Effects of deuterium oxide on cell growth and vesicle speed in RBL-2H3 cells

Cells

The RBL-2H3 cell line was purchased from ATCC. RBL-2H3 cells were maintained in minimal essential medium (MEM) supplemented with 10% fetal bovine serum (FBS), 1% Penicillin Streptomycin (Pen-Strep), and 1% L-glutamine (L-glut). For the experiment cells were grown under 5% CO₂ atmosphere in 6-well dishes for up to five days with media made from MEM powder containing 0 moles/L or 15 moles/L D₂O. Anti-DNP IgE was purchased from Sigma-Aldrich. Before microscopy, cells were IgE primed by incubation with 0.5 mg/mL of IgE overnight. After the addition of cells to the microscope imaging chamber, cells were stimulated with DNP-conjugated BSA (DNP₂₅-BSA) at 2 µg/mL for up to 30 min. For colchicine drug treatment, cells were treated with 100 µM colchicine for 45 min (Smith et al., 2003). Trypan blue stained cells were used to count viable (alive) and dead cells on a hemacytometer. The presented studies on this cell line have been approved by the University of Colorado at Colorado Springs Institutional Review Board, approval number IBC 13-001.

Supported lipid bilayers

Prior to use, microscope glass cover slips were cleaned of organic residues with a mixture of sulfuric acid and hydrogen peroxide (piranha solution). Supported lipid bilayers (Carroll-Portillo et al., 2010; Spendier et al., 2010) were made by spontaneous liposome fusion (Werner et al., 2009). Lipids, obtained from Avanti, were dissolved in chloroform, dried under air flow, and then placed under a vacuum for 1 h to remove traces of oxygen. The lipid film was then suspended in PBS to 1.3 mM and sonicated for 5 min using a probe sonicator in an ice bath. Laterally mobile bilayers were formed from 1-palmitoyl-2-oleoyl-sn-glycero-3-phosphocholine (POPC) and 5 mol% N-dinitrophenyl-aminocaproyl phosphatidylethanolamine (DNP-Cap PE) on piranha-cleaned cover glass for 15 min on a slide warmer at 37 °C. Each bilayer coated coverslip was kept water immersed during transfer to the imaging chamber. Prior to adding cells to the bilayer, the chamber was flushed with Hanks’s buffered saline solution.

DiI labeling

For fluorescent imaging of lipid membrane vesicles, approximately 3 × 10⁵ cells were suspended in 1 mL of Hanks’s buffered saline solution and incubated with 2 µL Vibrant DiI (Life Technologies, Grans Islan, NY) at 2 mM concentration for 2 min. After incubation, the cells were washed at least three times in Hanks’s buffered saline solution.

Flow cytometry measurements

RBL-2H3 cell cycle was analyzed by quantification of DNA content with a Beckman Coulter Cytomics FC 500 MPL Flow Cytometer using forward scatter (FS), side scatter (SS), and fluorescence 3 (FL3, 620 nm Band Pass Filter) channels. The brightness of the propidium iodide (PI) fluorescent dye was used to evaluate DNA content. A 488 nm laser beam was focused onto the flowing stream at a fixed point, illuminating the cells as they passed through. To prepare cell samples, cells were first trypsinated and then washed with MEM of the appropriate D₂O concentration. Cells were fixed and PI-stained by re-suspending cell pellets in 250 µL of MEM, 250 µL of ethanol and 100 µL of PI at 0.1 g/L. PI-stained cell samples were refrigerated overnight and analyzed the following day. Samples were run 100 s or until the count exceeded 10,000 cells. For analysis, the single cell population was first gated using FS vs. SS to remove obvious cell debris and non-cellular elements. This gate was then applied to FS vs. FL3 to reduce doublets, i.e., two cells stuck together that may register as a single cell. These two gates were then combined and applied to give the final PI histogram plots.

Fluorescent imaging

DiI stained RBL-2H3 cells were imaged using a total internal reflection fluorescence (TIRF) microscope. Cell samples were maintained at 37 °C using an objective heater. In TIRF microscopy, cells were allowed to settle onto a fluid lipid POPC bilayer with 5 mol% DNP lipid or on piranha-cleaned glass under gravity. Objective-based TIRF microscopy was performed with an S-TIRF module (Spectral Applied Research, Canada) attached to a Leica DMI3000 B inverted microscope with a 100 × and 1.47 N.A. oil immersion objective using a 561 nm laser (Coherent Inc.) excitation. A 1.5 × lens was also added to the excitation beam path resulting in a final magnification of 150 ×. The penetration depth of the evanescent wave for 561 nm excitation was calculated to be 200 nm. A 600/50 nm single-band bandpass filter (Chroma) was used to collect fluorescence. TIRF images were collected with an EMCCD camera (Evolve Delta; Photometrics) operated by Micro-Manager (Stuurman, Amodaj & Vale, 2007).

