A ratiometric two-photon probe for quantitative imaging of mitochondrial pH values

A ratiometric two-photon fluorescent probe for quantitative imaging of mitochondrial pH values in live cells and tissues was reported.


Introduction
The mitochondrion is the primary organelle for oxygen utilization to generate bioenergy. 1 While the cytosol and other organelles have neutral and acidic pH, the mitochondrial pH (pH mito ) is alkaline (pH $ 8.0) in resting cells. 2 Proton (H + ) extrusion via the respiratory electron transport chain provides a proton motive force (J H + ), which consists of a pH gradient across the mitochondrial membrane. 3 Mitochondria have also been shown to dynamically change in their morphology, cellular location, and distribution. They constantly undergo remodeling through ssion, fusion, biogenesis, and autophagy, as well as transport to specic sites with high energy demands. 4 The disruption of mitochondrial dynamics could be directly linked to apoptosis, cancer, aging, and neurodegenerative disorders. 5 As such, molecular imaging of mitochondrial dynamics could provide information in the study of human diseases and drug development. 6 However, the variation of pH mito values along with their subcellular distribution, specic location, and morphological change has not been well established. Therefore, a method for quantitative imaging of pH mito in living cells and tissues holds promise to elucidate mitochondria-related physiology and diseases.
To target the pH mito , pH-sensitive green uorescent protein (GFP) mutants have been developed and successfully utilized to conrm the alkaline pH of the mitochondria. 7 However, this approach suffers from variations in different cell types and technical difficulties when used with live tissues and animals. 8 Recently, small-molecule probes for pH mito based on the uorescence off-on response have been reported. 9 However, these probes utilized a single detection window and/or double excitation source that limit quantitative measurements due to the presence of experimental artifacts such as probe distribution, incident laser power, detection sensitivity, and photobleaching. Moreover, the pK a values of these probes are in the acidicneutral range (6. 2-7.4), which is far from the ideal measurement of pH mito (pH $ 8.0). In addition, live cell imaging with these probes required short excitation wavelengths that can cause photodamage and the articial production of reactive oxygen species (ROS). 10 Two-photon microscopy (TPM) is a fast growing domain in biomedical research, because it utilizes two near IR photons and offers higher spatial resolution with minimum background emission, deeper tissue penetration depth, and longer observation time. 11 In combination with TPM, an emission ratiometric probe with a single excitation wavelength would be an ideal method for quantitative imaging analysis of pH mito . The emission ratio simultaneously reects the population of the protonated and deprotonated forms of the uorescent emitter, which can cancel out the aforementioned experimental artefacts. Recently, a variety of small-molecule TP probes and their utilities in bioimaging applications have been reported. [12][13][14][15][16] However, no ratiometric TP probe for pH mito have been exploited. Hence, there is a strong need to develop an emission ratiometric TP probe for pH mito with a pK a value near 8.0.
Recently, we have developed a benzimidazole-derived ratiometric TP probe (pK a ¼ 5.82) for acidic pH detection, which has an electron donor-acceptor substituted dipolar character. 17 The probe showed red-shis in its spectrum upon protonation of the benzimidazole as a weak-base, owing to the increased electron withdrawing ability and the enhanced intramolecular charge transfer (ICT). However, this approach, utilizing protonation of a weak base, makes it difficult to modify the pK a to a value near 8.0. Thus, there is need to employ a weak acid for development of pH mito sensitive probe. In this work, we report a 2-naphthol-derived ratiometric TP probe (CMP1, Scheme 1) for pH mito that can quantitatively monitor subcellular pH mito values in living cells and tissues.

