lable at ScienceDirect

European Journal of Medicinal Chemistry 46 (2011) 2147e2151

Contents lists avai

European Journal of Medicinal Chemistry

journal homepage: http: / /www.elsevier .com/locate/ejmech

Original article

Lipid-like sulfoxides and amine oxides as inhibitors of mast cell activation

José Batista a,b,1, Georg Schlechtingen b,1, Tim Friedrichson b, Tobias Braxmeier b, Jürgen Bajorath a,*

aDepartment of Life Science Informatics, B-IT, LIMES, Program Unit Medicinal Chemistry and Chemical Biology, Rheinische Friedrich-Wilhelms-Universität Bonn, Dahlmannstr. 2,D-53113 Bonn, Germanyb JADO Technologies GmbH, Tatzberg 47-51, D-01307 Dresden, Germany

a r t i c l e i n f o

Article history:Received 11 November 2010Received in revised form21 February 2011Accepted 26 February 2011Available online 5 March 2011

Keywords:Mast cell activation and degranulationImmunoglobulin E receptorMembrane microdomainsLipid raftsLipid-like compoundsMembrane-directed inhibitorsComputational filteringMolecular similarity

* Corresponding author. Tel.: þ49 228 2699 306; faE-mail address: bajorath@bit.uni-bonn.de (J. Bajor

1 The contributions of these authors should be con

0223-5234/$ e see front matter � 2011 Elsevier Masdoi:10.1016/j.ejmech.2011.02.068

a b s t r a c t

The drug miltefosine is a prototypic lipid-like compound thought to modulate membrane environmentsand thereby indirectly prevent receptor-mediated signaling events. In addition to its primary therapeuticindications in cancer and leishmaniasis, miltefosine has also been shown to block immunoglobulin Ereceptor-dependent mast cell activation. Miltefosine and other alkylphospholipids that are active in mastcell degranulation assays contain a positively charged nitrogen and a phosphate group that are importantfor activity. In addition to alkylphospholipids, ceramides are also known to act on membrane environ-ments and inhibit mast cell activation. We have systematically searched a very large compound collec-tion for other lipid-like inhibitors of mast cell activation. Analogs of an initially identified screening hitwere synthesized and preliminary SAR information was collected, leading to the identification of sulf-oxide and amine oxide containing lipid-like compounds as new inhibitors of mast cell activation. Sulf-oxide and amine oxide derivatives were found to be only slightly less active than miltefosine.

� 2011 Elsevier Masson SAS. All rights reserved.

1. Introduction

Allergic diseases are associated with deregulated mast cell acti-vation [1]. Among other effects, mast cell activation and degranu-lationdependon signal transduction through the immunoglobulinEreceptor [2]. This receptor is known tobe concentrated inmembranemicrodomains having a characteristic lipid composition and a highdegree of lipid ordering [2]. These microdomains are often calledlipid rafts [3]. Mast cell activation can be prevented by blocking theimmunoglobulin E receptor, for example, through the use ofmonoclonal antibodies. Signaling through the immunoglobulin Ereceptor depends on specific receptoreligand interactions and alsoon the structural integrity of its lipid raft environment [2]. Hence, inaddition toblocking the receptoreligand interactions, an alternativebut as of yet only little explored route to the treatment of allergicdiseases might be to selectively alter the structure and organizationof immunoglobulin E receptor containingmembranemicrodomainsand thereby prevent effective immunoglobulin E receptor-depen-dent signaling. Targeting membrane structures instead of receptors

x: þ49 228 2699 341.ath).sidered equal.

son SAS. All rights reserved.

still is an unconventional approach to therapy. For this purpose,compoundswould be required to enter rafts and perturb the regularpacking of their characteristic phospho- and glycolipids, thus indi-rectly affecting receptor function. However, there is some priorevidence for this potential mechanism of action. The drug miltefo-sine (hexadecylphosphocholine), an alkylphospholipid, is currentlyapproved for the treatment of breast cancer metastasis and leish-maniasis [4,5]. Miltefosine (compound 1, Fig. 1) consists of a polarhead group that is similar to plasmalogen phospholipids. Plasmal-ogen phospholipids are known to be enriched in immunoglobulin Ereceptor containing lipid rafts [6]. It is thus likely that miltefosinemight alsoenter thesemembranedomains and therebycompromisetheir structural integrity. If so, then miltefosine would also beexpected to affect immunoglobulin E receptor-dependent signaling,although it had originally not been implicated in such effects.However, this hypothesis has been tested and it has indeed beenshown that miltefosine blocked immunoglobulin E receptor-dependent mast cell activation both in vivo and in vitro [7]. There-fore, miltefosine is thought to perturb the structure of immuno-globulin E receptor containing lipid rafts, although there currently isno detailed mechanistic information available.

