A description of how to calibrate Förster Resonance Energy Transfer integrated biological sensors (FIBS) for in situ metabolic profiling is presented. The FIBS can be used to measure intracellular levels of metabolites noninvasively aiding in the development of metabolic models and high throughput screening of bioprocess conditions.
In the era of computational biology, new high throughput experimental systems are necessary in order to populate and refine models so that they can be validated for predictive purposes. Ideally such systems would be low volume, which precludes sampling and destructive analyses when time course data are to be obtained. What is needed is an in situ monitoring tool which can report the necessary information in real-time and noninvasively. An interesting option is the use of fluorescent, protein-based in vivo biological sensors as reporters of intracellular concentrations. One particular class of in vivo biosensors that has found applications in metabolite quantification is based on Förster Resonance Energy Transfer (FRET) between two fluorescent proteins connected by a ligand binding domain. FRET integrated biological sensors (FIBS) are constitutively produced within the cell line, they have fast response times and their spectral characteristics change based on the concentration of metabolite within the cell. In this paper, the method for constructing Chinese hamster ovary (CHO) cell lines that constitutively express a FIBS for glucose and glutamine and calibrating the FIBS in vivo in batch cell culture in order to enable future quantification of intracellular metabolite concentration is described. Data from fed-batch CHO cell cultures demonstrates that the FIBS was able in each case to detect the resulting change in the intracellular concentration. Using the fluorescent signal from the FIBS and the previously constructed calibration curve, the intracellular concentration was accurately determined as confirmed by an independent enzymatic assay.
Metabolite monitoring has various applications in bioprocessing, including process development, media and feed design, and metabolic engineering. Various methods are available for concentration measurements through the use of enzymatic1, chemical2, or binding assays3. An interesting option is the use of fluorescent, protein-based in vivo biological sensors as reporters of the intracellular concentration of key metabolites. In ultra-low volume assays, fluorescence is a convenient tool as miniaturization actually improves the signal-to-noise ratio4,5 and protein-based sensors can be genetically encoded meaning that no exogenous reagents are necessary for metabolite analysis. Förster Resonance Energy Transfer (FRET) biosensors consist of two fluorescent proteins connected by a ligand binding domain. FRET is a nonradiative transfer of energy from a photo-excited donor to an acceptor fluorescent molecule located in close proximity (<100). Ligand cleavage or binding causes a conformational change in the sensor, which, in turn, induces a change in the proximity of the fluorophores, leading to a change in FRET efficiency measured by the change in the emission spectrum. FRET integrated biological sensors (FIBS) are constitutively produced within the cell line and their spectral characteristics change based on the concentration of metabolite within the cell. FIBS have fast response times making them ideal for taking measurements for dynamic models6. Previous applications include monitoring single7-9 and multiple10 metabolites and providing data on spatiotemporal distribution11. FIBS can be created in two configurations: Apo-Max, where ligand binding disrupts the proximity of the fluorophores lowering energy transfer and Apo-Min, where ligand binding brings the two fluorophores into closer contact (Figure 1).
In this work, a protocol is presented for constructing Chinese hamster ovary (CHO) cell lines that constitutively express a FIBS for a metabolite, glucose or glutamine. A methodology is established for the calibration of the sensor in vivo to enable future quantitative measurements of intracellular metabolite concentration, as presented in Figure 2. Based on this, the intracellular concentration of glucose or glutamine can be determined in fed-batch CHO cell cultures, to which the two nutrients were added in high concentrations in-process. The results demonstrate that using the fluorescent signal from the FIBS and the previously constructed calibration curve, accurately prediction of the intracellular concentration is possible, as confirmed by an independent enzymatic assay. This method offers substantial advantages over current analytical technologies because it is noninvasive, low-cost and fast, giving a real-time signal of the FIBS that can be monitored throughout the culture.
1. Cell Line Revival and Maintenance
2. Transfection of the Cell Line with the Plasmid Containing the Biosensor Gene
3. Batch and Fed-batch Cell Growth Curve
4. FRET Ratio Measurements
5. Metabolite Assays
An overview of the methodology is presented in Figure 2. In the work presented herein, CHO cells were transfected with the FIBS vector and stable cell lines were selected at an antibiotic pressure of 400 μg/ml zeocin. Two separate stable cell lines constitutively expressing the glucose and glutamine sensors were therefore created. Figure 1 depicts the configurations of the two biosensors used in this study. The glucose sensor is based on the Apo-Max principle and the corresponding plasmid is pRSET-FLIPglu600μΔ11Aphrodite. This is a fusion of the enhanced cyan fluorescent protein (ECFP) and the Aphrodite variant of the enhanced yellow fluorescent protein (EYFP) to the F16A mutant of the E. coli glucose/galactose periplasmic binding protein. 11 amino acid residues were removed from the linker region to improve FRET efficiency13. This vector was cloned into the pCDNA4/TO vector using the unique EcoRI and PstI restriction sites, which uses a Cytomegalovirus promoter and a SV40 enhancer.
