Codon-Optimized Luciola Italica Luciferase Variants for Mammalian Gene Expression in Culture and in Vivo

Abstract

Luciferases have proven to be useful tools in advancing our understanding of biologic processes. Having a multitude of bioluminescent reporters with different properties is highly desirable. We characterized codon-optimized thermostable green- and red-emitting luciferase variants from the Italian firefly Luciola italica for mammalian gene expression in culture and in vivo. Using lentivirus vectors to deliver and stably express these luciferases in mammalian cells, we showed that both variants displayed similar levels of activity and protein half-lives as well as similar light emission kinetics and higher stability compared to the North American firefly luciferase. Further, we characterized the red-shifted variant for in vivo bioluminescence imaging. Intramuscular injection of tumor cells stably expressing this variant into nude mice yielded a robust luciferase activity. Light emission peaked at 10 minutes post-D-luciferin injection and retained > 60% of signal at 1 hour. Similarly, luciferase activity from intracranially injected glioma cells expressing the red-shifted variant was readily detected and used as a marker to monitor tumor growth over time. Overall, our characterization of these codon-optimized luciferases lays the groundwork for their further use as bioluminescent reporters in mammalian cells.

BIOLUMINESCENCE IMAGING (BLI) using luciferase reporters has provided crucial information regarding many biologic processes, including tumorigenesis, bacterial pathogenesis, and transcription factor activation.^1–3 The major advantage of BLI compared to end-point analysis is that it provides real-time, noninvasive analysis of in situ biologic events, thereby giving a complete “picture” of the kinetics of an entire process. Great strides have been made since the seminal study by Contag and colleagues published in 1995, which was the first demonstration of in vivo BLI. ⁴ For example, as few as 10 cells expressing an optimized American firefly luciferase can be detected in mice. ⁵ Additional progress has been made by the discovery and codon optimization of the naturally secreted Gaussia princeps luciferase, allowing for sensitive detection of in vivo biologic processes by simple blood sampling. ⁶ The discovery of new luciferases and the genetic engineering of existing bioluminescence reporters that emit light at longer wavelengths, thus reducing tissue absorption of light, have enhanced the sensitivity of in vivo BLI.^7–9

The luciferase with the most biochemical characterization is from the North American firefly Photinus pyralis (Fluc). Fluc catalyzes a two-step reaction in which luciferin is first adenylated through an adenosine triphosphate (ATP)-dependent mechanism. Next, molecular oxygen is used to oxidize the adenylated intermediate yielding oxyluciferin, a molecule in an excited energy state. On decay of excited state oxyluciferin, a photon of light is emitted. ¹⁰ Owing to the high quantum yield of the luminescence reaction, Fluc was pursued as a reporter enzyme. Since its cloning and expression in mammalian cells in 1987, ¹¹ Fluc has become the “workhorse” of the molecular biology laboratory as a reporter for gene expression and ATP levels in cells^12,13 and noninvasive monitoring of in vivo processes using BLI technology. ¹⁴

Many other luciferases have been cloned and characterized for BLI in an attempt to increase detection sensitivity and dual parameter measurement. Luciferases that catalyze emission of red light (emission peak > 600 nm) ¹⁵ or red-shifted variants of existing luciferases ¹⁶ have improved the sensitivity of in vivo detection. ¹⁷ Renilla reniformis luciferase (Rluc) and Gaussia princeps luciferase (Gluc) use coelenterazine as a substrate, in contrast to Fluc; therefore, sequential imaging of Fluc and Gluc or Rluc in vivo can be used to monitor two different biologic processes. ¹⁸ Recently, the complementary deoxyribonucleic acid (cDNA) encoding the green-emitting luciferase from the Italian firefly Luciola italica (liFluc) was cloned by Branchini and colleagues, which has 64% amino acid identity with the luciferase from Photinus pyralis. ¹⁹ This luciferase has several advantages over Fluc, including thermostability and higher enzyme turnover. ¹⁹ Further, using standard mutagenesis, a red-shifted variant was characterized. ²⁰ liFluc has been expressed and purified from bacteria and used for cell-based assays ²⁰ ; however, it has not been established for mammalian gene expression and in vivo imaging. Herein, we characterize codon-optimized variants of both the green- and red-emitting liFluc for mammalian gene expression and validate the latter for in vivo BLI in small animals.

