Index Introduction Materials and Methods Optical set-up Cell culture, transfection, and imaging Results and Discussion Acknowledgements References

Introduction

The Ras GTPases operate as molecular switches that link extracellular stimuli to a diverse range of biological outcomes. It is now accepted that three Ras isoforms, H, N, and K, exhibit biologically significant differences, despite their high degree of homology (Bar-sagi 2001; Shields et al., 2000; Prior and Hancock, 2001; Rodriguez-Viciana and McCormick, 2006; Sternberg and Schmid, 1999). The main sequence divergence between the Ras isoforms is confined to the C-terminal hypervariable region (HVR) (Hancock, 2003). N-Ras and H-Ras are palmitoylated in their HVRs, but K-Ras is not. It is likely that these regions target the three isoforms to different microdomains within a cell. Most studies on Ras activation have been performed employing constitutively active isoforms. For example, the oncogenic versions of H-Ras and K-Ras were introduced into COS cells, where they displayed distinct effector activation profiles: Raf-1 was more effectively activated by K-Ras, while phosphatidylinositol 3-kinase (PI3K) was preferentially activated by H-Ras (Yan et al., 1998). Such differences in the consequences of H- and K-Ras activation have been assumed to result from the different localization of their isoforms. In contrast to this focus on the downstream effects of Ras activation, we were interested in its upstream regulation; that is, when and where H- and K-Ras isoforms are activated in response to extracellular stimuli.

Recent biochemical, fluorescence and electron microscopic observations detected an association of H-Ras with lipid rafts, and localization of K-Ras to the disordered plasma membrane (Bar-sagi, 2001; Shields et al., 2000; Prior and Hancock et al., 2001; Sternberg and Schmid, 1999; Prior et al., 2003). In fact, the dependence of H-Ras, but not K-Ras, on lipid raft association was demonstrated using dominant-negative caveolin (Roy et al., 1999). This differential compartmentalization of H- and K-Ras proteins is, however, still controversial, because immunolocalization of endogenous Ras proteins has not been studied thoroughly, and the observations are sensitive to the techniques used. In addition, a recent study has demonstrated that the endomembrane system, as well as the plasma membrane, is an important platform from which H-Ras regulates signaling (Chiu et al., 2002). More recently, rapid shuttling between the plasma membrane and the Golgi apparatus, through a constitutive deacylation/reacylation cycle, has been demonstrated for the palmitoylated Ras isoforms H-Ras and N-Ras (Rocks et al., 2005). In the present study, we have used Raichu-Ras (Mochizuki et al., 2001), a fluorescent indicator for the activation of Ras. The indicator consists of Ras, the Ras-binding domain (RBD) of Raf, cyan and yellow mutants of Green Fluorescent Protein (CFP and YFP, respectively), and can sense the local balance between the activities of guanine nucleotide exchange factors (GEFs) and GTPase activating proteins (GAPs).

Materials and Methods

Optical set-up

An inverted fluorescence microscope (Olympus, IX70) was modified by placing a digital micromirror device (DMD) at the location of the field iris in the excitation optical path (schematic diagram is shown in Fig. 1) (Fukano and Miyawaki, 2003). The DMD chip and controlling board were obtained from a commercially available digital light-processing (DLP) projector (PLUS, U3-1080). The DMD chip consisted of 1024×768 micromirrors. Each mirror measured 16 μm×16 μm, aligned vertically and horizontally with 1 μm gaps between them, to achieve a pixel size of 17 μm×17 μm. Because each micromirror could be set electronically to tilt at ±10 degrees with respect to the surface of the DMD, the device acted, with regard to intensity, as a reflection-type spatial light modulator. When the micromirrors were set to the +10 degree state, light from a lamp was directed to the microscope. When the micromirrors were set to the –10 degree state, light was deflected outside the microscope. We created software to model the light pattern projected onto a sample by the DMD. As switching between the two mirror states required 15 μsec, the intensity of the light was also controlled by an eight-bit pulse-width modulation. A plate of frosted glass (F) and a lens (L1) were placed to illuminate the DMD chip by scattering and diverging light (Fig. 1A). This configuration eliminated the diffraction caused by the periodic arrangement of micromirrors. The fluorescence emitted from the sample was captured by a 16-bit cooled CCD camera (Roper Scientific, Cascade 650) and transferred to the computer. The sample could be moved axially on a mechanical stage (Ludl Electronics Products, MAC2000) in minimum steps of 0.1 μm. Image capture, DMD pattern design, shutter control, and sample movement were coordinated by the computer software (Universal Imaging, MetaMorph 4.6). These experiments used a filter set comprising 440DF20 as the excitation filter, 455DRLP as the dichroic mirror, and 480AF30 and 535AF45 as the emission filters.

