A major barrier to the physical characterization and structure determination of membrane proteins is low yield in recombinant expression. To address this problem, we have designed a selection strategy to isolate mutant strains of Escherichia coli that improve the expression of a targeted membrane protein. In this method, the coding sequence of the membrane protein of interest is fused to a C-terminal selectable marker, so that the production of the selectable marker and survival on selective media is linked to expression of the targeted membrane protein. Thus, mutant strains with improved expression properties can be directly selected. We also introduce a rapid method for curing isolated strains of the plasmids used during the selection process, in which the plasmids are removed by in vivo digestion with the homing endonuclease I-CreI. We tested this selection system on a rhomboid family protein from Mycobacterium tuberculosis (Rv1337) and were able to isolate mutants, which we call EXP strains, with up to 75-fold increased expression. The EXP strains also improve the expression of other membrane proteins that were not the target of selection, in one case roughly 90-fold.
A primary barrier to the biochemical, physical, and structural characterization of many membrane proteins is the inability to obtain sufficient quantities by recombinant expression. This problem is illustrated in a study by Korepanova et al. wherein the Mycobacterium tuberculosis (MTb) α-helical membrane proteome was expression-tested in various Escherichia coli strains.1 Out of the 105 membrane protein targets tested, only 37 were over-expressed sufficiently to be detectable by Coomassie staining. Only nine of those 37, all less than 16 kD, were expressed to the membrane fraction and presumably correctly folded. Thus, the standard expression techniques used in this study resulted in high-level expression to the membrane for only 10% of the targets, a success rate that is not unusual for membrane proteins.2 It is unlikely that all membrane proteins can be overexpressed in E. coli, but because it is such a convenient host it is desirable to increase the success rate of E. coli expression.
There are myriad reasons why expression could fail. Potential problems common to all heterologous protein expression include codon usage, mRNA or protein stability, and cell physiology changes induced by the stress of recombinant expression.3 With membrane protein production, there are numerous additional possible failure points because insertion requires proper targeting of the nascent polypeptide chain and intimate interactions with the insertion machinery.4, 5 These complexities are well illustrated by an impressive proteomics study of the effects of membrane protein expression in E. coli by Wagner et al.6 Among many changes, they found alterations in chaperone machinery, evidence of energy stress, and impairment of native membrane protein as well as secretory protein expression. Thus, membrane protein expression can have dramatic consequences for many aspects of cell physiology. It remains unclear, however, what the critical factors are that limit proper expression and insertion, and the key barriers may be different for each membrane protein.
Current approaches to improving expression that are generally employed in an ad hoc fashion include altering growth media, temperature, or induction levels. Wagner et al. found that membrane protein expression could be improved considerably by fine tuning induction levels.7 In addition, fusion to other proteins can help expression, and fusion to Mistic appears to be particularly helpful and is thought to help chaperone the protein into the membrane.8 Also, the membrane protein itself can be mutated to generate more stable variants that express at higher levels, but the disadvantage is that the product may not reflect the native structure and function.9–11 Another approach is cell-free expression which can bypass deleterious changes in cell physiology.12 The produced protein, however, is not necessarily in a folded, functional form, and this is particularly troublesome for membrane proteins, which tend to be difficult to refold. Although these techniques can offer significant improvement in isolated cases, there have been no universal solutions and membrane protein production continues to be a considerable obstacle that must be addressed.
Because of the many possible points in membrane protein biogenesis that could go awry and prevent expression, a rational, hypothesis-driven approach to the problem could be difficult. An alternative approach is to select for genomic mutations that improve expression, obviating the need for a precise understanding of the barriers to protein production. Along these lines, Miroux and Walker isolated strains of E. coli that were resistant to the toxicity of membrane protein expression.13 The isolated strains, called C41 and C43, improve expression of many membrane proteins expressed from the T7 promoter and are now routinely used in expression screening. These mutant strains were recently shown to act by slowing expression from the strong T7 promoter.7 These results clearly indicate that it is possible to reengineer E. coli with an improved ability to express membrane proteins.
The techniques used to isolate C41 and C43 are limiting, however, because not all poorly expressed membrane proteins are toxic. Moreover, toxicity can be eliminated in ways that do not improve expression, and indeed, one way to prevent toxicity is to not produce the membrane protein at all. Thus, it would be useful to develop a selection that is more directly linked to membrane protein expression.
