PEDS Advance Access originally published online on July 20, 2006
Protein Engineering Design and Selection 2006 19(9):391-400; doi:10.1093/protein/gzl023
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Consensus guided mutagenesis of Renilla luciferase yields enhanced stability and light output
1 The CRUMP Institute for Molecular Imaging, Department of Molecular and Medical Pharmacology, Geffen School of Medicine at UCLA Los Angeles, CA 2 Molecular Imaging Program at Stanford, Department of Radiology and Bio-X Program, Stanford University Stanford, CA 3 Department of Bioengineering, Stanford University Stanford, CA 4 Department of Molecular and Cellular Physiology, Stanford University Stanford, CA, USA
5To whom correspondence should be addressed. Stanford University School of Medicine, Department of Radiology and Bio-X Program, The James H. Clark Center, 318 Campus Drive, Clark E150, Stanford, CA 94305-5427, USA E-mail: sgambhir{at}stanford.edu
 |
Abstract |
|---|
 Top Abstract IntroductionMaterials and methodsResultsDiscussionAcknowledgementsReferences |
|---|
Luciferases, which have seen expansive employment as reporter genes in biological research, could also be used in applications where the protein itself is conjugated to ligands to create probes that are appropriate for use in small animal imaging. As the bioluminescence activity of commonly used luciferases is too labile in serum to permit this application, specific mutations of Renilla luciferase, selected using a consensus sequence driven strategy, were screened for their ability to confer stability of activity in serum as well as their light output. Using this information, a total of eight favorable mutations were combined to generate a mutant Renilla luciferase (RLuc8) that, compared with the parental enzyme, is 200-fold more resistant to inactivation in murine serum and exhibits a 4-fold improvement in light output. Results of the mutational analysis were also used to generate a double mutant optimized for use as a reporter gene. The double mutant had half the resistance to inactivation in serum of the native enzyme while yielding a 5-fold improvement in light output.These variants of Renilla luciferase, which exhibit significantly improved properties compared with the native enzyme, will allow enhanced sensitivity in existing luciferase-based assays as well as enable the development of novel probes labeled with the luciferase protein.
Keywords: enzyme stability/mutagenesis/Renilla luciferase/reporter gene
|
Introduction |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionAcknowledgementsReferences |
|---|
Since the cloning of a luciferase from the firefly Photinus pyralis in 1985 (Wet et al., 1985
), luciferase genes have become essential components of biological research. They are used ubiquitously as reporter genes in cell culture experiments, and their use as reporters has been extended into the context of small animal imaging (Contag et al., 1997).
Recently, it has been proposed that the luciferase protein itself could be conjugated to other proteins such as antibodies or growth factors (Park and Gambhir, 2005), and these bioluminescently labeled ligands could then be used for imaging receptor targets in small animals. The advantage of using a bioluminescent entity to label a protein over similar fluorescent or radioactive approaches is that in the context of small animal imaging the bioluminescent approach has the potential to be more sensitive (Wu et al., 2001).
The beetle luciferases (e.g. firefly), however, are not optimal for employment as bioluminescent tags. These luciferases are not particularly small (
62 kDa) and are dependent on ATP, molecular oxygen, and magnesium for activity. The dependence on ATP especially would hinder the application of beetle luciferases as bioluminescent tags in vivo, as serum ATP concentrations are generally below 10 nM (Yegutkin et al., 2003).
Luciferases that use coelenterazine as their substrate are more appropriate for application as bioluminescent tags, as these enzymes are not ATP dependent and in general require only molecular oxygen in addition to coelenterazine for luminescence. From this group of proteins, the luciferase from the sea pansy Renilla reniformis (RLuc) (Matthews et al., 1977; Lorenz et al., 1991) is the best characterized as well as being of a size (36 kDa) more appropriate for use as a tag.
The limiting factor for use of RLuc as a bioluminescent tag is the rapid inactivation (
1/2 = 0.5–1 h) of its activity in murine serum at 37°C (see Results). A single point mutation of RLuc (C124A) that increases the enzyme's resistance to inactivation 6-fold has been reported (Liu and Escher, 1999), however, even this level of resistance is insufficient for the tagging of large proteins (e.g. antibodies) that require time scales on the order of days to sufficiently distribute. For this reason, we pursued a semi-rational, consensus sequence driven mutagenesis strategy (Lehmann et al., 2002) in order to identify mutations that could alter the stability of RLuc's activity in serum. Through the course of the present work, we also identified mutations that increased the light output of RLuc, particularly when used with analogs of coelenterazine. By combining these mutations appropriately, we were able to generate a mutant RLuc with enhanced light output and resistance to serum inactivation that is optimized for use as a bioluminescent tag, as well as a mutant RLuc with increased light output and accelerated inactivation that may find use as a reporter gene.
|
Materials and methods |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionAcknowledgementsReferences |
|---|
Materials
Coelenterazine was from Prolume (Pinetop, AZ, USA). Benzyl-coelenterazine (coelenterazine-h) was a generous gift from Dr Bruce Bryan. Coelenterazine-n and coelenterazine-cp were from Biotium (Hayward, CA, USA). Bisdeoxycoelenterazine (coelenterazine-400a, di-dehydro coelenterazine, DeepBlueC) was from Perkin Elmer (Boston, MA, USA). The chemical structures of these compounds are shown in Figure 1. Coelenterazine and the analogs were dissolved in propylene glycol and stored in small aliquots at –80°C.
|
Luminometer calibration
Light measurements were made using a Turner 20/20 luminometer and later using a Turner 20/20n luminometer (Turner Designs, Sunnyvale, CA, USA). The luminometers were calibrated to absolute units (photons/s) using the luminol light standard performed in dimethyl sulfoxide (DMSO) (Lee et al., 1966; O'Kane and Lee, 2000). No corrections were applied for the spectral sensitivity of the luminometer, as the spectral peak of luminol chemiluminescence in DMSO (486 nm) is close to the spectral peak of Renilla luciferase bioluminescence (482 nm).
