Transcription Factor Sensor System for Parallel ... - ACS Publications

Article pubs.acs.org/ac

Transcription Factor Sensor System for Parallel Quantification of Metabolites On-Chip Simon Ketterer,†,‡,∇ Désirée Hövermann,‡,§,∇ Raphael J. Guebeli,‡,§,∥ Frauke Bartels-Burgahn,‡,§ David Riewe,⊥ Thomas Altmann,⊥ Matias D. Zurbriggen,‡,§ Björn Junker,# Wilfried Weber,*,‡,§,∥ and Matthias Meier*,†,‡ †

Microfluidic and Biological Engineering, Department of Microsystems Engineering, University of Freiburg, Georges-Koehler-Allee 103, 79110 Freiburg, Germany ‡ Centre for Biological Signalling Studies, and §Faculty of Biology, University of Freiburg, Schänzlestrasse 18, 79104 Freiburg, Germany ∥ Spemann School of Biology and Medicine, University of Freiburg, Albertstrasse 19a, 79104 Freiburg, Germany ⊥ Department of Molecular Genetics, Leibniz Institute of Plant Genetics and Crop Plant Research, Corrensstrasse 3, 14471 Gatersleben, Germany # Institute of Pharmacy, Martin-Luther-University, Hoher Weg 8, 06120 Halle, Germany S Supporting Information *

ABSTRACT: Steadily growing demands for identification and quantification of cellular metabolites in higher throughput have brought a need for new analytical technologies. Here, we developed a synthetic biological sensor system for quantifying metabolites from biological cell samples. For this, bacterial transcription factors were exploited, which bind to or dissociate from regulatory DNA elements in response to physiological changes in the cellular metabolite concentration range. Representatively, the bacterial pyruvate dehydrogenase (PdhR), trehalose (TreR), and L-arginine (ArgR) repressor proteins were functionalized to detect pyruvate, trehalose-6-phosphate (T6P), and arginine concentration in solution. For each transcription factor the mutual binding behavior between metabolite and DNA, their working range, and othogonality were determined. High-throughput, parallel processing, and automation were achieved through integration of the metabolic sensor system on a microfluidic large-scale integration (mLSI) chip platform. To demonstrate the functionality of the integrated metabolic sensor system, we measured diurnal concentration changes of pyruvate and the plant signaling molecule T6P within cell etxracts of Arabidopsis thaliana rosettes. The transcription factor sensor system is of generic nature and extendable on the microfluidic chip.

A

family for analytical use are prokaryotic transcription factors with binding affinities to amino acids, carbohydrates, nucleotides, vitamins, and fatty acids. These metabolic sensors are evolutionary optimized proteins that naturally quantify key metabolites at the intersection of metabolic pathways. Upon changes in cellular metabolite concentration, the transcription factors either bind to or dissociate from a regulatory DNA element. In response to this molecular event transcription of mRNAs encoding for functional proteins associated with the metabolite homeostasis are either activated or repressed.9 Hallmarks of metabolite-responsive transcription factors include high binding specificity and affinity toward the corresponding metabolite and activation within a physiological concentration window of the metabolite.

dvances in metabolite analytics rely on sensitive detection and quantification methods of low molecular weight molecules from different cellular composites. Routinely, metabolites are quantified by using liquid chromatography coupled to mass spectrometry.1−3 Alternative quantitation methods exist for only a few of eukaryotic cell metabolites.4,5 Biosensors have been developed for the detection of metabolites by converting metabolic concentration information into analytically useful signals, e.g., fluorescence or electrochemical potential.6 The most prominent examples of a biosensor system are enzymatic assays for the determination of glucose levels in human blood samples.7,8 A central problem for biosensor development is the identification, selection, functionalization, and integration of suitable proteins for technical application. Recent functional genomic research has identified a vast variety of regulatory biomolecules of potential interest for analytical investigations of metabolites. One attractive protein © 2014 American Chemical Society

