Am J Physiol Cell Physiol 290: C524-C538, 2006.
First published August 31, 2005; doi:10.1152/ajpcell.00290.2005
0363-6143/06 $8.00
METHODS IN CELL PHYSIOLOGY
Spectral imaging microscopy demonstrates cytoplasmic pH oscillations in glial cells
Sergio Sánchez-Armáss,1,*
Souad R. Sennoune,2,*
Debasish Maiti,2
Filiberta Ortega,1 and
Raul Martínez-Zaguilán2,3
1Departamento de Fisiología y Farmacología, Facultad de Medicina, Universidad Autónoma de San Luis Potosí, San Luis Potosí, Mexico; and 2Department of Physiology and 3Southwest Cancer Center, Texas Tech University Health Sciences Center, Lubbock, Texas
Submitted 15 June 2005
; accepted in final form 24 August 2005
 |
ABSTRACT
|
|---|
Glial cells exhibit distinct cellular domains, somata, and filopodia. Thus the cytoplasmic pH (pHcyt) and/or the behavior of the fluorescent ion indicator might be different in these cellular domains because of distinct microenvironments. To address these issues, we loaded C6 glial cells with carboxyseminaphthorhodafluor (SNARF)-1 and evaluated pHcyt using spectral imaging microscopy. This approach allowed us to study pHcyt in discrete cellular domains with high temporal, spatial, and spectral resolution. Because there are differences in the cell microenvironment that may affect the behavior of SNARF-1, we performed in situ titrations in discrete cellular regions of single cells encompassing the somata and filopodia. The in situ titration parameters apparent acid-base dissociation constant (pK'a), maximum ratio (Rmax), and minimum ratio (Rmin) had a mean coefficient of variation approximately six times greater than those measured in vitro. Therefore, the individual in situ titration parameters obtained from specific cellular domains were used to estimate the pHcyt of each region. These studies indicated that glial cells exhibit pHcyt heterogeneities and pHcyt oscillations in both the absence and presence of physiological HCO3–. The amplitude and frequency of the pHcyt oscillations were affected by alkalosis, by acidosis, and by inhibitors of the ubiquitous Na+/H+ exchanger- and HCO3–-based H+-transporting mechanisms. Optical imaging approaches used in conjunction with BCECF as a pH probe corroborated the existence of pHcyt oscillations in glial cells.
proton gradients; proton waves; carboxyseminaphthorhodafluor-1; 2',7'-bis(2-carboxyethyl)-5(6)-carboxyfluorescein
IT HAS BEEN SUGGESTED THAT nonneuronal cells called glial cells contribute to information processing in the brain (19). This requires two-way communication between neurons and glial cells. There are three categories of glia: Schwann cells, oligodendrocytes, and astrocytes (19). Glial cells exhibit many of the same voltage-sensitive ion channels and neurotransmitter receptors as neurons (72). However, glial cells lack the membrane properties needed to trigger action potentials. Thus glial cells use chemical signals rather than electrical signals to communicate with each other and with neurons (13, 54). The chemical signals for communication between neurons and glial cells include ion fluxes, neurotransmitters, and other specialized molecules (12, 13, 54). Dynamic changes in cytosolic pH (pHcyt) are important in the regulation of many physiological functions, including cell growth, differentiation, exocytosis, and cell signaling.
In the cerebral cortex, stimulus-evoked glial depolarization is associated with a transient intracellular alkalinization (10). The regulation of pHcyt transients is unclear, but it may involve the coordinated action of several pHcyt regulatory mechanisms. Specifically, it has been found that glial cells, including the C6 glial cell line, utilize the Na+/H+ exchanger (NHE) as an important pHcyt regulatory mechanism (15, 22, 30, 63). Astrocytes also utilize both the Na+-independent and the Na+-dependent Cl–/HCO3– exchanger to regulate pHcyt (11, 33, 47, 63). The Na+-HCO3– cotransporter has also been found in glial cells, including astrocytes (11, 15).
The physiological significance of pHcyt gradients has been suggested (29, 32, 49). Specifically, cells with a polarized morphology (e.g., intestinal enterocytes and other epithelial cells) display a distinct pHcyt between apical and basolateral sites (23, 67). Cells with a distinct functional polarization (e.g., microvascular endothelial cells) display a distinct pHcyt in the lamellipodium (leading) vs. the trailing (lagging) cell regions (62). In polarized neutrophils, pHcyt waves moving from the lagging to the leading edge, as well as differences in viscosity among these cellular regions, have been described (58, 74). In some instances, the regional pHcyt differences appear to be related to the preferential localization of H+-coupled secondary transporters (23, 67) or H+-ATPases (62). It has also been suggested that there are near-membrane pH heterogeneities, possibly due to regional differences in membrane acid-base transporters and/or the location of charged sugars in the glycocalyx in the apical membrane patches of Madin-Darby canine kidney cells (34). In isolated snail neurons, pHcyt microdomains are induced by membrane depolarization (61). In contrast, studies in other cell types have not detected pHcyt heterogeneities (8, 25, 69, 73).
Most of the studies suggesting the existence of distinct pHcyt heterogeneities have not considered the possibility that the behavior of the fluoroprobes used to measure pHcyt may be distinct in discrete subcellular domains because of distinct microenvironments. If this were the case, it could hamper the interpretation of the data regarding pHcyt heterogeneities. This is critical, because pHcyt heterogeneities under steady-state conditions may determine distinct behavior of proteins or transporters or the enzymatic activities in different cellular domains. Furthermore, pHcyt gradients may have significance in several physiological functions, including exocytosis, neuronal differentiation, development of growth cones and neurites, regulation of pHcyt in dendritic spines, glial-neuron communication, learning, and memory.
In this study, we tested the hypotheses that 1) glial cells exhibit pHcyt oscillations under steady-state conditions that can be perturbed by alkalosis, acidosis, and inhibitors of pHcyt regulatory mechanisms and 2) the spectral properties of pH fluoroprobes such as carboxyseminaphthorhodafluor (SNARF)-1 and BCECF are affected by the intracellular microenvironment. Using spectral imaging microscopy, we found that the properties of the pH fluoroprobe SNARF-1 must be characterized by performing in situ titrations of discrete cellular domains to avoid errors in the interpretation of data. Optical imaging approaches using BCECF as a pH probe corroborated the idea that in situ titrations are needed to properly assign pHcyt values in discrete cellular domains. We concluded that glial cells exhibit pHcyt oscillations under steady-state conditions in the presence and absence of HCO3–. Furthermore, conditions that elicit cytosolic alkalosis or acidosis as well as inhibition of NHE- and HCO3–-based H+-transporting mechanisms affect the properties of pHcyt oscillations and the propagation of proton waves.
|
MATERIALS AND METHODS
|
|---|
Buffers.