Following up on these findings, we have previously reportedother alkylphospholipid structures that inhibit the activation ofRat Basophilic Leukemia clone 2H3 (RBL-2H3) cells, a commonly

Fig. 1. Known lipid-like inhibitors of mast cell activation. Compound 1 is miltefosine.Its head group on the left contains a quaternary nitrogen and a phosphate group.Compound 2 has a head group that is chemically distinct from miltefosine, but alsocontains a charged nitrogen. Compound 3 represents the ceramide (sphingolipid)chemotype.

J. Batista et al. / European Journal of Medicinal Chemistry 46 (2011) 2147e21512148

used in vitro cell culture model for mast cells [8]. RBL-2H3 cellswere also used as a cell culture model in our current study. Milte-fosine and other alkylphospholipids that are active in degranula-tion assays contain a positively charged quaternary amine anda negatively charged phosphate group (compound 1, Fig. 1), i.e. theyare betaines. Through a miltefosine-based pharmacophore search,we have also identified another lipid-like inhibitor of mast cellactivation with a head group that is chemically distinct fromalkylphospholipids [8] (compound 2, Fig. 1). This inhibitor alsocontains a quaternary nitrogen and, in this case, a delocalizedpositive charge, but no negatively charged group (that might mimica phosphate group). There is currently only one other chemotypeknown that is not a betaine, but active in RBL-2H3 cell degranula-tion assays, i.e. ceramide, a sphingolipid [9,10] (compound 3, Fig. 1).

In this study, we have searched for other lipid-like inhibitors ofmast cell activation that are chemically distinct from alkylphos-pholipids and ceramides as potential starting points for thegeneration of membrane-directed anti-allergic agents.

Fig. 2. Test compounds. Shown are four different lipid-like compounds that wereselected as candidate inhibitors. Assay results are reported for each compound.

2. Experimental

2.1. Computational analysis

Approximately 4.2 million in silico-formatted compoundscollected frommedicinal chemistry vendor sources were subjectedto a computational filtering protocol. Compound filtering wascarried out using theMolecular Operation Environment (MOE) [11].To ensure the desired lipid-like character of candidate structuresthe following rules were initially applied: (1) molecular size:between 18 and 50 non-hydrogen atoms, (2) lipid-likeness: pres-ence of a C11 or longer saturated alkyl chain, (3) charge state: nocharged atoms, (4) head group: hydrophobic substituent witha maximum of three atoms permitted. The last criterion ensuredthat the alkyl chain was the only major hydrophobic substructure.Only 1868 compounds passed these filters. These lipid-likecompounds were further filtered for the following undesiredchemical moieties: (a) polyether, (b) polyfluorine, (c) adamantane-like rings (d) nitrile, (e) nitrite, (f) azo-coupling groups, (g) alkylchain with 22 or more carbon atoms. A total of 956 compounds didnot contain any of these moieties and were further analyzed. The956 selected database compounds were encoded as MACCS

fingerprints, as provided by the MOE Software Toolkit, andcompared to compounds representing currently known inhibitorychemotypes including betaines and ceramides. Only compoundswith less than 70% 2D Tanimoto similarity to any known inhibitorwere further considered in order to focus the search on new che-motypes. A total of 179 compounds with 90 unique head groupsmet this dissimilarity criterion. These candidate molecules weresubjected to visual inspection and only four compounds wereultimately selected for experimental evaluation, as rationalized inthe results and discussion section. Fig. 2 shows these four selectedchemotypes. Compound 4 was acquired from Timtec, compound 5and 6 from SigmaeAldrich, and compound 7 from Chemdiv.