Received: August 19, 2014 Accepted: November 20, 2014 Published: November 20, 2014 12152

dx.doi.org/10.1021/ac503269m | Anal. Chem. 2014, 86, 12152−12158

Analytical Chemistry

Article

tion. Next, the composite was incubated at 4 °C for 15 min under shaking conditions before addition of 0.4 mL of H2O. After centrifugation of the extraction samples (12 500g, 10 min, 4 °C), the supernatant of the polar phase was isolated and dried using a Speed Vac concentrator. All samples were stored at −20 °C until use. For an experiment 0.01 g of dried extract samples was solved to a final concentration of 1 g FW/mL in dilution buffer (250 mM NaCl, 20 mM MgCl2, 1 mM DTT, 1% w/v BSA, 0.05% w/v Tween). For the determination of the diurnal concentration changes of pyruvate and T6P about 0.1 g FW of rosettes was collected in a time interval of 4 h. PDMS Chip Production and Run Process. Molds for rapid prototyping12 of the flow and control layer of the PDMS chip were fabricated by using SU-8 3025 (MicroChem, MA) and AZ 9260 (MicroChemicals, Germany), respectively. The feature heights of the control and flow mold were 27 ± 1 and 18 ± 1 μm, respectively. PDMS chips were manufactured following the standard procedure for multilayer devices.13 Pressure-driven flow and valve operations on-chip were automated using a custom-made pressure control unit, equipped with solenoid valves (SMC, Germany). The interface of the control unit was addressed with Matlab (Mathworks, MA). The surface chemistry to deposit the DNA onto the epoxy glass surface within the unit cells of the microfluidic chip has been described in detail before.14,15 The surface DNA concentration was adjusted by mixing the DNA with the binding sequence for each TF with the control DNA at different ratios. Both dsDNA types carried a biotin group at the 5′-end of the sense strand. Metabolites and TF were mixed in dilution buffer before introduction on the chip platform. The final concentrations of the PdhR, TreR, and ArgR TF were 0.74, 0.74, and 0.77 pg/mL, respectively. For the cross-reactivity the TFs were mixed before the metabolites were added. TFs bound to the deposited DNA on the pull-down area on-chip were detected with a mouse monoclonal anti-V5 antibody conjugated to Cy-3 fluorophore (excitation 549 nm) at a concentration of ∼2 μg/mL (Sigma-Aldrich) in dilution buffer. Image Analysis. Two fluorescence images at 494 and 549 nm excitation wavelength next to a bright-field image were acquired on a Zeiss-Axio Observer microscope for each unit cell of the chip. Fluorescence signals of the 640 pull-down areas from a microfluidic chip were extracted by image analysis with Matlab (Mathworks, MA). For each metabolic measurement four repeats were performed on-chip. Mean and error values were calculated from the repeats. We used the fluorescence signal of 32 control spots for the determination of the limit of detection (LOD).16

In this work we use a synthetic biological approach to engineer a general analytical sensor system for metabolites by exploiting the binding properties of transcription factors (TF) on a microfluidic large-scale integration (mLSI) chip.10,11 The poly(dimethylsiloxane) (PDMS) platform consists of 640 functional unit cells. In each unit cell a miniaturized pulldown assay for a metabolite responsive TF is constructed with sequential flush operations. Thereby, double-stranded DNA (dsDNA) containing the TF binding sequence is deposited onto a spatially defined surface area. TF binding to the cognate DNA is then exploited to quantitate metabolites in different analytical solutions. The study exemplarily integrates the TFs sensitive to pyruvate, trehalose-6-phosphate (T6P), and Larginine. In a first developmental step the mutual binding behavior between the TF’s metabolites and DNA sequences with the TF binding motif is determined in order to obtain the working ranges of the sensors. In a following step, the orthogonality of the three TFs in a multiplexed assay for the metabolites is verified. Finally, the mLSI integrated TF sensor system is applied to quantify the three metabolites in cell extracts from rosettes of the model organism Arabidopsis thaliana.

