A FRET based mechanical stress sensor for specific proteins in situ

Abstract

To measure mechanical stress in real time we designed a fluorescence energy transfer (FRET) cassette, noted stFRET, which could be inserted into structural protein hosts. The probe was made of a green fluorescence protein (GFP) pair, Cerulean and Venus, linked with a stable α-helix. We measured the FRET efficiency of the free cassette protein as function of the length of the linker, the angles of the fluorophores, temperature and urea denaturation and protease treatment. The linking helix was stable to 80C°, unfolded in 8M urea, and was rapidly digested by proteases, but in all cases the fluorophores were unaffected. We modified the α-helix linker by adding and subtracting residues to vary the angles and distance between the donor and acceptor, and assuming the cassette was a rigid body we calculated its geometry. We tested the strain sensitivity of stFRET by linking both ends to a rubber sheet subjected to equibiaxial stretch. FRET decreased proportionally to the substrate strain. The naked cassette expressed well in human embryonic kidney (HEK-293) cells and surprisingly was concentrated in the nucleus. However, when the cassette was located into host proteins such alpha-actinin, non-erythrocyte spectrin and filamin A, the labeled hosts expressed well and distributed normally in cell lines such as 3T3 where they were stressed at the leading edge of migrating cells and relaxed at the trailing edge. When COL-19 was labeled near its middle with stFRET, it expressed well in C. elegans, distributing similarly to hosts labeled with a terminal GFP and the worms behaved normally.

Mechanical stress is one of the most influential physical factors in biology and one of the least characterized. While it is obvious from molecular dynamics [1–4] and force spectroscopy [5–12] that forces deform molecules, the mechanics of cells is much more complicated, involving the interaction of heterogeneous polymers and membranes and their interaction with both two dimensional heterogeneous liquid membranes [13, 14] and three dimensional cytoplasmic solutions where signaling factors can vary in time and space [15–17]. Mechanical interactions at the levels of cells, organs and organisms are responsible for such familiar physiology as motor function, hearing [18], touch [19] and the regulation of blood pressure [20], but the interactions are also deeply embedded in the biochemistry of the cell affecting such varied process as the phenotype of stem cells [21], DNA transcription [22, 23],, translation of cellular components by motor proteins such as kinesin [24], stress induced changes of structure as occurs in shear stress modulation of the cytoskeleton of the endothelia [25, 26] and more general interactions due to the physical chemistry of concentrated protein solutions[27]. To dissect which stresses affect which functions, we need labels that are sensitive to mechanical stress and that can be attached to specific proteins.

To meet that need we designed a cassette (noted stFRET) that can be inserted into structural proteins and reports molecular strain via changes in FRET, and with appropriate calibration, molecular stress. The cassette consists of the GFP monomers Cerulean and Venus [28–32] linked by a stable α-helix [33]. This paper characterizes the properties of the probes, shows that they can be efficiently incorporated into structural proteins such as collagen-19, non-erythrocyte spectrin, alpha actinin and filamin A within living cells, and that the FRET from this cassette changes with stress in situ.

The efficiency of energy transfer for a FRET pair is E ∝1/(1 + (R/R_O)⁶) where R is the distance between the dipoles and R_O is the characteristic distance for 50% energy transfer [34]. The maximal sensitivity for changes in R occurs at R = R_O. For Venus and Cerulean R_O is ~5nm [35] so we linked them with a 5nm α-helix. The efficiency is affected by the angle between the transition dipoles as well as the distance between them, and estimated the probe geometry by varying the number of residues in the linker. Removing one residue caused a large change in angle with a small change in distance and adding or removing a full turn produced a change in distance with no change in angle. We used six mutants to solve for the three relevant angles of the dipoles assuming the cassette was rigid. stFRET was stable over temperature and mild urea denaturing conditions, but with 8M urea the linker unfolded and the fluorophores remained stable. Thus stFRET is robust.

stFRET expressed well in various biological systems including 3T3 and HEK-293 cells and in C. Elegans. After insertion into a variety of structural host proteins such as collagen, filamin, actinin and spectrin it distributed in the same manner as the same hosts with terminal GFP tags. stFRET changed FRET with the spontaneous movement of motile cells, decreasing efficiency in regions under tension and increasing in regions expected to be free of significant stress. By axially stretching C. Elegans we could demonstrate acute reversible changes in FRET associated with tension and relaxation. stFRET opens the door to studying in real time many physiological processes that are modulated or driven by mechanical stress.

Results

General configuration and FRET spectra of stFRET and its variants

Figure 2B shows the alignment of the DNA sequence of the linker with five modified versions (the predicted geometrical changes are in Table 1). As shown in Table 1, according to the general property of alpha-helices one amino acid residue deletion produces a change (−100°) in angle with negligible change (−1.5 Å) in length. A five residue deletion of the helix rotates the structure 360° but shrinks the helix 2.7nm. Deletion or addition of two and half turns of the helix twists the structure −180° or +180° and decreases or increases the length 1.35nm. Figure 2C gives the amino acid sequence and the segments of the helix linker we modified. Deletion of 18 amino acid residues eliminates five turns of the helix and a 9 amino acid deletion eliminates two and half turns. Figure 2A shows the general configuration of six stFRET variants. Inward arrows show the excitation wavelength and outward arrows show the emission wavelength. The width of arrows denotes light intensity.

Figure 2.