Recently, G-protein coupled receptor (GPCR) expression in E. coli was greatly improved using a rapid screening method.14, 15 The membrane protein target was fused to a C-terminal GFP marker, so that improved hosts could be directly identified by monitoring GFP fluorescence. Link et al. screened a panel of 10 genes for the ability to improve expression and found that overexpression of the protease FtsH enhanced expression of the GPCR targets tested.14 This group also used fluorescence-activated cell sorting to screen an E. coli transposon insertion library for variants expressing high levels of GPCR target fused to GFP, and found that deletion of the chaperone DnaJ resulted in increased expression.15 These studies utilize a more direct method to isolate improved strains and also lend further evidence that it is possible to engineer expression hosts for improved production of difficult to express membrane protein targets.
Here we introduce a simple, effective, and efficient strategy to select for mutants of E. coli that improve membrane protein expression, and we test it on a poorly expressed rhomboid family protein from MTb. The EXP strains isolated using this method improve expression of the target membrane protein up to 75-fold.
The selection system
To select for expression we fuse the targeted membrane protein to a C-terminal selectable marker that confers a drug resistance phenotype, so that as long as the C-terminus is in the cytoplasm, growth on selective media indicates expression of the target membrane protein. As there can be many ways for mutations to provide drug resistance that have nothing to do with expression of the fused membrane protein, we employ a dual selection strategy in which the same membrane protein target is fused to one of the two drug resistance markers on two separate plasmids. The probability of obtaining mutations that confer resistance to both drugs without increasing membrane protein expression should be extremely low. The two selection plasmids, named pSEL1 and pSEL2, are illustrated in Figure 1(A).
We initially tested the system by comparing growth on selective media of cells producing a well expressed membrane protein (GlpF) and those expressing a poorly expressed membrane protein (SPP). As shown in Figure 2, cells harboring GlpF in pSEL1 and pSEL2 and those harboring SPP in pSEL1 and pSEL2 both survived on inducing media without drugs, indicating that induction of neither protein is lethal to cells. In the presence of selecting drugs without induction, none of the constructs allowed survival. Under inducing conditions and in the presence of selection, only cells expressing the GlpF fusions survived. Thus, our selection system effectively and cleanly discriminated between cells expressing high levels of target membrane protein and those expressing little to no protein.
Curing of selected mutants
After mutant selection, it is necessary to remove the selection system plasmids, pSEL1 and pSEL2, from the strains. We found, however, that traditional curing methods were highly inefficient when applied to our strains and plasmids. We therefore developed a rapid and efficient curing method for this work.
In our curing method, the plasmids used during selection are eliminated by in vivo digestion with a rare-cutting endonuclease, the homing endonuclease I-CreI.16 As shown in Figure 1(A), the recognition site for I-CreI was introduced into pSEL1 and pSEL2. To remove the selection plasmids, we introduce a third plasmid called pCURE, which encodes I-CreI endonuclease and contains a temperature sensitive origin of replication [Fig. 1(B)]. The I-CreI expressed from pCURE digests the two pSEL plasmids, and pCURE is subsequently removed by growth at an elevated temperature. Using this method strains can reliably be cured in only 2 days.
Selection of strains that improve rhomboid-Rv1337 expression
With an effective selection system and a highly efficient curing system, we tested our ability to isolate E. coli mutants that improve membrane protein expression. We targeted the MTb alpha-helical inner membrane protein Rv1337, a rhomboid family protein, because it is a relatively large protein known from prior work to be expressed at low levels detectable by western blotting. In addition, rhomboid-Rv1337 has a cytoplasmic C-terminus, which is necessary for selection with the C-terminal selectable marker fusions used in pSEL1 and pSEL2. Expression of Rv1337 appears to be toxic to cells in that cell growth is dramatically reduced after induction.
Selection was performed in two steps. First, TOP10 cells harboring pSEL1 encoding rhomboid-Rv1337 were mutagenized with either the base analog 2-aminopurine (2AP) or the mutator gene mutD5, and colonies were selected for their ability to grow on media containing the drug trimethoprim. In the second step, the trimethoprim-resistant colonies were pooled and transformed with the pSEL2 construct encoding rhomboid-Rv1337. The cells, now harboring two plasmid constructs, were then selected on media containing both trimethoprim and kanamycin. The advantage of this two step procedure is that a vast number of mutagenized cells can be plated in the first step without losing mutants because of the low efficiency of plasmid transformation. The enriched pool can then be selected in the second step to isolate cells truly surviving due to expression of the target membrane protein. About 1 in 10,000 mutagenized cells survived the first trimethoprim selection (100–200 μg/mL), whereas roughly 1 in 1000 of the trimethoprim-selected colonies survived the second step and were able to grow on both kanamycin (10–80 μg/mL) and trimethoprim (100–200 μg/mL).
