A Robust Small-Molecule Microarray Platform for Screening Cell Lysates

A Robust Small-Molecule Microarray Platform for Screening Cell Lysates
of 12
All materials on our website are shared by users. If you have any questions about copyright issues, please report us to resolve them. We are always happy to assist you.
Related Documents
  Chemistry & Biology  13 , 493–504, May 2006  ª 2006 Elsevier Ltd All rights reserved DOI 10.1016/j.chembiol.2006.03.004 A Robust Small-Molecule Microarray Platformfor Screening Cell Lysates James E. Bradner, 1,2 Olivia M. McPherson, 1 Ralph Mazitschek, 1 David Barnes-Seeman, 4,7 John P. Shen, 1 Jasmeet Dhaliwal, 1 Kristen E. Stevenson, 3 Jay L. Duffner, 1 Seung Bum Park, 4,8 Donna S. Neuberg, 3 Paul Nghiem, 4,9 Stuart L. Schreiber, 1,4,5, *and Angela N. Koehler 1,6, * 1 Broad Institute of Harvard and MIT7 Cambridge Center Cambridge, Massachusetts 02142 2 Division of Hematologic Neoplasia 3 Department of Biostatistical ScienceDana Farber Cancer InstituteHarvard Medical School44 Binney StreetBoston, Massachusetts 02115 4 Howard Hughes Medical InstituteDepartment of Chemistry and Chemical BiologyHarvard University SummaryHerein we report the expanded functional group com-patibility of small-molecule microarrays to include im-mobilization of primary alcohols, secondary alcohols,phenols, carboxylic acids, hydroxamic acids, thiols,and amines on a single slide surface. Small-molecule‘‘diversity microarrays’’ containing nearly 10,000known bioactive small molecules, natural products,and small molecules srcinating from several diver-sity-oriented syntheses were produced by using anisocyanate-mediated covalent capture strategy. Se-lected printed bioactive compounds were detectedwith antibodies against compounds of interest. Thenewsurfaceofthediversitymicroarraysishighlycom-patible with approaches involving cellular lysates.This feature has enabled a robust, optimized screen-ing methodology using cellular lysates, allowing thedetection of specific interactions with a broad rangeofbindingaffinitybyusingepitope-taggedorchimericfluorescent proteins without prior purification. We be-lieve that this expanded research capability has con-siderable promise in biology and medicine.Introduction Natural products and products of diversity-orientedsynthesis (DOS) constitute a rich pool of small mole-cules from which specific ligands to proteins of interestmay be found [1]. Small-molecule microarrays [2–11] (SMMs) enable the discovery of previously unknownprotein-ligand interactions, resulting in small-moleculemodulators of protein function [12, 13]. To makeSMMs,stocksolutionsofcompoundsareroboticallyar-rayed onto functionalized glass microscope slides thatare incubated with proteins of interest. Microarray fea-tures representing putative interactions between pro-teins and small molecules are typically visualized withfluorescently labeled antibodies and a standard fluores-cence slide scanner.Several mild and selective coupling reactions havebeen used to capture covalently synthetic compoundsonto glass surfaces and include a Michael addition[10], addition of a primary alcohol to a silyl chloride [4], diazobenzylidene-mediated capture of phenols [2],1,3-dipolar cycloaddition [3], a Diels-Alder reaction [5], a Staudinger ligation of azides onto phosphane-modi-fied slides [7], and capture of hydrazide-linked com-pounds onto epoxide-functionalized glass and viceversa [8, 9]. Most of these surface capture methodstake advantage of a reactive functional group that is in-troduced as part of a solid-phase organic synthesis andbiases the orientation of the small molecule on the sur-face [7, 14]. Nonselective photoinduced crosslinkinghas also been used to immobilize a set of ten complexnatural products onto glass slides [6]. Noncovalent ap-proaches have also been employed, such as the hybrid-izationofpeptide-nucleicacidconjugatestooligonucle-otide arrays [15, 16].Usingselectiveapproaches,ourlaboratorieshaveim-mobilized over 50,000 products of diversity-orientedsynthesis pathways via capture through a primary or secondary alcohol onchlorinated slides or through cap-ture of phenols on diazobenzylidene-functionalizedslides [2, 4, 12]. Unfortunately, the previous approachesforced us to make separate microarrays for compoundsthat contained either a primary or secondary alcoholand compounds containing aryl alcohols. We hoped todevelop arrays that would capture all three types of alcohols on a common slide surface. Additionally, wehoped to include