Nature Methods,Optimization of membrane protein overexpression and purificati

wupfzhongguo

Optimization of membrane protein overexpression and purification using GFP fusions

David Drew1, 4, Mirjam Lerch1, 4, Edmund Kunji2, Dirk-Jan Slotboom3 & Jan-Willem de Gier1

1 Department of Biochemistry and Biophysics, Stockholm University, SE-106 91 Stockholm, Sweden.

2 MRC Dunn Human Nutrition Unit, Hills Road, CB2 2XY Cambridge, United Kingdom.

3 Department of Biochemistry, University of Groningen, Nyenborg 4, 9747 AG Groningen, the Netherlands.

4 These authors contributed equally to this work.
Correspondence should be addressed to Jan-Willem de Gier degier@dbb.su.se
Optimizing conditions for the overexpression and purification of membrane proteins for functional and structural studies is usually a laborious and time-consuming process. This process can be accelerated using membrane protein–GFP fusions1, 2, 3, which allows direct monitoring and visualization of membrane proteins of interest at any stage during overexpression, solubilization and purification (Fig. 1). The exceptionally stable GFP moiety of the fusion protein can be used to detect membrane proteins by observing fluorescence in whole cells during overexpression, with a detection limit as low as 10 mug of GFP per liter of culture, and in solution during solubilization and purification. Notably, the fluorescence of the GFP moiety can also be detected in standard SDS polyacrylamide gels with a detection limit of less than 5 ng of GFP per protein band (Fig. 2). In-gel fluorescence allows assessment of the integrity of membrane protein–GFP fusions and provides a rapid and generic alternative for the notoriously difficult immunoblotting of membrane proteins. With whole-cell and in-gel fluorescence the overexpression potential of many membrane protein–GFP fusions can be rapidly assessed and yields of promising targets can be improved. In this protocol the Escherichia coli BL21(DE3)-pET system—the most widely used (membrane) protein overexpression system—is used as a platform to illustrate the GFP-based method. The methodology described in this protocol can be transferred easily to other systems.

Figure 1. Flowchart illustrating optimization of membrane protein overexpression and purification using GFP fusions.




Figure 2. Monitoring overexpression of membrane protein GFP fusions using whole-cell and in-gel fluorescence.

(a) Indicated amounts of purified GFP-8His were run on a 12% SDS polyacrylamide gel. In-gel fluorescence was monitored (Steps 11–13) and then the gel was stained with Coomassie (left). Intensities of in-gel fluorescent signals after 0.5 s exposure were plotted versus the amounts of GFP-8His loaded (right). (b) A culture of BL21(DE3)pLysS cells harboring pYedZ-TEV-GFP-8His was grown as described in Steps 3–6. After induction of expression of the YedZ-TEV-GFP-8His fusion with 0.4 mM IPTG at 25 °C, samples were collected at indicated times. YedZ-TEV-GFP-8His expression was monitored by means of whole-cell fluorescence (Steps 7–9) and in-gel fluorescence (top; Steps 10–13). The whole-cell fluorescence signals were plotted versus the intensities of the in-gel YedZ-TEV-GFP-8His fluorescence signals (bottom). (c) To compare protein production in different culture volumes, 13 different membrane protein–GFP fusions were expressed in BL21(DE3)pLysS cells in 1-ml and 1-l cultures as described in Steps 3–6. Four hours after induction of expression, whole-cell fluorescence was monitored as described in Steps 7–9. The whole-cell fluorescence of cells from 1-ml cultures was plotted versus the whole-cell fluorescence of a 1-ml sample from the 1-l cultures. MW, molecular weight.


MATERIALS

Reagents
1,4-dithiothreitol (DTT; Sigma)
Buffer A: phosphate-buffered saline (PBS) with 0.1% n-dodecyl-beta-D-maltopyranoside (DDM; or other detergent at 5times critical micellar concentration; see Supplementary Table 1 online)
Buffer B: Buffer A with 500 mM imidazole
Deoxyribonuclease I from bovine pancreas Type IV lyophilized powder (Sigma)
E. coli BL21(DE3)–derived host strains (see Supplementary Table 2 online)
Ethylenediaminetetraacetic acid (EDTA; Sigma)
Purified GFP (Supplementary Methods online)
Solubilization buffer (SB): 200 mM Tris-HCl (pH 8.8), 20% Glycerol, 5 mM EDTA (pH 8.0), 0.02% bromphenol blue, make aliquots of 700 mul and keep at -20 °C. Before use, add 200 mul 20% SDS and 100 mul 0.5 M DDT
Tobacco etch virus (TEV) protease, His-tagged (see Supplementary Data online)

