2.1.

Reagents

DHA was obtained from Bide Pharmaceutical Corporation (Guangzhou, Guangdong Province, China). Working solutions were prepared by dissolving the compound in dimethyl sulphoxide (DMSO) before experiments. The final concentration of DMSO was less than 1% in all experiments. Lipofectamine2000 and Mitotracker Red 633 were purchased from Invitrogen (Carlsbad, California). $N$ -acetyl cysteine (NAC), Rhodamine123 (Rho123), Hoechst 33258, propidium iodide (PI), RNase A, SP600125, and SB203580 were obtained from Sigma (St. Louis, Missouri). Cell Counting Kit-8 (CCK-8) was purchased from Dojindo Laboratories (Kumamoto, Japan). Annexin V-FITC apoptosis detection kit was obtained from Bender Medsystems (Vienna, Austria). Caspase-3 substrate (Ac-DEVD-AFC), caspase-8 substrate (Ac-IETD-AFC), caspase-9 substrate (Ac-LEHD-AFC), staurosporine (STS), and U0126 were purchased from Alexis Biochemicals (Lausen, Switzerland). Mitochondria/cytosol fractionation kit was obtained from Calbiochem (Darmstadt, Germany). $2^{\prime },7^{\prime }$ -Dichlorodihydrofluorescin diacetate (DCFH-DA) was purchased from Wako Ltd (Osaka, Japan). Anti-cytochrome $c$ antibodies were obtained from Santa Cruz Biotechnology (Santa Cruz, California). Anti-Bcl-xL, anti-Bax, anti-caspase-3, anti- $\beta$ -actin, anti-phospho-ERK, anti-ERK, anti-phospho-JNK, anti-JNK, anti-phospho-p38, anti-p38 antibodies, and positive phospho-ERK/JNK/P38 control protein were all obtained from Cell Signaling (Beverly, Massachusetts). All the secondary antibodies were supplied by Molecular Probes (Eugene, Oregon).

2.2.

Cell Culture and Transfection

The ASTC-a-1 cell line was obtained from the Department of Medicine, Jinan University (Guangzhou, China), and cultured in Dulbecco’s modified Eagle’s medium (DMEM; Gibco, Grand Island, New York) supplemented with 10% fetal calf serum (FCS; Sijiqing, Hangzhou, China), $50\text{units}∕\mathrm{ml}$ penicillin, and $50\mu \mathrm{g}∕\mathrm{ml}$ streptomycin in 5% $\mathrm{C}{\mathrm{O}}_2$ at $37°\mathrm{C}$ in a humidified incubator. For fluorescence studies, cells were transiently transfected with plasmids using lipofectamine 2000 in a $35\text{-}\mathrm{mm}$ dish for $24\text{to}48\mathrm{h}$ . Cells stably expressing SCAT3 reporter were screened with $0.8\mathrm{mg}∕\mathrm{ml}$ G418, and positive clones were picked up with micropipettes.²⁹

2.3.

Cell Viability and Apoptosis Assay

Cell viability was assessed by CCK-8 assay, according to the supplier recommendations. Viable cells were counted by absorbance measurements at $450\mathrm{nm}$ using an auto-microplate reader (infinite M200, Tecan, Austria). The ${\mathrm{OD}}_{450}$ value was proportional to the degree of cell viability. All experiments were performed in quadruple on three separate occasions.

Apoptosis of cells was examined by two approaches: Hoechst 33258 staining and flow cytometric (FCM) analysis using Annexin V-FITC/PI staining were performed as we previously described,³ and for each FCM analysis, 10,000 events were recorded.

2.4.

Cell Cycle Analysis

The proportions of cells in sub- ${\mathrm{G}}_0$ (apoptosis), ${\mathrm{G}}_0∕{\mathrm{G}}_1$ , ${\mathrm{G}}_2∕\mathrm{M}$ , and S phases were determined by FCM analysis of DNA content. To evaluate the cell cycle profile, cells (about $1\times {10}^6$ cells) were harvested, washed twice with PBS, and fixed in ice-cold 70% (v/v) ethanol for $1\mathrm{h}$ at $4°\mathrm{C}$ . Prior to analysis, samples were washed again and incubated in PBS containing $10\mu \mathrm{g}∕\mathrm{ml}$ RNase A for $30\min$ , and then incubated with $5\mu \mathrm{g}∕\mathrm{ml}$ PI at $37°\mathrm{C}$ in the dark for $30\min$ . DNA content was determined using an FCM (FACS, Arla BD, San Jose, California), and data were analyzed by FCS Express version 3 software. For each analysis, 10,000 events were recorded.

2.5.

Measurement of Intracellular ROS Generation

ROS generation inside living cells was measured by confocal microscope imaging and FCM analysis using DCFH-DA, an oxidation-sensitive probe, which is cleaved intracellularly by nonspecific esterases and turns to highly fluorescent DCF upon oxidation by ROS. Untreated or treated cells were stained with $20\mu \mathrm{M}$ DCFH-DA for $30\min$ in the dark and subsequently assayed by FACScan. For each analysis, 10,000 events were recorded.

2.6.

Determination of Mitochondrial Size

The mitochondrial transversal section (width) is the most constant dimension of the organelle, and variations on this parameter reflect an alteration of the morphology of mitochondria.³³ Cells were cultivated on a coverslip of a $35\text{-}\mathrm{mm}$ chamber. After desired treatment, the cells were washed with PBS three times and incubated with $0.1\mu \mathrm{M}$ Mitotracker Red 633 for $30\min$ at room temperature in the dark. The cells were then washed three times with PBS and visualized under confocal microscope (LSM510/ConfoCor2, Zeiss, Jena, Germany). Mitotracker Red was excited at $633\mathrm{nm}$ , and the emitted light was recorded through a $650\text{-}\mathrm{nm}$ long-pass filter. For every condition tested, the size of 100 to 200 mitochondria in at least 50 different cells is from three independent experiments. The determination of the mitochondrial size was carried out using Zeiss Rel 3.2 image processing software (Zeiss, Jena, Germany) as we previously described.³⁴

2.7.

