PEDS Advance Access published online on November 21, 2008
Protein Engineering Design and Selection, doi:10.1093/protein/gzn069
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Evolving Lac repressor for enhanced inducibility
Center for Cellular and Molecular Biology, Council of Scientific and Industrial Research, Uppal Road, Hyderabad 500007, India
1 To whom correspondence should be addressed. E-mail: madhu{at}ccmb.res.in
 |
Abstract |
|---|
 Top Abstract IntroductionMaterials and methodsResults and discussionFundingAcknowledgementsReferences |
|---|
Lactose repressor (LacI) is one of the best studied prokaryotic transcriptional regulatory proteins till date. Detailed structural, biochemical and genetic studies are being carried out on LacI since four decades to understand its ligand binding properties and the basis of allosteric response. We applied directed evolution methods on LacI to generate mutants with altered allosteric properties. After testing several hosts and expression vectors, a robust expression and screening system was optimised for identifying LacI variants with altered allosteric properties. After two rounds of error prone PCR (polymerase chain reaction) and shuffling, four mutants were selected from several thousand mutants, for their ability to induce reporter gene expression at 1 µM of isopropyl β-D-1-thiogalactopyranoside (IPTG). The observed combination of mutations in these four improved LacIs was not reported earlier. The mutant Lac repressors seem to operate as very good molecular switches by inducing gene expression at 1 µM of IPTG and confer 2–10 times higher level of gene expression as compared with the WT (wild -type).
Keywords: allostery/directed evolution/inducer/LacI/operator
|
Introduction |
|---|
|
TopAbstractIntroductionMaterials and methodsResults and discussionFundingAcknowledgementsReferences |
|---|
Lactose repressor (LacI) is one of the most extensively investigated proteins in the class of deoxyribonucleic acid (DNA) binding regulatory proteins (Barkley and Bourgeois, 1978
; Bell and Lewis, 2001; Lewis, 2005; Wilson et al., 2007). The regulatory properties of LacI are based on its allosterically regulated dissociation from operator upon binding of ligands viz. allolactose or isopropyl β-D-1-thiogalactopyranoside (IPTG). The three-dimensional (3D) structure of the aporepressor, repressor complexed with the operator as well as with the inducer, IPTG, provided important clues in understanding the allosteric response in LacI (Lewis et al., 1996). Several mutations in Lac repressor and their effects have been reported (http://www.bioc.rice.edu/~ksm/). Among these, few mutations were found to have altered the allosteric properties of the repressor. Prominently investigated among these mutations are the position H74, S77, A81 K84, E100, A110 and L148 (Barry and Matthews, 1999; Swint-Kruse et al., 2003; Swint-Kruse et al., 2005). Substitution at just a single residue, A110K, altered the allosteric equilibrium to High operator, Low inducer affinity state or Low operator, High inducer affinity state in case of A110T. This has been explained by the stabilisation of operator-bound conformation in the former case, and inducer -bound conformation in the latter case (Muller-Hartmann and Muller-Hill, 1996). Chakerian et al. (1985) showed that A81V mutation was capable of locking the protein in the induced conformation in LacI. Based on mutational studies (Sukow et al., 1996) and targeted molecular dynamics (TMD) (Flynn et al., 2003) it is postulated that the conformational changes during allosteric transition upon inducer binding result in shifts in N-subdomains both with respect to each other and to the C-subdomain, while both the subdomains were structurally unaltered during the transition (Bell and Lewis, 2001; Lewis, 2005; Wilson et al., 2007). C-subdomain acts as a scaffold for the N-subdomain to rotate, where primarily the dimer interface plays a key role. The ‘end states’ revealed in the crystal structure and equivocal explanations of the role of amino acid substitutions limited our understanding of the transition pathway between promoter bound and free states of the LacI.
