2-color photobleaching experiments reveal distinct intracellular dynamics of two components of the Hsp90 complex

E XP E RI ME N TA L CE L L RE S E A RCH 3 1 2 ( 2 00 6 ) 3 9 4 9 –39 5 8

a v a i l a b l e a t w w w. s c i e n c e d i r e c t . c o m

w w w. e l s e v i e r. c o m / l o c a t e / y e x c r

Research Article

2-color photobleaching experiments reveal distinct intracellular dynamics of two components of the Hsp90 complex Didier Picard⁎, Elena Suslova, Pierre-André Briand Département de Biologie Cellulaire, Université de Genève, Sciences III, 30 quai Ernest-Ansermet, CH-1211 Genève 4, Switzerland

ARTICLE INFORMATION

ABS T R AC T

Article Chronology:

The abundant molecular chaperone Hsp90 functions in association with co-chaperones

Received 12 July 2006

including p23 to promote the folding and maturation of a subset of cytosolic proteins.

Revised version received

“Fluorescence recovery after photobleaching” (FRAP) experiments showed that the

21 August 2006

dynamics of p23 in live cells is dictated by Hsp90. Since Hsp90 is present in large excess

Accepted 26 August 2006

over p23, the mobility of Hsp90 could conceivably be quite different. To facilitate the analysis

Available online 3 September 2006

and to allow a direct comparison with p23, we developed a 2-color FRAP technique. Two test proteins are expressed as fusion proteins with the two spectrally separable fluorescent

Keywords:

proteins mCherry and enhanced green fluorescent protein (EGFP). The 2-color FRAP

Confocal microscopy

technique is powerful for the concomitant recording of two proteins located in the same

Kinetics

area of a cell, two components of the same protein complex, or mutant and wild-type

Interaction

versions of the same protein under identical experimental conditions. 2-color FRAP of Hsp90

In vivo

and p23 is virtually indistinguishable, consistent with the notion that they are both engaged

Molecular chaperone

in a multitude of large protein complexes. However, when Hsp90–p23 complexes are

Geldanamycin

disrupted by the Hsp90 inhibitor geldanamycin, p23 moves by free diffusion while Hsp90 maintains its low mobility because it remains bound in remodeled multicomponent complexes. © 2006 Elsevier Inc. All rights reserved.

Introduction Biochemical methods allow the characterization of many aspects of protein–protein interactions. However, extrapolating from the test tube to living cells is not straightforward and largely guesswork, particularly for complex protein networks and transient interactions. Advanced microscopy methods provide a powerful and more direct alternative to investigate macromolecular movements and interactions in living cells. Although these methods all involve rendering one or several proteins of interest, fluorescent typically by fusion to an intrinsically fluorescent protein, an impressive collection of

tools is now available [1,2]. One of the options to gain insights into the cellular biochemistry of a macromolecule such as a fluorescent protein is to monitor its “Fluorescence recovery after photobleaching” (FRAP) [2–5]. Using a confocal microscope, the fluorescent molecules within a small region of a live cell are bleached, and then the recovery of the fluorescence is recorded over time. Recovery can only come from molecules that move into the bleached area from outside and replace the bleached ones in macromolecular complexes within this area. Therefore, the kinetics of this recovery is a function of diffusion and macromolecular interactions of the labeled protein.

⁎ Corresponding author. Fax: +41 22 379 6928. E-mail address: [email protected] (D. Picard). 0014-4827/$ – see front matter © 2006 Elsevier Inc. All rights reserved. doi:10.1016/j.yexcr.2006.08.026

3950

E XP E RI ME N TA L CE LL RE S E A RCH 3 1 2 ( 2 00 6 ) 3 9 4 9 –39 5 8

The cytosolic isoform(s) of the ubiquitous molecular chaperone Hsp90 is essential in all eukaryotes even at normal physiological temperatures [6–8]. Metazoans also have distinct Hsp90 isoforms in the endoplasmic reticulum, and in mitochondria and chloroplasts, but their functions are less well understood. In the cytosol, the Hsp90 machine is required for folding and maturation of a subset of cytosolic proteins [9], many of which are signaling molecules that may need to be maintained in a metastable conformation until activated (for an updated list, see http://www.picard.ch/downloads/downloads. htm). Hsp90 acts in concert with a large cohort of co-chaperones that assist it by regulating its ATPase function and interaction with substrate proteins [6,8,10,11]. We recently analyzed the intracellular dynamics of the Hsp90 co-chaperone p23 using the FRAP technique [12]. We found that its movements and interactions are dictated by Hsp90. At steady-state, the bulk of p23 is associated with Hsp90. This intracellular behavior is interesting in light of the view that p23 binds the ATP-bound substratebinding form of Hsp90 [13–23]. The p23–Hsp90 interaction is targeted by a variety of Hsp90-specific anti-cancer drugs such as geldanamycin (GA) [16,24–26], and indeed, GA frees p23 from Hsp90 in living cells as judged by FRAP [12]. The intracellular dynamics of Hsp90 itself is unknown and possibly more complex since it engages in a large number of complexes with its cochaperones with and without p23. In turn, this chaperone machine interacts with a wide range of substrate proteins. FRAP experiments are perfectly suited to gain a global view of the intracellular behavior of Hsp90. FRAP experiments have almost exclusively been performed with a single color to investigate one fluorescent macromolecule at a time. For many analyses, it would be desirable to be able to track at least two different macromolecules at the same time, regardless of whether these macromolecules interact with one another or are simply present in the same cellular locale. With spectrally separable fluorescent proteins having become available [27], we developed a 2-color FRAP method that allows the simultaneous analysis of two different proteins, in general, and of Hsp90 and its co-chaperone p23, in particular.