Fluorescence Recovery after Photobleaching (FRAP) Analysis inside Single Living Cell

FRAP analysis was performed on a confocal microscope (LSM510/ConfoCor2, Zeiss, Jena, Germany). Here, to investigate the mobility of GFP-Bax and the integrity of mitochondria after different treatments, the GFP or Mitotracker Red 633 in the indicating regions of living cells were photobleached by scanning the region with the maximal $488\text{-}\mathrm{nm}$ or $633\text{-}\mathrm{nm}$ laser line, and subsequently, the entire cell was imaged at every $5\mathrm{s}$ with a low laser power (5% power) excitation for a duration of $500\mathrm{s}$ to monitor the recovery of fluorescence.

2.8.

Dynamic Monitoring of Bax Translocation and Cytochrome $c$ Release

A confocal laser scanning microscope (LSM510/ConfoCor2, Zeiss, Jena, Germany) was used to perform time-lapse fluorescence imaging of Bax translocation and cytochrome $c$ release inside single living cells. Thirty-six hours after transiently expressing GFP-Bax or GFP-cytochrome $c$ $(\mathrm{GFP}\text{-}\mathrm{Cyt}.c)$ and DsRed-Mito, cells were treated and then imaged by the confocal microscope at $37°\mathrm{C}$ . The images of cells expressing GFP-Bax or $\mathrm{GFP}\text{-}\mathrm{Cyt}.c$ and DsRed-Mito were collected using dual fluorescence channels. The excitation wavelengths were $488\mathrm{nm}$ for GFP and $543\mathrm{nm}$ for DsRed. The emission fluorescence channels were $500\text{to}550\mathrm{nm}$ for GFP and $600\text{to}650\mathrm{nm}$ for DsRed.

2.9.

Measurement of Mitochondrial Membrane Potential $\left(\Delta {\Psi }_m\right)$

Rho123, a potential-sensitive dye, was used to evaluate changes in $\Delta {\Psi }_m$ by FCM as previously described.³ Briefly, cells after treatment with DHA for 0 and $24\mathrm{h}$ in the presence or absence of NAC were harvested and stained by $1\mu \mathrm{M}$ Rho123 for $30\min$ at $37°\mathrm{C}$ in the dark. Cells were then washed with PBS twice and subsequently assayed by FACScan. Results were expressed as the proportion of cells with lost or low $\Delta {\Psi }_m$ , which was estimated by reduced fluorescence intensity from Rho123. For each analysis, 10,000 events were recorded.

The $\Delta {\Psi }_m$ in single living cells was also monitored in real-time by time-lapse confocal imaging, as previously described.³ Briefly, the fluorescence images of cells stained with $5\mu \mathrm{M}$ Rho123 were monitored in real time using a confocal microscope (LSM510/ConfoCor2, Zeiss, Jena, Germany) equipped with a device sustained culture condition ( $37°\mathrm{C}$ , 5% $\mathrm{C}{\mathrm{O}}_2$ ).

2.J.

Detection of Emission Spectra of SCAT3 inside Living Cells

The cell line stably expressing SCAT3 was used for the emission spectral analysis of caspase-3 activation inside living cells, as described in our previous work.^{3, 34} Each well of a 96-well plate contained $5\times {10}^5$ cells stably expressing SCAT3. The emission spectra of SCAT3 inside living cells were scanned using an auto-microplate reader (Infinite M200, Tecan, Austria). The excitation wavelength was $430\mathrm{nm}$ , and the scanning range of emission fluorescence was from $456\mathrm{nm}\text{to}600\mathrm{nm}$ . The scanning step was $2\mathrm{nm}$ .

2.K.

Confocal Imaging and Fluorescence Resonance Energy Transfer (FRET) Acceptor Photobleaching Technique

In this study, confocal imaging and FRET analysis were performed on a confocal microscope (LSM510/ConfoCor2, Zeiss, Jena, Germany). All the quantitative analysis of the fluorescence images was performed by Zeiss Rel 3.2 image processing software (Zeiss, Jena, Germany). For time-lapse imaging, culture dishes were mounted onto the microscope stage equipped with a temperature-controlled chamber (Zeiss, Jena, Germany).

For photobleaching acceptor studies, $458\text{-}\mathrm{nm}$ laser from an argon ion laser was used for CFP excitation. Images were acquired through CFP and YFP filter channels, respectively. Here, the filter sets used were CFP (bandpass BP $470\text{to}500\mathrm{nm}$ ) and YFP (long-pass LP $530\mathrm{nm}$ ). $514\text{-}\mathrm{nm}$ laser of an argon ion laser was used to photobleach the acceptor of SCAT9 and FRET-Bid, respectively. In order to quantitatively analyze the FRET efficiency of FRET plasmids, we defined two parameters: $\Delta I_{\mathit{CFP}}$ and $\Delta I_{\mathit{YFP}}$ , where $\Delta I_{\mathit{CFP}}$ is the increase of CFP intensity, and $\Delta I_{\mathit{YFP}}$ is the decrease of YFP intensity, as described in our previous report.³⁵ The condition ( $37°\mathrm{C}$ , 5% $\mathrm{C}{\mathrm{O}}_2$ ) was sustained during measurement. All the quantitative analysis of the fluorescence images was performed by Zeiss Rel3.2 image processing software.

2.L.

Caspase Enzymatic Activities

Activities of caspase-3, caspase-8, and caspase-9 were measured using fluorogenic substrates Ac-DEVD-AFC, Ac-IETD-AFC, and Ac-LEHD-AFC, respectively, according to the manufacturer’s instructions. PBS-washed cell pellets (derived from either the medium or the adherent cells) were resuspended in extract buffer [ $25\mathrm{mM}$ HEPES (pH7.4), 0.1% TritonX-100, 10% glycerol, $5\mathrm{mM}$ DTT, $1\mathrm{mM}$ phenylmethylsulfonyl fluoride, $10\mathrm{mg}∕\mathrm{ml}$ pepstatin, and $10\mathrm{mg}∕\mathrm{ml}$ Leupeptin] and vortexed vigorously. A proluminescence caspase-3, caspase-8, or caspase-9 substrate was added to the extract in each well of a 96-well plate at $100\mu \mathrm{M}$ final concentration for $2\mathrm{h}$ at $37°\mathrm{C}$ , and caspase activity was measured continuously by monitoring the release of fluorigenic AFC using an auto-microplate reader (Infinite M200, Tecan, Austria). Caspase-like activity is reported as the ratio of the fluorescence output in treated samples relative to untreated controls.

