2.1.

Experimental Setup

Figure 1(a) shows the experimental setup of the SHG microscope. The expanded beam of a mode-locked Ti:Sa laser with 100-fs pulse duration and 80-MHz repetition rate, model Mira 900F (Coherent, Santa Clara, California) is passed through a half-waveplate and an optical scanning system. This system is composed of two galvomirrors VM500 (GSI Lumunics, Billerica, Massachusetts) and a telescopic lens pair (L1 and L2 in the figure) that pivot the laser beam from the first galvomirror (Gx in the figure) to the second (Gy). This optical configuration was designed to implement the use of mirrors of reduced dimensions (6 mm diameter), enabling an increase of the scanning velocity with respect to conventional setups. Lenses L3 and TL pivot the laser beam in the back focal plane of the microscope objective, Plan Apo 60×/1.4 oil (Nikon, Tokyo, Japan). The SHG signal is collected in the forward direction. Since the excitation beam is focused by a high-numerical-aperture objective, the resulting SHG signal is dominantly emitted off-axis^{5 6} and the collector objective, Fluor 60×/1.0 water (Nikon, Tokyo, Japan), is chosen with an appropriate numerical aperture to collect all the SHG signal. The SHG is detected with a photomultipliertube (PMT) H7710-13 (Hamamatsu, Hamamatsu City, Japan), after blocking both transmitted laser light and concurrent two-photon excited fluorescence with IR-blocking and interference filters. In the experiment, we used a motorized achromatic half-waveplate [λ/2 in Fig. 1(a)] to control the polarization of the excitation laser beam. In the measurements of SHG sensitivity on membrane potential, for each recording, two orthogonally polarized images were acquired and summed to collect signal from all the dye molecules in the membrane.⁶ In the measurements of dye tilt angle, a series of images was acquired by varying the direction of linear polarization of the exciting beam over a range of 180 deg. The images were obtained with excitation wavelength of 880 nm and average laser power of 1 mW on the sample. The images dimensions are 50×50 μm with a resolution of 500×500 pixels. The integration time of each pixel is 5 μs.

Figure 1

Experimental setup: (a) schematic drawing of the optics of the SHG microscope and (b) close-up picture of the objectives (60× oil immersion on the bottom, 60× water immersion at the top) and patch-clamp electrode (shown on the left).

The SHG microscope was integrated with a patch-clamp device AXOPATCH-1C (Axon Instruments, Inc., Foster City, California). The microneedle was positioned between the two objectives [see Fig. 1(b)] by an electronic micromanipulator InjectMan NI2 (Eppendorf, Hamburg, Germany). The SHG microscope and patch-clamp were controlled by two synchronized PCs programmed using LabVIEW software (National Instruments, Austin, Texas).

2.2.

Cells Preparation and Staining for SHG Measurements

First, HEK 293 cells (human embryonic kidney) from clone 5, according to the method reported in Crociani et al.,¹² were transfected with herg1 (human eag related gene) cDNA, codifying for the HERG potassium channel.¹³ Briefly, the cells, cultured on 100-mm petri dishes, were transiently transfected with herg1 cDNA cloned into HindIII/BamHI sites of the pcDNA3.1 vector (Invitrogen) by the calcium phosphate method. Six hours before transfection, the medium was replaced once. The precipitation solution was then added to the cell cultures. The precipitation solution was 400 μl of 2× balanced electrolyte solution (BES)-buffered saline [50-mM BES, 280-mM NaCl, 1.5-mM Na ₂ HPO ₄⋅2H ₂ O (pH 6.96)] plus 400 μl of 0.25-M CaCl ₂ and 36 μg of cDNA construct. The medium was replaced 15 h later and the cells were cultured in Dulbecco’s modified Eagle medium (DMEM) supplemented with 10% fetal bovine serum, 1% glutamine (Euroclone) and geneticin (G418, Gibco) in air containing 5% CO ₂.