■

EXPERIMENTAL SECTION DNA Sequences. All DNA sequences used for binding experiments with the TF are given in Supporting Information Table S1. All sense oligonucleotides except for the control DNA were labeled with 6-FAM (excitation 494 nm), whereas all antisense oligonucleotides were labeled with biotin at the 5′end (Sigma-Aldrich, Germany). Expression Vector Construction. The expression plasmids pRG001, pRG006, and pRG016 for hexahistidine- and V5-tagged transcription factors PhdR, ArgR, and TreR were generated by introducing PCR-amplified gene sequences (see Supporting Information Table S2) out of chromosomal or plasmid DNA into the TOPO vector pET101/D-TOPO (Invitrogen GmbH, Germany), thereby resulting in bacterial expression vectors for PhdR, ArgR, and TreR under the control of the phage T7 promoter. Protein Expression and Purification. The expression vectors pRG001, pRG006, and pRG016 were transformed into E. coli BL21 STAR (DE3) (Invitrogen, U.S.A.). TF production was induced in LB medium at OD 600 = 0.8 with 1 mM isopropyl β-D-1-thiogalactopyranoside (IPTG) for 4 h at 37 °C. The cells were harvested by centrifugation (6000g, 7 min, 4 °C). The resulting cell pellets were resuspended in lysis buffer (50 mM NaH2PO4, 300 mM NaCl, 10 mM imidazole, pH 8.0) and disrupted using a French press (Albertslund, Denmark) at 1000 bar. Clear cell lysates were purified on a gravity flow Ni2+−NTA−agarose Superflow column (Qiagen, Germany) following the vendor’s description. The procedure for the analytical characterization of the TFs is given in the Supporting Information. Plant Material and Metabolite Extraction. Rosettes were harvested from 5 week old wild-type A. thaliana ecotype Col-0 plants grown in soil placed in a growth chamber at 22 °C. The light intensity during a daily 16 h light photoperiod was adjusted to 80 μmol m−2 s−1. For one extraction sample, total rosettes of 5−6 Arabidopsis plants were pooled and immediately shock-frozen in liquid nitrogen. Freeze-dried material was ground into powder with a mortar and pestle, weighed, and diluted (0.5 mL per 0.1 g fresh weight [g FW]) in extraction buffer (2.5:1:1 v/v MeOH/CHCl3/H2O) for enzyme inactiva-

■

RESULTS AND DISCUSSION For sensing pyruvate, T6P, and L-arginine, the previously identified transcriptional repressors for pyruvate dehydrogenase (PdhR),17,18 trehalose (TreR),19,20 and arginine (ArgR)21,22 were used, respectively. Upon increasing metabolite concentration, PdhR and TreR dissociate from an operator DNA element, whereas ArgR associates with its cognate DNA operator. Figure 1A conceptually illustrates how to translate the binding behavior of the TFs into a readout signal that is proportional to the metabolite concentrations in solutions. TFs are mixed with the analytical sample containing the metabolite. The composite is incubated over a surface area with deposited DNA molecules carrying the binding motif of the TFs. Depending on the metabolite concentration, the TFs bind to 12153