2.M.

Protein Extraction and Western Blot Analysis

After different treatments, ASTC-a-1 cells were washed twice in cold PBS and lysed in a buffer containing $50\mathrm{mM}$ Tris HCl pH 8.0, $150\mathrm{mM}$ NaCl, 1%TritonX-100, $100\mu \mathrm{g}∕\mathrm{ml}$ PMSF and Protease Inhibitor Cocktail Set I (Calbiochem, San Diego, California). Lysates were separated by SDS-PAGE and transferred to a nitrocellulose membrane (Roch). The membrane was first blocked with 5% skim milk for $0.5\text{to}1\mathrm{h}$ , followed by incubation with primary antibodies overnight at $4°\mathrm{C}$ . Primary antibody incubation was performed in 5% bovine serum albumin in TBS-T (Tris base saline, pH 7.4, 0.1% Tween 20). A designated secondary antibody was diluted at the recommended ratio in TBS-T and 5% bovine serum albumin for $2.0\text{to}3.0\mathrm{h}$ at room temperature. Washing with TBS-T was performed between all steps. Detection was performed using the Odyssey Scanning Infrared Fluorescence Imaging System (LI-COR, Lincoln, Nebraska).

2.N.

Statistical Analysis

Results were expressed as $\text{mean}\pm \text{standard}$ deviation (SD). A student’s $t$ -test was used to compare the mean differences between samples using the statistical software SPSS, version 10.0 (SPSS, Chicago). Throughout the work, $P$ values less than 0.05 were considered to be statistically significant. All charts were drawn using Origin software, version 6.0.

3.1.

DHA-Induced ROS-Dependent Cytotoxicity in ASTC-a-1 Cells

Our previous study showed that DHA effectively exerted cytotoxicity on ASTC-a-1 cells in a time- and concentration-dependent manner.³ For every experiment described here, a lethal concentration of $20\mu \mathrm{g}∕\mathrm{ml}$ was used. In the present study, we used the same cell line to test the hypothesis that DHA-induced cell death is initiated by ROS. Intracellular ROS generation detected by DCFH-DA, an oxidation-sensitive fluorescent probe, was assessed by confocal microscope imaging and FCM analysis for both control and DHA-treated cells. As can be seen in Fig. 1 , exposure of cells to DHA elicited a rapid elevation of cellular ROS level within $30\min$ , and the dynamics of DCF fluorescence corresponding to Fig. 1 are shown in Fig. 1. The consistent results from FCM analysis demonstrated that DHA-treated cells exhibited a marked increase of DCF fluorescence as ${\mathrm{H}}_2{\mathrm{O}}_2$ treatment (as a positive control), which was completely abrogated by the antioxidant NAC pretreatment [Fig. 1], suggesting that DHA treatment led to a rapid induction of ROS inside cells. Moreover, CCK-8 assay was performed to further investigate the contribution of ROS to DHA-induced cytotoxicity, and our results indicated that NAC significantly attenuated DHA-induced cytotoxicity [Fig. 1], suggesting that DHA-elicited ROS were essential for DHA to exert its cytotoxic activity.

Fig. 1

DHA-induced ROS-dependent cytotoxicity in ASTC-a-1 cells. (a) Dynamical fluorescence images of DHA-induced ROS generation in living cells. Cells were incubated with $20\mu \mathrm{M}$ DCFH-DA, an oxidation-sensitive fluorescent probe, for $30\min$ in the dark then treated with $20\mu \mathrm{g}∕\mathrm{ml}$ DHA. The levels of intracellular ROS were determined by the fluorescence of DCF inside living cells using a confocal microscope. Scale bar: $10\mu \mathrm{m}$ . (b) Mean traces of the dynamics of the DCF fluorescence corresponding to Fig. 1. (c) FCM analysis of DHA-elicited ROS under different conditions indicated. Untreated or treated cells were harvested and stained with $20\mu \mathrm{M}$ DCFH-DA for $30\min$ before being subjected to FCM. Control: ${\mathrm{H}}_2{\mathrm{O}}_2$ treatment as a positive control. (d) Effect of NAC on DHA-induced cytotoxicity assessed by CCK-8. Cells were treated with $20\mu \mathrm{g}∕\mathrm{ml}$ DHA for 0, 12, 24, and $48\mathrm{h}$ in the presence or absence of $5\mathrm{mM}$ NAC. Experiments were conducted in quadruple wells, and data analyzed with SPSS 10.0 software were expressed as $\text{mean}\pm \mathrm{SD}$ . ${}^{**}P<0.01$ , compared with control; ${}^{\#}P<0.05$ , ${}^{\#\#}P<0.01$ , compared with DHA treatment alone.

3.2.

ROS Involvement in DHA-Triggered ${\mathrm{G}}_2∕\mathrm{M}$ Arrest and Subsequent Apoptosis

It has been suggested that the effects of ROS on initiating checkpoint responses influence the cell cycle progression.³⁶ In this study, evaluation of DNA content by FCM analysis of PI staining demonstrated that the proportion of cells with sub- ${\mathrm{G}}_0$ DNA content, indicative of apoptotic cells, increased from 1.94% (control) to 14.4% and 34.0% at 24 and $48\mathrm{h}$ after DHA treatment, which was markedly attenuated by NAC pretreatment [Fig. 2 ]. As shown in Fig. 2, DHA-treated cells exhibited time-dependent increases in the proportion of cells in ${\mathrm{G}}_2∕\mathrm{M}$ phase, which also remarkably decreased in the case of pretreatment of NAC, indicating that DHA-elicited ROS played an important role in DHA-induced ${\mathrm{G}}_2∕\mathrm{M}$ arrest.