Equipment
1.5-ml polyallomer microcentrifuge tubes (Beckman)
腒TAprime or higher 膋ta system (GE Healthcare)
Beckman TLA100 bench-top ultracentrifuge equipped with Beckman TLA100 rotor
Centricon Centrifugal Filter Unit (Millipore); cutoff 30,000, 50,000 and 100,000 nominal molecular weight limit (NMWL) depending on size of protein and detergent
LAS-1000 charge-coupled device (CCD) camera system (Fujifilm)
Nunc 96-well optical bottom plate, black (Nunc)
Poly-Prep chromatography columns (Bio-Rad)
Shaking incubator with temperature control
SpectraMax Gemini EM microplate spectrofluorometer (Molecular Devices)
Superdex 200 10/300 GL Tricorn gel filtration column (GE Healthcare)
Thermomixer comfort (Eppendorf) equipped with thermoblocks for 2.0-ml or 1.5-ml microcentrifuge tubes
Tunair 2.5-liter baffled shaker flasks
Ultracentrifuge, for example Beckman Coulter Optima LE-80k equipped with Beckman Ti 70.1 rotor
Ultraviolet-visible (UV-Vis) spectrophotometer, for example UV-1601 (Shimadzu)
XK 16/20 column (GE Healthcare) or larger column
Additional reagents are listed in Supplementary Methods.

PROCEDURE
Construction of genes encoding membrane protein–GFP fusions
1. Before cloning the genes encoding the membrane proteins into the GFP-fusion vector, verify that the C termini of the membrane proteins are in the cytoplasm (Cin topology; Fig. 1).
In E. coli, the GFP moiety of a membrane protein–GFP fusion is fluorescent only if the fusion is integrated into the cytoplasmic membrane (that is, inclusion bodies are not fluorescent1, 3, 4) and has a Cin topology2. If the location of the C terminus of the membrane protein to be overexpressed is unknown, predict its topology using, for example, the online application TMHMM (http://www.cbs.dtu.dk/services/TMHMM). Approximately 80% of all helical membrane proteins have a cytosolic C terminus5, 6, 7.

2. For each membrane protein, clone the gene of interest into a standard pET28a(+)-derived GFP-8His fusion vector1, 8 that harbors a TEV protease recognition site for removal of the GFP-8His moiety during purification (see Supplementary Fig. 1 online). Note that we also use the abbreviation TEV to indicate the TEV protease recognition site between the membrane protein and GFP moiety; membrane protein–TEV-GFP-8His.
A library covering the vast majority of E. coli membrane proteins fused to GFP is available7.

Determining the overexpression potential of membrane protein–GFP fusions
3. Transform the expression vector encoding a membrane protein–GFP fusion into BL21(DE3)pLysS cells (see Supplementary Methods). Use a fresh colony of the transformed strain to set up an overnight culture in a 2-ml standard microcentrifuge tube containing 1 ml LB medium with 50 mug/ml kanamycin and 34 mug/ml chloramphenicol. Also set up a culture to be used as control in Step 9 to measure background fluorescence. This control can be a culture harboring the expression vector that will be uninduced in Step 6 or a culture harboring an 'empty' expression vector.
To target the most promising candidates, overexpression of several membrane proteins as GFP fusions can be tested simultaneously (Steps 3–13; Fig. 3a). Alternatively, 24-deep-well microtiter plates can be used for cultures in Steps 3–6, although their handling is more cumbersome.
Figure 3. Examples of method application.

