Capillary Electrophoretic Separation of Nuclei Released from Single

Jan 7, 2004 - Wausau, Wisconsin 54401 ... the plasma membrane-bound farnesylated enhanced green ... University of WisconsinsMarathon County...
0 downloads 0 Views 162KB Size
Anal. Chem. 2004, 76, 655-662

Capillary Electrophoretic Separation of Nuclei Released from Single Cells Nilhan Gunasekera,† Karen J. Olson,‡ Karin Musier-Forsyth,§ and Edgar A. Arriaga*,‡,§

Department of Chemistry and Department of Biomedical Engineering, University of Minnesota, 207 Pleasant Street S.E., Minneapolis, Minnesota 55455, and University of WisconsinsMarathon County, 518 South 7th Avenue, Wausau, Wisconsin 54401

We report here the first capillary electrophoresis analysis of intact nuclei released on-column from single cells. Expression of the nuclear-targeted protein nuDsRed2 and the plasma membrane-bound farnesylated enhanced green fluorescent protein in cultured human ∆H2-1 cells allowed fluorescent monitoring of the fate of these subcellular compartments upon injection of a single cell into the separation capillary. On-column treatment with digitonin allowed for the separation of the plasma membrane from the nucleus as indicated by their selective laser-induced fluorescence detection in two separate spectral regions. The data suggest that less than 0.1% of the plasma membrane remains bound to individually detected nuclei. In digitonin-treated cells, the electropherograms consisted of a prominent fluorescent peak attributed to nuDsRed2 localized to the nucleus and a collection of weakly fluorescent events (barely distinguishable from scattering) that seem to indicate additional localization of this protein to other subcellular regions. Taken together, this report points to the feasibility of studying intact organelles released from a single mammalian cell by capillary electrophoresis, which is a prerequisite to understanding the relevance of subcellular heterogeneity in biological systems. Understanding the function of a single cell in its natural environment is a formidable task, complicated by its microscopic dimensions, compartmentalization into organelles, the dynamic nature of cellular processes, the interactions with surrounding cells, and the wide range of chemical information stored within the cell. There are several methodologies that are becoming widely accepted for probing a single cell (or a subcellular region) and providing chemical, temporal, or subcellular distribution information.1 Capillary electrophoresis (CE) is one of the approaches that is compatible with the analysis of the chemical content of single cells.2-12 Unlike flow cytometry or microscopy, * To whom correspondence should be addressed. Phone: 612-624-8024. E-mail: [email protected]. † University of WisconsinsMarathon County. ‡ Department of Biomedical Engineering, University of Minnesota. § Department of Chemistry, University of Minnesota. (1) Stuart, J. N.; Sweedler, J. V. Anal. Bioanal. Chem. 2003, 375, 28-29. (2) Kennedy, R.; Oates, M.; Cooper, B.; Nickerson, B.; Jorgenson, J. Science 1989, 246, 57-63. (3) Chen, G.; Ewing, A. G. Crit. Rev. Neurobiol. 1997, 11, 59-90. (4) Sims, C. E.; Allbritton, N. L. Curr. Opin. Biotechnol. 2003, 14, 23-28. 10.1021/ac034916a CCC: $27.50 Published on Web 01/07/2004