General principles underlying complex properties such as allostery are unclear not only for LacI but for many other allosteric proteins as well. In the absence of structure-function correlations of proteins, directed evolution has emerged as a powerful tool in improving protein properties and also in generating unanticipated but vital mutations (Bloom et al., 2005). This approach is popular due to the minimalist requirements of the method to create variant library of the desired gene and a sensitive screening system. The majority of properties improved by directed evolution, however, belong to the commercially important properties, and proteins with regulatory properties were seldom attempted by this approach. We report results of directed evolution of the LacI towards altered allosteric properties. The random mutagenesis and screening of LacI mutant library for response towards low concentrations of inducer (IPTG) reported in this study resulted in mutants that nearly behaved as ideal molecular switches for regulating gene expression. Ability to respond to lower concentrations of inducer and a better On/Off switch behaviour are desirable in an ideal heterologous protein expression system. We show here that ligand binding properties of a complex regulatory oligomeric protein such as LacI can be evolved using simple in vivo screening tools such as inducibility and exploiting the natural context of the Lac operon.
|
Materials and methods |
|---|
|
TopAbstractIntroductionMaterials and methodsResults and discussionFundingAcknowledgementsReferences |
|---|
Materials
IPTG, ortho-nitrophenyl-β-galactoside (ONPG), 5-bromo-4-chloro-3-indolyl- beta-D-galactopyranoside (X-gal) and antibiotics were purchased from Sigma Co, MD, USA. Plasmid purification kits were from Qiagen, CA, USA. Enzymes were purchased from New England Biolabs, MA, USA and Stratagene, CA, USA. All other chemicals were of analytical grade. GJ 2427 [Escherichia coli (E.coli) MG1655 LacI::Tn10 (Kan)] was a kind gift from Dr J. Gowrishankar, CDFD, Hyderabad. Vector pLS2, a modified version of vector pLS1 was a kind gift from Dr Kathleen S. Matthews, Rice University, USA. Vector pET21d was purchased from Novagen.
Generation of mutant library of LacI by error prone PCR
Low copy number vector pPR9TT was used and E.coli host strain SL1 was used for analysing the β-gal activity of clones. Error prone polymerase chain reaction (PCR) product of LacI was generated using pPRWT as template with ppF (5'-AAAGCTGACTCTAGCTAGAGGATCTTC-3') and ppD (5'-TATCGATAAGCTTGATATCGAATTCCT-3') as primers with the final concentration of buffer containing 10 mM Tris–Cl pH 8.3, 50 mM KCl, 7 mM MgCl2, 0.01% gelatine and 0.5 mM MnCl2. The dNTP ratio was also altered with dCTP and dTTP having a final concentration of 1 mM while dATP and dGTP were at a final concentration of 0.2 mM. The cycling conditions were similar to normal PCR as shown below: 94°C – 3 min/94°C – 30 s/55°C – 1 min/72°C – 1 min/72°C – 5 min/4°C – Hold.
DNA shuffling was performed according to the protocol of (Stemmer, 1994). Lac repressor gene was amplified from all the twenty clones from the first generation in separate tubes and purified. DNA was quantified and mixed in equal amounts for DNAse digestion. DNAse was added to this mixture in a buffer containing 10 mM Tris–Cl pH 8.0, to a final concentration of 0.0016 units and incubated at 37°C for 30 min. The reaction was stopped by adding 0.5 M EDTA to a final concentration of 100 mM and subsequently run on a 1.8% agarose gel and stained with ethidium bromide. Smear having the size of 10–50 bp fragments was cut from the gel and eluted using Qiagen gel extraction kit. The eluted products were used as templates for the first recombination reaction without primers using the cycling conditions as in the case of error prone PCR but with a reduced ramp rate of 50% at the annealing step. Following this, the first recombination product was used as template in a 1/20 dilution in a second recombination reaction using the primers ppF and ppD. The cycling conditions for this reassembly reaction were exactly similar as earlier. The PCR products generated after second reassembly reaction were purified and digested with BstAPI and NotI as well as the vector pPRWT. These digested products were purified by gel extraction and ligated using T4 DNA ligase. The ligated products were transformed into DH5
+ pJO292 and the transformants were plated on X-gal + 1 µM IPTG plates.
