Methods 34 (2004) 354–363www.elsevier.com/locate/ymeth1046-2023/$ - see front matter
2004 Elsevier Inc. All rights reserved.doi:10.1016/j.ymeth.2004.03.024
The use of recombinant methods and molecular engineeringin protein crystallization
Zygmunt S. Derewenda
¤
Department of Molecular Physiology and Biological Physics, University of Virginia School of Medicine, Charlottesville, VA 22908, USA
Accepted 24 March 2004
Abstract
Recombinant techniques are routinely used for the preparation of protein samples for structural studies including X-ray crystal-lography. Among other bene
W
ts, these methods allow for a vast increase in the amount of obtained protein as compared to puri
W
ca-tion from source tissues, ease of puri
W
cation when fusion proteins containing a
Y
nity tags are used, introduction of SeMet forphasing, and the opportunity to modify the protein to enhance its crystallizability. Protein engineering may involve removal of
X
exi-ble regions including termini and interior loops, as well as replacement of residues that a
V
ect solubility. Moreover, modi
W
cation of the protein surface to induce crystal growth may include rational engineering of surface patches that can readily mediate crystal con-tacts. The latter approach can be used to obtain proteins of crystals recalcitrant to crystallization or to obtain well-di
V
racting crystalsin lieu of wild-type crystals yielding data to limited resolution. This review discusses recent advances in the
W
eld and describes a num-ber of examples of diverse protein engineering techniques used in crystallographic investigations.
2004 Elsevier Inc. All rights reserved.
1. Introduction
Crystallization is the key limiting step in macromolec-ular X-ray di
V
raction analysis, ironically more so todaythan during the few decades after J.D. Bernal and D.Hodgkin recorded the
W
rst X-ray di
V
raction from a pro-tein crystal [1]. The legacy of Sumner, Northrop, andother pioneers of protein biochemistry, who crystallizeda multitude of enzymes during the 1920s and 1930s, keptthe crystallographers busy for much of the subsequenthalf-century. Hemoglobin, pepsin, insulin, lysozyme, andmany other early targets of X-ray analysis had all beencrystallized well before full structural analysis becametechnically possible, and the availability of crystals wasnot an issue. However, as the number of crystal struc-tures tackled by crystallographers escalated in the early1980s, it became obvious that the prospects for progressare dramatically limited by the availability of target pro-teins, and—more importantly—by their propensity toform good quality di
V
racting crystals. Protein crystallog-raphers no longer had the luxury of picking up the nextavailable crystal from the shelf. As there was no way topredict crystallization conditions, and no way to alterthe nature of the puri
W
ed protein—except, perhaps forthe level of purity—the only way to crystallize a proteinwas to screen a wide range of possible conditions. Someof the ideas that even today constitute the core of high-throughput crystallization screens, such as the incom-plete factorial method, date from that period [2].With ine
Y
cient and labor-intensive puri
W
cation pro-tocols, and uncertain prospects for crystallization, onlythe most abundant and stable medium-size proteins,such as those found in bodily
X
uids or muscle, were con-sidered to be feasible targets. The whole
W
eld might havestalled prematurely, had it not been for the advent of contemporary recombinant methods and heterologousoverexpression. Shortly after the
W
rst successful expres-sion of recombinant proteins in
Escherichia coli
, includinginsulin [3] and somatostatin [4], protein crystallogra-phers turned to recombinant methods as the means toobtain samples for crystallization. Among the very
W
rstproteins crystallized using recombinant samples were
¤
Fax: 1-804-982-1616.
E-mail address:
Z.S. Derewenda / Methods 34 (2004) 354–363
355
insulin [5], human leukocyte interferon A [6,7], murineinterferon-
[8], and eglin C [9],With the dawn of recombinant methods, modi
W
cationof protein and using protein as another variable in thecrystallization experiments became also possible. Priorto that point, the only way to use the protein as a vari-able was to screen homologues from various species. Theconcept was originally used by Kendrew [10] in his origi-nal selection of sperm whale myoglobin as a target forX-ray analysis. Later on, Campbell et al. [11] reiteratedthe idea and suggested screening of various species as aroutine tool in search for a crystallizable protein homo-logue. The approach has been successfully used in manystudies, and is still often exploited in membrane proteincrystallography, where
W
nding a suitable and crystalliz-able variant of the target molecule is even more of achallenge.In spite of its obvious advantages, homologue screen-ing is cumbersome, and not useful if a speci
W
c structureis needed. In contrast, recombinant methods o
V
er arange of possibilities for modifying the target protein tohelp in the expression, puri
W
cation, and crystallization.As a consequence, most protein crystallography groupshave incorporated molecular biology techniques intotheir standard arsenal of methodologies, and ofteninvest much more time at the bench producing the pro-tein and crystallizing it, than solving the structure oncecrystals are in hand.The aim of this review is to provide the reader with anoverview of the contemporary protein engineeringapproaches used in a protein crystallography laboratoryto facilitate the preparation of high-quality crystals. Thesubject is broad, and a truly comprehensive review is notpossible within the limitations of this chapter. Thus, wehighlight the selected important aspects and provide rep-resentative examples of case applications for globularproteins. Particular attention is given to the advances incrystallization enhancement via surface mutagenesis. Wedo not discuss protein engineering of membrane proteins,as the options there are very limited due to the much lessadvanced stage of heterologous overexpression.