compounds from natural sources, notnecessarily bearing primary alcohols, secondary alco-hols, or phenols, alongside synthetic compounds in themicroarrays. We turned to nonselective photoinducedcrosslinking as a capture method and experiencedmixed results. Although we successfully printed anddetected several of the known ligands described byKanoh et al. [6], our attempts to print and screen micro-arrays of 6336 phenol-containing fused bicycles andtetracycles [2, 17] provided unacceptable numbers of false positives as judged by secondary binding assaysusing surface plasmon resonance. This experienceled us to pursue new capture strategies that wouldallow immobilization of several common functional *Correspondence: (S.L.S.); koehler@ (A.N.K.) 5 Labaddress:  6 Lab address: koehler/index.html 7 Present address: Novartis Institutes for Biomedical Sciences, 100MIT Technology Square, Cambridge, Massachusetts 02139. 8 Present address: School of Chemistry, Seoul National University,Seoul, 151-747, Korea. 9 Present address: Division of Dermatology, University of Washing-ton, 815 Mercer Street, Seattle, Washington 98109.  groups that are present in both synthetic and naturalcompounds.We have previously reported the use of SMMs to dis-cover ligands for calmodulin (calmoduphilins) [2], theyeast transcriptional corepressor Ure2p (uretupamines)[13], and the Hap3p subunit of the yeast HAP transcrip-tion factor complex (haptamides) [12]. Each of thesescreens involved SMMs in which only one DOS librarywas contained on a given slide. More recently, wesought to prepare an SMM that contains sublibrariesfrom various DOS synthetic routes in one array. Thegoal of preparing such an SMM is to allow researchersto sample the various sublibraries in one array andthen prioritize screens of the full DOS libraries basedon the initial screening results from the diverse subset.Here we report the use of isocyanate-functionalizedglass slides to make a small-molecule ‘‘diversity micro-array’’ containing several collections of DOS com-pounds coming from various solid-phase organic syn-thesis routes [18–24] and commercially availablebioactive compounds, including natural products, onthe same slide ( Figure 1 ). Isocyanates react with a num-ber of nucleophilic functional groups without leaving anacidicbyproduct[25]andanisocyanatesurfacetherebyincreases the diversity of small molecules, from naturalorsyntheticsources,thatcanbeimmobilizedontoasin-gle SMM. Isocyanate glass substrates have also beenprepared and used to immobilize oligonucleotides ina microarray format [26–29].Prior strategies aimed at ligand discovery usingSMMs have relied on incubation with a purified proteinof interest. Potential applications of these protocolshave been limited by challenges in protein biochemistryinvolving expression of large proteins, solubility, post-translational modification state, activity, and yield. Fur-thermore, without commercial availability of a proteintarget of interest, investigators without expertise in pro-tein biochemistry may be limited in their capacity toscreenSMMs.Herewedescribetheoptimizationofaro-bust, efficient SMM screening methodology which al-lowsthe detection ofspecific protein-small moleculein-teractions by using epitope-tagged target proteinsdirectly from cell lysates without purification. We dem-onstrate that the new attachment chemistry is compati-ble with detection of known interactions between vari-ous small molecules and FKBP12 [30, 31] obtaineddirectly from cellular lysates. Previous research report-ing the detection of specific interactions with complexlysates has typically involved the addition of known,purified proteins [32] or has required incubation in solu-tion with focused libraries of covalent probes conju-gated to nucleic acids prior to spatial arraying on an Figure 1. Schematic Design of the Diversity SMM Containing Bioactive Small Molecules and Products of Diversity-Oriented SynthesisReactive functional groups are colored. Representative bioactive small molecules printed in the diversity array include  1a , nigericin;  1b , ba-filomycin A1;  1c , doxorubicin;  1d , genistein;  1e , lactacystin;  1f , uvaol;  1g , D- erythro -sphingosine;  1h , gibberellic acid;  1i , ingenol;  1j , aloin. Rep-resentative scaffolds for DOS small molecules printed in the diversity array include  2a , dihydropyrancarboxamides [23];  2b , alkylidene-pyran-3-ones [18, 19];  2c , fused pyrrolidines [20];  2d , serine-derived peptidomimetics;  2e , shikimic acid-derived