Materials and methods

for human Trap1 into vector pEGFP-N1 (Clontech) between Bgl2 and EcoR1 sites. Since the Ensembl database entry ENSG00000126602 for human Trap1 indicates an N-terminus that begins with the peptide sequence MARELRA…, the fulllength Trap1 sequence was reconstituted from plasmid pBluescript-TRAP1 [28] (a gift from David B. Donner) and the EST clone IMAGE:6014494.

Cell culture and transfection Human HeLa cells were cultured in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% fetal calf serum. For coimmunoprecipitation experiments, cells were transfected by the calcium-phosphate coprecipitation technique. For FRAP experiments, transfections were done in 35 mm glass bottom culture dishes (MatTek Corporation) with FuGENE 6 (Roche) as directed by the manufacturer. Cells were used for FRAP experiments 28–30 h post-transfection.

Immunoprecipitation experiments Cells were washed and harvested with a rubber policeman in Tris-buffered saline about 42 h after transfection. They were lysed on ice in 10 mM Tris–HCl pH 7.4, 50 mM NaCl, 10 mM Namolybdate, 1 mM EDTA, 0.1% NP-40, 10% glycerol, 1 mM EDTA, 1 mM DTT and protease inhibitors. About 1 mg of cleared lysate was immunoprecipitated in the same buffer complemented with 0.25 mM ATPγS using the mouse monoclonal JJ3 directed against p23 (a kind gift from David O. Toft), a polyclonal rabbit antiserum against GFP (Molecular Probes), or the unrelated antibody 9E10 by tumbling for 2 h at 4°C. Antibody–antigen complexes were recovered with Protein G–Sepharose 4 Fast Flow (Amersham Biosciences) by tumbling for 2 h at 4°C and washed three times with ice-cold lysis buffer. Samples were boiled in sample buffer and electrophoresed on a 4–12% SDSpolyacrylamide gel. Following transfer to a nitrocellulose membrane, Hsp90 in the upper part of the membrane was revealed by probing with the monoclonal H90-10 (a kind gift from David O. Toft), whereas p23 and the EGFP-p23 fusion protein were probed for with JJ3 in the lower part of the membrane. To avoid overlap of the p23 band with that of the immunoglobulin light chain, a heavy-chain specific secondary antibody was used.

Plasmids Fluorescence microscopy A plasmid with the mCherry [27] coding region was generously provided by R. Y. Tsien. The open reading frame was cloned into the pEGFP-C1 (Clontech) backbone, replacing the EGFP coding sequence between the Nhe1 and Bgl2 sites, to generate the vector pmCherry-C1. It allows the expression of C-terminal mCherry fusion proteins in mammalian cells under the control of the CMV enhancer/promoter. Constructs pEGFP-p23 and pmCherry-p23 were generated by insertion of an EcoR1/BamH1 fragment containing the entire human p23 coding sequence into plasmids pEGFP-C1 and pmCherry-C1, respectively. Constructs pCherry.90β and pEGFP.90β for expression of mCherryHsp90β and EGFP-Hsp90β were obtained by insertion of the human Hsp90β coding sequence as a Bgl2–Sal1 fragment into pmCherry-C1 and pEGFP-C1, respectively. Construct Trap1EGFP was generated by insertion of the entire coding sequence

Standard epifluorescence microscopy was performed with a Zeiss Axiophot microscope connected to a Retiga CCD camera (QImaging). HeLa cells were seeded and transfected on glass cover slips. They were fixed with 3.7% formaldehyde in phosphate-buffered saline and mounted on slides with mowiol 4-88. EGFP and mCherry fluorescence was imaged with fluorescein and rhodamine filter sets, respectively.

FRAP experiments Experiments were performed with a Leica TCS SP2 AOBS confocal microscope with a 63× oil immersion objective. Stage and objective were placed in an environment box preequilibrated to 37°C. Until used for a FRAP experiment, cells in glass bottom

E XP E RI ME N TA L CE L L RE S E A RCH 3 1 2 ( 2 00 6 ) 3 9 4 9 –39 5 8

dishes were maintained in imaging medium (“Minimal Essential Medium Eagle” [MEM from Sigma], without phenol red, complemented with glutamine and buffered by adding 30 mM HEPES pH 7.4) in a 37°C incubator. Where indicated, GA was added to 1 μM from a 1 mM stock in DMSO about 15 min before FRAP experiments were initiated. The settings for scanning were as follows: bidirectional scanning at 1000 Hz, 6× zoom, image format 512 × 64, 4% and 5% laser power at 458 and 594 nm, respectively. Fluorescence emission from EGFP and mCherry was detected simultaneously in two channels with the ranges of photomultiplier tubes (PMTs) 2 and 3 set to 490– 552 nm and 601–707 nm, respectively (with a maximal gain of 750 and 2% offset). Bleaching was performed with a single pass with all laser lines between 458 and 594 nm set to maximal power. The bleach area was a circle with a radius of 1.9 μm, and average frame rates were 1 frame/87 ms. Average background values were determined outside of cells and subtracted from all raw data points before further analysis. Curve fitting was done with the software Prism (Graphpad).

