How to Measure Molecular Forces in Cells: A Guide to Evaluating Genetically-Encoded FRET-Based Tension Sensors
- 3.8k Downloads
The ability of cells to sense and respond to mechanical forces is central to a wide range of biological processes and plays an important role in numerous pathologies. The molecular mechanisms underlying cellular mechanotransduction, however, have remained largely elusive because suitable methods to investigate subcellular force propagation were missing. Here, we review recent advances in the development of biosensors that allow molecular force measurements. We describe the underlying principle of currently available techniques and propose a strategy to systematically evaluate new Förster resonance energy transfer (FRET)-based biosensors.
KeywordsMechanobiology Förster resonance energy transfer Biosensors Mechanotransduction Fluorescence lifetime imaging microscopy
Cells are exposed to a wide range of mechanical forces. Endothelial cells, for instance, are subject to high shear stress in arteries but low forces in veinous or lymphatic vessels,14 cardiomyocytes bear the rhythmic contractions of the heart,27 keratinocytes are stimulated by shear or tension in the skin,57 and chondrocytes sense forces from cartilage compression.51 Interestingly, cells respond to such mechanical stimuli—that may vary over orders of magnitude—with astonishing specificity suggesting that cell type-specific mechanisms exist, which convey fine-tuned mechanoresponses. Indeed, a range of subcellular structures mediating different aspects of mechanotransduction has been identified including mechanosensitive ion-channels,16 the plasma membrane,18 the cytoskeleton,26 the nucleus,24 and cell-adhesion complexes.21 Techniques such as traction force microscopy67 have greatly contributed to our understanding of force transduction across these subcellular structures.3 Yet, how forces propagate on the molecular level is still largely unknown.
Molecular Force Transduction Occurs in the Piconewton Range
The notion that mechanotransduction—the translation of mechanical information into a biochemical response—may occur at similar forces was supported by the observation that conformational changes in the adhesion protein talin can be induced by mechanical tension as low as 2 pN.15,77 Likewise, cleavage of the von Willebrand factor is facilitated by force-induced protein unfolding at 5 pN80 and collagen proteolysis is increased 100-fold upon application of 10 pN force.1 Interestingly, some receptor-ligand pairs form adhesive interactions, called catch bonds, which strengthen under pN forces.72 The linkage between P-selectin and monomeric P-selectin glycoprotein ligand-1, for instance, is characterized by catch bond behavior below 11 pN,42,71 and α5β1 integrin shows enhanced binding to fibronectin type III repeats at 10–30 pN.33 As the unfolding of whole protein domains usually requires higher forces—the green fluorescent protein (GFP) starts to unfold at about 35 pN17 and immunoglobulin (Ig) or fibronectin type III domains unfold at 80–300 pN50,58,59—it appears that important aspects of mechanotransduction do indeed occur in the low pN range (Fig. 1). But how can such low forces be measured in cells?