We screened 47 selected colonies, and based on western blotting, 17 demonstrated increased expression of rhomboid-Rv1337. We chose five clones, all from independently mutagenized cultures, that showed the greatest increase in protein production, and we cured them of the selection plasmids using the method described earlier. We refer to these strains as EXP-Rv1337-1, EXP-Rv1337-2, EXP-Rv1337-3, EXP-Rv1337-4, and EXP-Rv1337-5.
To validate the expression results, we retransformed the cured mutants with pSEL1 and pSEL2 encoding rhomboid-Rv1337. The increase in expression in the five selected, cured, and retransformed mutants is shown in Figure 3. Qualitative examination of the western blot shown in Figure 3(A) shows a clear enhancement of targeted membrane protein production. This increase was quantified by comparison with the intensities of known amounts of an added protein standard to the lanes. As shown in Figure 3(B), the selected strains improved the expression of the rhomboid-Rv1337 fusions up to 75-fold.
Expression without a fusion to a selectable marker
As our general selection system requires the attachment of the target protein to a fusion partner, we wanted to test whether the mutations were effective when the protein was expressed without a marker protein attached. Figure 4 shows expression of rhomboid-Rv1337 either with or without fusions to selectable markers. For the wild type and all the mutant strains, the levels of expression with the fusion were clearly lower than the expression without the fusion (see Fig. 4). Nevertheless, the EXP strains improved expression of rhomboid-Rv1337 when compared with wild type whether fused to a marker protein or not.
Performance of selected strains with a T7 promoter
We were interested to see if the mutant effects were specific to the arabinose promoter system we used in our selection process or if they were more generally effective. For example, the C41 and C43 effects were found to be specific to the T7 promoter.7 We therefore lysogenized all strains with λDE3 to introduce T7 RNA polymerase and tested the mutants for efficacy in a T7 promoter system. As shown in Figure 5, all of the mutants improved expression of Rv1337 behind the T7 promoter up to fourfold. Thus, improved expression seen in the mutants is not completely specific to any one promoter type, although EXP-Rv1337-5 seems more effective for expression from an arabinose promoter.
EXP mutants deliver protein to the membrane
We expected that the selection method would indirectly select for insertion into the membrane rather than inclusion bodies, as the fusion partner must be folded and active to confer drug resistance. To evaluate if the increased expression in the mutants corresponded to increased expression to the membrane, we performed differential centrifugation. The insoluble, soluble, and membrane fractions of each sample were isolated and subsequently analyzed by western blotting. Marker proteins for the various fractions were also included during the purification: (1) streptavidin, which is found in inclusion bodies, (2) maltose binding protein (MBP), which remains soluble, and (3) GlpF, a protein targeted to the membrane. The results are shown in Figure 6. In the wild-type strain, rhomboid-Rv1337 expressed almost completely to the membrane, with a small amount of protein detected in the insoluble fraction. In all of the mutant strains, however, rhomboid-Rv1337 appeared to be expressed exclusively to the membrane with no detectable component in the insoluble fraction. The EXP mutants, therefore, lead to insertion of the protein into the membrane rather than shunting of the protein into inclusion bodies, and in fact, insertion seems to be improved in the selected mutants.
Application of selected mutants to other membrane protein targets
Although we selected for improved expression of rhomboid-Rv1337, it is possible that the isolated strains could improve expression of other membrane proteins. To test this possibility, we expressed other MTb targets and a number of rhomboid constructs from various species in the wild-type and EXP mutant strains (Table I). With the exception of Rv2835, all of these targets appear to be toxic to cells based on the observation that cell growth was dramatically reduced after induction. As shown in Figure 7, the expression of these targets could be improved in one or more of the selected mutants. EXP-Rv1337-5 is particularly effective for M. jannaschii rhomboid, D. melangaster rhomboid, and MTb Rv2835, improving expression approximately 10-, 20-, and 90-fold, respectively.
Table I. Membrane Protein Targets Expression Tested in EXP Mutants
# TM helices
Rhomboid family protein
Rhomboid family protein
Rhomboid family protein
Rhomboid family protein
Relative plasmid copy number of mutants
We found that the expression of rhomboid-Rv1337 fused to a selectable marker was improved when expressed from a single plasmid rather than two (not shown). Thus, one obvious way to increase expression would be to decrease plasmid copy number. We therefore evaluated the copy number of the EXP mutants. As indicated in Figure 8, all mutants had a plasmid copy number comparable with the wild-type TOP10 strain, except for EXP-Rv1337-4 which showed a dramatically decreased copy number. It is likely that the reduced copy number in EXP-Rv1337-4 slows down expression to a level that the cells can more readily accommodate, much like the C41 and C43 strains do for expression from the T7 promoter.7, 13
We have developed a simple selection system that can be used to produce E. coli strains that improve the production of a targeted membrane protein, and we have employed the system to improve the expression of MTb rhomboid protease Rv1337. The isolated EXP strains can also improve the production of other membrane proteins that were not the target of selection, suggesting that there may be common defects that can be corrected by genomic mutations. One EXP mutant, EXP-Rv1337-5, is generally the most effective for all tested targets.