Computational prediction
A BLAST search (Altschul et al., 1997) was performed using the ‘nr’ subset of GenBank, and all sequences with an E-value of <1–10 were initially retained. Redundant sequences were then removed, as well as sequences that did not contain a conserved catalytic triad known to be required for activity in bacterial haloalkane dehalogenases. An alignment between RLuc and the 14 remaining sequences was then generated using CLUSTAL W (Thompson et al., 1994).
A homology model of RLuc was built with SWISS-MODEL (v3.5) (Schwede et al., 2003) using the default parameters (Figure 4a). In generating this homology model, SWISS-MODEL utilized several crystal structures of the haloalkane dehalogenase LinB from Sphingomonas paucimobilis (PDB files 1iz8, 1k63, 1k6e, 1iz7, and 1mj5).
Construction of Renilla luciferase mutants
The hrluc gene from the plasmid phRL-CMV (Promega, Madison, WI, USA) was used as the initial template for cloning. This gene is a human codon usage optimized version of rluc and encodes a protein identical to RLuc with the exception of a T2A substitution. To construct a bacterial expression plasmid, PCR was used to remove the stop codon and to replace the N-terminal methionine codon with a pelB leader sequence. The pelB leader sequence, consisting of the first 22 codons of the pectate lyase B gene from Erwinia carotovora (Lei et al., 1987), directs protein expression into the bacterial periplasm and is cleaved from the final protein product. Using NcoI and HindIII restriction sites, the PCR product was inserted into the pBAD/Myc-His A plasmid (Invitrogen, Carlsbad, CA, USA), which adds a Myc epitope, a 6xHis tag, and a stop codon to the C-terminus of the gene. In some later constructs, the plasmid's SalI site was used for insertion in order to remove the Myc epitope from the construct. Site-directed mutagenesis was performed using a QuikChange II XL kit (Stratagene, La Jolla, CA, USA). When needed, cytoplasmic expression plasmids were generated by reverting the pelB leader sequence back to a methionine codon using PCR. All constructs and mutations were confirmed by sequencing.
Protein production and purification
All proteins were produced from the periplasmic (pelB containing) expression plasmids unless otherwise noted. To express proteins periplasmically, plasmid-containing Escherichia coli LMG 194 cells were grown at 32°C in Terrific Broth. Cultures were allowed to reach an OD600 of 0.7 and then induced by addition of L-(+)-Arabinose to a final concentration of 0.2%. Cells were harvested 12–14 h later and the periplasm extracted by osmotic shock (Neu and Heppel, 1965).
The periplasmic fraction was brought to the same concentration as the wash buffer (WB: 300 mM NaCl, 20 mM HEPES, 20 mM imidazole, pH 8) using a 10x stock, and phenylmethylsulphonylfluoride (PMSF) was added to 1 mM. The solution was clarified by 0.2 µm filtration and passed over a nickel affinity column (Ni-NTA Superflow, Qiagen, Valencia, CA, USA). The column was washed with WB and eluted with elution buffer (EB: 300 mM NaCl, 20 mM HEPES, 250 mM imidazole, pH 8). Protein concentration measurements were made using the Bradford assay (Bradford, 1976) with human serum albumin (HSA: Baxter Healthcare Corporation, Glendale, CA, USA) as the standard. Aliquots were taken at this point for gel electrophoresis (Figure 2). To the remainder of the elution, HSA was added to 1% as a carrier protein. All samples were stored at 4°C.
|
For those instances in which cytoplasmic expression plasmids were used, the cells were grown as described above. After harvesting, the cell pellet was frozen; thawed in WB containing 1 mg/ml lysozyme, 10 µg/ml RNase A, and 5 µg/ml DNase I; sonicated; and centrifuged to remove debris. These lysates were then cleared by filtration and purified by nickel affinity chromatography as described above.
Characterization of Renilla luciferase mutants
Luciferase activity was measured by adding 1 µl of the sample (diluted as necessary in EB containing 1% HSA) to 100 µl room temperature 100 mM sodium phosphate buffer (pH 7) (Sørensen, 1909), adding 1 µl of 0.5 µg/µl coelenterazine or analog, manually mixing, and reading for 10 s in a luminometer. The time between the addition of the luciferin and the start of measurement was 4 s.