Next, SH-SY5Y cells (henceforth abbreviated as SY5Y) were maintained in RPMI 1640 medium (Euroclone) supplemented with 10% fetal calf serum (FCS) (Euroclone), 1% glutamine (Euroclone) in air containing 5% CO ₂, as previously described.¹⁴

Finally Balb/c3T3 murine fibroblasts and simian virus 40 (SV40)-transformed Balb/c3T3 cells (henceforth abbreviated as SV3T3) were grown in DMEM containing M4500 mg/L glucose (DMEM 4500) (GIBCO, Life Technologies, Italy) supplemented with 10% FCS (Boehringer Mannheim, Germany), at 37 °C in a 10% CO ₂ -humidified atmosphere.¹⁵

All cell types were seeded at 1.0×10⁶ cells per 100-mm dishes and propagated every 3 days by incubation with a 0.25% trypsin solution (GIBCO). After trypsinization, cells were seeded into 24×40-mm glass coverslips and cultured as already described. After 2 days the medium was removed and supplied with low potassium solution (see later) containing the dye RH 237 (Molecular Probes, Eugene, Oregen) at a 4-μM concentration and incubated for 10 min to reach equilibration of the dye in the cell membrane. The sample was then mounted on the experimental setup for SHG measurements.

2.3.

Solutions

2.4.

Controls on Effects of the Dye on Cell Electrophysiology

The cells were seeded on 35-mm petri dishes and the transmembrane potential was recorded using the whole cell configuration.¹⁶ Measurements of the membrane potential were performed in the current clamp. Pipette resistances were between 3 and 5 MΩ. Gigaseal resistances were in the range of 1 to 2 GΩ. Input resistances of the cells were in the range 2 to 6 GΩ. For data acquisition and analysis, the pClamp8 and Axoscope software (Axon Instruments, Inc., Foster City, California) and Origin (Microcal Software, Northampton, Massachusetts) were used. Comparisons were conducted between cells stained as described, but using a range of dye concentration from 0 to 12 μM.

3.1.

Sensitivity of the SHG

To explore the SHG sensitivity to V_M in a wide range of cell phenotypes, we chose to carry out our experiments using the following cell lines: (1) the herg-transfected human embryonic kidney cell clone HEK293 (Ref. 12), (2) the neuroblastoma adrenergic clone SY5Y (Ref. 17), (3) the normal Balb/c3T3, and (4) virus-transformed SV3T3 mouse fibroblast.¹⁵ In particular, Balb/c3T3 and SV3T3 were chosen for their suitability to explore differences of SHG sensitivity bound to changes produced by the neoplastic transformation within a properly controlled cellular model. SHG images relative to the preceding cell types, stained with RH 237, are shown in Fig. 2. Combining SHG imaging with the patch-clamp technique makes it possible to clamp the transmembrane potential of the cell and analyze the behavior of the variation of SHG intensity versus potential applied.

Figure 2

SHG images of live cells stained with RH 237: (a) herg-transfected HEK293 epithelial cells (a cluster of cells is shown but the SHG signal arising from the regions of contact between adjacent cells is very low), (b) SY5Y neuroblastoma cell, (c) Balb/c3T3 fibroblast cell, and (d) SV3T3 cell. The images were obtained with excitation wavelength of 880 nm and average laser power of 1 mW on the sample. The images dimensions are 50×50 μm with a resolution of 500×500 pixels. The integration time of each pixel is 5 μs. Each image is the sum of two frames acquired by exciting the sample with two orthogonal polarizations.