Physiological saline solution (PSS) contained (in mM) 140 NaCl, 5 KCl, 1.5 KH2PO4, 1.0 CaCl2, 1.2 MgCl2, 10 HEPES, 10 glucose, and 5 glutamine, pH 7.2, at 37°C. High-K+ titration solution (TS) contained (in mM) 140 KCl, 10 MES, 10 HEPES, 10 N,N-bis(2-hydroxyethyl)glycine (bicine), 0.007 nigericin, and 0.002 valinomycin, adjusted for pH with NaOH as needed, at 37°C (46).
Cell culture.
The C6 glial cell line was obtained from the American Type Culture Collection (no. CCL107). Cells were grown on circular coverslips (25 mm) in Ham's F-10 medium supplemented with 15% horse serum and 2.5% fetal bovine serum.
Spectral imaging microscopy.
Cells were loaded with 7 µM SNARF-1 in its AM form and incubated for 30 min at 37°C to allow for hydrolysis of the ester bond (44). The cells were washed and further incubated for 30 min with buffer to allow for hydrolysis of uncleaved dye. Coverslips with cells were placed into a thermostated chamber (PDMI-2; Medical Systems, Greenvale, NY) and then fastened onto the stage of an inverted fluorescence microscope (Olympus IX-70). The chamber was maintained at 37°C and was continuously superfused with PSS at a rate of 3 ml/min, maintaining a final volume of 3.0 ml. Fluorescence signaling was evaluated with a spectral imaging microscope (45, 46). The spectral imaging system consisted of a Spectra-Pro-300i spectrograph (Acton Research Instruments, Acton, MA) directly coupled to the side port of an Olympus IX70 inverted microscope through a C-mount adaptor (IX-TVAD). For spectral imaging, we used a high-dynamic-range frame transfer back-illuminated, cryogenically cooled, charge-coupled device (CCD) camera (Spec10:400 B, LN, 16 bit; Princeton Instruments, Trenton, NJ) that was controlled using a ST133 controller (Princeton Instruments). The CCD had a 1,340 x 400-pixel imaging array (pixel 20 x 20 µm). The spectrograph and the CCD camera settings were controlled using a personal computer equipped with WinSpec software (version 2.5.10.1
[EC]
; Roper Scientific, Trenton, NJ). The CCD temperature was maintained at –100°C for all the experiments. The entrance of the slit spectrograph was set at 200 µm throughout the experiments, except for the zero-order spectra, for which the slit was set at 3.0 mm. For cell imaging we used a x60, 1.4 numerical aperture (NA) oil-immersion lens objective. Thus, under conditions in which the slit was set at either 3.0 mm or 200 µm, the image projected onto the CCD camera corresponded to 50 µm x 100 µm or 15 µm x 100 µm, respectively. The length of the cell area delimited by the 200-µm slit width was divided into 16 discrete cellular regions of interest (ROIs) encompassing 
2.3–4.5 µm/ROI. This approach allowed us to acquire images with high temporal and spectral resolution. The optical filters used for excitation of SNARF-1 were a 488-nm narrow-band-pass excitation filter and a 550-nm long-bandpass dichroic mirror. This allowed collection of full SNARF-1 spectra from 560 to 750 nm at 0.4-nm resolution. To study the significance of pHcyt waves and gradients under physiological conditions, we used PSS supplemented with 24 mM HCO3– in equilibrium with 5% CO2 at extracellular pH (pHex) 7.4 (21). Cells were continuously perfused at 3.0 ml/min with PSS containing HCO3–, and steady-state pHcyt was monitored for 5 min. The perfusate was then exchanged to one containing 25 mM NH4Cl, and cells were perfused for 5 min to elicit alkalosis. Thereafter, the cells were perfused with buffer lacking NH4Cl to elicit cytosolic acidosis, and pHcyt was monitored for an additional 5 min.
SNARF-1 in situ calibration.
Simultaneous in situ calibrations of SNARF-1 at 16 different cellular regions (encompassing the soma and filopodia) were performed at the single-cell level. Briefly, PSS buffer was exchanged for high-K+ TS buffer, and the cells were subsequently incubated over 10 min to ensure total collapse of pHcyt gradients before the titration procedure was started. pHcyt was increased from 5.5 to 8.0 using stepwise increases in pHex in cells perfused with high-K+ TS for 3 min to allow for equilibration of pHex = pHcyt. The first-order spectra were acquired simultaneously from the 16 discrete cell regions.
SNARF-1 in vitro calibration.
To characterize the behavior of SNARF-1 in solution, we performed in vitro titrations in the microscope cell chamber with 3 ml of high-K+ TS containing 2 µM SNARF-1 potassium salt using the x60 objective at 200-µm slit width. The in vitro titrations were performed from pH 5.2 to pH 9.2 (45). The emission spectra of 16 different regions along the entrance slit at 5-µm intervals were acquired simultaneously. The pH value after each addition of NaOH was measured with a pH glass electrode in parallel titration curves. The pH of the titration curve in the presence of glycerol or BSA was measured as stated above. BSA alkalinized the initial pH of the titration solution and was therefore adjusted for pH. In vitro titrations were also performed in a SLM spectrofluorometer DMX-1000 with a quartz cuvette containing 2 ml of high-K+ TS plus 2 µM SNARF-1-free acid.
Spectral imaging and data analysis.
All data were corrected using background subtraction followed by flat field correction, which allowed us to divide out small nonuniformities in gain from pixel to pixel. For data analysis, we took advantage of the fact that SNARF-1 exhibits a clear isoemissive point at 609 nm. Importantly, after deprotonation of its phenolic group, it undergoes a significant spectral shift to longer wavelengths. For purposes of data analysis, we denominated the region below the isoemissive wavelength (i.e., 560–609 nm) the protonated SNARF-1 domain and the region above the isoemissive wavelength (609–750 nm) the deprotonated domain. Thus the protonated and deprotonated domains were fitted to the following logistic power peak function:
 | (1) |
where n = exp{[x + d ln (e) – c]/d}. This equation was solved iteratively with TableCurve 2D (version 5.0; Systat Software, Richmond, CA) to obtain the area under the curve of the protonated and deprotonated domains. With this equation, the correlation coefficient (r2) for fitting these curves is >0.99. The ratios of the area under the deprotonated to protonated domains were obtained with SigmaPlot software (version 8.0; SPSS, Richmond, CA) and used to estimate pHcyt from in situ titrations performed at the end of the experiment. This approach allowed us to use the complete spectral output of SNARF-1 to estimate pHcyt instead of the classic ratio values used for SNARF-1 at 644 nm (deprotonated) and 584 nm (protonated) (44, 46).