Fig. 2

DHA-induced ROS-mediated cell cycle arrest and apoptosis. (a) Determination of cell cycle profile by FCM. Cells were cultured with $20\mu \mathrm{g}∕\mathrm{ml}$ DHA for 0, 24, or $48\mathrm{h}$ with or without the addition of $5\mathrm{mM}$ NAC and then stained with $5\mu \mathrm{g}∕\mathrm{ml}$ PI before being analyzed by FCM. (b) The proportion of cells in the ${\mathrm{G}}_0∕{\mathrm{G}}_1$ , S, and ${\mathrm{G}}_2∕\mathrm{M}$ phases of the cell cycle were calculated using FCS express V3 software and shown as $\text{mean}\pm \mathrm{SD}$ from three independent experiments. ${}^{*}P<0.05$ , ${}^{**}P<0.01$ , compared with control; ${}^{\#}P<0.05$ , ${}^{\#\#}P<0.01$ , compared with DHA treatment alone. (c) Nuclear morphology for apoptosis assessed by Hoechst 33258 staining. Images were recorded using a digital camera with $1280\times 1280\text{pixels}$ resolution. Magnification 400. (d) FCM analysis of DHA-induced ROS-dependent apoptosis. Cells were treated with $20\mu \mathrm{g}∕\mathrm{ml}$ DHA for 0 and $24\mathrm{h}$ in the presence or absence of $5\mathrm{mM}$ NAC, stained with Annexin V-FITC/PI, and then analyzed by FCM.

In our previous study, DHA-provoked caspase-3-dependent apoptosis was confirmed.³ We next sought to determine whether ROS were implicated in DHA-induced cell apoptosis. Hoechst 33258 staining experiments showed that chromatin was evenly present in the whole nucleus in control cells, and cells after exposure to DHA for $24\mathrm{h}$ underwent typical apoptotic morphology changes such as chromatin condensation, margination, and shrunken nucleus, which were effectively blocked by NAC pretreatment [Fig. 2]. These results implied that DHA induced ROS-dependent cell death in an apoptotic fashion. Moreover, the early apoptotic characteristic of appearance of phosphatidyl serine (PS) on the extracellular side of the cell membrane was quantified by annexin V-FITC/PI staining. The consistent results from Fig. 2 showed that NAC pretreatment conferred protection against DHA-induced apoptosis, further demonstrating that DHA triggered ROS-mediated apoptosis.

3.3.

DHA-Induced ROS-Dependent Bax Translocation and Activation

In some models of mitochondrial-dependent apoptosis, ROS act upstream to cause the activation of Bax or Bak.¹⁴ Here, we addressed the question of whether ROS acted upstream of Bax translocation and activation in DHA-induced apoptotic process. The FRAP technique was used to measure the mobility of the GFP-Bax fusion protein in single living cells after different treatments. Strikingly, we noticed a rapid refilling of GFP-Bax in the photobleached area for control cells [Fig. 3 , upper panel], implying that GFP-Bax distributed freely throughout the cell. However, DHA treatment induced GFP-Bax puncta formation, in which the fluorescence recovery in the photobleached area was completely blocked [Fig. 3, middle panel], implying that DHA induced Bax activation and translocation. Upon pretreatment with NAC, the reoccurrence of fluorescence recovery was observed as a control cell [Fig. 3, lower panel], suggesting that ROS mediated the DHA-induced Bax translocation. Dynamics of FRAP from 15 to 20 cells in three independent experiments for control, DHA-treated, and DHA- and NAC-cotreated cells are shown in Fig. 3. Furthermore, confocal images and statistical results also confirmed that DHA triggered time-dependent Bax translocation, which was effectively inhibited by NAC pretreatment [Figs. 3 and 3]. Based on these data, we concluded that ROS-mediated Bax recruitment after DHA treatment irreversibly localized to certain organelle membranes such as mitochondria or endoplasmic reticulum.

Fig. 3

DHA-induced ROS-dependent Bax translocation. (a) Typical time-lapse imaging of the mobility of GFP-Bax fusion protein in living cells after photobleaching GFP under different treatments. Upper panel: control cell; middle panel: DHA treatment; lower panel: DHA and NAC cotreatment. Scale bar: $5\mu \mathrm{m}$ . (b) Dynamics of FRAP from 15 to 20 cells in three independent experiments for control, DHA-treated, and DHA and NAC–cotreated cells. Data were collected at $5\text{-}\mathrm{s}$ intervals during recovery. (c) Typical fluorescence images of GFP-Bax distribution in living cells under different treatments. Left panel: DHA treatment alone; right panel: DHA and NAC cotreatment. Scale bar: $20\mu \mathrm{m}$ . (d) Quantification of cells showing GFP-Bax translocation. Cells were treated with $20\mu \mathrm{g}∕\mathrm{ml}$ DHA for 0, 24, and $48\mathrm{h}$ in the presence or absence of $5\mathrm{mM}$ NAC. Results were presented as $\text{mean}\pm \mathrm{SD}$ , collecting from 200 to 300 cells per treatment in 15 to 20 randomly selected image frames from three independent experiments. ${}^{**}P<0.01$ , compared with control; ${}^{\#\#}P<0.01$ , compared with DHA treatment alone. (e) Time-lapse imaging of Bax translocation into mitochondria inside single living cell. Cells transiently coexpressing GFP-Bax and DsRed-Mito were treated with $20\mu \mathrm{g}∕\mathrm{ml}$ DHA, and GFP-Bax localization at mitochondria was determined based on the overlays of GFP-Bax and DsRed-Mito fluorescence images. Scale bar: $10\mu \mathrm{m}$ . (f) Dynamics of relative fluorescence intensity of GFP-Bax redistribution in the enriched mitochondrial area (indicated by white square) corresponding to Fig. 3. (g) Western blot analysis of Bax translocation into mitochondria. Cells were treated with DHA for $48\mathrm{h}$ in the presence or absence of NAC, and STS-treated cells were used as a positive control. The cytosolic and mitochondrial proteins were extracted using a mitochondria/cytosol fractionation kit and then analyzed by Western blot using antibodies against Bax and $\beta$ -actin.