compounds;  2f , 1,3-dioxanes [24]; 2g , spirooxindoles [22];  2h , macrocyclic lactones;  2i , ansa-seco steroid-derived compounds [21].Chemistry & Biology494  oligonucleotide array [15, 16]. The ability to detect se-lective interactions in cellular lysates without proteinpurification is appealing for ligand discovery, targetidentification, antibody and protein specificity profiling,as well as for future applications such as signaturediscovery for cellular states and diagnostic tooldevelopment. Results Smallmoleculescontaining nucleophiles witharange of reactivitieswerearrayedontoaweaklyelectrophilicsur-facethatreactsslowlywitheitherthesmallmoleculesor ambient moisture and yields no potentially deleteriousbyproducts such as an acid. As shown in Figure 2,  g -aminopropylsilane slides (  S1  ) were coated with a shortpolyethyleneglycol(PEG)spacerandcoupledto1,6-dii-socyanatohexane via a urea bond to generate putativeisocyanate-functionalized glass slides (  S2  ). Slidesprinted with compound stock solutions were thenplaced in a dry environment and exposed to a pyridinevapor that catalyzes the covalent capture ofsmall mole-cules onto the slide surface (  S3  ).Toevaluatethisapproach,aroboticmicroarrayerwasused to print a series of synthetic FKBP12 ligands [33]that were derivatized so as to present a primary alcohol(  3a ,  3o ,  3p ,  3q  ), secondary alcohol (  3b  ), tertiary alcohol(  3c  ), phenol (  3d  ), methyl ether (  3e  ), carboxylic acid (  3f  ),hydroxamic acid (  3g  ), methyl (  3h  ), thiol (  3i  ), primaryamine (  3j ,  3n  ), secondary amine (  3k   ), indole (  3l  ), or arylamine (  3m  ) onto the isocyanate-derivatized slides ( Fig-ures 3 A and 3B). The site of modification for eachFKBP12ligandhaspreviouslybeenshowntobetolerantto substitution as  3  is a parent structure for chemical in-ducers of dimerization [34]. The ligands were printed inserial 2-fold dilutions (10 mM to 20  m M) with DMF asa solvent. The printed slides were exposed to pyridinevapor, quenched with ethylene glycol, and washed ex-tensively with DMF, THF, and methanol. Dried slideswere probed with FKBP12-GST [30, 31], followed bya Cy5-labeled anti-GST antibody, and scanned for fluo-rescence at 635 nm with GenePix Pro 6.0 software (Mo-lecular Devices, Union City, CA). As shown in Figure 3,the intensity of fluorescent signals corresponding toFKBP12-GST varied according to both the functionalgroup presented for attachment and concentration of li-gand. Feature diameter was dependent on the concen-tration of ligand, and at higher concentrations the aver-age diameter was 250  m m. The primary amines, arylamine, and thiol appear to have the highest immobiliza-tionlevels.Fluorescenceintensitiesfortheprimaryalco-hols,phenol,hydroxamicacid,secondaryamine,andin-dole are also consistent with significant immobilization.The secondary alcohol, carboxylic acid, and tertiary al-cohol were immobilized in lower amounts. At 1.25 mM,a typical concentration for our compound stock solu-tions, trace levels of primary amides  3e  and  3h  were de-tectedwhereasthe N,N  -substitutedamide 3r ( Figure3D)was not. The addition of polyethylene glycol spacers of varying lengths to the ligand (  3n – 3q  ) did not make a sig-nificant impact on the feature morphology or fluores-cenceintensitywhenprobedwithpurifiedprotein.Addi-tionally, polyethylene glycol spacers of varying lengths(n = 0, 2, 4, 8, 70) were added to surface  S2  and com-pared (data not shown). Surfaces with shorter PEGchains (n = 2, 4, 8) were equivalent and displayed im-proved signal-to-noise ratios over the surface withoutPEG. The surface with longer PEG chains displayedthe lowest fluorescence levels in the binding assay andgave inconsistent spot morphologies.Fluorescence levels were significantly reduced whenpyridine vapor was omitted from the procedure ( Fig-ure 3D). Immobilization levels were slightly enhancedwhen the slides were exposed to pyridine at 37ºC(data not shown). To test the sensitivity of this capturemethod to moisture present in the compound stock so-lutions used for printing, 1 mM solutions of FKBP12 li-gands 3a , 3b , 3c ,and 3e in9:1DMF:ddH 2 Owerearrayedin triplicate onto isocyanate-derivatized slides ( Fig-ure 3E). Fluorescence intensities were equivalent tothose of compounds printed directly from DMF. Toler-ance to water is an important consideration for SMMpreparation because compound stock solutions in DMFand DMSO appear to take on water over time as they Figure 2. Vapor-Catalyzed Surface