Article

BB, except in the circular area of the button valve, which was actuated and thus protected during this step. After release of the button valve, biotinylated dsDNA with the TF binding sequence was introduced. Thus, a circular DNA pull-down area the same size of the button valve was obtained for the TF in each unit cell. The principle function of the metabolic sensor system on the mLSI chip platform was tested first in an experiment with PdhR. Here, double-stranded DNA with the PdhR binding sequence was deposited in unit cells of the chip. Then equal amounts of PdhR were introduced from the row direction in the presence of different pyruvate concentrations. After 10 min, the button valve was closed and the binding equilibrium in each unit cell was mechanically trapped and maintained. Unbound TFs and metabolites from the unit cells were removed with a buffer flush step. The cross-contamination of neighboring unit cells was avoided by leaving the button valve in a close state. The pressure of the button valve was released after 10 min for staining of the bound TF with fluorescently labeled antibodies. Figure 1D shows the corresponding fluorescence signals of the deposited dsDNA and bound PdhR of eight unit cells with different pyruvate concentrations. Increasing the pyruvate concentration from 0 to 10 mM led to a decrease in the fluorescence signal for the DNA bound PdhR, demonstrating the functionality of the miniaturized sensor assay. Control experiments, in which the button was released after binding of the detection antibody to the TFs at the end of the assay for more than 15 min, did not lead to a change of the fluorescence signal. The working range of the three TFs was determined by measuring the binding equilibriums between the TFs, dsDNA, and metabolites. In one chip run, the matrix arrangement of the unit cells was used to screen the two-dimensional binding space for a TF. The metabolite and dsDNA concentration varied within the rows and column elements, respectively, whereas the TF concentration was left constant. Variation of the DNA surface concentration on the pull-down areas was achieved by using different mole fractions of biotinylated dsDNA with and without the TF binding sequence (XDNA) during the build-up step of the pull-down assay. The relative surface concentration of the dsDNA with the TF binding sequence was determined by attaching a fluorescent label to the DNA. A linear dependence of the fluorescence signal on the pull-down area for the dsDNA with the TF binding sequence on its XDNA value in solution is given in Supporting Information Figure S1. A dilution series of the metabolites was then mixed with the TFs and introduced in separated row elements of the chip. The resulting isothermal binding curves of the PdhR, TreR, and ArgR are shown in Figure 2. The limit of detection (LOD) for each TF was determined in a control experiment with nonspecific dsDNA. The isothermal TF binding curves could be described with a heterotropic two-site linked binding model23 (solid line in Figure 2), which is derived in the Supporting Information. The calculated dissociation constants (Kd) from the binding model are given in Table 1. The Kd of PdhR and ArgR for pyruvate and L-arginine are in the millimolar range, whereas the Kd of TreR for T6P was in the micromolar range. This result is in agreement with previous determined TF−DNA dissociation constants for PdhR,18 TreR,19 and ArgR.22 The working ranges of the integrated metabolic sensor systems were obtained by calculating the derivative of the binding isotherms. The upper and lower boundary of the TF

Figure 1. Integrated metabolic sensor system on a microfluidic chip. (A) The molecular principle of metabolic sensor system. (B) Image of the mLSI chip platform. Flow and control channels of the multilayered PDMS chip are shown in blue and red, respectively. Flow channels on the chip exhibit a matrix design. Scale bar = 5 mm. (C) Operational unit cells of the mLSI chip. V1 and V2 denote for the pneumatic membrane valves to separate row and column elements of the matrix. BV denotes for the button valve. Scale bar = 100 μm. (D) Fluorescence signal of bound PdhR bound to its cognate DNA binding sequence in dependence of an increasing pyruvate concentration within a functionalized unit cell. Upper row: the fluorescence signal of the deposited DNA. Lower row: fluorescence signal of bound PdhR. Scale bar = 100 μm.

their cognate DNA sequence, where the degree of binding is directly proportional to the metabolite concentration. The amount of bound TF is then detectable with a fluorescently labeled antibody. For generalization of the detection step the TF included an antibody tag, i.e., a V5 tag at the C-terminal end of the proteins. It is clear that for each metabolite sensor system a different pull-down assay with matched TF and DNA binding sequence has to be constructed. In order to reduce the reagent materials per assay, and optimize parallelization and automation of the metabolite sensors system, we miniaturized the assay on an mLSI platform. The mLSI chip, based on multilayered PDMS, was bonded to an epoxy glass surface (Figure 1B). Fluidic microchannels were arranged in a matrix format with 16 rows and 40 columns on the chip platform. A functional unit cell was located at each cross section of the matrix. A magnified image of a microfluidic unit cell is shown in Figure 1C. All fluids are guided through the chip in the column or row direction upon actuation of integrated pneumatic membrane valves.13 All 16 rows have a single inlet port, whereas the 40 columns are addressed from 10 inlet ports. This chip design allows for the screening of two biophysical parameters in a 16 × 10 format with four replicates. Finally, each unit cell contains a round pneumatic membrane valve, named the button valve (BV), which protects or exposes a small circular area (Ø 100 μm) on the epoxy glass surface.11,14 The BV enables us to build up the surface pull-down assay in Figure 1A in sequential flush steps. In detail, we consecutively deposit biotinylated BSA (BB) and NeutrAvidin (NA) on the epoxy glass surface of all microchannels. The top reactive NA layer was then passivated with 12154



