2.1.

Cell Culture and mt-cpYFP Transfection

HeLa cells were used as the proof of principle demonstration in this study. Cells were cultured in Dulbecco’s modified Eagle’s medium (DMEM) supplemented with 10% fetal bovine serum and incubated at 37°C with 5% ${\mathrm{CO}}_2$. In experiments, cells were seeded in 35-mm Petri dishes with a 0.17-mm glass slide (Nest) for confocal microscopy and photostimulation. Mitochondrial matrix-targeting circularly permuted yellow fluorescent protein (mt-cpYFP) was used as a mitoflash biosensor.¹ Cells were transfected with mt-cpYFP plasmid at a concentration of $2.5\mu \mathrm{g}/\mathrm{mL}$ by polyethylenimine (PEI) ($1\mathrm{mg}/\mathrm{mL}$). The required amount of plasmid was dissolved in $100\mu \mathrm{L}$ DMEM, and then PEI was added in a volume to achieve PEI/DNA ratio at 5/1. The PEI-plasmid mixture was incubated at room temperature for 10 min and added to cell culture medium drop by drop. After transfection, cells were incubated for 24 h before experiments.

2.2.

Photostimulation and Confocal Microscopy

The photostimulation scheme, which enabled real-time observation and stimulation, was based on a confocal microscope (Olympus FV1000/IX81) coupled with a homemade Ti: Sapphire laser system (810 nm, 75 fs, and 80 MHz) as shown in Fig. 1(a). The fs-laser was collimated by a lens pair, directly reflected by a beam splitter (BS, $\text{reflection}=70\%$) to the back aperture of objective lens (Olympus, water immersion, $60\times$, $\mathrm{NA}=1.2$), which could be focused at a single mitochondrial tubule. A dichroic mirror (transmission: $>750\mathrm{nm}$; reflection: $<750\mathrm{nm}$) above the BS was used to split the fs-laser and the continuous wave (CW) laser inside the microscope. A CCD camera under the dichroic mirror was adopted for monitoring the focus of fs-laser. By the wild-field imaging of the CCD camera, the diameter of the laser focus was measured ($\sim 1.5\mu \mathrm{m}$). Therefore, unlike the scanning focus in two-photon microscopes, the focus of fs-laser was a fixed spot in the central position of the field of view (FOV) in this setup. The output power of fs-laser could be tuned by a round continuously variable metallic neutral density filter (Thorlabs). A mechanical shutter (GCI-73, China Daheng Group) was used to control the laser stimulation duration. In the vertical direction (the laser propagation direction vertical to the specimen), the focus of fs-laser was adjusted to the confocal scanning focal plane by tuning distances between the collimating lenses. In this way, any observed mitochondria in the imaging plane could be stimulated by the fs-laser. In experiments, the laser beam was originally cut off by the shutter. Once the targeted mitochondrion was selected, it was moved to the focal position of the fs-laser by adjusting the stage of the microscope (to the center of FOV). Then the shutter would open for a preset duration to deliver photostimulation during CW laser confocal microscopy (note that the photostimulation would not affect the confocal microscopy). In general, a select cell would be stimulated for only one time, which would excite only one tubular structure of a single mitochondrion inside.

Fig. 1

Photostimulation to mitochondria. (a) The photostimulation scheme based on confocal microscope system. BS, beam splitter; DM, dichroic mirror; and PMT, photomultiplier tube. (b) Multiple mitoflashes acquired from the stimulated mitochondrial tubule. The quantification of the mitoflashes is defined. Red arrow: photostimulation. Insets: time-lapse images of mitoflashes indicated by mt-cpYFP. Bar: $10\mu \mathrm{m}$.

For confocal scanning, mt-cpYFP was excited by a 473-nm laser and detected at 510 to 550 nm. The cells were observed continuously by the confocal microscopy in experiments. In general, for each time the photostimulation was delivered during the confocal imaging, the targeted cell was observed for a total duration of 440 s to provide an observation window as long as possible to acquire the most mitoflash events while minimizing information missing (such as spontaneous mitoflashes caused by long scanning time of the confocal microscopy). Images were acquired at a speed of one frame ($512\times 512\text{pixels}$) per 2.2 s.

2.3.

Statistics

The data were presented as mean ± standard error of mean (SEM) and analyzed with GraphPad Prism 7 software through the one-tailed paired $t$-test except where stated otherwise.

3.1.

