Meganuclease-Driven Targeted Integration in CHO-K1 Cells for the Fast Generation of HTS-Compatible Cell-Based Assays

Abstract

The development of cell-based assays for high-throughput screening (HTS) approaches often requires the generation of stable transformant cell lines. However, these cell lines are essentially created by random integration of a gene of interest (GOI) with no control over the level and stability of gene expression. The authors developed a targeted integration system in Chinese hamster ovary (CHO) cells, called the cellular genome positioning system (cGPS), based on the stimulation of homologous gene targeting by meganucleases. Five different GOIs were knocked in at the same locus in cGPS CHO-K1 cells. Further characterization revealed that the cGPS CHO-K1 system is more rapid (2-week protocol), efficient (all selected clones expressed the GOI), reproducible (GOI expression level variation of 12%), and stable over time (no change in GOI expression after 23 weeks of culture) than classical random integration. Moreover, in all cGPS CHO-K1 targeted clones, the recombinant protein was biologically active and its properties similar to the endogenous protein. This fast and robust method opens the door for creating large collections of cell lines of better quality and expressing therapeutically relevant GOIs at physiological levels, thereby enhancing the potential scope of HTS.

Keywords

meganuclease homologous gene targeting protein expression cell-based assay high-throughput screening

Introduction

Cell-based assays are used in high-throughput screening (HTS) approaches for the discovery and validation of new therapeutic molecules.^1,2 The establishment of cell lines expressing target molecules is a key aim, and a variety of methods are used to achieve this goal.³ The most common current strategy for generating stable cell lines is to introduce DNA sequences coding the gene of interest (GOI) into the host cell genome and to characterize individual clones after random integration of the transgene. A selection cassette, usually an antibiotic resistance gene, is introduced at the same time to select the transformants, which are then assayed for expression of the GOI. Once the potential clones are characterized, they can be tested using a functional assay and eventually validated. Several hurdles make this strategy tedious, expensive, and time-consuming, but the real bottleneck of this process is the production, maintenance, and characterization of numerous individual clones. Furthermore, plasmid concatenation may lead to genetic and epigenetic modifications of the integrated construct, leading to silencing. A more recent strategy relies on large-scale transient transfection by a classical method⁴ or using viral transduction.^5,6 Transient transfection offers compelling features, including flexibility and reproducibility, and alleviates the establishment of cell lines but, at least for the liposomal transfection procedures, can prove costly, tedious, poorly reproducible from batch to batch, and not easily automated. However, the major drawback of the approach is the nonphysiological levels of associated GOI expression, which sometimes results in toxicity and, more generally, challenges the physiological relevance of the assay.

Thus, dramatically different strategies have been developed to ensure the fast generation of homogeneous populations of GOI expressing cells. For example, the laser-enabled analysis and processing (LEAP) technology relies on elimination of nonexpressing clones.⁷ Another promising approach is targeted integration of the GOI at a predefined chromosomal location. This strategy has the advantage of inserting a single copy of each transgene into the same chromosomal environment. Thus, a very small number of clones need to be characterized, the vast majority of them being identical. Recombinases, such as yeast Flp recombinase that allows the integration of a circular plasmid into a chromosome, can be useful tools when specific sequences are present on both DNA partners.⁸ Therefore, targeted integration requires the prior integration of a target site into the chromosome. The recombinase-based Flp-In system commercialized by Life Technologies (Invitrogen, Carlsbad, CA) allows for the integration into a landing pad inserted into the target cell.^9,10 Recently, a dual system based on the PhiC31 and R4 integrases was used to integrate GOIs into pseudo att target sites into human cell lines, CHOS Chinese hamster ovary cells, strain S, and human embryonic stem cells.^11,12

The other option is to target foreign DNA integration via homologous gene targeting (HGT). HGT is usually inefficient in mammalian cell lines, but it is greatly stimulated if a DNA double-strand break (DSB) occurs in the vicinity of the targeted locus. Such DSBs can be delivered using meganucleases, sequence-specific endonucleases that recognize large DNA target sites (>12 bp).¹³ These proteins, because of their exquisite specificity, can cleave a unique chromosomal sequence without affecting the global genome integrity.¹⁴ Therefore, meganucleases represent potent stimulators of HGT and have been used to trigger targeted recombination events in a variety of organisms and cell types.^14-18 Here, we describe the cGPS CHO-K1 cell line, a CHO-K1-derived cell line that was engineered to optimize the targeted integration of transgenes.

To validate this system, a reporter gene (LacZ) and 4 relevant pharmacological target proteins (hATX, hMT₁, hMT₂, and hHCN4) were expressed using targeted and random strategies and characterized. Escherichia coli β-galactosidase, a robust cytosolic protein used in a vast number of expression systems, was used to set up the procedure and compare our results to other published expression systems. The human secreted enzyme, autotaxin (hATX), is a motility factor belonging to the ecto-nucleotide pyrophosphate phosphodiesterase (PDE) family. Recently, a second activity was described for ATX, lyso-phospholipase D (PLD).¹⁹ Autotaxin and lysophosphatidic acid (LPA) have been associated with early cancer progression and cancer cell motility.²⁰ To understand the catalytic activity of hATX, access to a mammalian source is essential. Several strategies have been developed for ATX expression, including transient transfection of mammalian cells and Escherichia coli production; however, all of these procedures were rather difficult and never resulted in high protein yield.^21-23 The melatonin receptors hMT₁ and hMT₂ are also major pharmacological targets because they are implicated in circadian rhythm regulation.²⁴ Finally, hyperpolarization-activated cyclic nucleotide-gated potassium channel 4 (hHCN4) is a human variant of a sarcolemmal ionic channel belonging to the hyperpolarization-activated cyclic nucleotide-gated channel family, and 4 isoforms (HCN1-4) have been identified.²⁵ These voltage-dependent channels comprising a tetramer of 6-transmembrane domain proteins are activated in the hyperpolarized range of potential and generate an inward current carried by both sodium and potassium HCN channels. They are involved in important electrophysiological processes such as automatism and cell excitability regulation in the heart or in the nervous system.²⁶ These voltage-dependant channels are considered attractive pharmacological targets for several pathologies²⁷ and represent good models for the validation of cGPS CHO-K1 cells in patch-clamp automated tests, which require not only faithful integrity of the channels but also a large number of channels in the plasma membrane.²⁸