Construction of stFRET protein and five variants. A. Schematic structure of stFRET. Cyan is the donor, Cerulean; yellow is the acceptor, Venus. The height of the β-can structure is 4.2nm. The black helix is the linker and it nominal length is 5.0nm. Incoming arrows indicate the excitation and outgoing ones the emission with wavelength marked adjacent and the width of the arrows is proportional to the light intensity. B. Alignment of the primary and modified linker DNA sequences. C. Modifications to the linker with DNA and amino acid sequences.

Table 1.

Changes in stFRET geometry caused by adding and deleting amino acids. Values with positive symbols indicate an increasing amount and negative symbol indicate a decreasing amount.

Number of AA added or subtracted	Change in length of linker (nm)	Chang in angle of linker (radians)
−1	−0.15	5π/9
−2	−0.3	10π/9
+9	+1.35	π
−9	−1.35	π
−18	−2.7	0

Figure 3A shows the emission spectrum of stFRET with excitation at 433nm. There are peaks at 475 and 527nm: with the 475 nm emission from the donor Cerulean and 527nm from the acceptor Venus with robust energy transfer. A 100μM solution of unlinked donor and acceptor (1:1 mixture, green filled squares and line with 433nm excitation) had a small emission at 527nm due to the bleed-through from Cerulean, the donor (Blue filled reversed triangle and line) and some direct excitation of the acceptor Venus by 433nm (Black triangle and line). The donor and acceptor mixture had E = 0 and D/Aratio = 2.47 ± 0.05 (Figure 3, B). However, for stFRET E = 44 ± 2.5% and D/Aratio = 0.47 ± 0.02, showing efficient energy transfer (for E and D/ARatio calculation see Methods).

Figure 3.

FRET efficiency and D/A ratio (mean ± SD). A. Spectra of stFRET, Cerulean and Venus monomers and Cerulean and Venus in 1:1 mixture. B. FRET efficiency and D/A Ratio of stFRET with Cerulean and Venus in 1:1 mixture. Data were taken with protein from three separate purifications. CV is Cerulean and Venus in a 1:1 mixture. Excitation 433nm, emission 460nm–550nm.

Calibration of three angles and κ²

Confident in the origin of stFRET energy transfer, we purified the other five variants and measured their fluorescence (Figure 4A). All mutants exhibited robust FRET (Table 2). stFRET itself had 44 ± 2.5% energy transfer and the 5T construct had the highest efficiency E = 56 ± 4.5%, the 2.5T construct increased E to 47 ± 2.1% while the 2.5I decreased E to 37 ± 0.9%. FT1AA and FT2AA, presumably only having their angles changed, decreased E to 29 ± 7.1% and 38 ± 4.3% (Figure 4B). Table 2 summarizes the apparent change of angles and distances obtained by modifying the linker and the corresponding energy transfer efficiency.

Figure 4.

Modification of the linker changes FRET efficiency of six constructs. A. fluorescence spectra of stFRET and its five variants (scan parameters as Figure 3). B. FRET efficiency of the six constructs. C. SDS-PAGE gel of purified the proteins, and Cerulean and Venus monomers. FT stands for stFRET; 5T and 2.5T are constructs with 5 or 2.5 turns deletion from the linker; 2.5I is the construct with a 2.5 turn insert; FT1AA, FT2AA are the constructs with one amino acid or two amino acid deletions. All values are means ± SD and the data were taken with proteins from three separate purifications.

Table 2.

Values of parameters in equation (6) of six stFRET variants as described in Figure 4.

Protein Constructs	Energy Transfer Efficiency (E)	r (linker length, nm)	Z (H/2 of β-Can, nm)	θA (Unknown parameter 1)	θD (Unknow n parameter 2)	Φ (Unknown parameter 3)
stFRET	44 ± 2.5%	5.0	2.1	θA	θD	Φ
5T	56 ± 4.5%	5.0–2.7 = 2.3				Φ
2.5T	47 ± 2.1%	5.0–1.35 = 3.65	No Change	No Change	No Change	Φ+π
2.5I	37 ± 0.9%	5.0+1.35 = 6.35				Φ+π
FT1AA	29 ± 7.1%	5.0–0.15 = 4.85				Φ+5π/9
FT2AA	38 ± 4.3%	5.0–0.3 = 4.7				Φ+10π/9

If we assume that a single residue alters the linker length by a translation of 0.15nm and 100 degrees, and that the structure is rigid, we can use the data in Table 2 to solve for the probe geometry (see Experimental procedures). The numerical solutions gave θ_A = 3.83,θ_D = −0.78 and Φ = 1.97 radians, and κ² = 0.86, which is 30% higher than 2/3, the κ² value one would obtain assuming of random rotation of the donor and acceptor (Figure 1). However, we wish to point out that a value of κ² ~2/3 does not necessarily imply the probes are moving randomly.

Figure 1.

Geometry of stFRET. D and A are donor and acceptor dipole vectors, r is the length of linker. The three angles ( θ_A, θ_D, Φ) are the unknown parameters. R_A–D is the distance between acceptor and donor chromophores.

Stability of the linker as perturbed by urea, temperature and proteinase K

We did a number of tests to assess linker integrity. If the linker were an α-helix, then melting will increase the end-end spacing and the efficiency will decrease. With urea as a denaturant [36, 37], Figure 5A shows that the efficiency of stFRET declined with concentration up to 8M and the previously quenched donor emission recovered. Remarkably, the fluorophore spectra were almost unaffected by urea with <10–15% change in amplitude (Figure 5 C and D). Figure 5B shows that 1 to 8M urea caused the D/ARatio to increase from 0.46 to 1.21 as expected if the helix unfolded into a random coil allowing the donor and acceptor to move further apart and reducing energy transfer (Figure 5E).