Some EXP strains are more effective for particular membrane proteins than others, suggesting that the mutants act by distinct mechanisms. This is also indicated by our finding that EXP-Rv1337-4 dramatically reduces plasmid copy number, whereas none of the other EXP strains display this phenotype. It should in principle be possible to map the relevant mutations in the EXP mutants, and the identified genes may not only provide information about mechanism but will also enable more focused mutagenesis to achieve additional increases in expression. Furthermore, the discovered mutations could be combined in one strain to improve expression to an even greater degree.
In addition to characterizing the current mutants, our future goals are to utilize the basic selection system to further improve expression and to apply the system to a broader range of targets. Finally, we note that our selection system could also be used to improve expression of soluble protein targets.
Materials and Methods
Miller LB media (BD Difco, Cat # 244610) was used for all liquid cell cultures. Miller LB agar (BD Difco, Cat# 244510) was used for plate growth in all cloning and in all I-CreI curing procedures. M9 Minimal Media Agar was used for plate growth in all selection procedures and was composed of 1.5% agar, 1× M9 minimal media salts (47.9 mM Na2HPO4, 18.7 mM NH4Cl, 8.55 mM NaCl, and 22.0 mM KH2PO4), 1 mM CaCl2, 1 mM MgCl2, 0.2% glucose, 0.1% casamino acids, and 1 μg/mL thiamine. Chloramphenicol was added to media where indicated to a final concentration of 34 μg/mL, and ampicillin was added to 100 μg/mL.
Cloning of plasmids used in the selection process
The construct pSEL2 was made as follows. First, the gene encoding kanamycin resistance (aminoglycoside 3′-phosphotransferase, NCBI accession # ABP57185) was amplified by PCR. Code for a flexible linker was added to the fragment with the following primers: forward 5′ CTA ATC GTA ACA GGT ACC CGG CCG ACC TCT GGC ACC TCT GGC ACC TCT GGC ACC TCT GGC ATG AGC CAT ATT CAA CGG GA 3′ and reverse 5′ ACG ATC ACT GCA AAG CTT TTA GAA AAA CTC ATC GAG CAT 3′. This fragment was then placed between the KpnI and HindIII sites in the MCS of pBAD/HisA (Invitrogen). Code for the MTb target Rv1337 (NCBI accession # NP_2158531) was amplified by PCR using the following primers: forward 5′ CTA GAT ACA CGA CTC GAG ATG GGC ATG ACC CCG CGC CGG 3′ and reverse 5′ ACT CCT GAC TGA CGG CCG TAA CTT CGG ATG CCC GGA ACG 3′. This fragment was then inserted between the XhoI site in the MCS of pBAD/HisA and the EagI site inserted through the primers in the previous step.
To make pSEL1 we first had to make a second vector that would be compatible with pBAD/HisA. To do so, we constructed a plasmid with the same multiple cloning site as pBAD/HisA, but with chloramphenicol resistance and the p15A origin of replication. This compatible construct, called pBAD/HisA/p15A, was made by first amplifying a fragment in the plasmid pBT (Invitrogen, NCBI taxonomy ID 158466) from the BspHI site at base pair 2701 to the SphI site at base pair 1580. This 2 kB fragment, which contained the p15A origin of replication and a chloramphenicol resistance gene, was ligated to the 1929 base pair SphI and BphI fragment of pBAD/HisA containing the pBAD/HisA MCS and araC region.
To make the plasmid pSEL1, a fragment encoding mouse DHFR (NCBI accession # NP_034179) was amplified by PCR using the following primers: forward 5′ CTA ATC GTA ACA GGT ACC CGG CCG ACC TCT GGC ACC TCT GGC ACC TCT GGC ACC TCT GGC ATG GTT CGA CCA TTG AAC TG 3′ and reverse 5′ ACG ATC ACT GCA AAG CTT TTA GTC TTT CTT CTC GTA GAC 3′. An Eag1 site and a flexible linker were added to the fragment through the primers. This fragment was then placed between the KpnI and HindIII sites of pBAD/HisA/p15A. Code for Rv1337 was then inserted as described for pSEL2.