Materials and Methods

cGPS CHO-K1 system

In the cGPS CHO-K1 system (Cellectis bioresearch, Romainville, France), the hygromycin resistance gene was under the transcriptional control of the human EF1α promoter. Roughly 1 kb of additional EF1α sequence, corresponding to the first 2 untranslated exons and first intron, was present between the promoter and hygromycin resistance gene. A cleavage site for the natural endonuclease I-CreI was inserted 16 bp upstream of the ATG in the hygromycin resistance gene. A promoterless neomycin resistance gene was positioned 94 bp downstream of the SV40 polyadenylation sequence, terminating the hygromycin resistance gene. The single copy of the construct was stably integrated into the CHO-K1 cell genome. The phenotype of the cGPS CHO-K1 cell line is hygromycin resistant, G418 sensitive. The integration matrix contained 2 DNA fragments homologous to the landing pad described above (0.8 kb upstream from the I-CreI cleavage site and 0.53 kb downstream). The downstream homologous sequence represents the first 530 bp of the neo gene coding sequence. Both homology arms are separated by (1) the puromycin resistance gene, which lacks a promoter in the plasmid; (2) a multiple cloning site, allowing the cloning of a full expression cassette; and (3) the pSV40 promoter. This plasmid cannot induce a puromycin and neomycin resistance phenotype on its own; the puro^R gene is promoterless, and the neo^R gene lacks 243 base pairs at its 3′ end. The I-CreI coding sequence was cloned into pcDNA6.2/V5-DEST (Invitrogen) using Gateway^® technology according to the manufacturer’s recommendations to create the meganuclease expression vector.

GOI cloning

The human melatonin receptors (hMT₁ and hMT₂), hATX variant 2 (hENPP2), and hHCN4 coding sequences were isolated by reverse transcriptase polymerase chain reaction (RT-PCR) of human brain or heart mRNA (Clontech, Palo Alto, CA) and subcloned into the NheI and NotI sites of the integration matrix (Cellectis bioresearch) and pcDNA3.1(+) vector (Invitrogen). Detail of constructions is available upon request.

Establishment of stably transfected cell lines

CHO-K1 cells (CCL61, American Type Culture Collection, Manassas, VA) were maintained in Ham-F12 medium supplemented with 10% (v/v) fetal calf serum, 2 mM L-glutamine, 500 U/mL penicillin, and 500 µg/mL streptomycin. The cGPS CHO-K1 cells were maintained in Ham-F12K medium (Kaighn’s modification) supplemented with 10% (v/v) fetal calf serum, 2 mM L-glutamine, 500 U/mL penicillin, 500 µg/mL streptomycin, and 600 µg/mL hygromycin. Nucleofection of cGPS CHO-K1 and CHO-K1 cells was performed according to the manufacturer’s instructions using the Nucleofector machine (Amaxa, Cologne, Germany). Adherent CHO cells were harvested at day 3 of culture, washed twice in cold phosphate-buffered saline (PBS), trypsinized, and gently resuspended in T nucleofection solution to a final concentration of 2 × 10⁶ cells/100 µL. Then, 1 µg of integration matrix/GOI vector and 5 µg of meganuclease expression vector were mixed with 0.1 mL of the cGPS CHO-K1 cell suspension, transferred to a 2.0-mm electroporation cuvette, and nucleofected using program U-023 and an Amaxa Nucleofector apparatus. Similarly, CHO-K1 cells were nucleofected with 5 µg of plasmid pcDNA3.1/GOI. Immediately after nucleofection, 0.5 mL of prewarmed CHO-K1 or cGPS CHO-K1 medium was added to the cuvette. Cells were transferred to a Petri dish containing 5 mL CHO medium using a plastic pipette provided in the Amaxa kit and cultured in a humidified 37°C, 5% CO₂ incubator. After 24 h, CHO-K1 cells were cultured in medium containing 800 µg/mL G418 (Invitrogen) for 12 days. After G418 selection, single colony clones were picked and seeded in 96-well plates containing complete medium supplemented with 0.4 mg/mL G418. After 24 h of transfection, cGPS CHO-K1 cells were incubated at 37°C under 5% CO₂ in Ham-F12K medium supplemented with 10% (v/v) fetal calf serum, 2 mM L-glutamine, 500 U/mL penicillin, 500 µg/mL streptomycin, and 0.6 mg/mL G418 (Invitrogen). Ten days after G418 selection, single colony clones were picked and seeded in 96-well plates in complete medium supplemented with G418 at 0.6 mg/mL and at 10 µg/mL puromycin (Sigma-Aldrich, St. Louis, MO). Six to 7 days later, double-resistant clones were amplified in complete medium supplemented with the 2 selection agents. Ten antibiotic-resistant cell clones were isolated and analyzed for each stably transfected cell line.

β-galactosidase enzymatic activity

The cGPS CHO-K1/lacZ clones expressing β-galactosidase protein were detached from the Petri dish by 0.05% trypsin/EDTA and resuspended in F12-K medium supplemented with 10% fetal bovine serum. Cells (200,000) from each clone were centrifuged at 1200 rpm for 5 min. The supernatant was aspirated and 100 µL PBS added to each pellet. The cells were incubated for exactly 10 min at 37°C in an atmosphere of 95% air/5% CO₂. Next, the cells were loaded with 100 µL of 2 mM Fluorescein Di β-galactopyranoside (FDG; Sigma-Aldrich) for exactly 1 min at 37°C in an atmosphere of 95% air/5% CO₂. After loading, 200 µL PBS supplemented with 2% fetal bovine serum was added and the vials left on ice until cytometric evaluation. Cells were analyzed on an EPICS XL/MCL flow cytometer (Beckman Coulter, Roissy, France) using a 488-nm ion-argon laser. The emitted fluorescence (emission wavelength 520 nm) was collected on the fluorescence 2 channel. Basal fluorescence was measured in 10,000 cGPS CHO-K1 cells per assay. Alternatively, trypsinized cells were resuspended in complete F-12K medium at a density of 50,000 cells/mL. Five thousand cells were aliquoted in triplicate in a white 96-well plate (PerkinElmer, Waltham, MA). β-Glo reagent (100 µL; Promega, Madison, WI) was added to each well. After a 30-min incubation, the plate was read on a luminometer (Viktor, PerkinElmer). Results are expressed as the ratio of signal to background (parental cGPS CHO-K1 cells).

