogen lipophilicamino tail group is accommodated (Figure3).2BoA The within addition the hydrophobic of a hydroxyethyl pocket. group Moreover, to PF-8380 it is in resultedclose proximity in FP-Cpd-3, to Thr209bond with and the has Trp275 hydrophobic amino interactionsgroup (Figure with 3). Leu213, The addition Phe273, of and a hydroxyethyl Phe274, as well group as ato hydrogen PF‐8380 which had an increasedOH solubility and was co-crystallisation with ATX (Table3)[ 65]. ATX inhibition Cl Cl resultedbond with in FPthe‐ CpdTrp275‐3,B which amino had group an increased (Figure 3). solubility The addition and was of coa hydroxyethyl‐crystallisation group with ATXto PF (Table‐8380 with PF-8380 wasHO shown to attenuate bleomycin-induced pulmonary fibrosis, owing to decreased resulted3) [65].LPA ATX in levels FP inhibition‐Cpd in plasma‐3, which with and PF hadN bronchioalveolar‐8380 an wasincreased shown solubility lavage to attenuate fluid, and together bleomycin was co‐ withcrystallisation‐induced a decrease pulmonary with in inflammation ATX fibrosis, (Table and N S owing3) [65].collagen toATX decreased inhibition deposition LPA with (Tablelevels PF ‐2in8380). plasma However, was shownand e bronchioalveolarffectiveness to attenuate3wax of bleomycin treatment 13lavage nM * fluid,‐ withinduced together this[63] pulmonary compound with a decrease variesfibrosis, in the owingin inflammation to decreased and LPA collagen levels indeposition plasmaN and (Table bronchioalveolar 2). However, lavage effectiveness fluid, together of treatment with a with decrease this literature [42,66]. Recently, treatmentO of mice on a high-fat diet with PF-8380 reduced plasma LPA compoundin inflammationlevels, resultingvaries and in the in collagen lower literature diet-induced deposition [42,66]. Recently, cardiac(Table 2). dysfunction treatment However, of and effectivenessmice inflammatory on a high of‐fat treatment response diet with [with 67PF].‐ 8380this reducedcompound plasma varies LPA in the levels, literature 3BoAresulting [42,66]. in lowerRecently, diet treatment‐induced cardiacof mice dysfunctionon a high‐fat and diet inflammatory with PF‐8380 The compound SBJ-Cpd-1 (IC50 = 520 nM, LPC) [7] (Table3) was obtained from an aminopyrimidine reducedresponse plasma [67]. LPA levels, resulting in lower diet‐induced cardiac dysfunction and inflammatory series that was further improved byCl theCl addition of the benzoxazolone moiety present in PF-8380. responseThe [67].compound SBJ‐CpdN ‐1 (IC50 = 520 nM, LPC) [7] (Table 3) was obtained from an This was crucialHO for synthesizing the far more active SBJ-Cpd-2 (IC50 = 2.5 nM, LPC), which reaches B N aminopyrimidineThe compound series 2+SBJ that‐Cpd was‐1 (ICfurtherS 50 = improved520 nM, LPC)by the [7]addition (Table of 3) the was benzoxazolone obtained from moiety an the active site ZnOH ions in a similar manner to that of PF-8380. Additionally, it also contacts Asp171, N 3way 22 nM * [63] presentaminopyrimidineAsp311, in PF His315‐8380. series This and His474thatwas wascrucial at further the for active synthesizing improved site, and by the Leu213, the far addition more Phe273, active of Phe274 the SBJ benzoxazolone‐Cpd and‐2 Trp275 (IC50 = at 2.5moiety the nM, pocket O 2+ LPC),present which in PF reaches‐8380. This the activewas crucial site Zn for ionssynthesizing in a similar the manner far more to activethat of SBJ PF‐‐Cpd8380.‐2 Additionally, (IC50 = 2.5 nM, it 1 (Figure3, Table1). This compound was further tested in rats by a single oral dose (10 mg kg − ), 2+ alsoLPC), wherecontacts which plasma reachesAsp171, LPA the Asp311, levelsactive4BoA His315 decreasedsite Zn and ions 80%His474 in upon a similarat the 12 active h manner treatment site, to and that (Table Leu213, of 2PF).‐8380. Lastly, Phe273, Additionally, another Phe274 relevant and it alsoTrp275 compoundcontacts at the Asp171, pocket shown (FigureAsp311, to behave 3, His315 Table type Iand1). inhibitors This His474 compound is at the the benzotriazole active was site,further and BI-2545 tested Leu213, (2.2in ratsPhe273, nM, by LPC) a Phe274 single [64] (Tableoraland 3), ‐1 Trp275dosewhich (10 at mg the was kg pocket able), where to (Figure reduce plasma 3, LPA Table LPA levels 1). levels bothThis decreased incompound vitro and 80% was in vivo.upon further 12 testedh treatment in rats (Table by a single2). Lastly, oral dose (10 mg kg‐1), where plasma LPA levels decreased 80% upon 12 h treatment (Table 2). Lastly,

Cancers 2019, 11, x 9 of 17

Cancers 2019, 11, x 9 of 17 another relevant compound shown to behave type I inhibitors is the benzotriazole BI‐2545 (2.2 nM, Cancers 2019, 11, x 9 of 17 LPC) [64] (Table 3), which was able to reduce LPA levels both in vitro and in vivo. CancersCancersanother 2019 2019 relevant, ,11 11, ,x x compound shown to behave type I inhibitors is the benzotriazole BI‐2545 (2.299 of ofnM, 1717 anotherLPC) [64] relevant (Table 3),compound which was shown able toto reducebehave LPA type levels I inhibitors both in is vitro the benzotriazole and in vivo. BI‐2545 (2.2 nM, 3.2.another Type relevant II Inhibitors compound shown to behave type I inhibitors is the benzotriazole BI‐2545 (2.2 nM, anotherLPC)Cancers [64] relevant 2019(Table, 11 , compound3), 1577 which was shown able toto behavereduce LPAtype levelsI inhibitors both inis vitrothe benzotriazole and in vivo. BI‐2545 (2.2 nM,9 of 17 LPC)LPC)3.2. Type Type[64] [64] II(Table (TableII Inhibitors ATX 3), 3), inhibitors which which was was owe able able their to to reduce effectreduce solely LPA LPA levels tolevels binding both both toin in thevitro vitro hydrophobic and and in in vivo. vivo. pocket, where they 3.2.obstruct TypeType LPCII II Inhibitors ATX accommodation. inhibitors owe As their a result, effect this solely competitive to binding mode to the of hydrophobic binding avoids pocket, interaction where withthey 3.2.3.2. Type 3.2.Type Type II II Inhibitors Inhibitors II Inhibitors theobstruct catalyticType LPC II zincATX accommodation. ions, inhibitors which owe may As their offer a result, effect selectivity this solely competitive advantages to binding mode overto the ofother hydrophobic binding inhibitors. avoids pocket, The interaction artificial where ATXwiththey obstructsubstrate,the catalyticTypeType TypeLPC II IIFS ATX zincATX‐ accommodation. II3, ATXwas inhibitorsions, inhibitors inhibitorsused which byowe owe may PharmAkea As owe their their offer a result, their effect effect selectivity e ff thistosolely ectsolely screen competitive solely advantagesto to binding bindingsmall to binding modemolecule to overto the the to ofother hydrophobic thehydrophobic binding compounds, hydrophobic inhibitors. avoids pocket, pocket, Thefrom interaction pocket, artificial where whichwhere where they withATXthey they obstructobstructtheidentifiedsubstrate, obstructcatalytic LPC LPC fourFS LPC zincaccommodation.‐ accommodation.3, indole wasions, accommodation. used‐ whichbased by analoguesmay As PharmAkeaAs aoffer a Asresult, result, a selectivitywith result, this thisto high screencompetitive this competitive inhibitoryadvantages competitive small mode modemoleculepotency. over mode of ofother binding binding Among ofcompounds, inhibitors. binding avoids avoidsthese, avoids The fromthreeinteraction interaction artificial interaction whichlead type with ATXwiththey II with thecompoundsthesubstrate,identified catalytic thecatalytic catalytic fourFS zinc zincwere‐3, indole ions,was zincions, reported, usedwhich ions,‐ whichbased by which namelymay analoguesmay PharmAkea offer offer may PAT selectivity selectivity owith‐ff078er to high selectivity (IC screen 50advantages advantagesinhibitory = 472 small advantagesnM; molecule potency. LPC),over over other otherPAT over Amongcompounds, inhibitors.‐ inhibitors.494 other (IC these,50 inhibitors. = The The20 fromthree nM; artificial artificial whichlead LPC), The type ATX artificialATX theyand II PAT‐352 (IC50 = 26 nM; LPC) [50] (Table 4). The structures of these compounds revealed common substrate,substrate,identifiedcompoundsATX substrate, FSfourFS ‐were3,‐3, indole waswas reported, FS-3, usedused‐based was byby namely analogues PharmAkea usedPharmAkea byPAT PharmAkeawith‐078 toto high (IC screenscreen50 inhibitory = 472 to smallsmall screen nM; molecule moleculepotency. LPC), small PAT molecule compounds,Amongcompounds,‐494 (IC these, compounds,50 = from 20 fromthree nM; which whichlead LPC), from type theythey and which II hydrophobicidentified four interactions indole‐based between analogues their with vinyl high‐nitrile inhibitory or hydantoin potency. moieties,Among these, and threeLeu213, lead Leu216, type II identifiedcompoundsPATthey‐352 identified four(IC were50 indole= 26 reported, four nM;‐based indole-basedLPC) analoguesnamely [50] (Table PAT analogues with‐078 4). high (ICThe 50 withinhibitory structures= 472 high nM; inhibitorypotency. LPC),of these PAT Among compounds potency.‐494 (ICthese, Among50 = revealedthree20 nM; these, lead LPC), common type three and II lead Phe274,compounds Trp275 were and reported, Tyr306 namely(Figure 3,PAT Table‐078 1). (IC 50 = 472 nM; LPC), PAT‐494 (IC50 = 20 nM; LPC), and compoundsPAThydrophobictype‐352 II(IC compounds were50 interactions= 26reported, nM; were LPC) betweennamely reported, [50] (TablePAT their namely‐078 vinyl4). (IC The‐50nitrile PAT-078 =structures 472 ornM; (IC hydantoin LPC), 50of= these472 PAT nM;moieties,compounds‐494 LPC), (IC 50and PAT-494= 20revealed Leu213, nM; (ICLPC), common 50Leu216,= and20 nM; PAT‐352 (IC50 = 26 nM; LPC) [50] (Table 4). The structures of these compounds revealed common PAThydrophobicPhe274,LPC),‐352 Trp275(IC and50 interactions= PAT-352 26and nM; Tyr306 (ICLPC) 50between (Figure= [50]26 nM;(Table 3,their Table LPC) 4).vinyl 1). [The50 ‐]nitrile (Tablestructures or4). hydantoin The of these structures compoundsmoieties, of these and revealed compounds Leu213, common Leu216, revealed Table 4. Type II ATX inhibitors based on their kinetic and crystallographic analysis. hydrophobicPhe274,hydrophobiccommon Trp275 interactions hydrophobicinteractions and Tyr306 betweenbetween interactions (Figure their3,their Table between vinylvinyl 1).‐ nitrile‐nitrile their or vinyl-nitrileor hydantoinhydantoin or moieties,moieties, hydantoin andand moieties, Leu213,Leu213, and Leu216,Leu216, Leu213, Phe274,Phe274,Leu216, Trp275 Trp275 Phe274,Table and and 4. Tyr306 Trp275Tyr306Type II (Figure ATX and(Figure Tyr306inhibitors 3, 3, Table Table (Figure based 1). 1). 3 on, Table their1 kinetic). and crystallographic analysis. Table 4. Type II ATX inhibitors based on their kinetic and crystallographic analysis. Type II Inhibitor PDB ID Activity (IC50) Reference TableTableTable 4. 4. Type Type 4. Type II II ATX ATX II ATXinhibitors inhibitors inhibitors based based based on on their their on their kinetic kinetic kinetic and and crystallographic and crystallographic crystallographic analysis. analysis. analysis. Type II Inhibitor PDB ID Activity (IC50) Reference