Results Experimental strategy for 2-color FRAP analyses

3951

mCherry [27] fulfil these criteria (see also below). Note that EGFP and mCherry are derived from two different species but have similar molecular weights. EGFP and mCherry are variants of the original fluorescent proteins from Aequorea victoria and Discosoma sp. and are 29 and 27 kDa, respectively. In addition, both fluorescent proteins are monomeric and bleachable [27,29]. To establish the 2-color FRAP technology, several technical issues needed to be addressed: (i) the two fluorescent proteins must be distinguishable at the same time with minimal or no spectral overlap; (ii) the “colors”, i.e. the fluorescent proteins, must be interchangeable between two test proteins; (iii) the results obtained with a 2-color FRAP analysis must be identical to those obtained with FRAP experiments performed separately for the two test proteins. Once established, we wanted to apply this new technology to new biological questions, and particularly to study the Hsp90 chaperone machine. We decided to explore three different situations in which 2-color FRAP analysis could be a powerful new tool: (i) two proteins in different but overlapping cellular compartments; (ii) two components of the same molecular complex; (iii) the same two components in the presence of an inhibitory drug that disrupts this complex. As a reference, we chose to use the Hsp90 co-chaperone p23 whose behavior we recently characterized in detail [12]. The other test proteins were the unfused fluorescent proteins EGFP and mCherry, the cytosolic Hsp90 isoform Hsp90β fused to either EGFP or mCherry and the mitochondrial Hsp90 isoform Trap1 [30] fused to EGFP.

To monitor the intracellular dynamics of proteins by FRAP experiments, they must be fluorescently labeled, for example, by expression as fusions with fluorescent proteins. Since we wanted to monitor two proteins at the same time, two different fluorescent proteins had to be chosen whose excitation and emission spectra would be well separated. Fig. 1 illustrates that EGFP and the recently developed red fluorescent protein

Initial characterization of the fluorescent Hsp90β and p23 fusion proteins

Fig. 1 – Spectral properties of EGFP and mCherry allow independent excitation and recording. Excitation and emission spectra are shown in blue and green, respectively, for EGFP, and yellow and red, respectively, for mCherry. Peak wavelengths are indicated above the graphs. Excitation (arrows indicate laser lines) and emission (colored bars indicate PMT settings) parameters used for FRAP experiments are schematically shown below the graphs. EGFP and mCherry spectra were adapted from https://www.omegafilters.com and [27], respectively. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

First, we determined the subcellular localization of Hsp90β and p23 as fluorescent fusion proteins. As we previously reported for a EYFP fusion protein of p23, EGFP-p23 distributes throughout the cell (Fig. 2). In contrast, the mCherry-Hsp90β fusion protein is predominantly cytoplasmic although some cells show clear nuclear fluorescence as well (see bottom panel in Fig. 2). This pattern agrees with the established localization of endogenous Hsp90 [31–33] and mirrors that seen for the other Hsp90 isoform, Hsp90α, as an EGFP fusion protein [34]. It is therefore not surprising that coexpressed EGFP-p23 and mCherry-Hsp90β largely colocalize as indicated by the yellow color in the merged micrographs of Fig. 2. Although these images seem to suggest an excess of EGFP-p23 in the nucleus, the protein levels cannot be directly compared based on the apparent fluorescence. Moreover, our previous investigation with the FRAP technology clearly demonstrated that the vast majority of p23 molecules are complexed with endogenous Hsp90 both in the cytoplasm and in the nucleus [12]. It should be noted that the reciprocal fusion proteins EGFP-Hsp90β and mCherry-p23 give exactly the same patterns (data not shown) and that cells with moderate expression of fusion proteins were selected to acquire both such micrographs and FRAP data. The immunoblot shown in Fig. 3 (lanes marked “input”) gives an impression of the expression levels of the “overexpressed” fluorescent proteins. Considering that only a fraction of cells transiently express the exogenous Hsp90 proteins, these are expressed at similar levels to endogenous Hsp90.

3952

E XP E RI ME N TA L CE LL RE S E A RCH 3 1 2 ( 2 00 6 ) 3 9 4 9 –39 5 8

Fig. 2 – Extensive colocalization of p23 and Hsp90 fluorescent fusion proteins. The fusion proteins schematically illustrated above the micrographs were transiently expressed in HeLa cells. Two representative micrographs of cells co-expressing the two fusion proteins were acquired by standard epifluorescence microscopy. The false color images were merged to reveal extensive colocalization in yellow (panel on the right). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Second, it was important to ascertain that the two fusion proteins are able to interact with each other. While we had previously shown that EGFP-p23 interacts with endogenous Hsp90 [12], we had no direct evidence that the N-terminal fluorescent moiety of mCherry-Hsp90β (or EGFP-Hsp90β) would

not interfere with binding to exogenous or endogenous p23. This was assessed by an immunoprecipitation (IP) experiment with proteins transiently expressed in HeLa cells. As can be seen in Fig. 3, an antibody to the EGFP moiety of p23 can immunoprecipitate both endogenous Hsp90 and exogenously expressed mCherry-Hsp90. In a control IP, an antibody to endogenous p23 also immunoprecipitates both fused and unfused Hsp90. The ratio of immunoprecipitated mCherry-Hsp90β to endogenous Hsp90 seems to be similar in the input lysates and both immunoprecipitates. Thus, complexes can readily be formed between p23 and Hsp90 molecules both carrying an N-terminal fluorescent protein of 27–29 kDa. This is consistent with the recently reported crystal structure of the yeast Hsp90– p23 complex that shows accessible and protruding N-termini [23].

2-color FRAP set-up

Fig. 3 – Fluorescent fusion proteins of p23 and Hsp90 can interact. mCherry-Hsp90β alone or in combination with EGFP-p23 was transiently expressed in HeLa cells. Immunoprecipitations (IP) were carried out with antibodies indicated below the bottom panel. Immunoblots to reveal proteins present in the total lysates (input) and immunoprecipitates were performed with antibodies against Hsp90 (top panel) and p23 (bottom two panels).