Autotaxin enzymatic assay

CHO-KI/hATX or cGPS CHO-K1/hATX clone cells were washed twice with PBS and then 3 times with serum-free F12-K medium supplemented with 1% glutamine to remove the serum (2 mL per well and per wash for a 6-well plate). The cells were then incubated in serum-free F12-K medium (1 mL/well) for 6 h at 37°C in a humidified atmosphere containing 7% CO₂. After incubation, the conditioned medium (CM) was separated from the cells, centrifuged to eliminate cell debris, and dialyzed overnight against 10 L of 20 mM HEPES (pH 7.4), 6 mM D(+)-glucose, 1 mM CaCl₂, and 1.2 mM MgSO₄ using Spectra-Por 1.7-mL/cm tubing (Pierce Chemicals, Interchim, Montluçon, France). After dialysis, the CM was concentrated roughly 15-fold using an Amicon Ultra 10,000 (Millipore, Billerica, MA). Concentrated conditioned media (CCM) were aliquoted and stored at −20°C until use.

The phosphodiesterase activity of recombinant ATX was measured with pNppp substrate using a modification of Razzell and Khorana’s method.²⁹ The recombinant ATX proteins were used as an enzyme source. Samples (10 µL) were incubated in 96-well plates at a final volume of 100 µL containing 5 mM p-nitrophenyl phenylphosphonate and 50 mM Tris-HCl (pH 9.0) buffer. After 30 min at 37°C, the reaction was stopped by adding 100 µL of 0.1 M NaOH. The production of p-nitrophenol was measured by following the absorbance at 410 nm using a Pherastar plate reader (BMG, Offenburg, Germany) and subtracting the blank. The initial rates of pNP formation, which were measured after 30 min of incubation using light absorbance at 410 nm, were plotted against increasing substrate concentrations and fit according to the Michaelis-Menten equation. Nonlinear regression analyses of the kinetic data with calculation of the V_max and K_m were performed using the statistical software package Prism 4.0 (GraphPad Software, La Jolla, CA).

Radioligand saturation with intact cells

CHO-K1 or cGPS CHO-K1 cells expressing hMT₁ or hMT₂ receptor were resuspended in 50 mM Tris/HCl (pH 7.4), 1 mM EDTA, and 5 mM MgCl₂ and aliquoted into 96-well polypropylene plates (13,000 cells/well). [¹²⁵I]-2-Iodomelatonin (5 pM to 1.5 nM) was added to determine the total binding, and control wells contained an additional 1 µM melatonin to determine nonspecific binding. Incubation was performed at 37°C for 2 h in a total volume of 250 µL. Cells were then transferred to unifilter GF/B plates (PerkinElmer) with a FilterMate cell harvester (PerkinElmer) and washed 3 times with 1 mL ice-cold Tris (50 mM). Microscint 20 (40 µL/well; PerkinElmer) was added before sealing the plates. The radioligand associated with the filter plates was evaluated by scintillation counting using TopCount (PerkinElmer). Experiments were conducted in triplicate, and data were expressed as femtomole of radioligand-specific binding sites (total minus nonspecific) per milligram of total protein. Graphic representations and data analyses were generated using PRISM 4.03 (GraphPad Software).

Cellular dielectric spectroscopy

CHO-K1 or cGPS CHO-K1 cells expressing hMT₁ or hMT₂ receptor were plated at a density of 80,000 cells/well on MDS Analytical (Concord, Canada) 96-well assay plates with embedded electrodes and incubated at 37°C in 6% CO₂ for 24 h, in the presence of 100 ng/mL pertussis toxin when specified. Prior to the measurements, the cells were washed 3 times with Hank’s balanced salt solution, 0.1% bovine serum albumin (BSA), and 20 mM HEPES (pH 7.4) and left to equilibrate at 28°C for 30 min. The impedance measurement was performed on an MDS Analytical CellKey system with a baseline recorded for 5 min before the online addition of melatonin and further kinetic measurements for 15 min. The resulting data were expressed as the maximal signal corrected for the baseline and represented as a percentage. Experiments were performed in duplicate.

Patch-clamp recording of hHCN4 current

CHO-K1 or cGPS CHO-K1 cells expressing the hHCN4 ion channel were seeded in 35-mm Petri dishes in culture medium with 5% fetal calf serum and 50 ng/mL nocodazole to stop cell division. The cells were used within 1 to 3 days. Patch-clamp experiments in the whole-cell configuration were carried out at the controlled temperature of 35 ± 1°C (Cell MicroControls). The extracellular bathing and recording solution was a modified Tyrode’s solution containing (in mM) the following: NaCl 110, KCl 30, MgCl₂ 0.5, CaCl₂ 1.8, HEPES 10, and glucose 20, adjusted to pH 7.4 with NaOH. The borosilicate pipettes were filled with an intracellular solution containing (in mM) the following: KCl 130, NaCl 10, MgCl₂ 0.5, EGTA 1, and HEPES 5, adjusted to pH 7.2 with KOH. Membrane capacitance was canceled and access resistance was compensated to approximately 70%. The currents, recorded by a RK-400 (Bio-Logic, Claix, France) or Axopatch 200-B (Axon Instruments, Sunnyvale, CA) amplifier, were filtered at 1 KHz, digitized at 5 KHz, and analyzed with pClamp 9.0 software (Axon Instruments). The hHCN4 current was elicited every 6 s by a 5-s hyperpolarizing pulse to −110 mV from a holding potential of −30 mV and deactivated by a depolarizing pulse at +20 mV for 600 ms. The current was measured as the difference between the instantaneous current at the beginning of the hyperpolarizing pulse and the value of the inward current at the end of the pulse. The membrane capacitance value was measured as the value corrected on the amplifier. Values were expressed as mean ± SEM. From 10 selected clones, 10 cGPS CHO-K1 exhibited an immunofluorescence signal, whereas only 2 of the 10 CHO-K1 were positive.