View Details

Fig. 1. Wide-field fluorescence microscopy using a DMD to acquire optically sectioned images using fringe projection. (A) Schematics of the optical configuration. F, frosted-glass plate; L, lens; M, mirror; DMD, digital micromirror device; EX, excitation filter; DM, dichroic mirror; OL, objective lens; EM, emission filter; PC, computer. (B) Excitation light was modulated sinusoidally by the DMD and projected onto the sample. The fluorescence signal in the focal plane was also modulated sinusoidally, whereas the signal out of the focal plane was uniform. To extract an optically sectioned image, three images with lateral phase-shifts of 0, π/2, and π were captured sequentially. From the set of three images, it was possible to obtain optically sectioned (OS) and wide-field images that respectively correspond to the amplitude and bias components in the schematic.

Cell culture, transfection, and imaging

COS cells were transfected with pRaichu-tH or pRaichu-tK using the Lipofectin method (Gibco BRL). For disruption of caveolae/rafts, cells were incubated in Hank’s Balanced Salt Solution (HBSS) containing 10 mM methyl-β-cyclodextrin (MβCD) (Sigma-Aldrich) for 30 min at 37°C (Prior et al., 2003; Roy et al., 1999). To block trafficking of Raichu proteins, cells were treated with 2 μg/ml brefeldin A (BFA, Sigma-Aldrich) for 20 h (Apolloni et al., 2000).

Results and Discussion

To understand more precisely the spatiotemporal patterns of H- and K-Ras activation within a whole cell, we localized Raichu-Ras (Mochizuki et al., 2001) conjugated to H-Ras- and K-Ras-specific hypervariable regions (HVRs). The hypervariable regions consist of two domains: a linker domain and a membrane-targeting domain. The latter comprises the C-terminal CAAX motif, common to all three Ras proteins, plus two palmitoylated cysteine residues in H-Ras, or a polybasic domain of six contiguous lysines in K-Ras. According to previous reports (Bar-sagi, 2001; Shields et al., 2000; Prior and Hancock et al., 2001; Sternberg and Schmid, 1999; Prior et al., 2003), K-Ras resides in the disordered, non-raft plasma membrane, irrespective of the presence of the linker domain or its activation state. We used the entire hypervariable region of K-Ras to generate Raichu-tK (Fig. 2A). H-Ras, on the other hand, is targeted to caveolae/rafts, but in the presence of its linker domain, activated H-Ras moves to the disordered plasma membrane (Prior et al., 2001). In order to localize the indicator to caveolae/rafts constitutively, we used only the membrane-targeting domain of H-Ras to create Raichu-tH (Fig. 2C). Thus, Raichu-tH and Raichu-tK were designed to detect Ras activation in caveolae/rafts and non-raft domains of the plasma membrane, respectively. Treatment of cells with 2 μg/ml brefeldin A (BFA) abolished the plasma membrane targeting of Raichu-tH, but not that of Raichu-tK (Fig. 3). Because H-Ras, but not K-Ras, is delivered from the ER to the plasma membrane via the Golgi apparatus in a BFA-sensitive pathway, this result showed that the two indicators follow the native pathways of their respective Ras isoforms before reaching the plasma membrane (Apolloni et al., 2000). Since Raichu-tH is also localized strongly to Golgi membranes, it can, in addition, be used to compare Ras activation between the plasma membrane and endomembrane system.