Figure 3 thumbnail

(a) Seven membrane protein–GFP fusions were screened for overexpression in the strain BL21(DE3)pLysS by means of whole-cell and in-gel fluorescence. Before loading onto gel the cell suspensions were twofold concentrated for hKDELr and twofold diluted for YciS. As controls, purified GFP-8His and a sample of uninduced cells (YciS, fivefold concentrated) were also loaded (U). (b) Whole-cell fluorescence from cells overexpressing YbaT-TEV-GFP-8His was monitored after 4 and 22 h of expression. Colored bars represent the different strains used: C41(DE3), green; C43(DE3), red; and BL21(DE3)pLysS, blue. IPTG induction: 0.1 mM (odd numbers) or 0.4 mM (even numbers). Error bars represent minima and maxima from three independent experiments carried out in duplicate. (c) Protein was detected by in-gel fluorescence on a 12% SDS gel. Numbers correspond to the numbers in b. (d) Screening different detergents for their efficiency to solubilize YciS-TEV-GFP-8His–containing membranes. Error bars represent minima and maxima from three independent screens. Inset, in-gel fluorescence of solubilized material. LDAO, lauryldimethylamine oxide; CHAPS, 3-[(3-cholamidopropyl)dimethylammonio]-1-propanesulfonic acid. (e) Purification of YedZ-TEV-GFP-8His fusion and recovery of YedZ from the fusion as analyzed after 12% SDS-PAGE by in-gel fluorescence (left) and Coomassie staining (right). The lanes were loaded as follows: 1, YedZ-TEV-GFP-8His–containing membranes (12 mug); 2, non–detergent-solubilized protein (12 mug); 3, solubilized protein (12 mug); 4, IMAC flowthrough (4 mug); 5, YedZ-TEV-GFP-8His eluate from IMAC (2 mug); 6, purifed YedZ-TEV-GFP-8His TEV digest (2.5 mug; note that GFP-8His and YedZ are not separated by this percentage of SDS-gel); 7, YedZ after gel filtration and removal of His-tagged TEV protease and GFP-8His by batch-binding to Ni-NTA resin; 8, GFP-8His (0.5 mug); 9, His-tagged TEV (0.5 mug). MW, molecular weight.


4. Incubate the culture overnight in a thermomixer at 37 °C at 900 r.p.m.

5. Dilute the overnight culture 50-fold into two 2-ml standard microcentrifuge tubes, each containing 1 ml LB medium with 50 mug/ml kanamycin and 34 mug/ml chloramphenicol.

6. Incubate the cultures in a thermomixer at 900 r.p.m. at 37 °C and designate one of the tubes to monitor the optical density (OD)600 of the culture. At an OD600 of 0.4–0.5 (after approx2 h) lower the temperature to 25 °C and induce expression of membrane protein–GFP fusion in the remaining tube with isopropyl beta-D-thiogalactopyranoside (IPTG; 0.4 mM final concentration).
Critical step

7. Four hours after induction, centrifuge the cells at 15,700g for 2 min in a benchtop centrifuge, remove the supernatant carefully and resuspend the cell pellet in 200 mul of PBS.

8. Transfer 100 mul of cell suspension to a black Nunc 96-well optical-bottom plate. (Set aside the remaining 100 mul of cell suspension for monitoring in-gel fluorescence in Steps 10–13).

9. Measure GFP fluorescence emission at 512 nm and excitation at 485 nm in a microplate spectrofluorometer. Select the option 'bottom read' for maximal sensitivity. Estimate membrane protein overexpression levels (in mg/l; see Supplementary Methods and Supplementary Fig. 2 online).
To assess background fluorescence levels in the system used, measure whole-cell fluorescence of an uninduced sample or cells containing an empty expression vector. Using the microplate spectrofluorometer described here, fluorescence counts for cells grown to an OD600 of 1.5 are twice that of PBS.
Troubleshooting

10. Centrifuge the 100-mul cell suspension (set aside in Step 8) in a bench-top centrifuge at 15,700g for 2 min and remove the supernatant carefully.
Pause Point The cell pellets can be stored at -20 °C for several days.

11. Based on the whole-cell fluorescence measurement (Step 9), resuspend the pellets in a volume of PBS to give a GFP fluorescence level equal to that of 5–10 ng/mul of purified GFP-8His (see Supplementary Methods). Add 10 mul of SB to 10 mul of each cell suspension and to 10 mul of purified GFP at a concentration of 5–10 ng/mul. Incubate the samples at 37 °C for 5 min.
Whole-cell fluorescence in the cell suspensions should be adjusted to roughly similar levels to ensure that in SDS-PAGE (Step 12) weak bands (that is, in case of degradation) can be detected without interference of a much stronger signal in the neighboring lanes. The most important advantage of using in-gel fluorescence is to verify that full-length protein is present. Quantification is also possible, but measuring fluorescence in solution is less time-consuming.
Critical step