SDS-PAGE separation and Western Blotting

Sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) 4% to 12% was performed according to Laemmli³⁰ followed by Sypro Ruby staining and Western blotting detection using chicken anti-autotaxin antibody²² followed by a horseradish peroxidase (HRP)–conjugated anti-chicken antibody (Sigma-Aldrich) before chemiluminescence detection of the immunocomplexes.

Results

Stable cell line generation

We developed the cGPS system that allows for the targeted integration of a GOI in a predefined locus in CHO-KI cells. A landing pad containing a hygromycin resistance gene under the control of an EF1α promoter-intron fusion, a recognition site for the I-CreI meganuclease, and a promoterless neomycin resistance gene was integrated into the genome of CHO-KI cells ( Fig. 1A ). The cGPS CHO-K1 cell line was initially hygromycin resistant. Upon cleavage of the I-CreI recognition site by the cognate meganuclease, high frequencies of homologous recombination were obtained with a targeting vector containing sufficient homology. The integration matrix includes 2 regions homologous to the cGPS locus: a promoterless Puro^R ORF and an SV40 promoter ( Fig. 1A ). GOIs can be cloned between the Puro^R ORF and the SV40 promoter. Homologous gene targeting places the Puro^R downstream from the EF1α promoter and upstream of the GOI and restores a functional SV40::Neo^R gene in the 3′ end, resulting in a selectable Puro^R, Neo^R, Hygro^S cell ( Fig. 1A ). In contrast, random integration of the integration matrix genes does not result in Puro^R clones unless the puromycin resistance gene is fortuitously integrated downstream of an endogenous promoter (promoter trap), and it never results in Neo^R clones (data not shown). Thus, double G418/puromycin selection should select only targeted clones.

Fig. 1.

cGPS principle overview and molecular characterization of clones. (A) Schematic representation of the cGPS landing pad containing the recognition site for the natural meganuclease I-CreI, the cGPS integration matrix containing 2 arms homologous to the cGPS landing pad, and the cGPS locus after meganuclease-driven targeted integration. Upon cleavage by the meganuclease, recombination between homologous sequences results in targeted integration of the GOI at the cGPS CHO-K1 locus and the creation of 2 functional antibiotic resistance cassettes upstream (puro) and downstream (neo) of the DNA double-strand break. Restriction sites and probes used for Southern blot analysis are indicated. (B) Southern blot analysis of cGPS CHO-K1/LacZ clones using BglII/EcoRV double-digested genomic DNA and radioactive probe A specific to the left homology arm. The parental cGPS CHO-K1 locus corresponds to a 10-kb band (Parental P) and appropriate targeting of LacZ in the cGPS CHO-K1 locus to a 4.9-kb band (Targeted T). (C, D) Southern blot analysis of cGPS CHO-K1/hMT₁ and hMT₂ clones using BglII/NotI double-digested genomic DNA and probe A or RsrII-digested genomic DNA and probe B. Probe A detects the upstream region of the cGPS CHO-K1 landing pad (>10 kb) in the cGPS CHO-K1 cell line (P). A 4.8-kb band is expected in targeted clones (T). Probe B detects the downstream region of the cGPS CHO-K1 landing pad (1.6 kb, P; 4.3 kb, T). pEF1α, elongation factor 1α promoter; pCMV, cytomegalovirus promoter; pSV40, simian virus 40 promoter; Hygro^R, hygromycin resistance gene; Neo^R, neomycin resistance gene; Puro^R, puromycin resistance gene; GOI, gene of interest; CHO-K1, Chinese hamster ovary cell line.

We also used a classical random integration method: GOIs were cloned into pCDNA3.1(+). The resulting plasmids were transfected into CHO-KI cells, and stable transformants were selected on G418. Clones were obtained in about 44 days. With the cGPS strategy, HGT of the GOI was induced with the I-CreI meganuclease, and clones were selected for G418 and puromycin resistance with a short 28-day protocol. Isolated clones were then analyzed at the genomic level, for protein expression and/or activity and for expression stability.

Molecular evidence for targeted integration in cGPS CHO-K1 clones

To monitor targeted integration events in the cGPS system, 20 cell clones obtained with the lacZ construct and 9 or 10 clones obtained with the other cassettes were analyzed by Southern blot. Targeted integration was analyzed by 1 probe monitoring the upstream regions of the cGPS landing pad and a second probe downstream of the construct for hMT₁ and hMT₂. The parental cGPS locus and the GOI (LacZ, hMT₁, or hMT₂) targeted cGPS loci, with relevant restriction sites and probe locations, are shown in Figure 1A , and the results are summarized in Table 1 .

Table 1.

Summary of Integration Analysis

	LacZ	hATX	hMT₁	hMT₂	Total, n (%)
Number of clones analyzed	20	9	10	9	48
Perfect targeted integration^a	18	6	8	4	36 (75.0)
Targeted and random^b	1	3	2	2	8 (16.6)
Others^c	1	0	0	3	4 (8.4)

Gene targeting with 2 perfect homologous junctions and no additional random integration.

Gene targeting with 2 perfect homologous junctions and additional random integration.

Targeted integration associated with further rearrangement of the targeted locus or conservation of 1 nonrearranged locus.

Gene targeting with 2 perfectly homologous junctions was observed in 19 of 20 clones with LacZ ( Fig. 1B ), 10 of 10 clones with hMT₁ ( Fig. 1C ), 8 of 9 clones with hMT₂ ( Fig. 1D ), and 9 of 9 clones with hATX (data not shown). Seven targeted clones also displayed additional bands in the presence of probe A and/or probe B. Such bands can be indicative of an additional, nontargeted integration event ( Fig. 1B , clone 16; Fig. 1C , clones 4 and 6; Fig. 1D , clones 5 and 7). However, in 2 hMT₂ clones (1 and 4), the additional band corresponded to the parental pattern. This last result was confirmed with 2 independent cell samples and DNA extractions, and it suggests a duplication of the chromosome or a mixed population of cells. Finally, 2 clones presented a complex pattern wherein loss of the parental pattern was not associated with the appearance of a targeted pattern ( Fig. 1B , clone 5; Fig. 1D , clone 2). Such clones are best interpreted as targeted integrations associated with further rearrangement of the targeted locus, as described in earlier studies of meganuclease-mediated gene targeting.^16,31 Thus, all of the recovered cells seem to have undergone a targeted event, and 75% (36/48) were perfectly targeted events, with 2 homologous junctions and no additional integration events.