Type IIF Inhibitor PDB ID Activity (IC50) Reference Type II Inhibitor PDB ID Activity (IC50) Reference TypeType II Inhibitor II Inhibitor PDB ID PDB Activity ID (IC Activity50) Reference (IC50) Reference F N F

F N O N 4zg6 472 nM [50] F

N N OOH 4zg6 472 nM [50] N Cl N N O OH 4zg6 472 nM [50] N Cl 4zg6 472 nM [50] O PAT‐078 N 4zg6 4zg6472 nM 472 nM[50] [50] O OH

Cl OH O OH PAT‐078 Cl Cl PAT‐078 ONH PAT-078 N PATPATN‐078‐078 ONH 4zga 20 nM [50] N F N O O ONH 4zga 20 nM [50] N N F NH NH N O 4zga 4zga20 nM 20 nM[50] [50] PATN ‐494 N N F O 4zga 20 nM [50] O 4zga 20 nM [50] F PAT‐494 F F PAT-494 O PAT‐494 N F PATN ‐494 HO NOPAT‐494 4zg9 26 nM [50] N F N O O HO NO F F 4zg9 4zg926 nM 26 nM[50] [50] N O N O HO N O O N 4zg9 26 nM [50] PATN ‐352N HO N N HO N 4zg9 26 nM [50] O O PAT-352 4zg9 26 nM [50] O O O PAT‐352F O O F PATF‐352 O F O CH3 PAT‐352F PAT‐352F N O F 5lia 1 nM [68] N 5lia 1 nM [68] N O H CH3 O FF N F O F N Cl NF 5lia 1 nM [68] N OF F H CH3 F N N O CH3 Cl 5lia 1 nM [68] CRT0273750O N CH3 N CRT0273750H N 5lia 1 nM [68] N N N N N H Cl 5lia 1 nM [68] H N CRT0273750N N Cl The artificial ATX substrate FS-3Cl was also used as readout for a high-throughput screen from The artificial ATX substrate FS‐3 was also used as readout for a high‐throughput screen from a a collection of 87,865 compounds.CRT0273750 Upon a preliminary selection of 1.2% best hits, the physiological collectionThe artificial of 87,865 ATX compounds. substrateCRT0273750 Upon FS‐3 wasa preliminary also used selectionas readout of for1.2% a besthigh hits,‐throughput the physiological screen from LPC a LPC choline release assayCRT0273750 was used, from which the imidazo[4,5-b] biyridine-derivative CRT0273750 cholinecollectionThe release artificial of 87,865 assay ATX compounds. was substrate used, from Upon FS‐ 3which wasa preliminary also the imidazo[4,5used selectionas readout‐b] of biyridine for1.2% a highbest‐derivative ‐hits,throughput the physiological CRT0273750 screen from (ICLPC 50a = 1 (ICnM,50 LPC;= 1 nM, 14 nM LPC; human 14 nM plasma human LPC) plasma [69] was LPC) identified [69] was (Table identified 4). The (Table crystal4). The structure crystal of structure ATX collectioncholineTheThe releaseartificial artificial of 87,865 assay ATX ATX compounds. was substrate substrate used, from UponFS FS‐3‐3 whichwas wasa preliminary also also the used imidazo[4,5used as selectionas readout readout‐b] of biyridine for for1.2% a a high highbest‐derivative‐ throughput‐hits,throughput the physiological CRT0273750 screen screen from from (ICLPC a 50a incollection complexof ATX of inwith 87,865 complex CRT0273750 compounds. with CRT0273750 indicated Upon a thatpreliminary indicated the compound that selection the compoundbinds of 1.2% at the best binds hydrophobic hits, at the the physiological hydrophobic pocket, as LPCwell pocket, collectioncholine= 1 nM, releaseLPC; of 87,865 14 assay nM compounds. humanwas used, plasma Uponfrom LPC)which a preliminary [69] the wasimidazo[4,5 identifiedselection‐b] of (Table biyridine 1.2% best4). ‐Thederivative hits, crystal the physiological CRT0273750structure of LPC ATX(IC50 as well as with the boundaries of the ATX tunnel, but 5 Å away from the active site. As a consequence, ascholine with releasethe boundaries assay was of used, the ATX from tunnel, which thebut imidazo[4,5 5 Å away from‐b] biyridine the active‐derivative site. As CRT0273750a consequence, (IC it50 choline=in 1 complex nM, release LPC; with 14 assay nMCRT0273750 washuman used, plasma indicated from whichLPC) that [69] the the imidazo[4,5was compound identified‐b] binds biyridine(Table at the4).‐ derivative Thehydrophobic crystal CRT0273750 structure pocket, ofas (IC ATXwell50 it establishes a hydrogen bond with Lys247, and hydrophobic interactions with Leu213, Phe248, Trp254, =establishes=inas 1 1 complexnM, withnM, LPC; LPC;the withaboundaries 14 14hydrogen nM nMCRT0273750 human human ofbond the plasma plasma indicated ATXwith LPC) tunnel,LPC)Lys247, that [69] [69] butthe wasand was compound5 Åidentified hydrophobicidentified away from binds (Table (Table theinteractions at 4). activethe4). The Thehydrophobic site.crystal crystal with As structure astructureLeu213, consequence,pocket, of Phe248, ofas ATX ATXwell it Phe273, Phe274 and Trp275 (Table1). In vitro data showed that this compound could inhibit migration in inasestablishes complex complexwith the with withaboundaries hydrogen CRT0273750 CRT0273750 ofbond the indicated indicated ATXwith tunnel, Lys247, that that the butthe and compound 5compound Åhydrophobic away frombinds binds the atinteractions at the activethe hydrophobic hydrophobic site. with As aLeu213, pocket,consequence, pocket, asPhe248, as well well it of 4T1 cells. Moreover, the compound was effective in reducing 18:1 LPA levels in vitro in human asestablishesas with with the the boundaries aboundaries hydrogen of ofbond the the ATX ATXwith tunnel, tunnel,Lys247, but but and 5 5 Å Å hydrophobicaway away from from the theinteractions active active site. site. with As As a a Leu213,consequence, consequence, Phe248, it it plasma, as well as in in vivo samples from MDA-MB-231-luc tumour bearing Balb-c nu/nu mice establishesestablishes aa hydrogenhydrogen bondbond withwith Lys247,Lys247, andand hydrophobichydrophobic interactionsinteractions withwith Leu213,Leu213, Phe248,Phe248, (Table2)[ 69]. Lastly, the short phosphonate lipid analogue GWJ-A-23 [68] was used in models for lung