We favor the selected approach over spectra shape analysis algorithms because the spectra of SNARF-1 exhibit spectral shifts associated with changes in spectral shape, amplitude, and intensity that are affected by the degree of protonation. These properties impede the use of a discrete series of spectra typically needed for spectral shape analysis, in which a library of discrete and distinct spectra must be created (18). Although this is achieved relatively easily when the position and shape parameters remain constant and the fluorescence intensity is linear, the spectral changes in SNARF-1 are not linear but sigmoidal. Furthermore, spectral shape analysis is facilitated when the amplitudes of individual components with quenching obey the Stern-Volmer law (9, 39). This is not the case for pH fluoroprobes, in which the ion-fluoroprobe chelate exhibits a nonlinear Henderson-Hasselbalch equilibrium. The log-normal functions used in spectral shape analysis have been used successfully to describe absorption and fluorescence spectra and multicomponent absorption/fluorescence spectra of molecules exhibiting mirror-symmetrical form (9, 48). However, as shown in Fig. 2, the SNARF-1 fluorescence spectrum is not symmetric and exhibits an infinite number of transitions between the protonated and deprotonated states, which complicated assignment of spectral shape to specific pH values. Thus the selected approach of fitting the area under the protonated/deprotonated domains to obtain a singular value of decomposition seemed to be the most appropriate. Furthermore, the use of all experimentally determined points in the input spectra for decomposition attenuated the noise due to heterogeneities in dye concentration among different cellular domains.
View larger version (28K):
[in this window]
[in a new window]
|
Fig. 2. SNARF-1 in situ titration. A: spectra of SNARF-1 (2 µM free acid) as a function of pH were obtained using stepwise increases in extracellular pH (pHex) from 5.2 to 9.2 as described in MATERIALS AND METHODS to show that there is a clearly defined isoemissive wavelength at 609 nm with protonated and deprotonated domains below and above the isoemissive point. B and C: in situ titrations of 2 cells loaded with SNARF-1. In situ titrations were performed in high-K+ buffer in the presence of nigericin and valinomycin as described in MATERIALS AND METHODS. Note that there are significant variations in the ratio values among different tracks at the same pH because the in situ calibration parameters (pK'a, Rmax, and Rmin) are different (see text for details). R2 for the fitted titration curves varied between 0.927 and 0.984. Ratios of the deprotonated-to-protonated areas were fitted using a modified Henderson-Hasselbalch equation to obtain the in situ titration parameters for SNARF-1. Data are representative of 30 in situ titrations.
|
|
pHcyt measurements using BCECF as a pH probe.
As an alternative pH probe to SNARF-1, we used the ratiometric pH probe BCECF (21). In these experiments, cells grown onto 25-mm round coverslips were loaded with 2 µM BCECF-AM for 30 min at 37°C. Thereafter, cells were washed and incubated with PSS buffer for an additional 30 min to allow for hydrolysis of dye. BCECF fluorescence was monitored with a cell imaging system consisting of a high-speed filter changer (Lambda DG4; Sutter Instruments, Novato, CA) to rapidly change excitation filters (440 and 505 nm) and a quartz fiber optic to deliver the light. The excitation light was reflected with a dichroic mirror (515 DRLPXR). We collected the fluorescence signal at an emission of 530 nm using a frame transfer CCD camera coupled to an intensifier (Princeton Instruments Intensifier Pentamax, ADC 5 MHZ). Cells were imaged using an inverted microscope (Olympus IX-70) with a x60 oil-immersion lens objective (1.4 NA). In this imaging system, the excitation filters can be changed as fast as 1.5 ms. For these experiments, we used 20-ms exposure to obtain a high signal-to-noise ratio. Images were collected and analyzed using Meta Imaging Systems software (version 4.5; Universal Imaging, West Chester, PA). Cells loaded with BCECF were transferred to the microscope cell chamber (PDMI-2; Medical Systems) and maintained at 37°C on the microscope stage. Cells were continuously perfused at 3.0 ml/min, and steady-state pHcyt was monitored for 5 min in the presence of 24 mM HCO3– in equilibrium with 5% CO2 at pHex 7.4, followed by NH4Cl treatment and removal to elicit cytosolic alkalosis and acidosis as described for SNARF-1 experiments. In situ calibrations were performed at the end of every experiment as described for SNARF-1 measurements. The BCECF in situ calibration parameters were obtained essentially as described for SNARF-1, except that this analysis used the BCECF pH-sensitive ratios (60).
Rheology measurements.
Viscosity measurements were performed at 37°C with a Brookfield programmable rheometer DV-111 V3.3 LV (Brookfield, Stoughton, MA) using cone and plate geometry (CP-40) within a shear rate range from 525 to 900 s–1.
Statistical analyses.
To evaluate the variation in the titration parameters of SNARF-1 among the different cellular domains, we used the coefficient of variation (CV; Ref. 24). For normal distribution, the statistical differences were determined using a t-test or ANOVA with either the Tukey or Bonferroni test for multiple comparisons. For nonparametric distribution, the statistical differences were determined using Kruskal-Wallis one-way ANOVA on ranks with Dunn's test for multiple comparisons or the Mann-Whitney rank-sum test. The Wilcoxon signed-rank test was used for paired data comparisons (SigmaStat version 2.03; Jandel Scientific). All statistical tests were considered significant at P < 0.05.
|
RESULTS
|
|---|
Spectral imaging microscopy allows determination of pHcyt heterogeneities.
