A streamlined approach to screening for the expression of recombinant membrane proteins in Escherichia coli based on fusion to green fluorescent protein is presented.
The production of recombinant membrane proteins for structural and functional studies remains technically challenging due to low levels of expression and the inherent instability of many membrane proteins once solubilized in detergents. A protocol is described that combines ligation independent cloning of membrane proteins as GFP fusions with expression in Escherichia coli detected by GFP fluorescence. This enables the construction and expression screening of multiple membrane protein/variants to identify candidates suitable for further investment of time and effort. The GFP reporter is used in a primary screen of expression by visualizing GFP fluorescence following SDS polyacrylamide gel electrophoresis (SDS-PAGE). Membrane proteins that show both a high expression level with minimum degradation as indicated by the absence of free GFP, are selected for a secondary screen. These constructs are scaled and a total membrane fraction prepared and solubilized in four different detergents. Following ultracentrifugation to remove detergent-insoluble material, lysates are analyzed by fluorescence detection size exclusion chromatography (FSEC). Monitoring the size exclusion profile by GFP fluorescence provides information about the mono-dispersity and integrity of the membrane proteins in different detergents. Protein: detergent combinations that elute with a symmetrical peak with little or no free GFP and minimum aggregation are candidates for subsequent purification. Using the above methodology, the heterologous expression in E. coli of SED (shape, elongation, division, and sporulation) proteins from 47 different species of bacteria was analyzed. These proteins typically have ten transmembrane domains and are essential for cell division. The results show that the production of the SEDs orthologues in E. coli was highly variable with respect to the expression levels and integrity of the GFP fusion proteins. The experiment identified a subset for further investigation.
It has been estimated that membrane proteins account for approximately 20-30% of the genes coded in all sequenced genomes 1, including the human genome 2. Given their critical role in many biological processes, for example transport of metabolites, energy generation and as drug targets, there is intense interest in determining their three dimensional structure. However compared to soluble proteins, membrane proteins are highly under-represented in the Protein Data Bank. In fact, of the 99,624 released structures in the PDB (http://www.rcsb.org/pdb/) only 471 (http://blanco.biomol.uci.edu/mpstruc/) are classified as membrane proteins. This reflects the technical difficulties of working with membrane proteins which are naturally expressed at relatively low levels; rhodopsin is one of a few exceptions3,4. Over-expression of recombinant versions in heterologous cells often results in misfolding and poor targeting to the membrane which limits the overall level of production. Selecting the best detergent for extraction and solubilization of a membrane protein once expressed makes subsequent purification difficult and requires careful optimization.
Escherichia coli remains the most commonly used expression host for membrane proteins accounting for 72% (http://blanco.biomol.uci.edu/mpstruc/) of structures solved to date which are almost exclusively from bacterial sources. Specific E. coli strains, C41(DE3) and C43(DE3), have been isolated which do not lead to the over-expression of membrane proteins and hence avoid the effects of toxicity observed for other BL21 strains5,6. Tuning the level of recombinant membrane protein expression appears to be critical for optimizing the level of production6,7.
In many cases successful production of membrane proteins for structure determination has required the evaluation of multiple versions including amino- and carboxy-terminal deletions, multiple orthologues to exploit natural sequence variation and different fusion proteins6-8. Therefore, a method for the expression screening of multiple membrane proteins in parallel is essential. One commonly used approach is to fuse the protein to green fluorescent protein (GFP) enabling expression to be tracked without the need to purify the protein. In this way, information about expression level, stability and behavior in different detergents can be assessed with small amounts of un-purified material9-12.
In this article we describe a streamlined protocol for cloning and expression screening of membrane proteins in E. coli using fusion to GFP as a reporter of production and subsequent characterization of detergent: protein complexes (Figure 1).
1. Construction of Expression Vectors
2. Transformation of E. coli Expression Strains
Note: Prior to performing this protocol, E. coli competent cells are made, aliquotted and stored frozen at -80 °C as described previously13. Membrane protein expression is routinely screened in two strains, Lemo21(DE3) and C41(DE3)pLysS. To aid interpretation of the results it can be helpful to include a positive control, e.g. a plasmid expressing GFP or a membrane protein fused to GFP known to express and an empty vector negative control.
3. Preparation of Overnight Starter Cultures
Note: Always prepare overnight cultures the day after transformation never from an old transformation plate.
4. Growth and Induction of Cultures
5. Analysis of Expression by In-gel Fluorescence
6. Scale-up of Cell Cultures
7. Preparation of Membranes
8. Solubilization and Analysis of the Membrane Fraction
The SEDS family of proteins (shape, elongation, division, and sporulation) is essential for bacterial cell division. These integral proteins typically have ten transmembrane domains with both amino and carboxy termini inside the cell. To exemplify the protocol described above, 47 SEDs orthologues were cloned into the vector pOPINE-3C-eGFP. A small scale expression screen of the constructs was carried out in two E .coli strains, Lemo21(DE3) grown in the presence of 0.25 mM rhamnose to block leaky expression and C41(DE3) pLysS, which are routinely used for the expression of membrane proteins5-8.