Cancers 2019, 11, x 10 of 17 Cancers 2019, 11, x 10 of 17 Trp254,Cancers 2019 Phe273,, 11, x Phe274 and Trp275 (Table 1). In vitro data showed that this compound could 10inhibit of 17 Cancers 2019, 11, x 10 of 17 CancersmigrationTrp254, 2019 Phe273, , of11 ,4T1 x Phe274 cells. Moreover, and Trp275 the (Table compound 1). In vitrowas effectivedata showed in reducing that this 18:1 compound LPA levels could in vitro 10inhibit of in17 Trp254,Cancers 2019 Phe273,, 11, x Phe274 and Trp275 (Table 1). In vitro data showed that this compound could inhibit10 of 17 humanmigration plasma, of 4T1 as cells. well Moreover, as in in vivo the compoundsamples from was MDA effective‐MB ‐in231 reducing‐luc tumour 18:1 bearingLPA levels Balb in‐ cvitro nu/nu in Trp254,migration Phe273, of 4T1 Phe274 cells. Moreover, and Trp275 the (Table compound 1). In vitrowas effectivedata showed in reducing that this 18:1 compound LPA levels could in vitro inhibit in Trp254,humanmice (Table plasma,Phe273, 2) [69]. asPhe274 wellLastly, andas thein Trp275 in short vivo (Tablephosphonate samples 1). Infrom vitro lipid MDA data analogue‐ MBshowed‐231 GWJ‐ lucthat ‐tumourA this‐23 compound[68] bearing was used Balbcould in‐ cmodels inhibitnu/nu migrationhumanTrp254, plasma, Phe273, of 4T1 as Phe274cells. well Moreover, asand in Trp275in vivo the (Table compoundsamples 1). fromIn wasvitro MDA effective data‐MB showed ‐in231 reducing‐luc that tumour this 18:1 compound LPAbearing levels Balb could in‐ cvitro nu/nuinhibit in migrationmicefor lung (Table fibrosis of 2) 4T1 [69]. cells.and Lastly, inflammation, Moreover, the short the phosphonate wherecompound it resulted was lipid effective analoguein a 50% in reducingdecreaseGWJ‐A‐23 18:1of [68] LPA LPA was concentrations levels used in vitromodels in humanmicemigration (Table plasma, of 2) 4T1 [69]. as cells. wellLastly, Moreover, as thein in short vivo the phosphonate samplescompound from was lipid MDA effective analogue‐MB‐ 231in GWJreducing‐luc ‐tumourA‐23 18:1 [68] bearing LPA was levelsused Balb inin‐ c models vitronu/nu in humanbronchioalveolarfor Cancerslung plasma, fibrosis2019, 11 ,as 1577 andlavage well inflammation, asfluid in in(Table vivo 2).wheresamples it fromresulted MDA in ‐aMB 50%‐231 decrease‐luc tumour of LPA bearing concentrations Balb‐c nu/nu 10in of 17 miceforhuman lung (Table plasma,fibrosis 2) [69]. asand Lastly,well inflammation, as the in inshort vivo phosphonatewhere samples it resultedfrom lipid MDA analoguein a‐MB 50%‐231 GWJdecrease‐luc‐ Atumour‐23 of [68] LPA bearing was concentrations used Balb in‐ modelsc nu/nu in micebronchioalveolar (Table 2) [69]. lavage Lastly, fluid the (Tableshort phosphonate 2). lipid analogue GWJ‐A‐23 [68] was used in models forbronchioalveolarmice lung (Table fibrosis 2) [69]. andlavage Lastly, inflammation, fluid the (Table short phosphonate2).where it resulted lipid in analogue a 50% decrease GWJ‐A‐ 23of [68]LPA was concentrations used in models in for3.3. lungType IIIfibrosis Inhibitors and inflammation, where it resulted in a 50% decrease of LPA concentrations in bronchioalveolarforfibrosis lung fibrosis and inflammation, lavageand inflammation, fluid where (Table it 2).where resulted it in resulted a 50% decrease in a 50% of LPAdecrease concentrations of LPA concentrations in bronchioalveolar in bronchioalveolar3.3. Type III Inhibitors lavage fluid (Table 2). 3.3.bronchioalveolar lavageTypeType III III fluid Inhibitors inhibitors (Table lavage 2 ).specifically fluid (Table occupy 2). the allosteric regulatory tunnel, modulating ATX activity 3.3.by non TypeType‐competitive III III Inhibitors inhibitors inhibition. specifically We haveoccupy reported the allosteric that 7‐α‐ regulatoryhydroxycholesterol tunnel, modulating (Table 5) isATX commonly activity 3.3. TypeType III III Inhibitors inhibitors specifically occupy the allosteric regulatory tunnel, modulating ATX activity bypresent3.3. non3.3. Type‐ Typecompetitivein III purified Inhibitors III Inhibitors inhibition.mammalian We ATX have structures, reported that but 7this‐α‐hydroxycholesterol does not have an (Tableobservable 5) is commonlyinhibitory by nonType‐competitive III inhibitors inhibition. specifically We haveoccupy reported the allosteric that 7‐α‐ regulatoryhydroxycholesterol tunnel, modulating (Table 5) isATX commonly activity activity.presentType inHowever, IIIpurified inhibitors other mammalian specifically steroids, ATXnamely, occupy structures, bile the salts allosteric butsuch this asregulatory tauroursodeoxycholicdoes not tunnel, have anmodulating observable acid (TUDCA) ATX inhibitory activity (IC50 by nonType‐competitiveType III IIIinhibitors inhibitors inhibition. specifically specifically We have occupy occupy reported the the allosteric that allosteric 7‐α‐ regulatoryhydroxycholesterol regulatory tunnel, tunnel, modulating (Table modulating 5) is ATX commonly ATX activity activity activity.present inHowever, purified other mammalian steroids, ATXnamely, structures, bile salts butsuch this as tauroursodeoxycholicdoes not have an observable acid (TUDCA) inhibitory (IC50 presentby=by 10.4 non non μ‐ ‐competitivecompetitiveinM; purifiedLPC) or inhibition. mammalianinhibition.ursodeoxycholic We We ATX have have acidstructures, reported reported (UDCA) thatthatbut (IC 7 7this‐α‐‐α‐50 =hydroxycholesterol hydroxycholesterol does8.8 μ notM; LPC)have exertan (Table observable(Table modest 5) 5) is is and commonly commonlyinhibitory partial activity.by non-competitive However, other inhibition. steroids, namely, We have bile reported salts such that as 7- tauroursodeoxycholicα-hydroxycholesterol acid (Table (TUDCA)5) is commonly (IC50 present= 10.4 μ M;in purifiedLPC) or mammalianursodeoxycholic ATX acidstructures, (UDCA) but (IC this50 = does8.8 μ notM; LPC)have exertan observable modest and inhibitory partial activity.(alwayspresent However,leavingin purified residual other mammalian steroids, activity) namely, ATXnon‐ competitivestructures, bile salts suchbutinhibition this as tauroursodeoxycholic does of lysoPLD not have activity an observable acid (Table (TUDCA) 5) inhibitory[26]. (IC The50 = 10.4present μM; inLPC) purified or ursodeoxycholic mammalian ATX acid structures, (UDCA) (IC but50 = this 8.8 does μM; notLPC) have exert an modest observable and inhibitorypartial activity.ATX(always co ‐ However,leavingcrystal, residualwith other 18:1 steroids, activity) LPA boundnamely, non‐competitive at bile the salts orthosteric suchinhibition as tauroursodeoxycholic site of andlysoPLD TUDCA activity bound acid (Table (TUDCA)at 5)the [26]. tunnel, (IC The50 =(alwaysactivity. 10.4activity. μ M;leaving However, LPC) However, residual or other ursodeoxycholic other steroids,activity) steroids, namely,non ‐acidcompetitive namely, bile(UDCA) salts bile inhibitionsuch salts(IC50 as such= tauroursodeoxycholic8.8 of μ as lysoPLDM; tauroursodeoxycholic LPC) activity exert modest acid(Table (TUDCA) and5) acid [26]. partial (TUDCA) (ICThe50 =ATX 10.4 co μ‐M;crystal, LPC) with or ursodeoxycholic 18:1 LPA bound acid at (UDCA)the orthosteric (IC50 = site8.8 μ andM; LPC)TUDCA exert bound modest at andthe tunnel,partial (alwaysconfirmed leaving the non residual‐competitive activity) mode non of‐competitive inhibition. inhibitionTUDCA50 interacts of lysoPLD with activity ATX, forming (Table 5)a hydrogen [26]. The ATX= 10.4(IC co50 μ‐crystal,=M;10.4 LPC)µ Mwith or; LPC) ursodeoxycholic 18:1 or LPA ursodeoxycholic bound acid at the(UDCA) acid orthosteric (UDCA) (IC = (IC site8.850 μ=andM;8.8 TUDCALPC)µM; LPC) exert bound exert modest modestat theand andtunnel, partial partial (alwaysbondconfirmed with leaving theTrp260 non residual‐ competitiveand hydrophobic activity) mode non stackingof‐competitive inhibition. with inhibitionTUDCA Trp254 andinteracts of lysoPLDPhe274 with (Figure activity ATX, forming3). (Table The mechanisma5) hydrogen [26]. The ATXconfirmed(always(always co‐ crystal,leaving the leaving non withresidual‐ residualcompetitive 18:1 activity) activity)LPA mode bound non non-competitive of‐competitive inhibition.at the orthosteric TUDCAinhibition inhibition siteinteracts of and lysoPLDlysoPLD TUDCAwith activity activityATX, bound forming (Table (Table at5 )[ athe5) hydrogen26 [26]. tunnel,]. The The ATX ATXunderlyingbond cowith‐crystal, Trp260its inhibition with and 18:1 hydrophobic may LPA occur bound by stacking counteracting at the withorthosteric Trp254 the modulatory siteand andPhe274 actionTUDCA (Figure of LPAbound 3). in The theat mechanismtunnelthe tunnel, [28]. confirmedbondATXco-crystal, withco‐crystal, theTrp260 withnon with‐ 18:1competitiveand 18:1 LPAhydrophobic LPA bound mode bound at stacking theof inhibition. orthostericat the with orthosteric TUDCA Trp254 site and interactsandsite TUDCA Phe274and with boundTUDCA (Figure ATX, at thebound forming 3). tunnel, The at amechanismthe confirmedhydrogen tunnel, the confirmedunderlying the its inhibitionnon‐competitive may occur mode by of counteracting inhibition. TUDCA the modulatory interacts actionwith ATX, of LPA forming in the atunnel hydrogen [28]. bondunderlyingconfirmednon-competitive with Trp260theits inhibition non ‐andcompetitive mode hydrophobic may of occur inhibition. mode by stacking ofcounteracting inhibition. TUDCA with interacts TUDCATrp254 the modulatory and withinteracts Phe274 ATX, action with forming (Figure ATX, of LPA aforming3). hydrogenin The the mechanism tunnela hydrogen bond [28]. with bond with Trp260Table 5. and Type hydrophobic III ATX inhibitors stacking based withon their Trp254 kinetic and and Phe274 crystallographic (Figure analysis.3). The mechanism underlyingbondTrp260 with and itsTrp260 inhibition hydrophobic and hydrophobicmay stacking occur by with stackingcounteracting Trp254 with and theTrp254 Phe274 modulatory and (Figure Phe274 action3). The(Figure of mechanism LPA 3). in The the tunnel underlyingmechanism [28]. its underlying itsTable inhibition 5. Type mayIII ATX occur inhibitors by counteracting based on their the kinetic modulatory and crystallographic action of LPA analysis. in the tunnel [28]. underlyinginhibition Tableits may inhibition 5. occur Type by IIImay ATX counteracting occur inhibitors by counteracting based the modulatory on their the kinetic modulatory action and ofcrystallographic LPA action in the of LPA tunnel analysis. in [the28 ]. tunnel [28]. Table 5. Type III ATX inhibitors based on their kinetic and crystallographic analysis. Table 5. Type III ATXType inhibitors III Inhibitor based on PDBtheir IDkinetic Activity and crystallographic (IC50) Reference analysis. TableTable 5. Type 5. Type III ATX III ATX inhibitors inhibitors based based on ontheir their kinetic kinetic and and crystallographic crystallographic analysis. analysis. Type III Inhibitor PDB ID Activity (IC50) Reference Type III Inhibitor PDB ID Activity (IC50) Reference

Type III Inhibitor PDB ID Activity (IC50) Reference HO Type III Inhibitor PDB ID Activity (IC50) Reference 50 TypeTypeHO III III Inhibitor Inhibitor PDB PDB ID IDActivity Activity (IC ) (ICReference50) Reference