2. Use of recombinant proteins in crystallization
The advantages of using recombinant proteins forcrystallization typically include, among others, a dra-matic increase in the yield of protein, ease of puri
W
cationusing a
Y
nity tags and fusion proteins, possibility of pro-tein modi
W
cation or use of isolated domains, and ease of labeling with SeMet for phasing purposes. A survey of crystallographic studies published last year in high-impact journals (excluding membrane proteins) indi-cates that over 90% of crystal structures are based onrecombinant material (Bielnicki, J., Janda, I., Derew-enda, Z.S., unpublished).Bacteria, and in particular
E. coli
, are preferred asprotein producers for biophysical studies [12]. It is thebest characterized host with a diversity of strains for spe-ci
W
c applications, such as expression of proteins withnon-optimal codon bias, protease-de
W
cient strains,strains containing chaperones for enhanced folding, etc[13]. Other advantages include ease of handling com-pared to eukaryotic cell cultures, multitude of vectors forexpression of numerous fusion proteins, and overall lowcost. However, in spite of these advantages, not all pro-teins of interest will be expressed with an adequate yield,in a properly folded soluble form, or at all.A simple method to increase the yield of protein pro-duction is to increase the culture density from the typicalOD of 0.6–2.0 or higher. Very high density can beachieved using disposable 2-L plastic bottles, which getwell aerated in shakers and which also minimize theoverall e
V
ort by disposing with sterilization [14]. Themedia composition may also be altered and both mini-mal media and rich media have been shown to be advan-tageous in speci
W
c cases. To further increase the yield forthose proteins for which expression levels are particu-larly low, it is necessary to resort to fermentors, althoughthe optimization of large-scale expression is often a sig-ni
W
cant e
V
ort in itself [15].Protein aggregation and inclusion body formation areoften a problem [16]. To monitor the solubility of theexpressed protein, one can use a fusion with a down-stream located GFP (green
X
uorescent protein). Underthose circumstances, increase in
X
uorescence indicatesproper folding of the target protein and can be used as adiagnostic [17]. If the target protein expresses well, butgoes into inclusion bodies, the di
Y
culties can often beresolved using a number of alternative strategies. Thesimplest remedy is to lower the IPTG inducer concentra-tion to
»
0.2mM and reduce the post-induction tempera-ture [18], so that the translation machinery slows downallowing the polypeptide chain enough time to fold. Forexample, reduction of the temperature from the typical30–37°C to 25°C resulted in a 2-fold increase in the sol-uble fraction of apoglobin in
E. coli
[19]. A similar resultwas obtained for the phage K11 RNA polymerase,which required a temperature of 25°C for proper foldingand solubility [20]. In the case of the hepatitis C virusNS3 serine protease, maximum solubility was achievedwhen the temperature was dropped to 15°C [21]. In ourlaboratory, we routinely bring the temperature to 10°Cwith very good results (Derewenda, U., unpublished).If the temperature does not help, it may be necessaryto use an
E. coli
strain co-expressing molecular chaper-ones belonging to the DnaK–DnaJ–GrpE and GroEL– GroES systems [22]. The 50kDa
E. coli
trigger factor, apeptidyl-prolyl
cis/trans
isomerase, can also signi
W
cantlyhelp to prevent protein aggregation, with or without theassistance of chaperones [23]. Finally, it should not beforgotten that some carrier proteins used in fusion
356
Z.S. Derewenda / Methods 34 (2004) 354–363
constructs for expression (see below) signi
W
cantlyincrease the amount of folded, soluble protein in expres-sion experiments.Low expression levels in
E. coli
may be caused by thepresence of rare codons within the target gene [24]. Thiscan be dealt with in two ways. The
W
rst option is to usestrains adjusted with additional copies of rare tRNAgenes. Among the most popular are BL21(DE3)RILcells containing extra copies of
argU
,
ileY
, and
leuW
genes, and the RP variant of the same strain which con-tains extra copies of the
argU
and
proL
tRNA to over-come expression problems of genes containing eitherAGG/AGA or CCC codons in CG-rich genes [25]. Insome cases, the yield of recombinant protein obtainedusing these cells can be over 100 times more than usingconventional BL21(DE3) cells [26]. The second option isto use wholly or partly synthetic genes, with optimizedcodons. For example, in the case of the gene encodingthe herpes simplex virus 1 protease the use of a syntheticgene with optimized codon usage leads to a 20-foldincrease in the expression yield [27].Interestingly, rare codons may not always be the rea-son behind low expression related to the cDNAsequence. In their studies of the poorly expressing S1dehydrofolate reductase (DHFR), Dale et al. [28] foundthat the replacement of all rare codons in the
