If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
Department of Biomedical Engineering, The Pennsylvania State University, University Park, PA 16802 USADepartment of Mechanical Engineering and Department of Surgery, The Pennsylvania State University, University Park, PA 16802 USA
Transcription factors are essential regulators of various physiological and pathological processes. However, detecting transcription factor-DNA binding activities is often time-consuming and labor-intensive. Homogeneous biosensors that are compatible with mix-and-measure protocols have the potential to simplify the workflow for therapeutic screening and disease diagnostics. In this study, we apply a combined computational-experimental approach to investigate the design of a sticky-end probe biosensor, where the transcription factor-DNA complex stabilizes the fluorescence resonance energy transfer signal of the donor-acceptor pair. We design a sticky-end biosensor for the SOX9 transcription factor based on the consensus sequence and characterize its sensing performance. A systems biology model is also developed to investigate the reaction kinetics and optimize the operating conditions. Taken together, our study provides a conceptual framework for the design and optimization of sticky-end probe biosensors for homogeneous detection of transcription factor-DNA binding activity.
Transcription factors initiate the transcription of genes and play a pivotal role in regulating various biomedical processes, such as stem cell differentiation and tissue regeneration. The self-renewal and differentiation capability of stem cells can be controlled by modulating transcription factor activities. The importance of transcription factor activities in stem cell differentiation is highlighted by the discovery that the induction of MYC, OCT3/4, SOX2, and KLF4 can convert somatic cells into induced pluripotent stem cells [
], at which transcription factors act as key biomarkers for characterizing the success of cell reprogramming. For example, the SOX9 transcription factor, with a known binding motif of AGAACAATGG [
]. Rapid detection of transcription factor activities is required to characterize and optimize the efficiency of stem cell therapies.
Conventional assays for detecting transcription factor-DNA binding, such as electrophoretic mobility shift assays, DNA footprinting assays, and Western blotting, are often cumbersome and not compatible with high throughput screening equipment [
] have been described. These assays allow separation-free detection of transcription factor binding activities with double-stranded DNA and are suitable for laboratory automation applications (e.g., inhibitor screening). However, the design of these homogeneous biosensors can be challenging due to the complex molecular design and a large number of operating parameters of the biosensors.
This study investigated the sticky-end probe biosensor design for detecting transcription factor-DNA binding (Fig. 1A). The primary objective was to establish a conceptual framework for understanding the performance of sticky-end probe biosensors in detecting transcription factor activities [
]. As a model system, we designed a homogeneous SOX9 biosensor, which was chosen due to its role as a differentiation regulator in enteroendocrine and corneal epithelial stem cells and its medium sized consensus sequence (10 base pairs) that provided a general example of uniform overhangs in the sensor design [
]. The biosensor employed the fluorescence resonance energy transfer (FRET) principle. We calibrated the sensor's sensitivity and tested its specificity against a negative control transcription factor, RUNX2. Finally, we developed a computational model in the MATLAB SimBiology platform to investigate the influences of model parameters and optimize the operating parameters. We optimized the acceptor-to-donor ratio to enhance the biosensor's performance. Our results emphasize the critical parameters in designing sticky-end probe biosensors for detecting transcription factor-DNA binding activities.
Fig. 1(A) Schematic of the sticky-end probe biosensor for homogenous detection of transcription factor-DNA binding activity. (B) Transcription factor binding stabilizes the energy transfer between the FRET pair. (C) Upon binding the transcription factor, the FRET signal indicated by the ratio between the emission peaks of 525 and 560 increases. Data represent mean ± s.d. (n = 3).
Probes were designed as shown in Table 1 and synthesized by Integrated DNA Technologies (San Diego, CA). The binding motif for Sox9 was obtained using multiple databases, such as JASPAR [
]. The biosensor consisted of a pair of double-stranded DNA probes with 5-base long complementary overhangs (sticky-ends). The fluorescein and CY3 fluorophores were chosen as donor and acceptor, respectively, and were incorporated as internal dT modifications. All probes were purified according to the manufacturer's recommendation. In the design, the fluorophores were separated by one base. The probe pairs were prepared individually by annealing the sequences at 94°C for 5 minutes and gradually cooling down until they reached room temperature. Varying ratios between probes were prepared, and a 1:6 donor-to-acceptor ratio resulted in similar emission peaks (i.e., approximately 100 RFU starting values).