HO OH 5dlt undetermined [26] H3C C HO H3C H3 OH CH3 C HO 5dlt undetermined [26] H3C CH3 H3C H3 HO OH CH3 5dlt undetermined [26] 3 C H3 H C CH3 C OH H3 CH3 7‐α‐hydroxycholesterolC 5dlt undetermined [26] H3C H3C CH3 H3 OH CH3 OH 5dlt undetermined [26] H3C 7‐α‐hydroxycholesterolC H3C 5dlt undetermined [26] CH3 H3 5dlt undetermined [26] H3C C H3OC CH3 H3 7‐α‐hydroxycholesterolCH3 CH3 HO 7‐α‐CH3 hydroxycholesterol O HO 7-α-hydroxycholesterol OH OH 7‐α‐hydroxycholesterol 7‐α‐Ohydroxycholesterol 5dlw 10.4 μM [26] O C HO H3C S N H3 OH OH H CH3 O HO 5dlw 10.4 μM [26] O O C H3C S N H3 HO OH OH CH3 H O C 5dlw 10.4 μM [26] O O H3C OH SOH N H3 H O CH3 5dlw 10.4 μM [26] O TUDCAC H3C 5dlw 10.4 µM[26] OS N O H3 OH OH H CH3 OH OH 5dlw 10.4 μM [26] O TUDCAC H3C O N H3 5dlw 10.4 μM [26] S 3 C H3 O H CCHH3 C S N H3 H TUDCACH3 O F O HO S CH3 TUDCAN TUDCA N F CH3 HO S N O TUDCA O N F TUDCACH3 F 4zg7 0.3 nM [50] HO S N O F O N CH3 HO S F 4zg7 0.3 nM [50] CHN3Cl O F O N HO F S F 4zg7 0.3 nM [50] N 4zg7 0.3 nM [50] O HO S Cl O N N PAT‐347F N 4zg7 0.3 nM [50] O Cl O O PAT‐347F O 4zg7 0.3 nM [50] Cl F 4zg7 0.3 nM [50] H3C PAT-347 PAT‐347Cl N Cl F H3 C PAT‐347 OH N S N N H3C F PAT‐347 OH N PAT‐347S N N O H3C F 5kxa 2 nM [70] F OH N S N O H3C N F OH 5kxa5kxa 2 nM 2 nM[70] [70] H3C NN F S ClN O N F OH 5kxa 2 nM [70] N S F F OH N ClN S O N 5kxa 2 nM [70] F O PATCl ‐505 O 5kxa 2 nM [70] F PAT-505 5kxa 2 nM [70] F PATCl ‐505 PATCl ‐505 Cl O PAT‐505CH3 N O PAT‐505CH3 5lqq 81 nM [71] OH PATS ‐505 N 5lqq 81 nM [71] O CH3 5lqq 81 nM [71] OH S N O CH3 Cl 5lqq 81 nM [71] OH S N O CH3 5lqq 81 nM [71] O CH3 Cl OH LM-CpdS 51N N 5lqq 81 nM [71] OH S Cl 5lqq 81 nM [71] OH S Cl The identification of type II indole-basedCl ATX inhibitors, as discussed above, also yielded a type Cl III compound, namely PAT-347 (IC50 = 0.3 nM; LPC) (Table5)[ 50]. The crystal structure of ATX bound

to both PAT-347 and 14:0 LPA confirmed its non-competitive binding mode. PAT-347 accommodates at the tunnel by π–π interactions between its indole moiety and Phe274 and His251. Moreover, its benzoic Cancers 2019, 11, x 11 of 17 Cancers 2019, 11, x 11 of 17 CancersCancers 2019 2019, 11, 11, x, x LM‐Cpd 51 1111 of of 17 17 LM‐Cpd 51 The identification of typeLMLM ‐IICpd‐ Cpdindole 51 51 ‐ based ATX inhibitors, as discussed above, also yielded a type III compound,The identification namely PATof type‐347 II (IC indole50 = 0.3‐based nM; ATXLPC) inhibitors, (Table 5) [50]. as discussed The crystal above, structure also ofyielded ATX bound a type The identification of type II indole‐based ATX inhibitors, as discussed above, also yielded a type IIIto bothcompound,The PAT identification‐347 namely and 14:0 PAT of LPA ‐type347 confirmed (IC II indole50 = 0.3‐ itsnM;based non LPC) ATX‐competitive (Table inhibitors, 5) [50]. binding as The discussed mode.crystal PAT above,structure‐347 also accommodates of yielded ATX bound a type III compound, namely PAT‐347 (IC50 = 0.3 nM; LPC) (Table 5) [50]. The crystal structure of ATX bound toat IIIboththe compound, tunnel PAT‐347 by πnamelyand–π 14:0interactions PAT LPA‐347 confirmed (IC between50 = 0.3 its nM;its non indole LPC)‐competitive (Tablemoiety 5) bindingand [50]. Phe274 The mode. crystal and PAT His251.structure‐347 accommodates Moreover, of ATX bound its to bothCancers PAT2019‐347, 11, 1577and 14:0 LPA confirmed its non‐competitive binding mode. PAT‐347 accommodates11 of 17 atbenzoic tothe both tunnel acid PAT makesby‐347 π –andπ anotherinteractions 14:0 LPA π–π confirmed interactionbetween itsits with nonindole‐ competitivePhe249, moiety and and binding it Phe274also contacts mode. and PATHis251. Lys248,‐347 Moreover, accommodates Trp254, and its at the tunnel by π–π interactions between its indole moiety and Phe274 and His251. Moreover, its Trp260.benzoicat the Moretunnelacid makes recently, by π –anotherπ theinteractions pharmacokinetic π–π interaction between propertiesitswith indole Phe249, moietyof twoand novelandit also Phe274 compounds, contacts and Lys248, His251. PAT‐ 505Trp254,Moreover, (IC50 and = 2 its benzoic acid makes another π–π interaction with Phe249, and it also contacts Lys248, Trp254, and nM;Trp260.benzoicacid LPC) More makes acidand recently, makesPAT another‐048 another the π(IC–π pharmacokinetic50interaction π= 1.1–π nM;interaction withLPC), properties Phe249, havewith Phe249,been and of assessedtwo it and also novel it contacts (Tablealso compounds, contacts 5) Lys248, [70,72–74]. Lys248, PAT Trp254,‐505 The Trp254, and(IC crystal50 Trp260. = and 2 Trp260. More recently, the pharmacokinetic properties of two novel compounds, PAT‐505 (IC50 = 2 structurenM;Trp260.More LPC) recently, of Moreand PAT PAT recently,‐505 the‐048 bound pharmacokinetic (ICthe 50to pharmacokinetic= ATX 1.1 nM;showed properties LPC), a very propertieshave of similar twobeen novel ofassessedbinding two compounds, novel mode (Table compounds, to 5) that PAT-505 [70,72–74]. of PAT PAT (IC‐347,50 The‐505= namely,2 crystal(IC nM50; LPC)= 2 nM; LPC) and PAT‐048 (IC50 = 1.1 nM; LPC), have been assessed (Table 5) [70,72–74]. The crystal bystructurenM; interactingand LPC) PAT-048 of PATand with ‐ (IC505PAT Lys248,50 bound‐048= 1.1 (IC toPhe249, nM;50 ATX = LPC),1.1 showed His251, nM; have LPC), aTrp254, beenvery have similar assessed Trp260 been binding assessed (Tableand Phe274, mode5)[ (Table70 to ,but72 that –5) 74also [70,72–74].].of ThePATSer81‐ crystal347, and The namely, Val277 structure crystal structure of PAT‐505 bound to ATX showed a very similar binding mode to that of PAT‐347, namely, (Tablebystructure interactingof PAT-505 1). Clinical of PAT boundwith assays‐505 Lys248, to bound ATX with showedPhe249, PATto ATX‐048, aHis251, showed very showed similar Trp254, a verybetter binding similarTrp260 pharmacodynamics mode bindingand to Phe274, that mode of PAT-347, but results,to thatalso of namely,andSer81 PAT this and‐347, by was interactingVal277 namely, used by interacting with Lys248, Phe249, His251, Trp254, Trp260 and Phe274, but also Ser81 and Val277 (Tableforby within interacting 1).vivo Lys248, Clinical assays Phe249,with assays with Lys248, His251,witha bleomycin PATPhe249, Trp254,‐048,‐ inducedHis251, showed Trp260 Trp254, mouse andbetter Phe274, pharmacodynamics Trp260model but forand also dermal Phe274, Ser81 fibrosis.results, and but Val277 also and Pharmacological Ser81 (Table this andwas1). Val277used Clinical (Table 1). Clinical assays with PAT‐048, showed better pharmacodynamics results, and this was used inhibitionfor(Table assaysin vivo 1). withof Clinicalassays ATX PAT-048, activity withassays showeda withmarkedlybleomycin PAT better‐ 048,attenuated‐induced pharmacodynamics showed mouse skin better fibrosis model pharmacodynamics results, uponfor dermal andtreatment this fibrosis. wasresults, with used Pharmacological10 and formg thisin kg vivo was‐1 PAT assaysused‐ for in vivo assays with a bleomycin‐induced mouse model for dermal fibrosis. Pharmacological inhibition048,forwith which in avivo bleomycin-inducedof related ATXassays activityto awith 75% markedlya inhibition bleomycin mouse model attenuated of‐ inducedATX for activity dermal skinmouse fibrosisafter fibrosis. model 24 hupon Pharmacological and for treatment>90%dermal at doublefibrosis. with inhibition 10dose Pharmacological mg (Table of kg ATX‐1 PAT 2). activity It‐ ‐1 inhibition of ATX activity markedly attenuated skin fibrosis upon treatment1 with 10 mg kg ‐PAT1 ‐ is048, inhibitionworthymarkedly which to related ofmention attenuated ATX to activity athat 75% skin five inhibition markedly fibrosis new patented uponof attenuated ATX treatment PharmAkea activity skin after with fibrosis compounds, 24 10 h mg andupon kg >90%− treatment specifically,PAT-048, at double with which PharmAkea dose 10 related mg(Table kg‐ to2).Cpd PAT a It 75% ‐ 048, which related to a 75% inhibition of ATX activity after 24 h and >90% at double dose (Table 2). It isA ‐048,worthyEinhibition (IC which50 =to < mentionrelated of0.5 ATX μM; to activitythat LPC)a 75% five have inhibition after new been 24 patented h andofrecently ATX> PharmAkea90% activity used at double in after metaboliccompounds, dose24 h and (Table disorder >90% specifically,2). Itat istreatment, double worthy PharmAkea dose to where mention (Table ‐theyCpd 2). that It is worthy to mention that five new patented PharmAkea compounds, specifically, PharmAkea‐Cpd showedA‐isEfive worthy(IC new 50a =decrease

5m0s 202 nM [65] 5m0s 202 nM [65]

5m0s5m0s5m0s 202202 202nM nM nM [65][65] [65] FP‐Cpd 5

Cl FP‐Cpd 5 OH FP-CpdFP‐Cpd 5 5 Cl FP‐Cpd 5 OH N Cl OH H3C Cl O N OH Cl CH3 N 5m0d 814 nM [65] H3C OH H3C O O N N Cl CH3 N 5m0d5m0d 814 nM 814 nM[65] [65] H3C O N H3C OH Cl OO N CH3 H3C 5m0d 814 nM [65] Cl CH3 H3C OH 5m0d 814 nM [65] H3C O FP‐Cpd 11 OH O FP-Cpd 11 FP‐Cpd 11 FPFP‐Cpd‐Cpd 11 11

5m0m5m0m 20 nM 20 nM[65] [65] Cancers 2019, 11, x 12 of 17 5m0m 20 nM [65]

FP-Cpd 17 5m0m5m0m 2020 nM nM [65][65]