Figure 5.

Melting the linker. A. Spectra from stFRET treated with 1 to 8M urea (scan parameters as Fig. 3B). B. D/ARatio of stFRET after different concentration urea treatments (means ± SD, n = 3 in each treatment), increasing D/A ratio indicates the recovery of donor emission and decrease of energy transfer. C. Cerulean monomer fluorescence with urea treatments (scan parameters as per Fig 3). D. Venus monomer fluorescence with urea (excitation at 515nm and scan 520–600nm). E. Urea melts the linker and leaves the donor and acceptor intact decreasing FRET energy transfer as donor emission recovering and D/ARatio increasing (definitions as per Figure 2 A).

As a second test of the helix stability we tried to melt stFRET at elevated temperatures but the protein proved stable up to 80 C°. Figure 6A shows the temperature dependence of fluorescence of 100μM stFRET protein excited at 433nm from room temperature to 80C°. Donor and acceptor emission both declined somewhat as the temperature increased probably due to a direct change in quantum efficiency but there was no significant change in transfer efficiency from 60C° to 80C°, the upper limit of our measurements so that the linker structure can be considered quite robust.

Figure 6.

Alpha-helix linker in stFRET is resistant to temperature melting. A. stFRET spectra under 60C° 2min, 60C° 5min, 70C° 5min and 80C° 5min temperature treatments (scan parameters as per Fig 3). B. stFRET D/A emission ratio after different temperature treatments (means ± SD, n = 3 in each treatment). Deg means degrees Celsius and roomtem means room temperature.

As a final test of linker integrity we digested stFRET with proteases that cut the linker but left the fluorophores intact. Figure 7 shows that proteinase K led to a rapid fall in efficiency that was complete within 1min. The D/ARatio changed from 0.42 to 1.95 over 30min (Figure 7B) compared a change from 0.46 to only 1.21 when the protein was treated with 8M urea (Figure 5B). Similar behavior was found for all six constructs (data not shown). The donor and acceptor fluorophore spectra were unaffected by proteinase K after 30min digestion (Figure 7 C and D). Figure 5E and 7E are diagrammatic models summarizing the energy transfer between donor and acceptor under different treatments (the width of the arrows represents signal intensity).

Figure 7.

Two Units Proteinase K (1 Unit/μl) digests the linker but not Cerulean or Venus. Spectra of stFRET protein (A), Cerulean (C), and Venus (D) digested for 20 seconds, 1, 2, 3, 5, 10, 15 and 30 minutes at room temperature with 200μl 100μM protein. B. Time course of D/ARatio for proteinase K digestion of stFRET. E. Proteinase K cleaved the linker and eliminated FRET in stFRET protein (PK refers to proteinase K, S, seconds and M, minute. n = 3).

In Vitro measurement of strain sensitivity

To verify the strain responsiveness we bonded the ends of derivatized stFRET to a silicone rubber sheet using StreptagII-Streptactin™ and stretched the sheet equibiaxially on the fluorescence microscope. When the C- and N-terminal ends of stFRET were derivatized so that it would be stretched with the sheet there was a reversible ~11% decrease in the D/A Ratio (Figure 8). As a control we measured FRET from stFRET that was derivatized at one end only so that it was simply immobilized but not stretched and there was no significant change in FRET with strain (Figure 8). Non-specific binding of double tagged stFRET to an untreated silicone surfaces also produced no significant change in FRET with strain. Thus, stFRET is sensitive to strain as expected from the solution assays and the design of the probe.

Figure 8.

Double Streptag II tagged stFRET shows a decrease in FRET ratio when stretched on silicone rubber disks. Single and double Streptag II tagged stFRET were allowed to bind to either untreated or Streptactin modified silicone disks. FRET ratio was monitored in 10 spots on each disk during application of the suction stimulus shown. Only the disks with Streptactin treated surfaces and stFRET proteins having Streptag tags at both the C- and N- termini showed a significant change in FRET ratio when stretched.

Eukaryotic expression and targeting property of stFRET

Before inserting stFRET into host proteins, we placed the gene under a eukaryotic promoter (Human cytomegalovirus) and transiently transfected HEK cells with stFRET alone. Control transfections with Venus or Cerulean monomers showed no preferential localization and obviously no energy transfer (Figure 9 A, B, C, D, E, F). Cells transfected with stFRET displayed significant energy transfer (Figure 9I). stFRET localized to the nucleus with an extremely high density in the nucleoli (Figure 9K).

Figure 9.

stFRET expressed in HEK-293 cells exhibits efficient FRET. Confocal reference image of Cerulean taken from the CFP channel (A) and the DIC channel (B) with the overlap in (C). Reference image of Venus from the YFP (D) and DIC channels (E) and the overlap in (F). Images of stFRET using the CFP channel (G), YFP channel (H), FRET channel (I) and the DIC channel (J) with the overlap of these four channels in image (K). FRET index was calibrated pixel by pixel using Xia’s method (L) [44]. Hollow black regions were excluded from the calculation because of intensity saturation. stFRET is localized in the nucleus and especially concentrated in the nucleoli (arrow heads).

Nuclear targeting proteins have a consensus amino acid sequence of Lysine/Arginine (K/R)_4–6 or smaller clusters separated by 10–12 amino acids: (K/R)₂X_10–12(K/R)3 [38]. The linker has multiple arginine clusters similar to nuclear targeting sequence, but simply removing one or the other fluorophores from stFRET produced a uniform cytoplasmic distribution showing that the linker’s sequence alone was not sufficient for targeting. These unexpected nuclear targeting properties of stFRET may provide a useful tool for understanding nuclear protein transport.