12. Analyze the samples from Step 11 by SDS-PAGE; include a molecular weight marker.

13. Rinse the gel with distilled water. To detect the fluorescent bands, expose the gel to ultraviolet light and capture images with a CCD camera system (Fig. 3a). Increase exposure time to desired band intensity.
Fluorescence intensities can be quantified using Image Gauge V 3.45 software or comparable software (Fig. 2a,b). If desired, the gel can be subsequently stained with Coomassie.
Troubleshooting

Optimization of overexpression of membrane protein–GFP fusions
14. Transform the expression vector encoding a membrane protein–GFP fusion selected from previous screen into BL21(DE3)pLysS, C41(DE3) and C43(DE3) strains (see Supplementary Methods and Supplementary Table 2).
These three strains give consistently good membrane protein overexpression yields in our laboratory. The strain that gives the best results for a particular membrane protein, however, must be determined empirically.

15. Set up overnight cultures using fresh transformants in standard 2-ml microcentrifuge tubes containing 1 ml of LB medium with appropriate antibiotic(s).
Critical step

16. Incubate overnight cultures in a thermomixer at 37 °C at 900 r.p.m.

17. Dilute overnight cultures 75-fold into four 50-ml Falcon tubes per strain, each tube containing 15 ml of LB medium with appropriate antibiotic(s), and incubate at 37 °C at 220 r.p.m.

18. Monitor the OD600 of the cultures, and upon reaching an OD600 of 0.25–0.35 (after approx2 h) shift the incubation temperature for two cultures to 30 °C and, for the other two cultures, to 25 °C.
Critical step

19. Grow the cultures at the lower temperatures for 30 min; then induce expression of the membrane protein–GFP fusion by adding IPTG. For each set of duplicate cultures grown at 30 °C, add IPTG to one culture to a final concentration of 0.1 mM, and, to the other culture, to a final concentration of 0.4 mM. Similarly, for each set of duplicate cultures grown at 25 °C, add IPTG to one culture to a final concentration of 0.1 mM, and to the other, to a final concentration of 0.4 mM. There are now 12 different conditions represented (host strain, temperature shift and IPTG concentration) as summarized below.


20. Grow the strains in the presence of IPTG for 4 h, then remove 1 ml of culture for whole-cell fluorescence measurements (see Steps 7–9). Incubate the remaining cultures overnight and repeat the whole-cell fluorescence measurements after approx22 h (Fig. 3b).
OD600 and in-gel fluorescence can be monitored as well (Fig. 3c). The optimization screen is done in 50-ml Falcon tubes instead of 2-ml tubes to provide enough volume for two measurements (at 4 h and 22 h). Furthermore, optimizing overexpression in 2-ml microcentrifuge tubes is unreliable for the overnight estimates because of oxygen depletion.
Troubleshooting

Isolation of membranes
21. Select the strain that gives the best overexpression as established by the overexpression optimization screen (Steps 14–20), and set up an overnight culture in a 200-ml shaker flask containing 20 ml LB medium with appropriate antibiotic(s) (see Supplementary Table 2).

22. Transfer the overnight culture into 1 l of LB medium with appropriate antibiotic(s) in a 2.5-liter baffled shaker flask. Incubate the culture at 37 °C at 220 r.p.m. and use the parameters established in the overexpression optimization screen to overexpress the membrane protein–GFP fusion (Fig. 2c). Before collecting the cells, remove a 1-ml sample for measuring whole-cell fluorescence.
OD600 and in-gel fluorescence can be monitored as well. The volume of the overnight culture in an appropriate shaker flask depends on the number of 1-l cultures to be inoculated (use 20 ml per liter).

23. Collect the cells by centrifugation at 6,200g at 4 °C for 15 min. Decant the supernatant and resuspend the cell pellet in 500 ml ice-cold PBS.
From this step on, even when not indicated, material should be kept on ice or at 4 °C.

24. Centrifuge the resuspended cells at 6,200g at 4 °C for 15 min and decant supernatant. Resuspend the cell pellet in 10 ml ice-cold PBS.
Pause Point Cell suspensions can be rapidly frozen in liquid nitrogen and stored at -80 °C for up to 6 months. Use screw-capped tubes for storage.