Current Approaches for Measuring Molecular Tension
Genetically-encoded tension sensors
Application by original group
Iwai et al. 29
Meng et al. 46
Meng et al. 45
1–6 pN; by single-molecule spectroscopy
Grashoff et al. 23
Conway et al. 12
Kuriyama et al. 35
Leerberg et al. 39
Borghi et al. 5
Chang et al. 9
Cai et al. 6
Krieg et al. 34
5–7 pN; by DNA springs
Meng and Sachs43
Rahimzadeh et al. 56
Verma et al. 74
Ichimura et al. 28
5–7 pN; by DNA springs
Meng and Sachs44
Synthetic tension sensors
Application by original group
PEG n a
Theoretical; WLC modelc
Stabley et al. 66
Jurchencko et al. 31
PEG n a
Theoretical; WLC modelc
Liu et al. 41
Grashoff et al. 23
Morimatsu et al. 47
Single molecule AFM
Wang and Ha76
A Brief Introduction to Förster Resonance Energy Transfer (FRET)
FRET Measurements in Cells
The FLIM approach is insensitive to fluorophore concentration and experimental settings but nevertheless requires rigorous controls and careful data analysis.75 Other imaging methods include acceptor photobleaching73 or anisotropy measurements,40 each with its own advantages and disadvantages. In general, life-cell FRET experiments are complicated by cellular auto-fluorescence, undesired photobleaching and the fact that fluorophore properties depend on environmental factors such as pH, ion concentration or temperature.52,61 Thus, an in-depth understanding of the limitations inherent to the different FRET analysis methods is essential.30,53,69,78
Genetically-Encoded Tension Sensor Modules for Measuring Intracellular Molecular Forces
Most of the existing tension sensor modules are based upon the initial observation that elastic molecules such as single-stranded DNA (ssDNA) can act as pN force sensors when inserted between two fluorescent dyes undergoing efficient FRET.65 Since the FRET efficiency inversely correlates with the D–A distance (Eq. 1), forces that extend the linker and thereby increase chromophore separation strongly reduce FRET (Fig. 2a). Therefore, the selection of an appropriate elastic element is critical and the following requirements have to be satisfied. First, the linker has to be short because currently available FRET pairs are characterized by a Förster distance R 0 ≈ 5–6 nm, at which the FRET efficiency is most sensitive to changes in fluorophore separation distance (Fig. 2a).53,69 Second, the increase in linker length has to be sufficiently large so that applied tension translates into measurable FRET efficiency differences. Finally, data interpretation is greatly facilitated if the linker follows a simple folding/unfolding pathway and quickly returns to its original conformation when forces subside.
Following these principles, a number of FRET-based tension-sensitive modules have been developed (Table 1; Fig. 2). The linker elements range from a comparably stiff α-helix46 and spectrin repeat43 to the elastic spider silk flagelliform peptide23 (Fig. 2a). An alternative approach was recently tested, in which the force-sensitive element does not change its length but rather conformation (Fig. 2b).44 In addition to the FRET-based approaches, a circularly permuted (cp) YFP has been generated that loses fluorescence under force28 (Fig. 2c). Similarly, proximity imaging (PRIM) has been used to correlate molecular strain with changes in the emission spectrum of an engineered GFP-dimer29 (Fig. 2d). Whether all these techniques will be useful for further applications in cells, however, remains to be determined.
Synthetic Tension Sensing Techniques for Measuring Forces at the Cell Surface
Measuring mechanical forces at the cell surface does not require genetic encoding of the tension sensing element but can be performed using mechanically well-described polymers. In addition, organic dyes can be employed which are more photostable than most genetically encoded fluorophores and rarely affect the functionality of the labeled molecules. Together with the versatile surface chemistry technologies that are available, these tools have enabled the development of highly sensitive methods to determine extracellular molecular forces (Table 2). For example, the molecular tension–based fluorescence microscopy (MTFM) approach uses polyethylene glycol (PEG) as a force-sensitive tether molecule to measure mechanical tension across growth factor66 and cell adhesion receptors31,41 (Fig. 2e). Similarly, functionalization of the flagelliform peptide23 with organic dyes and arginine–glycine–glutamine (RGD)-ligands allows the estimation of force across single integrin receptors47 (Fig. 2a). An addition to these synthetic sensors is the tension gauge tether (TGT) approach, where immobilized double-stranded DNA (dsDNA) is functionalized with cell surface receptor ligands so that force above a well-defined threshold can be easily detected76 (Fig. 2f).