H3C N FP‐Cpd 17

FPN ‐Cpd 17 S FP‐Cpd 17 CH3 N O FP‐Cpd N17 CH3 S N N CH3 O 5m7m5m7m 357 nM 357 nM[41] [41]

F AJ-CpdAJ‐Cpd 9

CH3

H3C N

N N N O S N N CH3 N N 5mhp 27 nM [51]

OH

F GLPG1690

We reported the structure‐guided design and chemical evolution of bile salts, from weak physiological non‐competitive inhibitors into potent competitive type IV compounds [65]. This was achieved by using the PF‐8380 dichlorocarbamate moiety as a pocket‐binding lipophilic portion. By design, these compounds did not interact with the active site residues, but still hampered LPC or LPA binding. The best lead compounds differed chiefly regarding spacers connecting the steroid and the dichrolocarbamate moieties, i.e., linear amide linker, piperazine, and piperidine for FP‐Cpd‐5 (IC50 = 202 nM; LPC), ‐11 (IC50 = 814 nM; LPC), and ‐17 (IC50 = 20 nM; LPC), respectively (Table 6) [65]. Moreover, a very short spacer yielded no measurable inhibitory activity, which explains why FP‐ Cpd‐11 exhibited a five‐fold lower activity than FP‐Cpd‐17. These compounds were co‐crystalized with rat ATX, showing how these types of inhibitors accommodate at both the tunnel and the pocket. Specifically, they all made hydrogen bonds with Tyr81 and Trp260 at the tunnel, and established π– π interactions with Phe273 and Trp274 in the hydrophobic pocket (Table 6, Table 1) [65]. Lastly, these type IV compounds were used in in vitro cell assays, where they showed a decrease of LPA‐driven downstream phosphorylation of Akt in BJeH fibroblasts [65]. The Belgo‐Dutch company Galapagos NV serendipitously discovered the only other confirmed type IV compound series. They used a high‐throughput screen, where they identified an imidazo [1,2‐a] pyridine series, from which the very potent compounds GLPG1690 (IC50 = 27 nM; LPC) [51] and AJ‐Cpd‐9 (IC50 = 357 nM; LPC) [41] were obtained by structure‐activity evolution (Table 6). Moreover, the crystal structures of these compounds confirmed their binding mode to ATX. GLPG1690 makes a hydrogen bond with Trp254, π‐π interactions with Phe274, and further hydrophobic interactions with Phe250 and Phe275. Conversely, AJ‐Cpd‐9 makes a hydrogen bond with Trp260 and hydrophobic interactions with Trp254, Phe250, Phe260 and Phe275 (Table 6, Table 1). After pharmacokinetics and pharmacodynamics analyses [76,77], in vivo experiments showed that administration of GLPG1690 to bleomycin‐treated mice resulted in a reduction of lung fibrosis, which was linked to a dose‐dependent reduction of plasma LPA 18:2 levels (90%) (Table 2) [41,51]. This compound has shown very promising results in advanced clinical trials against idiopathic lung fibrosis (IPF) and has recently progressed to phase III clinical trials for IPF [78].

4. Conclusions The different types of ATX inhibitors could have different outcomes that go beyond the simple orthosteric inhibition of catalytic activity. While we know that LPA binding to the allosteric site is likely increasing the catalytic rate of ATX, possible roles of the tunnel in delivering LPA to its receptors, perhaps in an interplay between surface integrins and proteoglycans, have not been exploited. The fact that EDG and non‐EDG receptors have different binding pocket characteristics could imply that LPA delivery through the tunnel could be receptor‐dependent.

Cancers 2019, 11, x 11 of 17

LM‐Cpd 51

The identification of type II indole‐based ATX inhibitors, as discussed above, also yielded a type III compound, namely PAT‐347 (IC50 = 0.3 nM; LPC) (Table 5) [50]. The crystal structure of ATX bound to both PAT‐347 and 14:0 LPA confirmed its non‐competitive binding mode. PAT‐347 accommodates at the tunnel by π–π interactions between its indole moiety and Phe274 and His251. Moreover, its benzoic acid makes another π–π interaction with Phe249, and it also contacts Lys248, Trp254, and Trp260. More recently, the pharmacokinetic properties of two novel compounds, PAT‐505 (IC50 = 2 nM; LPC) and PAT‐048 (IC50 = 1.1 nM; LPC), have been assessed (Table 5) [70,72–74]. The crystal structure of PAT‐505 bound to ATX showed a very similar binding mode to that of PAT‐347, namely, by interacting with Lys248, Phe249, His251, Trp254, Trp260 and Phe274, but also Ser81 and Val277 (Table 1). Clinical assays with PAT‐048, showed better pharmacodynamics results, and this was used for in vivo assays with a bleomycin‐induced mouse model for dermal fibrosis. Pharmacological inhibition of ATX activity markedly attenuated skin fibrosis upon treatment with 10 mg kg‐1 PAT‐ 048, which related to a 75% inhibition of ATX activity after 24 h and >90% at double dose (Table 2). It is worthy to mention that five new patented PharmAkea compounds, specifically, PharmAkea‐Cpd A‐E (IC50 = < 0.5 μM; LPC) have been recently used in metabolic disorder treatment, where they showed a decrease of fasting blood glucose levels in mice fed with a high‐fat diet (Table 2) [49]. Recently, another series of indole‐derived type III compound has been reported, resulting from a structure‐activity evolution of a compound reported by Amira Pharmaceuticals [75], which led to the identification of LM‐Cpd 51 [71] (Table 5). The co‐crystal structure with rat ATX showed that the compound binds in the tunnel, establishing stacking interactions with Trp254, as well as hydrophobic interactions with Phe249, Trp260 and Phe274 (Table 1).

3.4. Type IV Inhibitors Type IV compounds occupy the binding pocket and the tunnel, but do not contact the catalytic Cancerssite. 2019Such, 11 compounds, x have been discovered either by design, fusing parts of a type I and a12 type of 17 III

inhibitor,Cancers 2019 such, 11, 1577as in FP‐Cpd‐17 [65], or by serendipity, during a high‐throughput screen followed12 by of 17 specific structure‐basedH3C Ndesign, such as in GLPG1690 [51] (Table 6).

N

S CH3 N O Table 6. Type IV ATX inhibitorsN basedTable on their 6. Cont. kinetic and crystallographic analysis. CH3 S N N CH3 O 5m7m 357 nM [41]

F TypeType IV IV Inhibitor Inhibitor PDBPDB ID ID Activity Activity (IC (IC50)50 )Reference Reference AJ‐Cpd 9

CH3

H3C N

N N N O S N N 5m0s 202 nM [65] CH3 N 5mhp 27 nM [51] N 5mhp 27 nM [51] OH FP‐Cpd 5 F

Cl OH GLPG1690

N

O N H3C We reportedCl the structure-guidedCH3 design and chemical5m0d evolution814 nM of bile[65] salts, from weak We reported the structure‐guided designH3C andOH chemical evolution of bile salts, from weak O physiologicalphysiological non non-competitive‐competitive inhibitors inhibitors into into potent potent competitive competitive type type IV compounds IV compounds [65]. [65 This]. This was was achievedachieved by using by using the PF the‐8380 PF-8380FP dichlorocarbamate‐Cpd dichlorocarbamate11 moiety moiety as a pocket as a pocket-binding‐binding lipophilic lipophilic portion. portion. By design,By design, these compounds these compounds did not did interact not interact with the with active the activesite residues, site residues, but still but hampered still hampered LPC or LPC LPAor binding. LPA binding. The best The lead best compounds lead compounds differed di chieflyffered chieflyregarding regarding spacers spacers connecting connecting the steroid the and steroid and the dichrolocarbamate moieties, i.e., linear amide linker, piperazine, and piperidine for FP-Cpd-5 the dichrolocarbamate moieties, i.e., linear amide linker, 5m0mpiperazine, 20and nM piperidine [65] for FP‐Cpd‐5 (IC50(IC = 50202= nM;202 nMLPC),; LPC), ‐11 (IC -1150 (IC = 81450 = nM;814 LPC), nM; LPC), and ‐ and17 (IC -1750 (IC = 2050 nM;= 20 LPC), nM; LPC), respectively respectively (Table (Table 6) [65].6)[ 65].

Moreover,Moreover, a very a very short short spacer spacer yielded yielded no no measurable inhibitoryinhibitory activity, activity, which which explains explains why why FP-Cpd-11 FP‐ Cpdexhibited‐11 exhibited a five-fold a five‐fold lower FPlower‐ activityCpd activity 17 than than FP-Cpd-17. FP‐Cpd‐17. These These compounds compounds were were co-crystalized co‐crystalized with withrat rat ATX, ATX, showing showing how how thesethese types of of inhibitors inhibitors accommodate accommodate at at both both the the tunnel tunnel and and the thepocket. pocket.

Specifically,Specifically, they they all made all made hydrogen hydrogen bonds bonds with withTyr81 Tyr81 and Trp260 and Trp260 at the at tunnel, the tunnel, and established and established π– π interactionsπ–π interactions with Phe273 with Phe273 and Trp274 and Trp274in the hydrophobic in the hydrophobic pocket (Table pocket 6, (Table Table 61), Table[65]. Lastly,1)[ 65 ].these Lastly, typethese IV compounds type IV compounds were used were in usedin vitro in in cell vitro assays,cell assays, where where they showed they showed a decrease a decrease of LPA of LPA-driven‐driven downstreamdownstream phosphorylation phosphorylation of Akt of Akt in BJeH in BJeH fibroblasts fibroblasts [65]. [65 ]. TheThe Belgo Belgo-Dutch‐Dutch company company Galapagos Galapagos NV NVserendipitously serendipitously discovered discovered the theonly only other other confirmed confirmed typetype IV compound IV compound series. series. They They used used a high a high-throughput‐throughput screen, screen, where where they they identified identified an imidazo an imidazo [1,2[1,2-a]‐a] pyridine pyridine series, series, from from which which the the very very potent potent compounds compounds GLPG1690 GLPG1690 (IC(IC5050 = 27 nM; LPC)LPC) [ [51]51] and andAJ-Cpd-9 AJ‐Cpd‐9 (IC (IC50 50= =357 357 nM; nM; LPC) LPC) [41 ][41] were were obtained obtained by structure-activity by structure‐activity evolution evolution (Table 6(Table). Moreover, 6). Moreover,the crystal the structures crystal structures of these compounds of these compounds confirmed theirconfirmed binding their mode binding to ATX. mode GLPG1690 to ATX. makes GLPG1690a hydrogen makes bond a withhydrogen Trp254, bondπ-π interactionswith Trp254, with π‐π Phe274,interactions and further with hydrophobicPhe274, and interactions further hydrophobicwith Phe250 interactions and Phe275. with Conversely, Phe250 and AJ-Cpd-9 Phe275. makes Conversely, a hydrogen AJ‐Cpd bond‐9 with makes Trp260 a hydrogen and hydrophobic bond withinteractions Trp260 and with hydrophobic Trp254, Phe250, interactions Phe260 with and Trp254, Phe275 Phe250, (Table6 ,Phe260 Table1). and After Phe275 pharmacokinetics (Table 6, Table and 1). Afterpharmacodynamics pharmacokinetics analyses and pharmacodynamics [76,77], in vivo experiments analyses [76,77], showed in thatvivo administration experiments showed of GLPG1690 that administrationto bleomycin-treated of GLPG1690 mice to resulted bleomycin in a reduction‐treated mice of lung resulted fibrosis, in a which reduction was linked of lung to fibrosis, a dose-dependent which wasreduction linked to of a plasmadose‐dependent LPA 18:2 levels reduction (90%) of (Table plasma2)[ 41LPA,51 ].18:2 This levels compound (90%) has(Table shown 2) [41,51]. very promising This compoundresults in has advanced shown clinicalvery promising trials against results idiopathic in advanced lung fibrosisclinical (IPF)trials and against has recentlyidiopathic progressed lung fibrosisto phase (IPF) III and clinical has recently trials for progressed IPF [78]. to phase III clinical trials for IPF [78].