Article

chip thermodynamic binding behavior of the TFs (see Supporting Information Figure S2). Next, we sought to increase the multiplexing degree of the integrated metabolic sensor system. While in the first binding experiments single TFs were used, it was favorable to solve several TFs in one solution and measure the metabolites in parallel. Therefore, the TF had to be orthogonal. Although the three TFs are not reported to interact, a different ionic strength or pH, as in the cellular microenvironment, can influence the orthogonality and binding thermodynamics of the TFs. For this, the orthogonality of the metabolic binding reactions of the TFs was tested on-chip by systematically determining the crossreactivity of the metabolites, dsDNA binding sequences, and TFs. Here, the column elements of the microfluidic matrix were used to deposit the dsDNA with the TF binding sequences on the pull-down areas of the unit cells. More precisely, the columns were used to pattern four repeats for each TF binding sequence at three different molar surface ratios (XDNA). For a control one column with unspecific control DNA was included. In the following the individual and all combinatorial mixtures of TFs with different metabolites were introduced from the row elements of the matrix. The concentration of the TFs was left constant, whereas the concentration of the metabolites was set above the upper limit of the working range of the TF to obtain the maximal signal change compared to samples without metabolites. Figure 3 shows the result of the cross-correlation experiment between the FFs and metabolites. The correlation coefficient between the fluorescence signals obtained for the three TFs solved individually and within the composites was 0.99. In conclusion, the metabolic sensor system is orthogonal and can be used for the parallel detection of the three metabolites in one solution. In the last step, we demonstrate the applicability of the integrated metabolic sensor system by determining metabolite concentrations within a cell extract of rosettes from the model plant organism A. thaliana. Multiplexing was achieved by patterning the chip in the column direction with the three TF DNA binding sequences at different XDNA ratios (0.25, 0.5, and 0.75). For a control one column was patterned with nonspecific DNA. From the row direction of the chip the analytical samples containing the TF and unknown amount of the metabolite were introduced. Here, metabolites were extracted from 1 g FW of rosettes of 5 week old ecotype Col-0 A. thaliana plants (see Materials and Methods in the Supporting Information). Dried extracts were dissolved in dilution buffer and stepwise diluted by factors of f = 100, 50, 10, 8, 6, 5, 4, 3, and 2, which were performed outside of the microfluidic chip platform. The three TFs were added to each dilution. All samples were spiked with 220 μM pyruvate, 22 μM T6P, and 600 μM L-arginine. The spiked-in metabolite concentration calibrates the fluorescence signal of the metabolic sensor systems to the lower limit of the working range. Therefore, small changes to the TF−DNA equilibrium due to the metabolites in the cell extract at even lower dilutions are detectable without prior knowledge of the metabolite concentration in the sample. An additional calibration curve for the TFs was prepared by including seven samples with a constant TF concentration and defined metabolite concentrations, c, covering the whole working range of each TF. Figure 4, parts A and B, shows the fluorescence signal of the calibration samples exemplarily for T6P and pyruvate after binding to the dsDNA binding sequences. The calibration curve for the TFs is shown as solid black line in Figure 4, whereas the spiked-in metabolite

Figure 2. Binding isotherms of PdhR, TreR, and ArgR to the cognate DNA sequences depending on DNA and metabolite concentration. Each data set was recorded within one mLSI chip run. Solid lines are the fit results of a general heterotropic two-site binding model to the binding isotherms (see the Supporting Information for derivation). The corresponding dissociation constants of the TF are given in Table 1. The working ranges of the metabolic sensor systems are highlighted in green, whereas the dashed lines denote the limit of detection (LOD).

Table 1. Dissociation Constants and Working Ranges of the TFa sensor

K1/10−10 mol dm−2

K2/10−5 mol/L

working range/10−5 mol/L

PdhR TreR ArgR

2.23 ± 0.05 1.46 ± 0.03 5.00 ± 0.10

32.0 ± 4 1.5 ± 0.1 101.0 ± 7

16−206 0.6−25.9 22−233

a

All thermodynamic binding parameters are derived from data shown in Figure 2. K1 and K2 denote the dissociation constants of the TF to the cognate DNA and metabolite, respectively (see also the Supporting Information). (For the calculation of the working ranges, see main text.)

working range was defined by 15% of the maximum signal of the derivative. This value ensured that the metabolite concentration yielded a detectable fluorescence signal change within the binding equilibrium between the TF, dsDNA, and metabolite (see Table 1). Single binding isotherms were confirmed off-chip at lower resolution in well-plate experiments. All obtained results were in close agreement to the on12155