Results and discussion
We initiated this study using 2-naphthol, which has a pK a value of 9.63 and shows a red-shied emission maximum with pH change from acidic to basic (357 to 409 nm). 18 The spectral red-shi under basic conditions can be attribute to an enhanced ICT process due to the stronger electron donating ability of the deprotonated form (O À ) than the OH (s + ¼ À2.3 for O À vs. À0.92 for OH) (Scheme 1). 19 We anticipated that the pK a value of 2-naphthol derivatives could be optimized to the mitochondrial pH by employing an electron withdrawing group and extended p-conjugated system. These translations would stabilize the conjugated base (O À ), giving a lower pK a value than that of 2-naphthol, along with emission ratiometric character. Therefore, we rstly prepared compound 1 containing a benzothiazolyl electron withdrawing group (Scheme 1). Benzochromene-2-one derivative (2) was also synthesized as a linear p-conjugating system with the expectation of more effective ICT than 1. 14b The synthetic route for these compounds is described in the ESI. † Photophysical properties of 1 and 2 were measured in buffer solutions of different pH. Under acidic conditions (4.0), 1 displayed an absorption maximum (l abs ) at 344 nm and a uorescence emission maximum (l  ) at 480 nm with a large Stokes shi of 136 nm (Table 1). When the pH was changed from acidic (4.0) to basic (10.0), the l abs and l  of 1 were shied to the longer wavelengths at 380 nm and 530 nm, respectively, and with an increased uorescence quantum yield ( Fig. 1a and Table 1). These red-shis might reect the presence of the phenolate form at basic pH, as a stronger electron donor than the phenol form, thereby boosting the ICT processes. The pK a value was calculated from the titration curve of the emission ratios (I basic /I iso ) at the isoemission point (I iso ) and l  at pH 10.0 (I basic ). The pK a value for 1 was 9.04 AE 0.07, which is a markedly reduced value as compared to 2-naphthol. At pH 4.0, 2 showed more red-shied l abs and l  than those for 1, presumably due to the enhanced ICT. As the pH of the solution was changed from acidic to basic, the intensity of uorescence of 2 at 520 nm was gradually decreased with a simultaneous increase in intensity at 585 nm (Fig. 1b). The pK a value of 2 is found to be 8.41 AE 0.07, lower than the value for 1.
Next, we utilized 2 to prepare the mitochondria targeting probe (CMP1) by attaching triphenylphosphonium salt (TPP), a well-known mitochondria staining anchor, 20 through an amide linkage (Scheme 1). Upon changing the pH from acidic to basic, the spectral shis for CMP1 were nearly identical to those of 2, except that both l abs and l  of CMP1 and the phenolate form were shied to the longer wavelength range (Fig. 1c, Table  1). These observations are likely due to the combined effects of the stronger electron withdrawing ability of the amide group than the carboxyl group (s À ¼ 0.31 for CO 2 À vs. 0.61 for CONH 2 ) 19 and the inductive effect of the phosphonium cation. Consistently, the pK a value of CMP1 is 7.95 AE 0.05, a lower value than for 2 and well-matched to mitochondrial pH (8.0). Further, the plot of the emission ratio with pH indicated that this probe would be suitable for detecting over the pH range of 6.0-9.0 ( Fig. 1d). The ratio (I 604 /I 540 ) of CMP1 was robustly cycled back and forth in the pH changes between pH 5.0 and 9.0 ( Fig. S8, ESI †), indicating the reversible sensing ability. In addition, the Scheme 1 Structures of 2-naphthol, 1, 2, and CMP1, and proposed mechanism of the equilibrium on 2-naphthol and CMP1 and their deprotonated form. Table 1 Photophysical data for 1, 2, and CMP1 in buffer solution The numbers in parentheses are molar extinction coefficients in M À1 cm À1 . c l max of the one-photon emission spectra in nm. d Fluorescence quantum yield. e pK a values measured by one-photon mode. f l max of the two-photon excitation spectra in nm. g Two-photon absorption cross-section in 10 À50 cm 4 s per photon (GM). ratio of CMP1 in buffer solutions was unperturbed in the presence of biologically relevant metal ions, thiol species, ROS, reactive nitrogen species, and hydrogen sulde (Fig. S9, ESI †).
Consequently, CMP1 showed a marked yellow-to-red emission color change in response to a change in pH from acidic to basic and an appropriate pK a value for assessing the change of pH mito without interference from other physiological species. We then tested the ability of CMP1 as a TPM imaging probe. The maximum two-photon absorption cross section (d max ) values of 1, 2, and CMP1 at pH 4.0 and 10.0 were in the range of 20-70 GM, which are comparable to the values of existing smallmolecule TP probes (Table 1 and Fig. S10, ESI †). 12b Further, the TPM images of the CMP1-labeled HeLa cells and primary cultured astrocytes showed bright tubule-like networks, which might be the mitochondria. This probe displayed high photostability under the imaging conditions (Fig. S11, ESI †) and low cytotoxicity as determined by MTS and CCK-8 assays (Fig. 2).
To demonstrate the mitochondrial staining ability of the probe, we conducted co-localization experiments with mCherrymito-7, a well-known red-emissive protein for mitochondria. 21 The TPM image of CMP1 merged well with the mCherry-mito-7 image (Fig. 3a-c). The Pearson's co-localization coefficient (A) was 0.95 AE 0.05, which indicated that CMP1 predominantly resided in the mitochondria. A similar result was also obtained when the co-localization experiment was performed with the molecular marker, MitoTracker Red (A ¼ 0.88 AE 0.05) (Fig. S12, ESI †). Upon treatment of CCCP (carbonyl cyanide m-chlorophenyl hydrazone), a protonophore that promotes the release of intra-mitochondrial cations and causes rapid acidication by collapsing the mitochondrial membrane potential (DJ m ), 22 the A values between the images of CMP1 and mCherry-mito-7 were found to be 0.94 AE 0.05 (Fig. 3d-f). Because it has been wellestablished that the staining ability of mCherry-mito-7 does not depend on the DJ m , 21 this result conrmed the robust staining ability of CMP1 for the mitochondria regardless of the alteration of DJ m (Fig. 3).