Host proteins of stFRET with normal expression showing stress sensitivity

We have inserted stFRET into various host proteins including collagen-19 (Figure 10G), non-erythrocyte spectrin (Figure 10E), filamin A (Figure 10C), alpha-actinin (Figure 10A), and expression systems including HEK-293, 3T3, and C. Elegans, and the insertion locations were optimized to obtain protein distributions similar to those observed for the host protein C-terminal tagged by GFP or Cerulean (Figure 10B, D, F and H). Inserting stFRET into host proteins eliminated nuclear targeting. The fluorescence of stFRET in cultured cells was located in the cytoplasm and/or the cell membrane depending upon the host (Figure 10A, C and E). We have expressed the construct of the most abundant collagen in C. Elegans, COL-19, and the protein was properly assembled showing the typical striated pattern and the worms behaved normally. When we stretched the worm with micromanipulators, the labeled COL-19 showed a decrease in FRET efficiency with stretch, and in convex regions as they actively wiggled (Figure 10G and H).

Figure 10.

Normal expression of stFRET in various host proteins. Alpha-actinin-stFRET (A), alpha-actinin-GFP (B), filamin A-stFRET (C), filamin A-CFP (D), spectrin-stFRET (E) and spectrin-CFP (F) in 3T3 fibroblast cells; Collagen-19-stFRET (G) and collagen-19-GFP (H) in C. elegans (with assistance of Dr. R. Gronostajski ); Arrow heads indicate the striated expression pattern and central line in the worm cuticle.

Figure 11 indicates stFRET integrated into actinin and filamin can sense tension in situ. Migrating 3T3 cells have a characteristic leading and lagging edge and Figure 11A, B and C shows the donor, acceptor and FRET images from three confocal microscopy channels. stFRET was distributed evenly across the cytoplasm as visualized with a 16 color pseudo-color map (Figure 11C,D). Transfection with Actinin-stFRET revealed that during migration, the lagging edge showed higher energy transfer than the leading edge (Figure 11E and F), i.e., it was relaxed. We measured the efficiency of various domains in the lagging and leading edges from fourteen confocal image stacks. The lagging edges (the red outline domain) nearly doubled the FRET efficiency compared to the leading edge (blue and green outline domains). Multiple cells had the same behavior but because of the complexity of the various shapes it was difficult to arrive at any useful statistic for frequency. We have shown a typical cell with different domains as an internal control. The same phenomena was observed in Filamin-stFRET transfected 3T3 cells (Figure 11G, H, I, J, K and L). Figure 11G, H and I are three confocal image channels, and Figure 11J is the pseudo-color image of stFRET protein distribution. Figure 11K is the FRET efficiency image in which three domains were selected. The efficiency in the red outlined domain is twice as high as that in the blue and green domains (Figure 11L). These data suggest tension in both actinin and filamin is lower in domains close to the lagging edge (where adhesion to the substrate is released), and higher at the leading edge where adhesions pulling the cell forward.

Figure 11.

stFRET senses the strain change in actinin and filamin. Actinin-stFRET transfected 3T3 fibroblast confocal images were taken in three channels: CFP donor channel (A), FRET channel (B), YFP acceptor channel (C). Actinin-stFRET protein expression levels were displayed by applying image J lookup table (LUT) 16-color color map to the YFP acceptor channel image. Arrows show the leading edge, lagging edge and the cell domains with missing filopodia (D). FRET efficiency was calculated by E = nF/(nF + I_D) in which nF is the net FRET from FRET channel and I_D is donor intensity from donor channel. E value was showed by image J lookup table (LUT) 16-color color map (E). Three cell domains were selected for statistical analysis of E and fourteen confocal stacks of each domain were measured and analyzed (F). Histogram bars were assigned same colors as the related domains in (E). Filamin-stFRET confocal images (G, H, I, J, K and L). Three scan channels (G, H and I) are as actinin-stFRET images. Arrangements and statistics of filamin-stFRET images (J, K and L) are as in actinin-stFRET images.

Discussion

Designed to be an in situ stress sensor, stFRET has a robust and predictable energy transfer both in vitro and in vivo. We were able to explore the geometry of stFRET by perturbing the linker length and terminal angles using the known properties of alpha helices. FRET efficiency changed in a predictable manner with the postulated geometry suggesting that the fluorophores are not free to rotate. A recent molecular dynamics simulation study of FRET in lysozyme found that κ and R_O could be correlated by as much as 0.8, so that FRET measurements that assume random rotational freedom are likely to be in error [39]. The ability to change angle and distance by varying the linker can be used in vivo to examine the effect of host proteins on probe geometry. Regardless of the coupling of the fluorophores to the linker, all of the host proteins we studied were coiled-coiled dimers or trimers so that the fluorophores of stFRET would not be able to rotate freely.

Figure 2 A shows the predicted mean structure of free stFRET. The three unknown angles of equation (6) were solved using data for the six mutants using the least squares equation solver in MAPLE. The solutions were stable to perturbations of the starting values suggesting that we were measuring a constrained system. Our final solution: θ_A = 3.83, θ_D = −0.78, Φ = 1.97 yielding κ² =0.86. There will be bending and flexing motions of the structure in solution, but we obtained consistent answers from the overdetermined set of equations suggesting that the calculated mean values are at least self consistent. The geometric values we have calculated would represent mean values weighted by the efficiency. Fluctuations that bring the dipoles closer are more heavily weighted than those that move them further away although the probability of occupancy of these conformations is another weighting factor. A detailed MD simulation would be useful, but that is not essential to the use of stFRET as a probe of molecular stress since the most important variables are the differences in efficiency, i.e. the gradients of stress.