© 2004 American Chemical Society

CE allows for the electrophoretic separation of cellular contents, simplifies and eliminates artifacts associated with bulk analysis such as the activation of unwanted enzymatic activity,13,14 and can provide information on cellular heterogeneity.15 Previous studies have included the total analysis of single cells selected using a microscope and introduced into the capillary using either an electrokinetic or a hydrodynamic injection. The cellular contents are then released by lysing the whole cell inside the separation capillary, and an electric field is applied to perform the CE separation. Sensitive detection such as laser-induced fluorescence (LIF) is then used to monitor the separation of soluble fluorescent components.12,13,16,17 In most of the single-cell CE analyses mentioned above, subcellular information is lost. However, there have been several reports that have adapted single-cell analyses to obtain subcellular distribution information.6,13,18-20 CE has provided spatial information by sampling subcellular portions of large single cells such as neurons.13,21,22 For instance, cytoplasmic regions from a Planorbus corneus neuron have been sampled by etching a capillary tip that then was inserted into the cell22 or by disrupting a neuronal subregion of a rat pheochromocytoma cell using a laser pulse and simultaneously introducing the released contents into a capillary.13 (5) Tong, W.; Yeung, E. S. J. Chromatogr., B: Biomed. Sci. Appl. 1996, 685, 35-40. (6) Tong, W.; Yeung, E. S. J. Chromatogr.. B: Biomed. Sci. Appl. 1997, 689, 321-325. (7) Li, H.; Yeung, E. S. Electrophoresis 2002, 23, 3372-3380. (8) Yeung, E. S. J. Chromatogr., A 1999, 830, 243-262. (9) Cannon, D. M., Jr.; Winograd, N.; Ewing, A. G. Annu. Rev. Biophys. Biomol. Struct. 2000, 29, 239-263. (10) Han, F.; Lillard, S. J. Anal. Biochem. 2002, 302, 136-143. (11) Hu, S.; Jiang, J.; Cook, L. M.; Richards, D. P.; Horlick, L.; Wong, B.; Dovichi, N. J. Electrophoresis 2002, 23, 3136-3142. (12) Krylov, S. N.; Arriaga, E. A.; Chan, N. W.; Dovichi, N. J.; Palcic, M. M. Anal. Biochem. 2000, 283, 133-135. (13) Li, H.; Sims, C. E.; Wu, H. Y.; Allbritton, N. L. Anal. Chem. 2001, 73, 46254631. (14) Krylov, S. N.; Arriaga, E.; Zhang, Z.; Chan, N. W.; Palcic, M. M.; Dovichi, N. J. J. Chromatogr., B: Biomed. Sci. Appl. 2000, 741, 31-35. (15) Xue, Q.; Yeung, E. S. J. Chromatogr., A 1994, 661, 287-295. (16) Chang, H.; Yeung, E. S. Anal. Chem. 1995, 67, 1079-1083. (17) Zabzdyr, J. L.; Lillard, S. J. Anal. Chem. 2001, 73, 5771-5775. (18) Rubakhin, S. S.; Garden, R. W.; Fuller, R. R.; Sweedler, J. V. Nat. Biotechnol. 2000, 18, 172-175. (19) Miao, H.; Rubakhin, S. S.; Sweedler, J. V. Anal. Bioanal. Chem. 2003, 377, 1007-1013. (20) Kristensen, H. K.; Lau, Y. Y.; Ewing, A. G. J. Neurosci. Methods 1994, 51, 183-188. (21) Cruz, L.; Moroz, L.; Gillette, R.; Sweedler, J. V. J. Neurochem. 1997, 69, 110-115. (22) Wallingford, R.; Ewing, A. G. Anal. Chem. 1988, 60, 1972-1975.

Analytical Chemistry, Vol. 76, No. 3, February 1, 2004 655

Subcellular information may also be obtained by using probes that localize to specific subcellular environments. Allbritton’s group has used nuclear-targeted peptides, which are kinase substrates, to investigate the nuclear localization and activity of these enzymes.13 Another report describes the cytoplasmic localization of green fluorescent protein (GFP) using single-cell CE analysis.23 Differential lysis of a single cell inside a capillary prior to CE-LIF separation of the soluble forms of dopamine20 and insulin6 has also been used to investigate the subcellular localization of these bioactive compounds. The plasma membrane and the dopamine storage vesicles lysed on-column at different rates. The early release of cytoplasmic dopamine caused its earlier migration and detection in comparison to dopamine found in storage vesicles. Thus, differences in migration time suggested the relative abundance of this neurotransmitter in two distinct subcellular compartments.20 Similarly, in a second study, cytoplasmic insulin was initially released from an intact βTC3 cell by digitonin permeabilization of the plasma membrane.6 The cytoplasmic insulin was separated from residual insulin released after total lysis of the cell. These studies, based on the analysis of the solubilized contents of single cells, clearly point to the feasibility of developing CE approaches capable of providing combined chemical content and subcellular distribution information. The capability to release intact organelles from a single cell followed by CE analysis is a powerful approach to describe the chemical content of individual organelles and to further the development of other techniques based on individual organelle measurements (e.g., DNA analysis in individual mitochondria). Previously, we have reported the analysis of organelles, including mitochondria, lysosomes, and nuclei, by CE-LIF but relied on bulk fractionation to obtain fractions enriched in these organelles.24-26 Preparation of these fractions required cell disruption based on mechanical homogenization or nitrogen cavitation, approaches that are not compatible with single-cell analysis. Fortunately, there are other disruption strategies, such as electrical disruption27 and partial lysis with nonionic surfactants,28,29 that promise to be more effective for selectively disrupting the cell and releasing its subcellular compartments. In particular, digitonin has been reported to interact with cholesterol compromising the structural integrity of cholesterol-rich membranes,30 such as the plasma membrane.29,31,32 In this report, we describe the use of digitonin to partially disrupt on-column the plasma membrane of single cells for CE analysis of their nuclei. Organelle intactness during the separation is maintained by using an isoosmolar sucrose-based buffer at biological pH (7.4). A nuclear-targeted fluorescent protein, nuD(23) Malek, K.; Khaledi, M. Anal. Biochem. 1999, 268, 262-269. (24) Fuller, K.; Arriaga, E. Anal. Chem. 2003, 75, 2123-2130. (25) Duffy, C.; Fuller, K.; Malvey, M.; O’Kennedy, R.; Arriaga, E. Anal. Chem. 2002, 74, 171-176. (26) Gunasekera, N.; Musier-Forsyth, K.; Arriaga, E. Electrophoresis 2002, 23, 2110-2116. (27) Han, F.; Wang, Y.; Sims, C. E.; Bachman, M.; Chang, R.; Li, G.; Allbritton, N. L. Anal. Chem. 2003, 75, 3688-3696. (28) Ramsby, M. L.; Makowski, G. S.; Khairallah, E. A. Electrophoresis 1994, 15, 265-277. (29) Temkin, R. J.; So, D. Y.; Lea, P. J. Microsc. Res. Tech. 1993, 26, 260-271. (30) Miller, R. Biochim. Biophys. Acta 1984, 774, 151-157. (31) Fiskum, G.; Craig, S. W.; Decker, G. L.; Lehninger, A. L. Proc. Natl. Acad. Sci. U.S.A. 1980, 77, 3430-3434. (32) Cook, G.; Gattone, V.; Evan, A.; Harris, R. Biochim. Biophys. Acta 1983, 763, 357-367.