We next tested the probe to determine the pH mito value in live cells using TPM. Live cell imaging was carried out at 37 C using an incubating chamber for maintaining humidity and pH. To nd the optimal detection window for ratiometric imaging, the TP excited uorescence (TPEF) spectra were collected from ionophore-treated HeLa cells labeled with CMP1 at pH 4.0 and 10.0. Upon excitation at 820 nm, the TPEF spectra at pH 4.0 and pH 10.0 were similar to those measured in buffer solution, in which the l  maxima (505 and 569 nm at pH 4.0 and 10.0) are slightly blue-shied from those measured in buffer solution (Fig. S13, ESI †). These results indicated that the polarity of the probe environment within the cells was rather homogeneous and slightly more hydrophobic than in the buffer solution. 17 This consequence allowed ratiometric imaging (F red / F green ) for pH mito by using 450-500 nm (F green ) and 550-600 nm   (F red ) as most sensitive windows. The pH calibration curve was obtained by using F red /F green of ionophore-treated and CMP1labeled HeLa cells (Fig. 4a). The pK a of CMP1 in cells was 7.86 AE 0.05, a nearly identical value as that measured in buffer solution. In addition, the plots of F red /F green vs. pH value indicated that CMP1 is suitable for measuring the pH mito in the range of pH 6.0-9.0.
More importantly, the ratiometric TPM image (F red /F green ) of CMP1-labeled HeLa cells clearly displayed various pH mito values ranging from 7.4 to 8.2, indicating the heterogeneity of pH mito (Fig. 4b and c). In addition, there was a denser population of mitochondria around the nucleus than in the periphery. Interestingly, Fig. 4b-d shows that the mitochondria in the perinuclear position had higher pH values in comparison to mitochondria at the periphery of cells. This difference might reect a mitochondrial region-specic function. 23 It has been well established that perinuclear mitochondria are mainly involved in mitochondria-metabolized ATP generation by regulating nuclear function by the assimilated phosphotransfer network between mitochondria and the nucleus. 23b The higher pH values in the perinuclear region may be linked to the local metabolic state, thus indicating to the need for future studies to elucidate its physiological function. Similar results were observed with primary cultured astrocytes (Fig. 7).
We next monitored the change of pH mito with carbonyl cyanide m-chlorophenyl hydrazone (CCCP) treatments and under starvation conditions. The average value of pH mito of CMP1-labeled HeLa cells was 7.82 AE 0.08. Upon treatment with CCCP for 1 h and 6 h, the average values of pH mito decreased to 7.09 AE 0.16 and 6.75 AE 0.27, respectively ( Fig. 5a-d). In addition, the ratio TPM imaging clearly displayed that the perinuclear mitochondria preferentially adopted a globular morphology with acidic pH, compared to the peripheral mitochondria ( Fig. 5b and c) and the process extended to the periphery. Moreover, higher magnication images clearly showed the pH values of globular structures ranged from 7.40 to 6.36 (Fig. 5c), which were signicantly decreased values by 0.7-1.8 pH units as  compared to those for long and highly interconnected structures. Similar results were detected in the cell starvation process. It is well-established that the damaged mitochondria can be eliminated by cell degradation processes, i.e., autophagy, a process in which mitochondria that have attained an acidic pH are entrapped in acidic lysosomes followed by hydrolytic digestion. 24 Upon nutrient deprivation for 2-6 h ( Fig. 5e-h), a globular morphology was noted in mitochondria near the nucleus appeared and this population of altered mitochondria increased with time; during this time, the pH values ranged from 7.1 to 6.2 ( Fig. 5e-g).
To conrm these processes, HeLa cells were co-labeled with CMP1 and BLT-blue, 25 a TP LysoTracker that emits separate TPEF from CMP1 in the cells (Fig. S14, ESI †). The dual-color TPM images (Fig. 6) clearly exhibited the autophagy processes. The Pearson's co-localization coefficient (A) was found to be 0.14 AE 0.5 before stimulating the cells (Fig. 6a, e and i). When the cells were starved in nutrient-deprived buffer solution, the A value increased to 0.24 to 0.38 to 0.55 aer starvation for 2 h, 4 h and 6 h, respectively (Fig. 6). Hence, these outcomes clearly demonstrated that CMP1 is capable of monitoring the change of pH mito values in live cells while dening the dynamic shapes of mitochondria at specic positions.
We further utilized CMP1 to measure the pH mito values in Parkinson's disease (PD) model astrocytes such as DJ-1, Parkin, and Pink gene knockout (KO) astrocytes. These astrocytes contain a PD-related gene that has multiple functions. The ratiometric TPM imaging with CMP1 exhibited the average pH mito value of 7.78 AE 0.09 in wild-type (WT) astrocytes (Fig. 7a). Moreover, the distribution of pH mito were similar between PD models and WT astrocytes, with the exception that the average pH mito values of PD models are slightly lower (ca. 0.1-0.2 pH unit) than those for WT astrocytes (Fig. 7b-d). This feature suggests the potential use of CMP1 in clinical applications.
Finally, we investigated the utility of CMP1 in live tissue imaging. The ratiometric TPM images were obtained from a fresh slice of rat hippocampus, a region of the brain that is important for learning and memory. The accumulated 360 TPM images collected at 90-210 mm depths provided the overall pH mito distribution through the regions of CA1, CA3, and DG (dentate gyrus) (Fig. 8). They showed the average pH mito values of these regions in 7.86-7.88 (Fig. 8), in which higher pH values of approximately 8.2 (red spots) are mainly appeared in the DG ( Fig. 8a and d). Moreover, the higher magnication images at these regions revealed the heterogeneous pH mito distribution at the tissue level.