Analysis of the expression of GOIs’

To monitor protein activity levels with hMT₁, hMT₂, hATX, and hHCN4, 10 cGPS CHO-K1 clones and 10 CHO-K1 clones were randomly chosen and analyzed for the biological activity of the expressed recombinant proteins ( Fig. 2 ). Analysis included enzymatic assays for LacZ and hATX and a specific ligand binding assay for hMT₁ and hMT₂. hHCN4 expression was first analyzed by immunostaining, and positive clones were further analyzed quantitatively using a patch-clamp assay (see below). In addition, 10 LacZ cGPS CHO-K1 clones were analyzed for β-galactosidase expression. All cGPS targeted clones (10 clones for each of the 5 chosen genes) were shown to express a functional protein ( Fig. 2A-E ), although only 17.5% (7 of 40) of randomly inserted clones expressed functional protein ( Fig. 2F-I ). Furthermore, the expression/activity was found to be generally homogenous from clone to clone ( Fig. 2A-E ; LacZ, 9 of 10 clones with similar activity; hATX, 7 of 10 clones; hMT₁, 5 of 10 clones; hMT₂, 9 of 10 clones; hHCN4, all). Interestingly, the level of clone expression/ activity was close to the expression level of a polyclonal double-resistant population ( Fig. 2B-E ). Such populations were directly recovered by pooling all double-resistant cells from the same 10-cm plate (~100-200 clones). Inversely, expression analysis of the polyclonal cell population for the classical strategy showed a marked decrease compared to the cGPS CHO-K1 polyclonal population, which is explained by the huge variability in expression among clones ( Fig. 2F-I ).

Fig. 2.

Recombinant protein expression in targeted (cGPS CHO-K1) and random transfection of CHO-K1. β-Galactosidase activity was measured with Fluorescein Di-β-galactopyranoside (emission wavelength 520 nm) using flow cytometry and 10 transfected cGPS CHO-K1 cells (A). Phosphodiesterase (PDE) activity of autotoxin enzyme was determined by the production of p-nitrophenol (absorbance at 410 nm) in culture medium from 10 transfected (B) cGPS CHO-K1 or (F) CHO-K1 cells. hMT₁ and hMT₂ expression was monitored in (C) 10 cGPS CHO-K1/hMT₁ clones, (D) 10 cGPS CHO-K1/hMT₂ clones, (G) 10 CHO-K1/hMT₁ clones, and (H) 10 CHO-K1/hMT₂ clones, using radioligand saturation curves. Figures display the maximum number of binding sites reported per total protein. Clones expressing hHCN4 were first identified by immunochemistry (see text), and the expression of recombinant hHCN4 was then estimated by the percentage of cells that presented a current between 500 and 3000 pA. (E) Ten cGPS CHO-K1/hHCN4 clones and (I) 10 CHO-K1/hHCN4 clones were tested. All cGPS CHO-K1/hHCN4 clones were found to be positive by immunostaining, whereas only 2 CHO-K1/hHCN4 clones (2 and 4) were positive and further characterized with the patch-clamp assay. The current was measured as the difference between the instantaneous current at the beginning of the hyperpolarizing pulse (−110 mV) and the value of the inward current at the end of the pulse. P, polyclonal; NT, nontransfected cGPS CHO-K1 or CHO-K1 cells. The individual clones are presented by their ID number.

Functional characteristics of recombinant hATX

The analysis of recombinant hATX by SDS-PAGE and Western blot revealed only 1 protein (110 kDa), which corresponded to the expected protein size of endogenous hATX ( Fig. 3A ). All of the cGPS CHO-K1 targeted clones expressed HATX, whereas only 2 randomly integrated CHO-K1/HATX clones (clones 6 and 8) produced hATX. The kinetic analysis of hATX expression in cGPS CHO-K1/hATX clone 10 showed that the hATX expression level in the cell culture medium increased in a time-dependent manner over a 24-h period (data not shown). Recombinant hATX obeys the Michaelis-Menten model. Therefore, the K_m was determined using the Hill representation and V_max by direct double inverse plotting ( Fig. 3B ). The kinetic values of recombinant hATX were in accordance with the K_m and V_max of endogenous hATX (K_m = 11.3 ± 0.2 mM for pNppp; V_max = 1.9 ± 0.1 nmol pNppp/min).²² Similar to endogenous hATX, the recombinant hATX exhibited allosteric behavior with an n_h of 1.7 ( Fig. 3B ).

Fig. 3.

Functional characteristics of recombinant proteins expressed by the cGPS CHO-K1 system. (A) Expression of hATX by sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) and Western Blot detection. A total of 40 µg protein from cell culture medium concentrated 15-fold was loaded on each well. The 10 selected cGPS CHO-K1 clones exhibited an autotoxin expression. For CHO-KI, 10 clones were selected, but only 2 (6 and 8) of them exhibited an autotaxine expression in Western blot detection (top panel). Molecular weight of recombinant hATX corresponds to 110 kDa. (B) Kinetic constants of recombinant hATX were determined in cGPS CHO-K1/hATX clone 10. Phosphodiesterase activity was assayed with various concentrations ranging from 0.38 to 45 mM of pNppp as substrate. Enzymatic values V_max and K_m were determined graphically using Prism software. The K_m substrate (pNppp) was 16 mM. P, polyclonal; NT, nontransfected cGPS CHO-K1 or CHO-K1 cells; MW, molecular weight markers. (C) Determination of the level of hMT₁ receptor expression using [¹²⁵I]-2-iodomelatonin saturation experiments and represented on a direct plot using the Scatchard linear regression (inset). Clone 10 is used as an example and provided the following data: B_max = 82 fmol/mg total protein, K_D = 27 pM. (D) Assessment of the functional response of hMT₁ using CDS technology after stimulating the cells with melatonin at the concentrations noted. Data are expressed as the variation of extracellular impedance (Ziec) and plotted against the concentration of melatonin. Inset: the kinetic trace obtained with each clone using 100 nM melatonin. The potency of melatonin obtained in this assay was EC₅₀ = 75 pM. (E) Example of hHCN4 current recorded in a targeted cGPS CHO-K1 clone (clone 9) and a nontargeted CHO-K1 clone (clone 2). (F) hHCN4 current densities for 9 tested clones obtained with the targeted system. The currents were measured at the end of a 5-s hyperpolarizing pulse at −110 mV and normalized to the cell capacitance.