656

Analytical Chemistry, Vol. 76, No. 3, February 1, 2004

sRed2, transiently expressed in mammalian ∆H2-1 cells was used to monitor the mobility of the intact nucleus. Use of a farnesylated enhanced green fluorescent protein (EGFP-f), which localizes to the plasma membrane, indicated that this membrane is largely separated from the nucleus. This work outlines exciting opportunities for the manipulation of individual organelles and for furthering the analysis of the chemical, spatial, and temporal information contained within the microcosm of a single cell. EXPERIMENTAL SECTION Chemicals. Tris(hydroxymethyl)aminomethane (tris),(N-(2hydroxyethyl)piperazine-N-ethanesulfonic acid) (HEPES), phosphate buffered saline (PBS), Dulbecco’s modified Eagle’s medium, calf serum, poly(L-lysine) (MW, 70 000-150 000), and 0.5% trypsin were purchased from Sigma (St. Louis, MO). Trypan blue stain was purchased from BioWhitaker (Walkersville, MD). Magnesium chloride and sucrose were purchased from Fisher (Fair Lawn, NJ). Fluorescein and ethidium homodimer were purchased from Molecular Probes, Inc. (Eugene, OR). Green fluorescent microspheres (0.75 ( 0.02 µm i.d) were purchased from Polysciences, Inc. (Warrington, PA). Milli-Q filtered water was used to make buffer A (250 mM sucrose, 10 mM HEPES, pH 7.4). Cell Culture. Human ∆H2-1 cells (a kind gift from Dr. Carlos Moraes, Department of Neurology and Cell Biology & Anatomy, University of Miami, FL) were cultured at 37 °C and 5% CO2 by splitting cells 1:6 every 3-4 days in minimum essential medium (MEM) supplemented with 10% fetal bovine serum. Cell Viability Test Using Trypan Blue. ∆H2-1 cells were collected by trypsinization and suspended in buffer A. A stock of digitonin (100 mg/mL) was prepared and serially diluted to obtain 0.1, 0.5, 1.0, 5.0, and 10.0 mg/mL solutions. An aliquot of the cells (200 µL) was incubated with each digitonin solution (200 µL). Samples from each aliquot were collected at different time points (0.75, 1.5, 3, 5 min) and centrifuged (14000g) for 20 s to remove the digitonin. Then, each sample was incubated at a 1:1 ratio with 0.2% Trypan blue at room temperature. The cells that are stained dark blue have permeabilized plasma membranes. The percentage of dark cells in each sample was determined by counting cells using a hemocytometer (Hausser Scientific, Horsham, PA). Fusion Proteins. The nuclear localized fluorescent protein, nuDsRed2, was expressed from a commercially available plasmid, pnuDsRed2 (BD Biosciences, Palo Alto, CA). This plasmid contains the nuclear localization signal from the SV-T40 antigen, and consequently, the nuDsRed2 fusion protein selectively localizes to the nucleus of mammalian cells. The fluorophore in this fusion protein is DsRed2. A fusion protein containing the enhanced green fluorescent protein (EGFP) was used to label the plasma membrane. This protein (EGFP-f) contains a farnesylation signal that directs it to the inner face of the plasma membrane. The plasmid encoding this protein (pEGFP-f) was a kind donation from Dr. Largaespada of the Department of Lab Medicine and Pathology at the University of Minnesota. Transfection. Lipofection with DMRIE-C was used to transfect the cells according to the manufacturer’s instructions.33 Briefly, plasmids are suspended in serum-free medium and incubated with the lipid at room temperature for 30 min. The adherent cells are washed with PBS before the lipid-DNA complex is layered over (33) www.lifetech.com/content/sfs/manuals/10459014.pdf.