25. Add Pefabloc SC (1 mg/ml final concentration), DNase (20–100 U/ml final concentration) and MgCl2 (1 mM final concentration) and break the cells with a French press at 10,000 p.s.i. for at least two passes at 4 °C. Most cells are broken when the suspension has turned from turbid to transparent.
Alternatively, other methods of cell disruption can be applied, such as sonication in combination with EDTA-lysozyme treatment, homogenization and cell disruption using disruptors from Constant Systems.

26. Remove the unbroken cells and debris by centrifugation at 24,000g at 4 °C for 12 min and collect the supernatant containing the membranes. Repeat this centrifugation step to clear the supernatant of any residual cells and debris.

27. To collect the membranes, centrifuge the cleared supernatant at 150,000g at 4 °C for 45 min. Remove the supernatant and resuspend the pellet in 10 ml ice-cold PBS using a disposable 10-ml syringe with a 21-gauge needle. Fill centrifugation tubes with ice-cold PBS to avoid collapsing of tubes during ultracentrifugation in Step 28.

28. Collect the membranes by repeating centrifugation at 150,000g at 4 °C for 45 min. Resuspend the pellet-containing membranes in 5 ml ice-cold PBS as described in Step 27, and measure total amount of protein in the membrane suspension using the BCA (bicinchoninic acid) protein assay kit.
If any EDTA was used in Step 25, it will be washed away and will not interfere with immobilized metal ion affinity chromatography (IMAC) in Step 36.
Pause Point Membrane suspensions can be rapidly frozen in liquid nitrogen and stored at -80 °C for up to 6 months. Note, however, that some membrane protein crystallographers avoid freezing and storing membranes and continue with purification immediately.
Troubleshooting

Detergent screen
29. Adjust the membrane suspension to a protein concentration of 3.75 mg/ml. Transfer 800-mul aliquots of the suspension into 1.5-ml polyallomer microcentrifuge tubes.

30. Select a range of different types of detergents to test for membrane solubilization (see Supplementary Table 1). Add 200 mul of a selected detergent in PBS to each of the 1.5-ml tubes containing 800 mul of membrane suspension. See Supplementary Table 1 for the final percentage of each detergent (the final protein concentration is 3 mg/ml). Incubate the mixtures at 4 °C for 1 h with mild agitation.

31. Centrifuge the nonsolubilized material in a bench-top ultracentrifuge at 100,000g at 4 °C for 45 min. Collect the supernatant and measure GFP fluorescence in 100 mul of the supernatant containing the solubilized membrane protein to estimate the solubilization yields (see Supplementary Methods and Fig. 3d).
GFP fluorescence changes maximally plusminus3% in the presence of the detergents tested. The integrity of extracted membrane protein–GFP fusions can be analyzed with the in-gel fluorescence assay as described in Steps 11–13 (Fig. 3d). The percentage of detergent solubilization can be estimated by comparing the fluorescence in the detergent-solubilized membranes to that of the fluorescence left in the nonsolubilized pelleted membranes resuspended in the same volume of buffer.
Troubleshooting

32. Determine the optimal protein:detergent ratio by repeating Steps 29–31 with the most effective detergent (as established in Step 31), at a constant percentage with increasing amounts of protein (that is, 3–10 mg/ml protein).
Establishing the point at which an increase in protein still yields a linear increase in GFP fluorescence (optimal protein:detergent ratio) is important for enriching the solubilized membranes with the membrane protein–GFP fusion.

Purification of membrane protein–GFP fusions
33. Using the protein:detergent ratio established in Step 32, solubilize the membranes for purification by incubating the membrane-detergent mixture at 4 °C for 1 h with mild agitation.

34. Remove the unsolubilized material by centrifugation at 100,000g at 4 °C for 45 min. Remove a 200-mul sample of the supernatant and measure fluorescence as described in Step 31. Set aside the remaining 100 mul for subsequent analysis of the purification by SDS-PAGE as described in Step 43 (Fig. 3e).

35. Pack an XK 16/20 column, using approx1 ml of Ni-NTA resin per milligram of membrane protein–GFP fusion to be purified, and equilibrate the Ni-NTA column with five column volumes of Buffer A.