2.2. Degranulation assay

Inhibition of degranulationwas determined as percent reductionof b-hexosaminidase release compared to antigen-stimulatedrelease used as a positive control after subtraction of unspecificrelease as a negative control. Relative inhibitionwas calculated as: %Inhibition ¼ 100 � (1 � (test compound � negative control)/(posi-tive control � negative control)). The assay was performed on RBL-2H3 cells. Cells were seeded in 24 well plates in a volume of 1 mlmedium (70% Minimum Essential Medium with Earle’s Salts, 20%RPMI 1640 Medium, 10% FBS (fetal bovine serum) and 2 mM L-glutamine). After incubation the mediumwas replaced with 400 mlfresh medium containing 0.4 mg/ml anti-DNP IgE. The next day cellswere washed with 800 ml pre-warmed Tyrode’s buffer (TyB). Afterwashing 380 ml TyB containing a test compound or no compoundwere added and the cells were incubated for 1 h at 37 �C. Cells wereactivated by addition of 20 ml DNP-HSA (dinitrophenol-humanserumalbumin; 0.1 mg/ml inTyB) for 15min. Plateswere centrifugedat 4 �C and transferred to an ice bath. Duplicate aliquots (25 ml) ofsupernatants were transferred to a black 96-well plate. 100 ml MUGsubstrate solution (2.5 mM 4-methylumbelliferyl-N-acetyl- b-D-glucosaminide (Calbiochem, no. 474502) in 0.05 M citrate) wasadded and plates were incubated for 30 min at 37 �C. The reactionwas terminated by addition of 150 ml stop solution (0.1 M NaHCO3/0.1 M Na2CO3, pH 10). Fluorescence was measured at 365 nm exci-tation and 440 nm emission wavelengths. As a negative control,supernatant of non-stimulated cells without anti-DNP IgE wasmeasured to detect unspecific b-hexosaminidase release. As

J. Batista et al. / European Journal of Medicinal Chemistry 46 (2011) 2147e2151 2149

a control for total b-hexosaminidase content, the remaining cellswere treated with 400 ml lysis buffer (25 mM Tris$HCl, pH 7.5,150 mM NaCl, 5 mM EDTA and 1% (w/v) Triton X-100) at roomtemperature. Antigen-stimulated b-hexosaminidase release servedas positive control. IC50 values were determined from concentra-tion-response data (concentration vs. inhibition of degranulation)with a non-linearfive point parameter curvefitting procedure usingSigmaPlot software [10].

2.3. Toxicity assay

Themaximum tolerated concentration (Mtc) was determined asthe highest dose of a test compound that did not cause toxic cellulartoxicity determined by lactate dehydrogenase release usinga commercially available cytotoxicity test (Promega Cytotox-Onecat. #67891).

2.4. Synthesis

Dodecyl methyl sulfoxide (compound 8, Fig. 3) was purchasedfrom SigmaeAldrich (no. 641588) and N,N-dimethyltetradecyl-amine-N-oxide (compound 14, Fig. 3) from Anatrace, Ohio (no.T360). Sulfoxides were prepared from the corresponding thioethersby oxidation with H2O2 in hexafluoroisopropanol (HFIP) [12],occasionally with addition of CH2Cl2 to improve solubility. Thismethod yielded clean monooxygenation without any furtheroxidation to sulfones. Except for dithiane (compound 16, Fig. 3),thioethers were prepared by alkylation of commercial thiolsfollowing known procedures (Fig. 3). Experimental details andanalytical data are provided as Supplementary information.

Disulfoxides were obtained as mixtures of stereoisomers. In thecase of compound 13 (Fig. 3), oxidation of racemic trans-thioether(compound 19, Fig. 3) yielded two equipotent stereoisomers (S]Oepimers), which were separated by flash chromatography. Theavailable spectroscopic data did not allow for unequivocal assign-ment of configuration at the S-atom. The bioassay results presentedfor compound 13 (Fig. 3) refer to the faster-eluting, more solublestereoisomer, which was the main product of this synthesis.

Fig. 3. Synthesis of sulfoxides 9e13. i: dodecyl iodide, NaOH/EtOH (75%), ii: EtBr,NaOH/EtOH (88%), iii: H2O2/HFIP, iv: BuLi; dodecyl iodide, �78 �C; 0 �C (47%), v: MeI,K2CO3 (98%), vi: dodecane thiol, NEt3. Yields of oxidation vii: 9 (33%), 10 (54%), 11 (55%),12 (86%), 13 (64%).

3. Results

In order to search for lipid-like inhibitors that depart from thealkylphospholipid and ceramide structures, we have carried outa rule-based computational filtering procedure of a very largecompound collection that was augmented by molecular similarityanalysis. Through filtering and exclusion of compounds withobvious 2D structural similarity to alkylphospholipids and ceram-ides, we ultimately reduced w4.2 million database compounds to179 lipid-like structures with 90 unique head groups (the lipid-likeness constraint eliminated the majority of databasecompounds). We then subjected this relatively small set of candi-date head groups to visual inspection in order to eliminatecompounds that had no H-bond acceptor functions, which arelikely to be important for activity in light of currently availableactive chemotypes, or that had only limited accessibility to H-bondacceptors (e.g. because of proximal ring systems). We also onlyconsidered compounds that were commercially available. Theseadditional criteria further reduced the number of candidates, andwe finally selected only four compounds for experimental evalua-tion in RBL-2H3 cell degranulation assays (shown in Fig. 2).