Figure 1A shows the zero-order spectra of a single C6 glial cell intracellularly loaded with SNARF-1. We used SNARF-1 as a ratiometric pH probe in the spectral imaging system because it exhibits distinct spectral output in the protonated and deprotonated states. It should be noted that the cell exhibits clearly defined somata and filopodia. Decreasing the slit width from 2,000 to 200 µm (Fig. 1, A–C) allowed us to obtain the first-order spectra and the complete emission of SNARF-1 from 550 to 700 nm at 0.4-nm spectral resolution (Fig. 1D). For this experiment, we analyzed the spectral domains from somata to filopodia encompassing 16 ROIs at 4.5 µm each in the vertical direction and 2 µm each in the horizontal direction (numbered 1–16 in Fig. 1E) on images of the cell shown in Fig. 1A obtained at time 0. These data reveal that the spectral intensities vary in different cellular domains (Fig. 1E). Figure 1F shows selected ROIs from Fig. 1E to demonstrate that there are significant differences in spectral shape between cellular domains (e.g., domain 1 vs. domain 8 and domain 2 vs. domain 7). For purposes of data analysis, we normalized the spectra from these cellular domains. It should be noted that, notwithstanding the different fluorescence intensities in the spectra from cellular domains 1 and 2, their spectral shapes are similar. The same statement applies to cellular domains 7 and 8 (Fig. 1G). These data suggest that the changes in spectral shape shown in Fig. 1, E–G, are due to differences in either the pHcyt or the microenvironment of those cellular domains. These data also indicate that cellular domains 7 and 8 exhibited a more alkaline pHcyt than cellular domains 1 and 2. The steady-state SNARF-1 deprotonated-to-protonated ratios for the cellular domains from somata to filopodia show significant variations in ratios among the 16 different domains (CV = 6.2%; Fig. 1H). Specifically, cellular domains 1 and 2 (similar spectral shape) had lower ratios (i.e., more acidic pHcyt) than did cellular domains 7 and 8 (P < 0.05; also with similar spectral shape). It is also evident that cellular domain 5, located in the soma, had the same ratio as cellular domain 12, located in the filopodium; however, cellular domain 12 was sixfold dimmer than cellular domain 5 (cf. Fig. 1E). These data demonstrate that, within these boundary conditions, the deprotonated-to-protonated ratio can correct for differences in dye concentration. Therefore, the observed variations in spectra and ratios among the subcellular regions are likely due to differences in pHcyt or in the microenvironment in those cell regions.
Spectral characterization of SNARF-1 in situ.
To assign pHcyt values for each cellular domain, we performed in situ titrations. The spectral output was collected from 16 different cellular domains along the vertical slit by binning the data corresponding to 4.5 µm of the projected image at 200-µm slit width. It should be noted that SNARF-1 exhibited a clear isoemissive wavelength at 609 nm and that the fluorescence signals below and above this wavelength decreased and increased, respectively, as pH was increased (Fig. 2A). We fitted the complete protonated and deprotonated SNARF-1 domains as a function of pH using Eq. 1 to obtain a singular value decomposition. The deprotonated-to-protonated domain ratios were then plotted as a function of pH, and the data were fitted using a modified Henderson-Hasselbalch equation. This allowed us to determine the SNARF-1 in situ titration parameters that describe this behavior. Figure 2, B and C, shows the variation in the deprotonated-to-protonated SNARF-1 ratio as a function of pH in two distinct cells with a fully collapsed pHcyt gradient with H+ ionophores. The mean CVs (MCVs) for the in situ titration parameters for the 16 cellular domains of 25 single C6 cells were about six times larger (pK'a MCV = 2%, Rmax MCV = 17%, and Rmin MCV = 9%) than those observed in the in vitro titrations. These data suggest that, notwithstanding the fact that all pHcyt gradients were collapsed during the titration procedure, the differences in ratio among cellular domains persisted, probably because of the interactions of the dye with the cytoplasmic microenvironment. These data also indicate that the use of regional pK'a and Rmax for assignment of pHcyt in discrete cell regions is necessary, because error propagation analysis indicated that if a single mean pK'a or Rmax were applied to a single ratio value, there would be an under- or overestimation of pHcyt by 0.15 and 0.08 pH units, respectively. These data emphasize the need to use in situ calibration parameters for each cellular domain.
Once we established that the behavior of SNARF-1 in situ was heterogeneous in different cellular domains, we performed spectral imaging analysis in cells whose pHcyt gradients had not been collapsed by ionophores. In these studies, we selected cells with clearly defined filopodia and somata. Steady-state deprotonated-to-protonated SNARF-1 ratio values for the 16 selected cellular domains were obtained every 30 s for up to 5 min. These data indicate that the variation in steady-state deprotonated-to-protonated SNARF-1 ratio values for each of the 16 selected cellular domains vs. time was small (MCV = 0.38%). However, among different cellular domains, there was a significant (P < 0.05) variation in steady-state SNARF-1 ratios (MCV = 10%, n = 11; cf. Fig. 3, A and F). This value should be compared with the in vitro variations in ratio among the 16 discrete domains for a single pH value (i.e., pH 7), which had a MCV of 1.8% (n = 6; data not shown). Cells that did not exhibit defined filopodia displayed the same pattern, suggesting that even in cells without an apparent morphological polarization, there were heterogeneous SNARF-1 ratios and therefore varying pHcyt values. To calculate the pHcyt values in distinct cellular domains, we used the unique in situ titration parameter values corresponding to each cellular domain (Fig. 3, B–D and G–I). As shown in Fig. 3, there were significant variations (P < 0.05) in the values of pK'a, Rmax, and Rmin parameters among different cell regions. There seemed to be no predictable pattern in the distinct domains.
View larger version (31K):
[in this window]
[in a new window]
|
Fig. 3. In situ calibration parameters and steady-state pHcyt values from different cellular domains in 2 glial cells. Glial cells loaded with SNARF-1 were handled as described in Fig. 1 to perform spectral imaging microscopy. A and F: SNARF-1 steady-state ratio profiles from 2 glial cells in 16 discrete tracks. In situ titration was performed at the end of the experiment to obtain the in situ calibration parameters for each of these cellular domains: pK'a (B and G), Rmax (C and H), and Rmin (D and I). E and J: steady-state ratio values from each cell (A and F) were converted to pHcyt with individual pK'a, Rmax, and Rmin for each cellular domain to assign specific pHcyt. Note that there seems to be no predictable pattern in the variations of either the cytoplasmic titration parameters or the basal ratio or cytoplasmic pH values. Data are representative of 11 experiments.