Bright-field images of the cells were observed to identify a suitable cell based on its morphology (cells well adhering to the glass and displaying their normal shape, depending on the cell type) and then establish on it the patch-clamp seal. During this operation a PC with LabVIEW code provided diagnostics such as seal resistance.¹⁶ When a seal resistance of the order of 1 GΩ was obtained, the patch was ruptured and V_M was clamped at its resting value (V _rest ). The sample was then excited with laser light and a series of SHG images were acquired with a software that triggered the patch-clamp to switch between V _rest and V _rest −ΔV_M every three images. Because SHG signal arises exclusively from dye molecules partitioned into the cell membrane⁶ (see also Fig. 2), it is possible to reliably quantify the signal by simply summing the greylevel of all the pixels in each image. Figure 3(a) shows the behavior of the SHG intensity (black dots) in a HEK293 cell to which a variation ΔV_M of 100 mV was imposed (dashed line). In all measurements three data points at V _rest were acquired before and after the ΔV_M perturbation. This enabled an evaluation of possible small variations of SHG intensity over time due to flip-flop, photobleaching, and some residual nonequilibrium diffusion of the dye into the cell membrane. When a positive or negative drift was measurable above noise in the signal before and after the ΔV_M perturbation, a correction was applied to the data by linear interpolation.¹⁰ In the experiment shown in the figure, we measured a relative SHG intensity change of 9±1%. The spread within a triplet of points acquired in one condition represents the noise of our measurements, due to different possible sources (background noise, laser power fluctuations, etc.). With the same experimental procedure, we also analyzed the relative SHG intensity changes versus voltage variation (ΔV_M). Figure 3(b) shows a linear dependence of the variation of SHG on the voltage change in HEK293 cells. The slope of a linear fit to the data [dashed line in Fig. 3(b)] is 0.10±0.01%/mV. These measurements were performed on at least five different cells for each ΔV_M, enabling us to sample different cell shapes and possible spatial and temporal heterogeneities in dye incorporation.

Figure 3

SHG measurements of transmembrane potential in herg-transfected HEK293 cells. (a) Recording of the total SHG intensity (left axis) measured on a single HEK293 cell while switching the V_M between −20 and −120 mV using the patch clamp (the trace of voltage versus time is shown as a dashed line and the voltage scale is shown on the right axis). The total SHG signal is the integral of gray levels over the whole image acquired. (b) SHG signal change versus variation in transmembrane potential (ΔV_M). Each point is calculated from the variation in SHG measured as shown in (a). For each ΔV_M, the measurement was repeated at least five times (each on a different cell) and the mean value is reported in the graph. Error bars represent the largest deviation from the mean among all recorded data points. The dashed line is a linear regression through the data, yielding a slope of 0.10±0.01%/mV (R²=0.997).

With the same methodology, we analyzed the SHG intensity changes in SY5Y cells [Fig. 4(a)]. In these cells, the SHG change versus voltage variation was also linear [see Fig. 4(b)] but with a linear coefficient of 0.17±0.01%/mV. These measurements show a significant difference between the two kinds of cells. Hypothesizing that this should be due to the difference in the membrane structure and lipid and protein composition, we measured the SHG sensitivity in other kinds of cells: Balb/c3T3 and SV3T3, where the effects of the transformation on the membrane composition are well studied.¹⁵ In these cells, the change of SHG was measured only for a ΔV_M of 100 mV. The SHG response to this change of potential was 12±1% for Balb/c3T3 and 15±1% for SV3T3. Also in this case, we measured a significant difference of SHG sensitivity. Figure 5 summarizes the results from measurements in all four cell types investigated. It is interesting to notice that each cell type displays a characteristic SHG sensitivity, and the differences between cell types are significant within the precision of our measurements. The possible physical origins of these differences were investigated with measurements of the dye tilt angle in the membrane as described next.

Figure 4

SHG measurements of transmembrane potential in SY5Y cells: (a) the same type of measurement as in Fig. 3(a) is shown here on SY5Y cells and (b) see legend in Fig. 3 (b). The dashed line is a linear regression through the data, yielding a slope of 0.17±0.01%/mV (R²=0.999).

Figure 5

Comparison of SHG sensitivity in four different cell types. Each bar represents the SHG sensitivity measured in the cell type indicated on the x axis. The values reported are the slope (and its uncertainty) of the regressions shown in Figs. 3(b) and 4(b) for HEK293 and SY5Y respectively. Sensitivity values for Balb/c3T3 and SV3T3 were calculated from a set of 20 measurements on each cell type with ΔV_M fixed at 100 mV; the error bars represent the largest deviation from the mean among all recorded data points.

3.2.

