A Fluorescence-Detection Size-Exclusion Chromatography-Based Thermostability Assay for Membrane Protein Precrystallization Screening

A Fluorescence-Detection Size-Exclusion Chromatography-Based Thermostability Assay for Membrane Protein Precrystallization Screening
of 9
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
  Structure  14 , 673–681, April 2006 ª 2006 Elsevier Ltd All rights reserved DOI 10.1016/j.str.2006.01.013 Fluorescence-Detection Size-ExclusionChromatography for PrecrystallizationScreening of Integral Membrane Proteins Toshimitsu Kawate 1,3 and Eric Gouaux 1,2,3, * 1 Department ofBiochemistry andMolecular BiophysicsColumbia University650 West 168 th StreetNew York, New York 10032 2 Howard Hughes Medical Institute SummaryFormation of well-ordered crystals of membrane pro-teins is a bottleneck for structure determination byX-raycrystallography.Nevertheless,onecanincreasetheprobabilityofsuccessfulcrystallizationbyprecrys-tallizationscreening,aprocessbywhichoneanalyzesthe monodispersity and stability of the protein-deter-gent complex. Traditionally, this has required micro-gram to milligram quantities of purified protein anda concomitant investment of time and resources.Here, we describe a rapid and efficient precrystalliza-tion screening strategy in which the target protein iscovalently fused to green fluorescent protein (GFP)and the resulting unpurified protein is analyzed byfluorescence-detection size-exclusion chromatogra-phy (FSEC). This strategy requires only nanogramquantities of unpurified protein and allows one toevaluate localization and expression level, the degreeof monodispersity, and the approximate molecularmass. We show the application of this precrystalliza-tion screening to four membrane proteins derivedfrom prokaryotic or eukaryotic organisms.Introduction  X-ray crystallography is currently the most powerfultechnique for determining atomic resolution structuresof biological macromolecules ( Hendrickson, 2000 ). Theresulting atomic structures, in turn, not only provide in-sight into mechanism, but they may also acceleratethe discovery of therapeutic agents ( Blundell et al.,2002; Kuhn et al., 2002 ). In particular, the structuresof membrane proteins, including receptors, channels,and transporters, are especially important becausethese molecules are the targets of most drugs currentlyprescribed ( Zambrowicz and Sands, 2003 ). Unfortu-nately,atomicresolutionstructuralinformationonmem-brane proteins is limited, as they are more difficult toexpress and crystallize compared to water-soluble pro-teins( Tate,2001;Walianetal.,2004 ).Inordertoincreasethe likelihood of obtaining crystals of a membrane pro-tein, it is advantageous to not only optimize the expres-sion conditions, but to also characterize the homogene-ity of the protein prior to crystallization.Inatypicalscenario,oneembarksonanefforttocrys-tallizeanewmembraneproteinbyfirstexaminingdiffer-ent expression systems and purification strategies, withthe aim of obtaining milligram quantities of purified pro-tein. With purified protein in hand, one may then assessthe homogeneity of the sample, both in terms of chem-ical composition and oligomerization state, by a varietyof techniques that might include SDS-PAGE, massspectrometry, light scattering, and size-exclusion chro-matography (SEC). If the protein proves to be homoge-nous, then one proceeds to crystallization. However, if the protein is not homogenous or fails to crystallize,then one may vary a number of parameters that mightinclude the detergent and/or lipids used in purificationand crystallization ( Garavito and Ferguson-Miller, 2001;le Maire et al., 2000; Long et al., 2005; Seddon et al.,2004 ) and the nature of the protein construct ( Cohenet al., 1995 ), perhaps also exploring the properties of proteins from other species ( Kendrew, 1948, 1950; Ken-drew and Parrish, 1956; Kendrew and Pauling, 1956 ).Here, we refer to the aforementioned experiments as‘‘precrystallization screening’’ ( Figure 1 ).In traditional precrystallization screening, one typi-cally monitors the presence of the protein by absor-bance at 280 nm and by staining on SDS-PAGE gels( Figure 1 A).To dothis, however, microgram tomilligramquantities of the protein are required for reliable detec-tion. Because almost all proteins, as well as nucleicacids, absorb at 280 nm, the target protein must alsobe free of major contaminants. Thus, a substantial in-vestment in time and resources must be made in order to bring a target molecule from the cloning stage to pre-crystallization screening. Moreover, because many tar-get molecules fail precrystallization