As indicated above, the excitation and emission spectra of EGFP and mCherry suggested that fluorescence from the two proteins could be monitored without overlap (Fig. 1). Details are given in the Materials and methods section, but it should be pointed out that EGFP has to be excited with the 458 nm laser line. For EGFP, the 488 nm laser line would be more optimal, but it noticeably excites mCherry as well. At 458 nm and moderate laser power, there is no excitation of mCherry at all. There is no excitation of EGFP fluorescence by the 594 nm laser line used for mCherry. Moreover, we did not detect any excitation of mCherry by the 458 nm line in any of the protein combinations used in this work (data not shown).

E XP E RI ME N TA L CE L L RE S E A RCH 3 1 2 ( 2 00 6 ) 3 9 4 9 –39 5 8

Colors are interchangeable Using p23 fusion proteins and the corresponding free fluorescent proteins, we set out to determine whether the two colors can be combined both ways. In contrast to free mCherry and

3953

EGFP, which could be expected to move by free diffusion, we knew that an EYFP-p23 fusion protein is subject to more complex kinetics [12]. The FRAP experiments of Fig. 4A confirm these predictions and demonstrate that two proteins with different “colors” can be monitored simultaneously. A model for free diffusion can be fitted to the FRAP data of EGFP as first described for a circular bleach spot by Soumpasis [35]: frap(t) = e−τD/2t[I0(τD / 2t) + I1(τD / 2t)], where τD = w2/D (w, radius of bleach spot; D, diffusion constant). In contrast, the kinetics of mCherry-p23 FRAP is controlled by protein–protein interactions rather than free diffusion. Since the intracellular mobility of mCherry-p23 is not only determined by its interaction with Hsp90, but also by the interactions of the Hsp90–p23 complex with yet other co-chaperones and substrates, it is too complex to be analytically modeled. For the convenience of fitting a curve, we chose the “two-phase exponential association” equation frap(t) = 1− C1 * e−k1off *t − C2 *e−k2off *t in all cases where a free diffusion model could not account for the observed data. This equation, which describes the binding of a test protein to two other proteins or complexes, is only used because its flexibility provides for easy fitting to a variety of raw data. For EGFP and mCherry-p23, the two models fit the data with R2 values of 0.9247 and 0.9397, respectively (a value of 1 indicating a perfect fit). The reciprocal experiment with p23 fused to EGFP, and free mCherry gives the expected reciprocal result (Fig. 4B). Fig. 4C illustrates that the fitted curves of the two reciprocal FRAP experiments overlap perfectly.

2-color FRAP faithfully recapitulates separate classical FRAP

Fig. 4 – Reciprocal 2-color FRAP experiments with EGFP and mCherry fusion proteins. (A) Graph of simultaneous FRAP of mCherry-p23 (gray triangles) and EGFP (black squares). The minimal fluorescence values after bleaching at time 0 and the average values at the post-bleach plateau were arbitrarily standardized to 0 and 1, respectively. The data points are averages obtained with 5 nuclei each. The Soumpasis model (free diffusion) (yellow line) and a model for a two-phase exponential association (red line) were used to fit curves to the data. (B) Graph for reciprocal FRAP experiment with mCherry (gray triangles) and EGFP-p23 (black squares). (C) Overlap of the fitted curves of the two reciprocal FRAP experiments shown in panels A (yellow and red lines) and B (black lines). The inset shows a blow-up of the first 3 s of recovery. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

2-color FRAP experiments must yield the same results as a classical FRAP experiment performed separately for two proteins of interest. This was already an underlying assumption for the aforementioned experiments. Note that this could not only be experimentally verified (data not shown), but that the results obtained by 2-color FRAP corresponded exactly to what was expected for the two kinds of proteins (see above). Nevertheless, we decided to investigate this issue more thoroughly with two components of the same protein complex, the Hsp90– p23 complex. FRAP was monitored for fluorescent versions of p23 and Hsp90 expressed either separately or together in the same cells. In both cases, cells were selected that displayed moderate and similar levels of fluorescence, and thus presumably levels of expression of the exogenous fusion proteins. The parameters for recording FRAP were identical for both sets of experiments, which confirmed once again that there is no bleedthrough between the two channels corresponding to EGFP and mCherry (data not shown). Fig. 5 demonstrates that the results are virtually identical for the two sets of experiments. This is particularly clear in panel C of Fig. 5, which shows an almost perfect overlap of the fitted curves. Thus, FRAP of p23 and Hsp90 can be monitored in the same place (bleach area of a cell) and at the same time. Particularly strong support for this conclusion is provided by the fact that it is true for FRAP under two entirely different conditions. In the absence of GA, the dynamics of p23 and Hsp90 are almost identical whereas in the presence of GA, FRAP of p23 is considerably accelerated relative to Hsp90. The Soumpasis

3954

E XP E RI ME N TA L CE LL RE S E A RCH 3 1 2 ( 2 00 6 ) 3 9 4 9 –39 5 8

Fig. 5 – 2-color FRAP experiments faithfully reproduce results obtained with single color experiments. (A) FRAP of EGFP-Hsp90β and mCherry-p23 expressed separately, with and without prior treatment with GA. The data points and fitted curves of the experiments that were performed separately were overlapped for comparison with the graphs of panel B. The plots show all data points from 5 different cells with a bleach area in the cytoplasm. A model for a two-phase exponential association (dark green, olive, and light blue lines) and the Soumpasis model (free diffusion) (light yellow line) were used to fit curves to the data. (B) FRAP of EGFP-Hsp90β and mCherry-p23 coexpressed in the same cells. Data points are from concomitant 2-color recordings from 5 different cells. A model for a two-phase exponential association (red, orange, and blue lines) and the Soumpasis model (free diffusion) (green line) were used to fit curves to the data. (C) Overlap of the fitted curves of the FRAP experiments shown in panels A and B. The inset on the left shows a blow-up of the first 2 s of recovery. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

model for free diffusion fits the p23 FRAP data obtained in the presence of GA with R2 values of 0.7842 and 0.8097 for panels A and B, respectively. Indeed, as we previously demonstrated [12], the recovery of p23 following disruption of the Hsp90–p23 complex by GA indicates a diffusion-controlled process.