Materials and Methods

Cell Culture and Reagents

For this study, 293T human kidney fibroblast cells, U87 human glioma cells (both from American Type Culture Collection, Manassas, VA), and Gli36 human glioma cells (kindly provided by Dr. Anthony Capanogni, University of California at Los Angeles, Los Angeles, CA) were main tained in Dulbecco's Modified Eagle's Medium (DMEM) supplemented with 10% fetal bovine serum (Sigma, St. Louis, MO), 100 U/mL penicillin, and 0.1 mg/mL streptomycin (Sigma), referred to as complete DMEM. All cells were grown at 37°C in a 5% CO₂ humidified atmosphere.

Lentivirus Vector Construction

Human codon-optimized cDNA variants of both green-emitting Luciola italica luciferase (G-liFluc) and red-emitting (R-liFluc) were kindly provided by Dr. Rampyari Walia (Targeting Systems, El-Cajon, CA). These cDNAs were cloned into a lentivirus vector plasmid, CSCW. ²¹ Transgene expression in this plasmid is driven by a cytomegalovirus (CMV) promoter with the inclusion of an internal ribosomal entry site (IRES) for coexpression of mCherry red fluorescent protein to allow for titering and transduction confirmation. ⁶ The lentivirus encoding North American firefly luciferase (Fluc) and mCherry has been previously described. ²² Both G-liFluc and R-liFluc coding sequences were amplified by polymerase chain reaction using Pfu polymerase (Agilent Technologies, Santa Clara, CA) and the following primers: forward primer, 5′ATAGC-TAGCGATCCATGGAAACAGAAAG3′; reverse primer, 5′TA-CTCGAGACTACCCACCTGCTTGAGGT3′. The thermal cycler conditions were 1 cycle of 94°C for 2 minutes; 30 cycles of 94°C for 45 seconds, 60°C for 45 seconds, 72°C for 90 seconds; and 1 cycle of 72°C for 10 minutes. The forward primer was synthesized with an NheI site and the reverse with an XhoI restriction site for ligation with similarly digested CSCW-ImCherry plasmid. Constructs were named CSCW-GliFluc-ImCherry and CSCW-R-liFluc-ImCherry. A lentivirus vector encoding the Gaussia luciferase (Gluc) and the cerulean fluorescent protein (CFP) separated by an IRES (CSCW-Gluc-ICFP) was described previously. ²³ Lentivirus vectors were produced as described before, ²¹ and vector titers (transducing units/mL) were determined by performing serial dilutions of vector stocks on 293T cells followed by counting the number of mCherry or CFP-positive 293T cells 3 days later using fluorescence microscopy.

Lentivirus Vector Transduction

For side-by-side comparison of G-liFluc, R-liFluc, and Fluc luciferases, Gli36 and U87 human glioma cells were engineered to stably express both Gluc and one of these luciferases. Around 350,000 cells were plated in a six-well plate. The next day, cells were first transduced with the CSCW-Gluc-ICFP lentivirus using a multiplicitiy of infection of 50 in the presence of 10 μg/mL polybrene in 3 mL. The plate was centrifuged at 1,800 rpm for 1 hour at room temperature. Afterward, cells were cultured overnight. Seventy-two hours later, the same protocol was applied to transduce the same cells with either CSCW-RliFluc-ImCherry or CSCW-GliFluc-ImCherry. The CFP (a marker for cell number) and mCherry (marker for transduction efficiency) fluorescence intensities in both cell lines were quantified using a FlexStation 3 microplate reader (Molecular Devices, Sunnyvale, CA).

In Vitro Gaussia Luciferase Assay

For in vitro Gaussia luciferase assay, 10⁴ cells/well were plated in a 96-well plate. After 24 hours, cells were washed once with 50 μL phosphate-buffered saline (PBS) and lysed on the plate in 50 μL lysis buffer (Targeting Systems, El Cajon, CA). From each well, 20 μL cell lysates were transferred to a standard white opaque 96-well plate for end-point luminescence measurement using the FlexStation 3 microplate reader. Fifty microliters of 10 μM coelenterazine in PBS was injected into each well, and photon count was measured immediately for 500 ms.