Measurements of inactivation in serum were carried out by mixing 0.5 µl dilute luciferase with either 20 µl mouse serum or 50 µl rat serum (Equitech-Bio, Kerrville, TX, USA), placing the sample in a 37°C incubator, and removing aliquots for activity testing. To calculate the serum inactivation half-life, a mono-exponential decay model was fit to the data using a Nelder/Mead Simplex non-linear least squares minimization algorithm provided by the Octave numerical programming language (Eaton, 1997). Emission spectra at ambient temperature were measured using a Triax 320 (Horiba Jobin Yvon, Edison, NJ, USA), which incorporates an optical grating device with a liquid N2-cooled CCD detector.
Protein size and monodispersity were confirmed using a Superdex 200 analytical grade gel-filtration column (GE/Amersham Biosciences, Piscataway, NJ, USA) followed by in-line multi-angle light scattering and refractive index detectors (DAWN EOS and Optilab DSP, Wyatt Technologies, Santa Barbara, CA, USA). A dn/dc value of 0.185 ml/g was assumed in all calculations, and all processing was performed using the ASTRA software package (Wyatt Technologies).
For quantum yield measurements, separate 1 µl drops of protein (
2 pmol) and substrate (0.2 pmol) were placed in a tube, 100 µl of 100 mM sodium phosphate buffer (pH 7) was injected by the luminometer to mix, and the total light output was integrated (generally 5–10 min). For coelenterazine-n, the protein amount was increased 10-fold and the acquisition time lengthened to ensure that the reaction approached completion.
Kinetics
Kinetics were assessed by injecting 100 µl of 100 mM sodium phosphate buffer (pH 7) containing coelenterazine onto 1 µl of protein (diluted appropriately in EB containing 1% HSA), and recording the light output for 20 min. The final coelenterazine concentrations tested were 118, 24, 4.7, 0.94, 0.19, and 0.038 µM. The final luciferase concentrations were in the range of 1–7 pM. Coelenterazine absorbance was corrected for, although this was only significant for the highest concentration (10% attenuation). The values were converted from photons/s to molecules/s using the data from the quantum yield measurements, converted from flux units to mass units via integration, and processed using the kinetic curve fitting program Dynafit (Kuzmic, 1996).
Mammalian expression
In order to construct mammalian expression vectors, bacterial expression vectors containing the desired mutations were used as templates for PCR, with primers designed such that a methionine codon replaced the N-terminal pelB sequence and a C-terminal stop codon replaced the Myc epitope and 6xHis tag. The primers also contained appropriate NheI and HindIII restriction sites to allow insertion of the product into the pcDNA 3.1 plasmid (Invitrogen). The resultant plasmids were transiently transfected using SuperFect (Qiagen) into 293T cells (DuBridge et al., 1987) growing in 24-well plates following the manufacturer's protocol. Transfection medium was replaced with fresh medium (Dulbecco's modified Eagle medium containing 10% fetal bovine serum) after 3 h. At several time points, cells were lysed using passive lysis buffer (Promega), measured for total protein content using the Bradford assay, and assessed for luciferase activity using coelenterazine in the same manner as described above for bacterially expressed luciferase. Intracellular stability of the luciferases was assessed by adding cycloheximide to 100 µg/ml and lysing cells at several time points thereafter. Westerns were run on lysates with a monoclonal antibody to RLuc (MAb 4400, Chemicon, Temecula, CA, USA) in order determine the luciferase protein content, with purified bacterially produced RLuc8 used as the standard.
|
Results |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionAcknowledgementsReferences |
|---|
Computational predictions for Renilla luciferase
Via sequence similarity searches, RLuc was predicted to contain a characteristic 
/ß-hydrolase fold from around amino acid 71–301 (Marchler-Bauer et al., 2003) and was found to have a high level of similarity to a number of proteins, most of which are known or putative haloalkane dehalogenases. An alignment between these sequences is shown in Figure 3, and a homology model is presented in Figure 4a.
|
|
0.05 using a two-tailed t-test with the incorporation of a Bonferroni–Holms correction for multiple comparisons (Holm, 1979Mutagenesis of Renilla luciferase and screening
In the hope of further enhancing the resistance to inactivation of RLuc beyond that achieved with the C124A mutation [denoted C152A in (Liu and Escher, 1999)], a number of further mutations were explored. Candidate mutations were chosen from the alignment data at positions where RLuc most clearly diverged from the consensus sequence. For instance, the candidate mutation A55T was chosen because RLuc harbors the aliphatic amino acid alanine at position 55, while nearly all the other proteins harbor a hydroxylic residue of either threonine or serine at this site. Similarly, S287L was chosen as a candidate because RLuc contains a hydroxylic residue at this position, differing from the consensus aliphatic residue. Some of the candidates, such as M253L, are less obvious. This mutation substitutes an aliphatic residue for another aliphatic, but brings the RLuc sequence into consensus with the highly conserved local sequence near this position.
Complete results with respect to activity, inactivation in serum, and emission spectra peaks are summarized in Table I for 25 initial mutations, performed on a background of RLuc with the C124A mutation. Note that activity was defined as a 10 s integration of the light output curve in order to disfavor mutations that merely increased the burst value at the expense of total light output. Representative data for inactivation in serum and emission spectra are shown in Figure 4b and c, respectively.
|
The assayed values for RLuc reported in Table I corresponded well to previous values reported in the literature. In terms of the stability of activity under serum-like conditions, the values reported here for recombinant RLuc (
1/2 = 0.4–0.9 h) are in line with Liu et al. (1997), who reported a half-life of 0.6 h for recombinant RLuc in hamster blood at 37°C, as well as Lorenz et al. (1993), who reported a half-life of 0.5 h for recombinant RLuc in a high ionic strength buffer. The measured emission peak for RLuc with coelenterazine (482 nm) corresponded exactly to the previously published value of 482 nm for RLuc purified directly from R.reniformis (Hart et al., 1979).