screening, due topolydispersity or instability, the time and resources in-vested in the moderate- to large-scale expression andpurification of these proteins are wasted.To obviate the requirement for moderate- to large-scale expression and purification, we have developedapproachesthatallowonetocarryoutprecrystallizationscreening on nanogram quantities of unpurified proteinobtained from whole-cell lysates or crude membranepreparations of prokaryotic or eukaryotic cells ( Fig-ure 1B). Here, the target proteins are covalently fusedto green fluorescent protein (GFP) ( Chalfie et al., 1994;Shimomuraetal.,1962;Zhangetal.,2002 ).Theresultantfusion proteins are monitored first for expression leveland pattern in whole cells by epifluorescence micros-copy (eukaryotic cells) or batch fluorescence measure-ments (prokaryotic cells). After solubilization of wholecells or crude membranes, SEC profiles are monitoredby fluorescence spectroscopy. SEC is one of the mostuseful tools for monitoring the monodispersity and sta-bility of the target protein; a monodisperse and foldedprotein will generally yield a single symmetrical Gauss-ian peak, while a polydisperse, unstable, or unfoldedprotein will typically yield multiple asymmetric peaks( Barth et al., 1994; Ricker and Sandoval, 1996 ).Inthisarticle,wepresentthemethodologyofprecrys-tallization screening by epi- and batch fluorescence, to-gether with FSEC, followed by application of thesemethods to a eukaryotic, oligomeric ion channel protein *Correspondence: 3 Present address: The Vollum Institute at Oregon Health and Sci-ence University, 3181 SW Sam Jackson Park Road, Portland, Ore-gon 97239.  and a bacterial transport protein. Subsequently, we de-scribe the application of FSEC to the screening of tagposition and of stability in different detergents. Finally,we evaluate Gaussian peak fitting of FSEC peak profilesas a method for quantitatively evaluating the mono- or polydispersity of a protein sample. Results and DiscussionCovalent GFP Fusions Our precrystallization screening methodology has twofacets.Thefirstinvolvesaseriesofnewexpressionvec-tors, forbacterial and mammalian cells, in which the tar-get gene is covalently linked to GFP. Fused to the termi-nus of GFP is a polyhistidine tag for affinity purification,and inserted between the target protein and GFP isathrombinsiteforproteolyticcleavageofthetargetpro-tein from GFP ( Figure 2 ). Enhanced green fluorescentprotein (EGFP) is chosen for eukaryotic expression,and GFPuv ( Crameri et al., 1996 ) is chosen for bacterialexpression in order to (1) maximize the stability of thechromophore in each expression system, (2) exploitthe stronger fluorescence signals in comparison to thewild-type counterparts, and (3) utilize genes that havecodons optimized for each expression system ( Cramerietal.,1996;Haasetal.,1996;Heimetal.,1995 ).ToreduceGFP-mediated dimerization that might confound FSECanalysis, alanine 206 in both GFP variants is mutated tolysine ( Zacharias et al., 2002 ). In all expression vectors,multiple cloning sites (MCSs) are located at the 5 0 or 3 0 side of the GFP coding sequence so that the target pro-tein can be tagged with GFP at its N or C terminus, re-spectively.BecausecompatibleMCSsareusedforbothN- and C-terminal GFP fusion vectors for each expres-sion system, one can readily screen N- and C-terminal-tagged constructs by using the same PCR product.The covalently fused GFP constructs allow one to vi-sually inspect subcellular localization of proteins ineukaryotic cells by fluorescence microscopy and to de-termine protein expression in bacterial cells by batchfluorescence. Moreover, polyhistidine and thrombinsites in the GFP fusion vectors allow one to purify andcharacterize the proteins from a small number of cellsand take advantage of the robust fluorescence fromGFP. These features profoundly benefit precrystalliza-tion screening of integral membrane proteins whose ex-pression levelsare usuallysignificantly lowerthanthoseof soluble proteins. Figure 1. Flow Chart of Precrystallization Screening(A) Traditional precrystallization screening. Variants of a target pro-tein are expressed and purified on mid- to large scales to producemicrogram to milligram quantities. The resulting purified proteinsarecharacterizedformonodispersityand stability byaseriesof bio-chemical assays including immunostaining (IS), Western blotting(WB), size-exclusion chromatography (SEC), and light scatteringexperiments. This precrystallization screening is continued untila promising construct is found.