36. Add imidazole (10 mM final concentration) to the solubilized membranes (supernatant from Step 34) and load onto the Ni-NTA column at a slow flow rate (0.3–0.5 ml/min).

37. Wash the column with approx20 column volumes of 4% Buffer B at a flow rate of 1 ml/min.

38. Deliver a gradient of 4–25% Buffer B over 20 column volumes at a flow rate of 1 ml/min and collect fractions.
The fraction volume to be collected is proportional to the size of the column; for example, for a 5-ml column usually 1-ml fractions are collected. Once the wash and elution conditions have been established, step gradients can be used instead of continuous gradients: wash the column with 20 column volumes at 2% less than the highest percentage of Buffer B at which protein was still bound to the column.

39. Elute the fusion protein with 50% Buffer B at a flow rate of 1 ml/min and collect all fractions. Save 100-mul samples from the flowthrough, wash and elution fractions.

40. Measure GFP emission (see Steps 8 and 9) in the different fractions and determine the amount of membrane protein–GFP fusion (see Supplementary Methods). Estimate any losses in each step (for example, in flowthrough).
The amount of fusion in the eluate should be determined by measuring the GFP fluorescence, as the BCA assay measures total protein (including contamination) and is affected by cross-reactivity with imidazole at concentrations >50 mM.
Troubleshooting

Removal of GFP moiety from the membrane protein fusion
41. Add equimolar His-tagged tobacco etch virus (TEV) protease to membrane protein–GFP fusion and adjust DTT and EDTA to a final concentration of 1 and 5 mM, respectively and incubate at 4 °C for 10 h or overnight (see Supplementary Data).
For commonly used detergents, such as n-dodecyl-beta-D-maltopyranoside (DDM) and Triton X-100, equimolar amounts of TEV protease typically suffice for a complete overnight digest at 4 °C. Using small amounts of membrane protein–GFP fusion and TEV or any other site-specific protease, optimal cleavage conditions can be identified rapidly with in-gel fluorescence.

42. Measure total protein concentrations in the different fractions with the BCA assay for SDS-PAGE analysis (Step 43).

43. Analyze the solubilzed membranes (Step 34), IMAC flowthrough (Step 36), wash fractions (Steps 37–38), eluate (Step 39) and TEV-digest reactions (for completeness of the digest; Step 41) by SDS-PAGE, adding an appropriate amount of protein in a 10-mul volume to 10 mul of SB. Process as described in Steps 12–13.
Troubleshooting

44. If digest is complete (from Step 41), concentrate it in Centricon concentrators (the cutoff used depends on the size of the protein).

45. Separate the proteins by standard gel filtration using a Superdex 200 10/30 column. Remove 100 mul from each of the (different) protein absorbance peaks and process as described in Steps 8–9 to establish which fractions contain GFP-8His.
Note that GFP-8His and His-tagged TEV have a similar retention time.

46. If the membrane protein and GFP-8His are in the same fractions, remove GFP-8His and His-tagged TEV by adding Ni-NTA resin equilibrated in the buffer used for gel filtration. Use approx1 ml of resin per 5 mg of total protein.

47. Transfer the mixture to an empty Poly-Prep chromatography column and collect the flowthrough (membrane protein). Wash with 3 column volumes to recover membrane protein remaining in the dead volume. Process the different fractions as in Steps 12–13 (Fig. 3e).
Troubleshooting

TROUBLESHOOTING
Problem: Expression yields are less than 200 mug per liter of culture.
[Step 9]
Solution: Improve signal to noise ratio by increasing the amount of cells analyzed; set up 5-ml cultures and resuspend the cell pellet in 100 mul of PBS for fluorescence measurements.

Problem: There is severe proteolysis of membrane protein–GFP fusion.
[Step 13]
Solution: Induce expression for 1–2 h at 37 °C or express overnight at low temperature (20 °C) (Step 6).

Problem: Expression yields are low.
[Step 20]
Solution: Try expression at different temperatures (20 °C and 37 °C), in different media (for example, Terrific broth, 2times YT, minimal medium), in different strains (for example, BL21-CodonPlus(DE3))9 or change to homologs of the protein.