Estimation of β-galactosidase activity
The β-galactosidase assay was done according to the method of Miller (1979) with minor modifications as detailed later. A secondary culture was inoculated using 1% overnight grown primary inoculum and grown in the absence as well as presence of IPTG. After the secondary cultures have reached mid log phase, OD600 was measured and the cultures chilled on ice for 30 min. Meanwhile, 900 µl of assay buffer was dispensed in assay tubes following which 100 µl of chilled culture was added. Sodium dodecyl sulphate (SDS) to a final concentration of 0.35% and a drop of chloroform were added to lyse the cells by vortexing for 30 s. To this, 200 µl of a 4 mg/ml ONPG solution was added and the time noted down. The tubes were incubated at room temperature till sufficient yellow colour developed. The reactions were stopped after the time was noted down for the colour to develop by adding 500 µl of 1M sodium carbonate. Subsequently, OD420 and OD540 were measured for all the reactions. β-Galactosidase assays were performed according to the protocol suggested by Miller in the presence as well as absence of IPTG and the results expressed as induction ratio which is defined as β-galactosidase levels in the presence of IPTG/β-galactosidase levels in the absence of IPTG. Plasmid was isolated from the first-generation and second-generation clones using Qiagen plasmid extraction kit and sequenced in-house using ABI sequencer 3730 and 3700 with various primers spanning the whole gene.
Selection of positive clones by β-galactosidase activity
In each of the first and second generation, the positive clones were selected on the basis of an intense blue colour of the colonies with the prescribed concentration of IPTG. Such blue colonies from the petri plates were picked and inoculated into individual wells of a micro titer plate containing 200 µl 2XYT. This primary inoculum was grown overnight at 37°C, 200 r.p.m. A secondary culture in the absence as well as presence of IPTG was inoculated with 2.5% primary inoculum. These plates were again grown at 37°C till the cultures reached a mid log phase characterised by an OD 600 of 0.4–0.6. In a fresh plate, mixed equal volumes of 2XYT + 0.7% SDS with the log phase culture (100 µl of each) and mixed thoroughly. This mixture was incubated at 37°C for 30 min. to facilitate cell lysis. This plate after 30 min, was centrifuged at 4000 r.p.m. for 30 min to pellet down the cell debris. The β-galactosidase was estimated as described earlier. Plasmid was isolated from the first-generation and second-generation clones using Qiagen plasmid extraction kit and sequenced in-house using ABI sequencer 3730 and 3700 with various primers spanning the whole gene.
|
Results and discussion |
|---|
|
TopAbstractIntroductionMaterials and methodsResults and discussionFundingAcknowledgementsReferences |
|---|
Optimising the screen
Directed evolution of LacI towards ligand binding affinity implies that the variants could be improved in the direction of either enhanced DNA binding ability or inducer (IPTG) binding ability, or improved allosteric response. DNA binding affinity of LacI is in the picomolar range while the inducer binding affinity is in the micromolar range, hence providing scope for further improvement in the inducer binding ability (Riggs et al., 1970; Friedman et al., 1977). We chose to employ the components of naturally occurring Lac operon system present in the E.coli cell to examine the LacI activity by monitoring the reporter gene, β-galactosidase enzyme. The reporter activity can be monitored using X-gal and ONPG, synthetic substrates, for solid phase and solution phase assays, respectively. The selection criterion for positive clones in this study is based on induction ratio calculated as (β-galactosidase activity with IPTG/β-galactosidase activity without IPTG) where the former represents induced level and the latter signifies repressed level of the operon. Thus, white colour colony on X-gal plates without IPTG and blue colour on X-gal + IPTG plates would represent the desired level of repression and induction levels and an optimal induction ratio. The host SL1, which does not contain endogenous LacI, was constructed to prevent subunit poisoning (see Supplementary Fig. 1 available at PEDS online). To optimise the induction ratio for the plate-based assay, vector backbones of different copy numbers and two promoters of different strengths exhibiting varied levels of repression and induction were tested. pLS2, vector expressing the LacI under high strength promoter LacIq and pet21D, a medium copy vector expressing the repressor under LacI promoter were employed. In case of pLS2 the repression was too high resulting in white coloured colonies in the presence of IPTG and in the pet21D case, repression was too low resulting in blue coloured colonies even in the absence of IPTG. However, under high strength LacIq promoter on a low copy number vector pPR9TT (Santos et al., 2001), the expression of LacI as estimated by the β-galactosidase expression was responsive to variations in IPTG concentration (Supplementary Fig. 2, available at PEDS online). Thus, the most important feature that emerged during the establishment of a vector system was that in vivo expression levels of LacI (in this case varied due to plasmid copy number) is crucial for the efficiency of the screen.