(B) FSEC-based precrystallization screening. Variants of a targetprotein are expressed as GFP fusionson a small scale and are char-acterizeddirectlybyFSECwithoutpurification. Here,thefusionpro-teins are analyzed for expression level, monodispersity, approxi-mate molecular mass, and stability.Figure 2. Maps of the GFP Fusion Vectors(A) Eukaryotic expression vectors (pNGFP-EU and pCGFP-EU). Transcription is drivenby a cytomegalovirus promoter (CMV) andterminated by SV40 polyadenylation se-quences.Sequencesencodingapolyhistidinetag, a thrombin proteolysis site (Th), and en-hanced green fluorescent protein (EGFP) arelocated at either the 5 0 end or the 3 0 end of the multiple cloning sites (MCS). The restric-tion sites in the MCS, the stop codons, the A206K mutation in EGFP, and the translationinitiation site (Kozak-ATG) are indicated.(B) Bacterial expression vectors (pNGFP-BCandpCGFP-BC).Forthebacterialexpressionvectors, transcription is directed by a T7 pro-moter and terminated by a T7 terminator se-quence. The coding sequence is designedin the same way as for the eukaryotic expres-sion vectors, except that a variant of uvGFP,which has codons optimized for bacterial ex-pression, is used instead of EGFP.Structure674Structure674  Fluorescence-Detection Size-ExclusionChromatography The second facet of our precrystallization screeningmethodology involves a chromatography system fittedwith an SEC column and a fluorescence detector. Withthis setup, one can monitor the elution of GFP fusionproteins in the context of whole-cell lysates or solubi-lized crude membranes ( Figure 3 ). We call this methodfluorescence-detection SEC (FSEC). Here, we will focusour attention on integral membrane proteins, althoughone can carry out FSEC precrystallization screening onwater-soluble proteins as well.For membrane proteins, crude membranes from bac-terial cells or intact tissue culture cells are solubilized ina detergent-containing solution, followed by a high-speed centrifugation step ( Figure 3 A). The supernatantis then directly applied to an SEC column equilibratedin a detergent-containing solution, and the column isconnected to a fluorometer fitted with a flow cell ( Fig-ure 3B). FSEC is a powerful screening method becausethe peak areas, profiles, and elution volumes providein-formation on (1) the expression level, (2) the degree of monodispersity, and (3) the approximate molecular mass of the fusion protein, respectively. Because FSECexploits the unique fluorescence signal of GFP, neither protein purification nor large-scale culture is required;readily obtainable fluorometers can detect  w 10 ng of GFP. In the following sections, we describe the applica-tion of our precrystallization screening strategy to four different membrane proteins expressed in bacterial or eukaryotic cells. Precrystallization Screening of P2X Receptors P2X receptors are eukaryotic integral membrane pro-teins that form ion channels gated by ATP ( Khakh,2001; North, 2002 ). There are seven P2X receptor sub-types (P2X  1–7  ), and all subtypes except P2X  6  form func-tional channels when expressed in human embryonickidney (HEK) 293 cells. To determine whether any of the rat P2X receptor subtypes would be suitable for crystallization trials, we carried out precrystallizationscreening with GFP fusion constructs. PCR productsof P2X  1–5,7  genes were subcloned into either pCGFP-EUor pNGFP-EU, and the resulting plasmids were trans-fected into HEK293 cells. Two days after transfection,the subcellular localizations of the P2X receptors werechecked by fluorescence microscopy ( Figure 4 A). Onthe basis of visual inspection, the C-terminally-taggedP2Xconstructs(C-P2Xs)expressedmorerobustlycom-pared to the N-terminally-tagged variants (N-P2Xs). InthecaseofP2X  3 ,boththeN-andC-terminal variantsre-sulted not only in fluorescence at the plasma mem-brane, but also in diffuse fluorescence in the cytoplasm,thelatterofwhichwaspresumablyduetofreeGFPgen-eratedbyadventitiousproteolysis.ForP2X  4 ,theC-P2X  4 variantappearedtoexpressatthehighestlevelandwasfoundinintracellularpuncta,whiletheN-P2X  4 constructwas found primarily on the cell surface. Although it ispossible that fusion of GFP to the receptor’s termini al-ters trafficking or ion channel function, we note thattheapproximatesubcellularlocalizationsoftheC-termi-nally-tagged P2X receptors observed in the presentstudy are consistent with previous observations ( Boba-novic et al., 2002 ). Moreover, electrophysiological andligand binding experiments have shown that the P2X  4 -GFPfusionspossessessentiallywild-typebehavior( Ka-wate, 2005 ). Therefore, these data indicate that GFP fu-sionsarerelativelybenignperturbationstothereceptor.To more quantitatively