Detergent-protein and detergent-lipid interactions : implications for two-dimensional crystallization of membrane proteins and development of tools for high throughput crystallography

摘要

2.1 Scope of this ThesisudThis thesis represents an attempt to enlighten the role of the detergent in reconstitution and more specificallyudin two-dimensional (2D) crystallogenesis of membrane proteins. The construction of a tool for preciseudand routine measurements of detergent concentrations provided a valuable tool for better understandingudand controlling the detergent issue. Additionally, a novel approach for detergent removal in 2D crystallization,udi.e. the use of cyclodextrins was explored and a nanoliter dispensing high throughput tool wasuddeveloped allowing for profound and sophisticated screening of optimal conditions for protein reconstitutionudand crystallization.udud2.2 Combining Electron Microscopyudand Atomic ForceudMicroscopyudAlthough electron crystallography has proven to beuda powerful approach to structure determination ofudmembrane proteins (for a recent example see (Gonenudet al., 2005)) successes are somehow restrictedudto certain classes of membrane proteins (e.g., outerudmembrane porins, aquaporins, naturally occurringudcrystalline proteins). This is mainly due to the stabilityudof these proteins with respect to biochemicaludmanipulation. One can not exclude however, thatudthese are simply more amenable to crystallizationuddue to the nature of their molecular surfaces.ud2D crystallization exhibits several advantagesudcompared to 3D crystallization of membrane proteins:udThe simple fact that the proteins are allowedudto reside in a native-like environment, i.e.,udthe membrane and that their function is not impairedudby the lateral crystal contacts is of considerableudinterest. If structural investigations shall notudbe restricted to static snapshots of different conformationsudand moreover structure-function relationshipsudshall be established, then electron microscopyud(EM) in combination with atomic force microscopyud(AFM) surely represent a valuable approach.udIn Chapter 2 the combination of such data hasudbeen successfully applied to the ammonium transporterudAmtB from Escherichia coli. The aim was touddetermine the crystal packing of the double-layeredud2D crystals of AmtB by AFM in order to process theudcryo EM data. Additionally, the AFM images, dueudto their outstanding signal-to-noise ratio, enabledudthe direct visualization of trimers in the reconstitutedudmembranes. The topographical data fromudthe AFM allowed the assessment of a single layerudwithin the double layered crystals.ud2.3 Investigating the Role of theudDetergentudIn Chapter 3 the development of a fast and preciseudmethod for detergent concentration determinationudis presented. The robustness and wide applicationudrange of this method has been demonstratedudby comparing concentrations of radioactivelyudlabeled dodecyl-[beta],D-maltoside (DDM) withudmeasured contact angles, by measuring the amountudof DDM bound to the proton/galactose symporterudGalP from E. coli, by measuring the effectsudof 100 mM NaCl on the cmc of dodecyl-N,Ndimethylamine-udN-oxide, by characterizing the surfaceudenergy of Parafilm, and finally by revealingudthe stoichiometry of complex formation betweenudmethyl-[beta]-cyclodextrin (MBCD) and different de-ududtergents. The possibility of performing such measurementsudroutinely in membrane biochemistry isudunique compared to all other methods available touddate.udChapter 4 addresses the major aspects of detergentuduse in membrane protein purification andudcrystallization. First, the stability of GalP in differentuddetergents is assessed, unveiling profounduddifferences in the capacity of detergents to keepudthe protein in solution. Second, it is demonstrated,udthat the amount of a detergent, i.e., dodecyl-�,Dmaltoside,udbound to a protein can be controlledudduring purification. At last the amount of differentuddetergents for solubilization of E. coli lipids isuddetermined, showing differences in the mechanismsudby which detergents promote solubilization.udBanerjee et al. (Banerjee et al., 1995) examinedudthe preferential affinity of detergents for differentudlipids in mixed membranes (such as biological membranes).udThey showed that different detergents extractudthe serotonin 5-HT1A receptor from nativeudmembranes along with different lipids. The effectudis considerable and might explain why different detergentsudexhibit such a different ability to keep audprotein in its native state, because some might simplyudnot be able to co-solubilize native lipids essentialudfor the stability (and function) of the protein.udThe amount of detergent bound to a protein isudof special interest when using dialysis or dilutionudfor detergent removal. Furthermore, in