Peak light flux from recombinant RLuc was determined to be (1.2 ± 0.2) x 1023 photons/s/mole enzyme when in the presence of 24 µM coelenterazine. This value corresponds acceptably to the value of 6.5 x 1022 photons/s/mole enzyme reported for RLuc purified directly from R.reniformis (Matthews et al., 1977) and of 9 x 1022 photons/s/mole enzyme reported for recombinant RLuc (Lorenz et al., 1993).
Combining mutations for a luciferase resistant to serum inactivation
For the purpose of generating a mutant RLuc more appropriate for use as a bioluminescent tag in small animal imaging applications, the initial mutations were judged for their ability to confer resistance to serum inactivation as well as their light output. After excluding A54P due to its strong negative effect on light output and F116L as it yielded only a small increase in resistance to serum inactivation, the seven remaining mutations that exhibited either increased light output or enhanced resistance to serum inactivation were combined along with the C124A mutation into a single protein designated as ‘RLuc8’. The eight mutations present in RLuc8 are A55T, C124A, S130A, K136R, A143M, M185V, M253L, and S287L. Since the Myc epitope was removed during the cloning of RLuc8, a C124A mutant was constructed without the Myc epitope to facilitate a valid comparison (C124A-
Myc). The activity, resistance to serum inactivation, and spectra peak values for these two enzymes are shown in Table I. When compared with the native enzyme, RLuc8 exhibited a greater than 4-fold enhancement in activity, a 200-fold increased resistance to serum inactivation, and a small but measurable 5 nm red shift in the emission spectrum. Compared with the C124A mutant, RLuc8 showed a 3-fold increase in activity and at least a 13-fold improved resistance to inactivation in murine serum.
Light scattering results suggest that RLuc8 exists as a monomer in solution, as molar mass moment calculations based on the multi-angle scattering indicate a molecular weight of 33.8 kDa (error: 7%) with a relatively low polydispersity across the gel-filtration elution profile (11%).
Comparison with cytoplasmically expressed protein
In order to assess what effects the oxidative environment of the bacterial periplasm may be having on the proteins, RLuc and RLuc8 were expressed in the reducing environment of the bacterial cytoplasm using expression plasmids that did not include the pelB leader sequence. Compared with the periplasmically expressed RLuc presented in Table I, cytoplasmically expressed RLuc had 25% greater activity with similar serum inactivation half-lives (0.8 h and 0.4 h for mouse and rat serum, respectively). Cytoplasmically expressed RLuc8 was also purified and had a similar activity (within 10%) and serum inactivation half-lives (290 and 65 h in mouse and rat serum, respectively) to the periplasmically expressed RLuc8.
Quantum yield and kinetic parameters of mutants
To understand the basis for RLuc8's higher activity, both quantum yield and kinetic measurements were undertaken. The results shown in Table II indicated that RLuc8 had a 30% improvement in quantum yield for native coelenterazine and a 30-fold increase in quantum yield for bisdeoxycoelenterazine. A Michaelis–Menten model was fit to initial reaction velocity data for coelenterazine concentrations in the range of 0.038–24 µM. The results for RLuc, the C124A mutant, and RLuc8 were Km = 2.9 ± 1.0, 2.7 ± 0.8, 1.6 ± 0.2 µM, and kcat = 3.9 ± 0.4, 4.7 ± 0.4, 4.9 ± 0.1 s–1, respectively, with the errors presented representing the formal standard errors of the fitted parameters. The results for RLuc are roughly consistent with a previously published Km value of 2 µM for RLuc in the presence of benzyl-coelenterazine (Matthews et al., 1977).
|
Mutations to test proposed active site
Based on the catalytic triad of residues known to be critical for activity in the haloalkane dehalogenases, it was predicted that D120, E144, and H285 would be required for activity in Renilla luciferase as well. The locations of these residues in a homology model of Renilla luciferase are shown in Figure 4a. To test the hypothesis that these residues comprise a portion of the enzyme's active site, further mutations were made at these sites on the RLuc8 construct, with the results shown in Table I. With respect to maintaining luciferase activity, mutations at these proposed active site residues were deleterious.
Combining mutants for a luciferase prone to inactivation
In order to construct brighter mutants that were less resistant to serum inactivation, the initial double mutants were compared with the single mutant C124A to identify mutations that led to increased activity without increasing resistance to serum inactivation (e.g. M185V) or increased serum inactivation without affecting initial light output (e.g. Q235A, S257G). Combining these mutations in the absence of C124A resulted in the mutants M185V and M185V/Q235A (Table I) that showed both increased inactivation in serum as well as enhanced activity in comparison with RLuc (Table I).