To determine whether DHA induced Bax translocation into mitochondria, we next monitored the dynamics of Bax translocation inside single living cells co-expressing GFP-Bax and DsRed-Mito using time-lapse confocal fluorescence microscope. We found that the diffuse distribution of GFP-Bax, after DHA treatment, became punctuated and colocalized with mitochondria, as visualized by a mitochondrial marker, DsRed-Mito [Fig. 3]. The increase of relative fluorescence intensity of GFP-Bax in the enriched mitochondrial area (indicated as a white square) corresponding to Fig. 3 also confirmed that translocation of cytosolic Bax to mitochondrial membrane occurred during DHA-induced apoptotic cell death [Fig. 3]. Last, Western blot analysis further demonstrated that DHA-provoked Bax translocation into mitochondria was significantly prevented by NAC pretreatment [Fig. 3]. Collectively, these findings suggested that DHA-elicited ROS promoted Bax activation and translocation into mitochondria.

3.4.

ROS-Mediated Mitochondrial Dysfunction during DHA-Induced Apoptosis

Mitochondrial dysfunction—in particular, the induction of the mitochondrial membrane potential $\left(\Delta {\Psi }_m\right)$ depolarization and ultrastructural changes—has been implicated in the cascade of events involved in the induction of apoptosis via the mitochondrial pathway.³⁷ In this study, the dynamical loss of $\Delta {\Psi }_m$ in single living cells stained with potential-sensitive dye Rho123 was monitored by imaging the reduction of Rho123 fluorescence using a time-lapse confocal microscope. The time-lapse images of cells stained with Rho123 and the mean dynamical traces of fluorescence intensity of these three cells under different treatments are shown in Figs. 4 and 4 , respectively. We found that compared with control cells, DHA treatment induced a significant decrease of Rho123 fluorescence, which was almost completely blocked by NAC pretreatment. Moreover, we used FCM to further confirm the loss of $\Delta {\Psi }_m$ by measuring the fluorescence of Rho123 under various treatments. At $24\mathrm{h}$ after DHA treatment, the percentage of cells with lost or low Rho123 fluorescence increased from 3.8% (control) to 38.8%, which reduced to 21.5% in the presence of NAC [Fig. 4]. These results revealed that DHA induced ROS-dependent loss of $\Delta {\Psi }_m$ .

Fig. 4

DHA-induced ROS-mediated mitochondrial dysfunction. (a) Time-lapse imaging of cells stained with potential-sensitive dye Rho123. Cells, after treatment with DHA for $24\mathrm{h}$ in the presence or absence of NAC, were harvested, and then stained with $5\mu \mathrm{M}$ Rho123 for $30\min$ in the dark. Scale bar: $10\mu \mathrm{m}$ . (b) Quantitative analysis of fluorescence intensities of Rho123 in the indicated cells corresponding to Fig. 4. (c) FCM analysis of DHA-induced ROS-dependent loss of $\Delta {\Psi }_m$ . Cells were cultured with DHA for $24\mathrm{h}$ with or without the addition of $5\mathrm{mM}$ NAC and then stained with $1\mu \mathrm{M}$ Rho123 before being analyzed by FCM. (d) Typical fluorescence images of mitochondria under different treatments. Upper panel: control cell; middle panel: DHA treatment; lower panel: DHA and NAC cotreatment. Scale bar: $10\mu \mathrm{m}$ . (e) Spatial distribution of mitochondrial size indicated as the white line corresponding to Fig. 4. (f) Quantification of mitochondrial width under different treatments. For every condition tested, data for the width were collected from 200 to 300 mitochondria in at least 50 different cells from three independent experiments. ${}^{**}P<0.01$ , compared with control; ${}^{\#\#}P<0.01$ , compared with DHA treatment alone. (g) Time-lapse imaging of mitochondria inside living cells after photobleaching Mitotracker Red 633 under different treatments. Upper panel: control cell; middle panel: DHA treatment; lower panel: DHA and NAC cotreatment. Images were recorded at $5\text{-}\mathrm{s}$ intervals during recovery. Cells, after treatment with DHA for $48\mathrm{h}$ in the presence or absence of NAC, were washed three times and then stained by $0.1\mu \mathrm{M}$ Mitotracker Red 633 for $30\min$ in the dark. Scale bar: $10\mu \mathrm{m}$ . (h) Quantitative analysis of fluorescence recovery of Mitotracker Red in indicated cells corresponding to Fig. 4. (Color online only.)

Recent studies have indicated that the loss of $\Delta {\Psi }_m$ leads to mitochondrial swelling, which is often associated with cell injury.³⁸ Thus, we measured mitochondrial width by imaging the mitochondria stained with Mitotracker Red 633 inside living cells. Figure 4 represents the typical confocal images of mitochondrial morphology, displaying that in contrast to the normal long tubular mitochondria in control cells [Fig. 4, upper panel], mitochondria inside DHA-treated cells were punctate and severely swollen [Fig. 4, middle panel], which can be significantly attenuated by NAC pretreatment [Fig. 4, lower panel]. Figure 4 shows the spatial distribution of mitochondrial size, indicated as the white line corresponding to Fig. 4. Statistical results of 200 to 300 mitochondria of at least 50 different cells from three independent experiments showed that at $48\mathrm{h}$ after DHA treatment, the size of mitochondria increased from $0.29\pm 0.03\mu \mathrm{m}$ (control) to $0.68\pm 0.08\mu \mathrm{m}$ , which reduced to $0.45\pm 0.11\mu \mathrm{m}$ in the case of pretreatment with NAC [Fig. 4], surely suggesting that ROS were implicated in DHA-triggered mitochondrial swelling.

Importantly, mitochondrial networks become extensively fragmented into smaller networks during apoptosis.³⁷ Rapid recovery of fluorescence after photobleaching is indicative of mitochondrial connectivity, while a failure to recover fluorescence demonstrates mitochondrial discontinuity or fragmentation. Using this method, we found a rapid fluorescence refilling in the long tubular mitochondria after photobleaching Mitotracker Red 633 for control cells [Fig. 4, upper panel], while the fluorescence failed to refill for DHA-treated cells [Fig. 4, middle panel]. However, fluorescence in the photobleached area recovered significantly for the cells co-treated with DHA and NAC [Fig. 4, lower panel], which was consistent with the FRAP curves of quantitative analysis of fluorescence intensities recorded during photobleaching corresponding to Fig. 4 [Fig. 4], further demonstrating that DHA-induced mitochondrial fragmentation, associated with the impairment of mitochondrial motility, was also mediated by DHA-elicited ROS.