W
rst 66bases did little to help expression. However, replacementof the N-terminal 18 out of 22 amino acids by the corre-sponding codons from the SaDHFR, which is 80% tothe type S1 enzyme, led to a dramatic increase in expres-sion levels. It is possible that subtle issues involvingRNA secondary structure are at play.Finally, increasing intrinsic protein solubility via pro-tein mutagenesis of surface residues (see below) can alsosigni
W
cantly boost expression levels and yields. Forexample, four di
V
erent mutations of interferon-
increased the relative amount of the protein in the solu-ble fraction by 1-fold [29]. A similar result was reportedfor the S1 DHFR, as we discuss below [28].In spite of all the available options, many eukary-otic proteins cannot be produced in bacterial hostsdue to a multitude of factors, including but not limitedto diverse post-translational modi
W
cations. The solu-tion is to use more labor-intensive and time-consum-ing eukaryotic hosts such as insect cells, baker's yeast,
Pichia pastoris
or CHO (Chinese hamster ovary) cells.These systems are signi
W
cantly more labor-intensive,and a limited scope of this review precludes theirdetailed presentation.Finally, progress is being made on cell-free systemsfor large-scale protein production [30]. Although at thetime when this review was written, these systems didnot yet constitute an alternative. Due in part to a veryhigh cost, they are expected—when optimized—to rev-olutionize the
W
eld in the possibly not too distantfuture.
3. The use of fusion proteins for expression andpuri
W
cation
Our literature survey indicated that approximately75% of proteins for crystallization are expressed asfusion constructs [31] using a number of small proteinsand tags for a
Y
nity puri
W
cation as well as for solubilityenhancement. A wide choice of systems is available forbacterial hosts, and a smaller but ever expanding selec-tion is available for eukaryotic cells. Among the fusionpartners particularly popular among structural biolo-gists are the hexaHis tag [32], GST (glutathione-
S
-trans-ferase) [33], and MBP (maltose-binding protein) [34].Some of the less commonly used include thioredoxin[35], Z-domain from protein A [36], NusA [37] GB1-domain from protein G [38], and a number of others [13].None of the tags is universally superior, and often paral-lel expression experiments determine what the best strat-egy is. Families of vectors allowing for parallel assays of this kind have been developed [39,40].A survey of recent crystallographic studies revealsthat nearly 60% used one of the several incarnations of the polyHis tag, making it by far the
W
rst choice amongcrystallographers. The popularity is due to the relativesimplicity of Ni-a
Y
nity chromatography and the factthat the tag itself need not necessarily pose a majorobstacle in crystallization trials. While some groupsadvocate the need to remove proteolytically the tag priorto crystallization, others argue against it or use a two-tier screening process whereby the
W
rst screen usestagged proteins, and the second screen is performed aftertag removal only for those proteins that failed to crystal-lize during the
W
rst attempt. Many examples of proteinscrystallized with the His-tag left intact are reported inthe literature. The His-tag works well with soluble, smallto medium size proteins, but is known in many cases tolower the solubility. Interestingly, a comparative studyof several N-terminal fusion tags and 32 human proteinsrevealed that most protein tags are superior to the hexaHistag with respect to conferring solubility, and two—thio-redoxin and MBP (maltose-binding protein)—appear tobe particularly powerful [41]. Thus, if the His-taggedprotein is not adequately expressed as a soluble protein,the investigator should immediately consider changingthe fusion protein.A distant second in popularity among crystallogra-phers as a fusion tag is GST. This tag is useful as a solu-bility enhancer, but it also creates some problems due toits dimeric state, and is not appropriate for expression of homo-oligomers because of the possibility of formingaggregates. Finally, MBP has recently received a lot of attention due to its potential to yield soluble proteinsotherwise di
Y
cult or impossible to express [42]. However,MBP is not very e
V
ective as a puri
W
cation tag, becauseMBP-fusion proteins often fail to bind to the amyloseresin [43]. Thus, the use of supplementary a
Y
nity tags is
Search History:
Result 00 of 00
00 results for result for
p.
Use of Recombinant Methods and Molecular Engineering in Protein Crystallisation
1.0K
Reads
0
Readcasts
1
Embed Views
Notes
Load more