Applications of FRET-Based Molecular Tension Sensors
The genetic tension sensor modules described above have been applied to a range of proteins (Fig. 2; Table 1) in different cell types and even whole organisms such as C. elegans 34 or D. melanogaster. 6 The targeted molecules include actin-binding proteins such as α-actinin,43, 44, 45, 46,74 filamin,45,46 and spectrin34,43,46 as well as cell adhesion molecules like cadherin,5,6,12 PECAM-1,12 and vinculin.23 These measurements confirmed the long held assumption that many cytoskeletal proteins bear pN forces and are an ideal starting point for a more detailed analysis. The use of a vinculin tension sensor, for instance, revealed an average force of about 2.5 pN across vinculin. More interestingly, however, the vinculin transduced tension strongly depends on the cell adhesion state, with highest tension occurring in assembling focal adhesions but low forces in disassembling complexes23; this indicates that vinculin stabilizes cell adhesions under mechanical force. In another study, a β-spectrin tension sensor revealed constitutive tension of about 1.5 pN across this cytoskeletal adaptor protein. Interestingly, genetic manipulations decreasing β-spectrin pre-stress correlate with impaired touch sensation suggesting that cytoskeletal pre-tension is critical for efficient mechanosensation in neurons.34 These examples illustrate that the true power of FRET-based tension sensors lies not only in the force measurement itself but also in the possibility to unravel molecular mechanisms that are currently inaccessible to other techniques.
A Guide to Evaluating Genetically-Encoded FRET-Based Tension Sensors
A detailed understanding of a tension sensor’s biophysical properties is crucial. In which force range is the tension sensor module applicable? How does the linker unfold in response to force, and how large is the dynamic FRET range? These kind of questions need to be answered before meaningful experiments can be performed. Furthermore, effects of tension sensor module integration into the protein of interest (POI) need to be carefully evaluated and the FRET experiments must be properly controlled. While every novel genetically-encoded biosensor will require its specific evaluation strategy, we propose here a series of experimental controls which, in our opinion, are indispensable for any FRET-based tension sensor characterization.
Tension Sensor Design: Which Forces are to be Measured?
Before the experiment, a number of obvious (but not trivial) questions should be addressed. Which molecule should be targeted, what are the expected mechanical forces and do tension sensor modules that are sensitive to these forces exist? It is important to note that molecular tension sensors are unsuitable to measure forces across subcellular structures in general (in fact, this is precisely what they do not do), but specifically report tension across the POI. Our previously published vinculin tension sensor, for instance, can be efficiently used to determine vinculin tension but is unsuitable to measure focal adhesion forces in general.23 So, it is also worth asking: Are we interested in forces across distinct proteins or across whole subcellular structures?
Once a target protein has been identified, it is necessary to carefully evaluate whether the tension sensor module can be inserted into the POI without significantly affecting its function. We find that structural information is often helpful to identify possible insertion sites, which are preferably unstructured and flexible. In case of the vinculin tension sensor, for example, the chosen integration site is located in a flexible linker region between two well-defined structural domains and vinculin function is preserved after tension sensor module integration.10,23 If little structure information is available for the POI, we recommend testing several integrations in parallel.
Characterizing the Tension Sensor Module: What is the Sensor’s Force Sensitivity?
Biosensor Characterization I: Are the Fluorophores Functional After Integration into the POI?
Next to the in vitro calibration of the tension sensor module, its functionality after integration into the POI needs to be validated. Steric constraints, for instance, could impair fluorophore folding. Furthermore, forces of about 35 pN17 are sufficient to partially unfold GFP-like proteins, which might influence fluorescence.61 Therefore, we recommend comparing the properties of individual donor (D) and acceptor (A) fluorophores terminally fused to the target protein (X) (Fig. 3a, X-DC or X-AC) with fluorophores that have been integrated into the POI (Fig. 3a, X-DI, X-AI). Alternatively, integrated tension sensor modules harboring one non-fluorescent mutant fluorophore (Fig. 3a, X-mTSI(D), X-mTSI(A)) may be used for a comparison. Fluorescence lifetime as well as absorption or emission spectra are useful parameters to determine whether properties of internally placed fluorophores are affected (Fig. 3c).
Biosensor Characterization II: Is the POI Functional After Tension Sensor Module Integration?