|
|
Once the specific set of in situ titration parameters was calculated for every cellular domain, the deprotonated-to-protonated SNARF-1 ratios (cf. Fig. 3, A and F) were transformed to pHcyt. Figure 3, E and J, shows the distribution of steady-state pHcyt values for the cells in Fig. 3, A and F. It is interesting to note that in these cells, regions with similar basal ratios exhibited different pH values [see cellular domains at 10 and 70 µm (Fig. 3, F and J) or cellular domains at 10 and 30 µm (Fig. 3, A and E)], emphasizing the relevance of using individual in situ pK'a, Rmax, and Rmin to assign pHcyt values for each cellular domain. However, in other instances, the pHcyt differences resembled the variations in basal ratio (see cellular domains from 10 to 60 µm in Fig. 3, A and E). For these cells, the pHcyt values for the distinct cellular domains varied between 6.95 and 7.25 in one case (Fig. 3E) and between 7.00 and 7.25 in the second case (Fig. 3J). Each of the cells studied had a unique deprotonated-to-protonated SNARF-1 ratio and therefore a unique pHcyt profile in distinct cellular domains. A summary of pK'a and estimated pHcyt in five selected cells is shown in Table 1. The data in Table 1 show examples of the behavior of pK'a and pHcyt from soma to dendrite. Analysis of 49 cells in terms of pHcyt behavior from soma to dendrite with linear regression analysis indicated that 42% of the cells exhibited a 0.1- to 0.2-pH unit decrease from soma to filopodium.
To further evaluate the difference in pHcyt from soma to filopodium, we used a distinct pH probe. We selected BCECF because it is a ratiometric pH probe that has been used widely and exhibits a pK'a of 7.0 (21). Similarly to the procedure used with SNARF-1, we performed in situ titrations using BCECF to characterize the behavior of this probe from soma to filopodium. Figure 4 shows an example of an in situ titration in different cellular regions encompassing these cellular domains. Increasing pH from 6.0 to 8.0 resulted in increases in ratio. The pK'a in these different cellular regions varied from 6.87 to 7.05; however, the most remarkable variability was in Rmax. Table 2 shows the behavior of pK'a and estimated pHcyt values from soma to filopodium in five selected cells. Although there is not an immediately apparent trend in the data, linear regression analysis of 60 cells indicated that 40% of the cells were 0.15–0.2 pH units more acidic in dendrites than in somata. These data corroborate the data obtained using SNARF-1 to indicate that there are distinct pHcyt values associated with discrete cellular domains. Altogether, these data indicate that the distinct pHcyt in each cellular domain is not necessarily associated with a unique pK'a, but that the pK'a variations in different cellular domains, as well as in Rmax and Rmin values, determine the unique pHcyt values for each cellular domain.
View larger version (24K):
[in this window]
[in a new window]
|
Fig. 4. In situ titration of BCECF in different cellular domains encompassing soma and filopodium in a single glial cell. Glial cells were loaded with BCECF as described in MATERIALS AND METHODS, and the probe was excited at 440 and 505 nm with a high-speed filter changer. The emission was collected at 505 nm using a CCD camera coupled to an intensifier. Cells were visualized using a x60 objective [1.4 numerical aperture (NA)]. Discrete regions of interest (ROI; 10 x 10 pixels) were located from soma to filopodium to obtain in situ calibration parameters. For purposes of data presentation, we show only 10 ROIs from a single cell. In situ titration was performed at the end of the experiment to obtain the in situ calibration parameters for each of these cellular domains: pK'a, Rmax, and Rmin. It is apparent from this figure that Rmax is remarkably distinct in each domain, although there are also differences in pK'a and Rmin. A summary of pK'a values for 5 selected cells is shown in Table 2. Data are representative of 11 experiments.
|
|
Viscosity and protein environment affect SNARF-1 fluorescence properties.
The variation in the steady-state deprotonated-to-protonated SNARF-1 ratio may be due to actual pHcyt differences, to variations in the regional cytoplasmic microviscosity (45, 59, 68, 74), or even to different proportions of dye bound to cytoplasmic proteins (4). To evaluate these possibilities, we performed in vitro SNARF-1 calibration curves at 37°C. In these experiments, SNARF-1 was dissolved in solutions containing either glycerol (15% wt/vol) to change the viscosity of high-K+ TS from 0.648 to 0.967 mPa·s at 37°C or BSA (1%) to allow the binding of SNARF-1 to BSA in addition to increasing the viscosity from 0.648 to 0.889 mPa·s at 37°C. The addition of glycerol not only modified the viscosity but also linearly decreased the solvent dielectric constant (28), thus affecting the SNARF-1 pK'a, because the acid-base equilibrium is sensitive to changes in the solvent dielectric constant (3). Viscosity values for the leading (3.8 mPa·s), trailing (0.5 mPa·s), and cell body (0.5 mPa·s) regions of locomoting neutrophils have been documented (14). Thus the viscosities selected in this study were well within the physiological range. Figure 5 shows that both glycerol and BSA significantly (P < 0.05) increased the pK'a of SNARF-1 compared with the control (high-K+ TS) from 7.84 (SD 0.02) to 7.93 (SD 0.02) and 7.98 (SD 0.06), respectively. Glycerol also induced a large increase in Rmax (P < 0.05), whereas BSA decreased Rmax significantly (P < 0.05), compared with the control solution. Error propagation analysis to evaluate the relevance of these effects in pK'a and Rmax caused by glycerol or BSA indicated that these variations in pK'a or Rmax, when applied to a constant ratio, under- or overestimate pH by 0.10 or 0.20 pH units, respectively. In contrast, a change of 14% in Rmin produces an error of only 0.003 pH units. Similar effects of protein and viscosity on the in vitro calibration parameters of BCECF were also observed.
View larger version (22K):
[in this window]
[in a new window]
|
Fig. 5. Effects of viscosity and protein on SNARF-1 properties. SNARF-1-free acid was dissolved in high-K+ buffer and titrated with NaOH in the presence of glycerol (15%) to increase viscosity or in the presence of BSA (1%) to alter viscosity and dye binding to BSA. The data were fitted to a Henderson-Hasselbalch equation to obtain the in vitro calibration parameters of SNARF-1. The curves are representative of 4 in vitro experiments. The summary of the data at bottom represents the mean values; numbers in parentheses represent standard deviation (SD). *P < 0.05 glycerol vs. control and BSA vs. control.
|
|
The absolute viscosities measured for high-K+ TS or PSS at 37°C were consistent with the absolute viscosity value of water at 37°C of 0.6915 mPa·s (40). The influence of viscosity on the SNARF-1 in vitro titration parameters was statistically significant, despite the fact that the absolute viscosity values for 15% glycerol and 1% albumin in high-K+ TS are somewhat lower than the values found for fluid-phase viscosity in the cytoplasm, i.e., 1.2–1.4 mPa·s (20). These data suggest that the regional differences in cytoplasmic viscosity and/or the interactions between cytoplasmic proteins and the fluorescent dye contributed to the regional variations in ratio, pK'a, Rmax, and possibly Rmin. These results emphasize that conversion of subdomain fluorescence ratios to pHcyt requires the use of specific regional calibration parameters to compensate for cytoplasmic microenvironment differences.