3.5.

Involvement of Bcl-xL in the DHA-Induced Apoptosis and Inhibition of the DHA-Induced Bax Translocation by Bcl-xL Overexpression

It is not clear why mitochondrial networks fragment during apoptosis, but it has been proposed that this may be due to neutralization of the function of one or more members of Bcl-2 family.³⁹ Here, we considered whether Bcl-xL, a strongly anti-apoptotic member of the Bcl-2 family, was implicated in the regulation of apoptosis by DHA. As can be seen in Fig. 5 , pretreatment with HA14-1, a Bcl-xL inhibitor, significantly increased DHA-induced cytotoxicity compared with DHA treatment alone, suggesting that Bcl-xL protected cells against DHA-induced cell death.

Fig. 5

Overexpression of Bcl-xL prevented Bax translocation into mitochondrial membrane. (a) Inhibition of HA14-1, a Bcl-xL inhibitor, on cell viability assessed by CCK-8. Cells were cultured with DHA for 0 or $24\mathrm{h}$ with or without the addition of $10\mu \mathrm{M}$ HA14-1. Experiments were conducted in quadruple wells, and data analyzed with SPSS 10.0 software were expressed as $\text{mean}\pm \mathrm{SD}$ . ${}^{**}P<0.01$ , compared with control; ${}^{\#\#}P<0.01$ , compared with DHA treatment alone. (b) Western blot analysis of the inhibition of DHA on Bcl-xL expression. Cells were treated with DHA for 0, 24, and $48\mathrm{h}$ , and cotreated with DHA and NAC for $48\mathrm{h}$ . Bcl-xL expression in control, STS-treated (as a positive control), DHA-treated, and DHA and NAC–cotreated cells were detected by Western blot using antibodies against Bcl-xL and $\beta$ -actin. (c) Inhibition of Bcl-xL overexpression on Bax translocation. Cells transiently coexpressing CFP-Bax and YFP-Bcl-xL were treated with $20\mu \mathrm{g}∕\mathrm{ml}$ DHA, and the dynamics of Bax distribution were monitored using time-lapse confocal imaging. Scale bar: $5\mu \mathrm{m}$ . (d) Quantification of cells showing the inhibition of Bcl-xL overexpression on Bax translocation. Cells expressing GFP-Bax or coexpressing CFP-Bax and YFP-Bcl-xL were treated with $20\mu \mathrm{g}∕\mathrm{ml}$ DHA. Results were presented as $\text{mean}\pm \mathrm{SD}$ , collecting in 200 to 300 cells per treatment in 15 to 20 randomly selected image frames from three independent experiments. ${}^{**}P<0.01$ , compared with GFP-Bax-transfected control; ${}^{\#\#}P<0.01$ , compared with GFP-Bax-transfected cells after DHA treatment.

To explore whether ROS might mediate the effect of DHA on Bcl-xL, cells were treated with DHA in the presence or absence of NAC, and the corresponding effects on Bcl-xL expression are shown in Fig. 5. DHA caused a time-dependent decrease in Bcl-xL expression as well as STS treatment, whereas pretreatment of the cells with NAC markedly inhibited the decrease in Bcl-xL expression level [Fig. 5]. These findings implied that ROS were responsible in great part for DHA-induced Bcl-xL down-regulation.

Furthermore, to provide a mechanism insight into how Bcl-xL prevented DHA-induced cell death, we assessed the effect of Bcl-xL overexpression on Bax translocation in DHA-treated cells. For the cells co-expressing CFP-Bax and YFP-Bcl-xL, time-lapse images showed that after DHA treatment for $30\mathrm{h}$ , Bax still evenly distributed throughout the cells even in a dead cell (indicated by a white arrow) [Fig. 5], which was quite different from the punctuated distribution of Bax in the DHA-treated cells expressing GFP-Bax alone [Fig. 3], implying that Bcl-xL overexpression prevented Bax translocation. The percentage of cells showing Bax translocation for the cells expressing GFP-Bax alone and co-expressing both CFP-Bax and YFP-Bcl-xL at 0 and $24\mathrm{h}$ after DHA treatment is shown in Fig. 5. DHA treatment induced an increase of Bax translocation from $4.85\pm 1.5\mathrm{\%}$ (control) to $29\pm 2.1\mathrm{\%}$ , whereas the percentage of cells with DHA-induced Bax translocation markedly decreased to $16.8\pm 3.0\mathrm{\%}$ in the case of Bcl-xL overexpression. These results presented the evidence that Bcl-xL protected cells by preventing DHA-induced Bax translocation, also confirming that Bax activation was of paramount importance in the DHA-provoked apoptotic cell death.

3.6.

Requirement of ROS for DHA-Triggered Cytochrome $c$ Release and Caspase-9 Activation

The activation of Bax or Bak triggers apoptosis by decreasing the $\Delta {\Psi }_m$ , which resulted in the release of cytochrome $c$ from the mitochondrial intermembrane space into the cytosol.²² We first monitored the dynamics of cytochrome $c$ release inside single living cell co-expressing $\mathrm{GFP}\text{-}\mathrm{Cyt}.c$ and DsRed-Mito using a time-lapse confocal fluorescence microscope. Figure 6 shows a typical record of real-time images, where the DHA-treated cell was undergoing the release of cytochrome $c$ , which became diffusely distributed throughout the cytoplasm. In contrast, pretreatment with NAC markedly postponed cytochrome $c$ release in response to DHA (data not shown). Next, Western blot analysis further confirmed that the release of cytochrome $c$ was significantly inhibited by NAC pretreatment [Fig. 6], implying that ROS were required for cytochrome $c$ release in this apoptotic process.