Immobilization Scheme g -aminopropylsilane (GAPS) slides (  S1  ) are coated with a short Fmoc-protected polyethylene glycol spacer. After removal of the Fmoc groupwith piperidine, 1,6-diisocyanatohexane is coupled to the surface via urea bond formation to generate putative isocyanate-functionalized glassslides (  S2  ). Slides printed with compound stock solutions are then placed in a dry environment and exposed to a pyridine vapor that catalyzesthe covalent capture of small molecules onto the slide surface (  S3  ).Small-Molecule Microarrays and Cellular Lysates495  Figure 3. Comparison of Functional Group Reactivity with Isocyanate-Functionalized Glass(A) Parent structure of AP1497 derivatives  3a – 3q .(B) AP1497 derivative array with FKBP12 ligands  3a – 3q  printed in serial 2-fold dilutions (10 mM to 20  m M) onto isocyanate-derivatized slides.The slides were exposed to pyridine vapor to catalyze the attachment of printed compounds. Washed slides were probed with FKBP12-GSTfollowed by a Cy5-labeled anti-GST antibody. An image for a microarray scanned for fluorescence at 635 nm is shown. The functional groupspresented for surface capture are shown at the top of the array.(C) Total fluorescence intensity was computed within 300  m m spots centered on each microarray feature with GenePix Pro 6.0 microarray anal-ysis software. The capture of small molecules is catalyzed in the presence of pyridine vapor and is tolerant of moisture in compound stocksolutions.(D) Solutions of FKBP12 ligands  3a ,  3d ,  3e ,  3r , and  3s  (1 mM) in DMF were arrayed in triplicate onto surface  S2  and the slides were incubatedeither under an atmosphere of N 2  (bottom) or in the presence of pyridine vapor under an atmosphere of N 2  (top).(E) Solutions of FKBP12 ligands  3a ,  3d ,  3h , and  3s  (1 mM) in DMF (top row) or 9:1 DMF:ddH 2 O (bottom row) were arrayed in triplicate ontoisocyanate-derivatized slides.Chemistry & Biology496  move in and out of freezer storage [35]. Small moleculesprinted from DMSO were also captured by using thismethod with smaller feature diameters (  w 100–150  m m)than compounds printed from DMF (  w 250–300  m m)(data not shown).Toinvestigate thesuitability of ourapproachforprint-ing compounds that have not been intentionally synthe-sized with appendages for covalent capture, more than300 commercially available bioactive compounds wereprinted onto isocyanate-functionalized slides. Wescreened these bioactive microarrays with rabbit pri-mary antibodies against corticosterone, digitoxin, and17 b -estradiol, followed by a fluor-labeled goat anti-rab-bit secondary antibody. The signal-to-noise ratio (SNR)was determined by calculating intensity at 635 nm andadjusting for local background for each feature on repli-cate arrays, and data were compared to replicate con-trol arrays incubated with the labeled secondary anti-body alone ( Figure 4 ). Six bioactives, with SNR ratios>3.0,werefoundinreplicatearraystobindtothelabeledpolyclonal secondary antibody alone. None of the com-poundswereautofluorescentat635nmasjudgedbyar-raysprobedwithPBSbufferalone(datanotshown).Hy-gromycin B, an aminoglycoside antibiotic, gave thehighest adjusted SNR (mean 47.6). Three quinolone an-tibiotics, norfloxacin, ciprofloxacin, and pipemidic acid,displayed mean adjusted fluorescent intensities greater than 3.0 in at least one experiment. In the anti-cortico-sterone antibody binding profile, hydrocortisone (meanSNR 68.9), beclomethasone (63.3), and corticosterone(59.2), corticosteroids related in structure, scored aspositives. Gitoxigenin (mean SNR 62.5), convallatoxin(52.7), lanatoside C (24.0), digoxin (17.8), and digitoxin(15.1), all cardioactive steroid glycosides, likewisescored as positives in replicate anti-digitoxin antibodyexperiments. 17 b -estradiol (mean SNR 9.0), estriol(8.7), and estrone (7.3), primary estrogenic hormonesvarying in the number of reactive groups for capture,scored as positives in the anti-17 b -estradiol bindingprofile. The antibody binding profiles demonstrate thatsmall molecules with multiple nucleophilic functionalgroups can be printed and detected by using isocya-nate-mediatedcapture.Additionally,thesedatademon-strateafacileapproachforprofilingthespecificityofim-munoglobulins for small molecules.We aimed to expand the scope of this method to in-clude the detection of interactions between small mole-cules and target proteins expressed in mammalian cellswithout prior purification. Toward this end, a screeningprotocol was