Problem: The yield is less than 60 mg of total protein per liter (indicative of poor cell breakage).
[Step 28]
Solution: Add EDTA (1 mM final concentration) and lysozyme (0.5 mg/ml final concentration) to the cell suspension and incubate for 15–30 min on ice before breaking the cells. If cells are treated with EDTA and lysozyme before disruption, 2 mM MgCl2 rather than 1 mM MgCl2 should be used in Step 25.

Problem: There is severe proteolysis of membrane protein–GFP fusion.
[Step 31]
Solution: Use commercially available protease inhibitor cocktails rather than Pefabloc SC only in Step 25. Add ligand to increase stability of protein.

Problem: Not all of the protein binds to the column as evident from the presence of GFP fluorescence in the flowthrough (see Step 43).
[Step 40]
Solution: Increase the bed volume and, if necessary, use a column of a larger diameter (Steps 35–39).

Problem: There is degradation of the purified fusion.
[Step 43]
Solution: See Troubleshooting, Step 31. Do not freeze cells or membranes.

Problem: There is nonspecific binding of membrane protein to Ni-NTA resin.
[Step 47]
Solution: Add imidazole to a final concentration of 5–10 mM to buffer used in batch binding (Step 46).

  Top

CRITICAL STEPS
Determining the overexpression potential of membrane protein–GFP fusions, Step 3 Always use freshly transformed BL21(DE3)pLysS cells (that is, not older than 2–3 d) and medium with freshly added antibiotics. Do not use glycerol stocks of transformed BL21(DE3)pLysS cells as starting material because this can cause significant losses in expression levels10.

Determining the overexpression potential of membrane protein–GFP fusions, Step 6 Because OD600 measurements are highly dependent on the photospectrometer used, it is strongly recommended that the same instrument is used for the different experiments. Diluting samples to OD600 values less than 0.3 increases accuracy of OD600 measurements substantially.

Determining the overexpression potential of membrane protein–GFP fusions, Step 11 Samples should be heated to 37 °C rather than 95 °C. Heating membrane proteins at 95 °C often causes aggregation, and GFP loses fluorescence after incubation at 95 °C. If frozen cells are used for the in-gel fluorescence assay, add MgCl2 (1 mM final concentration) and DNase (1–5 U per 10 mul of cell suspension) and incubate for 15 min on ice before adding SB.

Optimization of overexpression of membrane protein–GFP fusions, Step 15 Always use freshly transformed BL21(DE3)pLysS cells (that is, not older than 2–3 d) and medium with freshly added antibiotics. Do not use glycerol stocks of transformed BL21(DE3)pLysS cells as starting material because this can cause significant losses in expression levels10.

Optimization of overexpression of membrane protein–GFP fusions, Step 18 For C41(DE3) and C43(DE3) the OD600 at induction is crucial; significantly decreased overexpression yields have been observed if cells continued to grow at 37 °C to cell densities higher than 0.5 before induction. Expression in BL21(DE3)pLysS, however, is less sensitive to variations in the OD600 of induction (note the smaller error bars in BL21(DE3)pLysS compared to C41(DE3) and C43(DE3) in Fig. 3b), making this strain the vehicle of choice for initial overexpression screening (Steps 3–13).

Comments
The GFP moiety of a membrane protein–GFP fusion can be used for direct, rapid and quantitative detection of membrane proteins during overexpression and isolation. In this protocol the E. coli BL21(DE3)-pET system is used as a platform to illustrate the method. It should be noted that the protocol can be easily modified and extended: for example, different strains, expression vectors, culture media, buffer systems for protein isolation or other site-specific proteases can be used. An important prerequisite for using this method in E. coli is that the membrane protein should have a cytosolic C terminus, as GFP can only fold and become fluorescent in the cytoplasm2, 11. Fortunately, approx80% of the multispanning membrane proteins have a cytosolic C terminus5, 6, 7. If expression hosts other than E. coli (for example, Lactococcus lactis3, Saccharomyces cerevisiae or Pichia pastoris12) are used for the overexpression and isolation of membrane protein–GFP fusions, the GFP variant most suited for that particular host should be used.

Thus far, it has not been possible to predict how well a membrane protein can be overexpressed7, making the development of methods with which membrane protein overexpression can be rapidly and accurately monitored a bare necessity to improve the throughput of membrane protein research. The GFP moiety is not only a time saver: the sensitivity and accuracy with which it can be monitored both in solution and in gels makes it a superior alternative to the notoriously unreliable and time-consuming immunoblotting of membrane proteins.