Cytosolic and mitochondrial Hsp90 behave differently from p23 Next, we applied the new 2-color FRAP technology to three different questions as outlined above. The mitochondrial Hsp90 isoform Trap1 “colocalizes” with the cytosolic p23 wherever

E XP E RI ME N TA L CE L L RE S E A RCH 3 1 2 ( 2 00 6 ) 3 9 4 9 –39 5 8

mitochondria are located (data not shown). Whereas one can see that Trap1-EGFP is concentrated in mitochondria, we assume that mCherry-p23 is only in the cytosol surrounding these mitochondria since there is no evidence for p23 inside mitochondria. We performed a 2-color FRAP experiment to compare the dynamics of the two proteins. To this end, we selected bleach areas with prominent mitochondrial Trap1-EGFP fluorescence. The results for the two proteins are strikingly different. In contrast to p23, Trap1 fluorescence recovers much more slowly, and only to an average of 47% (compared to 94% for p23) within the time-frame of the experiment (Fig. 6A). The discontinuous and heterogeneous nature of mitochondria within a comparatively large bleach area exceeding the diameter of individual mitochondria probably accounts for the relatively large data spread. When entire mitochondria are

3955

bleached, additional new fluorescent Trap1 molecules may only come from slower processes such as mitochondrial import of de novo translated molecules and fusions of bleached mitochondria with unbleached ones. The present experiment was aimed at illustrating that such studies can be done. Further experiments will be necessary to explore the intramitochondrial mobility of Trap1 in more detail. One of the salient conclusions of our previous FRAP analysis of p23 was that the mobility of p23 is dictated by Hsp90 unless this interaction is blocked by GA [12]. The experiment of Fig. 5 affords the opportunity to examine this protein complex from the Hsp90 angle. The graph shown in Fig. 6B facilitates the reading of the data by displaying only the fitted curves. In the absence of the inhibitory drug GA, the mobility of p23 and Hsp90 is very similar and consistent with complex protein– protein interactions that are graphically well approximated by an equation for a two-phase exponential association (see Discussion). Unlike Trap1 but similarly to p23, the vast majority of the Hsp90 fluorescence (98%) recovers, indicating that almost all Hsp90 (and p23) molecules are not engaged in irreversible or extremely stable interactions within the bleach area and that there is a sufficient pool of highly mobile Hsp90 molecules outside of the bleach area. In the presence of GA, the dynamics of p23 can be explained by free diffusion. In contrast, the intracellular mobility of Hsp90 remains essentially unchanged. It should be noted that the diffusion constant and half-time of recovery for freely pffiffiffiffiffi diffusing molecules are proportional to 3 M , where M is the molecular mass. Thus, the 2.36-fold larger molecular size of EGFP-Hsp90β compared to mCherry-p23 would only translate into a 1.33-fold longer half-time of FRAP. Fitting the Soumpasis model (free diffusion), a one-phase exponential association and a two-phase exponential association to these Hsp90 data yields R2 values of 0.8048, 0.8996 and 0.9304, respectively. This demonstrates that the mobility of Hsp90 in the presence of GA is still subject to complex molecular interactions and clearly not diffusion-controlled.

Discussion

Fig. 6 – The 2-color FRAP technique reveals differential dynamics of Hsp90 isoforms in different compartments and by comparison to p23 upon treatment with GA. (A) 2-color FRAP of cytosolic mCherry-p23 and mitochondrial Trap1-EGFP. The plots show all data points from 5 different cells with a bleach area in the cytoplasm containing EGFP fluorescence in mitochondrial structures. In contrast to the data in Figs. 4 and 5, these data were standardized to the average pre-bleach values set to 1. A two-phase exponential association equation was used to fit curves to the data. (B) The dynamics of two components of the Hsp90 complex, EGFP-Hsp90β and mCherry-p23, are differentially affected by GA. Fitted curves from Fig. 5 were overlapped using the same color code. The inset shows a blow-up of the first 4 s of recovery.

We have described the development of a new FRAP technique that allows the simultaneous analysis of the intracellular kinetics of the interactions of two different proteins. With the 2-color FRAP technique, two different proteins can be monitored at the same time, in the exact same intracellular locale and under identical experimental conditions. We applied the 2-color FRAP method to compare the behavior of different components of the cytosolic and mitochondrial Hsp90 molecular chaperone machines. Whereas Hsp90 and its co-chaperone p23 display very similar dynamics under normal circumstances, their mobility diverges upon disruption of the complex with the Hsp90 inhibitor GA.