The robust nature of stFRET was clear from the melting experiments. stFRET was thermally stable up to at least 80°C with the FRET efficiency virtually unchanged. Melting the linker with urea (Figure 5) [40] left the fluorophores untouched (Figure 5 C, D), but decreased the energy transfer consistent with unfolding of the linker (Figure 5 B, E). Two models have been proposed for urea-induced protein denaturation: the binding model, in which the denaturant binds weakly but specifically to sites exposed by the unfolded proteins [41], and a solvent exchange model in which the interaction of the solvent and the denaturant is a one-for-one substitution reaction at particular sites [42]. stFRET might serve as a useful probe to examine these alternatives.

The sensitivity of stFRET to protease cleavage has both positive and negative implications. If proteases are accidentally present in situ, they could cleave stFRET and provide misleading results. We saw no evidence of protease activity in HEK or 3T3cells or C. Elegans. However, the presence of intracellular proteases has been associated with acute pancreatitis, proposed to arise from trypsin over-activation in large endocytotic vacuoles of acinar cells [43]. Thus to study pancreatitis, stFRET may be a useful probe (cf., Figure 7).

Figure 8. Double Streptag II tagged stFRET shows a decrease in D/A Ratio when stretched on silicone rubber disks. stFRET immobilized with a single Streptag II tag did not respond to strain, nor did double tagged stFRET bound to a non-derivatized rubber disk. The D/A Ratio was monitored at 10 spots on each disk during equibiaxial strain and the results averaged. Only the disks with Streptactin treated surfaces and stFRET proteins having Streptag tags at both the C- and N- termini showed a significant change in FRET ratio when stretched.

Having established the basic physical properties of stFRET, we expressed it in HEK cells (Figure 9) and evaluated the energy transfer by Xia’s method [44] using confocal microscopy. The surprising localization of stFRET to the nucleus was proved not a result of the linker possessing a consensus nuclear targeting sequence since deletion of either fluorophore from the construct destroyed localization. This adaptability suggests stFRET can serve as a useful probe of nuclear targeting.

Knowing how native stFRET itself distributes, we incorporated it into host proteins including non-erythrocyte spectrin, filamin A, alpha-actinin (Figs 10A, C and E) in 3T3 cells and the collagen COL-19 in C. Elegans (Fig 10G). The distribution of the probe depended upon where the cassette was placed within the host. When inserted toward the middle, the fluorescent distribution appeared similar to that of the host protein tagged by GFP or CFP at the C terminus (Figure 10B, D, F and H). Insertion the cassette towards the termini of the host led to different spatial distributions. There is no gold standard for the proper localization of proteins in cells since fixation and exposure to various tracer ligands can produce changes in structure, but to first order the stFRET probes placed in the middle of the hosts appeared to constitute a minimal perturbation.

Under physiological conditions FRET efficiency varied in different regions of the cells (Figure 11) and these seemed to be correlated with the anticipated distribution of stress. Efficiency should be reduced when the host is under tension. Actinin-stFRET and filamin-stFRET generally showed lower efficiency than free stFRET (Figure 3B) suggesting that those proteins were normally under tension (Figure 11E, F, K and L, greenline and blueline domains). However, as cells migrate the stress in the leading and trailing edges changes. Connections to the extracellular matrix in the lagging edge must be disengaged and the connections at the leading edge put under tension. Figure 11E, F, K and L show increased FRET efficiency in the lagging edge as the filopodia were released from the substrate and tension decreased (redline domains), and decreased efficiency associated with increased tension in the leading edge as the cell is pulled forward.

To turn stFRET from a strain sensor into a stress sensor, we need to measure its force-distance properties. At the current time we only have estimates from published AFM data on the stretching the coiled coil myosin II [45]. Schweiger et al obtained a three phase force distance relationship: a linear phase of ~1mN/m, a plateau of ~25pN, and a wormlike chain phase as the helices were stretched closer to the contour length. The presence of a force plateau implies that if monomeric stFRET we subject of to force >10–25pN it would unfold in an all-or-none manner for about 3nm producing a large drop in FRET. We don’t see that, probably in part because the in situ probes are not homomers, but are coiled coils where the stress is shared with labeled and unlabelled neighbors. It may be possible to knockdown the background hosts to at least create homogenous labeled hosts. In addition, stress is shared between different proteins within the cell and at the current time we are only probing one of those components.

stFRET can be applied to any biological system with large covalently bonded proteins. It is possible to examine the role of stress in selected proteins within cells or even within free ranging organisms. With organ targeting in small organisms such as C. Elegans and zebrafish, it should be possible to develop high contrast video images of specific parts of the organism during controlled or natural behavior. We look forward to finding out how mechanical stress is coupled to biochemistry and to cell biology.