Glial cells exhibit proton gradients and proton waves in the presence of physiological HCO3–.
To directly address the physiological relevance of spatial pHcyt heterogeneities, we performed experiments in the presence of 24 mM HCO3– in equilibrium with 5% CO2 at pHex 7.4 (21). This was necessary because the absence of HCO3– in the previous experiments may have altered the ability of the cells to regulate pHcyt. In these experiments, the buffers were maintained in equilibrium at pHex 7.4 by continuously bubbling the perfusate with 5% CO2. Figure 6 shows the zero-order spectra of the cell imaged throughout the slit's spectrograph. In these experiments, we binned spectral data from somata to filopodia at 4-µm intervals (total of 15 different ROIs) using an exposure time of 20 ms. We acquired 20 spectra per burst with a delay of 30 s for up to 5 min. Figure 6, A–C, shows that under steady-state conditions there are apparent pHcyt oscillations during the time frame of these experiments. The magnitude of these pHcyt fluctuations was apparent at the millisecond time frame as shown for three distinct bursts measured at time 0 and 1.5 and 3.5 min. Because the steady-state pHcyt values for each cellular domain were corrected for microenvironment differences with individual pK'a, Rmax, and Rmin parameters, these data indicate that the heterogeneities in regional pHcyt values were associated with physiological pHcyt differences in which each cellular domain exhibits a unique mosaic of pHcyt values.
View larger version (38K):
[in this window]
[in a new window]
|
Fig. 6. Glial cells exhibit proton gradients and pHcyt oscillations under physiological HCO3– conditions. Cells were loaded with SNARF-1 as described in MATERIALS AND METHODS and transferred to the microscope stage for spectral imaging microscopy as described in Fig. 1. For purposes of data presentation, only the zero-order spectrum of the cell studied is shown at left. The slit width was decreased sequentially as described in Fig. 1, and the SNARF-1 deprotonated-to-protonated ratios were converted to pHcyt using individual in situ titration parameters obtained for each region of the cell studied at 4-µm intervals from filopodium to soma (i.e., 15 ROIs). Note that there are apparent pHcyt fluctuations and pHcyt gradients during the time frame of this experiment in the presence of physiological HCO3– maintained in equilibrium with 5% CO2 at pHex 7.4. The selected time points at t = 0 (A), t = 1.5 min (B), and t = 3.5 min (C) indicate that there are pHcyt fluctuations in the millisecond time frame. These types of data were used to determine the frequency distribution for pHcyt amplitude and the intervals of pHcyt waves shown in Figs. 7–9.
|
|
Figure 7A shows the analysis of the properties of pHcyt oscillations and proton wave propagation from experiments similar to those shown in Fig. 6. These data indicate that under steady-state conditions, in the presence of HCO3–, 35% of the pHcyt oscillations exhibited amplitudes of 0.05–0.07 pH unit, whereas <5% exhibited pHcyt amplitudes of 0.12–0.24 pH unit. Treatment of cells with NH4Cl in the presence of HCO3– elicited an increase in pHcyt from resting to 7.7–7.9, which was associated with a significant decrease in the amplitude of the pHcyt fluctuations (P < 0.05; Fig. 7A). In contrast, cytosolic acidification elicited by removal of NH4Cl to acidify the cytosol to pHcyt 6.5–6.7 in the presence of HCO3– significantly decreased the frequency of the pHcyt oscillations without affecting the amplitude of the pHcyt oscillations (cf. Fig. 7B). The NH4Cl experiments were performed under isosmotic conditions. Further analyses of the characteristics of the pHcyt fluctuations indicated that under steady-state conditions, the proton waves appeared at intervals of 80 ms, with a conspicuous second harmonic wave appearing at 160-ms intervals. Cytosolic alkalinization and acidification elicited by NH4Cl treatment resulted in significant increases in the frequency of proton waves appearing at 160-ms intervals that were associated with a significant decrease in the frequency of proton waves that appeared at 80-ms intervals (Fig. 7, C and D).
To obtain further insight regarding the pHcyt regulatory mechanisms involved in the generation of proton wave propagation, we used inhibitors for two ubiquitous pHcyt regulatory mechanisms: NHE- and HCO3–-based H+-transporting mechanisms. These experiments were performed in the presence of 24 mM HCO3– in equilibrium with 5% CO2. Figure 8A shows that treatment of cells with hexamethylene amiloride (HMA) to inhibit NHE significantly decreased the amplitude of the pHcyt oscillations (P < 0.05). Likewise, acute treatment of cells with DIDS to inhibit HCO3–-based H+-transporting mechanisms also significantly decreased the amplitude of the pHcyt oscillations (P < 0.05; Fig. 8B). The magnitude of these effects in decreasing the amplitude of the pHcyt oscillations was larger for DIDS than for HMA (P < 0.05). Furthermore, HMA significantly increased the frequency of proton waves appearing at 160-ms intervals while decreasing the frequency of proton waves appearing at 80-ms intervals (P < 0.05 compared with steady state; Fig. 8C). In contrast, DIDS treatment significantly increased the frequency of proton waves appearing at both 80- and 160-ms intervals compared with the frequency of proton waves under steady-state conditions (P < 0.05; Fig. 8D). Altogether, these data indicate that both the amplitude and the frequency of proton waves were affected by inhibitors of either NHE- or HCO3–-based H+-transporting mechanisms.
View larger version (41K):
[in this window]
[in a new window]
|
Fig. 8. Inhibition of Na+/H+ exchanger (NHE)- and HCO3–-based H+-transporting mechanisms decreases the amplitude of the proton waves (A and B) and increases the interval between proton waves (C and D). As in Fig. 6, these experiments were performed in the presence of 24 mM HCO3– in equilibrium with 5% CO2. Data were derived from experiments similar to those shown in Fig. 6, except that after a steady-state pHcyt was reached, the perfusate was exchanged for one containing either 50 µM hexamethylene amiloride (HMA) to inhibit NHE-based H+-transporting mechanisms (A and C) or 50 µM DIDS to inhibit HCO3–-based H+-transporting mechanisms (B and D). In these experiments, we analyzed a total of 310 waves for HMA experiments (n = 5) and 300 wave events for DIDS experiments (n = 5). Frequency plots and data analyses were performed as described in Fig. 7. Statistical analyses indicated that either HMA or DIDS significantly decreased the amplitude of the pHcyt waves and increased the interval between proton waves compared with steady state (P < 0.05). Data were derived from 5 experiments.
|
|
BCECF, a widely used pH probe, corroborates that glial cells exhibit pHcyt oscillations.