In Vitro Firefly Luciferase Assays

A 20 μL aliquot from cell lysates was transferred to a white opaque 96-well plate. Eighty microliters of FLAR-1 Luciferase Assay Kit reagent (Targeting Systems) was added, and total luminescence was measured for 500 ms.

Luciferin Dosing Studies

Eighty microliters of different D-luciferin concentrations (Gold Biotechnology, St. Louis, MO; diluted in PBS containing 2 mM Mg-ATP) ranging from 0.5 μM to 50 mM were added into wells containing 20 μL of lysates from Gli36 cells expressing either G-liFluc, R-liFluc, or Fluc. Photon count was acquired for 500 ms using the microplate reader.

Stability of Luciferases

Five × 10⁵ cells plated in a well of a six-well plate were lysed in 500 μL of cell lysis buffer. A 65 μL aliquot of cell lysates was frozen immediately. Lysates were incubated for 30 minutes and 1, 2, 3, 4, 5, and 6 hours at 37°C. In triplicate, 20 μL aliquots from each time point were transferred into an opaque 96-well plate and 80 μL luciferase substrate buffer (0.8 mM D-luciferin, 2 mM Mg-ATP, 1× PBS, pH7.2) was injected per well. Luminescence was acquired as described above.

Kinetic Assays

Twenty-microliter aliquots of lysates from Gli36 cells stably expressing G-liFluc, R-liFluc, or Fluc (10⁴ cells lysed in 50 mL) were mixed with 80 μL luciferase substrate buffer. Luminescence was acquired for 500 ms every 20 seconds for 5 minutes using the microplate reader.

Spectral Analysis of G-liFluc and R-liFluc Light Emission

Eighty microliters of FLAR reagent was added to 20 μL of cell lysates, and light emission was measured using the microplate reader every 10 nm using the device's emission monochromator.

In Vivo Experiments

All animal experiments were approved by the Massachusetts General Hospital Subcommittee on Research Animal Care. Athymic nude mice were anesthesized with a mixture of ketamine (100 mg/kg) and xylazine (5 mg/kg) in 0.9% sterile saline. For intramuscular injection (n 5 4), 3 × 10⁶ (in 25 μL) Gli36 cells expressing R-liFluc and Gluc were mixed with equal volume of Matrigel (BD Biosciences, San Jose, CA) and injected using an insulin syringe. For the brain tumor model, 10⁵ U87 cells expressing R-liFluc and Gluc (in 1 μL PBS) were intracranially injected in the left midstriatum of nude mice (n = 4) using the following coordinates from bregma in millimeters: anterior-posterior, +0.5; mediolateral, +2.0; and dorsoventral, −2.5. These injections were performed using a Micro 4 Microsyringe Pump Controller (World Precision Instruments, Sarasota, FL) attached to a Hamilton syringe with a 33-gauge needle (Hamilton, Rena, NV) at a rate of 0.2 μL/min.

In Vivo BLI

Bioluminescence images were obtained at different time points postimplantation of tumor cells. Mice were anesthesized as above. Initially, 20 μL of blood was drawn from the tail vein of each mouse and mixed with 2 μL 20 mM of ethylenediaminetetraacetic acid (EDTA), which was stored on ice until assayed. Mice were then intraperitoneally injected with 150 μL of D-luciferin (150 mg/kg body weight). Ten minutes later, photon count was acquired using a cryogenically cooled charge-coupled device (CCD) camera for 1 minute as described. ²⁴ For in vivo light emission kinetics, imaging was performed every 5 minutes for 1 hour, starting immediately on substrate injection. CMIR-Image software, developed by the Center for Molecular Imaging Research at Massachusetts General Hospital, was used to process the data and quantify signal intensity. Total bioluminescent signals, representing the sum of 1-minute photon counts, are shown in a pseudocolored photon count manner. Images were fused with a grayscale white light image to allow anatomic localization.

Gluc Blood Assay

The Gluc blood assay was performed as described. ²⁵ Briefly, 5 μL of blood was transferred into a white opaque 96-well plate, 100 μL of 100 μM coelenterazine was injected into each well, and the total luminescence was acquired for 10 seconds using a luminometer (Dynex, Richfield, MN).