Conclusions
We have developed a ratiometric two-photon probe (CMP1) for quantitative measurement of pH mito in live cells and tissues. This probe, designed with the aim of controlling ICT in 2naphthol, has a pK a value of 7.86 AE 0.05 in the cells, and shows a marked yellow to red emission color change in response to pH alterations from 6.0 to 9.0. CMP1 exhibits easy cell loading, robust staining ability of mitochondria, low cytotoxicity, and bright TPEF in situ, thereby allowing quantitative detection of the pH mito in live cells and tissues. The ratiometric TPM imaging studies by using CMP1 clearly reveal the heterogeneity of pH mito values with respect to the specic location of mitochondria within the cells, along with their different shapes. Further, we found that the pH mito values of mitochondria in the perinuclear region are higher than those at the periphery of cells. The average pH mito values of PD model astrocytes are slightly more acidic than those for WT astrocytes. These ndings demonstrate that CMP1 will be useful as a quantitative imaging probe to study pH mito in biomedical research.

Spectroscopic measurements
Absorption spectra were recorded on S-3100 UV-Vis spectrophotometer and uorescence spectra were obtained with Fluo-roMate FS-2 uorescence spectrophotometer with a 1 cm standard quartz cell. The uorescence quantum yield was determined by using Coumarin 307 (F ¼ 0.95 in MeOH) and Rhodamine 6G (F ¼ 0.95 in MeOH) as the reference by the literature method. 26 pK a value Fluorescence pH titrations were performed in buffer solution [pH 3.0, pH 4.0, pH 5.0 (0.1 M KH phthalate buffer solution), pH 6.0, pH 6.5, pH 7.0, pH 7.2, pH 7.5, pH 7.8, pH 8.0, pH 8.2, pH 8.5 and pH 8.8 (0.1 M KH 2 PO 4 buffer solution), pH 9.0, pH 9.2, pH 9.5, pH 9.8 and pH 10.0 (0.025 M sodium borate buffer solution), pH 11.0 (0.05 M NaHCO 3 buffer solution)]. For 1, 2 and CMP1, a 3.0 mL of the stock solution of probe in DMSO (1 Â 10 À3 M) was added to cuvette containing 3.0 mL of buffer solution to prepared 1.0 mM of probe solution and the spectral change in the uorescence were measured as a function of pH (3.0-11.0), pK a values of 1, 2 and CMP1 were calculated by linear regression analysis uorescence data to t the following equation. 27 where R is the observed ratios (I 529 /I iso for 1, I 587 /I iso for 2 and I 604 /I iso for CMP1, respectively) at given pH. R max and R min are the maximum and minimum limiting values of R and c is the slope. I a /I b is the ratio of the uorescence intensity in acid (pH 3.0) to the intensity in the base (pH 11.0) at the wavelength chosen for the denominator of R. In these cases, this correction vanishes by using the sharp isoemissive point (472 nm for 1, 554 nm for 2 and 572 nm for CMP1).