4. Conclusions 4. Conclusions The different types of ATX inhibitors could have different outcomes that go beyond the simple The different types of ATX inhibitors could have different outcomes that go beyond the simple orthosteric inhibition of catalytic activity. While we know that LPA binding to the allosteric site is orthosteric inhibition of catalytic activity. While we know that LPA binding to the allosteric site is likely increasing the catalytic rate of ATX, possible roles of the tunnel in delivering LPA to its receptors, likely increasing the catalytic rate of ATX, possible roles of the tunnel in delivering LPA to its perhaps in an interplay between surface integrins and proteoglycans, have not been exploited. The fact receptors, perhaps in an interplay between surface integrins and proteoglycans, have not been that EDG and non-EDG receptors have different binding pocket characteristics could imply that LPA exploited. The fact that EDG and non‐EDG receptors have different binding pocket characteristics delivery through the tunnel could be receptor-dependent. could imply that LPA delivery through the tunnel could be receptor‐dependent. Type I inhibitors would effectively stop LPA production from LPC, but would not affect any independent function of the allosteric site. Type IV inhibitors, however, would also abolish any

Cancers 2019, 11, 1577 13 of 17 functionality of the allosteric site, expelling bound LPA. We suggest that the clinical success of the type IV compound, GLPG1690, is not independent from its mode of binding and inhibition, interfering with the allosteric tunnel. Albeit some ATX inhibitors have shown promising results in animal cancer models [55,79,80], none have progressed to clinical trials, at least to our knowledge. One should consider the different types of inhibitors in the context of cancer therapy. We suggest that a type IV inhibitor occupying the low-affinity allosteric tunnel could prove much more effective, in the context of a tumour with high local LPA concentration that could displace the inhibitor from the high-affinity orthosteric site. It is our expectation that Autotaxin inhibitors will have a dynamic comeback in the context of cancer therapy.

Funding: Funding for ATX research has been provided by NWO-TOP (700.10.354) and by the Oncode Institute to AP. Acknowledgments: The authors would like to thank Wouter Moolenaar for critically reading the manuscript. Conflicts of Interest: The authors declare no conflicts of interest.

References

1. van Corven, E.; Groenink, A.; Jalink, K.; Eichholtz, T.; Moolenaar, W. Lysophosphatidate-induced cell proliferation: Identification and dissection of signaling pathways mediated by G proteins. Cell 1989, 59, 45–54. [CrossRef] 2. Jalink, K.; Van Corven, E.J.; Moolenaar, W.H. Lysophosphatidic acid, but not phosphatidic acid, is a potent Ca2+-mobilizing stimulus for fibroblasts: Evidence for an extracellular site of action. J. Biol. Chem. 1990, 265, 12232–12239. [PubMed] 3. Wang, J.; Sun, Y.; Qu, J.; Yan, Y.; Yang, Y.; Cai, H. Roles of LPA receptor signaling in breast cancer. Expert Rev. Mol. Diagn. 2016, 16, 1103–1111. [CrossRef][PubMed] 4. Kaffe, E.; Katsifa, A.; Xylourgidis, N.; Ninou, I.; Zannikou, M.; Harokopos, V.; Foka, P.; Dimitriadis, A.; Evangelou, K.; Moulas, A.N.; et al. Hepatocyte autotaxin expression promotes liver fibrosis and cancer. Hepatology 2017, 65, 1369–1383. [CrossRef] 5. Tager, A.M.; LaCamera, P.; Shea, B.S.; Campanella, G.S.; Selman, M.; Zhao, Z.; Polosukhin, V.; Wain, J.; Karimi-Shah, B.A.; Kim, N.D.; et al. The lysophosphatidic acid receptor LPA1 links pulmonary fibrosis to lung injury by mediating fibroblast recruitment and vascular leak. Nat. Med. 2008, 14, 45–54. [CrossRef] 6. Kremer, A.E.; Martens, J.J.W.W.; Kulik, W.; Ruëff, F.; Kuiper, E.M.M.; van Buuren, H.R.; van Erpecum, K.J.; Kondrackiene, J.; Prieto, J.; Rust, C.; et al. Lysophosphatidic acid is a potential mediator of cholestatic pruritus. Gastroenterology 2010, 139, 1008–1018. [CrossRef] 7. Thirunavukkarasu, K.; Tan, B.; Swearingen, C.A.; Rocha, G.; Bui, H.H.; McCann, D.J.; Jones, S.B.; Norman, B.H.; Pfeifer, L.A.; Saha, J.K. Pharmacological characterization of a potent inhibitor of autotaxin in animal models of inflammatory bowel disease and multiple sclerosis. J. Pharmacol. Exp. Ther. 2016, 359, 207–214. [CrossRef] 8. Barbayiannia, E.; Kaffeb, E.; Aidinisb, V.; Kokotos, G. Autotaxin, a secreted lysophospholipase D, as a promising therapeutic target in chronic inflammation and cancer. Prog. Lipid Res. 2015, 58, 76–96. [CrossRef] 9. Benesch, M.G.K.; Tang, X.; Venkatraman, G.; Bekele, R.T.; Brindley, D.N. Recent advances in targeting the autotaxin-lysophosphatidate-lipid phosphate phosphatase axis in vivo. J. Biomed. Res. 2016, 30, 272–284. 10. Chrencik, J.E.; Roth, C.B.; Terakado, M.; Kurata, H.; Omi, R.; Kihara, Y.; Warshaviak, D.; Nakade, S.; Asmar-Rovida, G.; Mileni, M.; et al. Crystal structure of antagonist bound human lysophosphatidic acid receptor 1. Cell 2015, 161, 1633–1643. [CrossRef] 11. Taniguchi, R.; Inoue, A.; Sayama, M.; Uwamizu, A.; Yamashita, K.; Hirata, K.; Yoshida, M.; Tanaka, Y.; Kato, H.E.; Nakada-Nakura, Y.; et al. Structural insights into ligand recognition by the lysophosphatidic acid receptor LPA6. Nature 2017, 548, 356–360. [CrossRef][PubMed] 12. Zhang, H.; Pluhackova, K.; Jiang, Z.; Böckmann, R.A. Binding characteristics of sphingosine-1-phosphate to APOM hints to assisted release mechanism via the APOM calyx-opening. Sci. Rep. 2016, 6, e30655. [CrossRef][PubMed] 13. Stanley, N.; Pardo, L.; Fabritiis, G. De The pathway of ligand entry from the membrane bilayer to a lipid G protein-coupled receptor. Sci. Rep. 2016, 6, 1–9. [CrossRef][PubMed] Cancers 2019, 11, 1577 14 of 17