The glutamine sensor is based on the Apo-Min principle (Figure 1) and the corresponding plasmid was derived from an insertion of the yellow fluorescent protein citrine into the E. coli glutamine periplasmic binding protein between amino acids 98 and 99 with an ECFP attached to the C-terminus. The structure has been rigidified by deletion of residues in the linking region, while the amino acid substitution D157N has been made to the QBP to increase affinity for glutamine, followed by several mutations to optimize FRET efficiency8. The glutamine FRET construct was supplied in the pUTKan plant expression vector and was digested with BamHI and SalI to release the insert. The latter was cloned into the corresponding restriction sites in the vector pET41a in frame with the N-terminal purification tags. The same insert was ligated into the pCDNA4/TO vector, after it had been digested with BamHI and XhoI using the compatible sticky ends produced by SalI and XhoI, to produce the mammalian expression vector.
Plasmid expression was confirmed by quantifying the mRNA levels of the biosensors using qRT-PCR. Both FIBS were found to be actively transcribed at 25-44 mRNA copies per cell assuming β-actin at 3,000 transcripts per cell at the mid-exponential level14.
Batch overgrow cultures of each cell line were maintained taking daily samples for the in vivo calibration of each FIBS. The viable cell growth profile of the two cell lines is shown in Figure 3 and the corresponding FRET measurements and intracellular metabolite concentrations determined by enzymatic assays are presented in Table 1. A fluorescence plate reader was used to measure fluorescence of samples at an excitation wavelength of 430/35 nm and emission wavelengths of 465/435 nm for cyan and 520/510 nm for yellow. It was found that low cell numbers present in the first 4 days of culture led to unreliable FRET signal. Similarly, the high level of cell lysated present after day 8 of culture produced a high amount of light scattering. Only data for days 4-8 is therefore used to construct the calibration curves for the glucose and glutamine FIBS, the equations for which are shown in Table 1. The results show that the FIBS signal produces a reliable correlation with the intracellular concentrations between 1-5 mM for glucose and 0.3-2 mM for glutamine.
Following this and to validate the above findings, fed-batch overgrow cultures of the two cell lines were performed. A feed containing high substrate concentration (glucose in the case of the glucose FIBS and glutamine in the case of the glutamine FIBS) was supplemented on day 6 of the cell culture to elevate the extracellular concentrations of the substrate. In the case of the glucose-fed culture the extracellular glucose concentration was increased back up to 36 mM, while for the glutamine-fed culture, the glutamine concentration was increased to 4 mM. Figure 4 shows representative results from this study. Figures 4A and 4C depict the FRET ratios on each day, which clearly respond to feeding by reversing the trend they followed up to day 6. Specifically, in Figure 4A, the FRET ratio for the glucose FIBS CHO cell line decreases on day 7 in response to glucose addition, since the glucose FIBS follows the APO-On configuration. Similarly, in Figure 4C, the FRET ratio for the glutamine FIBS CHO cell line increases in response to glutamine addition, in line with the principle of Apo-Off sensors.
These FRET measurements were finally used to calculate the corresponding intracellular glucose and glutamine concentrations based on the aforementioned calibration curves and the results are shown in Figure 4B for glucose and Figure 4D for glutamine. They are compared to the actual intracellular concentrations of these substrates as determined with the glucose and glutamine enzymatic assays and show adequate agreement.
Table 1. Results for in vivo calibration of FRET ratio for glucose and glutamine FIBS. The calibration curves are described by the following equations that can be used to make predictions: y= -1.232x + 8.034 for glucose (R2=0.961), and y= 1.892x + 0.332 for glutamine (R2=0.879), where y is the FRET ratio and x is the concentration in mM.
Figure 1. Principles of the FIBS used in this study. The glutamine FIBS (left) follows the Apo-Min principle, by which glutamine binds to the sensing domains and generates a conformational shift that brings both fluorescent proteins closer together, thus increasing FRET. The glucose FIBS (right) follows the Apo-Max principle, by which glucose binds to the sensing domain and generates a conformational change that separates the fluorescent proteins, thus decreasing FRET.
Figure 2. Overview of the methodology for FIBS calibration and subsequent noninvasive quantification of intracellular glucose and glutamine concentrations.
Figure 3. Viable cell concentration profile for batch overgrow cultures of glucose FIBS- and glutamine FIBS-expressing CHO cell lines (n=2 biological repeats with three measurements per cell line).