Mitoflash Excitation

To manually initiate mitoflashes, mitochondria labeled with mt-cpYFP were randomly selected and then stimulated by the fs-laser at 16 to 32 mW (at cell) for 66 to 150 ms. In our experiments, we acquired a very high stimulation–response ratio (90 responding mitoflashes from 92 stimulated mitochondria in 92 different cells of 15 independent experiments). However, unlike the randomness of spontaneous mitoflashes, fs-laser initiated mitoflashes were only found in those stimulated mitochondria during the observation period. Generally, around 10 s after the photostimulation, the stimulated mitochondria showed mitoflash events. As control, we counted the spontaneous mitoflashes in cells without any stimulation. The spontaneous mitoflash events were observed 33 times in total in 262 cells during the 440-s observation duration for each cell.

Intriguingly, multitime mitoflashes could be observed in the stimulated mitochondrial tubule after a single stimulation [Fig. 1(b)]. To describe the properties of mitoflashes in a quantitative manner, we defined parameters as shown in Fig. 1(b). The time interval between the photostimulation (fs-laser exposure) and the first mitoflash peak was defined as the response duration ($T_s$). The time interval ($\mathrm{\Delta }T$) between two adjacent mitoflashes was defined by the duration between the half peak of the mitoflashes. The time width of each mitoflash was defined by the full duration at half maximum (FDHM). Though the morphology of the mitochondria might change (fragmented and soon recovered or swelled without recovery)^19,20 after fs-laser stimulation, the actual displacement of the targeted mitochondrion is negligible during the whole observation. Therefore, the statistical fluorescent intensity of the original tubular structure would fully represent the slightly changing mitochondrion.

Fluorescent images [insets in Fig. 1(b)] showed photostimulated mitoflashes in cells labeled with mt-cpYFP. The mitochondrial morphology did not change and the cpYFP fluorescence had no photobleaching after stimulation (stimulation: 16 mW for 100 ms, observation: 440 s later). The viability of stimulated cells was validated by the proliferation rate (cell amount afterward/original cell number). To be strict, cells were stimulated at 40 mW from 100 ms to 1 s. The proliferation rate was then acquired after 12 and 24 h, respectively (the data of each stimulation time were acquired from five independent tests and presented as $\text{mean}\pm \mathrm{SEM}$), as shown in Table 1. It could be found that the proliferation rate of those stimulated cells was all close to the control (without any stimulation). When the photostimulation duration was increased to 5 s, the proliferation rate decreased dramatically.

Table 1

The proliferation rate of the stimulated cells (at 40 mW).

The amplitude of a mitoflash, which was defined as the maximum ratio of fluorescence $F$ over the initial baseline $F_0$ ($F/F_0$), was $1.324\pm 0.012$ ($n=150$ flashes in 90 stimulated mitochondria) in photostimulated cells, which was close to spontaneous mitoflashes in rat cardiac myocytes ($F/F_0\sim 1.41\pm 0.02$),¹ neural progenitor cells ($F/F_0\sim 1.29\pm 0.11$),¹⁴ HeLa cells ($F/F_0\sim 1.32$),¹⁵ and in our experiment ($F/F_0\sim 1.303\pm 0.024$). This result suggests the mt-cpYFP indication as a stable and universal method for quantitative study of mitoflash.

3.2.

Quantitative Analysis

In order to quantitatively present this laser-based method, we investigated the properties of mitoflashes excited by photostimulation at different laser powers. With a universal 100-ms duration, we triggered mitoflashes using three sets of moderate power (16, 24, and 32 mW at cell) fs-laser stimulation. Each mitochondrion was stimulated for only one single time and thereafter would be observed for 440 s. If at least one-time mitoflash was observed, the excitation would be counted as one effective trial. The efficiency (stimulation–response ratio) of mitoflash excitation was then acquired as 94.44% at 16 mW (incident energy: 1.6 mJ, $n=18$), 95.24% at 24 mW (incident energy: 2.4 mJ, $n=21$), and 100% at 32 mW (incident energy: 3.2 mJ, $n=17$). To check if the mitoflash excitation was energy dependent, we tuned the stimulation duration (exposure time) with a fixed fs-laser power to create varied incident fs-laser energy. The stimulation was performed at 24 mW for 67 ms (24 mW-S) and 150 ms (24 mW-L), respectively, to provide the incident energy at 1.6 and 3.2 mJ, the same energy level as the photostimulation at 16 and 32 mW. In this way or alike, we would be able to compare the mitoflashes at different stimulation powers with the same incident laser energy, and at different energies with the same power.