developed whereby SMMs incubatedwith cellular lysates bearing overexpressed epitope-tagged proteins of interest are compared with controlSMMs incubated with mock-transfected cellular lysates( Figure 5 A). First attempts at this approach were unsuc-cessful due to an unfavorable interaction between theslide surface and cellular lysates prepared from a phos-phate-buffered RIPA lysis buffer, yielding a uniform,high fluorescent background. By varying buffer condi-tions, we identified optimal signal-to-noise ratios by us-ing an MIPP lysis buffer. These initial experiments high-light the importance of nonfluorescent detergents andbuffer ionic strength, such that a balance between effi-cient cellular lysis and nonspecific surface interactionsis achieved. Following lysis and clarification by centrifu-gation, cellular lysates were incubated on SMMs. Sub-sequently, the arrays were serially incubated with a pri-mary anti-epitope antibody and a Cy5-conjugatedsecondary antibody. A brief wash with PBST and mildagitation followed each incubation. Fluorescence inten-sity was detected and SNR was calculated, compared,and averaged for corresponding features on replicatearrays.We explored this approach by screening the array of  AP1497 derivatives (as in Figure 3B) against HEK-293Tlysates prepared from mammalian cells transientlytransfected with a construct expressing FLAG-FKBP12. Optimization experiments were undertakenwith a stepwise introduction of variation to identify pa-rameters maximizing protocol robustness. Arrays werederivedfromthesameprintingseriesandwerescannedforfluorescencebyusingidenticallaserpowerandpho-tomultiplier tube gain. Experimental variables werecompared by using mean SNR for ligands arrayed ata uniform, standard concentration of 1.25 mM, as de-picted in Figure 5B. To determine whether the total pro-tein concentration affects ligand detection, SMMs wereincubated with lysates varying in concentration from 0.1to 1.0  m g/  m l. Maximum fluorescence intensity and SNRforeachfeatureprovedoptimalat0.3 m g/  m l.Blockingin-cubations are commonly employed in protocols involv-ing SMMs. Given the complex milieu of cellular lysates,we were interested in exploring whether blocking prior to sample incubation is required. Blocking with BSA was found to diminish both the maximum signal inten-sity and SNR when incubating SMMs with cellular ly-sates. Interactions between printed ligands and macro-molecules may be enhanced with the introduction of apolymericpolyethyleneglycol(PEG)spacer,whichad-ditionally may minimize nonspecific protein adsorption.To investigate the effect of spacer length on fluorescentdetection and SNR, PEG spacer length was varied inprinted AP1497 derivative SMMs. A marked decreasein the SNR was observed for each printed feature witha long (n w 70) PEG spacer compared to a substantiallyshorter spacer (n = 2). Additional optimization experi-ments and the detailed, optimized screening protocolfor SMMs with cellular lysates are presented in Supple-mental Data available with this article online.Recognizing the high affinity of AP1497 for FKBP12(K  D =8.8nM),wewereinterestedinassessingtheabilityof this technique to identify lower affinity interactions asmay be detected in screening experiments. Focused ar-rays of two ligands with disparate affinity for FKBP12( Figure6 A)wereprintedwithcontrolbioactives.Theop-timized screening protocol allowed the specific identifi-cation of ligands with K  D  as a high as 2.6  m M ( Figure 6B)[36]. To determine whether this method would allow thedetectionoflow-affinityinteractionsbetweensmallmol-ecules and chimeric fluorescent proteins, SMMs wereincubatedwithlysatesfrommammaliancellstransientlytransfectedwithavectorencodinganEGFP-FKBP12fu-sion protein. Incubated slides were washed briefly withPBST and scanned for fluorescence at 488 nm. Identifi-cation of ligands with low binding affinity was observedwithout the requirement of primary and fluorescently la-beledsecondaryantibodies( Figure6C).Transienttrans-fection of cells in tissue culture with protein expressionconstructs typically results in protein overexpression, Small-Molecule Microarrays and Cellular Lysates497
Similar documents
View more...
Related Search
We Need Your Support
Thank you for visiting our website and your interest in our free products and services. We are nonprofit website to share and download documents. To the running of this website, we need your help to support us.

Thanks to everyone for your continued support.

No, Thanks