Experimental procedures

Gene constructing and protein purification

pEYFP-C1 Venus and pECFP-C1 Cerulean plasmids are generous gifts from Dr. David W Piston [46]. Cerulean gene was sub-cloned from pECFP-C1 by primers 5′-GCAGGTGTGAATTCCATGGTGAGCAAGGGCGAGGAGC-3′ and 5′-CCAGATCGCGGCCGCCTTGTACAGCTCGTCATGCCGAGAG-3′; EcoRI and ApaI restriction enzyme sites were introduced into 5 prime and 3 prime of Cerulean DNA fragment. This DNA fragments was inserted into multiple cloning site of pEYFP-C1 Venus by EcoRI and ApaI digestion then ligation. The resulting vector has Venus followed closely by Cerulean and between them there are two restriction enzyme sites BglII and EcoRI which then were employed to insert the alpha-helix linker. The alpha-helix linker DNA, 5′-GGCCTGCGCAAGCGCTTACGAAAATTTAGAAACAAGATTAAAGAAAAGCTTAAAAAAATT GGTCAGAAAATCCAGGGTTTCGTGCCGAAACTTGCAGGTGT-3′, was synthesized by Operon (Huntsville, Alabama, USA) then amplified by PCR and BglII and EcoR I sites were introduced into 5 prime and 3 prime ends. The final construct with alpha-helix connecting Venus and Cerulean was named stFRET and ready for eukaryotic expression. In order to purify the protein, stFRET gene was sub-cloned into prokaryotic expression vector PinPoint Xa-3 (Promega, Madison, WI, USA) using BamHI and NotI restriction sites which were introduced into FRET DNA fragment by using the following primers: 5′-GCTTCAGCTGGGATCCGGTGGTATGGTGAGCAAGG-3′; 5′-CCAGATCGCGGCCGCTTAGTGGTGATGATGGTGGTGATGATGCTTGTACAGCTCGTCC-3′. Following 8-histidine tag TAA stop codon was inserted in front of NotI site to make sure that his-tag located in C-terminal and well-exposed to solution. By modifying the linker in PinPoint-stFRET constructs we created five other constructs and named them based on the modification. They are: 5T with 5 turns of peptide chain truncated off the alpha-helix, 2.5T with 2.5 turns truncated off, 2.5I with 2.5 turns of the linker duplicated and inserted back into alpha-helix, FT1AA and FT2AA with one and two amino acid residues deleted from the linker. The primers used for PCR are as following: 5T sense primer 5′-GCGCAAGCGCTTACGAAAATTCGTGCCGAAACTTGCA-3′, anti-sense primer 5′-TTTTCGTAAGCGCTTGCGCTGCAAGTTTCGGCACGAA-3′; 2.5T sense primer 5′-GCGCAAGCGCTTACGACTTAAAAAAATTGGTCAGAAAATCCAGG-3′, anti-sense primer 5′-CCTGGATTTTCTGACCAATTTTTTTAAGTCGTAAGCGCTTGCGC-3′, 2.5I sense primer 5′-GAAACAAGATTAAAGAAAAGAAAATTTAGAAACAAGATTAAAGAAAAGCTTAAAAAAAT TGGTCAGAAAATC-3′, 2.5I anti-sense primer 5′-GATTTTCTGACCAATTTTTTTAAGCTTTTCTTTAATCTTGTTTCTAAATTTTCTTTTCTTTAAT CTTGTTTC3; FT1AA sense primer 5′-GATTAAAGAAAAGCTTAAAATTGGTCAGAAAATCC-3′, FT1AA anti-sense primer 5′-GGATTTTCTGACCAATTTTAAGCTTTTCTTTAATC-3′; FT2AA sense primer 5′-CAAGATTAAAGAAAAGCTTATTGGTCAGAAAATCC-3′, FT2AA anti-sense primer 5′-GGATTTTCTGACCAATAAGCTTTTCTTTAATCTTG-3′. All the insertions and deletions were performed with site-directed mutagenesis kit from Stratagene (La Jolla, CA, USA). As host proteins for stFRET, collagen-19 gene were subcloned into Pinpoint Xa-3 vector and filamin A, alpha actinin and non-erythrocytic spectrin gene were subcloned into pEYFP-C1 vector in which YFP gene was deleted. Different sites in these host proteins were tested to maximally remain their function after integrating stFRET into them. All constructs were confirmed by sequencing data from Roswell-Park Cancer institute (Buffalo, NY, USA).

Plasmid DNA of six proteins constructs and Venus Cerulean monomers were transformed into E. Coli cells (BL21(DE3pLacI) from Novagen, Gibbstown, NJ, USA) for expression. Proteins were purified as described [46]. 500ml Luria-Bertani (LB) broth containing 50mg/ml ampicillin were inoculated with 5ml over night cell culture from a single colony of each construct. Cells were cultured under 37°C and 250 r.p.m orbital shaking until optical density value reaches 0.6. 1mM final concentration isopropyl-β-D-thi-galactopyranoside (Sigma, St. Louis, MO, USA) was applied to the culture to induce protein expression and temperature was adjusted to 30°C for overnight expression. The cells were harvested by centrifuging at 4000g, 10 minutes under 4°C. The pellets were stored under −20°C for later use or proceeded directly to lysis step. 5ml bugbuster protein extraction reagent (Novagen) containing 25 units/ml Benzonase (Novagen), 1,000 units/ml rLysozyme(Sigma), 1mM phenylmethylsulfonyl fluoride (PMSF), 10μg/ml pepstatin and 20 μg/ml leupeptin was used for protein extraction from each gram cell pellet. Cells were sat in room temperature 30 minutes for the lysis. Soluble proteins were separated by centrifuging 30 minutes at 10,000g, 4°C. Ni-NTA his-tag elution buffer (250mM imidazole, 300mM NaCl, 50mM Na2HPO4, 0.2% Tween-20, PH8.0) was added to the protein solution to make the imidazole final concentration 20mM. 1ml Ni-NTA His.bind slurry (Novagen) was used per 4ml clear lysate and gently mixed by shaking under 4°C for 60minutes. The solution was loaded on a column and washed with 10 bed volumes of washing buffer (20mM imidazole, 300mM NaCl, 50mM Na2HPO4, 0.2% Tween-20, PH8.0) by gravity flow. Proteins retained on the column were washed of by elution buffer. Protein concentration was determined by BCA protein kit (Pierce, Rockford, IL, USA) and measured by ND-1000 spectrophotometer (Nanodrop, Wilmington, DE, USA). SDS-PAGE gel analysis was used to check the protein purity. Proteins that have >95% purity would be proceeded for further essay, otherwise proceeded to dialyze proteins into Tris-HCl buffer (10mM Tis-HCl, 1mMDTT, 50mM NaCl, 0.2% Tween-20, PH7.4) and then repeat Ni-NTA his-tag purification procedure to make the purity reach 95%. All proteins purified were finally exchanged into 10mM Tris-HCl buffer by Spectra/Pro Dispodyalyzer (Spectrum, molecule weight cut off 10,000) for further spectroscopy measurements.