Table 1Sticky-end probe design for SOX9
TF
Probe Type
Sequence & Fluorophore
Length
SOX9
Donor
5′-AACAA/iFluorT/GGTTTGCTGGATTCAG-3’
21
Donor complimentary
5′-/CTGAATCCAGCAAACC/-3′
16
Acceptor sequence
5′-A/iCy3/TGTTCTATTGCTACGTT-3′
18
Acceptor complimentary
5′-/AACGTAGCAATAG/-3′
13
iFluorT and iCy3 represent internal fluorescein dt and Cy3 dT respectively.
Calibration experiments were run by serial dilution of SOX9 protein (Abcam). Negative control was kept with no protein to determine the background level. RUNX2 was applied to test the specificity of the biosensor (Abcam). The mixtures were kept in an incubator at 37°C for 30 minutes. Then, 30 μL of the mixture was pipetted into a 384-well glass bottom plate (Cellvis) and measured by a FlexStation® 3 multi-mode microplate reader. The donor was excited at 488 nm. The FRET signal was determined by the ratio of emission at 560 nm and 525 nm.
SimBiology modeling
The kinetic model was implemented in the MATLAB SimBiology environment (MathWorks SimBiology Version 6.3). The model was solved by ode15s, which is designed for stiff models. The simulation was performed for a duration of 60 minutes, which was sufficient for the binding reaction to reach equilibrium. Initial conditions were chosen based on the experimental settings. The rate constants were estimated by the experimental data. The parameters in the base model are shown in Table 2. Notably, the parameters in the base model were chosen to study the general trend of the data and were not intended to represent the biosensor quantitatively. The goal of the qualitative model was to examine the characteristics of the binding reactions and improve understanding of the key parameters. Sensitivity analyzes were performed to evaluate the sensitivity of model parameters.
Table 2Numerical values used in the base model for analyzing the characteristics of the sticky-end probe biosensor.
Experiments were generally performed in triplicates on multiple days. Data are presented as mean ± standard derivation. For the calibration curves, the ANOVA test was performed to determine the statistical significance of the data. p <0.05 was considered significant. The curve fitting function in Excel was applied to determine the best-fit curve and R-square value. The limit of detection was defined as the background intensity plus two standard derivations.
], the binding motif of the consensus DNA sequence is separated into two probes with sticky ends (Fig. 1A). The probes are labeled with either a donor or acceptor. The binding of the overhangs brings the donor-acceptor pair in proximity. The presence of the transcription factor stabilizes the binding of the donor and acceptor probes and enhances the FRET signal. In this study, we designed a biosensor based on the SOX9 consensus sequence with the fluorescein and Cy3 FRET pair. Fig. 1B shows the emission spectra of the biosensor with and without SOX9 protein. The emission peaks of fluorescein and Cy3 at 525 nm (Em1) and 560 nm (Em2) were measured. Ratiometric measurement (i.e., Em2/Em1) was performed as the output FRET signal of the biosensor. The Em2/Em1 value increased upon mixing with the transcription factor and typically reached equilibrium in approximately 30 minutes (Fig. 1C).
Performance of the SOX9 biosensor
The sensitivity of the SOX9 biosensor was characterized by performing a serial dilution experiment (Fig. 2A). We normalized the data by subtracting the background noise (i.e., signal at zero target concentration). The FRET signal generally increased (p = 0.0030) with the concentration of SOX9 protein in the semi-log plot and is most sensitive in the nM range. A calibration curve (R-square value 0.9631) was generated based on the data (black dotted line in Fig. 2A). We also estimated the limit of detection of the biosensor by the background plus two standard deviations (red dotted line in Fig. 2A). The intersection of the lines indicated a limit of detection of approximately 3 nM, which is compatible with other homogeneous transcription factor sensors. Specificity of the SOX9 sensor was tested using RUNX2 protein as a negative control (Fig. 2B). Results yielded no significant signal output (ANOVA p = 0.7423) with the non-specific RUNX2 protein, indicating that the sensor signal was associated with SOX9 binding.
Fig. 2(A) Calibration of the sticky-end probe sensor for SOX9 detection. (B) Specificity of the SOX9 sensor against RUNX2. Data are normalized to the background with zero targets (green lines). Data represent mean ± s.d. (n = 3). Black dotted lines represent curve fitting by regression analysis. Red dotted lines indicate the limit of detection (background + 2 S.D.).