Generally, the concentration used for inducing heterologous protein expression using a LacI–operator system is in the range of 0.5–1 mM IPTG. In the first round of screening of the mutants generated by error prone PCR, a suboptimal concentration of 200 µM was employed. In the first round of screening on X-gal + IPTG plates, 5000 clones were screened and 800 clones were selected based on the solid phase assay. The 800 selected clones were assayed for induction ratio in micro titer plates, and finally, 20 clones were selected based on their improvement in induction ratio over the wild-type (WT) repressor. As shown in Fig. 1A, WT LacI showed an induction ratio of 15 while the best performing clones 1-5A6, 1-1F8 and 1-2B10 showed an induction ratio of 
60, 48 and 50 respectively. The remaining clones also showed an increase in induction ratio at least 2-fold higher than the WT. A close inspection of the sequencing results of the first-generation clones illustrated that the mutations generated in first round of screening were scattered throughout the core domain (N- and C-domain) and both at the primary as well as in 3D structure. It is possible that clustering of mutations in the primary structure reflects bias in the generation of random mutations while clustering in the 3D structure reflects the importance of those regions in the property (Supplementary Fig. 3A and B, available at PEDS online). Incidence of mutations throughout the core protein, which do not participate directly in ligand binding as observed in our study suggests that these may be involved indirectly in the functioning of LacI.
|
The strategy used to create initial diversity in a library and the path taken for further diversification plays a very important role in the success of a directed evolution experiment (Bloom et al., 2005
). To generate second-generation mutant library two possibilities existed. The best clone from the first generation could be subjected to further rounds of random mutagenesis or a set of promising clones from the first generation could be shuffled to create further diversity. Since, the mutational spectrum of first-generation clones did not highlight any particular region of the protein we decided to shuffle all the twenty clones from the first generation instead of improving a single clone further. Such a strategy would be beneficial by evading Muller's ratchet also (Muller, 1964).
Shuffling of the 20 clones from first generation was performed by the method of Stemmer (Stemmer, 1994). Screening of the shuffled clones from first generation was done at 1 µM IPTG, which is 200 times lower than the concentration used in first generation. Five thousand clones were screened on solid phase assay and 500 clones were picked for micro titer plate assays. From these 47 clones were taken for confirmatory β-galactosidase assays and finally four mutants were identified based on their induction ratios. These four mutants showed inducibility of 4–8-fold higher than the WT, which remained uninducible at 1 µM IPTG (Fig.1B, Table I). An interesting observation with the second-generation clones was that almost all the clones showed high β-galactosidase levels in the absence of IPTG in addition to inducibility with IPTG (Table I). As expected, the second-generation mutations were scattered in the entire gene but their combinations were found to be different from the first generation (Supplementary Fig. 3C and D, available at PEDS online). The first-generation clones therefore showed normal repression but altered allosteric response, whereas the second-generation clones have altered repression as well as allosteric response. Thus, by evolving the LacI through a single round of error prone PCR followed by DNA shuffling, we could obtain mutant LacIs showing 4–8 times higher inducibility than the WT at 1 µM of IPTG concentration in the designed host–vector system.