Cell culture and transfection

Human embryonic kidney (HEK 293) and 3T3 fibroblast cells were cultured in Dulbecco’s Modified Eagle Medium (Gibco) supplemented with 10% fetal bovine serum and antibiotics. Cells were spread on 3.5mm coverslips and allowed to grow 24h, 1.0 μg plasmid DNA for each coverslip was delivered into cells by Fugene6 kit (Roche). After 24–36h growing, cells displaying significant fluorescence protein expression were used for confocal microscopy.

Confocal microscopy and data analysis

Cerulean, Venus and stFRET transfected HEK cells were visualized by LSM510 META confocal microscope (Carl Zeiss) 48h after transfection. Ar laser (458, 514 nm) lines were employed for excitation of FRET donor and acceptor. One multi-channel stack of the confocal images of Cerulean, Venus and FRET were obtained with oil-merged 63×, 1.4 numerical aperture apochromat objective lens (Carl Zeiss) and CFP, YFP or FRET filter sets, meanwhile, DIC (differential interference contrast) images were taken. Sensitized Emission method was used to collect the images from donor channel, acceptor channel and FRET channel. Data acquisition and processing were performed by FRET plus macro with Xia’s method [44]. Normalized FRET index was calculated pixel by pixel with equation: $N_{\mathit{FRET}}=\frac{I_{\mathit{FRET}}-I_{\mathit{YFP}}\times a-I_{\mathit{CFP}}\times b}{\sqrt{I_{\mathit{YFP}}\times I_{\mathit{CFP}}}}$ in which the numerator is the net fluorescence energy transfer nFRET and constant a and b are the ratio of bleed-through of YFP signal into FRET channel and CFP signal into FRET channel [47].

In vitro fluorescence energy transfer measurement

We used a fluorescence spectrometer (Aminco, Bowman Series 2) to measure the fluorescence of purified proteins in solution. All purified proteins were exchanged into 10mM Tris-HCl buffer before processing. The efficiency was usually measured at room temperature with 200μl of 100μM protein. The spectrometer was set to: bandpass, 4nm, 1nm step size, emission scan range, 450–550nm for measuring FRET; 450–500nm for Cerulean monomer and 520–600nm for Venus monomer. Cerulean excitation was at 433nm and Venus at 515nm.

FRET Efficiency

We have used two indexes of energy transfer: (1) $E=\frac{I_{\mathit{DFree}(475\mathit{Emission},433\mathit{Excitation})}-I_{DA(475\mathit{Emission},433\mathit{Excitation})}}{I_{\mathit{DFree}(475\mathit{Emission},433\mathit{Excitation})}}$ [48] in which I_DFree is the signal intensity of free donor, I_DA is the donor fluorescence intensity when connected to acceptor. (2) Donor/acceptor ratio $D/\mathit{ARatio}=\frac{I_{D(475\mathit{Emission},433\mathit{Excitation})}}{I_{A(527\mathit{Emission},433\mathit{Excitation})}}$ [49]. In equation (1) I_DFree was obtained from Cerulean and Venus 1:1 mixture; I_DA is the fluorescence intensity of the donor when it is linked to the acceptor; in equation 2 I_D is the donor emission at 475nm and I_A is the acceptor emission at 527nm. Excitation and emission wavelengths for data acquisition are shown in parenthesis.

Modeling and calibration

For an α-helix, deletion of one amino acid changes the length by 0.15nm and the angle between the termini by 100 degrees. Assuming that the cassette was a rigid body, we made six mutants that would have known spacing and angles to calculate the probe geometry. Figure 1 is a diagram of stFRET geometry as deduced from the procedure that follows. HC→e donor dipole vector and GB→ is the acceptor dipole vector. R_D–A ( BC→ ) is the distance between donor and acceptor and r ( GH→ ) is the length of the linker. θ_A is the angle between acceptor and the linker axis and θ_D is the angle between donor and linker axis. θ_T is the angle between the acceptor and donor. Φ is the dihedral angle between plane (D. r) and plane (A. r). In the diagram, BI→ is equal and parallel to FE→ and both are perpendicular to line GH→. BF→ is parallel and equal to IE→. CE→ is also perpendicular to EI→. Let CH = GB = Z, then Z is the half height of the Cerulean and Venus β-cans. After some trigonometry and algebra, we found that:

$$R_{A–D}=2Z^2(1-sin{\theta }_{A}sin{\theta }_{D}cos\phi +cos{\theta }_{A}cos{\theta }_D)+2rZ(cos{\theta }_D+cos{\theta }_A)+r^2$$ (1)