To understand the performance of the biosensor, we developed a computational model in the MATLAB SimBiology environment. The SimBiology graphical user interface implements ordinary differential equations that capture the binding reactions (Fig. 3A). In particular, the first reaction involves the reversible binding of the donor probe (DP) and acceptor probe (AP) to form the DPAP complex. The DPAP complex can further bind to the target transcription factor (TP) to form the TFDPAP complex. At the onset of a typical reaction, the donor probe and acceptor probe rapidly hybridized with each other to form the DPAP and TFDPAP complexes (Fig. 3B). Increasing the concentration of the target transcription factor resulted in a general decrease in the donor probe and DPAP complex and an increase in the TFDPAP complex (Fig. 3C). To compare with the FRET signal, we calculated the normalized signal of (TFDPAP+DPAP)/DP, which approximately represented the EM2/EM1 ratio. The normalized signal increased with the concentration of the target transcription factor, similar to the experiment trend (Fig. 3D). Direct comparison between the experimental and computational data revealed non-linearity of the data, suggesting a calibration curve should be obtained if quantitative data are required.
Fig. 3(A) SimBiology representation of the sticky-end probe binding reactions. The reversible binding of the donor probe (DP) and acceptor probe (AP) creates the DPAP complex, which further binds with the transcription factor to create the TFDPAP complex. (B) Representative time traces of the major reactants and products. (C) Analysis of the effect of the transcription factor concentration on the reaction. (D) Comparison of the normalized signal between experiment measurement and computational simulation.
We further explored the influences of the binding rate constants (Fig. 4). The formation of DPAP and TFDPAP complexes was plotted to evaluate the sensor performance. The changes of TF, DP, and AP were not shown and could be estimated based on the values of DPAP and TFDPAP complexes by considering the mass conservation. We first examined the effects of the forward and reverse binding rate constants of the reversible binding reaction of DP and AP (Fig. 4A-B). The reaction kinetic was most sensitive to the forward rate constant (Kf_S), which determined the overall reaction time, suggesting the initial binding event could represent a rate-limiting step of the sensing process. In contrast, the reverse rate constant (Kr_S) had a strong effect on the formation of the complexes. A large reverse rate constant significantly reduced the formation of both DPAP and TFDPAP complexes. Furthermore, the transcription factor reaction rate constants (i.e., Kf_TF and Kr_TF) directly influenced the formation of the TFDPAP complex. Increasing Kf_TF and decreasing Kr_TF had similar effects on the reaction kinetics and equilibrium values. In the base model, the reaction kinetics was limited by the DP-AP binding reaction. The influence of transcription factor binding could be described by an equilibrium constant in the computational model.
Fig. 4(A-D) Sensitivity analysis of the influence of the reaction constants on the binding kinetics. The reaction rate constants were increased or decreased 10-fold from the based model, while all other parameters remained constant. Only DPAP and TPDPAP complexes were shown as DP and TP could be determined by mass conservation.
The SimBiology model allowed us to investigate the influence of the operating parameters, such as the acceptor-to-donor ratio (AP:DP). In principle, a large AP:DP enhances the chance of forming the DPAP complex for detecting the target transcription factor. However, a large AP:DP also creates a substantial background signal, which reduces the overall signal-to-noise ratio. A large amount of unbound acceptor probe could also increase the noise associated with fluorescence bleed-through in FRET measurement [
]. To optimize the AP:DP ratio, we experimentally measured the FRET signal with and without the target transcription factor (Fig. 5A). The amount of DP remained constant at 12 nM while the amount of AP increased from 6 to 60 nM (an AP:DP ratio from 0.5 to 5). The ratio of EM2/EM1 values with and without the target was considered the signal-to-noise ratio of biosensor. The maximum signal-to-noise ratio was obtained at AP:DP around 2.5 in the experiment.
Fig. 5(A-D) Effect of the acceptor-to-donor ratio (AP:DP) on the signal-to-noise ratio. The concentrations of donor probe and target protein (SOX9) were 12 nM and 150 nM, respectively. The concentration of the acceptor probe was adjusted from 6 to 60 nM to generate a donor-to-acceptor ratio between 0.5 to 5. The value was normalized by the signal without target (i.e., 0 nM) to obtain the signal-to-noise ratio. (C-D) Traces of DPAP and TPDPAP complexes with (150 nM) and without target transcription factor.