evaluate the expression levelandthedegreeofmonodispersityoftheP2Xconstructs,the N- and C-terminal GFP fusions were expressed intransiently transfected HEK293 cells; one 35 mm dishwas used for each construct. The cells were solubilizedin a buffer containing the nonionic detergent n-dodecyl- b -D-maltoside (C 12 M), and the resulting supernatantwas analyzed by FSEC. As shown in Figure 4B, the fluo-rescence peak associated with the C-P2X  4  constructwas much larger than the peaks from the other con-structs, thus confirming the initial observation that C-P2X  4  expressed at a higher level than the other  Figure 3. Flow Chart of PrecrystallizationScreening with GFP Fusion Proteins andFSEC(A) A gene of interest is amplified by PCR andsubclonedintooneoftheGFPfusionvectors.Cells are either transfected or transformedwith the expression vectors and solubilizedwith a detergent-containing buffer.(B) The resulting crude cell lysate, after cen-trifugation, is loaded directly onto an SECcolumn. The SEC column eluent is thenpassed through a flow-cell in a fluorometer set to detect GFP fluorescence. In this sche-matic,theFSECsetupincludesaUVdetector and a fraction collector, elements that areuseful for running standards or for purifyinga fusion protein based on its fluorescenceprofile. The panel labeled ‘‘Fluorescence’’ isa hypothetical elution profile of a GFP fusionprotein detected by GFP fluorescence. Thepanel labeled ‘‘UV absorbance’’ representsa model of a typical UV absorbance patternfrom a crude cell lysate.Rapid Precrystallization Screening by FSEC675  constructs. For the C-P2X  4  construct, the fluorescencepeak was nearly symmetric, the elution position of thepeak was suggestive of an oligomer, and there wasonly a small peak at the void volume of the column.The expression level of C-P2X  4  was estimated as w 1  m g per 10 6 cells by using a standard curve derivedfrom known concentrations and fluorescence yields of recombinant GFP.The other fusion proteins, C-P2X  5 , C-P2X  7 , N-P2X  4 ,and N-P2X  5 , had reasonably symmetrical peak shapes,but they all expressed at much lower levels ( Figure 4B).TheC-P2X  2 constructgaverisetoasmallbutsignificantpeak at the void volume, suggesting that it had a ten-dency to form high-molecular weight aggregates. Inter-estingly,theC-P2X  3 supernatantcontainedasubstantialamount of free GFP ( Figure 4B, bottom panel), whichwas consistent with the previous fluorescence micros-copy ( Figure 4 A). Taken together, inspection of trans-fected cells by fluorescence microscopy and analysisofsolubilizedcellsbyFSECsuggestthattheratP2X  4 re-ceptor is a promising molecule for crystallization trials. Precrystallization Screening of BacterialHomologs of Na +  /Cl 2 -DependentNeurotransmitter Transporters Na +  /Cl 2 -dependentneurotransmittertransporters(NSS)areintegralmembraneproteinsthatuseiongradientstodrivetheuptakeofabroadarrayofsubstrates,includingthe biogenic amines, amino acids, and osmolytes (beta-ine and creatine), into cells ( Masson et al., 1999 ). In aneffort to obtain crystals of a bacterial homolog of anNSS protein, genes corresponding to orthologs fromsix prokaryotic organisms (genes 1–6) were subclonedinto either pCGFP-BC (C-1, C-2, . C-6) or pNGFP-BC(N-1, N-2, . N-6) vectors and screened by FSEC after solubilizationofcrudemembranesinabuffercontainingC 12 M. As shown in Figure 5 A, the expression levels andthe degree of monodispersity of the C-terminal GFP fu-sions varied substantially. Moreover, the expressionlevels of constructs C-2 and C-6 were much greater than those of C-1, C-3, C-4, and C-5. Interestingly, theproteins that were more abundantly expressed yieldedmore symmetric peaks, whereas those that were poorlyexpressedgavelesssymmetricpeaks,suggestiveofag-gregation, misfolding, or heterogeneity in subunit stoi-chiometry. The differences of the calculated molecular masses of the fusion proteins are within 10%, yet pro-teinsC-1andC-3elutedsignificantlylaterthantheother proteins. One explanation for this behavior is that pro-teins C-1 and C-3 were binding to the resin, perhapsdue to misfolding or partial unfolding. The FSEC tracesfor the N-terminal fusions showed lower expressionlevels compared to the C-terminal variants, suggestingthat tagging at the N terminus had a deleterious effecton expression (data not shown).On the basis of the FSEC screening, the target pro-teins of the C-2 and C-6 constructs were subjected tocrystallization trials with proteins that were expressedas non-GFP fusions. After affinity purification, the puri-fied C-6 product was subjected to SEC detected by ab-sorbanceat280nm( Figure5B).Thequalitativesimilarityof the FSEC ( Figure 5 A) and the SEC profiles indicatesthat the GFP tag, aswell asthe purity and concentrationof the target protein, did not substantially affect the Figure 4. Precrystallization Screening of P2X Receptors by Epifluorescence and FSEC(A)Fluorescence microscopic images of HEK293 cells expressing P2X-GFP fusion proteins. The images were taken 48 hr after transfection.Thescale bar is 100  m m.