2-color FRAP The development of the 2-color FRAP technique only became seriously conceivable with the arrival of mCherry in the cell biologist's tool box of fluorescent proteins [27]. It offers spectral characteristics that allow a clear separation of excitation and

3956

E XP E RI ME N TA L CE LL RE S E A RCH 3 1 2 ( 2 00 6 ) 3 9 4 9 –39 5 8

emission of two different proteins by confocal microscopy (Fig. 1). We used EGFP fusion proteins as partners of choice, but blue-shifted versions such as ECFP could possibly be used as well, provided sufficiently strong laser lines are or become available to bleach ECFP efficiently and rapidly. Alternatively, recombinant proteins could be chemically labeled with compatible fluorochromes and reintroduced into cells. In single cell assays such as these, this could be achieved by microinjection. The use of two spectrally separable proteins is a key feature of the 2-color FRAP technique because it allows the simultaneous acquisition of images in cases with relatively rapid kinetics. For p23 and Hsp90, the half-times of recovery are less than 1 s so that maximal image acquisition rates must be achieved. When the spectra overlap, for example when ECFP is combined with EYFP, images must be acquired one after the other with the multitracking feature that is available on current confocal microscopes. This may be acceptable when recovery times are comparatively slow (see for example [36]), but not for processes as fast as the ones investigated here. It remains to be seen whether the challenging spectral imaging and linear unmixing method, which allows the separation of the signals from fluorescent proteins with overlapping spectra [37,38], can be used as an additional alternative for multicolor FRAP experiments. However, even with the 2-color FRAP technique, there is one possible pitfall of imaging two colors at the same time that must be carefully avoided. With EGFP and mCherry, there is the potential of fluorescence resonance energy transfer (FRET). The emission from EGFP could excite mCherry if the two fluorescent protein moieties are sufficiently close and properly positioned. The FRET signal would erroneously contribute to the FRAP signal observed for the mCherry fusion protein. This potential problem can only arise for proteins in the same complex and can easily be controlled for by imaging with the 458 nm laser line alone without the 594 nm laser line, which is used to excite mCherry directly. There are a number of applications for which the 2-color FRAP technique is particularly useful. The ones explored here include the investigation of the intracellular dynamics of two proteins in different but overlapping cellular compartments and of two proteins of the same complex. The latter can also be monitored after subjecting the cells to various experimental conditions. Moreover, wild-type and mutant versions of a protein can be studied together, with the wild-type providing an internal reference.

Distinct intracellular dynamics of Hsp90 and p23 To follow up on our previous study on the Hsp90 co-chaperone p23 [12], we applied the 2-color FRAP technique to Hsp90 itself. The concomitant recording of the FRAP of p23 and Hsp90 reveals unambiguously that these two proteins have almost identical intracellular dynamics in the absence of GA. It should be emphasized that this does not mean that they are involved in identical protein–protein interactions. However, it is consistent with the view that both Hsp90 and p23 are engaged in large and only partially overlapping protein complexes. In the presence of GA, p23 cannot bind Hsp90 and diffuses freely whereas Hsp90 continues to be associated with other cochaperones and even substrate proteins. The following paragraphs provide more details on this interpretation.

Biochemical experiments [17–19] and the recently reported crystal structure of the yeast Hsp90–p23 complex [23] show that two molecules of p23 bind a dimer of Hsp90. Thus, the binding equilibrium 2*p23 + Hsp902 ⇔ p232•Hsp902 would describe this reaction if no other co-chaperones and no substrate proteins were involved. As we previously discussed [12], both theoretical considerations and p23 FRAP experiments demonstrate that the vast majority of p23 molecules are Hsp90-bound at steady-state. Therefore, from a p23 perspective, and if the ordered binding of other co-chaperones is ignored, the reaction becomes 2*p23 + Hsp902 ⇔ p232•Hsp902 + Xn ⇔ p232•Hsp902•Xn, where Xn stands for a multitude of Hsp90 substrates. This can account for the FRAP data for p23, but it is inappropriate to explain the seemingly identical Hsp90 data. Most Hsp90 molecules are not associated with p23 since the former is present in large excess over the latter (for discussion, see [12]). Nevertheless, we expect a large fraction of Hsp90 to be associated with one or several of the other co-chaperones such as Hop, Cdc37, Aha1, or the immunophilins [6,8,10,11]. Some of these interactions might preferentially involve Hsp90 that is not in a complex with p23 [20,39,40]. Despite the fact that p23 binds and stabilizes the ATP-bound and substratebinding form of Hsp90 [19,21,22,41], Hsp90 most likely binds substrates with lower affinity even without p23 at some point during the folding cycle. Moreover, it is not clear to what extent certain types of Hsp90 substrates are entirely p23-independent. In the presence of GA, Hsp90 complexes are remodeled since Hsp90 adopts a conformation that is similar to that induced by ADP [42] and fails to bind p23. One of several prominent co-chaperones that can bind GA-bound Hsp90 is Hop, which is typically considered an “early” co-chaperone in the assembly pathway of Hsp90–substrate complexes. Hop binds both Hsp70 and Hsp90 and is therefore thought to promote a substrate handover from the Hsp70 molecular chaperone system to Hsp90 [43–47]. These early effects of GA can even stabilize the interaction of Hsp90–Hop complexes with certain substrates [48] although ultimately GA induces the degradation of most Hsp90 substrates [7,24]. Unlike p23, whose GA-induced intracellular mobility is characterized by free diffusion commensurate with its molecular size, Hsp90 continues to be involved in a multitude of high molecular weight complexes. These are heterogeneous and yet distinct from those that Hsp90 engages in without the drug. Discriminating between these multitudes of complexes, notably if their sizes are roughly similar, is beyond the resolution of the FRAP technique, be it single or 2-color.