Detergent solubilized lysates of induced cultures were analyzed by SDS-polyacrylamide electrophoresis. Using in-gel fluorescence to visualize expressed GFP fusion proteins showed that between the two strains 38 out of the 47 SEDs constructs were produced. Interestingly there were differences between the two expression strains. Of the 38 expressed constructs, only 18 were produced in both strains, 14 only expressed in Lemo21(DE3) and 6 only in C41(DE3) pLysS. Overall, there was a higher success rate for the Lemo21(DE3) compared to the C41(DE3) pLysS (38 vs 24). However the observation that some proteins appeared to be strain specific means that screening in both is useful.
Representative data for 24 of the 47 SEDs constructs are shown in Figure 1. The intensity of the bands gives an indication of the level of expression, e.g. the fusion protein in well C4 in Figure 1A and 1B is expressed well in both strains however the additional lower molecular weight bands suggest that the protein is partially degraded. In many lanes there is a band which corresponds in size to GFP alone. A qualitative assessment of the integrity of the protein can be carried out by looking at the intensity of fusion protein relative to the free GFP; e.g. G4 and H4 are completely broken down to free GFP in C41(DE3) pLysS (Figure 1B) whereas in the Lemo21(DE3) there is some intact protein (Figure 1A). For 14 out of the 38 expressed proteins the intensity of the free GFP band is similar to that of the fusion protein, for 21 the intensity of the free GFP band is greater than that of the fusion protein and a minority of the fusion proteins (3/38 expressed) e.g. F6 and G6 in Figure 1A have relatively little free GFP compared to the fusion protein.
Two of the constructs that expressed well in the primary screen B5 (high intensity bands for free GFP and fusion proteins) and G6 (little free GFP) were expressed and a total membrane fraction prepared from 500 ml cell culture. Resuspended membranes were divided into aliquots and analyzed by fluorescence-detected size exclusion chromatography (FSEC). The FSEC profiles are presented in Figure 2A and 2B respectively and show that the two SEDs proteins behave differently in the four detergents tested. In one case (Figure 2A: sample B5) the protein only gave a symmetrical peak with little aggregation in DDM which would therefore be the detergent of choice for subsequent purification. By contrast the other fusion protein (Figure 2B: sample G6) gave a similar profile in all detergents tested.
Figure 1: Diagram of workflow for screening membrane protein expression in E. coli using a GFP reporter.
Figure 2: Primary screen of membrane protein expression using in-gel fluorescence. Detergent lysates of E. coli cells were analyzed by SDS-PAGE and gels imaged using Blue Epi illumination and a 530/28 filter. The results are shown for 24 constructs (labeled A4-H6 based on their position in a 96-well plate). The positions of the bands for free GFP and GFP fusion proteins are indicated. (A) Proteins expressed in Lemo21(DE3). (B) Proteins expressed in C41(DE3) pLysS.
Figure 3: Secondary screen of membrane protein expression using Fluorescence-detected size exclusion chromatography (FSEC). Total membrane fractions were extracted in four different detergents and the solubilized membrane proteins analyzed by size exclusion chromatography using a FPLC system coupled to a fluorescence detector. FSEC profiles for (A) membrane protein B5 and (B) membrane protein G6. The position of the peaks corresponding to aggregated protein, solubilized GFP fusion protein and free GFP are indicated. The detergents tested are indicated below the profiles.
In the protocol described in this article, ligation independent cloning into a GFP reporter vector is combined with fluorescence detection to rapidly screen the relative expression of multiple membrane proteins in E. coli. Vectors can be made in four working days with primary expression screening taking a further week. In-gel fluorescence of detergent lysates of whole cells is used to rank expression hits for further analysis in a secondary screen. The primary expression screen as presented does not require any specialist equipment apart from a shaker incubator suitable for growing cells in deep well blocks. All other liquid handling involving 96-well SBS plates can be carried out using multi-channel pipettes. The protocol can be readily modified to include other E.coli strains grown under different expression conditions (e.g. lower temperature). In the case of the LEMO21(DE3) titrating the level of rhamnose (from 0 to 2.0 mM) added to modulate the rate of transcription can increase the level of expression of some membrane proteins7 . Detergents other than DDM could be used for extraction in the primary screen though harsher detergents may solubilize proteins from inclusion bodies and therefore give false positives. Clearly, the more elaborate the initial screen becomes, the more time-consuming the experiment.