Fig. 6

DHA-induced ROS-dependent cytochrome $c$ release and caspase-9 activation. (a) Typical time-lapse imaging of cytochrome $c$ release inside a single living cell. Cells transiently coexpressing $\mathrm{GFP}\text{-}\mathrm{Cyt}.c$ and DsRed-Mito were treated with $20\mu \mathrm{g}∕\mathrm{ml}$ DHA, and the release of $\mathrm{GFP}\text{-}\mathrm{Cyt}.c$ from mitochondria to cytosol was determined based on the overlays of $\mathrm{GFP}\text{-}\mathrm{Cyt}.c$ and DsRed-Mito fluorescence images. Scale bar: $5\mu \mathrm{m}$ . (b) Western blot analysis of the release of cytochrome $c$ from mitochondria to cytosol. Cells were treated with DHA for $48\mathrm{h}$ in the presence or absence of NAC, and STS-treated cells were used as a positive control. The cytosolic and mitochondrial proteins were extracted using a mitochondria/cytosol fractionation kit and then analyzed by Western blot using antibodies against cytochrome $c$ and $\beta$ -actin. (c) Typical fluorescence images and the corresponding dynamics of fluorescence intensity of both ECFP and Venus channels for control cell (left panel) and DHA-treated cell (right panel) transiently expressing with SCAT9 before and after photobleaching Venus. The photobleaching laser is $514\mathrm{nm}$ , and the FRET excitation laser is $458\mathrm{nm}$ . $\Delta I_{\mathit{CFP}}$ is the increase of ECFP intensity, and $\Delta I_{\mathit{\text{Venus}}}$ is the decrease of Venus intensity. All initial intensity is normalized to 100. Scale bar: $5\mu \mathrm{m}$ . (d) Quantitative analysis of the ratio of $\Delta I_{\mathit{CFP}}∕\Delta I_{\mathit{\text{Venus}}}$ as described in Sec. 2. Data were obtained from 15 to 20 cells in three independent experiments. ${}^{**}P<0.01$ , compared with control. (e) Detection of caspase-9 activation by fluorometric assay. Cells were treated with DHA for $48\mathrm{h}$ in the presence or absence of NAC, and STS-treated cells were used as a positive control. Caspase-9 activity was measured using the fluorescence substrate Ac-LFHD-AFC, as described in Sec. 2, and the activation index was determined as the ratio between the activity in extracts of treated cells and that measured in extracts of control cells. Data were analyzed with SPSS 10.0 software from three independent experiments. ${}^{**}P<0.01$ , compared with control; ${}^{\#\#}P<0.01$ , compared with DHA-treated cells.

Due to the fact of cytochrome $c$ release, it was possible to speculate whether the activation of caspase-9 was involved in DHA-provoked apoptosis, which was detected by FRET acceptor photobleaching analysis and fluorometric assay. First, we used FRET acceptor photobleaching analysis to assess the caspase-9 activation inside single living cells transiently expressing SCAT9. Our data showed that after photobleaching Venus with a concomitant decrease of fluorescence intensity $\left(\Delta I_{\mathit{\text{Venus}}}\right)$ in the Venus channel, the increase of fluorescence intensity $\left(\Delta I_{\mathit{CFP}}\right)$ in the ECFP channel for DHA-treated cells [Fig. 6, right panel] was much less than that for control cells [Fig. 6, left panel]. Thus, the ratio of $\Delta I_{\mathit{CFP}}∕\Delta I_{\mathit{\text{Venus}}}$ from at least 15 to 20 cells in three independent experiments had a significant decrease at $48\mathrm{h}$ after DHA treatment compared with the control [Fig. 6], indicating that DHA induced caspase-9 activation. Furthermore, we also used fluorometric assay to assess the role of ROS in DHA-induced caspase-9 activation. The proteolytic activity that fluorogenic polypeptide Ac-LEHD-AFC can be cleaved by caspase-9-like proteases was associated with caspase-9 activation. STS-treated cells were used as a positive control. After incubation with Ac-LEHD-AFC for $2\mathrm{h}$ , an obvious increase of caspase-9 activity was observed at $48\mathrm{h}$ after DHA treatment compared with control, and the caspase-9 activation was almost completely attenuated in the presence of NAC pretreatment [Fig. 6], indicating that ROS mediated the DHA-induced caspase-9 activation. Collectively, these results demonstrated that ROS acted upstream of DHA-induced cytochrome $c$ release and caspase-9 activation.

3.7.

DHA-Induced ROS-Dependent Caspase-3 Activation

It is reported that the activated caspase-9 then cleaves and activates other downstream caspases such as caspase-3,⁴⁰ and we recently found that DHA induced caspase-3-dependent apoptosis in ASTC-a-1 cells.³ In order to verify whether the DHA-induced caspase-3 activation was mediated by ROS generation, we used fluorometric assay, fluorescence emission spectra, and Western blot analysis to assess the effect of NAC on the DHA-induced caspase-3 activation. STS-treated cells were used as a positive control. For fluorometric assay, activated caspase-3, but not pro-caspase-3, does exert proteolytic activation on the Ac-DEVD-AFC substrate. As shown in Fig. 7 , compared with the control, DHA induced a nearly twofold increase of caspase-3 activity that could be almost completely blocked by NAC pretreatment, implying that ROS mediated the DHA-induced caspase-3 activation. Emission spectral analysis in living cells stably expressing SCAT3 showed that bimodal emission peaks of SCAT3 were observed around $481\mathrm{nm}$ and $524\mathrm{nm}$ in control cells, but in DHA-treated cells, the increase at $481\text{-}\mathrm{nm}$ and the decrease at $524\text{-}\mathrm{nm}$ emission peak were indicative of caspase-3 activation [Fig. 7]. However, NAC pretreatment significantly attenuated the DHA-induced decrease at the $524\text{-}\mathrm{nm}$ peak [Fig. 7], demonstrating that ROS played a critical role in DHA-triggered caspase-3 activation. Moreover, Western blot analysis further confirmed these findings that NAC effectively blocked the DHA-induced pro-caspase-3 cleavage [Fig. 7]. Our data suggested that DHA-triggered caspase-3 activation was ROS-dependent.