View Details

Fig. 2. Ras activation within caveolae/rafts (Raichu-tH) and non-raft domains (Raichu-tK) of the plasma membrane, and within the endomembrane system (Raichu-tH). (A and C) Overall domain architecture of Raichu-tK and Raichu-tH, respectively. Raf RBD: Ras binding domain of Raf. The amino acid sequences (single-letter code) of the hypervariable regions at the C termini are shown in bold. The cysteine residue in the CAAX motif is indicated by *. In (A), the linker domain of K-Ras is drawn in green, and the polybasic domain comprising six contiguous lysines in K-Ras is underlined. In (C), palmitoylated cysteine residues in H-Ras are indicated in red. (B and D) Optically sectioned fluorescence images of resting COS cells expressing Raichu-tK (B) and Raichu-tH (D) at four different z-levels, and the corresponding wide-field images. Each wide-field image is accompanied by cross-sectional images at the horizontal and vertical lines drawn on the wide-field image. Scale bar, 10 μm for the wide-field images and 4 μm in the cross-sectional images.

View Details

Fig. 3. Localization of Raichu-tK and Raichu-tH in COS cells treated with brefeldin A (BFA) for 20 h. Stacked images along the z-axis are shown, and cross-sectional images at the horizontal and vertical lines indicated. Scale bar, 10 μm for the stacked images and 4 μm for the cross-sectioned images.

Many fluorescent indicator systems have been developed by coupling Green Fluorescent Protein (GFP)-based fluorescence resonance energy transfer (FRET) technology with intracellular signaling cascades (Miyawaki, 2003); most of these systems use CFP and YFP as the FRET donor and acceptor, respectively. To detect more precisely the subcellular localization of the fluorescent signals, it is necessary to acquire optically sectioned images, which is usually done by laser scanning confocal microscopy (LSCM). The use of lasers, however, limits the possible excitation wavelengths, and requires both special instrumentation and expertise. Selective excitation of CFP relative to YFP requires a helium/cadmium-ion (442 nm), Ar⁺ (458 nm), or Kr⁺ (413 nm) line, or a diode-pumped solid-state laser (430~440 nm), none of which is commonly found yet in cell biology laboratories. To monitor the FRET signals of Raichu-Ras with sufficiently high resolution, we used our novel microscopy system. This incorporates a fringe projection technique (Neil et al., 1997), which in the past has been implemented using a wide-field fluorescence microscope with a grating. However, our new system contains a computer-controlled digital micromirror device (DMD) (Fig. 1A), which increases greatly the flexibility of the technique. One advantage of this system over the grating-based one is that the optical sectioning strength is tunable (Fukano and Miyawaki, 2003). To obtain optically sectioned images, we employ a three-step phase-shift method (Kreis, 1996) using a phase step of 90 degrees (Fig. 1B). Samples are illuminated with a sinusoidal waveform. Three laterally phase-shifted images are captured sequentially, and the optically sectioned (OS) image and wide-field image are calculated and displayed simultaneously.

Optically sectioned fluorescence images were taken at different z-levels in COS cells expressing Raichu-tK; four representative images are shown in Fig. 2B. The right panel shows a wide-field image. In addition, two cross-sectional images at the horizontal and vertical lines indicated were constructed from the z-stacked images, and illustrate clearly that Raichu-tK was targeted to the plasma membrane. On the other hand, Fig. 2D shows that Raichu-tH resided in both the plasma membrane and the Golgi membranes. Similar pictures for Raichu-tK and Raichu-tH were obtained using LSCM; illumination of Raichu-tK at 488 nm using an Ar⁺ laser in the point-scan mode confirmed its localization to the plasma membrane (Fig. 4).

View Details

Fig. 4. Fluorescence images of COS cells expressing Raichu-tK (A) and Raichu-tH (B) acquired by LSCM (laser scanning confocal microscopy) using a 488 nm laser line. For each, confocal images at four different z-levels and a z-stacked image are shown. The stacked image is accompanied by cross-sectional images at the horizontal and vertical lines indicated. Scale bar, 10 μm for the Z-stacked images and 4 μm for the cross-sectional images.