Measurement of two-photon cross section
The two-photon cross section (d) was determined by using femto second (fs) uorescence measurement technique as described. 28 1, 2 and CMP1 (1.0 Â 10 À6 M) were dissolved in buffer solution (pH 4.0 and 10.0) and the two-photon induced uorescence intensity was measured at 720-950 nm by using Rhodamine 6G as the reference, whose two-photon property has been well characterized in the literature. 29 The intensities of the two-photon induced uorescence spectra of the reference and sample emitted at the same excitation wavelength were determined. The TPA cross section was calculated by using following equation where the subscripts s and r stand for the sample and reference molecules. The intensity of the signal collected by a CCD detector was denoted as S. F is the uorescence quantum yield. 4 is the overall uorescence collection efficiency of the experimental apparatus. The number density of the molecules in solution was denoted as c. d r is the TPA cross section of the reference molecule. Fig. S9, ESI † represents the two-photon spectra of 1, 2 and CMP1 in different buffer solution conditions.

Cell culture
All the cells were passed and plated on glass-bottomed dishes (NEST) before imaging for two days. They were maintained in a humidied atmosphere of 5/95 (v/v) of CO 2 /air at 37 C. The cells were treated and incubated with 2 mM CMP1 at 37 C under 5% CO 2 for 30 min, washed three times with phosphate buffered saline solution (PBS; Gibco), and then imaged aer further incubation in colorless serum-free media for 30 min. The culture mediums for each cells are as below. HeLa human cervical carcinoma cells (ATCC, Manassas, VA, USA): MEM (WelGene Inc, Seoul, Korea) supplemented with 10% FBS (WelGene), penicillin (100 units per mL), and streptomycin (100 mg mL À1 ). Primary astrocytes were cultured from the cortex of DJ-1-KO, Parkin-KO or WT mice brains. In brief, cortexes were removed and triturated in DMEM (Invitrogen, Carlsbad, CA, USA) containing 10% FBS (HyClone, Logan, UT, USA), plated in 75 cm 2 T-asks (0.5 hemisphere per ask), and incubated for 2-3 weeks. Microglia were detached from asks by mild shaking, ltered through a nylon mesh to remove cell clumps, and cultured in DMEM containing 10% FBS. 30 Astrocytes remaining in the ask were harvested with 0.1% trypsin and cultured in DMEM containing 10% FBS.