Results

Expression of Codon-Optimized Variants of the Luciola italica Luciferase in Mammalian Cells

To ascertain the utility of the codon-optimized green- and red-emitting variants of the Luciola italica luciferase as mammalian cell reporters, we cloned the cDNA of either variant into a lentivirus vector plasmid. Cells were first transduced with a lentivirus vector encoding Gaussia princeps luciferase (Gluc) and the cyan fluorescent protein separated by an IRES element (Figure 1, A and B). The expression of Gluc and CFP is used as an internal control to normalize for cell number both in vitro and in vivo as previously described. ⁶ These cells were then engineered using lentivirus vectors to stably express either the green-emitting G-liFluc or red-shifted R-liFluc as well as the mCherry red fluorescent protein separated by an IRES element (see Figure 1A). mCherry is used as a marker for transduction efficiency for the liFluc construct (see Figure 1B). Two weeks posttransduction, cells were plated in a 96-well plate. First, the mCherry fluorescence intensity was measured using a microplate reader, which showed that Gli36 cells had 1.5-fold higher transduction efficiency with R-liFluc compared to G-liFluc (Figure 1C). Then aliquots of lysates from the same cells were assayed for Gluc activity to normalize for cell number (Figure 1D). Finally, the same lysates were analyzed for liFluc activity. Luciferase activity was 3.6 × 10⁴ relative light units (RLU) for G-liFluc and 4.4 × 10⁴ RLU for R-liFluc (Figure 1E). When the liFluc values were normalized to transduction efficiency (mCherry; see Figure 1C) and cell number (Gluc; see Figure 1D), we obtained 2.2 × 10⁴ RLU for G-liFluc and 1.2 × 10⁴ RLU for R-liFluc (Figure 1F). This 1.83-fold difference was found to be statistically significant (p 5 .018).

Figure 1.

Expression of the codon-optimized Luciola italica luciferase green-emitting (G-liFluc) and red-shifted (R-liFluc) variants in mammalian cells. A, Lentivirus expression cassettes used in this study. Gli36 human glioma cells were first transduced with a lentivirus vector encoding Gaussia luciferase (Gluc) and CFP (i) followed by either G-liFluc (ii) or R-liFluc (iii) and mCherry encoding vectors. B, Confirmation of successful transduction with these vectors using fluorescence microscopy. Shown are overlays of bright field and mCherry fluorescence (i, ii) and bright field and CFP fluorescence (iii–iv). Scale bar = 100 μm. C, Transduction efficiency of Gli36 with vectors encoding G-liFluc or R-liFluc was determined by mCherry fluorescence analysis. D, Gluc activity, a marker for cell number, from lysates of same cells was assayed using coelenterazine. E, Luciola italica luciferase assay on separate aliquot of the same cellular lysates in D using D-luciferin. F, Luciola italica luciferase activity normalized for transduction efficiency (mCherry levels from C) and cell number (Gluc levels from D).

Emission and Stability of liFluc Variants Compared to Photinus pyralis Luciferase (Fluc)

We first confirmed the spectrum of light output from G-liFluc and R-liFluc in mammalian cells and compared it to the commonly used firefly luciferase (Fluc) from Photinus pyralis. As expected, we obtained emission spectra with a peak at 550 and 610 nm for G-liFluc and R-liFluc, respectively, which are similar to values reported for non–codon-optimized variants purified from bacteria ²⁰ (Figure 2A). Fluc displayed a peak emission at 550 nm, as expected (see Figure 2A). A 37°C thermostability assay was then performed for these variants and was compared to Fluc. Both G-liFluc and R-liFluc showed an increased thermostability compared to Fluc (Figure 2B). Fluc displayed a rapid decrease in stability with a decay to 5% of initial luminescence by 2 hours of incubation at 37°C (see Figure 2B). In contrast, both G-liFluc and R-liFluc maintained greater than 70% of initial luminescence up to 6 hours postincubation at 37°C (see Figure 2B).

Figure 2.

Characterization of Luciola italica luciferase activity in mammalian cells. A, Emission spectra of G-liFluc, R-liFluc, and the North American firefly luciferase Fluc. B, Luciola italica luciferase stability at 37°C. Lysates from cells expressing either G-liFluc, R-liFluc, or Fluc were incubated at 37uC and assayed for luciferase activity at different time points. C, D-Luciferin dose response of G-liFluc, R-liFluc, and Fluc was assayed on lysates from B. D, Light emission kinetics of liFluc variants. D-Luciferin was added to a well containing lysates from cells expressing either G-liFluc, R-liFluc, or Fluc. Light emission was measured every 30 seconds over 5 minutes.