Computationally, we also observed a dependency of the signal-to-noise ratio on the AP:DP ratio (Fig. 5B). While the model did not predict the absolute value of the signal-to-noise ratio (suggesting other noise contributors exist), the model was able to qualitatively capture the trend of the signal-to-noise ratio by considering the increase in background signal and fluorescence bleed-through. The data revealed several factors contributing to the signal-to-noise ratio (Fig. 5C-D). Without a target, the amount of DPAP increased with the concentration of AP, which contributed to the amount of background signal. With a target, the background level and the signal increased due to the formation of DPAP and TFDPAP. Notably, a high concentration of AP increased not only the signal due to the formation of DPAP and TFDPAP complexes but also the amount of unbound AP, which increased the noise level due to fluorescence bleed-through. These data support that the signal-to-noise ratio of the sticky-end probe biosensor can be optimized by the AP:DP ratio. Our data also provided a qualitative explanation of the experimental data and provided insights into designing the sticky-end probe biosensors.
Discussion
This study reports a combined experimental and computational analysis of the sticky-end probe biosensors [
]. Compared to conventional biosensing methods, the homogeneous biosensor allows the mix-and-measure format, which is attractive for laboratory automation and screening applications [
], the design is relatively simple as long as the binding motif is available and does not require additional competitive binding reactions. A FRET-based design, which allows “signal-on” sensing (i.e., the signal increases with the target), was investigated. A SOX9 biosensor was demonstrated for detecting the SOX9-DNA binding activity, which could potentially be useful for studying stem cell differentiation [
] in the future, as it could be used to assess the transcription factor activities. The biosensor was also shown to be selective against RUNX2. A computational model was developed to assist in the interpretation of the results. Limitations of the study include that only a SOX9 sensor was designed as a model system. Similarly, the limit of detection, although appropriate for the creation of the computational model requires further optimization for applying this sensor to lowly expressed transcription factors. The computational model was qualitative and did not fully represent all aspects of the biosensor, such as the binding rates and noise contributors.
Our results unscored several challenges in the design and application of the sticky-end probe biosensor for measuring transcription factor-DNA binding activity and provided helpful insight into the design of the sticky-end probe biosensor. Firstly, the biosensor is based on the equilibrium binding principle, which does not necessarily create a linear response. The ratiometric measurement also introduces additional non-linearity. A calibration curve should be generated if a linear response is required for the intended application [
]. Secondly, the signal-on FRET scheme, which is easier for result interpretation, can be complicated by fluorescence bleed-through. Careful fluorophore selection and bleed-through correction are warranted to minimize the measurement artifact [
]. Our results also highlighted the importance of optimizing the AP:DP ratio. Our initial base experiment was performed at a 6:1 AP:DP ratio to obtain a similar value of EM2 and EM1. While spectral and ratiometric imaging is often operated at a similar intensity level, our data suggest that a 2.5:1 AP:DP was optimal for the experiment, suggesting the signal-to-noise ratio is contributed by multiple factors. These issues should be considered in the future design of sticky-end probe biosensors for detecting transcription factor-DNA binding activities.
We envision applications of the homogeneous biosensor in live cell sensing and high-throughput screening in the future. The washing-free binding scheme and mix-and-measure nature of the sticky-end biosensor can potentially be applied in live cells for monitoring stem cell differentiation and cell reprogramming [
]. To achieve intracellular transcription factor sensing, additional biosensors should be designed for detecting important transfection factors. These probes should be transfected into live cells. Future investigations will be required to optimize the transfection protocol and enhance the stability of the biosensors in live cells.
Declaration of Competing Interest
The authors declare the following financial interests/personal relationships which may be considered as potential competing interests:
Pak Kin Wong reports financial support was provided by NSF.
Acknowledgment
This work was supported by the National Science Foundation RECODE program (2033673).
Reference
Takahashi K.
Yamanaka S.
Induction of pluripotent stem cells from mouse embryonic and adult fibroblast cultures by defined factors.
00033-X&id=ga1.jpg){kind=link}
00033-X&id=gr1.jpg){kind=link}
00033-X&id=gr2.jpg){kind=link}
00033-X&id=gr3.jpg){kind=link}
00033-X&id=gr4.jpg){kind=link}
00033-X&id=gr5.jpg){kind=link}