|
LacI expression levels and inducibility
It has been earlier demonstrated that the relative concentrations of repressor, operator and the inducer determines the On/Off status of the operon (Glascock and Weickert, 1998). Concentration of the inducer, IPTG, and operator (present on host chromosome) are the fixed parameters in this study and hence the only variable that can be varied is the repressor concentration inside the cell. During optimisation of the screen we observed that inducibility of β-galactosidase was critically dependent on the concentrations of LacI. Low expression of the LacI from low copy number plasmid may have caused leaky expression of the reporter gene, hence we enhanced the in vivo LacI concentration by expressing LacI from a high copy number plasmid. This was demonstrated with one of the second-generation clones 2-5D2 and the WT LacI on two high copy number vectors, pBSSK and pUC 18. In case of 2-5D2 the induction ratio has increased from 8 (pPRWT) to 25 (pBSSK or pUC18), where the significant effect of the plasmid was on the background levels than induced levels (Fig. 2). Our results indicate that expression from a high copy number vector is able to check leaky background expression effectively both in case of WT and the mutant 2-5D2. This reaffirms that the levels of intracellular repressor concentrations determine the leaky expression (Studier and Moffatt, 1986; Williams et al., 1998; Grabherr et al., 2002).
|
We estimated the β-galactosidase expression level of the mutants at increasing concentration of IPTG in the background of LacI expression from a high copy number vector, pBSSK, in the host SL1. The results as shown in Supplementary Fig. 4, available at PEDS online, suggest that β-galactosidase expression was switched Off in the absence of IPTG both for WT and mutants. Low levels of background expression (absence of IPTG) of the β-galactosidase in the mutants in the high copy number plasmid were a crucial observation. However, in the presence of IPTG, WT shows a gradual increase in expression of β-galactosidase which saturates above 250 µM IPTG. In the case of mutants, expression starts at 10 µM IPTG and saturates at 50 µM IPTG itself and further addition of IPTG does not increase gene expression levels. Thus, as compared with WT, which shows an IPTG concentration dependent increase in expression of β-galactosidase, the mutants show a distinctly altered and more sensitive induction profile.
Most of the E.coli based protein over-expression schemes are based on the LacI–IPTG system, wherein 0.5–1 mM IPTG is the typical concentration range used for inducing protein expression (Studier and Moffatt,1986; Dubendorff and Studier, 1991). The mutant LacIs, however, can induce gene expression at 100 times lower than the normal IPTG concentration, i.e. 10 µM IPTG. Apart from induction at low IPTG concentration, these mutant repressors also confer higher levels of induced gene expression as compared with the WT. The highest induced level of gene expression seen with the WT is 250 Miller units whereas all the mutants were able to show at least 2–10 times higher level of expression. Mutant 2-5D2 shows only two times higher expression while the mutant 2-1A5 shows the highest induced β-galactosidase levels of >2500 units, which is 10 times higher than WT. The shift of the induction profiles to lower IPTG concentrations suggests that the allosteric equilibrium may have been altered towards low operator and high inducer affinity state in case of the mutants (Supplementary Figs 5 and 6, available at PEDS online). In addition to altered binding with IPTG, the mutations may influence the turnover of the LacI in vivo, thereby altering the cellular concentrations.
The implication that these mutants possess allosteric properties is demonstrated by 2–10-fold higher expression of β-galactosidase at 10 µM IPTG and supported by the absence of background expression of reporter gene. Poor binding of these mutant repressors with the operator might have led to the shift in the conformational equilibrium towards inducer bound form. The allosteric behaviour of LacI explained using the Monod Wyman Changeux model suggests that the allosteric constant L is an indication of the two populations of R and T states existing at any given time point (Kercher et al.,1997). It also seems to suggest that DNA binding affinity and inducer binding affinity are at the two ends of a spectrum and improvement in one property often leads to a compromise in the other property. The reported crystal structures of the repressor complexed with the DNA (operator) and inducer (IPTG) describe the end states of conformational transition (Lewis et al., 1996). These have provided invaluable insight into the role of subdomains of the repressor during induction. The pathway of conformational transition between these two states and the amino acids involved in the transition are incompletely understood but few mutations were identified to influence the conformational transition (Barkley and Bourgeois, 1978; Bell and Lewis, 2001; Lewis, 2005; Wilson et al., 2007). At the same time there are mutations that do not seem to be a part of this seesaw model of allostery. Interestingly, in one case, a combination of IPTG bias mutation (L185F) and operator bias mutations (Q60G) resulted in a repressor with enhanced inducer sensitivity (Swint-Kruse et al., 2003).