Acknowledgments We are very grateful to David B. Donner, David O. Toft, and Roger Y. Tsien for gifts of reagents, and to Timo Zimmermann for discussions in the early phase of the project. We thank Sophie Carascossa for critical reading of the manuscript. FRAP experiments were performed at the bioimaging platform of the NCCR “Frontiers in Genetics” at the Université de Genève with kind help from Christoph Bauer and his collaborators. The work was supported by the Canton de Genève and the Swiss National Science Foundation.

E XP E RI ME N TA L CE L L RE S E A RCH 3 1 2 ( 2 00 6 ) 3 9 4 9 –39 5 8

REFERENCES

[1] B.N. Giepmans, S.R. Adams, M.H. Ellisman, R.Y. Tsien, The fluorescent toolbox for assessing protein location and function, Science 312 (2006) 217–224. [2] J. Lippincott-Schwartz, G.H. Patterson, Development and use of fluorescent protein markers in living cells, Science 300 (2003) 87–91. [3] B.L. Sprague, J.G. McNally, FRAP analysis of binding: proper and fitting, Trends Cell Biol. 15 (2005) 84–91. [4] B.L. Sprague, R.L. Pego, D.A. Stavreva, J.G. McNally, Analysis of binding reactions by fluorescence recovery after photobleaching, Biophys. J. 86 (2004) 3473–3495. [5] R.N. Day, F. Schaufele, Imaging molecular interactions in living cells, Mol. Endocrinol. 19 (2005) 1675–1686. [6] D. Picard, Heat-shock protein 90, a chaperone for folding and regulation, Cell. Mol. Life Sci. 59 (2002) 1640–1648. [7] L. Whitesell, S.L. Lindquist, HSP90 and the chaperoning of cancer, Nat. Rev., Cancer 5 (2005) 761–772. [8] L.H. Pearl, C. Prodromou, Structure and mechanism of the hsp90 molecular chaperone machinery, Annu. Rev. Biochem. 75 (2006) 271–294. [9] D.F. Nathan, M.H. Vos, S. Lindquist, In vivo functions of the Saccharomyces cerevisiae Hsp90 chaperone, Proc. Natl. Acad. Sci. U. S. A. 94 (1997) 12949–12956. [10] A.J. Caplan, What is a co-chaperone? Cell Stress Chaperones 8 (2003) 105–107. [11] D. Picard, Chaperoning steroid hormone action, Trends Endocrinol. Metab. 17 (2006) 229–235. [12] D. Picard, Intracellular dynamics of the Hsp90 co-chaperone p23 is dictated by Hsp90, Exp. Cell Res. 312 (2006) 198–204. [13] J.L. Johnson, T.G. Beito, C.J. Krco, D.O. Toft, Characterization of a novel 23-kilodalton protein of unactive progesterone receptor complexes, Mol. Cell. Biol. 14 (1994) 1956–1963. [14] S.J. Felts, D.O. Toft, p23, a simple protein with complex activities, Cell Stress Chaperones 8 (2003) 108–113. [15] J. Johnson, R. Corbisier, B. Stensgard, D.O. Toft, The involvement of p23, hsp90, and immunophilins in the assembly of progesterone receptor complexes, J. Steroid Biochem. Mol. Biol. 56 (1996) 31–37. [16] J.L. Johnson, D.O. Toft, Binding of p23 and hsp90 during assembly with the progesterone receptor, Mol. Endocrinol. 9 (1995) 670–678. [17] A.J. Weaver, W.P. Sullivan, S.J. Felts, B.A. Owen, D.O. Toft, Crystal structure and activity of human p23, a heat shock protein 90 co-chaperone, J. Biol. Chem. 275 (2000) 23045–23052. [18] C. Prodromou, B. Panaretou, S. Chohan, G. Siligardi, R. O'Brien, J.E. Ladbury, S.M. Roe, P.W. Piper, L.H. Pearl, The ATPase cycle of Hsp90 drives a molecular ‘clamp’ via transient dimerization of the N-terminal domains, EMBO J. 19 (2000) 4383–4392. [19] K. Richter, S. Walter, J. Buchner, The co-chaperone Sba1 connects the ATPase reaction of Hsp90 to the progression of the chaperone cycle, J. Mol. Biol. 342 (2004) 1403–1413. [20] G. Siligardi, B. Hu, B. Panaretou, P.W. Piper, L.H. Pearl, C. Prodromou, Co-chaperone regulation of conformational switching in the Hsp90 ATPase cycle, J. Biol. Chem. 279 (2004) 51989–51998. [21] W.P. Sullivan, B.A. Owen, D.O. Toft, The influence of ATP and p23 on the conformation of hsp90, J. Biol. Chem. 277 (2002) 45942–45948. [22] S.H. McLaughlin, H.W. Smith, S.E. Jackson, Stimulation of the weak ATPase activity of human Hsp90 by a client protein, J. Mol. Biol. 315 (2002) 787–798. [23] M.M. Ali, S.M. Roe, C.K. Vaughan, P. Meyer, B. Panaretou, P.W.