Measurements of the Dye Tilt Angle

Moreaux et al.⁸ developed a mathematical model enabling the determination of the dye tilt angle (with respect to the axis perpendicular to the membrane surface) from SHG measurements. In their work, Moreaux et al. used giant unilamellar vesicles (GUVs) and excited SHG with a linearly polarized beam; then, they analyzed the intensity of SHG versus the angle between the normal to the membrane surface and the exciting polarization. Taking advantage of the spherical shape of GUVs, this method enables the sampling of all dye angles from one image. In our experiments, the cells were nonspherical and the different angles between the dipole of the dye and the exciting polarization were sampled by choosing a region on the cell and imaging it with varying exciting polarizations. These measurements were performed on two cell types (HEK293 and SY5Y) clamping the membrane potential alternatively at 0 and −100 mV. A typical result of these measurements is shown in Fig. 6. For the quantitative measurements [shown in Fig. 6(b)] a rectangular region of interest of 1×40 pixels was selected with the long axis perpendicular to the surface of the membrane. This choice was operated to minimize the effects of irregularities in the orientation of the cell membrane over distances of the order of a pixel. In the direction perpendicular to the membrane, on the other hand, a larger number of pixels was chosen to collect signal from the whole membrane and compensate for small fluctuations in the cell position induced by the patch clamp electrode. The intensity data were fit using

Figure 6

Measurements of dye tilt angle. (a) Images of one cell stained with 4-μM RH237 and illuminated with light linearly polarized along the direction indicated by the arrow. The size of the images is 50×50 μm. The gray arrowhead indicates the region of interest selected for quantitative analysis as described in the text. (b) SHG intensity of the selected region of interest versus polarization angle. The dashed line represents a fit to the data using the model described in the text, yielding the following parameter: ξ=0.92±0.01, χ˜²=0.7.

The measurements performed on HEK293 and SY5Y cells at 0 and −100 mV did not show a significant difference in the dye tilt angle: in all cases (with n=7 for each measurement), the values of ξ measured were spread over a range between 1.0 and 0.9 (corresponding to α ranging from 0 to 25 deg). The noise in these measurements arises from several factors, including variability between cells, photobleaching, possible diffusion, and flip-flop of the dye. Note also that the physical and biochemical differences existing between GUVs and living cells could give raise to systematic uncertainties on the fitted parameters. In particular, the cell membrane has an irregular profile (especially, at the subresolution level) and allows the presence of dye molecules in the two opposite lipid layers, whereas unilamellar vesicles do not pose these problems.

However, within the precision of our measurements, there was no observable dependence of the dye tilt angle on the cell type or the voltage. The implications of these observations are discussed further in the following.

3.3.

Electrophysiology Controls

Since the styryl dye is an amphyphylic molecule that intercalates in the bilayer structure of the cell membrane, it could interfere with the normal electrophysiological properties of the cell. An effect of this kind would compromise the most attractive applications of this technique. Therefore, we performed control measurements aimed at assessing whether the dye used in this work alters the membrane physiological properties of the cells studied. In particular, we tested whether V _rest changes in the presence or absence of the dye. Figure 7 shows the experimental results of V _rest measurements obtained in the different cell types at four different dye concentrations, using an extracellular solution with low potassium. The data clearly show that, at the concentration used in the SHG measurements (4 μM), the dye does not modify the V _rest of the cells. We can see that significant changes in V _rest could be detected only at much higher concentrations (12 μM). At the highest concentration, the cells degenerated within 5 to 10 min, whereas in the range 0 to 8 μM the cells were perfectly viable for at least 30 min.

Figure 7

Controls on the effects of RH 237 dye on cell electrophysiology; V _rest is shown for each cell type stained with RH 237 dye at four different concentrations, as indicated. The values shown are mean and standard deviation. Statistics were accumulated over at least 10 samples for each cell line and dye concentration.

Additionally, the changes of transmembrane potential in response to a variation of the external concentration of potassium were compared in the different cell types with 0 and 4 μM dye. Also in these measurements, there was no observable effect induced by the presence of the dye in the membrane (data not shown).