The secondary screen involves preparing a total membrane fraction from the expression hits identified in the primary screen. This is a critical step as it will confirm the localisation of the fusion proteins to the membrane of the E. coli cells. Subsequent extraction of the GFP fusion protein in different detergents is monitored by fluorescence-detected size exclusion chromatography of solubilized material to assess the mono-dispersity of the detergent-protein complexes. This is another critical step since it identifies the most appropriate detergent for solubilization of the membrane protein for subsequent downstream processing. However, experience shows that further optimization of the choice of detergent(s) may still be required during purification and crystallization. Evidence of uniform behavior in different detergents including ones with a relatively small micelle size (e.g. LDAO) appears to be good indicator of a propensity to form well diffracting crystals at least for prokaryotic membrane proteins14. In this respect the profile shown for the membrane protein in Figure 2B appears highly favorable.
The main advantage of using a GFP reporter is that it gives a rapid read-out of expression in un-purified samples. A disadvantage is that the GFP fusion protein has to be removed during the purification of the protein for crystallization. In the case of the vectors used here, a rhinovirus 3C protease site enables cleavage of the GFP. Alternatively, it is straightforward to re-clone candidate proteins, identified in the screen, into vectors which only add a polyhistidine or other affinity tag for subsequent purification. This assumes that fusion to GFP is not necessary for expression. The main alternative to using fusion to GFP for detection of membrane protein expression and solubilization is to add only a histidine tag. However this approach typically requires starting with a membrane fraction15 which for the evaluation of many variants would become very time-consuming.
In summary, the main purpose of the protocol described in this article is to enable a large number of membrane protein sequence variants to be rapidly evaluated in terms of expression and detergent solubilization and prioritized for deeper analysis.
The authors have nothing to disclose.
The OPPF-UK is funded by the Medical Research Council, UK (grant MR/K018779/1) and the Membrane Protein Laboratory by the Wellcome Trust (grant 099165/Z/12/Z).
Name of Material | Company | Catalog Number |
pOPIN-N-GFP | addgene (www.addgene.org) | |
pOPINE-3C-GFP | addgene (www.addgene.org) | |
C41(DE3)pLysS | Lucigen (http://lucigen.com/ | 60444-1 |
LEMO21(DE3) | New England Biolabs | C2528H |
In-Fusion HD EcoDry Cloning System, 96 Rxns | Clontech/Takara | 639688 |
Power broth | Molecular Dimensions (http://www.moleculardimensions.com/) | MD12-106-1 |
Protease Inhibitor tablets | Roche | 11697498001 |
cOmplete, Mini, EDTA-Free; Protease Inhibitor Cocktail | Roche | 11836170001 |
L-Rhamnose monohydrate | Sigma-Aldrich | 83650 |
Protease Inhibitor Cocktail | Sigma-Aldrich | P8849 |
DNAse 1 | Sigma-Aldrich | DN25 |
Lysozyme | Sigma-Aldrich | L7651 |
Cholesterol hemisuccinate | Sigma-Aldrich | C6512 |
CYMAL-6 ( 6-Cyclohexyl-1-Hexyl-β-D-Maltoside) | Anatrace | C326 |
DDM ( n-Dodecyl ß-maltoside ) | Generon | D97002 |
DM (n-Decyl ß-maltoside) | Generon | D99003 |
LDAO ( N,N-Dimethyl-n-dodecylamine N-oxide) | Generon | D71111 |
E1 ClipTip Eq384 8 channel 1-30 ul | Thermo Scientific | 4672030 |
Rainin Pipette Pipet-Lite XLS 6ch 100-1200µL LTS Spacer | Anachem | LA6-300XLS |
Rainin Pipette Light XLS+ 8ch 2-20µL LTS | Anachem | L8-20XLSPLUS |
Pipette Pipet-Lite XLS 8ch 100-1200µL LTS Tip | Anachem | L8-1200XLS |
24 well V bottom deep well blocks | Porvair sciences | 360115 |
BenchMark Fluorescent Protein Standard | Life Technologies | LC5928 |
NuPAGE Novex 10% Bis-Tris Midi Protein Gels, 26 well | Life Technologies | WG1203BOX |
XCell4 SureLock Midi-Cell | Life Technologies | WR0100 |
Vibramax 100 | Heidolph | 544-21200-00 |
Tension rollers attachment | Heidolph | 549-81000-00 |
ptima L-100 Ultracentrifuge running 45-Ti rotor | Beckman Coulter | 393253 |
Optima Max-XP ultracentrifuge running TLA-55 rotor | Beckman Coulter | 393315 |
Floor standing centrifuge, Avanti J-26S XPI running JA-17 and JS-5.3 rotors | Beckman Coulter | B14535 |
Prominence UFLC with RF-20A Fluorescence detector | Shimadzu | |
Sepharose 6 10/300 GL size exclusion column | GE Healthcare | 17-5172-01 |
SSI5R-HS SHEL LAB Floor Model Shaking Incubator, 5 Cu.Ft. (144 L), 30-850 RPMs | Shel Lab | SSI5R-HS |
Innova44R incubator | New Brunswick /Eppendorf | Innova44R |
TS SERIES 0.75KW DISRUPTER 230V 50HZ 1PH (40KPSI) | Constant systems | PL083016 |
L-Rhamnose monohydrate | Sigma | 83650 |