14. Albers, H.M.H.G.; Dong, A.; van Meeteren, L.A.; Egan, D.A.; Sunkara, M.; van Tilburg, E.W.; Schuurman, K.; van Tellingen, O.; Morris, A.J.; Smyth, S.S.; et al. Boronic acid-based inhibitor of autotaxin reveals rapid turnover of LPA in the circulation. Proc. Natl. Acad. Sci. USA 2010, 107, 7257–7262. [CrossRef][PubMed] 15. Tomsig, J.L.; Snyder, A.H.; Berdyshev, E.V.; Skobeleva, A.; Mataya, C.; Natarajan, V.; Brindley, D.N.; Lynch, K.R. Lipid phosphate phosphohydrolase type 1 (LPP1) degrades extracellular lysophosphatidic acid in vivo. Biochem. J. 2009, 419, 611–618. [CrossRef][PubMed] 16. Tang, X.; Benesch, M.G.K.; Dewald, J.; Zhao, Y.Y.; Patwardhan, N.; Santos, W.L.; Curtis, J.M.; McMullen, T.P.W.; Brindley, D.N. Lipid phosphate phosphatase-1 expression in cancer cells attenuates tumor growth and metastasis in mice. J. Lipid Res. 2014, 55, 2389–2400. [CrossRef] 17. Jasinska, R.; Zhang, Q.; Pilquil, C.; Singh, I.; Xu, J.; Dewald, J.; Dillon, D.A.; Berthiaume, L.G.; Carman, G.M.; Waggoner, D.W.; et al. Lipid phosphate phosphohydrolase-1 degrades exogenous glycerolipid and sphingolipid phosphate esters. Biochem. J. 1999, 340, 677–686. [CrossRef] 18. van Meeteren, L.A.; Moolenaar, W.H. Regulation and biological activities of the autotaxin-LPA axis. Prog. Lipid Res. 2007, 46, 145–160. [CrossRef] 19. Aoki, J.; Taira, A.; Takanezawa, Y.; Kishi, Y.; Hama, K.; Kishimoto, T.; Mizuno, K.; Saku, K.; Taguchi, R.; Arai, H. Serum lysophosphatidic acid is produced through diverse phospholipase pathways. J. Biol. Chem. 2002, 277, 48737–48744. [CrossRef] 20. Umezu-Goto, M.; Kishi, Y.; Taira, A.; Hama, K.; Dohmae, N.; Takio, K.; Yamori, T.; Mills, G.B.; Inoue, K.; Aoki, J.; et al. Autotaxin has lysophospholipase D activity leading to tumor cell growth and motility by lysophosphatidic acid production. J. Cell Biol. 2002, 158, 227–233. [CrossRef] 21. Matralis, A.N.; Afantitis, A.; Aidinis, V. Development and therapeutic potential of autotaxin small molecule inhibitors: From bench to advanced clinical trials. Med. Res. Rev. 2019, 39, 976–1013. [CrossRef][PubMed] 22. Tokumura, A.; Majima, E.; Kariya, Y.; Tominaga, K.; Kogure, K.; Yasuda, K.; Fukuzawa, K. Identification of human plasma lysophospholipase D, a lysophosphatidic acid-producing enzyme, as autotaxin, a multifunctional phosphodiesterase. J. Biol. Chem. 2002, 277, 39436–39442. [CrossRef][PubMed] 23. Jansen, S. Proteolytic maturation and activation of autotaxin (NPP2), a secreted metastasis-enhancing lysophospholipase D. J. Cell Sci. 2005, 118, 3081–3089. [CrossRef][PubMed] 24. Hausmann, J.; Keune, W.-J.; Hipgrave Ederveen, A.L.; van Zeijl, L.; Joosten, R.P.; Perrakis, A. Structural snapshots of the catalytic cycle of the phosphodiesterase autotaxin. J. Struct. Biol. 2016, 195, 199–206. [CrossRef] 25. Hausmann, J.; Kamtekar, S.; Christodoulou, E.; Day, J.E.; Wu, T.; Fulkerson, Z.; Albers, H.M.H.G.H.G.; Van Meeteren, L.A.; Houben, A.J.S.S.; Van Zeijl, L.; et al. Structural basis for substrate discrimination and integrin binding by autotaxin. Nat. Struct. Mol. Biol. 2011, 18, 198–204. [CrossRef] 26. Keune, W.; Hausmann, J.; Bolier, R.; Tolenaars, D.; Kremer, A.; Heidebrecht, T.; Joosten, R.; Sunkara, M.; Morris, A.; Matas-Rico, E.; et al. Steroid binding to autotaxin links bile salts and lysophosphatidic acid signalling. Nat. Commun. 2016, 7, e11248. [CrossRef] 27. Nishimasu, H.; Okudaira, S.; Hama, K.; Mihara, E.; Dohmae, N.; Inoue, A.; Ishitani, R.; Takagi, J.; Aoki, J.; Nureki, O. Crystal structure of autotaxin and insight into GPCR activation by lipid mediators. Nat. Struct. Mol. Biol. 2011, 18, 205–212. [CrossRef] 28. Salgado-Polo, F.; Fish, A.; Matsoukas, M.-T.; Heidebrecht, T.; Keune, W.-J.; Perrakis, A. Lysophosphatidic acid produced by Autotaxin acts as an allosteric modulator of its catalytic efficiency. J. Biol. Chem. 2018, 293, 14312–14327. [CrossRef] 29. Saunders, L.P.; Cao, W.; Chang, W.C.; Albright, R.A.; Braddock, D.T.; De La Cruz, E.M. Kinetic analysis of autotaxin reveals substrate-specific catalytic pathways and a mechanism for lysophosphatidic acid distribution. J. Biol. Chem. 2011, 286, 30130–30141. [CrossRef] 30. Jethwa, S.A.; Leah, E.J.; Zhang, Q.; Bright, N.A.; Oxley, D.; Bootman, M.D.; Rudge, S.A.; Wakelam, M.J.O. Exosomes bind to autotaxin and act as a physiological delivery mechanism to stimulate LPA receptor signalling in cells. J. Cell Sci. 2016, 129, 3948–3957. [CrossRef] 31. Thumser, A.E.A.; Wilton, D.C. The binding of natural and fluorescent lysophospholipids to wild-type and mutant rat liver fatty acid-binding protein and albumin. Biochem. J. 1995, 307, 305–311. [CrossRef] 32. Leblanc, R.; Lee, S.C.; David, M.; Bordet, J.C.; Norman, D.D.; Patil, R.; Miller, D.; Sahay, D.; Ribeiro, J.; Clezardin, P.; et al. Interaction of platelet-derived autotaxin with tumor integrin alphaVbeta3 controls metastasis of breast cancer cells to bone. Blood 2014, 124, 3141–3150. [CrossRef][PubMed] Cancers 2019, 11, 1577 15 of 17

33. Wu, T.; Vander Kooi, C.; Shah, P.; Charnigo, R.; Huang, C.; Smyth, S.S.; Morris, A.J. Integrin-mediated cell surface recruitment of autotaxin promotes persistent directional cell migration. FASEB J. 2014, 28, 861–870. [CrossRef][PubMed] 34. Fulkerson, Z.; Wu, T.; Sunkara, M.; Vander Kooi, C.; Morris, A.J.; Smyth, S.S. Binding of autotaxin to integrins localizes lysophosphatidic acid production to platelets and mammalian cells. J. Biol. Chem. 2011, 286, 34654–34663. [CrossRef][PubMed] 35. Pamuklar, Z.; Federico, L.; Liu, S.; Umezu-Goto, M.; Dong, A.; Panchatcharam, M.; Fulerson, Z.; Berdyshev, E.; Natarajan, V.; Fang, X.; et al. Autotaxin/Lysopholipase D and lysophosphatidic acid regulate murine hemostasis and thrombosis. J. Biol. Chem. 2009, 284, 7385–7394. [CrossRef][PubMed] 36. Houben, A.J.S.; Van Wijk, X.M.R.; Van Meeteren, L.A.; Van Zeijl, L.; Van De Westerlo, E.M.A.; Hausmann, J.; Fish, A.; Perrakis, A.; Van Kuppevelt, T.H.; Moolenaar, W.H. The polybasic insertion in autotaxin α confers specific binding to heparin and cell surface heparan sulfate proteoglycans. J. Biol. Chem. 2013, 288, 510–519. [CrossRef][PubMed] 37. Castagna, D.; Budd, D.C.; Macdonald, S.J.F.F.; Jamieson, C.; Watson, A.J.B.B. Development of Autotaxin Inhibitors: An Overview of the Patent and Primary Literature. J. Med. Chem. 2016, 59, 5604–5621. [CrossRef] 38. Nikolaou, A.; Kokotou, M.G.; Limnios, D.; Psarra, A.; Kokotos, G. Autotaxin inhibitors: A patent review (2012–2016). Expert Opin. Ther. Pat. 2017, 27, 815–829. [CrossRef] 39. Lynch, K.R.; Macdonald, T.L. Phosphonate Derivatives as Autotaxin Inhibitors. Patent US 8378100, 19 February 2013. 40. Kato, K.; Ikeda, H.; Miyakawa, S.; Futakawa, S.; Nonaka, Y.; Fujiwara, M.; Okudaira, S.; Kano, K.; Aoki, J.; Morita, J.; et al. Structural basis for specific inhibition of Autotaxin by a DNA aptamer. Nat. Struct. Mol. Biol. 2016, 23, 395–401. [CrossRef] 41. Joncour, A.; Desroy, N.; Housseman, C.; Bock, X.; Bienvenu, N.; Cherel, L.; Labeguere, V.; Peixoto, C.; Annoot, D.; Lepissier, L.; et al. Supporting information Discovery, Structure-Activity Relationship and Binding Mode of an Imidazo [1,2-a] pyridine Series of Autotaxin Inhibitors. J. Med. Chem. 2017, 60, 7371–7392. 42. Ninou, I.; Kaffe, E.; Müller, S.; Budd, D.C.; Stevenson, C.S.; Ullmer, C.; Aidinis, V. Pharmacologic targeting of the ATX/LPA axis attenuates bleomycin-induced pulmonary fibrosis. Pulm. Pharmacol. Ther. 2018, 52, 32–40. [CrossRef][PubMed] 43. Ninou, I.; Magkrioti, C.; Aidinis, V. Autotaxin in pathophysiology and pulmonary fibrosis. Front. Med. 2018, 5, 1–11. [CrossRef][PubMed] 44. Bhave, S.R.; Dadey, D.Y.A.; Karvas, R.M.; Ferraro, D.J.; Kotipatruni, R.P.; Jaboin, J.J.; Hallahan, A.N.; DeWees, T.A.; Linkous, A.G.; Hallahan, D.E.; et al. Autotaxin inhibition with PF-8380 enhances the radiosensitivity of human and murine glioblastoma cell lines. Front. Oncol. 2013, 3, 1–11. [CrossRef] 45. Nikolaou, A.; Ninou, I.; Kokotou, M.G.; Kaffe, E.; Afantitis, A.; Aidinis, V.; Kokotos, G. Hydroxamic acids constitute a novel class of autotaxin inhibitors that exhibit in vivo efficacy in a pulmonary fibrosis model. J. Med. Chem. 2018, 61, 3697–3711. [CrossRef] 46. Lee, S.; Fujiwara, Y.; Liu, J.; Yue, J.; Shimizu, Y.; Norman, D.; Wang, Y.; Tsukahara, R.; Szabo, E.; Patil, R.; et al. Autotaxin, LPA receptors (1 and 5) exert disparate functions in tumor cells versus the host tissue microenvironment in melanoma invasion and metastasis. Mol Cancer Res 2015, 13, 174–185. [CrossRef] 47. Oikonomou, N.; Mouratis, M.A.; Tzouvelekis, A.; Kaffe, E.; Valavanis, C.; Vilaras, G.; Karameris, A.; Prestwich, G.D.; Bouros, D.; Aidinis, V. Pulmonary autotaxin expression contributes to the pathogenesis of pulmonary fibrosis. Am. J. Respir. Cell Mol. Biol. 2012, 47, 566–574. [CrossRef] 48. Park, G.Y.; Lee, Y.G.; Berdyshev, E.; Nyenhuis, S.; Du, J.; Fu, P.; Gorshkova, I.A.; Li, Y.; Chung, S.; Karpurapu, M.; et al. Autotaxin production of lysophosphatidic acid mediates allergic asthmatic inflammation. Am. J. Respir. Crit. Care Med. 2013, 188, 928–940. [CrossRef] 49. Evans, J.F. Methods and Compositions for the Treatment of Metabolic Disorders. Patent WO 2016028686, 25 February 2016. 50. Stein, A.J.; Bain, G.; Prodanovich, P.; Santini, A.M.; Darlington, J.; Stelzer, N.M.P.; Sidhu, R.S.; Schaub, J.; Goulet, L.; Lonergan, D.; et al. Structural basis for inhibition of human autotaxin by four potent compounds with distinct modes of binding. Mol. Pharmacol. 2015, 88, 982–992. [CrossRef] Cancers 2019, 11, 1577 16 of 17