Saturation assays and functional characteristics of recombinant melatonin receptors expressed in cGPS CHO-K1 targeted cells

From cGPS CHO-K1/hMT₁ and cGPS CHO-K1/hMT₂ cells, an analysis of the saturation data using 1-site and 2-site binding hyperbola fits (F-test, GraphPad Prism) revealed the presence of a single high affinity binding site for 2-[¹²⁵I]-melatonin ( Fig. 3C ). A Scatchard plot of the saturation data gave a K_D of 0.034 ± 0.09 nM (pK_D = 10.47 ± 0.07) and a number of B_max sites of 45 ± 11 fmol/mg total protein in cGPS CHO-K1/hMT₁ cells ( Fig. 3C ). A similar pK_D was obtained from CHO-K1/hMT₁ clones 2 and 7 (pK_D = 10.50 ± 0.07). A Scatchard plot of the saturation data gave a K_D of 0.076 ± 0.015 nM (pK_D = 10.12 ± 0.1) and a number of B_max sites of 33 ± 15 fmol/mg total protein in the cGPS CHO-K1/hMT₂ cells (data not shown). A similar pK_D was obtained from CHO-K1/hMT₂ clone 2 (pK_D = 10.06). The pK_D obtained from cGPS CHO-K1/hMT₁ and cGPS CHO-K1/hMT₂ clones were similar to the pK_D described for human endogenous hMT₁ and hMT₂ (pK_D = 10.85 ± 1.15 and pK_D = 9.9 ± 0.9, respectively).

The functional activity of recombinant melatonin receptors hMT₁ and hMT₂ was studied using cellular dielectric spectroscopy. This technology is based on the change in intercellular impedance, mainly the resistance to an external electric current, of a monolayer of cells.^32,33 Upon stimulation of the receptor with 100 nM melatonin, the signaling cascade eventually leads to minute changes in cell adherence and/or cell shape, resulting in a change in the intercellular impedance. Interestingly, the kinetics and values of the impedance signal are dependent on the G-protein subtype involved in the cascade (see Verdonk et al.³² for examples and theory). We obtained typical G_i-coupling kinetics upon melatonin stimulation of cGPS CHO-K1/hMT₁ cells, corroborating that the recombinant hMT₁ receptor is G_i-coupled ( Fig. 3D ). The stimulation EC₅₀ of melatonin in this system is 0.75 nM, which is in agreement with the high subnanomolar affinity of melatonin for its receptors. Similar results were obtained with hMT₂ (data not shown). In addition, after a 24-h treatment of cells with pertussis toxin, a G_i-protein inhibitor, the melatonin response was abolished (data not shown). The hMT₁ and hMT₂ receptors were mainly coupled to G_i-proteins when expressed in cGPS CHO-K1 cells, as no Gs or Gq signals were detected in these functional studies (data not shown). The results obtained with the recombinant hMT₁ and hMT₂ receptors were similar to those for the endogenous melatonin receptors (EC₅₀ = 0.65 nM and EC₅₀ = 0.43 nM, respectively).

Functional characteristics of the recombinant hHCN4 channel expressed in cGPS CHO-K1 targeted cells

Immunostaining of fixed and permeabilized cGPS CHO-K1/hHCN4 cells with fluorescent anti-hHCN4 antibody revealed major expression of the recombinant hHCN4 channel at the plasma membrane (data not shown), whereas no fluorescent signal was detected in cGPS CHO-K1 cells transfected with the vector alone.

However, the patch-clamp technique represents the gold-standard method for recording ionic channel activity and testing drug efficacy on them. In the whole-cell configuration, it allows the recording of transmembrane ionic current generated by all the channels embedded in the sarcolemmal membrane of a single cell.³⁴ Thus, the size of the signal is directly proportional to the number of channels expressed by the cell. One of the main difficulties encountered with this time-consuming technique is to obtain a maximal number of cells expressing a satisfactory amount of functional channels. This problem is further enhanced with the emerging use of automated-based patch-clamp devices for higher throughput data generation.²⁸ An ideal cell line should therefore provide a sufficient level of channel expression with the smallest cell-to-cell variation, ensuring a good signal-to-noise ratio of the recordings, a good reproducibility, and a high success rate of the experiments.

hHCN4 expressed in cGPS CHO-K1 and CHO-K1 cells led to the generation of comparable hyperpolarization-activated inward currents with biophysical properties similar to those described in the literature ( Fig. 3E ). The cGPS CHO-K1 targeted strategy had very good reproducibility for obtaining stable cell lines with a similar level of expression. The current density that reflects the number of channels per unit of membrane surface was comparable between all tested clones ( Fig. 3F ), and the large majority of the cells (>83%) exhibited a current in the ideal range for whole-cell patch-clamp recordings, between 500 and 3000 pA ( Table 2 ). These results show that the cGPS CHO-K1 system is particularly suitable for heterologous ionic channel expression for patch-clamp recording in whole-cell mode. The reproducibility and homogeneity on the cell-to-cell and clone-to-clone level of expression make it a very interesting tool for channel study with an automated patch-clamping system.

Table 2.

Current Measurement in Clones Having Targeted or Nontargeted Integration of hHCN4

		Number of Cells Displaying Current
Clone	Tested Cells	<500 pA	500-3000 pA, n (%)	>3000 pA
1	27	1	25 (92.6)	1
2	27	5	21 (77.8)	1
3	26	2	22 (84.6)	2
4	28	2	24 (85.7)	2
5	29	1	27 (93.1)	1
6	26	4	22 (84.6)	0
7	26	4	21 (80.8)	1
8	24	7	16 (69.6)	0
9	24	4	20 (83.3)	0
10	118	8	106 (89.8)	4
R2	15	12	3 (20.0)	0
R4	12	4	8 (66.6)	0

Clones have been obtained from the cGPS CHO-KI system (1-10) or from classical transfection of CHO-K1 (R2, R4).