Dose Response of Green- and Red-Emitting liFluc to D-Luciferin

To gain insight into the performance of each variant under different substrate concentrations, we carried out a dose-response analysis of D-luciferin ranging from 5 μM to 50 mM on lysates from Gli36 cells expressing either G-liFluc, R-liFluc, or Photinus pyralis Fluc. All three luciferases displayed a similar dose/activity profile, although the levels of luminescence varied with each luciferase (Figure 2C). Luminescence increased with increasing dose of D-luciferin, reaching a plateau at a 5 mM concentration. At this concentration, Fluc displayed a 9-fold higher and a 34-fold higher normalized luminescent intensity compared to G-liFluc and R-liFluc, respectively (see Figure 2C).

Light Emission Kinetics for liFluc Variants

A 5-minute light emission kinetic assay after D-luciferin addition was determined for G-liFluc, R-liFLuc, and Fluc in lysates from Gli36 cells expressing either luciferases. All luciferases displayed rapid decay in luminescent signal during the first 30 seconds of the assay, characteristic of firefly luciferase in the absence of stabilizers ²⁶ (Figure 2D). Thereafter, the signal decay slowed for all luciferases, with a difference in the degree of the decline. For example, from 30 seconds to 5 minutes, R-liFluc declined from 26% to 20% of the starting signal, similar to Fluc (Figure 2D). For the same time interval, G-liFluc declined from 17% to 5% of the starting luminescence (see Figure 2D).

R-liFluc as a Reporter for In Vivo BLI

Given that light of longer wavelengths is known to have much lower absorption by pigmented molecules such as hemoglobin and melanin as well as scattering by mammalian tissues, we characterized the red-shifted Luciola italica luciferase variant for in vivo deep tissue BLI. Nude mice were injected into the hamstring with 3 × 10⁶ Gli36 cells expressing R-liFluc. Fourteen days postinjection, mice were injected intraperitoneally with 150 mg/kg D-luciferin and imaged using a cooled CCD camera. Light emission was readily detected from these muscle tissues (4.03 × 10⁵ photons/min; Figure 3A). To determine the optimum acquisition time for imaging R-liFluc in deep tissues, we imaged mice immediately and at 5-minute intervals for 1 hour after D-luciferin injection. Bioluminescence signal was detected immediately after injection and peaked at 10 minutes post–substrate injection (Figure 3B). Signal began to slowly decline at 30 minutes, although it never dropped below 60% for the 1-hour duration (see Figure 3B).

Figure 3.

Red-emitting Luciola italica luciferase variant for in vivo imaging. The hamstrings of nude mice were injected with 3 × 10 Gli36 cells expressing R-liFluc. A, Two weeks later, mice were injected intraperitoneally with D-luciferin (150 mg/kg body weight) and imaged 10 minutes later using a cooled CCD camera. Signal from tumor was quantified using the CMIR-Image program. Data shown are mean ± SD (n = 4). B, Light emission kinetics after injection of D-luciferin. The same experiment in A was repeated, but the luciferase activity was acquired immediately and every 5 minutes over 1 hour post–substrate injection. Data shown are from two representative animals.

Having confirmed its utility for imaging in muscle tissues, we next tested R-liFluc as a reporter to image brain tumors, which requires adequate light to pass through 2.5 mm of brain tissue and ≈ 1 mm of skull. We used U87 human glioma cells cotransduced with lentivirus vectors encoding Gluc-ICFP and R-liFluc-ImCherry as this cell line is well established for use in orthotopic xenograft models of glioma. We first confirmed the R-liFluc and Gluc luciferase activity in lysates and conditioned medium from these cells, respectively (Figure 4A). Next, 10⁵ U87 glioma cells stably expressing R-liFluc-ImCherry and Gluc-ICFP were injected into the striata of nude mice. At 2 and 3 weeks postinjection, mice were anesthetized and blood was collected and assayed for Gluc activity ⁶ ; then mice were intraperitoneally injected with 150 mg/kg D-luciferin and imaged for R-liFluc 10 minutes later. At 2 weeks post–tumor cell implantation, luminescence was detected at an average of 1.23 × 10³ photons/min (Figure 4B). This signal increase to 8.58 × 3 10³ photons/min at week 3 post–tumor injection resulted in a sevenfold increase in signal between the two time points, proving that R-liFluc can be used as a reporter to monitor biologic processes, including tumor growth, over time (see Figure 4B). A similar fold increase was observed for Gluc activity in the tumor as assessed by the Gluc blood assay at the same time points (Figure 4C).