The mutations observed in our study are a total of thirteen mutations in four mutants. The distribution of the mutations is as follows: four mutations in DNA-binding domain, six in C-domain, one in N-domain and two in the junction of DNA-binding/N-domain and N-domain/C-domain. Occurrence of six mutations in our study in C-domain is interesting. Though a number of amino acids of C-domain bind to the inducer, C-domain is expected to act as a scaffold for the rotational twist that occurs in N-domain upon inducer binding. F161 and P188 positions were demonstrated to elicit Is phenotype, i.e. they have lost their ability to bind to the inducer or incapable of transmitting the allosteric signal (Chou et al., 1989). E277G is a significant mutation, i.e. by absence of charge, may affect the dimer interface. Its adjacent amino acid, D278 has been extensively investigated for its role in dimerisation (Barry and Matthews,1999). Extensive interaction of the DNA binding domain with the N-subdomain are seen in the crystal structure (Lewis et al., 1996; Bell and Lewis, 2001). N46, a position mutated in 2-5D2, hydrogen bonds with T141 of the N-domain implicating its role in conformational transition (Lewis, 2005). TMD study on the repressor, involving the two end states, operator-bound and inducer-bound states, provided three possible interconnected pathways by which the conformational transition may be routed (Flynn et al., 2003). Biochemical and mutational data is in good agreement with these pathways, thus providing a framework to test the mechanistic basis of the transition. F161, implicated in the pathway 3 in TMD, was one of the positions that shows I– and Is phenotype and present in the hydrophobic core pivot region. F161S would considerably alter the hydrophobicity thereby altering the mobility of the N-subdomain. Most of the mutants identified in our directed evolution effort could affect the allosteric signal, though they do not belong to the enumerated amino acids in contact with the inducer (Lewis, 2005; Chou et al., 1989). In an interesting structure based mutational study, affinity of maltose binding protein (MBP) to its ligand was improved 100-fold by identifying the sites in MBP that do not bind the ligand but whose interactions differ in the open and closed states (Marvin and Hellinga, 2001). Similarly, some of the mutations reported in our study though not involved directly in ligand binding might still be able to influence the conformational equilibrium possibly due to their different interactions in the R and T states. Absence of background expression and complete induction by 10 µM IPTG in the four mutants indicate that these mutant repressors are behaving as excellent molecular switches.
In conclusion, our studies on LacI demonstrated that directed evolution can successfully identify mutants with favourable allosteric properties, i.e. induction at lower concentrations of inducer in case of LacI. Mutant LacI induce gene expression in the presence of 100 times lower than the normal IPTG concentration and also confer 2–10-fold higher levels of expression. The mutations identified in our study are not reported in earlier studies. Interestingly these mutations, especially in C-domain, are not in contact with the inducer but could alter the allosteric response nevertheless. The mutations may be eliciting their effect by altering the equilibrium between the R and T states. Thus, engineering allostery in LacI not only provides variants to probe the basis of allosteric response but can also be exploited for industrial use as most of the commercial E.coli based over-expression systems are based on LacI-IPTG based regulatory system.
|
Funding |
|---|
|
TopAbstractIntroductionMaterials and methodsResults and discussionFundingAcknowledgementsReferences |
|---|
O.S. thanks Council of Scientific and Industrial Research for the Research Fellowship.
|
Footnotes |
|---|
Edited by Dr Jacques Fastrez
|
Acknowledgements |
|---|
|
TopAbstractIntroductionMaterials and methodsResults and discussionFundingAcknowledgementsReferences |
|---|
Authors thank Dr Manjula Reddy, Centre for Cellular and Molecular Biology, for discussions and helpful comments at various stages of the work.
|
References |
|---|
|
TopAbstractIntroductionMaterials and methodsResults and discussionFundingAcknowledgementsReferences |
|---|
Barkley M.D., Bourgeois S. The Lac Operon—Miller J.H., Reznikoff W.S., eds. (1978) Cold Spring Harbor Laboratory.