Cloning of the pSEL1 and pSEL2 plasmids containing E. coli GlpF, signal peptide pedtidase (SPP) from Archaeoglobus fulgidus, and the TB targets Rv2746 and Rv2835 was performed as described earlier for Rv1337.
Cloning of pBAD constructs expressing rhomboid
Code for a truncated version (amino acids 1–227 out of 249) of MTb rhomboid Rv0110 was amplified from genomic DNA, a gift of D. Eisenberg (University of California, Los Angeles) using the following primers: forward 5′ GCT AGG CAT GGT ACC ATG CAG ATA ACA CGG CCC ACA GGC and reverse 5′ GCC GTG ACG AAG CTT TCA TAA CTT CGG ATG CCC GGA ACG. This fragment was inserted between the KpnI and HindIII sites of a modified pBAD construct with the following multiple cloning site: MGHHHHHH-KpnI-HindIII.
Methanococcus jannaschii rhomboid (MJR) was expressed as a C-terminal fusion to SUMO (small ubiquitin-related modifier) with a TEV protease site in between. Code for rhomboid from Methanococcus jannaschii was amplified from the plasmid pcDNA3.1-MJR, a gift from M. Freeman (MRC Laboratory of Molecular Biology, Cambridge, UK), using the following primers: forward 5′ GATCG GGG CCC ATG ATT AAC ATT TTA ATA GTG GGG and reverse 5′ GATCG AAG CTT TTA CTC GAG ATA GTA TCT TAC ATC CAT TTT CCT. This fragment was inserted between the ApaI and HindIII sites of a modified pBAD construct with the following multiple cloning site: MGHHHHHH-KpnI-ApaI-HindIII. A fragment encoding SUMO was amplified with the following primers, which included code for a TEV protease site: forward 5′ GATCG GGT ACC ATG TCT GAC CAG GAG GCA AAA CCT and reverse 5′GATCG GGG CCC CTG AAA ATA CAG GTT TTC ACC CCC CGT TTG TTC CTG ATA AAC. This fragment was then inserted in a second step between the KpnI and ApaI sites.
The Drosophila melanogaster rhomboid 1 (Rho 1) construct was made exactly as described earlier for the SUMO-TEV-MJR construct, with the exception that an S-tag was added to the N-terminus of SUMO through the primers. Rho 1 was amplified from the plasmid pcDNA3.1-Rho-1, a gift from M. Freeman (MRC Laboratory of Molecular Biology, Cambridge, UK), using the following primers: forward 5′ GATCG GGG CCC ATG GAG AAC CCA ACG CAG AAT GT 3′ and reverse 5′ GATCG AAG CTT TTA GGA CAC TCC CAG GTC GTG C 3′. A fragment encoding S-tagged SUMO followed by a TEV protease site was amplified with the following primers: forward 5′ GATCG GGTACC AAA GAA ACC GCT GCT GCT AAA TTC GAA CGC CAG CAC ATG GAC AGC ATG TCT GAC CAG GAG GCA AAA CCT 3′ and reverse 5′GATCG GGG CCC CTG AAA ATA CAG GTT TTC ACC CCC CGT TTG TTC CTG ATA AAC 3′.
Cloning of controls used in cellular fractionation
E. coli GlpF was inserted into pBAD/HisA/p15A behind the XhoI site using the polymerase incomplete primer extension (PIPE) cloning method.17 The primers used to amplify the insert are as follows: forward 5′ GGA TCC GAG CTC GAG ATG AGT CAA ACA TCA ACC TTG AAA CGC 3′ and reverse 5′ CAA AAC AGC CAA GCT TTT ACA GCG AAG CTT TTT GTT CTG AAG G 3′. The primers used to amplify the vector are as follows: forward 5′ CGG CCG ACC TCT GGC ACC TCT GGC ACC 3′ and reverse 5′ AAG CTT GGC TGT TTT GGC GGA TGA GAG 3′.
pMALc2X (NEB) was modified to express the MBP with a C-terminal 6-histidine tag and without a fusion to β-galactosidase. This construct was made using the PIPE cloning method.17 The primers used to amplify the vector and which included code for an N-terminal 6×His tag and stop codon are as follows: forward 5′ CAC CAC CAC CAC CAC CAC TGA AAC AAC AAC AAC AAT AAC AAT 3′ and reverse 5′ TCA GTG GTG GTG GTG GTG GTG CGA GCT CGA ATT AGT CTG CGC 3′.