To corroborate the idea that there are pHcyt oscillations in glial cells, we used a distinct pH probe and measured pHcyt using an alternative approach to spectral imaging analysis. We selected BCECF because, in contrast to SNARF-1, a ratiometric dual-emission pH probe, BCECF is a ratiometric dual-excitation pH probe that has been used widely to study pHcyt in different cell types. Furthermore, the pK'a of BCECF is 7.0, whereas the pK'a of SNARF-1 is >7.4 (21). These differences in the pK'a of pH fluoroprobes may result in distinct compartmentalization in cellular environments and distinct H+ buffering. BCECF fluorescence was monitored using an optical approach distinct from the spectral imaging approach used to measure pHcyt with SNARF-1. In the ratiometric optical approach, we used a rapid excitation filter changer and collected full-frame cell images with a frame transfer CCD camera coupled to a CCD intensifier. The use of full-frame images prevented us from collecting data as fast as we did using spectral imaging, because the readout of a full frame takes 600 ms. Regardless of this limitation in temporal resolution, analyses of pHcyt in selected ROIs corresponding to a surface area of 2.5–5 µm2, similar to the ROIs used in spectral imaging analysis, indicated that glial cells exhibited pHcyt fluctuations determined using BCECF, although they were smaller in magnitude than those observed with SNARF-1 (Fig. 9A). Furthermore, similar to our observations with SNARF-1, there was a significant decrease in the amplitude of the pHcyt oscillations after either alkalinization or acidification elicited by NH4Cl treatment compared with steady-state pHcyt (P < 0.05; Fig. 9B). Analysis of the frequency of proton waves indicated that NH4Cl-induced cytosolic acidification resulted in a significant decrease in the frequency of proton waves, whereas NH4Cl-induced cytosolic alkalinization did not affect the intervals of the proton waves (Fig. 9, C and D). Because these data were derived from in situ calibration parameters in discrete cellular domains, it is unlikely that differences in pHcyt were due to the effect of the microenvironment on the pH fluoroprobe. Altogether, these data indicate that glial cells exhibit pHcyt oscillations that are observable using either BCECF or SNARF-1.
View larger version (39K):
[in this window]
[in a new window]
|
Fig. 9. Cytosolic alkalosis and acidosis affect the amplitude of the pHcyt waves (A and B) as well as the interval of proton waves (C and D). Cells were loaded with BCECF to monitor pHcyt using a ratiometric optical approach. Alkalosis and acidosis were elicited by NH4Cl treatment and removal, respectively, as described for Figs. 6 and 7. The amplitude of the pHcyt waves was determined essentially as described in Fig. 7 from data plots similar to those shown in Fig. 6. The proton wave intervals were also determined as described in Fig. 7. Data were derived from 2,000 wave events for steady state and 1,600 wave events for either alkalosis and acidosis. Statistical analyses were performed as described in Fig. 7. These data indicate that cytosolic alkalosis and acidosis decreased the amplitude of the pHcyt waves (A and B) and increased the proton wave intervals (C and D) compared with steady state (P < 0.05). Data were derived from 10 experiments.
|
|
|
DISCUSSION
|
|---|
The existence of pHcyt heterogeneities in eukaryotic cells was suggested previously. This has relevance because many metabolic processes, including glycolysis, protein synthesis/degradation, cell motility, exocytosis, and ion channel gating, among many other cellular processes, exhibit an optimal pH. Thus a mosaic of pHcyt values across the cytoplasm, regardless of the degree of morphological polarization, could allow for precise functional compartmentalization based on distinct pHcyt values for discrete cellular domains. To address the issue of distinct pHcyt domains, we have systematically evaluated whether the behavior of the pH fluoroprobes is affected by discrete heterogeneity in the intracellular microenvironment. To do this, we used spectral imaging microscopy and SNARF-1, a ratiometric pH probe that exhibits a clearly defined isoemissive wavelength and remarkable spectral shifts from the protonated to the deprotonated state. By analyzing the full spectral output of SNARF-1 in the protonated and deprotonated domains below and above the isoemissive wavelength, we were able to determine changes in spectral shape between different cellular domains and thereby determine changes in pHcyt. This approach allowed us to study pHcyt from soma to filopodium with high spatial (i.e., every 2–4 µm) and spectral (0.4 nm) resolution. In principle, we can evaluate spectra at the optical limit of resolution in the lateral direction (i.e., 300 nm at x60 magnification and 1.4 NA). Thus, in as little as 5–100 ms, we were able to discern the spectral output of SNARF-1 in discrete cellular domains. These data indicated that there were remarkable pHcyt differences in discrete cellular domains. Because the buffer used in these earlier studies was HCO3– free, it is possible that the absence of HCO3– may have exacerbated the differences in steady-state pHcyt, which could explain the discrepancies between the present study and others that have not found pHcyt heterogeneities (8, 69, 73). However, the previous studies used global in situ calibration parameters. Furthermore, the temporal resolution of these studies was significantly less than that obtained in this study because of the use of a cryogenically cooled frame transfer CCD camera. Therefore, it was critical to perform these experiments in the presence of physiological HCO3– concentration with a high temporal resolution. Under these conditions, we noted that there were pHcyt oscillations in the presence of HCO3–.