While compound 4was found to be inactive, compounds 5 and 6displayed weak inhibition of degranulation at 25 mM concentration.By contrast, compound 7 showed more than 70% inhibition underthese conditions and only very little toxicity (Mtc 100 mM). Hence,this compound was considered an attractive hit. Its head groupcontained a cyclohexyl acetate and a sulfoxide group, representinga previously unobserved inhibitory chemotype lacking chargedatoms. Compound 7 also had the shortest (C12) saturated alkyl tailamong the four test compounds.

In order to better understand key features of this head group,several analogs were synthesized and tested, as shown in Fig. 4.Because the oxygen-rich test compounds 5 and 6 in Fig. 2 were onlyvery weakly active, we first investigated the role of the sulfoxidegroup by generating compound 8 where the head group wasreduced to only a sulfoxide (Fig. 4). Interestingly, this compoundalready showed 66% inhibition at 25 mM. Thus, it was only slightlyless active than compound 7, which confirmed the importance ofthe sulfoxide group. For sulfoxides, we varied the length of thesaturated alkyl “lipid tail” from C10 to C16 and found that C12analogs were most active. Additional C12 analogs were generatedin order to investigate the presentation of sulfoxide group(s) indifferent chemical environments. For example, adding a secondsulfoxide group in a six-atom alkyl chain head group reducedinhibition (compound 9, Fig. 4). However, presenting a secondsulfoxide group in an aliphatic six-membered ring (compound 10,Fig. 4) further increased the inhibitory activity observed forcompound 7 to 80% at 25 mM. By contrast, presenting single or dualsulfoxide groups in head groups containing an aromatic ring againreduced inhibition (compounds 11 and 12, Fig. 4). The comparisonof compounds 12 and 13 in Fig. 4 demonstrates that the presence ofan aromatic instead on an aliphatic ring was unfavorable. Takentogether, the analogs of our initial hit shown in Fig. 4 revealed somepreliminary SAR information. With 80% inhibition at 25 mM, cor-responding to an IC50 value of 10.7 mM, compound 10was the mostactive sulfoxide derivative we identified. Similar to compound 7,compound 10 also displayed only very little cellular toxicity in ourassays.

Considering the previously observed relevance of nitrogencontaining betaine structures for inhibition, we also reasoned thatit should be interesting to replace the active sulfoxide head groupby an amine oxide. These head groups differ only in a single sulfur/nitrogen atom and have comparable polarity. We initially testeda C12 amine oxide, which displayed weak but detectable activity inRBL-2H3 cell degranulation assays, with 40% inhibition at 25 mM. In

Fig. 4. Analogs of compound 7. Assay results are reported. “Mtc” gives the maximumtolerated compound concentration in toxicity assays. Compound 10 was the mostactive sulfoxide derivative.

J. Batista et al. / European Journal of Medicinal Chemistry 46 (2011) 2147e21512150

this case, we also varied the length of the saturated alkyl chain fromC10 to C18 and found that the C14 amine oxide analog was highlyactive (compound 14, Fig. 5), with 95% inhibition at 25 mM and anIC50 value of 10.6 mM. Both compounds 10 and 14were only slightlyless active in RBL-2H3 cell degranulation assays than compound 2(IC50 ¼ 5.5 mM) and miltefosine (IC50 ¼ 4.6 mM) that were tested ascontrols. Thus, these findings suggest that lipid-like compoundswith sulfoxide and amine oxide containing head groups shouldmerit further exploration as membrane-directed inhibitors of mastcell activation.

4. Discussion

Evidence is accumulating that well-structured membranemicrodomains of specific composition, termed lipid rafts, providean environment for certain receptors that contributes to the regu-lation of their signaling functions [3]. However, the approach oftargeting membrane microdomains that are enriched with a ther-apeutically relevant receptor using lipid-like synthetic compoundsis still in its infancy. Although miltefosine is a marketed drug, themolecular mechanisms by which so-called raftophilic compoundsmight act are just beginning to be understood [3]. We have set outto target lipid rafts containing the immunoglobulin E receptor inorder to control signaling events leading tomast cell activation thatare implicated in allergic diseases. Compound classes that are

Fig. 5. Lipid-like amine oxide. Shown is the structure of an amine oxide that wasdesigned on the basis of compound 8. This amine oxide strongly inhibited mast celldegranulation and also showed only very little toxicity.