The pET21a construct expressing “alive” streptavidin was a gift from A. Y. Ting (Massachusetts Institute of Technology, Cambridge) and is described by Howarth et al.18
Cloning of plasmids used in I-CreI curing
A plasmid expressing the homing endonuclease I-CreI (NCBI accession # P05725), called pCURE, was made by first amplifying the region containing the arabinose promoter and code for I-CreI from the plasmid pAE.16 The primers used are as follows: forward 5′ TCA GCA CTA GAC CTC GAG ATC GAT GCA TAA TGT GCC TGT C 3′ and reverse 5′ ACG TCA CTG ACG CAG CTG TTA CGG GGA CGA TTT C 3′. This amplified region was then inserted between the XhoI and PvuII sites of plasmid pSC101. The construct was then made temperature sensitive by mutating the residue Ala56(gct) to Val56(gtt) in the repA gene of pSC101 through site-directed mutagenesis using Stratagene's QuikChange site-directed mutagenesis protocol.19
pSEL1 and pSEL2 plasmids containing the I-CreI target site (5′ CAA AAC GTC GTG AGA CAG TTT GGT 3′) were created as follows.16 The I-CreI insert was prepared by annealing the phosphorylated primers (5′ Phos AGC TCA AAA CGT CGT GAG ACA GTT TGG T 3′) and (5′ Phos AGC TAC CAA ACT GTC TCA CGA CGT TTT G 3′) and then inserting this fragment into the HindIII site of the pSEL1 and pSEL2 plasmids. The primers were annealed by briefly boiling a solution containing 20 μM of each primer and 1 mM NaCl and then allowing the mixture to cool down to room temperature.
Cloning of rhomboid-Rv1337 without a fusion to a selectable marker
Rhomboid-Rv1337 was amplified using the following primers: forward 5′ CTA GAT ACA CGA CTC GAG ATG GGC ATG ACC CCG CGC CGG 3′ and reverse 5′ ACT CCT GAC TGA AAG CTT TCA TAA CTT CGG ATG CCC GGA 3′. This fragment was inserted into the MCS of pBAD/HisA or pBAD/HisA/p15A between the XhoI and HindIII sites.
Cloning of pET construct for expression in T7 lysogens
The code for Rv1337 with a C-terminal fusion to kanamycin resistance was amplified from pSEL2 encoding rhomboid-Rv1337 using the following primers: forward 5′ GCT ACT AGT ATC GGA TCC ATG GGC ATG ACC CCG CGC CGG 3′ and reverse 5′ GCT GAC CGT CGG AAG CTT TTA GAA AAA CTC ATC GAG CAT 3′. The amplified sequence was then inserted between the BamHI and HindIII sites in the MCS of pET28a (Novagen).
Cloning of pSC101ts/mutD5
The mutD5 gene was amplified from the mutator strain 1187, which was a gift from R. C. Johnson (UCLA, Los Angeles), and contains the single base change C to T at position 967 in the dnaQ gene, changing Thr-15 to Ile-15.20 Code for mutD5 was then inserted into pSC101 using the PIPE cloning method.17 The primers used to amplify the insert are as follows: forward 5′ AAC CTG TAC TTC CAG CGC CTC CAG CGC GAC AAT AGC GGC CAT C 3′ and reverse 5′GAG TTA ATT AAG TCG CGC CGA CTG AAC TAC CGC TCC GCG TTG TG 3′. The primers used to amplify the vector are as follows: forward 5′ CGC GAC TTA ATT AAC TCC GTC GAA AAC TGC ATG CAT TGC 3′ and reverse 5′ CTG GAA GTA CAG GTT CGT TTA ATC ACC AGA ACG GTG GTG 3′. This construct was then made temperature sensitive as described earlier for the plasmid pCURE.
Test of selection system
TOP10 cells harboring pSEL1 and pSEL2 encoding GlpF and TOP10 cells harboring pSEL1 and pSEL2 encoding SPP were streaked onto minimal media agar containing chloramphenicol and ampicillin and the following three selection conditions: no kanamycin and no timethoprim with 0.2% arabinose, 40 μg/mL kanamycin, and 1 μg/mL trimethoprim with no arabinose, and 40 μg/mL kanamycin and 1 μg/mL trimethoprim with 0.2% arabinose. Plates were grown at 37°C for 2 days.
Mutagenesis with 2AP
Mutagenesis with the base analog 2AP (Sigma-Aldrich catalogue # A3509) was carried out as described by Miller.21 Mutants EXP-Rv1337-1, −2, −3, and −4 were produced using 2AP.