Expression stability in cGPS CHO-K1 targeted clones

The β-galactosidase activity in cGPS CHO-K1/lacZ cells was measured 2 ways. First, cGPS CHO-K1/LacZ clones were analyzed by fluorescence-activated cell sorting (FACS) using a fluorescent substrate for β-galactosidase over 20 weeks. This method allows appreciation of the homogeneity of expression in cells belonging to the same clonal population. A symmetric and narrow fluorescent signal was observed ( Fig. 4A ), indicating that the β-galactosidase expression was homogeneous. Furthermore, the same kind of narrow signal was observed after 2 (passage 4) or 20 weeks (passage 40) of culture and was observed for all tested cGPS CHO-K1/LacZ cell clones (data not shown). Second, the level of β-galactosidase expression for 4 cGPS CHO-K1/LacZ clones was measured over at least 30 passages in the presence or absence of selection agents ( Fig. 4B,C ). The expression of β-galactosidase was remarkably stable, even after a long culture period. Interestingly, the presence of the selection agents was not necessary to ensure long-lasting expression of the transgene because the reporter expression stability was equivalent whether the targeted clones were cultivated with or without selection agents. Analysis of current density in cGPS CHO-K1/hHCN4 clone 10 revealed good hHCN4 channel number stability at the cell surface over time. Less than 10% variability was observed between the 7th and 20th passages ( Fig. 4D ).

Fig. 4.

Homogeneity and stability of β-galactosidase and HCN4 expression in cGPS CHO-K1 clones. (A) β-Galactosidase activity was measured by fluorescence-activated cell sorting (FACS) over 20 weeks. FACS displays for weeks 2 and 20 are shown. The β-galactosidase activity of cGPS CHO-K1/LacZ clone 7 (red line) is compared with the parental cGPS CHO-K1 cell line (gray line). (B, C) β-Galactosidase expression was assayed by β-glo luminometric test for 4 independent cGPS CHO-K1/LacZ clones (ratio-to-signal/background) over (B) 45 passages in the presence of selection agents or (C) 30 passages without selection agents. (D) Passage effect on the hHCN4 current density assessed on 1 clone obtained with the targeted system. Current density in the cell membrane of cGPS CHO-K1/HCN4 clone 9 was analyzed over 20 passages. Channel density was calculated by normalizing the size of the recorded current (pA) to the cell capacitance (pF).

Discussion

Homologous gene targeting is a powerful method for integrating different constructs at the same locus displaying the required properties for transgene expression, but the efficacy of the process is usually low. The use of meganucleases, which strongly stimulate the process, results in the frequencies of targeted events.¹⁴ Furthermore, the double selection process we devised appeared to efficiently identify such targeted events; among 48 clones characterized at the genomic level, all appeared to be a targeted event in Southern blots, as shown by the loss of the parental pattern and/or gain of a specific band corresponding to targeting the landing pad locus. Moreover, 75% were “pure” targeting events with no additional transgene integration and no further rearrangement of the targeted locus, as described previously.^16,31 Compared the classical cell transformation that we conducted in parallel, this process was not only more precise but also faster; the targeted clones were obtained in 28 days instead of 44. Thus, there was no trade-off between the precision of the approach and the time to implement it but rather a double advantage for the cGPS process. This speed is comparable with what has been described with other targeted methods based on recombinases or integrases or with the LEAP approach (see above). However, in contrast with other systems, the cGPS system allows for the specific integration of an expression cassette without the plasmid backbone ( Fig. 1 ), avoiding the integration of unwanted long stretches of bacterial sequences, which could interfere with stable expression.

Functional characterization of targeted clones also provided convincing arguments for the potential of cGPS technology. All cGPS targeted clones expressed a functional protein, compared only 17.5% of randomly inserted clones. Furthermore, the expression/activity is generally homogeneous from clone to clone and close to the level of expression in a polyclonal double-resistant population. Given the high percentage of perfect targeting events among clones and the relatively homogeneous expression of the GOI, working with a polyclonal population resulting from double selection with G418 and puromycin seems to be sufficient ( Fig. 2 ) and would bypass the isolation of clones, further shortening the cell line derivation process. In the case of pharmaceutical targets, achieving a physiological level of expression compatible with the relevant functional assay is important. We found that proteins expressed in the cGPS CHO-K1 cells have hATX enzymatic constants (K_m and V_max), hMT₁ and hMT₂ pK_D and EC₅₀ values, and hHCN4 subcellular localization and function close to the published values for their wild-type endogenous counterparts. We could also show that 2 GOIs coding for a heterodimeric antibody can be inserted in the same locus in cGPS and efficiently expressed (J. P. Cabaniols, unpublished results). These results suggest that the cGPS system could address another important issue linked with random integration techniques, for unbalanced expression of 2 monomers expressed from 2 different insertion sites can have far-reaching consequences. Altogether, these results, which show that cGPS can easily match other systems based on integrases,^9-12 demonstrate that the landing pad we selected is appropriate for expression of GOIs in cell-based assays.

Finally, one of the most important drawbacks of randomly inserted transgenes is the instability of expression after a long period of culture. This aspect is important because a period of cell amplification is necessary before performing the HTS, and cells should be able to sustain a constant level of expression. In most cases of random insertion, the GOI is silenced because of multiple copy insertion, adding another level of complexity. Moreover, having a stable cell sustaining GOI expression without a selection agent during the amplification step is an invaluable advantage before running the cell-based assay. Here, we demonstrate that targeting genes in the cGPS CHO-K1 locus allow for long and sustainable expression, even in the absence of selection drugs.

Altogether, these data demonstrate that the cGPS system provides a fast, convenient, and efficient method of deriving stable clones with genes of therapeutic interest for HTS studies, as well as other research use.

References

Moore

Rees

: Cell-based versus isolated target screening: how lucky do you feel? J Biomol Screen 2001;6:69-74.

Pereira

Williams

: Origin and evolution of high throughput screening. Br J Pharmacol 2007;152:53-61.

Cabaniols

Mathis

Delenda

: Targeted gene modifications in drug discovery and development. Curr Opin Pharmacol 2009;9:657-663.

Chen

Lake

Sabet

Niforatos

Pratt

Cassar

: Utility of large-scale transiently transfected cells for cell-based high-throughput screens to identify transient receptor potential channel A1 (TRPA1) antagonists. J Biomol Screen 2007;12:61-69.