A critical step in the development of a genetically-encoded biosensor is the insertion of the tension sensor module into the POI; quite obviously, this involves the risk of altering the target protein’s function. Therefore, a detailed evaluation of the biosensor is critical and requires the generation of genetic control constructs (Fig. 3a), for which protocols have been described before.2 To evaluate the biosensor’s biological functionality, these constructs should be expressed in cells depleted of the endogenous protein, which has several advantages (Fig. 3d). First, overexpression artifacts can be avoided by adjusting biosensor expression to physiological levels. Second, it can be easily tested whether the biosensor is able to functionally replace the endogenous protein. The β-spectrin tension sensor, for example, rescues the paralysis phenotype of spectrin mutant C. elegans to wild type behavior34 and an E-cadherin tension sensor was shown to rescue the migration defect in E-cadherin–depleted border cells in D. melanogaster. 6 Finally, force measurements are likely to be more accurate as the total amount of force distributes only across biosensor molecules.
A typical evaluation experiment includes the reconstitution of knockout (or knockdown) cells with the tension sensor construct (Fig. 3a, X-TSI) and the N- or C-terminally tagged POI (Fig. 3a, X-DC or X-AC). This is followed by confirmation of proper subcellular localization using fluorescence microscopy methods as well as the evaluation of expression levels by western blotting. Depending on the POI, functionality may be further tested by fluorescence recovery after photobleaching (FRAP) analysis, where the subcellular dynamics of X-TSI and X-DC can be easily compared (Fig. 3d).
Controlling the FRET Experiment: Are Effects of Intermolecular FRET or conformation changes significant?
As described above, FRET experiments are complex because energy transfer does not only depend on the chromophore separation distance and orientation but also on the biophysical properties of the individual fluorophores (Eqs. 2 and 3).53 Therefore, FRET-based tension sensor experiments need to be carefully controlled.
To ensure that differences in FRET are caused by mechanical tension across the biosensor and are not a result of changes in the microenvironment (such as pH, temperature, etc.), we emphasize the need to use a zero-force control, which can be easily generated by fusing the tension sensor module to either end of the POI (Fig. 3a, X-TS0). This control should show identical subcellular localization as the biosensor (X-TSI) but should not display changes in FRET as no significant tension can be applied across the module. The second possibly confounding factor in a tension sensor FRET experiment is energy transfer between adjacent molecules (so-called intermolecular FRET) that can significantly contribute to the overall FRET in compact subcellular structures such as focal adhesions or cell–cell contacts. Intermolecular FRET can be easily estimated using a pair of control constructs in which either the individual fluorophores23 (Fig. 3a, X-DI, X-AI) or tension sensor modules with one non-fluorescent mutant fluorophore12 (Fig. 3a, X-mTSI(D), X-mTSI(A)) are integrated into the POI. Co-expression of such constructs in one cell and subsequent FRET measurement in the relevant subcellular structure allow calculation of intermolecular FRET. Furthermore, potential effects of protein conformation changes on FRET need to be considered. As this strongly depends on the molecule of interest, however, these control experiments are not generalizable. Nevertheless, conformation controls should be included to ensure that changes in FRET are reflective of differences in mechanical tension and not κ 2 artifacts. Finally, the notion that FRET changes actually reflect changes in tension may be reinforced by experiments in which external forces are rapidly applied using mechanical stretch5 or fluid shear flow.12
Data Analysis and Interpretation: What Do the FRET Efficiency Differences Mean?
At the end, proper data analysis is critical. We highly recommend the use of quantitative techniques such as fluorescence lifetime imaging microscopy (FLIM) allowing the calculation of FRET efficiencies instead of FRET ratios (Fig. 3e). Moreover, automated data analysis software to determine transfer rates in subcellular compartments greatly facilitates data interpretation. The evaluation experiments described above should be followed by additional controls that will depend on the individual context (such as inhibition of intracellular contractility or application of external forces). Together, this experimental strategy will allow a straightforward evaluation of new tension-sensitive biosensors. We wish to emphasize that insufficiently characterized tension sensors should not be utilized by the scientific community as their application all too often results in misleading interpretations and confusion.