[24] [25]

[26] [27]

[28]

[29] [30]

[31]

[32]

[33]

[34]

[35]

[36]

[37]

[38]

[39]

[40]

[41]

[42]

3957

Piper, C. Prodromou, L.H. Pearl, Crystal structure of an Hsp90-nucleotide–p23/Sba1 closed chaperone complex, Nature 440 (2006) 1013–1017. L. Neckers, Hsp90 inhibitors as novel cancer chemotherapeutic agents, Trends Mol. Med. 8 (2002) S55–S61. J.S. Isaacs, W. Xu, L. Neckers, Heat shock protein 90 as a molecular target for cancer therapeutics, Cancer Cell 3 (2003) 213–217. P. Workman, Overview: translating Hsp90 biology into Hsp90 drugs, Curr. Cancer Drug Targets 3 (2003) 297–300. N.C. Shaner, R.E. Campbell, P.A. Steinbach, B.N. Giepmans, A.E. Palmer, R.Y. Tsien, Improved monomeric red, orange and yellow fluorescent proteins derived from Discosoma sp. red fluorescent protein, Nat. Biotechnol. 22 (2004) 1567–1572. H.Y. Song, J.D. Dunbar, Y.X. Zhang, D.Q. Guo, D.B. Donner, Identification of a protein with homology to hsp90 that binds the type 1 tumor necrosis factor receptor, J. Biol. Chem. 270 (1995) 3574–3581. R.Y. Tsien, The green fluorescent protein, Annu. Rev. Biochem. 67 (1998) 509–544. S.J. Felts, B.A. Owen, P. Nguyen, J. Trepel, D.B. Donner, D.O. Toft, The hsp90-related protein TRAP1 is a mitochondrial protein with distinct functional properties, J. Biol. Chem. 275 (2000) 3305–3312. P.M. van Bergen en Henegouwen, G. Berbers, W.A. Linnemans, R. van Wijk, Subcellular localization of the 84,000 dalton heat-shock protein in mouse neuroblastoma cells: evidence for a cytoplasmic and nuclear location, Eur.J. Cell Biol. 43 (1987) 469–478. G.A. Berbers, R. Kunnen, P.M. van Bergen en Henegouwen, R. van Wijk, Localization and quantitation of hsp84 in mammalian cells, Exp. Cell Res. 177 (1988) 257–271. G. Akner, K. Mossberg, K.G. Sundqvist, J.-Å. Gustafsson, A.C. Wikström, Evidence for reversible, non-microtubule and non-microfilament-dependent nuclear translocation of hsp90 after heat shock in human fibroblasts, Eur.J. Cell Biol. 58 (1992) 356–364. T. Langer, S. Rosmus, H. Fasold, Intracellular localization of the 90 kDA heat shock protein (HSP90α) determined by expression of a EGFP-HSP90α-fusion protein in unstressed and heat stressed 3T3 cells, Cell Biol. Int. 27 (2003) 47–52. D.M. Soumpasis, Theoretical analysis of fluorescence photobleaching recovery experiments, Biophys. J. 41 (1983) 95–97. D.L. Stenoien, A.C. Nye, M.G. Mancini, K. Patel, M. Dutertre, B.W. O'Malley, C.L. Smith, A.S. Belmont, M.A. Mancini, Ligand-mediated assembly and real-time cellular dynamics of estrogen receptor α-coactivator complexes in living cells, Mol. Cell. Biol. 21 (2001) 4404–4412. T. Zimmermann, Spectral imaging and linear unmixing in light microscopy, Adv. Biochem. Eng. Biotechnol. 95 (2005) 245–265. T. Zimmermann, J. Rietdorf, R. Pepperkok, Spectral imaging and its applications in live cell microscopy, FEBS Lett. 546 (2003) 87–92. G.M. Scholz, K. Cartledge, N.E. Hall, Identification and characterization of Harc: a novel Hsp90 associating relative of Cdc37, J. Biol. Chem. 18 (2001) 18. A. Harst, H. Lin, W.M. Obermann, Aha1 competes with Hop, p50 and p23 for binding to the molecular chaperone Hsp90 and contributes to kinase and hormone receptor activation, Biochem. J. 387 (2005) 789–796. W.B. Pratt, D.O. Toft, Steroid receptor interactions with heat shock protein and immunophilin chaperones, Endocr. Rev. 18 (1997) 306–360. S.M. Roe, C. Prodromou, R. O'Brien, J.E. Ladbury, P.W. Piper, L.H. Pearl, Structural basis for inhibition of the hsp90

3958

E XP E RI ME N TA L CE LL RE S E A RCH 3 1 2 ( 2 00 6 ) 3 9 4 9 –39 5 8

molecular chaperone by the antitumor antibiotics radicicol and geldanamycin, J. Med. Chem. 42 (1999) 260–266. [43] K.D. Dittmar, K.A. Hutchison, J.K. Owens-Grillo, W.B. Pratt, Reconstitution of the steroid receptor˙hsp90 heterocomplex assembly system of rabbit reticulocyte lysate, J. Biol. Chem. 271 (1996) 12833–12839. [44] S. Chen, V. Prapapanich, R.A. Rimerman, B. Honoré, D.F. Smith, Interactions of p60, a mediator of progesterone receptor assembly, with heat shock proteins Hsp90 and Hsp70, Mol. Endocrinol. 10 (1996) 682–693. [45] B.D. Johnson, R.J. Schumacher, E.D. Ross, D.O. Toft, Hop

modulates hsp70/hsp90 interactions in protein folding, J. Biol. Chem. 273 (1998) 3679–3686. [46] O.O. Odunuga, V.M. Longshaw, G.L. Blatch, Hop: more than an Hsp70/Hsp90 adaptor protein, BioEssays 26 (2004) 1058–1068. [47] D.F. Smith, Tetratricopeptide repeat cochaperones in steroid receptor complexes, Cell Stress Chaperones 9 (2004) 109–121. [48] D.F. Smith, L. Whitesell, S.C. Nair, S. Chen, V. Prapapanich, R.A. Rimerman, Progesterone receptor structure and function altered by geldanamycin, an hsp90-binding agent, Mol. Cell. Biol. 15 (1995) 6804–6812.