The relative orientation factor κ² and cos _T are defined in reference [50] as:

$${\kappa }^2={(cos{\theta }_T-3cos{\theta }_{D}cos{\theta }_A)}^2$$ (2)

$$cos{\theta }_T=sin{\theta }_{D}sin{\theta }_{A}cos\phi +cos{\theta }_{D}cos{\theta }_A$$ (3)

The distance R_o, at which E = 50% is given implicitly by [51] :

$$R_o^6=\frac{9000(ln10){\kappa }^2{\phi }_d}{128{\pi }^{5}N{n}^4}{\int }_0^{\infty }f(\lambda )\epsilon (\lambda ){\lambda }^{4}d\lambda ={\kappa }^{2}C$$ (4)

where C is a constant characteristic of the spectral properties of Cerulean and Venus and R_o = 4.9 nm [35] so that C=R_o⁶/(2/3)=4.9⁶(2/3). Since

$$E=\frac{{R_o}^6}{R_o^6+{R_{A-D}}^6}$$ (5)

substituting equations 1, 2, 3, and 4 into 5 yielded:

$$E_{st\mathit{FRET}}=\frac{{(sin{\theta }_{D}sin{\theta }_{A}cos\phi -2cos{\theta }_{D}cos{\theta }_A)}^{2}C}{{(sin{\theta }_{D}sin{\theta }_{A}cos\phi -2cos{\theta }_{D}cos{\theta }_A)}^{2}C+{[{(\text{Z}sin{\theta }_{\text{D}})}^2+{(Zsin{\theta }_{\text{A}})}^2-2{\text{Z}}^{2}sin{\theta }_{\text{D}}sin{\theta }_{\text{A}}cos\phi +{(\text{r}+\text{Z}cos{\theta }_{\text{A}}+\text{Z}cos{\theta }_{\text{D}})}^2]}^3}$$ (6)

There are only three unknowns in Eq. 6, θ_A, θ_D and Φ which determine the orientation factor κ² as well as the global configuration of the protein. With six linker mutants we had six equations to calculate the three unknowns that we did with a least squares solution in Maple.

Testing of the linker in purified stFRET

Purified proteins were subjected to proteinase K digestion, temperature and or urea melting. One unit of proteinase K (500units/ml) was used to digest 200μl protein solution (100μM) in HEPES buffer (100mM HEPES, 100mM NaCl, 10mM Na2HPO4, PH 7.4) for 20 seconds, 1 minute, 2 minutes, 5 minutes, 10 minutes or 30 minutes. Cerulean and Venus monomers are also treated by proteinase K under the same conditions as controls. Urea treatment used 10μl protein (10mg/ml in 10mM Tris-HCl buffer) diluted into 200μl 8M, 7M, 6M, 5M, 4M, 3M, 2M, 1M or HEPES buffer only and incubated at room temperature for 10 minutes. Thermal melting was done by heating the stFRET solutions to 60, 70 and 80C° for 2–5min and immediately measuring the fluorescence energy transfer by spectrometer.

Stretching stFRET on a silicone rubber sheet

Silicone rubber disks with amino modified surfaces (Flexcell International, Hillsborogh, NC) were converted to carboxyl groups with 0.1 mM Methyl N-succinimidyl adipate (MSA) in phosphate buffered saline (PBS) for 1 hr at RT. These groups were converted to a crosslinkers by treating them with 2 mM 1-Ethyl-3-[3-dimethylaminopropyl]carbodiimide hydrochloride (EDC) and 5 mM N-hydroxy-succinimide (NHS) in 0.1 M MES buffer pH 6 for 15 min. Streptactin (a streptavidin variant, IBA, St. Louis, MO) at 1 mg/ml in PBS was then crosslinked to the surface for 2 hrs at room temperature.

Derivatized stFRET proteins were constructed with Streptag II linked to the C-terminal for single tagged stFRET and to the C- and N- terminals for double tagged stFRET. These proteins were allowed to bind overnight to untreated and to Streptactin modified, disks in PBS/0.05% Tween-20/1% BSA. The disks were then washed 3X for 2 hrs with PBS/0.05% Tween-20/1% BSA. Membranes were placed on a modified StageFlexer (Flexcell International, Hillsborogh, NC) which allowed us to apply equibiaxial strain to the disks with suction from an HSPC-1 Pressure Clamp (ALA Instruments, Westbury, NY) controlled by Axon Instruments pClamp software (Molecular Devices, Sunnydale, CA). Application of −200 mmHg of suction produced 10–15% strain (strained diameter/initial diameter). CFP-YFP emission intensities were monitored on an Axiovert 135 microscope (Zeiss, Germany) equipped with a Dual-View DV-CC beam splitter (Photometrics, Germany) with CFP-YFP splitter optics and an iXon DV-887 EM cooled CCD camera (Andor, Northern Ireland). The FRET ratio was determined using ImageJ software (NIH) to analyze and process the video data of the stretched membrane. FRET ratio = (I₅₃₅–I_{535 CFP Bleed})/I₄₇₀, where I₄₇₀ = emission intensity at 470 nm, I₅₃₅ = emission intensity at 535 nm and I_{535 CFP Bleed} = calculated fractional bleed of CFP fluorescence (0.9 × I₄₇₀) into the 535 nm channel.

The equibiaxial strain was measured by placing fiducial marks on the rubber and measuring the resulting strain.

Abbreviations

FRET

fluorescence energy transfer

HEK

human embryonic kidney

GFP

green fluorescence protein

D/ARatio

Donor emission to acceptor emission ratio

E

FRET energy transfer efficiency