(B) FSEC traces from P2X-GFP fusion proteins. The top panel shows the FSEC profiles of C-terminally tagged P2X  1–5, 7  (C-P2X  1–5, 7  ), and thebottom panel shows those of N-terminally tagged P2X  3–5  (N-P2X  3–5  ), including C-P2X  3  and C-P2X  5 . The arrows indicate the estimated elutionposition of the void volume, a P2X oligomer (ca. trimer to hexamer), a monomeric P2X subunit, and free GFP, respectively. Note the differencein scales for the top and bottom panels.Structure676  monodispersity of the C-6 construct. In fact, the targetproteinfromtheC-6constructcrystallizedreadily,yield-ing crystals that diffracted beyond 1.7 A ˚ resolution ( Fig-ures 5C–5E) (  Yamashita et al., 2005 ). Screening Tag Position by FSEC Using traditional approaches to precrystallizationscreening, we had previously obtained diffraction-qual-ity crystals of the trimeric glutamate transporter homo-log from  Pyrococcus horikoshi   (Glt Ph  ) expressed witha C-terminally-tagged construct ( Figures 6B and 6C)(  Yernool et al., 2004 ). Interestingly, N-terminally-taggedvariants of Glt Ph  did not express as well as C-termi-nally-tagged constructs and did not yield crystals(data not shown). At the molecular level, inspection of the Glt Ph  crystal structure shows that the C terminus isprojecting away from the protein; thus, it appears thatthe protein can accommodate a C-terminal tag. A mo-lecular understanding of the difficulties encounteredwith N-terminal tags is less clear, in part because thefirst few residues of the protein cannot be reliably posi-tioned in electron density. However, there are electrondensity features that suggest that the N terminus makescontactwiththeproteincore,andthismaybewhyN-ter-minal fusions are not tolerated.To test whether FSEC can provide data to determineoptimal tag location, we cloned the Glt Ph  gene into thepCGFP-BC and pNGFP-BC  E. coli   expression vectors,yielding the C-Glt Ph  and N-Glt Ph  constructs. Analysis of crude solubilized membranes by FSEC ( Figure 6 A)showed that while C-Glt Ph  had a narrow and symmetricpeak, N-Glt Ph  yielded a smaller and asymmetric peak,suggestive of heterogeneity in subunit stoichiometryand/or incomplete assembly. These results are consis-tent with our previous observations made prior to thedevelopment of FSEC technology. Therefore, the stud-ies of the N- and C-terminal fusions of Glt Ph  highlightthe importance of screening fusions at both ends of the target protein, and they emphasize how FSEC pre-crystallization analysis can provide important informa-tion rapidly and easily. Detergent Screening by FSEC Successful crystallization of a membrane protein is of-ten critically dependent on the detergent, and, in manycases, the most well-ordered crystals are formed in Figure 6. Screening Tag Position by FSEC(A)FSECtracesfromeitherN-orC-terminallytaggedGlt Ph ,aeukaryoticglutamatetransporterhomologfrom P.horikoshii  .Thetransportergenewas expressed with pCGFP-BC or pNGFP-BC, and the behaviors of these fusion proteins were examined by FSEC.(B) Hexagonal crystals of Glt Ph  protein are shown. The bar indicates 200  m m.(C) A diffraction image from the hexagonal crystal. These crystals diffract anisotropically to ca. 3.2 A ˚ along c* and to 3.8 A ˚ along a*.Figure 5. Precrystallization Screening andStructural Determination of a Bacterial Ho-molog ofNa +  /Cl 2 -DependentNeurotransmit-ter Transporters(A) FSEC traces from C-terminally taggedbacterial homologs. Six different genes (1–6)were screened by using the pCGFP-BC ex-pression vector. The arrows indicate thevoid volume and the elution volumes corre-sponding to the molecular weights.(B) SEC trace of the gene 6 product detectedbyUVabsorbanceat280nm.Afteraffinitypu-rification, the target protein without GFP wasconcentrated to w 2 mg/ml and subjected toSEC.(C)Rod-shaped crystals ofthegene6proteinare shown. The length of the bar is 200  m m.(D) A diffraction image from a rod-shapedcrystal that diffracted beyond 2.8 A ˚resolu-tion.(E) A ribbon diagram of the gene 6 protein, asodium-dependent leucine transporter from  Aquifex aeolicus  (  Yamashita et al., 2005 ).Rapid Precrystallization Screening by FSEC677
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