The photostimulation could activate almost immediate mitoflashes in the stimulated mitochondria. We quantified the response time $T_s$ of mitoflash at different stimulation powers. As shown in Fig. 2(a), with an increasing laser power, the response duration $T_s$ dropped significantly ($7.29\pm 1.07$ s at 16 mW, $n=17$; $5.22\pm 0.67\mathrm{s}$ at 24 mW-S, $n=18$; $4.10\pm 0.42\mathrm{s}$ at 24 mW, $n=20$; $3.22\pm 0.33\mathrm{s}$ at 24 mW-L, $n=18$; $3.35\pm 0.56\mathrm{s}$ at 32 mW, $n=17$, respectively). With the ascending stimulation laser power, it would take shorter time to observe the first mitoflash in the stimulated mitochondria. But it should be noted that the minimal $T_s$ is around 3.28 s (with enough-high incident energy).

Fig. 2

Quantitative analysis of mitoflashes excited by different laser powers. (a) The response duration of mitoflash could be observed after the photostimulation. (b) The average number of mitoflashes observed in each stimulated mitochondrion. (c) The amplitude and (d) pulse width of each mitoflash. (e) The time interval between multitime mitoflashes observed from a single mitochondrion. (f) The first flash width and the amplitude of second flash were sensitive to different stimulations. $x$-axis label; upper, incident energy level; below, incident power level; 24-S, 24 mW for 66 ms; and 24-L, 24 mW for 150 ms. *$P<0.05$, **$P<0.01$ (independent $t$-test).

Once the first fs-laser stimulated mitoflashes were initiated, one or more subsequent mitoflashes could be observed in the following $\sim 440\text{-}\mathrm{s}$ observation window [Figs. 2(b) and 2(c)]. However, the third mitoflash was relatively rare [Fig. 2(c)]. Interestingly, multiflashes were more likely to be induced by weak stimulations. It was very difficult to observe the third mitoflash at 24 or 32 mW. The average number of mitoflashes in the stimulated mitochondrion observed in the window decreased significantly if the laser power increased to 32 mW [Fig. 2(b)].

Next, we quantified the amplitude ($F/F_0$) and duration (FDHM) of the multiple mitoflashes excited by a single-time photostimulation as shown in Figs. 2(c) and 2(d). On weak stimulation (16 mW and 24 mW-S) conditions, the amplitude of the first and second mitoflashes was the same ($1.346\pm 0.031$, $n=17$ and $1.345\pm 0.052$, $n=12$ at 16 mW; $1.36\pm 0.042$, $n=18$ and $1.355\pm 0.058$, $n=10$ at 24 mW-S). The third one showed a little decrease at 16 mW ($1.275\pm 0.069$, $n=5$) and no change at 24 mW-S ($1.365\pm 0.062$, $n=6$). At 24 mW, the first mitoflash amplitude ($1.348\pm 0.031$, $n=20$) was also quite close to that at 16 mW. But the second flash significantly decreased [$1.217\pm 0.029$, $n=10$, Fig. 2(c)]. When facing stronger simulations (24 mW-L and 32 mW), the amplitude of the first mitoflash still remained the same ($1.336\pm 0.04$, $n=18$ at 24 mW-L and $1.336\pm 0.034$, $n=17$ at 32 mW). But the second amplitude showed a little decrease at 24 mW-L ($1.288\pm 0.034$, $n=9)$, and a significant decrease at 32 mW ($1.174\pm 0.05$, $n=5$) compared with the first or the second mitoflash excited by 16 mW. Moreover, the total number of the second or third mitoflash events under photostimulations at higher laser power also decreased significantly compared with ones at weak stimulation.

The width of mitoflashes, determined by the mPTP modulation, was even more sensitive to different photostimulations. As shown in Fig. 2(d), the width variation of the first mitoflash was much larger compared with the amplitude. Generally, with an ascending fs-laser power, the width of the first mitoflash decreased significantly ($15.35\pm 1.875\mathrm{s}$, $n=17$ at 16 mW; $14.06\pm 2.139\mathrm{s}$, $n=18$ at 24 mW-S; $11.95\pm 1.206\mathrm{s}$, $n=20$ at 24 mW; $10.44\pm 1.149\mathrm{s}$, $n=18$ at 24 mW-L; and $9.59\pm 0.515\mathrm{s}$, $n=17$ at 32 mW, respectively). Under each power circumstances, the width of the second mitoflash all increased dramatically compared with the first one ($27.42\pm 7.464$, $19.8\pm 4.669$, $27.5\pm 5.33$, $25\pm 3.037$, and $16.4\pm 7.075\mathrm{s}$, respectively).

The time interval ($\mathrm{\Delta }T$) between two adjacent flashes decreased at high-intensity stimulation [Fig. 2(e)]. Notably, at 16 mW, $\mathrm{\Delta }{T}_2$ was shorter than $\mathrm{\Delta }{T}_1$. $\mathrm{\Delta }{T}_1$ also decreased along with the increase of stimulation power.