most casesudthe protein must not be exposed to excess detergentudwhich anyway fails to satisfactorily mimic theudnative bilayer. As pointed out in the discussion ofudChapter 4, protein reconstitution is facilitated whenudthe detergent collar that is present around the hydrophobicudregion of membrane proteins in solutionudis near its solubility limit (Psol).udThe same is true for the lipid: Reconstitution isudlikely to happen when liposomes are forming, thereforeudan excess of detergent is not desirable either.udAdditionally, even detergents known to have adverseudeffects on protein stability can be used forudlipid solubilization, given that they are present at audminimal concentration. The use of detergent mixturesudin crystallization can also have the effect ofudreducing the size of the detergent collar around theudprotein. Moreover, the free detergent concentrationudin detergent mixtures is altered by the presence ofudthe second species and can be crucial to the formationudof crystals in some cases (Koning, 2003).udWhen using minimal amounts of detergent in audcrystallization mixture, special care should be takenudwith respect to the formation of ternary micelles.udIdeally, equilibration of the ternary mixtures priorudto detergent removal needs to be completed.ud2.4 The Use of Cyclodextrins forudHigh Thorughput 2D Crystallizationudof Membrane ProteinsudChapter 5 demonstrates the feasibility of theudcyclodextrin-based detergent removal for twodimensionaludcrystallization. The possibility ofudchoosing different kinetics, simply by adding differentudamounts of cyclodextrin at various time intervalsudis one of the major advantages of this method.udBy implementing optical spectroscopy, it would beudpossible to slow down the detergent removal rate atudthe onset of proteoliposome and 2D crystal formation.udAs pointed out by Lichtenberg et al. (Lichtenbergudet al., 2000) the rate of detergent removal hasudto be slow enough to allow for detergent-inducedudvesicle size growth, a process which is usually quiteudslow. This aspect is important to keep in mind asudone defines the rate of detergent neutralization (inudcontrast to dialysis). At a first glance one mightudthink that in this respect the cyclodextrin approachudbears no advantage compared to dialysis. However,udthe rate of low-cmc detergent removal using dialysisudcan be too slow, thereby keeping the protein outudfrom its native environment for too long, ultimatelyudpromoting its precipitation.udIn Chapter 6 we present an apparatus for paralleludquantitative reconstitution and 2D crystallizationudof membrane proteins. Cyclodextrin provides audunique opportunity for high throughput implementationudcompared to other methods available today.udProtein concentrating through controlled evaporationudwith concomitant detergent neutralization (toudprevent detergent concentrating) is advantageousudcompared to commercially available protein concentratinguddevices which very often concentrate detergentudmicelles too. Moreover, the possibility of usingudone protein preparation for wide screening ensuresudthat inconsistencies in results arising from preparativeuddifferences are excluded. Often, the detergentudand lipid concentration of the purified protein areudill characterized, and this variability may be a causeudfor much of the irreproducibility and failure in crystallizationud(Wiener, 2004).udSo far the use of wide screening matrices (sparseudmatrix design) in 2D crystallography was restrictedudby the enormous number of experiments andududamount of protein needed for a rigorous screening.udThe presented machine makes it possible to partiallyudcompensate for the first bottleneck in proteinudstructure elucidation, which is the over-expressionudof membrane proteins.udFig. 2.1 summarizes the screening strategy basedudon the criteria discussed in Chapter 6 and above.udScreening efficiency is provided by the subdivisionudof the problem into multiple subproblems and byudtheir sequential screening.udWith the high throughput approach however, audnew bottleneck arises as one will produce a largeudnumber of crystallization trials, which have to beudscreened for their outcome. Therefore –in analogyudto the x-ray community– the development of automatedudsample preparation and automated electronudmicroscopic analysis would provide substantial supportudto the 2D crystallographer.udCombining step-by-step identification of key valuesudnecessary for crystallization (and/or efficient reconstitution)udtogether with high throughput screeningudmatrices opens up new prospects in the enududdeavor to membrane protein structure and functionuddetermination. Now it is possible to apply audsemi-rational screening strategy and this might contributeudto transform 2D crystallization from art toudscience (Jap et al., 1992).