the cells. After 4 h, cell culture medium (MEM containing 20% fetal bovine serum) is added. The cells were analyzed starting 24-h posttransfection. The transfection efficiency of pnuDsRed2 in ∆H2-1 cells was determined by fluorescence microscopy to be 34 ( 10% (n ) 4). Prior to CE-LIF analysis, the cells were washed with PBS, treated with 0.5% trypsin in PBS for 5 min, pelleted at 600g, and washed with and resuspended in CE buffer A. Imaging of Cells and Nuclear Species. Fluorescent images were collected using a Nikon TE300 fluorescence microscope using a 60× objective lens (Fryer Co. Inc., Huntley, IL). The dichroic filter cubes blue B-2E/C (excitation 480 ( 15 nm; dichroic 505 nm; emission 535 ( 20 nm), green G-2E/C (excitation 540 ( 12 nm; dichroic 565 nm; emission 620 ( 30 nm), and dual GFP HY (dual excitation and emission) were used to visualize expression of GFP-f alone, nuDsRed2 alone, or both proteins simultaneously, respectively. A CCD camera (model KX85, Apogee Instruments Inc., Auburn, CA) calibrated for the red filter setting was used to collect all images. The camera was cooled to 0 °C, and exposure lasted 40 ms. The collected images were pseudocolored using Image J software.34 Briefly, the image file was changed to a RGB format and the image was colorized to show the appropriate red or green by adjusting the color using the magenta setting. For the duallabeled cells, the edges of the red and green areas were demarcated and colorized separately. The background was adjusted to a similar shade in order to facilitate easy comparison among the images. Single-Cell Injection and Disruption. The procedure for hydrodynamic siphoning injection of a cell into the capillary has been described elsewhere.35,36 Briefly, 5 µL of cell culture is deposited on a microscope slide that is mounted on the stage of a fluorescence microscope. Next, the injection end of an uncoated, 50-µm-i.d., 150-µm-o.d., fused-silica capillary (Polymicro Technologies, Phoenix, AZ) is positioned within several micrometers of the sample with its axis perpendicular to the microscope slide. A fluorescing cell is selected using the fluorescent microscope equipped to detect DsRed2 or EGFP fluorescence, or both. This cell is injected into the capillary by positioning the injection end of the capillary directly above the cell and creating a pressure difference (11 kPa for 1 s) between the injection and detection ends of the capillary. After each injection, fluorescence microscopy was used to confirm that the cell was inside the capillary. Once the cell is injected, a plug of the nonionic detergent digitonin (1 mg/mL in buffer A) is electrokinetically injected at 400 V/s for 5 s. After digitonin addition, the injection end of the capillary is inserted into a vial containing buffer A and incubated for a total of 1.5 min allowing the detergent to permeabilize the plasma membrane. Capillary Electrophoresis with Laser-Induced Fluorescence Detection. After incubating a cell with digitonin inside the capillary, electrophoresis is initiated at 400 V/cm for at least 15 min. At the end of each separation, the capillary was reconditioned by pressure flushing. Briefly, the capillary was first flushed with buffer A for 1 min, methanol for 1 min, and then again with buffer (34) http://rsb.info.nih.gov/ij/index.html. (35) Anderson, A. B.; Gergen, J.; Arriaga, E. A. J. Chromatogr., B: Anal. Technol. Biomed. Life Sci. 2002, 769, 97-106. (36) Krylov, S. N.; Starke, D. A.; Arriaga, E. A.; Zhang, Z.; Chan, N. W.; Palcic, M. M.; Dovichi, N. J. Anal. Chem. 2000, 72, 872-877.