Mutagenesis with mutD5
Based on the mutagenesis method of Selifonova et al., pSC101ts/mutD5 and pSEL1 encoding Rv1337 were transformed to wild-type TOP10 cells, and cells were outgrown at 30°C for 2 or 3 days on LB agar containing tetracycline (5 μg/mL).22 Cell growth was washed off of plates with LB broth, the OD600 was normalized to 0.2 with LB, and cells were outgrown for 1 h at 42°C to remove pSC101ts/mutD5. The mutation rate was determined by monitoring the level of rifampicin resistant mutants, and after 3 days of growth in the presence of mutD5 there was about a 300-fold increase in the mutation rate.21 Mutant EXP-Rv1337-5 was produced by using mutD5.
Mutagenized cells harboring pSEL1 encoding Rv1337 were washed once with sterile PBS (137 mM NaCl, 2.7 mM KCl, 10.1 mM Na2HPO4, 1.8 mM KH2PO4) and then suspended in sterile PBS. A total of 100 μL of washed cells were plated onto minimal media agar containing chloramphenicol, 0.02% arabinose, and a gradient of trimethoprim (0, 100, or 200 μg/mL). Plated cells were outgrown for 2–3 days at 37°C. Colonies surviving on high trimethoprim (100 μg/mL or 200 μg/mL) were collected by scraping. Cells were diluted to an OD600 of 0.05 in 200 mL LB containing 34 μg/mL chloramphenicol, and then grown to an OD600 of about 0.6. Cells were then prepared to be chemically competent and were transformed with pSEL2 encoding Rv1337. The transformed cells were spun down and then washed and resuspended in sterile PBS. Washed cells were then plated on minimal media agar containing chloramphenicol, ampicillin, 0.02% arabinose, trimethoprim (0, 100, or 200 μg/mL), and kanamycin (0, 10, 20, 40, or 80 μg/mL). The plates were incubated at 37°C for 2 days. Colonies surviving on the highest levels of trimethoprim and kanamycin were isolated by culturing overnight in LB media containing chloramphenicol and ampicillin.
Small-scale expression testing
Two milliliters of LB containing ampicillin and chloramphenicol were inoculated with 50 μL of a saturated overnight culture of a selected colony. Cultures were outgrown at 37°C to an OD600 of about 0.5–0.6 and then induced with 0.2% arabinose for 3 h at 37°C. One milliliter cells were spun down and brought up in 200 μL SDS protein sample buffer (125 mM Tris, 20% v/v glycerol, 4% w/v SDS, 0.2% w/v bromophenol blue, 10% v/v BME, pH 6.75). Expression was then analyzed by western blotting.
Whole cell lysates in SDS protein sample buffer were centrifuged for 10 min at 13,000 rpm, and a normalized amount of each sample, based on OD600 at the end of induction, was then loaded onto an Invitrogen NuPAGE Novex 4%–12% Bis-Tris gel. A twofold dilution series (1000 ng to 3.9 ng) of purified His-tagged biotin ligase (BirA) was also loaded as a protein quantification standard. The gel was run at 200 V in MOPS SDS running buffer (50 mM MOPS, 50 mM Tris base, 0.1% SDS, 1 mM EDTA, pH 7.7) using an Invitrogen XCell SureLock Mini-Cell electrophoresis system. The gel was then transferred to a nitrocellulose membrane in transfer buffer (25 mM Tris pH 8.3, 192 mM glycine) at 25V for 2 h in the Invitrogen Invitrogen XCell SureLock Mini-Cell with the XCell II blot module. The membrane was then probed using QIAgen's Anti Penta-His HRP conjugate kit (Cat# 34460) with a 1/2000 dilution of antibody. The blot was visualized using ECL Plus western blotting detection reagents (GE Healthcare catalogue # RPN2132) and then exposed to film.
Densitometry and quantification of expression
Adobe Photoshop CS2 was used to quantify the intensity of bands on western blots. A scanned image of the western blot was converted to grayscale and then dark and light were inverted. A band was selected using the lasso tool. The histogram function in Photoshop was used to obtain the value “Mean,” which is the average band brightness, and the value “Pixels,” which is the number of pixels selected. The values “Mean” and “Pixels” were multiplied to determine a band's absolute intensity. The absolute intensity values of the loaded BirA protein standard were used to generate a standard curve, which was then used to determine the amount of protein expressed in each test sample on that same blot. The fold increase in expression of each mutant over wild type was determined by dividing the sample's absolute intensity by that of the wild-type band found on the same western blot. For all charts, the average of three trials is shown and the standard deviation is reported.
Curing of mutant EXP-Rv1337-1 by outgrowth with the curing agent apramycin
Mutant EXP-Rv1337-1 was cured by outgrowth in the absence of selection and in the presence of the curing agent apramycin according to the method of Denap et al.23 Mutant EXP-Rv1337-1 was outgrown for approximately 1 month before a cured colony was isolated.