Testing of mutants in mammalian expression
In order to determine whether the in vitro data gathered for the RLuc mutants and RLuc8 would translate into the context of a mammalian reporter gene, expression vectors were constructed for RLuc, C124A, C124A/M185V, M185V, M185V/Q235A, and RLuc8 in a pcDNA 3.1 backbone. These mammalian expression plasmids were then transiently transfected into 293T cells. Measurements of light output over time, as shown in Figure 4d with respect to the RLuc plasmid, demonstrated that the mutations conferred increased light output following transfection in mammalian cells. A cycloheximide study was performed to assess the resistance to inactivation for the luciferase variants in the context of the mammalian cytoplasm. As shown in Figure 4e, the relative differences in inactivation resistance, but not the absolute differences, were consistent with the serum inactivation experiments. Through densitometry measurements of western blots performed on the cell lysates, the amount of luciferase was estimated for the different conditions and used to calculate the specific activity values shown in Table III. These values were roughly consistent with the in vitro data for bacterially expressed protein.
|
|
Discussion |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionAcknowledgementsReferences |
|---|
The amino acid sequence of R.reniformis luciferase, along with that of the closely related Renilla mülleri luciferase, contains a characteristic
/ß-hydrolase fold sequence (Holmquist, 2000). This fold pattern is found in enzymes that catalyze a diverse range of chemical reactions in all kingdoms of life. Interestingly, within the /ß-hydrolase family RLuc shows a high level of similarity to the bacterial haloalkane dehalogenases, enzymes that catalyze the hydrolytic detoxification of halogenated compounds. This similarity even extends to the conserved haloalkane dehalogenase catalytic triad, present as D120, E144, and H285 in RLuc, being required for appreciable luciferase activity. The level of similarity is unexpected as RLuc is an oxygenase and R.reniformis is not a bacterium.
The evolution of a coelenterazine using oxygenase from an enzyme that catalyzes an unrelated reaction would not, in fact, present a considerable challenge for evolution. Coelenterazine chemiluminesces easily in aprotic solutions, and an initial enzyme would have to provide little more than a hydrophobic environment for coelenterazine to achieve some low level of bioluminescence (Rees et al., 1998). The high level of primary sequence similarity between RLuc, the bacterial haloalkane dehalogenases, and the protein sequence derived from the Strongylocentrotus purpuratus genome is harder to explain. As the bacteria, sea pansies (e.g. Renilla species), and sea urchins (e.g. S.purpuratus) diverged at least 500 million years ago, horizontal gene transfer could be one possible explanation for this similarity. However, this would seem to imply two separate gene transfer events for the ancestral luciferase gene, which seems rather unlikely. On another note, it has not escaped our attention that the gene from S.purpuratus could potentially be a luciferase, a finding that might explain the blue bioluminescence emission that has been observed in the fertilized eggs of this species (Hewes, 1977).
In a 1999 paper, Liu and Escher reported results from sequentially mutating the three cysteines to alanine in a version of Renilla luciferase engineered for mammalian cell secretion (Liu and Escher, 1999). Their hypothesis was that, in a secreted version of Renilla, these cysteines would have a propensity to form inactivating disulfide bonds in the oxidizing environment of the protein secretion pathway. They reported a complete loss of activity when mutating out the second cysteine residue (C73). Interestingly, many of the aligned proteins contain a cysteine near this location (Figure 3). They also reported enhanced stability of the enzyme when the third cysteine (C124) was replaced and suggested that this was due to the blocking of unintended disulfide bond formation in oxidative environments. The homology model, however, indicates that the third cysteine is buried and removal of a potential disulfide bond is unlikely to explain the increased stability seen for this mutant after protein folding. More likely, the C124A mutation increases stability by allowing better packing of the hydrophobic core. Interestingly, the alignment data shows, that an alanine is favored at this position, coincidentally the amino acid chosen to substitute by Liu and Escher.
Liu and Escher also suggested that a disulfide bond formed under oxidative conditions between C24 and C73 could increase the resistance of the enzyme to inactivation and showed a number of experiments consistent with this hypothesis. We, however, did not see any significant changes with respect to inactivation between protein produced in the reducing environment of the cytoplasm and the oxidative environment of the periplasm. Cytoplasmically produced RLuc was found to have a 25% higher activity than the same protein produced periplasmically, but this is most likely due to some amount of activity loss during the comparatively longer periplasmic purification process. We have been unable to reconcile the discrepancies between our results and those of Liu and Escher, although they may in part arise due to differences between the mammalian secreted expression system used by Liu and Escher and the bacterial expression systems used in the present study.
In any case, the interpretation of the Liu and Escher results for the C124A mutation in light of the protein alignment data led us to examine a mutation strategy in which candidate mutations would be picked at locations where RLuc clearly diverged from the consensus sequence. This strategy was initially predicated upon the hypothesis that mutations would accumulate in the ancestral RLuc gene that are either neutral only in the specific context of the Renilla lumisomes or required for the association with and resonance energy transfer to the green fluorescent protein (GFP) homo-dimer that is the normal light emitter in Renilla (Ward and Cormier, 1978).