Single-vesicle tracking

Image processing was conducted in MATLAB (MathWorks, Inc., Natick, MA), in conjunction with the DIPImage image processing toolbox (Delft University of Technology). Image backgrounds were averaged and subtracted to reduce noise. Fluorescent particles of size within the microscope resolution limit (Gaussian variance of 200 nm) were identified as vesicles. Vesicle coordinates were identified in each frame by a direct Gaussian fit algorithm and a cost function was employed to link coordinates together into trajectories (Andrews et al., 2008). Trajectories were histogrammed by average hop speed for each vesicle trajectory. Hop speed was calculated as a function of distance/frame, where each frame represents a 20 ms time interval. Any distinct fluorescent clusters of low speed, defined as less than 10⁻² µm/s, were removed to separate vesicles from stationary cellular structures. The localization uncertainty for individual trajectory positions was within 100 nm.

The effect of deuterium oxide on RBL-2H3 cell growth and viability

It is known that D₂O can reduce cell growth in human pancreatic tumor cells (Hartmann et al., 2005). To confirm that a similar trend is observed in RBL-2H3 cells, cell growth and cell viability was monitored over a period of 5 days. RBL-2H3 cells were cultured in 6 well plates in cell media containing 0 moles/L D₂O or 15 moles/L D₂O. The plates were seeded at the same time to ensure similar initial cell seeding concentration for each well. Cells were allowed to divide for 24 h before the first measurement. Figure 1A shows that when cells were cultured in 15 moles/L D₂O (open circles) a 7-fold decrease in RBL-2H3 cell growth was observed after five days compared to cells cultured in 0 moles/L D₂O (closed circles). Cell viability also dropped by 20% for cells cultured in D₂O (open circles) as shown in Fig. 1B. These results are qualitatively consistent with studies on human pancreatic tumor cells (Hartmann et al., 2005).

The observed reduction in cell growth and viability may be caused by differences in pH between the two growing conditions (Lardner, 2001). For cell media containing 0 moles/L D₂O and 15 moles/L D₂O, the average pH was 7.58 ± 0.29 and 7.48 ± 0.12, respectively during a 5 day period. The error is represented by the standard deviation. No statistically significant difference was observed and therefore we conclude that changes in pH did not cause the observed decrease in cell growth.

Finally, to start investigating the effect of D₂O on RBL-2H3 cell cycle, flow cytometry measurements were performed to monitor the RBL-2H3 cell cycle by measuring DNA content (Darzynkiewicz, Robinson & Crissman, 1994). The cell cycle has four distinct phases that can be recognized in a proliferating cell population: the G1-(growth phase), S-(DNA synthesis phase), G2-(growth phase and preparation for mitosis) and M-phase (mitosis). Here, PI was used as a DNA probe to investigate the cell cycle, in particular the G0–G1-, S- and G2–M-phases. Each of these phases can be identified by DNA content. G0–G1 is diploid and has a normal complement of DNA. In S-phase, DNA synthesis occurs and in the G2–M-phase, twice the amount of DNA is found in the cells than in the G0–G1 phase. Therefore, a typical DNA content frequency histogram of proliferating (multiplying) cells shows two clearly separated peaks, a G0–G1-peak and a lower G2–M-peak separated by the cell subpopulation in the S-phase.

Figure 2 depicts DNA content frequency histograms for RBL-2H3 cells in untreated cultures (Fig. 2A) and in cultures treated with D₂O (Fig. 2B). The presented data was smoothed using a moving average and the area underneath the curve was normalized to unity. For 0 moles/L D₂O treated cells, the G2–M-phase (black arrow) is recognizable in one-day (black line in Fig. 2A) and four-day (red line in Fig. 2A) old cultures. In other 0 moles/L D₂O cell cultures (data not shown) the G2–M peak was more pronounced. In one-day old D₂O treated cultures, the G2–M-phase can also be identified (black line in Fig. 2B). The number of cells entering the G2–M-phase in four-day old D₂O cultures decreases, as shown by a weakened G2–M-phase and an increased height of the G0–G1 peak (red line in Fig. 2B). The loss of the G2–M peak reflects that RBL-2H3 cells have not doubled their DNA content. This loss of DNA content may be caused by cells failing to enter mitosis and/or is a sign that cells are in the early stages of apoptosis. Both mechanisms were reported previously in human pancreatic tumor cells in which D₂O induced apoptosis in PANC-1 and AsPC-1 cells and arrested PANC-1 and BxPC-3 cells in the G2–M-phase of the cell cycle (Hartmann et al., 2005). We also note that the G0–G1 peaks for both D₂O treated and untreated cultures experience a shift of approximately the same magnitude to the left within four days. Such shifts can be attributed to daily fluctuations in the flow cytometry excitation and detection system.