51. Desroy, N.; Housseman, C.; Bock, X.; Joncour, A.; Bienvenu, N.; Cherel, L.; Labeguere, V.; Rondet, E.; Peixoto, C.; Grassot, J.-M.; et al. Discovery of 2-[[2-Ethyl-6-[4-[2-(3-hydroxyazetidin-1-yl)-2-oxoethyl] piperazin-1-yl]-8-methylimidazo [1,2-a] pyridin-3-yl] methylamino]-4-(4-fluorophenyl) thiazole-5-carbonitrile (GLPG1690), a first-in-class autotaxin inhibitor undergoing clinical Ev. J. Med. Chem. 2017, 15, 3580–3590. [CrossRef] 52. Study to Assess Safety, Tolerability, Pharmacokinetic and Pharmacodynamic Properties of GLPG1690. Available online: https://clinicaltrials.gov/ct2/show/NCT02738801 (accessed on 8 October 2019). 53. Benesch, M.G.K.; Tang, X.; Maeda, T.; Ohhata, A.; Zhao, Y.Y.; Kok, B.P.C.; Dewald, J.; Hitt, M.; Curtis, J.M.; McMullen, T.P.W.; et al. Inhibition of autotaxin delays breast tumor growth and lung metastasis in mice. FASEB J. 2014, 28, 2655–2666. [CrossRef][PubMed] 54. Benesch, M.G.K.; Zhao, Y.Y.; Curtis, J.M.; McMullen, T.P.W.; Brindley, D.N. Regulation of autotaxin expression and secretion by lysophosphatidate and sphingosine 1-phosphate. J. Lipid Res. 2015, 56, 1134–1144. [CrossRef] [PubMed] 55. Baker, L.D.; Fujiwara, Y.; Pigg, K.; Tsukahara, R.; Kobatashi, S.; Murofushi, H.; Uchiyama, A.; Murakami-Murofushi, K.; Murph, E.; Mills, G.B.; et al. Carba analogs of cyclic phosphatidic acid are selective inhibitors of autotaxin and cancer cell invasion and metastasis. J. Biol. Chem. 2006, 281, 22786–22793. [CrossRef][PubMed] 56. Durgam, G.G.; Virag, T.; Walker, M.D.; Tsukahara, R.; Yasuda, S.; Liliom, K.; van Meeteren, L.A.; Moolenaar, W.H.; Wilke, N.; Siess, W.; et al. Synthesis, structure activity relationships, and biological − evaluation of fatty alcohol phosphates as lysophosphatidic acid receptor ligands, activators of PPARγ, and inhibitors of autotaxin. J. Med. Chem. 2005, 48, 4919–4930. [CrossRef] 57. Nikitopoulou, I.; Kaffe, E.; Sevastou, I.; Sirioti, I.; Samiotaki, M.; Madan, D.; Prestwich, G.D.; Aidinis, V. A Metabolically-stabilized phosphonate analog of lysophosphatidic acid attenuates collagen-induced arthritis. PLoS ONE 2013, 87, e70941. [CrossRef] 58. van Meeteren, L.A.; Brinkmann, V.; Saulnier-Blache, J.S.; Lynch, K.R.; Moolenaar, W.H. Anticancer activity of FTY720: Phosphorylated FTY720 inhibits autotaxin, a metastasis-enhancing and angiogenic lysophospholipase D. Cancer Lett. 2008, 266, 203–208. [CrossRef] 59. Ferry, G.; Moulharat, N.; Pradere, J.-P.; Desos, P.; Try, A.; Genton, A.; Giganti, A.; Beucher-Gaudin, M.; Lonchampt, M.; Bertrand, M.; et al. S32826, A Nanomolar Inhibitor of Autotaxin: Discovery, Synthesis and Applications as a pharmacological tool. J. Pharmacol. Exp. Ther. 2008, 327, 809–819. [CrossRef] 60. Albers, H.M.H.G.; Hendrickx, L.J.D.; van Tol, R.J.P.; Hausmann, J.; Perrakis, A.; Ovaa, H. Structure-based design of novel boronic acid-based inhibitors of autotaxin. J. Med. Chem. 2011, 54, 4619–4626. [CrossRef] 61. Jones, S.B.; Pfeifer, L.A.; Bleisch, T.J.; Beauchamp, T.J.; Durbin, J.D.; Klimkowski, V.J.; Hughes, N.E.; Rito, C.J.; Dao, Y.; Gruber, J.M.; et al. Novel autotaxin inhibitors for the treatment of osteoarthritis pain: Lead optimization via structure-based drug design. ACS Med. Chem. Lett. 2016, 7, 857–861. [CrossRef] 62. Gierse, J.; Thorarensen, A.; Beltey, K.; Bradshaw-pierce, E.; Cortes-burgos, L.; Hall, T.; Johnston, A.; Murphy, M.; Nemirovskiy, O.; Ogawa, S.; et al. A novel autotaxin inhibitor reduced lysophosphatidic acid levels in plasma and the site of inflammation. J. Pharmacol. Exp. Ther. 2010, 2, 310–317. [CrossRef] 63. Kawaguchi, M.; Okabe, T.; Okudaira, S.; Nishimasu, H.; Ishitani, R.; Kojima, H.; Nureki, O.; Aoki, J.; Nagano, T. Screening and X-ray crystal structure-based optimization of autotaxin (ENPP2) inhibitors, using a newly developed fluorescence probe. ACS Chem. Biol. 2013, 8, 1713–1721. [CrossRef] 64. Kuttruff, C.A.; Ferrara, M.; Bretschneider, T.; Hoerer, S.; Handschuh, S.; Nosse, B.; Romig, H.; Nicklin, P.; Roth, G.J. Discovery of BI-2545: A novel autotaxin inhibitor that significantly reduces LPA levels in vivo. ACS Med. Chem. Lett. 2017, 8, 1252–1257. [CrossRef] 65. Keune, W.J.; Potjewyd, F.; Heidebrecht, T.; Salgado-Polo, F.; Macdonald, S.J.F.; Chelvarajan, L.; Abdel Latif, A.; Soman, S.; Morris, A.J.; Watson, A.J.B.; et al. Rational design of autotaxin inhibitors by structural evolution of endogenous modulators. J. Med. Chem. 2017, 60, 2006–2017. [CrossRef] 66. Katsifa, A.; Kaffe, E.; Nikolaidou-Katsaridou, N.; Economides, A.N.; Newbigging, S.; McKerlie, C.; Aidinis, V. The bulk of autotaxin activity is dispensable for adult mouse life. PLoS ONE 2015, 10, 1–14. [CrossRef] 67. Weng, J.; Jiang, S.; Ding, L.; Xu, Y.; Zhu, X.; Jin, P. Autotaxin/lysophosphatidic acid signaling mediates obesity-related cardiomyopathy in mice and human subjects. J. Cell. Mol. Med. 2019, 23, 1050–1058. [CrossRef] 68. Jiang, G.; Madan, D.; Prestwich, D.G. Aromatic phosphonates inhibit the lydophosphodiesterase D activity of autotaxin. Bioorg. Med. Chem. Lett. 2011, 21, 5098–5101. [CrossRef] Cancers 2019, 11, 1577 17 of 17

69. Shah, P.; Cheasty, A.; Foxton, C.; Raynham, T.; Farooq, M.; Gutierrez, I.F.; Lejeune, A.; Pritchard, M.; Turnbull, A.; Pang, L.; et al. Discovery of potent inhibitors of the lysophospholipase autotaxin. Bioorg. Med. Chem. Lett. 2016, 26, 5403–5410. [CrossRef] 70. Bain, G.; Shannon, K.E.; Huang, F.; Darlington, J.; Goulet, L.; Prodanovich, P.; Ma, G.L.; Santini, A.M.; Stein, A.J.; Lonergan, D.; et al. Selective inhibition of autotaxin is efficacious in mouse models of liver fibrosis. J. Pharmacol. Exp. Ther. 2017, 360, 1–13. [CrossRef] 71. Miller, L.M.; Keune, W.J.; Castagna, D.; Young, L.C.; Duffy, E.L.; Potjewyd, F.; Salgado-Polo, F.; García, P.E.; Semaan, D.; Pritchard, J.M.; et al. Structure-activity relationships of small molecule autotaxin inhibitors with a discrete binding mode. J. Med. Chem. 2017, 60, 722–748. [CrossRef] 72. Black, K.E.; Berdyshev, E.; Bain, G.; Castelino, F.V.; Shea, B.S.; Probst, C.K.; Fontaine, B.A.; Bronova, I.; Goulet, L.; Lagares, D.; et al. Autotaxin activity increases locally following lung injury, but is not required for pulmonary lysophosphatidic acid production or fibrosis. FASEB J. 2016, 30, 2435–2450. [CrossRef] 73. Castelino, F.V.; Bain, G.; Pace, V.A.; Black, K.E.; George, L.; Probst, C.K.; Goulet, L.; Lafyatis, R.; Tager, A.M. An Autotaxin-LPA-IL-6 amplification loop drives scleroderma fibrosis. Arthritis Rheumatol. 2016, 68, 2964–2974. [CrossRef] 74. Roppe, J.R.; Parr, T.A.; Stock, N.S.; Volkots, D.; Hutchinson, J.H. Autotaxin Inhibitors and Uses Thereof. Patent WO 2012166415, 6 December 2012. 75. Hutchinson, J.H.; Parr, T.A.; Roppe, J.R.; Stock, N.S.; Volkots, D. Heterocyclic autotaxin inhibitors and uses thereof. Patent US 2013029948, 31 January 2013. 76. Blanque, R.D.N.; Dupont, S.; Cottereaux, C.; Marsais, F.; Lepescheux, L.; Monjardet, A.; Wakselman, E.; Laenen, W.; Russell, V.; van der Aar, E.; et al. Pharmacological Profile and Efficacy of GLPG1690, a Novel ATX Inhibitor for COPD Treatment. 2015. Available online: http://files.glpg.com/docs/website_1/Poster_ ERS_2015_final.pdf (accessed on 8 October 2019). 77. Maher, T.M.; van der Aar, E.M.; Van de Steen, O.; Allamassey, L.; Desrivot, J.; Dupont, S.; Fagard, L.; Ford, P.; Fieuw, A.; Wuyts, W. Safety, tolerability, pharmacokinetics, and pharmacodynamics of GLPG1690, a novel autotaxin inhibitor, to treat idiopathic pulmonary fibrosis (FLORA): A phase 2a randomised placebo-controlled trial. Lancet Respir. Med. 2018, 6, 627–635. [CrossRef] 78. Galapagos Announces ISABELA Phase 3 Programin IPF. Available online: http://hugin.info/133350/R/ 2183965/843660.pdf (accessed on 8 October 2019). 79. Zhang, H.; Xu, X.; Gajewiak, J.; Tsukahara, R.; Fujiwara, Y.; Liu, J.; Fells, J.I.; Perygin, D.; Parrill, A.L.; Tigyi, G.; et al. Dual activity lysophosphatidic acid receptor pan-antagonist/autotaxin inhibitor reduces breast cancer cell migration in vitro and causes tumor regression in vivo. Cancer Res. 2009, 69, 5441–5449. [CrossRef] 80. Gupte, R.; Patil, R.; Liu, J.; Wang, Y.; Lee, S.C.; Fujiwara, Y.; Fells, J.; Bolen, A.L.; Emmons-Thompson, K.; Yates, C.R.; et al. Benzyl and naphthalene methylphosphonic acid inhibitors of autotaxin with anti-invasive and anti-metastatic activity. Chem. Med. Chem. 2011, 6, 922–935. [CrossRef]

© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).