Curing of mutants EXP-Rv1337-2, −3, −4, and −5 by digestion with I-CreI
Selected cells were prepared to be competent and transformed with the plasmid pCURE. Transformed cells were then plated on LB agar containing 5 μg/mL tetracycline and 0.02% arabinose and outgrown overnight at 30°C to induce I-CreI expression and digest pSEL1 and pSEL2, both of which contained the I-CreI target site. To subsequently cure the cells of pCURE, an isolated colony was streaked to single colonies on LB agar without drug and grown overnight at 42°C. To confirm that curing was complete, colonies were replica plated on LB agar in this order: LB without drug, LB with ampicillin, LB with chloramphenicol, LB with 5 μg/mL tetracycline, and, finally, LB without drug.
Validating cured cell lines
Selected and cured mutant cell lines and the wild-type TOP10 cell line were transformed with pSEL1 and pSEL2 encoding rhomboid-Rv1337. Small-scale expression testing of cured mutants and western blotting was performed as described earlier.
Cellular fractionation by differential centrifugation
A total of 100 mL LB were inoculated with 1 mL of an overnight culture of wild-type TOP10 or mutant cells expressing rhomboid-Rv1337 in pSEL1 and pSEL2. In addition, 100 mL cultures were inoculated with 1 mL of overnight cultures of TOP10 cells expressing GlpF in pBAD/HisA/p15A, BL21(DE3)pLysS cells expressing “alive” streptavidin from a pET21a, or BL21(DE3) cells expressing MBP from pMALc2x. Cells were outgrown at 37°C to an OD600 of about 0.5–0.6. Wild-type and mutant TOP10 cultures were then induced with 0.2% arabinose, BL21(DE3)pLysS expressing streptavidin was induced with 0.085 mM IPTG, and BL21(DE3) expressing pMALc2x was induced with 0.1 mM IPTG. After 3 h of induction at 37°C, a normalized amount of cells based on OD600 were pelleted by centrifugation and stored at −80°C. Cells expressing rhomboid-Rv1337 in pSEL1 and pSEL2 were resuspended in 5 mL Buffer A (200 mM NaCl, 40 mM sodium phosphate buffer pH 7.5) with Protease Inhibitor Cocktail (Roche Complete P.I.C., Cat # 11697498001), and each sample was spiked with each of the three marker cell cultures resuspended in buffer A with P.I.C. Resuspended cells were then lysed by sonication with a microtip for four cycles (Level 4, 50%) of 1 min each, with about 10 min in between each cycle. The lysate was then centrifuged at 6000 rpm using a type SS-34 rotor for 30 min at 4°C to pellet the insoluble fraction, and this pellet was subsequently washed to remove any unlysed cells from the inclusion body fraction according to the protocol detailed in Palmer and Wingfield.24 The supernatant containing the membrane and soluble components was centrifuged at 30,000 rpm using a type 45 Ti rotor for 1 h at 4°C to pellet the membrane fraction. All samples were suspended in SDS protein sample buffer and were analyzed by western blotting as described earlier.
λDE3 Lysogenization and expression testing utilizing the T7 promoter
The Novagen λDE3 lysogenization kit (catalog # 69734-3) was used to introduce T7 polymerase to the genome of wild-type and mutant cell lines. Two milliliter of LB with chloramphenicol and kanamycin (40 μg/mL) were inoculated with 50 μL of an overnight culture of the lysogenized cell line harboring the pET28a/Rv1337 construct and pLysS. At an OD600 of about 0.5–0.6, cultures were induced with 1 mM or 0.1 mM IPTG for 3 h. Small-scale expression screening and western blotting was performed as described previously.
Relative copy number determination
The relative plasmid copy numbers of the wild-type TOP10 strain and all mutant strains were determined by growing cells harboring rhomboid-Rv1337 in pSEL1 and pSEL2 to an OD600 of about 0.6. Samples were normalized based on OD600. Plasmid DNA was then purified using the QIAgen QIAprep Spin Miniprep Kit (Cat #27104). An internal standard of HindIII-digested pUC19 plasmid DNA was added at the point at which cell lysate was added to the QIAprep column. Ten microliters of each purified DNA sample was loaded on a 1% agarose gel containing ethidium bromide to monitor relative copy number.
The authors thank the assistance of Christine Chen in plasmid cloning and Keith Gendel for figures. We thank Nathan Joh, Mary Jane Knight, Tyler Korman, Tracy Mitchell, Frank Pettit, Megan Plotkowski, Ryan Stafford, and Jijun Yuan for a critical reading of the manuscript.