Two-photon uorescence microscopy
Two-photon uorescence microscopy images of CMP1 labeled cells and tissues were obtained with spectral confocal and multiphoton microscopes (Leica TCS SP8 MP) with Â10 dry, Â40 oil and Â100 oil objectives, numerical aperture (NA) ¼ 0. 30, 1.30, and 1.30. The two-photon uorescence microscopy images were obtained with a DMI6000B Microscope (Leica) by exciting the probes with a mode-locked titanium-sapphire laser source (Mai Tai HP; Spectra Physics, 80 MHz, 100 fs) set at wavelength 820 nm and output power 2901 mW, which corresponded to approximately 6.26 Â 10 8 mW cm À2 average power in the focal plane. Live cell imaging was performed using the live cell incubator systems (Chamlide IC; Live Cell Instrument) for stable cell environment by maintaining proper temperature, humidity and pH over the long term. To obtain images at 450-500 nm (F green ) and 550-600 nm (F red ) range (Fig. S13a, ESI †) internal PMTs were used to collect the signals in an 8 bit unsigned 512 Â 512 and 1024 Â 1024 pixels at 400 and 200 Hz scan speed, respectively. Ratiometric image processing and analysis was carried out using MetaMorph soware.

Cell viability
To evaluate the cytotoxic effect of CMP1 in HeLa cells, MTS (cell Titer 96H; Promega, Madison, WI, USA) and CCK-8 kit (Cell Counting Kit-8; Dojindo, Japan) assay were performed according to the manufacture's protocol. The results are shown in Fig. 2, which revealed that the CMP1 has low cytotoxicity at its different concentration in our incubation condition.

Co-localization experiments
Co-localization experiments were conducted by co-staining the HeLa cells and astrocytes with appropriate combinations of CMP1 (2 mM), MTR (1 mM), mCherry-mito-7 and BLT-blue (0.5 mM) for 30 min. TPM and OPM images were obtained by collecting the emissions at 400-450 (BLT-blue), 450-500 and 550-600 (CMP1) and 650-700 (MTR, mCherry-mito-7), respectively. The two different detection windows for CMP1 were adopted upon requirement for colocalization experiments. The background images were corrected, and the distribution of pixels in the OPM and TPM images acquired in the green and red channels, respectively, was compared by using scatter gram. The Pearson's co-localization coefficients (A) were calculated by using AutoQuant X2 program.

Cell calibration
A pH calibration curve was generated by F red /F green of ionophores-treated and CMP1-labeled HeLa cells. The cells were treated and incubated with 2 mL of CMP1 in DMSO sock solution (1.0 mM) at 37 C under 5% CO 2 for 30 min, and then the extracellular media was replaced with 1 mL of calibration buffer (125 mM KCl, 20 mM NaCl, 0.5 mM CaCl 2 , 0.5 mM MgCl 2 , 5 mM nigericine, 5 mM monensin, and 25 mM buffer; MES for pH 6.0 and 6.5, HEPES for pH 6.5-8.0 and Tris for pH 8.5-9.0). 7a The cells were treated with the calibration buffer for 15-20 min at room temperature. The TPEF intensity at 450-500 nm (F green ) and 550-600 nm (F red ) of CMP1 (Fig. S13a, ESI †) well changed with pH, and we obtained pH calibration curve the plots of F red /F green versus pH values.

Preparation and staining of fresh rat hippocampal slices
Rat Hippocampal slice were prepared from the hippocampi of 2 weeks-old rat (SD) according to an approved institutional review board protocol. Coronal slices were cut into 400 mM-thicks using a vibrating-blade microtome in articial cerebrospinal uid (ACSF; 138.6 mM NaCl, 3.5 mM KCl, 21 mM NaHCO 3 , 0.6 mM NaH 2 PO 4 , 9.9 mM D-glucose, 1 mM CaCl 2 , and 3 mM MgCl 2 ). Slices were incubated with 20 mM CMP1 in ACSF bubbled with 95% O 2 and 5% CO 2 for 1 h at 37 C. Slices were then washed three times with ACSF and transferred to glassbottomed dishes (NEST) and observed in a spectral confocal multiphoton microscope. The TPM images obtained at about 90-210 mm depth.