Barry J.K., Matthews K.S. Biochemistry (1999) 38:3579–3590.
Bell C.E., Lewis M. Curr. Opin. Struct. Biol. (2001) 11:19–25.[CrossRef][Medline]
Bloom J.D., Meyer M.M., Meinhold P., Otey C.R., MacMillan D., Arnold F.H. Curr. Opin. Struct. Biol. (2005) 15:447–452.[CrossRef][Web of Science][Medline]
Chakerian A.E., Pfahl M., Olson J.S., Matthews K.S. J. Mol. Biol. (1985) 183:43–51.[Medline]
Chou W.Y., Bieber C., Matthews K.S. J. Biol. Chem. (1989) 264:18309–18313.
Dubendorff J.W., Studier F.W. J. Mol. Biol. (1991) 219:45–59.[CrossRef][Web of Science][Medline]
Flynn T.C., Swint-Kruse L., Kong Y., Booth C., Matthews K.S., Ma J. Protein Sci. (2003) 12:2523–2541.[CrossRef][Web of Science][Medline]
Friedman B.E., Olson J.S., Matthews K.S. J. Mol. Biol. (1977) 111:27–39.[CrossRef][Web of Science][Medline]
Glascock C.B., Weickert M.J. Gene (1998) 223:221–231.[CrossRef][Medline]
Grabherr R., Nilsson E., Striedner G., Bayer K. Biotechnol. Bioeng. (2002) 77:142–147.[CrossRef][Medline]
Kercher M.A., Lee P., lewis M. Curr. Opin. Struct. Biol. (1997) 7:76–88.[CrossRef][Web of Science][Medline]
Lewis M., ed. The Lac Repressor (2005) Vol. 328. C.R. Biologics. 521–548.
Lewis M., Chang G., Horton N.C., Kercher M.A., Pace H.C., Schumacher M.A., Brennan R.G., Lu P. Science (1996) 271:1247–1254.[Abstract]
Marvin J.S., Hellinga H.W. Struct. Biol. (2001) 8:795–798.
Miller J.H. J. Mol. Biol. (1979) 131:249–258.[CrossRef][Medline]
Muller H.J. Mutat. Res. (1964) 106:2–9.[Medline]
Muller-Hartmann H., Muller-Hill B. J. Mol. Biol. (1996) 257:473–478.[CrossRef][Medline]
Riggs A.D., Bourgeois S., Cohn M. J. Mol. Biol. (1970) 53:401–417.[CrossRef][Web of Science][Medline]
Santos P.M., Di Bartolo I., Blatny J.M., Zennaro E., Valla S. FEMS Microbiol. Lett. (2001) 195:91–96.[CrossRef][Web of Science][Medline]
Stemmer W.P. Nature (1994) 370:389–391.
Studier F.W., Moffatt B.A. J. Mol. Biol. (1986) 189:113–130.[CrossRef][Web of Science][Medline]
Suckow J., Markiewicz P., Kleina L.G., Miller J., Kisters-Woike B., Muller-Hill B. J. Mol. Biol. (1996) 261:509–523.[CrossRef][Web of Science][Medline]
Swinte-Kruse L., Zhan H., Fiarbanks B.M., Maheshwari A., Matthews K.S. Biochemistry (2003) 42:14004–14016.
Swint-Kruse L., Zhan H., Matthews K.S. Biochemistry (2005) 44:11201–11213.
Williams S.G., Cranenburgh R.M., Weiss A.M., Wrighton C.J., Sherratt D.J., Hanak J.A. Nucleic Acids Res. (1998) 26:2120–2124.
Wilson C.J., Zhan H., Swint-Kruse L., Matthews K.S. Cell Mol. Life Sci. (2007) 64:3–16.[CrossRef][Web of Science][Medline]
Received June 9, 2008; revised October 18, 2008; accepted October 22, 2008.
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  R. Daber and M. Lewis Towards evolving a better repressor Protein Eng. Des. Sel., November 1, 2009; 22(11): 673 - 683. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||