Previous studies have utilized similar consensus guided semi-rational mutagenesis strategies (Steipe et al., 1994; Wirtz and Steipe, 1999; Lehmann et al., 2002; Amin et al., 2004), albeit with differing experimental methods utilized as measures of stability (e.g. thermal denaturation, chemical denaturing, proteolytic cleavage). The common rationale behind all these studies, equally applicable to the study here, is that evolution tends to disfavor amino acids that destabilize a protein (Kimura and Ota, 1974). By looking at a family of similar proteins, one can pick candidate mutations that have already been screened by nature and are therefore less likely to be deleterious to the protein than a residue picked at random.
Three of the mutations, K136R, M185V, and S287L, showed sizable increases in the light output of the enzyme. In the case of M185V, a portion of the increase in light output can be explained by enhanced quantum yield, particularly for the coelenterazine analogs, with the difference assumed to arise from enhanced kinetics. The disproportionate increase in light output seen with M185V for several of the coelenterazine analogs leads us to speculate that the light increase and specificity decrease are related and that M185 may be positively selected for in RLuc and R.mülleri luciferase to ensure specificity of the reaction. This trade off between substrate recognition and light output could arise if the 185 position was important in substrate recognition. The alignment data gives some credence to our hypothesis, as M185 is located in the ‘cap’ of the enzyme, a domain often used for substrate specificity in the haloalkane dehalogenases (Holmquist, 2000). The location of M185 in the homology model lends further support, as its placement atop the presumptive catalytic site (formed by D120, E144, and H285) would be an appropriate position for conferring substrate specificity to the luciferase.
S287L, while not located in the active site, is close enough to the active site residue H285 that slight alterations in the structure of the enzymatic pocket induced by S287L could explain the modest increase in quantum yield and the larger increase in light output seen with this mutation.
K136, on the other hand, is located on the surface of the protein 20 Å distant from the presumptive active site. It is still possible that the increased light output seen with the K136R mutation could arise through long-range perturbations in the folding of the active site, especially as K136 lies in a loop between the active site residues D120 and E144. Incidentally, R.mülleri luciferase contains an arginine at this position.
An alternative hypothesis for the increases in light output, especially for M185 and S287, is that these are residues are somehow involved in the transfer of energy to the GFP homo-dimer that interacts with RLuc in nature. As such, these residues have been selected for their role in energy transfer to the GFP fluorophore and not for their ability to optimize the quantum yield of the luciferase on its own.
Using our standard assay, RLuc8 displayed a 4-fold increase in light output versus RLuc. For RLuc8, combining the increases in quantum efficiency with the enhanced Michaelis–Menten parameters would only predict an 70% increase in light output over RLuc. In addition, the Michaelis–Menten model could not satisfactorily fit the reaction progress curves, or the initial reaction velocity at the highest coelenterazine concentration tested (118 µM), even when product inhibition was incorporated into the model. Previous attempts at elucidating a satisfactory kinetic model for RLuc have failed (Matthews, 1976). We speculate that a complete kinetic model would need to factor in coelenterazine-dependent inactivation of RLuc, a phenomenon known to occur rapidly in anoxic conditions (Matthews et al., 1977).
Transient transfection of plasmids containing RLuc, C124A, C124A/M185V, M185V, M185V/Q235A, and RLuc8 into mammalian cells demonstrated that the basic trends derived from the in vitro mutation analysis are applicable in the context of a mammalian cell as well. The rate of RLuc inactivation in the mammalian cells (Figure 4e) was an order of magnitude slower than that observed in serum (Table I), which is not surprising given that RLuc is an intracellular eukaryotic enzyme in its native environment. However, the relative inactivation trends between the different luciferase variants were consistent for both sets of experiments. For instance, M185V/Q235A inactivated approximately twice as fast as RLuc whether the experiment was performed in serum or in mammalian cells.
The specific activities measured from the mammalian transfection experiment (Table III) were roughly consistent (within 30–50%) with the activities measured for the bacterially expressed proteins. The differences in the absolute values between the two sets of experiments are most likely due to errors in estimating the amounts of luciferase in the mammalian cellular lysates and due to the fact that the mammalian cellular lysates include a mix of active and inactivated protein accumulated during the course of the experiment.
Luciferases are extraordinarily useful in a variety of experiments that require reporter genes. In instances where the reporter gene is constitutively expressed [e.g. cell trafficking studies (Beilhack et al., 2005)], RLuc8 should be advantageous because of its greatly increased light output compared with RLuc in mammalian cells (Figure 4d). This increased light output stems from both RLuc8's increased specific activity (Table III) and its decreased rate of inactivation (Figure 4e).
In most reporter gene experiments, however, the investigator wishes to follow the dynamics of gene induction and suppression. In these contexts, a slow rate of inactivation of the reporter would be a detriment to the experiment, as the stability of the signal would obscure transient changes in gene expression. The single mutant M185V and the double mutant M185V/Q235A should be of utility in these cases, as both these mutants show an increase in specific activity as well as an increase in the rate of protein inactivation relative to RLuc.