EGF-induced changes in indicator signal were clearly detected in images optically sectioned using our wide-field technique. A series of optically sectioned ratio images for K-Ras activation was acquired. They are presented in the intensity-modified mode (Fig. 5A, top) together with the wide-field ratio images (Fig. 5A, bottom). While the optically sectioned images map K-Ras activation on membranes in detail, the wide-field images show global development of the activation inside cells. Using the fringe projection technique, these two kinds of data can be obtained simultaneously. To observe H-Ras activation on the plasma membrane and the endomembranes, optically sectioned ratio images were taken at different z-levels in COS cells expressing Raichu-tH (Fig. 5B). These observations have revealed differences in the spatio-temporal patterns of development of Raichu signals of K-Ras vs. H-Ras and the plasma membrane vs. endomembranes. A typical time course of the Raichu-tK signal is shown in Fig. 5C (green open squares). After one or two minutes of stimulation with 50 ng/ml EGF, the emission ratio began to increase. This latency time before the EGF-induced increase in the emission ratio of Raichu-tK was observed uniformly in 19 cells (1.5±0.4 min). On the other hand, the Raichu-tH signal on the plasma membrane changed much sooner after EGF application (Fig. 5C, red open circles), with an average latency time of 0.26±0.06 min (n=17 cells). The immediate increase in the signal for Ras activation in caveolae/rafts would suggest the preferential distribution of upstream regulators of H-Ras, such as the EGF receptor (Furuchi and Anderson, 1998) in these specialized microdomains; however, the distribution of these signaling molecules across caveolae, lipid rafts, and disordered plasma-membrane microdomains has not been resolved conclusively (White and Anderson, 2001). As reported previously (Chiu et al., 2002), the Raichu-tH signal on the endomembranes changed slowly, reaching a maximum in about 30 min (Fig. 5C, blue closed circles).

View Details

Fig. 5. Time-lapse imaging of activation of K-Ras and H-Ras using OSFM (optically sectioned fluorescence microscopy). (A) Optically sectioned ratio images (top) and wide-field ratio images (bottom) for signals of Raichu-tK after stimulation with 50 ng/ml EGF. Scale bar, 10 μm. (B) Optically sectioned ratio images for signals of Raichu-tH on the plasma membrane (top) and on the endomembranes (bottom) after stimulation with 50 ng/ml EGF. Scale bar, 10 μm. (C and D) Time courses of the EGF-induced increase in the emission ratio of Raichu-tK at several sites in cells. R/R0 is the ratio between the observed value [R] and the pre-stimulus value [R0]. 50 ng/ml EGF was applied at time=0 (indicated by vertical arrows). Raichu-tK signal, green open squares; Raichu-tH signal on the plasma membrane, red open circles; Raichu-tH on the emdomembranes, blue closed circles. (D) Cells pretreated with methyl-β-cyclodextrin (MβCD). inset, An optically sectioned image of Raichu-tH in the MβCD-treated cell. No substantial changes were observed in the distribution of fluorescence signals. Scale bar, 10 μm.

Next, we pretreated cells with methyl-β-cyclodextrin (MβCD) to disrupt caveolae/rafts (Prior et al., 2003; Roy et al., 1999). While the signal of Raichu-tH on the plasma membrane was greatly reduced, the Raichu-tK signal was less attenuated (Fig. 5D). These results indicate that the onset of the plasma membrane Raichu-tH signal is more dependent on the presence of caveolae/rafts. It is thus possible that the EGF receptor resides in caveolae/rafts, and, under these conditions, the Ras proteins in the microdomains are preferentially activated.

There has been no firm consensus on the properties of lipid rafts, and the most popular operational definition of rafts, their association with detergent-resistant membranes (DRMs), is now regarded as inadequate. On the other hand, our study has revealed that the Ras GTPases are activated in spatially and temporally unique patterns across caveolae/rafts and non-raft microdomains in response to stimulation with growth factors.

Acknowledgements

This work was partly supported by grants from CREST of JST (Japan Science and Technology), the Japanese Ministry of Education, Science and Technology, and Special Coordination Fund for the promotion of Ministry of Education, Culture, Sports, Science and Technology, the Japanese Government, NEDO (the New Energy and Industrial Technology Development Organization), HFSP (the Human Frontier Science Program), and RIKEN Strategic Research Program.