Functional membrane proteins can be recovered easily from membrane protein-GFP fusions by cleavage with a site-specific protease3 (Fig. 3e). Finally, it should be noted that the GFP-based methodology described in this protocol is also applicable to soluble proteins fused to GFP.

  Top
Example of application
We expressed seven different membrane proteins as GFP fusions in the E. coli strain BL21(DE3)pLysS, and monitored their expression and integrity using a combination of whole-cell and in-gel fluorescence (Steps 3–13; Fig. 3a). All proteins are E. coli cytoplasmic membrane proteins unless noted otherwise: YedZ (putative integral flavocytochrome3), YciS (unknown function), YbaT (putative amino acid transporter3), ProW (component of high-affinity transport system for glycine, betaine and proline), hKDELr (human KDEL endoplasmic reticulum protein retention receptor), GlpT (glycerol-3-phosphate transporter) and AmpG (involved in peptidoglycan recycling). Note that membrane proteins usually run faster on SDS-PAGE than expected, typically at approx70–85% of their expected molecular weight. In the experimental setup used, GFP runs at a molecular weight of 22 kDa rather than 28 kDa because it remains properly folded.

As determined by in-gel fluorescence, GlpT is a membrane protein that is not stably overexpressed; the full-length protein (band marked with an asterisk) is only a minor component. This is in keeping with published observations that GlpT has to be overexpressed in the presence of its substrate to prevent proteolytic cleavage13.

We were able to improve the overexpression yields of the membrane protein YbaT more than 10-fold using an experience-based matrix screen (varying parameters: temperature, IPTG concentration, duration of overexpression, host strain) as outlined in Steps 14–20 (Fig. 3b). To detect degradation of overexpressed material, we loaded protein corresponding to equal amounts of fluorescence and analyzed the in-gel fluorescence from each culture condition by SDS-PAGE, and subsequently exposed the gel to ultraviolet light for varying amounts of time. To calculate the amount of degradation, we also loaded two different amounts of GFP-8His standard onto the same gel (in this example, <5% of total fusion was degraded). It should be stressed that the overexpression screen shown in Figure 3b,c is an example. It is our experience that the behavior of a membrane protein in the screen cannot be predicted, that is, the &#39;optimal conditions&#39; for overexpression of a particular membrane protein must be established experimentally.

Once established, production of the membrane protein–GFP fusion can be reliably scaled up (Steps 21–22 and Fig. 2c), and membranes can be isolated (Steps 23–29). The fluorescence from the GFP moiety can conveniently be used for screening the solubilization efficiency of the membrane protein–GFP fusion into different detergents as outlined in Steps 30–32. A detergent screen for YciS-TEV-GFP-8His is illustrated in Figure 3d. After spinning down the nonsolubilized material, we calculated the solubilization efficiency (Step 31); in this example, LDAO is 90% efficient (once the detergent is selected, the best protein:detergent ratio can also be determined as described in Step 32). We assessed the integrity of the detergent-solubilized material by measuring in-gel fluorescence (Fig. 3d).

Detection of in-solution and in-gel fluorescence as well as visual detection is convenient during purification of membrane protein–GFP fusions. Analysis of the purification of YedZ-TEV-GFP-8His and the recovery of YedZ from the fusion by SDS-PAGE combined with in-gel fluorescence and Coomassie staining is illustrated in Figure 3e. We solubilized membranes containing YedZ-TEV-GFP-8His fusion with 1% DDM, at a protein concentration of 8 mg/ml with a solubilization efficiency of 72%. For IMAC, we washed the column with 10 column volumes of 10% Buffer B and eluted the fusion in a step gradient of 50% Buffer B (we recovered 90% of starting material). We cleaved off the GFP-8His moiety by overnight incubation with an equimolar amount of His-tagged TEV protease at 4 °C and further purified the protein by gel filtration. As the TEV-protease and the clipped-off GFP are His-tagged, any carryover contamination can be simply removed by batch-binding isolated membrane protein to Ni-NTA resin (see Steps 46–47).

At every stage during the overexpression and isolation of a membrane protein–GFP fusion it is possible to calculate the amount of membrane protein by measuring the GFP fluorescence and comparing it to a purified GFP-8His standard (Supplementary Methods).