One issue with the use of coelenterazine catalyzing luciferases for reporter gene assays in mammalian cells is that coelenterazine is a substrate for MDR1 P-glycoprotein (Pgp) (Pichler et al., 2004). While the resultant transport of coelenterazine out of mammalian cells can be used to measure levels of Pgp, in most studies this phenomenon leads to an inadvertent modulation of signal intensity. For this reason, there has been interest in the coelenterazine analogs coelenterazine-cp and coelenterazine-n as they are not substrates for Pgp (Pichler et al., 2004). These analogs, however, suffer from reduced light output when used with RLuc (2-fold for coelenterazine-cp, 4-fold for coelenterazine-n, see Table I) as well as 4-fold higher background rates of auto-chemiluminescence (Zhao et al., 2004). Combined, these factors lead to a drop in the signal-to-background ratio of 8- and 16-fold for coelenterazine-cp and coelenterazine-n, respectively. The M185V mutation greatly reduces the disadvantages of these alternative substrates. In the case of coelenterazine-cp, the M185V mutation leads to a signal-to-noise ratio that it is only a factor of 2 lower than that of the native substrate. RLuc8 is not as effective at using coelenterazine-cp as the M185V mutation alone, most likely because the A55T mutation present in RLuc8 decreases its ability to use this substrate.
Bisdeoxycoelenterazine has been proposed as a better analog to use with bioluminescence resonance energy transfer (BRET) studies because of the increased separation between the bioluminescence and the fluorescence spectra (Jensen et al., 2002). Bisdeoxycoelenterazine, however, suffers from extraordinary low light output when used with native RLuc (Table I) because of poor quantum yield (Table II). Both RLuc8 and the M185V mutation should be of great utility in these BRET assays, as they confer a 20- to 60-fold increase in light output with bisdeoxycoelenterazine. Interestingly, although C124A alone does not improve utilization of bisdeoxycoelenterazine, it appears to facilitate the M185V mutation, as C124A/M185V has an 2-fold better light output with this substrate compared with M185V alone.
In summary, we have characterized mutants of RLuc with respect to resistance to serum inactivation as well as light output and have used these results to develop luciferases optimized for different purposes. An eight mutation form of RLuc (RLuc8) was created that has greatly improved characteristics for use as a bioluminescent label. Compared with the native enzyme, RLuc8 exhibited a 200-fold improvement in resistance to murine serum inactivation, a 4-fold improvement in light output, and a 5 nm red shift in the emission spectrum. The enhancement in light output arises from a combination of increases in quantum yield and improved kinetics. A double mutant of RLuc (M185V/Q235A) was created that has improved performance as a reporter gene. Compared with the native enzyme it has twice the rate of inactivation, as measured in murine serum, while incorporating a close to 5-fold improvement in light output. These optimized Renilla luciferases represent significant improvements that will increase the sensitivity of luciferase-based assays for both in vitro experiments and in vivo imaging.
|
Footnotes |
|---|
Edited by Andrew Bradbury
|
Acknowledgements |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionAcknowledgementsReferences |
|---|
The authors would like to thank Drs Brad Rice and Olivier Coquoz of Xenogen Corporation for help with the spectra measurements, Dr Mark Breidenbach for assistance with the sequence alignment and homology modeling, Dr Dennis O'Kane for help with the luminol standardization, and Zachary Walls for comments on the manuscript. This work was supported in part by a Department of Defense NDSEG Fellowship (A.M.L.), a Stanford Bio-X Graduate Fellowship (A.M.L.), NIH 5 R01 CA082214 [GenBank] -06 (S.S.G.), and NIH ICMIC P50 CA114747 [GenBank] (S.S.G.).
|
References |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionAcknowledgementsReferences |
|---|
Altschul S.F., Madden T.L., Schaffer A.A., Zhang J., Zhang Z., Miller W., Lipman D.J. (1997) Nucleic Acids Res. 25:3389–3402.
Amin N., Liu A.D., Ramer S., Aehle W., Meijer D., Metin M., Wong S., Gualfetti P., Schellenberger V. (2004) Protein Eng. Des. Sel. 17:787–793.
Beilhack A., Schulz S., Baker J., Beilhack G.F., Wieland C.B., Herman E.I., Baker E.M., Cao Y.A., Contag C.H., Negrin R.S. (2005) Blood 206:1113–1122.
Bradford M. (1976) Anal. Biochem. 72:248–254.[CrossRef][ISI][Medline]
Contag C.H., et al. (1997) Photochem. Photobiol. 66:523–531.[ISI][Medline]
DuBridge R.B., Tang P., Hsia H.C., Leong P.M., Miller J.H., Calos M.P. (1987) Mol. Cell. Biol. 7:379–387.
Eaton J.W. (1997) GNU Octave: A High-Level Interactive Language for Numerical Computations Third edition , Network Theory, Clifton, UK.
Hart R.C., Matthews J.C., Hori K., Cormier M.J. (1979) Biochemistry 18:2204–2210.[CrossRef][Medline]
Hewes C.S. (1977) PhD Thesis, Sonoma State University.
Holm S. (1979) Scand. J. Stat. 6:65–70.
Holmquist M. (2000) Curr. Protein Pept. Sci. 1:209–235.[CrossRef][Medline]
Jensen A.A., Hansen J.L., Sheikh S.P., Bräuner-Osborne H. (2002) Eur. J. Biochem. 269:5076–5087.[ISI][Medline]
Kimura M. and Ota T. (1974) Proc. Natl Acad. Sci., 71:2848–2852.
Kuzmic P. (1996) Anal. Biochem 260–273.
Lee J., Wesley A.S., Ferguson J.F. III, Seliger H.H. (1966) In In Johnson F.H. and Haneda Y. (Eds.). Bioluminescence in Progress. , Princeton University Press: Princeton, NJ pp. 35–43.