currently known to be active in RBL-2H3 cell degranulation assays,used as a model system for mast cell activation, include alkyl-phospholipids and ceramides. The characteristic feature of activealkylphospholipids is the presence of a charged nitrogen anda phosphate group. From the comparison of different alkylphos-pholipids, it was possible to derive a simple pharmacophore [8]. Inorder to increase the probability of identifying other active che-motypes, we deliberately did not require the presence of a nega-tively charged acceptor function when defining thepharmacophore. This simple query was then utilized to discovera new active chemotype containing a charged nitrogen (compound2, Fig. 1) [8]. In order to further expand the search for other che-motypes as potential inhibitors of mast cell activation, we imple-mented a computational filtering protocol in this study that wasaugmented by molecular similarity calculations to eliminatecandidate compounds that were structurally similar to knowninhibitory chemotypes. Furthermore, we analyzed candidatecompounds based on the hypothesis that sterically well-accessiblehydrogen bond acceptors should generally be important for activityin lipid raft environments. A compound with a sulfoxide containinghead group was found to display considerable activity in RBL-2H3cell degranulation assays. Given the activity of betaines and thesulfoxide head group, both of which are strong dipoles, wereasoned that amine oxidesmight also be interesting candidates forevaluation as mast cell degranulation inhibitors. We then indeedfound that amine oxides were active in RBL-2H3 cell degranulationassays. However, in this case, most active analogs were obtained forC14 saturated alkyl chains, whereas sulfoxides were most activewith C12 alkyl chains. Analyzing the origins of different alkyl chainlength preferences for chemically simple sulfoxide and amine oxidehead groups should provide interesting opportunities for furtherstudies.

5. Conclusions

In this study, we have identified sulfoxide and amine oxidecontaining lipid-like molecules as new membrane-directed inhib-itors of mast cell activation. By preparing analogs of an initiallyidentified hit (compound 7) we have established that a sulfoxidehead group was sufficient for inhibition. However, sulfoxide groupspresented in an aliphatic chain showed reduced inhibition and theaddition of aromatic rings to sulfoxide containing head groups wasalso unfavorable. By contrast, two sulfoxide groups present in analiphatic ring yielded the currently most active analog (compound10). Similar to compound 7, this analog caused only very littlecellular toxicity and was only slightly less active than miltefosine indegranulation assays. Furthermore, we found that compound 14containing an amine oxide head group was also comparably active.

Appendix A. Supplementary material

Supplementary data associated with this article can be found, inthe online version, at doi:10.1016/j.ejmech.2011.02.068.

References

[1] E. Monroe, Cutis 76 (2005) 118e126.[2] D. Holowka, B. Baird, Semin. Immunol. 13 (2001) 99e105.[3] D. Lingwood, K. Simons, Science 327 (2010) 46e50.[4] H. Eibl, C. Unger, Cancer Treat. Rev. 17 (1990) 233e242.[5] L. Kedzierski, A. Sakthianandeswaren, J.M. Curtis, P.C. Andrews, P.C. Junk,

K. Kedzierska, Curr. Med. Chem. 16 (2009) 599e614.[6] Z. Surviladze, K.A. Harrison, R.C. Murphy, B.S. Wilson, J. Lipid Res. 48 (2007)

1325e1335.[7] K. Weller, M. Artuc, G. Jennings, T. Friedrichson, S. Guhl, R.V. dos Santos,

C. Sünder, T. Zuberbier, M. Maurer, J. Invest. Dermatol. 129 (2009) 496e498.

J. Batista et al. / European Journal of Medicinal Chemistry 46 (2011) 2147e2151 2151

[8] J. Batista, T. Friedrichson, G. Schlechtingen, T. Braxmeier, G. Jennings,J. Bajorath, Eur. J. Med. Chem. 45 (2010) 2700e2704.

[9] Braxmeier T., Friedrichson T., Fröhner W., Jennings G., Schlechtingen G.,Schroeder C., Knölker H.J., Simons K., Zerial M., Kurzchalia T., 2004. patent WO2006/002909A2.

[10] A. Gidwani, H.A. Brown, D. Holowka, B. Baird, J. Cell. Sci. 116 (2003)3177e3187.

[11] Molecular Operating Environment (MOE), Version 2007.09. ChemicalComputing Group Inc., Montreal, Canada, 2007.

[12] K.S. Ravikumar, J.-P. Begue, D. Bonnet-Delpon, Tetrahedron Lett. 39 (1998) 3141e3144.