DiI-stained RBL-2H3 cells for vesicle tracking

As pointed out above, one possible mechanism for failure to double DNA is cell arrest near the G2–M-phase of the cell cycle. G2–M-cell cycle arrest can be associated with problems in the mitotic spindle structure (Holy, 2002). The mitotic spindle is a structure composed of microtubules which segregates chromosomes into the daughter cells during mitosis. Besides structural support of microtubules, they also act as highways within a cell for trafficking a wide variety of cargo by molecular motors. Here we focus on secretory granules as cargo that are transported by molecular motors (kinesins) towards the cell plasma membrane for exocytosis during mast cell activation via IgE-FcϵRI crosslinking (Smith et al., 2003). It has been reported previously that stable microtubules result in increased Kinesin-1 motor speeds, since motors that move along dynamic microtubules could rapidly run off the end of the microtubule (Cai et al., 2009). Experiments also showed that D₂O has a microtubule stabilization effect in microtubule gliding assays (Panda et al., 2000) and on isolated tubulin proteins (Houston et al., 1974; Das et al., 2008). Therefore vesicles that are transported by molecular motors in a microtubule-dependent manner should experience different transport speeds in untreated and D₂O treated cultures. Specifically, the mean vesicle hop speed of tracked secretory vesicles in D₂O treated cultures is expected to increase. To mark these secretory vesicles for the studies presented here, diI was used as fluorescent lipophilic membrane stain (Anantharam et al., 2010). DiI is weakly fluorescent until incorporated into lipid membranes. It labels all cell membrane barriers, including the cell plasma membrane and membrane-bound organelles such as the nucleus, endoplasmic reticulum, golgi apparatus, lysosomes, endosomes, and mitochondria (see Fig. S1).

TIRF microscopy was employed to reduce background fluorescence from the cell plasma membrane and fluorescence from membrane-bound organelles. In TIRF microscopy, the excitation laser beam is totally internally reflected at the glass-water interface and only an evanescent wave traveling parallel to the interface penetrates into the sample. The evanescent wave decays rapidly with a perpendicular distance from the interface, and selectively excites the sample within a distance of about 200 nm from the surface. Besides reduction of background fluorescence, the advantage of using TIRF microscopy is selective excitation of vesicles that are within the evanescent field. In mast cells, membrane-bound granules are approximately 300–400 nm in diameter (Martynova et al., 2005). Therefore in the experiments outlined below, vesicles were identified as fluorescent particles of size within the microscope resolution limit. Since secretory granules are transported in a direct manner towards the cell plasma membrane for exocytosis, TIRF microscopy makes it possible to track these vesicles.

To test the ability of diI to label secretory vesicles, diI-stained, IgE-loaded RBL-2H3 cells cultured in 0 moles/L D₂O were allowed to settle onto glass over a 5 min period, and then stimulated with multivalent DNP₂₅-BSA for 30 min. Cells were imaged before the addition of DNP₂₅-BSA and during the incubation period using TIRF microscopy at 20 frames/s. Movie S1 clearly shows sustained vesicle motion 26 min after DNP₂₅-BSA stimulation (right panel in top row of Movie S1). Some vesicle motion, albeit slower, was also observed in unstimulated RBL-2H3 cells which is consistent with previous work by Smith et al. (2003). To shorten cell-settling time to less than 1 min and achieve good cell-substrate contact required for TIRF imaging, cells were allowed to settle onto a supported lipid bilayer with 5 mol% DNP-Cap PE for the single-vesicle tracking experiments outlined below. One of the authors showed previously that RBL-2H3 cells in contact with liganded-bilayers composed of 5 mol% DNP-Cap PE exhibit degranulation levels similar to un-stimulated cells (Carroll-Portillo et al., 2010).

Effect of D₂O on vesicle motion in RBL-2H3 cells

As pointed out above, diI labels all membrane barriers including vesicles that are not transported on microtubules. However, if a large number of vesicles in DNP₂₅-BSA activated cells are tracked within the cell-substrate contact area, a significant population of these vesicles is expected to be secretory, i.e., vesicles that are transported on microtubules for exocytosis. Therefore, over 1,500 individual vesicles were tracked using TIRF microscopy. For each vesicle-trajectory, the mean hop speed was calculated. The average of these mean vesicle hop speeds for cells cultured in 0 moles/L D₂O was 3.6 ± 0.1 µm/s. This average speed was significantly slower than the average of 8.5 ± 0.4 µm/s computed from the mean vesicle hop speeds for cells cultured in 15 moles/L D₂O. This difference was statistically significant as determined by the Student’s t-test with a p-value of less than 0.05.