A for 1 min, using a syringe fitted to the capillary through an adapter (Valco Instruments Co., Inc., Houston, TX). The LIF detector and its use have been previously described.37 For excitation, the 488-nm argon ion laser line output was set at 7 mW (model 532-BS-A04, Melles Griot, Carlsbad, CA). Scattering was reduced by using a long-pass filter (495 AELP, Omega Optical Inc., Brattleboro, VT). Fluorescein and GFP-f fluorescence in the 518-552-nm range was selected with a band-pass filter (535DF35, Omega Optical Inc.). Alternatively, the nuDsRed2 fluorescence in the 608-662-nm range was selected with a band-pass filter (635DF55, Omega Optical Inc.). After spectral filtering, fluorescence was detected with a R1471 photomultiplier tube (Hamamatsu, Bridgewater, NJ) biased at 700 V. The output of the photomultiplier tube was digitized at 100 Hz using a NiDaq I/O board (PCI-MIO-16XE-50, National Instruments, Austin, TX), and the data were saved as a binary file. For dual-channel detection, a dichroic beam splitter (550 nm, DRLP, X28, Omega Optical Inc.) oriented at 45° was placed after the long-pass filter to reflect the green region to the 535DF35 filter (green channel) and to transmit the red region to the 635DF55 filter (red channel). Fluorescence at each spectral region was monitored with two independent R1471 photomultiplier tubes biased at 700 V. The detector was aligned using a 10-9 M solution of fluorescein. Briefly, during a continuous electrokinetic flow of fluorescein at 400 V/cm through the capillary, the position of the sheathflow cuvette housing the capillary was adjusted until the signal from fluorescein was maximized. The limit of detection for fluorescein was 2.3 (1000 V photomultiplier tube bias, one channel) and 9.4 zmol (600 V photomultiplier tube bias, two channels). The reproducibility of the optical configuration (one channel) in detecting individual events was 22%, a value that corresponds to the relative standard deviation of the individual fluorescence intensity of fluorescent microspheres. For microscopy imaging of fluorescent species after CE separation, the fluorescence detector was not used. Instead, species eluting from the capillary were deposited onto a polylysinecoated microscope slide. Data Analysis. The procedures for data analysis have been described previously. 37 Using an Igor-Pro (Wavemetrics Lake Oswego, OR) algorithm, signals corresponding to individual organellar events (peak width at half-maximum: 40 ( 1 ms, n ) 111) can be separated from molecules that appreciably diffuse during the separation. A second routine (PickPeaks) was used to select and tabulate those narrow events that had a signal-to-noise ratio larger than five times the standard deviation of the background. For each event, PickPeaks determines the migration time and the signal intensity. Determination of Apparent Electrophoretic Mobility. The total apparent electrophoretic mobility (µT,i) of a detected event (i) can be calculated using the separation voltage (V), capillary length (L), and its migration time (tm) as follows:

µT,i ) L2/(Vtm)

(1)

(37) Duffy, C.; Gafoor, S.; Richards, D. P.; Ahmadzadeh, H.; O’Kennedy, R.; Arriaga, E. Anal. Chem. 2001, 73, 1855-1861.

Analytical Chemistry, Vol. 76, No. 3, February 1, 2004

657

Figure 1. Images of ∆H2-1 cells expressing nuDsRed2. Cells were transfected with pnuDsRed2 as described in the Experimental Section. The bright-field (panel A) and fluorescence (panel B) images were obtained after a posttransfection incubation in the tissue culture medium for 24 h at 5% CO2 and 37 °C. “N” denotes the nucleus of a cell (panel A), and the white arrowheads point to the nucleoli found within the nucleus (panels A, B). The magnification used was 60×; the bar on the bottom left denotes 5 µm.

The electroosmotic flow was estimated at 5.1 × 10-4 cm2/V‚s by a previously established current-monitoring method.26 This value was used as a correction for calculating the apparent electrophoretic mobility for each individual event. RESULTS AND DISCUSSION Nuclear Localization of nuDsRed2 in Single Cells. Figure 1 shows microscopic images of the nuDsRed2 protein expressed in ∆H2-1 cells. The bright-field image (panel A) shows that the nucleus (a representative nonfluorescent nucleus is marked N) occupies a relatively large fraction of the cell volume in these cells. Although fluorescence is dimly detected in the entire nucleus of the successfully transfected cells, it is more intense in the nucleoli (marked by unlabeled arrowheads), indicating that nuDsRed2 preferentially accumulates in this subnuclear compartment (panel B). Such nucleolar localization has been previously observed and apparently results from the nuclear localization signal (SV40 T antigen) used in the design of this fusion protein.38 Comparison of the left and right panels in Figure 1 also shows that only a fraction (34 ( 10%, n ) 4) of the cells are expressing nuDsRed2. These imaging experiments indicate that nuDsRed2 is a good nuclear probe to investigate the fate of the nucleus in single-cell CE-LIF analyses. Electrophoretic Fate of the Plasma Membrane. Digitonin is a nonionic detergent that has been shown to selectively lyse cells by permeabilizing the plasma membrane, while not affecting the intactness of the nuclear membrane.39 It binds to cholesterol in the plasma membrane, forming cholesterol-digitonin complexes that bud out from the membrane, leading to the disruption of the membrane.30 Thus, the selectivity of this disruption method depends on the relative cholesterol composition of the plasma membrane compared to organellar membranes. In general, it is (38) www.clontech.com/archive/JAN02UPD/pdf/DsRed2.pdf. (39) Bronfman, M.; Loyola, G.; Koenig, C. S. Anal. Biochem. 1998, 255, 252256. (40) Gray, G.; Yardley, H. J. Invest. Dermatol. 1975, 64, 423-430. (41) www.probes.com/media/pis/mp03224.pdf. (42) Pomorski, P.; Grebeca, L.; Grebecki, A.; Makuch, R. Biochem. Cell Biol. 2000, 78, 487-494.