Fig. 7

DHA-induced ROS-dependent caspase-3 activation. (a) Detection of caspase-3 activation by fluorometric assay. Cells were treated with DHA for $48\mathrm{h}$ in the presence or absence of NAC, and STS-treated cells were used as a positive control. Caspase-3 activity was measured using the fluorescence substrate Ac-DEVD-AFC, as described in Sec. 2, and the activation index was determined as the ratio between the activity in extracts of treated cells and that measured in extracts of control cells. Data were analyzed with SPSS 10.0 software from three independent experiments. ${}^{**}P<0.01$ , compared with control; ${}^{\#\#}P<0.01$ , compared with DHA-treated cells. (b) Emission spectral analysis of caspase-3 activation in living cells. Cells stably expressing SCAT3 were treated with $20\mu \mathrm{g}∕\mathrm{ml}$ DHA for 0 or $48\mathrm{h}$ with or without the addition of NAC. (c) Western blot analysis of DHA-induced ROS-mediated caspase-3 activation. Cells were treated with DHA for $48\mathrm{h}$ in the presence or absence of NAC, and STS-treated cells were used as a positive control. Protein levels of caspase-3 in control, STS-treated, DHA-treated, and DHA and NAC–cotreated cells were examined by Western blot using antibodies against caspase-3 and $\beta$ -actin.

3.8.

DHA-Triggered ROS-Dependent Caspase-8 and Bid Activation

In addition to the mitochondrial death (intrinsic) pathway activated caspase-8 amplifies the apoptotic signal in the extrinsic pathway, either by directly activating the downstream caspase-3 or by cleaving Bid to form t-Bid.^{32, 41} Hence, we next explored whether caspase-8 and Bid activation were involved in DHA-triggered apoptosis, in which DHA-elicited ROS acted as an upstream modulator. First, the activation of caspase-8 was evaluated by determining fluorogenic AFC release. Ac-IETD-AFC, which can be cleaved by caspase-8-like proteases, was associated with caspase-8 activation. STS-treated cells were used as a positive control. As shown in Fig. 8 , compared with the control, DHA induced a nearly 1.8-fold increase of caspase-8 activity, which was significantly but not completely suppressed by the NAC pretreatment, indicating that ROS partially mediated the DHA-triggered caspase-8 activation. Second, we used FRET acceptor photobleaching analysis to assess the effect of DHA on Bid cleavage in single living cells transiently expressing YFP-Bid-CFP (FRET-Bid). Our data showed that after photobleaching YFP with a concomitant decrease of fluorescence intensity $\left(\Delta I_{\mathit{YFP}}\right)$ in the YFP channel, the increase of fluorescence intensity $\left(\Delta I_{\mathit{CFP}}\right)$ in the CFP channel for DHA-treated cells [Fig. 8, lower panel] was much less than that for control cells [Fig. 8, upper panel]. Thus, the ratio of $\Delta I_{\mathit{CFP}}∕\Delta I_{\mathit{YFP}}$ from at least 15 to 20 cells in three independent experiments had a significant decrease at $48\mathrm{h}$ after DHA treatment compared with the control [Fig. 8], suggesting that DHA induced Bid cleavage, which was probably due to DHA-triggered caspase-8 activation. Third, the dependence of tBid translocation on ROS generation was next investigated in single living cells co-expressing CFP-Bid and DsRed-Mito. Figure 8 shows the typical images of Bid distribution for control cells and DHA-treated cells. CFP-Bid had a diffuse and cytoplasmic localization in control cells, while DHA-treated cells showed Bid translocation from cytoplasm to mitochondria, which was remarkably blocked by NAC pretreatment (data not shown), as revealed by the overlaps of CFP-Bid and DsRed-Mito fluorescence images. Statistical results of the percentage of cells showing Bid translocation in at least 150 to 200 cells in three independent experiments are shown in Fig. 8, also demonstrating that DHA induced time-dependent Bid translocation to mitochondria, which was effectively attenuated by NAC pretreatment.

Fig. 8

DHA-triggered ROS-dependent caspase-8 and Bid activation. (a) Detection of caspase-8 activation by fluorometric assay. Cells were treated with DHA for $48\mathrm{h}$ in the presence or absence of NAC, and STS-treated cells were used as a positive control. Caspase-8 activity was measured using the fluorescence substrate Ac-IETD-AFC, as described in Sec. 2, and the activation index was determined as the ratio between the activity in extracts of treated cells and that measured in extracts of control cells. Data were analyzed with SPSS 10.0 software from three independent experiments. ${}^{**}P<0.01$ , compared with control; ${}^{\#\#}P<0.01$ , compared with DHA-treated cells. (b) Typical fluorescence images and the corresponding dynamics of fluorescence intensity of both CFP and YFP channels for control cell (upper panel) and DHA-treated cell (lower panel) transiently expressing FRET-Bid before and after photobleaching YFP. The photobleaching laser is $514\mathrm{nm}$ , and the FRET excitation laser is $458\mathrm{nm}$ . $\Delta I_{\mathit{CFP}}$ is the increase of CFP intensity, and $\Delta I_{\mathit{YFP}}$ is the decrease of YFP intensity. All initial intensity is normalized to 100. Scale bar: $10\mu \mathrm{m}$ . (c) Quantitative analysis of the ratio of $\Delta I_{\mathit{CFP}}∕\Delta I_{\mathit{YFP}}$ , as described in Sec. 2. Data were obtained from 15 to 20 cells in three independent experiments. ${}^{**}P<0.01$ , compared with control. (d) Representative images of Bid translocation into mitochondria. Cells transiently coexpressing CFP-Bid and DsRed-Mito were treated with $20\mu \mathrm{g}∕\mathrm{ml}$ DHA for 0, 24, and $48\mathrm{h}$ , and the translocation of Bid from cytoplasm to mitochondria was determined based on the overlays of CFP-Bid and DsRed-Mito fluorescence images. Scale bar: $5\mu \mathrm{m}$ . (e) Quantification of cells showing CFP-Bid translocation into mitochondria. Cells were treated with $20\mu \mathrm{g}∕\mathrm{ml}$ DHA for 0, 24, and $48\mathrm{h}$ in the presence or absence of $5\mathrm{mM}$ NAC. Results were presented as $\text{mean}\pm \mathrm{SD}$ , collecting in 150 to 200 cells per treatment in 15 to 20 randomly selected image frames from three independent experiments. ${}^{**}P<0.01$ , compared with control; ${}^{\#}P<0.05$ , ${}^{\#\#}P<0.01$ , compared with DHA treatment alone.

Taken together, these results consistently indicated that DHA-elicited ROS acted upstream of the caspase-8 and Bid activation.