To test whether some of the diI-labeled vesicles tracked with TIRF microscopy are secretory, a known microtubule destabilizing drug called colchicine was used which is expected to slow down secretory vesicle transport (Smith et al., 2003). For cells cultured in 15 moles/L D₂O the average of mean vesicle hop speeds for colchicine treated cells decreased significantly to 2.4 ± 0.3 µm/s as expected for microtubule-dependent vesicle transport. However, for cells cultured in 0 moles/L D₂O the average of all mean hop speeds for colchicine treated cells did not significantly change and was measured to be 4.2 ± 0.3 µm/s. Possible reasons for not measuring a decrease in vesicle hop speed for colchicine treated cells cultured in 0 moles/L D₂O are as follows. Firstly, the camera projects a three-dimensional path of a moving object onto a two-dimensional plane. Therefore, the component of vesicle velocity parallel to the substrate is not the full vesicle velocity vector. If only a small parallel component of an outward (from cell center towards cell-substrate contact zone) directed vesicle velocity vector is seen, there is no reason to insist that the parallel component by itself be large enough to generate a statistically detectable difference. 15 moles/L D₂O treated cells appear to have larger vesicle velocities than 0 moles/L D₂O treated cultures. Cells with larger overall vesicle velocities also have larger parallel components, and those larger parallel components show a statistically significant change with the application of colchicine. Secondly, the reported average hop speed includes both microtubule-dependent and microtubule-independent vesicle transport. Therefore, the colchicine treatment may have a different effect on these two subpopulations. Microtubules extend from the nucleus to cell surface. The disruption of microtubules results in a loss of these structures that may alter the transport of non-secretory vesicles. Once this structure is disrupted, non-secretory vesicles may be able to move more freely (diffuse faster), which results in an increase of the average hop speed for this specific population. This change in vesicle speed for non-secretory vesicles can counteract or dominate the expected decrease in microtubule-dependent vesicle speeds. By culturing cells in D₂O, it appears that microtubule-stabilization is the dominant factor that increases the overall speed of vesicle transport. Finally, vesicles with low hop speed of less than 10⁻² µm/s were classified as stationary structures and therefore were not used to calculate the average of mean vesicle hop speeds. This proportion of stationary vesicles is expected to increase after treating cells with colchicine. Indeed a ten-fold increase from 0.043 to 0.418 and from 0.065 to 0.580 in 0 moles/L D₂O and 15 moles/L D₂O cultures was observed, respectively. This analysis shows a clear effect of colchicine on vesicle traffic in both D₂O and non-D₂O treated RBL-2H3 cultures.

Figure 3 shows four box plots of mean hop speeds obtained from analyzing over 1,500 individual vesicle trajectories. Going from left to right, the first and second plot represent mean vesicle hop speeds in cells cultured in 0 moles/L D₂O without colchicine (Fig. 3, H₂O) and with colchicine (Fig. 3, H₂O with colchicine) treatment, respectively. The third and fourth plot represent mean vesicle hop speeds in cells cultured in 15 moles/L D₂O without colchicine (Fig. 3, D₂O) and with colchicine (Fig. 3, D₂O with colchicine) treatment, respectively. The median for mean vesicle hop speeds for cells cultured in 0 moles/L D₂O was 2.3 µm/s. This median speed was lower than the median of 5.3 µm/s computed from the mean vesicle hop speeds for cells cultured in 15 moles/L D₂O. For cells cultured in 15 moles/L D₂O the median for colchicine treated cells decreased to 1.4 µm/s as expected. However, for cells cultured in 0 moles/L D₂O the median of mean vesicle hop speeds for colchicine treated cells increased to 3.6 µm/s . This comparison also indicates that colchicine treatment may have different effects on microtubule-dependent and microtubule-independent vesicle transport. It is likely that the outliers of the distributions (red crosses in Fig. 3) for colchicine untreated cultures represent mean vesicle hop speeds for microtubule-dependent transport. This claim is supported by the following. Firstly, cargo transport by kinesin motors along microtubules can be faster than microtubule-independent (diffusive) protein transport (Koon, Koh & Chiam, 2014). Secondly, these outliers decrease significantly for colchicine treated cells cultured in both 0 moles/L and 15 moles/L D₂O. However, to fully investigate deuterium oxide’s microtubule stabilization effect on secretory vesicle transport, future experiments that clearly identify secretory vesicles are warranted. For example, one can follow Smith et al. (2003) and transfect RBL-2H3 cells with a green fluorescent protein-Fas ligand fusion protein (GFP-FasL) to study the transport of GFP-labeled secretory vesicles in more detail. Such investigations were beyond the scope of the presented study. Nevertheless, the reported results can be used as a springboard for future investigations.