658 Analytical Chemistry, Vol. 76, No. 3, February 1, 2004

Table 1. Percent of Permeabilized Human ∆H2-1 Cells upon Digitonin Treatment incubation time (min)

0.1

0.75 1.5 3 5

25 ( 6 37 ( 4 36 ( 13 50 ( 10

digitonin concentration (mg/mL)a 0.5 1.0 5.0 31 ( 10 44 ( 10 51 ( 9 80 ( 17

56 ( 6 78 ( 6 85 ( 14 nd

ndb nd nd nd

10.0 nd nd nd nd

a Numbers indicate the percentage of cells that were determined to be permeabilized using the Trypan blue test, as explained in the Experimental Section. b nd indicates that whole cells could not be detected under these conditions.

accepted that mammalian nuclear membranes have a lower cholesterol composition than the plasma membrane. For example, in human epidermal cells, the nuclear membrane has a lower cholesterol/phospholipid molar ratio (0.22) than the overall cell membrane (0.38).40 Previously, it was shown that the nucleus remains intact while the plasma membrane is highly permeabilized when mammalian hepatocyte cells are incubated with digitonin (0.8 mg/mL for 1 min).32 This work suggests that it is feasible to select conditions for digitonin treatment that would selectively compromise the plasma membrane making it more susceptible to damage when an electric field is applied, without causing any disruptive effect to the nucleus. Using bulk cell cultures, we monitored the percentage of ∆H2-1 cells whose plasma membrane is permeabilized by digitonin using the standard Trypan blue cell viability test (Table 1). Low digitonin dosage (0.1 mg/mL) required up to a 5-min treatment to obtain an acceptable fraction of permeabilized cells (50%). In contrast, whole cells could not be detected in cultures treated with the two highest digitonin concentrations (5.0 and 10.0 mg/mL), indicating total cell lysis under these conditions. A dosage of 1 mg/mL digitonin was selected for our studies; this dosage permeabilized 78% of the cells in 1.5 min. To test the permeabilization conditions selected above inside a CE capillary, a cell was treated with ethidium homodimer,

Figure 2. Fluorescent images of a ∆H2-1 cell, dual transfected with pnuDsRed2 and pEGFP-f. ∆H2-1 cells were dual transfected with pnuDsRed2 and pEGFP-f plasmids. Panels I-III and panels IV and V correspond to images of a cell before and after CE separation, respectively. A cell that successfully expressed both of the fusion proteins was selected for these images. Panel I is a bright-field image. The arrowhead marks the plasma membrane, while the nucleus is marked “N”. Panel I shows a brighter halo on the periphery of the cell attributed to weakly fluorescing EGFP. Panel II shows both EGFP fluorescence (green halo) and nuDsRed2 fluorescence in the nucleus (red) as observed with the dual GFP HY dichroic cube. Panel III shows nuDsRed2 fluorescence alone, observed with the G-2E/C dichroic cube. In panels IV and V, a nuDsRed2 and EGFP-F expressing single cell was injected into a bare fused-silica capillary (39.7 cm), followed by an electrokinetic injection (400 V/cm, 5 s) of digitonin (1 mg/mL). The released cellular components were separated at 400 V/cm in buffer A. The species eluting from the detection end of the capillary were collected onto a microscope slide coated with polylysine and imaged using the G-2E/C dichroic cube (panel IV) and the B-2E/C dichroic cube (panel V). Images were pseudocolored using Image J software. The magnification used was 60×; the bar on the bottom left of panel I denotes 5 µm.