Lehmann M., Loch C., Middendorf A., Studer D., Lassen S.F., Pasamontes L., van Loon A.P., Wyss M. (2002) Protein Eng. 15:403–411.
Lei S.P., Lin H.C., Wang S.S., Callaway J., Wilcox G. (1987) J. Bacteriol. 169:4379–4383.
Liu J. and Escher A. (1999) Gene 237:153–159.[CrossRef][ISI][Medline]
Liu J., O'Kane D.J., Escher A. (1997) Gene 203:141–148.[CrossRef][ISI][Medline]
Lorenz W.W., Gray J.P., Cormier M.J., Gibson B.G., O'Kane D.J. (1993) Bioluminescence and Chemiluminescence: Status Report; 7th International Symposium on Bioluminescence and Chemiluminescence. , Banff, Canada pp. 191–195.
Lorenz W.W., McCann R.O., Longiaru M., Cormier M.J. (1991) Proc. Natl Acad. Sci. 88:4438–4442.
Marchler-Bauer A., et al. (2003) Nucleic Acids Res. 31:383–387.
Matthews J.C. (1976) PhD Thesis, University of Georgia.
Matthews J.C., Hori K., Cormier M.J. (1977) Biochemistry 16:85–91.[CrossRef][Medline]
Matthews J.C., Hori K., Cormier M.J. (1977) Biochemistry 16:5217–5220.[CrossRef][Medline]
Neu H.C. and Heppel L.A. (1965) J. Biol. Chem. 240:3685–3692.
O'Kane D.J. and Lee J. (2000) Methods Enzymol. 305:87–96.[ISI][Medline]
Park J. and Gambhir S. (2005) Proc. IEEE, 93:771–783.[CrossRef]
Pichler A., Prior J.L., Piwnica-Worms D. (2004) Proc. Natl Acad. Sci. 101:1702–1707.
Rees J.F., de Wergifosse B., Noiset O., Dubuisson M., Janssens B., Thompson E.M. (1998) J. Exp. Biol. 201:1211–1221.[Abstract]
Schwede T., Kopp J., Guex N., Peitsch M.C. (2003) Nucleic Acids Res. 31:3381–3385.
Sørensen S.P.L. (1909) Biochem. Z. 22:352–356.
Steipe E., Schiller B., Plückthun A., Steinbacher S. (1994) J. Mol. Biol. 240:188–192.[CrossRef][ISI][Medline]
Thompson J.D., Higgins D.G., Gibson T.J. (1994) Nucleic Acids Res. 22:4673–4680.
Ward W.W. and Cormier M.J. (1978) Photochem. Photobiol., 27:389–396.
Wet J.R.D., Wood K.V., Helinski D.R., DeLuca M. (1985) Proc. Natl Acad. Sci. 82:7870–7873.
Wirtz P. and Steipe B. (1999) Protein Sci., 8:2245–2250.[Abstract]
Wu J.C., Sundaresan G., Iyer M., Gambhir S.S. (2001) Mol. Ther. 4:297–306.[CrossRef][ISI][Medline]
Yegutkin G.G., Samburski S.S., Jalkanen S. (2003) FASEB J. 17:1328–1330.
Zhao H., Doyle T.C., Wong R.J., Cao Y., Stevenson D.K., Piwnica-Worms D., Contag C.H. (2004) Mol. Imaging 3:43–54.[CrossRef][Medline]
Received October 5, 2005; revised March 3, 2006; accepted May 14, 2006.
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  Z. F. Walls, M. Puttaraju, G. F. Temple, and S. S. Gambhir A Generalizable Strategy for Imaging pre-mRNA Levels in Living Subjects Using Spliceosome-Mediated RNA Trans-Splicing J. Nucl. Med., July 1, 2008; 49(7): 1146 - 1154. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 J. Woo, M. H. Howell, and A. G. von Arnim Structure-function studies on the active site of the coelenterazine-dependent luciferase from Renilla Protein Sci., April 1, 2008; 17(4): 725 - 735. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 C. T. Chan, R. Paulmurugan, O. S. Gheysens, J. Kim, G. Chiosis, and S. S. Gambhir Molecular Imaging of the Efficacy of Heat Shock Protein 90 Inhibitors in Living Subjects Cancer Res., January 1, 2008; 68(1): 216 - 226. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 E. du Plessis, J. Theron, E. Berger, and M. Louw Evaluation of the Staphylococcus aureus Class C Nonspecific Acid Phosphatase (SapS) as a Reporter for Gene Expression and Protein Secretion in Gram-Negative and Gram-Positive Bacteria Appl. Envir. Microbiol., November 15, 2007; 73(22): 7232 - 7239. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
A. De, A. M. Loening, and S. S. Gambhir An Improved Bioluminescence Resonance Energy Transfer Strategy for Imaging Intracellular Events in Single Cells and Living Subjects Cancer Res., August 1, 2007; 67(15): 7175 - 7183. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
P. Ray, R. Tsien, and S. S. Gambhir Construction and Validation of Improved Triple Fusion Reporter Gene Vectors for Molecular Imaging of Living Subjects Cancer Res., April 1, 2007; 67(7): 3085 - 3093. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||