Article

Figure 3. Cross-reactivity test to enable multiplexing of the metabolic sensor system. (A) Bar charts show the fluorescence signal of individual solved TFs bound to DNA on the pull-down area in unit cells of the chip in absence and presence of its cognate metabolite. For each TF, three tests with different molar surface ratios of the DNA with the corresponding TF binding sequence, XDNA, were performed: 1, 2, and 3 on the X-axis denote for XDNA values of 0.25, 0.5, and 0.75, respectively. (B) Bar charts show the fluorescence signal of the mixed TFs bound to the DNA on the pull-down area in one unit cell in the presence of different metabolites. In samples with pyruvate, T6P, or L-arginine the concentration was 5, 1, and 5 mM, respectively. DNA binding of TFs in composites (right) resembled the binding of individual solved TF (left).

concentration is marked with a dotted line. A least-square fit of the fluorescence signals of the diluted samples IS(f) to the calibration curve IC(c) yielded concentrations values of 2 μmol/ g FW for pyruvate and 0.261 μmol/g FW for T6P. During the fit procedure, the known dilution factors, f, between the analytical samples were fixed. Experiments repeated on different chips and the same plant extract showed a correlation coefficient of 0.98 (see Supporting Information Figure S3). On the other hand L-arginine could not be detected. To exclude that the extraction method for the metabolites biased the metabolite quantification, we tested the three comment metabolite extraction protocols (see Supporting Information Table S5) from rosettes of A. thaliana plants grown under the same conditions. While the pyruvate and T6P concentrations were in agreement with each other, L-arginine could not be detected in any extraction sample. In fact, within the more concentrated cell extract samples the spiked-in Larginine concentration could not be detected. Only upon dilution of the cell extract the fluorescence signal of the ArgR sensor reached the expected spiked-in value (see Supporting Information Figure S4). This is an indication that the cell extract sample from the rosettes exhibits an inhibitory response on the ArgR sensor. Previous measured absolute values for T6P and pyruvate within rosettes of Arabidopsis Col-0 leafs were between 0.02−0.120 and 0.07−0.11 nmol/g FW, respectively.24,25 The determined absolute concentrations of pyruvate and T6P are higher than the reported literature values. Therefore, we confirmed the absolute concentration of pyruvate in the leaf extract with an enzymatic assay. The physiological concentration of L-arginine within the A. thaliana leaves was reported to be in the same order.26 Compounds of a structure similar to that of the analyte can inhibit or competitively bind to the sensor, which is an inherent disadvantage of all protein-based sensors. This is exemplified on TreR, which binds next to T6P also trehalose although with a clearly lower affinity (KD ∼ 0.3 mM).20 Therefore, the specificity of biosensors has to be tested. To do so we determined the diurnal cycle of T6P concentration within rosette of Arabidopsis plants with the TreR sensor on-chip. During the light and night phase T6P is accumulated and degraded in leaf cells, respectively, whereas the trehalose concentration stays constant.27 Figure 4C shows the T6P concentrations determined within leaf cells collected during different light/dark phases. In accordance with the previously

reported values we observed a relative increase of T6P concentration of 30% during the light compared to the dark phase. All values were determined within one chip run. Pyruvate concentrations were measured in parallel and showed the expected diurnal change.25