injected into the CE capillary (11 kPa, 1 s), chased by a digitonin plug injection (1 mg/mL), and incubated for 1.5 min. Ethidium homodimer is a fluorescent DNA intercalating dye that can enter the cell only if the plasma membrane is damaged. Thus, only cells with permeabilized plasma membranes will be fluorescent when viewed under a microscope.41 The results showed that the optimized conditions (1 mg/mL digitonin, 1.5-min incubation inside the capillary) indeed lead to fluorescent labeling of cells and permeabilization of the plasma membrane (data not shown). Thus, these conditions were used for the disruption of this membrane during CE separation of the contents of a single cell. Although digitonin permeabilizes the plasma membrane, this treatment alone may not fully separate this membrane from the nucleus. Attachment of membrane buds to the nucleus via nonnuclear material such as remnants of the cytoskeleton has been reported after digitonin treatment.31,42 To monitor the fate of the plasma membrane during and after the CE separation of a single cell, we transfected ∆H2-1 cells with the pEGFP-f plasmid, which results in expression of a farnesylated EGFP that selectively localizes to the inner plasma membrane.43 Similarly, ∆H2-1 cells were doubly transfected with pnuDsRed2 and pEGFP-f plasmids, which express the red fluorescent DsRed2 protein localized in the nucleus and the EGFP-f protein in the plasma membrane. If the plasma membrane is successfully removed from a cell upon digitonin treatment and electrophoresis, then EGFP-f fluorescence will not be detected in the released nucleus. Figure 2 shows microscopy images of a cell expressing both fusion proteins (panels I-III). This cell, which is normally adherent, appears more round (panel I) since it has been removed from the culture flask by trypsinization and has been suspended in buffer A. The border of the relatively large nucleus (N) is hard to discern once these adherent cells are suspended in buffer A. Panel II is a fluorescent image of the same cell obtained with the dual GFP HY dichroic cube that allows for visualization of both EGFP-f and nuDsRed2 fluorescence. As expected, when both nuDsRed2 and EGFP-f are detected, the red fluorescence is in the nucleus and the green fluorescence is present on the plasma membrane. In contrast, only nuDsRed2, which selectively stains the nucleus, is observed when the G-2E/C dichroic cube (panel III) is used.

To determine whether the nucleus has bound plasma membrane after the electrophoretic separation, we introduced a dualtransfected cell into a capillary, treated it with 1 mg/mL digitonin for 1.5 min, applied a high electric filed of 400 V/cm, collected the electrophoresed species from the detection end, and imaged by microscopy. Figure 2, panel IV, shows the red fluorescence from the nucleus, while no green fluorescence attributed to the plasma membrane is detected (panel V). Panels IV and V also indicate that an intact nucleus is capable of electromigrating through the capillary in a 400 V/cm electric field. The observed irregular nuclear surface may be due to a change in osmotic pressure upon loss of the plasma membrane, alterations in the cytoskeleton scaffold that contributes to nuclear shape, damage due to shearing or wall interactions during the CE separation, or the effect of buffer evaporation once the nucleus is deposited onto the microscope slide after CE separation. In future work, direct observation of the nuclear morphology during the CE separation using fast confocal microscopy is expected to help us identify the causes of the observed changes in nuclear morphology. Since fluorescence microscopy may not be sufficiently sensitive (Figure 2, panel V) to detect plasma membrane residues containing EGFP-f that may be still attached to the nucleus after the CE separation, CE-dual LIF detection experiments were also carried out. Using fluorescence microscopy, a single ∆H2-1 cell expressing EGFP-f alone was selected for CE-dual LIF analysis (Figure 3A). The corresponding electropherogram of the undisrupted (i.e., no digitonin treatment) cell consists of a narrow spike (40 ms wide at half-maximum) in the green channel and a corresponding ghost spike of much lower intensity in the red channel (11% relative intensity). The latter is caused by residual scattering, spectral cross-talk (∼2%), and autofluorescence. Autofluorescence is known to interfere with analytical studies of fluorescent proteins.44 Therefore, for dual-transfection CE-LIF studies performed here, only cells that were clearly expressing both fluorescent proteins when observed under the microscope (cf. Figure 2, panels II and III) were chosen for further analysis. Figure 3B shows a typical electropherogram of a doubly transfected ∆H2-1 cell, not treated with digitonin, and expressing both EGFP-f and nuDsRed2. The coincident migration time (366.3 s) for the peaks in both the green and red channels confirm that the cell is

(43) www.clontech.com/techinfo/vectors/vectorsE/pEGFP-F.shtml.

(44) Billinton, N.; Knight, A. W. Anal. Biochem. 2001, 291, 175-197.

Analytical Chemistry, Vol. 76, No. 3, February 1, 2004

659

Figure 3. CE-dual LIF of untreated (panels A and B) and digitonin-treated (panels C and D) single cells. In each figure, the upper trace is the green channel sensitive to EGFP-f fluorescence and the lower trace is the red channel sensitive to nuDsRed2. In (A) and (C), cells were expressing only EGFP-f; in (B) and (D), cells were expressing both EGFP-f and nuDsRed2. The selected cell was injected into a bare fusedsilica capillary (37.0 cm long) followed by an electrokinetic injection (400 V/cm, 5 s) of digitonin (1 mg/mL) (panels C and D only). The released cellular components were separated at 400 V/cm in buffer A. For clarity, upper traces have been offset (0.2 V in (A) and (C); 1.5 V in (B) and (D)).