Figure 4. Fed batch culture of FIBS-producing CHO cells and corresponding FRET measurements. (a) Daily measurements of FRET ratio of fed-batch CHO cell cultures expressing glucose FIBS. The arrow indicates the addition of glucose to the cell culture. (b) Actual intracellular glucose concentration profile measured using a glucose assay kit against values predicted based on in vivo glucose calibration curve (Table 1). (c) Daily measurements of FRET ratio of fed-batch CHO cell cultures expressing glutamine FIBS. The arrow indicates the addition of glutamine to the cell culture. (d) Actual intracellular glutamine concentration profile measured using a glutamine assay kit against values predicted based on in vivo glutamine calibration curve (Table 1). In all cases, n=2 biological repeats with three measurements per cell line.
The FIBS enable in vivo and in situ quantitation of key molecules, in this case growth-limiting nutrients, removing uncertainties arising from quenching and extraction methods. The findings suggest that there is a good correlation between the FIBS signal and the intracellular concentrations in the range of 1-5 mM for glucose and 0.3-2 mM for glutamine. In batch CHO cell culture, these concentrations are encountered in the exponential, stationary, and early decline phases. The exponential and stationary phases are the most critical in terms of the design of appropriate feeding strategies; the use of these biosensors for estimating the metabolic requirements of the cells is therefore industrially relevant. The results of the fed-batch experiment confirm the FIBS’ suitability to provide estimates of intracellular glucose and glutamine concentrations. Deviations exhibited at earlier time points may have arisen because of lower viable cell numbers and/or because of interference by other metabolites, e.g. by galactose in the case of the glucose FIBS, and by other amino acids in the case of the glutamine FIBS. Specifically, it has been shown that a reduction of up to 20% in the FRET ratio can occur in the presence of aspartate, glutamate, glycine, threonine, valine, and tyrosine15. Overall, these results suggest that error in the FIBS signal is fairly low.
Although this methodology can only monitor a limited number of molecules at a time, it can provide useful information for cell line and process development, or even data acquisition for the validation of mathematical models. Given the diversity of periplasmic binding proteins, similar biosensors can be developed for an array for small molecules including other amino acids, sugars, and ions such as sulfate and phosphate16, making this a versatile method for metabolite quantification in real-time in living cells.
A key benefit of the use of FIBS is the ability to obtain online measurements potentially in a continuous fashion. The method therefore lends itself for high-throughput applications which can reduce manual handling and expedite development efforts. With the increased availability of microreactor environments with built-in functionality for fluorescence detection, it is envisaged that a FIBS-enabled metabolite profiling platform can be used for cell screening, media and process optimization, and process scale-up.
The authors have nothing to disclose.
We thank Professor Wolf Frommer (Carnegie Institution for Science, Stanford University) for kindly providing us with the glucose FRET plasmid and Dr. Uwe Ludewig (Hohenheim University) for kindly supplying the glutamine FRET construct in the pUTKan plant expression vector. AB is funded by the BBSRC Targeted Priority Studentships program. Both CK and KP are supported by RCUK Fellowships in Biopharmaceuticals Processing. CK also wishes to thank Lonza Biologics for their financial support. The Centre for Synthetic Biology and Innovation is generously supported by the EPSRC.
CHO-S Cells | Life Tecnologies | 11619-012 | Cell line will vary depending on the goals of the study |
pCDNA4/TO vector | Life Tecnologies | ||
TransIT-PRO Transfection Reagent | Mirus Bio | MIR 5700 | Transfections can be accomplished using any method suitable for the cell line under study |
Zeocin | Invivogen | ||
CD-CHO medium | Life Technologies | 10743-011 | Cell growth medium is dependent upon the cells under study |
100X HT Supplement | Life Technologies | 11067-030 | |
L-Glutamine 200 mM (100X), Liquid | Life Technologies | 25030032 | |
InfinitePRO 200 plate reader | Tecan | FLx800TBI | Any 96-well fluorescence plate reader that can access the required wavelengths can be substituted |
Filters for plate reader | Tecan | 30000463 | |
(520NM BW 10NM) | |||
30022786 | |||
(430NM BW 35NM) | |||
30022787 | |||
(465NM BW 35NM) | |||
Maxiprep Plasmid Purification Kit | Qiagen | 12163 | Any suitable kit can be substituted |
Amplex Red glucose/glucose oxidase assay kit | Invitrogen | A22189 | Any suitable kit can be substituted |
EnzyChrom Glutamine Assay Kit | BioAssay systems | EOAC-100 | Any suitable kit can be substituted |
Improved Neubauer haemocytometer | Fisher Scientific | MNK-420-010N | |
Incubator | Nuaire | NU-5510E |