Jenkinson

McCoy

Kerner

Ferris

Lawrence

Clay

: Development of a novel high-throughput surrogate assay to measure HIV envelope/CCR5/CD4-mediated viral/cell fusion using BacMam baculovirus technology. J Biomol Screen 2003;8:463-470.

Wang

Walsh

Payne

Modha

Scott

: Development of a high-throughput cell-based assay for 11beta-hydroxysteroid dehydrogenase type 1 using BacMam technology. Mol Biotechnol 2008;39:127-134.

Koller

Hanania

Stevens

Eisfeld

Sasaki

Fieck

: High-throughput laser-mediated in situ cell purification with high purity and yield. Cytometry A 2004;61:153-161.

Birling

Gofflot

Warot

: Site-specific recombinases for manipulation of the mouse genome. Methods Mol Biol 2009;561:245-263.

Nair

Avila

Wang

Lennartz

Darnay

: A novel high-throughput screening system identifies a small molecule repressive for matrix metalloproteinase-9 expression. Mol Pharmacol 2008;73:919-929.

10.

Zhu

Appel

Johnson

Chavin

Markowitz

: Role of carboxylesterase 1 and impact of natural genetic variants on the hydrolysis of trandolapril. Biochem Pharmacol 2009;77:1266-1272.

11.

Liu

Lakshmipathy

Ozgenc

Thyagarajan

Lieu

Fontes

: hESC engineering by integrase-mediated chromosomal targeting. Methods Mol Biol 2010;584:229-268.

12.

Lieu

Machleidt

Thyagarajan

Fontes

Frey

Fuerstenau-Sharp

: Generation of site-specific retargeting platform cell lines for drug discovery using phiC31 and R4 integrases. J Biomol Screen 2009;14:1207-1215.

13.

Chevalier

Stoddard

: Homing endonucleases: structural and functional insight into the catalysts of intron/intein mobility. Nucleic Acids Res 2001;29:3757-3774.

14.

Paques

Duchateau

: Meganucleases and DNA double-strand break-induced recombination: perspectives for gene therapy. Curr Gene Ther 2007;7:49-66.

15.

Choulika

Perrin

Dujon

Nicolas

: Induction of homologous recombination in mammalian chromosomes by using the I-SceI system of Saccharomyces cerevisiae . Mol Cell Biol 1995;15:1968-1973.

16.

Rouet

Smih

Jasin

: Introduction of double-strand breaks into the genome of mouse cells by expression of a rare-cutting endonuclease. Mol Cell Biol 1994;14:8096-8106.

17.

Redondo

Prieto

Munoz

Alibés

Stricher

Serrano

: Molecular basis of xeroderma pigmentosum group C DNA recognition by engineered meganucleases. Nature 2008;456:107-111.

18.

Cabaniols

Paques

: Robust cell line development using meganucleases. Methods Mol Biol 2008;435:31-45.

19.

Tokumura

Majima

Kariya

Tominaga

Kogure

Yasuda

: Identification of human plasma lysophospholipase D, a lysophosphatidic acid–producing enzyme, as autotaxin, a multifunctional phosphodiesterase. J Biol Chem 2002;277:39436-39442.

20.

Brindley

: Lipid phosphate phosphatases and related proteins: signaling functions in development, cell division, and cancer. J Cell Biochem 2004;92:900-912.

21.

Clair

Lee

Liotta

Stracke

: Autotaxin is an exoenzyme possessing 5′-nucleotide phosphodiesterase/ATP pyrophosphatase and ATPase activities. J Biol Chem 1997;272:996-1001.

22.

Giganti

Rodriguez

Fould

Moulharat

Coge

Chomarat

: Murine and human autotaxin alpha, beta, and gamma isoforms: gene organization, tissue distribution, and biochemical characterization. J Biol Chem 2008;283:7776-7789.

23.

Stracke

Krutzsch

Unsworth

Arestad

Cioce

Schiffmann

: Identification, purification, and partial sequence analysis of autotaxin, a novel motility-stimulating protein. J Biol Chem 1992;267:2524-2529.

24.

Audinot

Mailliet

Lahaye-Brasseur

Bonnaud

Le Gall

Amosse

: New selective ligands of human cloned melatonin MT1 and MT2 receptors. Naunyn Schmiedebergs Arch Pharmacol 2003;367:553-561.

25.

Ludwig

Zong

Hofmann

Biel

: Structure and function of cardiac pacemaker channels. Cell Physiol Biochem 1999;9:179-186.

26.

Wahl-Schott

Biel

: HCN channels: structure, cellular regulation and physiological function. Cell Mol Life Sci 2009;66:470-494.

27.

Maher

Guo

Dubin

Chaplan

Wickenden

: HCN channels as targets for drug discovery. Comb Chem High Throughput Screen 2009;12:64-72.

28.

Dunlop

Bowlby

Peri

Vasilyev

Arias

: High-throughput electrophysiology: an emerging paradigm for ion-channel screening and physiology. Nat Rev Drug Discov 2008;7:358-368.

29.

Razzell

Khorana

: Studies on polynucleotides: III. Enzymic degradation: substrate specificity and properties of snake venom phosphodiesterase. J Biol Chem 1959;234:2105-2113.

30.

Laemmli

: Cleavage of structural proteins during the assembly of the head of bacteriophage T4. Nature 1970;227:680-685.

31.

Smih

Rouet

Romanienko

Jasin

: Double-strand breaks at the target locus stimulate gene targeting in embryonic stem cells. Nucleic Acids Res 1995;23:5012-5019.

32.

Verdonk

Johnson

McGuinness

Leung

Chen

Tang

: Cellular dielectric spectroscopy: a label-free comprehensive platform for functional evaluation of endogenous receptors. Assay Drug Dev Technol 2006;4:609-619.

33.

Peters

Knappenberger

Wilkins

Sygowski

Lazor

Liu

: Evaluation of cellular dielectric spectroscopy, a whole-cell, label-free technology for drug discovery on Gi-coupled GPCRs. J Biomol Screen 2007;12:312-319.

34.

Hamill

Marty

Neher

Sakmann

Sigworth

: Improved patch-clamp techniques for high-resolution current recording from cells and cell-free membrane patches. Pflugers Arch 1981;391:85-100.