intact as it travels through the dual-LIF detector. Also, the intensity of the red signal (lower trace) is 80% of the intensity of the EGFP-f signal (upper trace), indicating that the latter is not simply a ghost peak (see Figure 3A). The fate of the plasma membrane and the nucleus after digitonin permeabilization of a single cell within the capillary is shown in Figure 3C and D. The electropherogram of a ∆H2-1 cell expressing EGFP-f treated with digitonin (Figure 3C) suggests that the membrane is not bound to a single cellular component (e.g., the whole cell or the nucleus). Multiple spikes widely spread over the migration time range and two broad bands at 307 and 318 s (upper trace, Figure 3C) indicate that the cell has been permeabilized and the plasma membrane has been disaggregated from the cell. The lower trace in Figure 3C indicates ghost features (i.e., peaks with intensities ranging from 7 to 15% of the corresponding peak in the green channel) that are not relevant to the analysis (see above). Figure 3D shows an electropherogram corresponding to a ∆H2-1 cell expressing both nuDsRed2 and EGFP-f that was treated on-column with digitonin. The data in the green channel (upper trace) suggest that the EGFP-f is distributed among multiple species that likely represent either membrane aggregates (spikes) 660

Analytical Chemistry, Vol. 76, No. 3, February 1, 2004

or a soluble membrane fraction (broad band at 345-360 s). The intense spike at 389.7 s in the red channel (lower trace) is likely due to nuDsRed2 localized to the nucleus (see below). A corresponding spike is detected in the green channel (upper trace), which represents only 5.2% of the total intensity of all the spikes in this channel. However, if the fluorescence intensity of the broad band (345-360 s) attributed to a solubilized membrane fraction is taken into account, then the spike at 389.7 s corresponds to less than 0.1% of the total intensity attributable to EGFP-f. Thus, dual-LIF detection suggests that only a small fraction of plasma membrane remains bound to the nuclear core, which is in agreement with the microscopy results shown in Figure 2, panel V. Other observations related to the dual-transfection studies described above include low reproducibility of the migration time for intact cells (i.e., no digitonin treatment) and the relatively low fraction of digitonin-treated cells that were adequately permeabilized and resulted in electropherograms such as the one shown in Figure 3D. The low reproducibility in migration time of whole cells is quite common and may result from heterogeneity among cells.45 A heterogeneous cell population is expected to be responsible for the efficiency of cell disruption upon digitonin

Figure 4. CE-LIF of single cells. The electropherograms correspond to the analysis of a single nuDsRed2 expressing cell (trace A) and an untransfected cell (trace B). The selected cell was injected into a bare fused-silica capillary (39.7 cm), followed by an electrokinetic injection (400 V/cm, 5 s) of digitonin (1 mg/mL). The released cellular components were separated at 400 V/cm in buffer A. For clarity, the event intensity axis is plotted in the log scale starting at 0.001.

Figure 5. Combined plot of event intensity versus apparent electrophoretic mobility for events detected in six single-cell CE-LIF analyses. Panel A corresponds to events from nuDsRed2 expressing cells (n ) 3), and panel B corresponds to events from unlabeled single cells (n ) 3). The CE conditions were similar to those reported in Figure 4.

treatment (cf. Table 1). Higher concentrations of digitonin may lead to a more effective permeabilization and disruption of the outer plasma membrane, but the risk of damaging other organelles increases.32 Identification of the Nucleus Using nuDsRed2-Transfected Cells. Despite the advantages of dual-LIF detection, this detector configuration is subject to spectral cross-talk, which may result in ambiguous interpretation of the data. Therefore, we chose ∆H2-1 cells and single-channel LIF detection to unambiguously identify the electropherogram peak corresponding to the nucleus. Fluorescence microscopy was used to select a cell expressing nuDsRed2, and CE-LIF analysis was carried out following oncolumn treatment with digitonin. The results are presented in Figure 4A, with the peak intensity axis shown in the log scale to emphasize the presence of low-intensity peaks. In addition to the most intense peak at 312.3 s, which is associated with the nucleus of the cell, 39 peaks of lower intensity were detected. These narrow peaks (peak width, 40 ( 1 ms, n ) 111) do not correspond to molecules or other ionic species that are able to diffuse in solution during the separation, as they have a constant peak width indicative of their traveling time through the laser beam. These events are tabulated in Table 2, which summarizes data obtained for cells expressing nuDsRed2 (Figure 4A), as well as for untransfected control cells (Figure 4B). Only transfected cells have (45) Mehrishi, J. N.; Bauer, J. Electrophoresis 2002, 23, 1984-1994.

Table 2. Number of Events Per Single Cella Analyzed by CE-LIF peak intens (V) >7.9 7.9 V) attributed to the nucleus. On the other hand, Figure 4 and Table 2 show numerous lowintensity events (signal