Figure 4.

Bioluminescence imaging of brain tumors expressing red-emitting Luciola italica luciferase. A, Luciferase activity from U87 cells stably expressing R-liFluc and Gluc. B, The striatum of nude mice were stereotactically injected with 10⁵ U87 human glioma cells expressing both R-liFluc and Gluc. At 2 and 3 weeks post–tumor cell implantation, mice were intraperitoneally injected with D-luciferin (150 mg/kg body weight) and imaged for Luciola italica luciferase activity 10 minutes later using a cooled CCD camera. Intracranial brain tumor associated R-liFluc signal (bottom) as quantified using the CMIR-Image program (top). C, Prior to D-luciferin injection, 5 μL blood was collected and assayed for Gluc activity after addition of 100 μL of 100 μM coelenterazine and acquiring photon counts using a luminometer. Data shown are mean ± SD (n = 5 4).

Discussion

Luciferases catalyze light-producing chemical reaction by oxidizing their substrate, luciferin, while emitting photons. This process, known as bioluminescence, is a natural phenomenon found in many lower forms of life, including fungi, bacteria, insects, and marine copepods. The North American firefly luciferase from Photinus pyralis (Fluc) is the best characterized luciferase for luminescence applications and has proven invaluable as a biologic reporter protein.^27,28 Oxidation of D-luciferin by Fluc results in a high quantum yield of light emission leading to a sensitive detection. Less than 10⁴ molecules of Fluc protein can be detected using a standard luciferase assay. ²⁹ However, the yellow-green color of light emission is suboptimal for in vivo imaging. Although there is a single report of a red-shifted variant of Fluc with enhanced in vivo luminescence properties, ¹⁶ the availability of other luciferases expands the potential of advancing our understanding of biologic processes. Our results broaden the characterization of the Luciola italica luciferase as a promising biologic reporter in mammalian cells. Using codon-optimized green- and red-emitting liFluc variants, we observed a robust luciferase activity in cultured mammalian cells.

One important characteristic of luciferases for molecular biologic applications is thermostability. For measurement of luciferase activity in cultured cells or tissue homogenates, cells are lysed, and the luciferase protein is subject to degradation/inactivation by proteases, pH changes, and temperature fluctuations.^30,31 This has implications for storage conditions of samples post–cell lysis. We found that both the G-liFluc and R-liFluc variants had greatly increased thermostablity in cell lysates after incubation at 37°C compared to Fluc. This may lower the amount of signal decay between sample harvest and freezing/assaying, allowing for higher number of samples to be processed simultaneously. For a luciferase/luciferin combination to be suited for in vivo imaging, it needs to overcome certain barriers, such as light absorption by pigmented molecules (eg, hemoglobin and melanin) and scattering by deep tissues. Given that most light below 600 nm cannot penetrate mammalian tissues, red-emitting luciferases would be highly beneficial. Both G-liFluc and R-liFluc showed significantly lower activities compared to Fluc in culture, which will likely translate into lower sensitivity for both variants for cell detection in vivo. Despite this lower sensitivity, the codon-optimized red-emitting liFluc characterized here with a peak emission at 610 nm proved to be a robust tool for imaging tumor cells in deep tissues, as observed by both intramuscular and intracranial tumor models. Further, this disadvantage could be elevated because G-liFluc and R-liFluc have two distinct emission spectra, making them highly useful for multicolored applications using spectral unmixing.^9,32 These reporters can be used together to monitor dual biologic processes simultaneously and can be combined with other luciferases using different substrates, such as Gaussia or Renilla luciferase, for triple reporter systems. ²⁰

Conclusion

We have characterized two human codon-optimized thermostable variants of the Luciola italica luciferase for bioluminescence applications in mammalian cells and showed that the red-emitting variant is useful for deep tissue imaging. These variants may be useful alternative luciferases where increased thermostability is desired and can be combined together for multiple color applications using spectral unmixing.