■

CONCLUSION Miniaturized chip assays for the quantification of DNA and proteins are currently revolutionizing analytical technologies. However, a multiplexed quantitative system for detecting metabolites is still lacking. We have presented an approach to functionalize natural proteins for the detection of small molecules on microfluidic chip platforms. In comparison to electronic, optical, or thermal sensors, the development of protein-based sensor systems is more complex. Therefore, the developmental phase of protein-based biosensors requires higher throughput platforms to characterize the sensors’ functions. With the presented mLSI platform, we established massive parallelization of metabolite−protein binding measurements, which allowed us to characterize and model the thermodynamic binding reactions of three TFs, their limit of detection, the working range of pyruvate, T6P, and L-arginine, and their orthogonality. Within the real-world application, two out of the three tested metabolic sensors were functional, i.e., the pyruvate and T6P sensor. In particular, the sugar molecule, T6P, is of growing interest due to its signaling function in diverse plants. Current detection methods of T6P from cell extracts are still limited, and thus, the presented TF-based system can be an attractive alternative. The L-arginine sensor exhibited a robust signal under the less complex test conditions but was inhibited within the cell extract from Arabidopsis rosettes, which contains more than 10K metabolites. The utilization of TF for metabolite detection is not limited to pyruvate, T6P, or L-arginine. At minimum, 40 other TFs with binding specificities to key metabolites from bacteria, plants, or humans exist and can be integrated using the same principles described here.28 The technical barrier of producing TFs is balanced by a low consumption rate of about 0.25 fmol TF per data point, which includes dead volumes of the microfluidic chip. The sensitivity of the sensor system for metabolites is dependent on the native binding constant of the TF and, thus, is less sensitive than a mass spectrometry assay. Nevertheless, the mode of multiplexing makes the chip platform competitive to other assays. On the current platform, 16 (samples) × 3 12156



Article

material is available free of charge via the Internet at http:// pubs.acs.org.

■

AUTHOR INFORMATION

Corresponding Authors

*E-mail: [email protected]. *E-mail: [email protected]. Author Contributions ∇

S.K. and D.H. contributed equally to this work.

Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS This study was supported by the Excellence Initiative of the German Federal and State Governments (EXC-294 and GSC4), the German Research Foundation (Emmy-Noether Grant ME3823/1-1), and the European Research Council under the European Community’s Seventh Framework Program (FP7/ 2007-2013)/ERC Grant agreement no. 259043-CompBioMat.

■

(1) Theodoridis, G. A.; Gika, H. G.; Want, E. J.; Wilson, I. D. Anal. Chim. Acta 2012, 711, 7−16. (2) Buescher, J. M.; Moco, S.; Sauer, U.; Zamboni, N. Anal. Chem. 2010, 82, 4403−4412. (3) Lisec, J.; Schauer, N.; Kopka, J.; Willmitzer, L.; Fernie, A. R. Nat. Protoc 2006, 1, 387−396. (4) Psychogios, N.; Hau, D. D.; Peng, J.; Guo, A. C.; Mandal, R.; Bouatra, S.; Sinelnikov, I.; Krishnamurthy, R.; Eisner, R.; Gautam, B.; Young, N.; Xia, J.; Knox, C.; Dong, E.; Huang, P.; Hollander, Z.; Pedersen, T. L.; Smith, S. R.; Bamforth, F.; Greiner, R.; McManus, B.; Newman, J. W.; Goodfriend, T.; Wishart, D. S. PLoS One 2011, 6, e16957. (5) Zurbriggen, M. D.; Moor, A.; Weber, W. J. Biotechnol. 2012, 160, 80−90. (6) Rapp, B. E.; Gruhl, F. J.; Länge, K. Anal. Bioanal. Chem. 2010, 398, 2403−2412. (7) Updike, S. J.; Hicks, G. P. Nature 1967, 214, 986−988. (8) Yoo, E.-H.; Lee, S.-Y. Sensors 2010, 10, 4558−4576. (9) Weber, C. C.; Link, N.; Fux, C.; Zisch, A. H.; Weber, W.; Fussenegger, M. Biotechnol. Bioeng. 2005, 89, 9−17. (10) Melin, J.; Quake, S. R. Annu. Rev. Biophys. Biomol. Struct. 2007, 36, 213−231. (11) Meier, M.; Sit, R. V.; Quake, S. R. Proc. Natl. Acad. Sci. U.S.A. 2013, 110, 477−482. (12) McDonald, J.; Whitesides, G. Acc. Chem. Res. 2002, 35, 491− 499. (13) Unger, M. A.; Chou, H. P.; Thorsen, T.; Scherer, A.; Quake, S. R. Science 2000, 288, 113−116. (14) Maerkl, S. J.; Quake, S. R. Science 2007, 315, 233−237. (15) Meier, M.; Sit, R.; Pan, W.; Quake, S. R. Anal. Chem. 2012, 84, 9572−9578. (16) Tholen, D. W.; Linnet, K.; Kondratovich, M.; Armbruster, D. A. Protocols for Determination of Limits of Detection and Limits of Quantitation; Approved Guideline; NCCLS Document EP17-A; NCCLS: Wayne, PA, 2004; Vol. 24. (17) Quail, M. A.; Haydon, D. J.; Guest, J. R. Mol. Microbiol. 1994, 12, 95−104. (18) Ogasawara, H.; Ishida, Y.; Yamada, K.; Yamamoto, K.; Ishihama, A. J. Bacteriol. 2007, 189, 5534−5541. (19) Horlacher, R. J. Biol. Chem. 1997, 272, 13026−13032. (20) Hars, U.; Horlacher, R.; Boos, W.; Welte, W.; Diederichs, K. Protein Sci. 1998, 7, 2511−2521. (21) Tian, G. L.; Lim, D. B.; Oppenheim, J. D.; Maas, W. K. J. Mol. Biol. 1994, 235, 221−230. (22) Schaumburg, C. S.; Tan, M. J. Bacteriol. 2006, 188, 919−927.