While the development of molecular tension sensors has already made significant contributions to a deeper understanding of force transduction, further improvements will be necessary to further elucidate molecular mechanisms. For instance, more calibrated tension sensor modules are required to evaluate distinct force ranges. Also, probes with increased dynamic range, quantum yield and photostability would be useful to perform intracellular single-molecule measurements that unravel the heterogeneity and dynamics of molecular processes. In this context, the development of orthogonal labeling techniques using genetically encoded proteins,13 peptides, and non-natural amino acids37 is promising, as they allow site-specific labeling of intracellular proteins with organic dyes. Another approach that seems worth pursuing is genomic integration of biosensors into the locus of the target proteins by the recently developed CRISPR/Cas9 technique.62 This strategy should ensure physiological expression levels of a biosensor and avoid the time-consuming generation of knockout (or knockdown) cell lines. Finally, other approaches to determine forces in cells could be combined with FRET-based tension sensors. These may include optical tweezers methods that can be applied to individual molecules within cells,49 the specific functionalization of micro-droplets, which were recently used to determine mechanical forces on the cellular level in embryonic tissue,7 or traction force microscopy techniques allowing the simultaneous measurement of traction forces and molecular forces in cell adhesions. In fact, while this manuscript was under revision, two new, synthetic tension sensor techniques that use hairpin-DNA as force sensitive linkers and fluorescence quenching as read-out were published,4,79 which allow molecular traction force microscopy.
In summary, properly characterized molecular tension sensors provide a powerful tool to gain insight into cellular mechanotransduction. Further improvements that will allow experiments at single-molecule resolution within cells, the application of tension sensors to a wider range of proteins, and the combination of biosensors with other quantitative techniques may pave the way to a better understanding of how cells sense and respond to their mechanical environment.
The authors thank Andrea Freikamp for help with the figures. C. Grashoff is supported by the German Research Council (DFG) through the Emmy Noether program grant (EN GR3399-2-1), the Collaborative Research Center SFB63 (B9), and a Paul Gerson Unna Research Group of the Max Planck Society (MPG).
Conflict of interest
A.-L. Cost, P. Ringer, A. Chrostek-Grashoff, and C. Grashoff declare that they have no conflicts of interest.
No human or animal studies were carried out by the authors for this article.
- 4.Blakely, B. L., C. E. Dumelin, B. Trappmann, L. M. McGregor, C. K. Choi, P. C. Anthony, V. K. Duesterberg, B. M. Baker, S. M. Block, D. R. Liu, and C. S. Chen. A DNA-based molecular probe for optically reporting cellular traction forces. Nat. Methods 11:1229–1232, 2014Google Scholar
- 5.Borghi, N., M. Sorokina, O. G. Shcherbakova, W. I. Weis, B. L. Pruitt, W. J. Nelson, and A. R. Dunn. E-cadherin is under constitutive actomyosin-generated tension that is increased at cell–cell contacts upon externally applied stretch. Proc. Natl. Acad. Sci. USA 109:12568–12573, 2012.CrossRefGoogle Scholar
- 14.Davies, P. F. Flow-mediated endothelial mechanotransduction. Physiol. Rev. 75:519–560, 1995.Google Scholar
- 40.Lidke, D. S., P. Nagy, B. G. Barisas, R. Heintzmann, J. N. Post, K. A. Lidke, A. H. A. Clayton, D. J. Arndt-Jovin, and T. M. Jovin. Imaging molecular interactions in cells by dynamic and static fluorescence aniosotropy (rFLIM and emFRET). Biochem. Soc. Trans. 31:1020–1027, 2003.CrossRefGoogle Scholar
- 77.Yao, M., B. T. Goult, H. Chen, P. Cong, M. P. Sheetz, and J. Yan. Mechanical activation of vinculin binding to talin locks talin in an unfolded conformation. Sci. Rep. 4:4610, 2014.Google Scholar
- 79.Zhang, Y., C. Ge, C. Zhu, and K. Salaita. DNA-based digital tension probes reveal integrin forces during early cell adhesion. Nat. Commun. 5, 2014.Google Scholar
Open AccessThis article is distributed under the terms of the Creative Commons Attribution License which permits any use, distribution, and reproduction in any medium, provided the original author(s) and the source are credited.