Figure 4. Determination of the pyruvate and T6P concentration within an extract of cells from A. thaliana rosettes. Fluorescent signal of DNA bound (A) TreR and (B) PdhR in the calibration (○) and dilutions of the analytical sample (▲) plotted against the metabolite concentrations of the calibration samples, respectively. T6P and pyruvate concentrations of 22 and 220 μmol/L, respectively, were spiked into all analytical samples. The T6P concentrations for the calibration curve were c = 1, 10, 22, 44, 110, 220, 440 μM. For the pyruvate calibration curve a 10-fold higher concentration series was used. (C) Diurnal changes of the T6P and pyruvate concentration within rosettes from 5 week old A. thaliana plants measured within one chip experiment. Yellow and gray areas in the plot denote for the light and dark phase, respectively.

(TF) metabolites can be determined within approximately 1.5 h excluding chip preparation. Future integration of the sample dilution workflow onto the mLSI chip can further improve the utility of the technology. The combined biochemical and microfluidic approach demonstrates promise in the field of applied metabolic engineering science by increasing the speed and throughput of analytics, as, for example, for sugar analysis or signaling molecules in plant organisms.

■

REFERENCES

ASSOCIATED CONTENT

S Supporting Information *

Additional material and methods, the TF binding model, results of the control experiments, and oligonucleotide sequences. This 12157



Article

(23) Wyman, J.; Gill, S. J. Binding and Linkage: Functional Chemistry of Biological Molecules. Mill Valley; University Science Books: Mill Valley, CA, 1990. (24) Lunn, J.; Feil, R.; Hendriks, J.; Gibon, Y.; Morcuende, R. Biochem. J. 2006, 1, 139−148. (25) Voll, L.; Hausler, R. E.; Hecker, R.; Weber, A.; Weissenbock, G.; Fiene, G.; Waffenschmidt, S.; Flugge, U.-I. Plant J. 2003, 36, 301−317. (26) Shi, H.; Ye, T.; Chen, F.; Cheng, Z.; Wang, Y.; Yang, P.; Zhang, Y.; Chan, Z. J. Exp. Bot. 2013, 64, 1367−1379. (27) Carillo, P.; Feil, R.; Gibon, Y.; Satoh-Nagasawa, N.; Jackson, D.; Sing, O. E. B.; Stitt, M.; Lunn, J. E. Plant Methods 2013, 9, 1−1. (28) Ramos, J. L.; Martínez-Bueno, M.; Molina-Henares, A. J.; Terán, W.; Watanabe, K.; Zhang, X.; Gallegos, M. T.; Brennan, R.; Tobes, R. Microbiol. Mol. Biol. Rev. 2005, 69, 326−356.

■

NOTE ADDED AFTER ASAP PUBLICATION This paper was originally published ASAP on December 5, 2014, with an error in the values for the previously measured absolute values for T6P and pyruvate within rosettes of Arabidopsis Col-0 leafs. The corrected version was reposted on December 16, 2014.

12158


Transcription Factor Sensor System for Parallel ... - ACS Publications

Recommend Documents