To address the possibility that the observed pHcyt oscillations might have been due to preferential compartmentalization of SNARF-1 in unique environments because of its pK'a or to dye binding to proteins or other types of interactions, we used BCECF as an alternative ratiometric pH probe in conjunction with an optical imaging approach distinct from spectral imaging. Using the optical imaging approach, we could collect one full frame at 20 ms for long periods of time (minutes to hours). However, the readout of a full frame requires 600 ms, thus compromising temporal resolution compared with spectral imaging. Our data indicate that regardless of this limitation in temporal resolution, pHcyt oscillations also were present in association with BCECF, although the oscillations were of smaller amplitude than those determined using SNARF-1. We also determined that the proton wave intervals monitored using BCECF occurred at longer intervals (i.e., seconds) than those observed when using SNARF-1 and spectral imaging, conditions in which data cannot be acquired for more than 1 s because of readout limitations. Specifically, acquiring spectral imaging data at 20-ms resolution continuously for 1 s requires 10 min to read out and write data to the hard disk. Thus long-term kinetic measurements performed with BCECF are not feasible with spectral imaging. Nevertheless, using spectral imaging, we obtained data continuously for 400 ms with 20-ms exposures and microsecond delays between exposures but with a significant delay of 1.5 min (needed for readout of the data set) before the next acquisition. Importantly, the combined data from both SNARF-1 and BCECF allow us to conclude that there are pHcyt oscillations in glial cells. Furthermore, the combined data from both fluoroprobes indicate that the intervals of proton waves occur within the order of hundreds of milliseconds (80- and 160-ms resolution) as well as within thousands of milliseconds. The slow proton wave intervals are associated with pHcyt amplitudes that are smaller than the pHcyt amplitudes observed in the faster wave intervals. This suggests a model in which fast proton waves with high amplitude travel along slow waves with low amplitude.
The pHcyt oscillations appear under steady-state conditions, both in the presence and in the absence of HCO3–. These data suggest that the pHcyt oscillations are a true phenomenon. Furthermore, pharmacological approaches to inhibiting NHE- and HCO3–-based H+ transport under steady-state conditions also affect the magnitude and frequency of the pHcyt oscillations. These data suggest that pHcyt oscillations are regulated by ubiquitous pHcyt regulatory systems. The NH4Cl experiments were performed to evaluate whether conditions of acidosis or alkalosis alter the frequency of these pHcyt oscillations. Our data show that acidosis and alkalosis indeed significantly altered pHcyt oscillations. Because it is known that alterations in NHE- or HCO3–-based H+ transport play an important role in regulating cell volume in many different cell types, it is not possible on the basis of our present studies to discard the possibility that changes in cell volume determine pHcyt oscillations, because the impact of intracellular pH on cell volume regulatory mechanisms is unclear. However, changes in cell volume are associated with changes in pHcyt. Specifically, cell swelling leads to cytosolic acidification (26, 36, 41, 50), which has been explained by the exit of HCO3– through anion channels (36), by release of H+ from acidic intracellular compartments, and by enhancement of Cl–/HCO3– exchange due to decreasing cellular Cl– activity (41). This latter mechanism, however, should be impaired by inhibition of the exchanger at acidic pHcyt (35), and cellular acidosis is not inhibited by removal of extracellular Cl– (50). After cell swelling, the cytosolic acidification may impair the activation of volume-regulatory K+ channels (43) and may contribute to inhibition of glycolysis and thus to the decreased release of lactic acid (36). The alkalinization after cell shrinkage could stimulate glycolysis (36). Furthermore, cell shrinkage may cause cytosolic alkalinization, which is at least partially due to activation of volume-regulatory NHE. Among the cloned members of the NHE family (76), NHE1 (16), NHE2 (16), and NHE4 (7) are stimulated, whereas NHE3 (16) is inhibited by cell shrinkage. The putative volume-sensitive site at the NHE1 molecule has been identified and is distinct from the sites regulated by Ca2+ and growth factors (5). The cloned anion exchanger AE2, but not AE1, is postulated to participate in regulatory cell volume increase (31). Altogether, these studies suggest that there is a delicate interplay between changes in cell volume and pHcyt regulation. However, the question of whether changes in pHcyt determine changes in cell volume requires further investigation. Importantly, the behavior of these proton waves as well as pHcyt amplitude can be altered by acidosis, alkalosis, and inhibition of NHE- and HCO3–-based H+-transport. The physiological significance of these findings warrants further investigation because pHcyt oscillations may encode specific information needed for cell function and cell signaling.
To interpret the differences in SNARF-1 and BCECF protonated-to-deprotonated ratios properly, we have taken into account the behavior of the pH fluoroprobes in the cytoplasm because it is heterogeneous in terms of composition and organization. Regional intracellular microenvironments may differ in viscosity, which could result in distinct behavior of the fluoroprobes (53). Indeed, it is known that ion-sensitive fluoroprobes may display spectral differences not only between in vitro and in situ environments (6, 57) but also within distinctive intracellular organelles (1, 17, 55, 70). Important effects of proteins on in vitro calibration parameters have been described for calcium fluoroprobes (4) and pH fluoroprobes (65). However, it has been suggested that SNARF-1 binds not to BSA but to a contaminant present in commercially available SNARF-1 (75). Furthermore, in the cytoplasm of cardiac myocytes, a major fraction of the fluoroprobes (0.5–0.9) appears to be bound to proteins (4). Other variables that could cause variations in the in situ titration parameters include partition of the dye between cytoplasm and endomembranous compartments (51, 71), the amount of dye bound to proteins (65), quenching agents, and inner filter phenomena (6, 64).
The data shown in Fig. 5 exemplify the behavior of SNARF-1 in vitro under conditions in which the protein content, viscosity, and ionic strength of the buffers were precisely controlled. These data indicate that in vitro calibration parameters such as pK'a, Rmax, and Rmin of SNARF-1 were affected by viscosity and protein. pK'a under all of these in vitro conditions varies from 7.84 to 7.98. Fortunately, as shown in the Table 1 data from in situ studies, the pK'a of SNARF-1 was as low as 7.08 and as high as 7.9. Also as shown in Table 1, the pK'a of SNARF-1 varied in different cellular domains, from soma to filopodium, in an unpredictable pattern. For example, the pK'a for SNARF-1 in situ varied from 7.08 to 7.78 in cell E in Table 1. Importantly, the differences between steady-state pHcyt and pK'a varied from 0.24 to 0.7 (average 0.49 pH unit). It is generally assumed that a useful pH indicator should be at pK'a ±1 pH unit (3, 56). Thus the average difference of 0.49 pH unit between pHcyt and pK'a in our studies using SNARF-1 was within the accepted range of 1 pH unit.
To further alleviate concerns related to the reliability of these pHcyt measurements, we used BCECF as an alternative pH probe (21, 60). It is generally accepted that the pK'a of BCECF is 7.0 in vitro (21). As shown in Table 2, similarly to our data obtained using SNARF-1, the in situ pK'a values for BCECF varied widely from 6.74 to 7.26. The pHcyt measurements in soma and filopodium also varied and were as low as 6.93 and as high as 7.54 in the different cells shown in Table 2. The differences between steady-state pHcyt and pK'a varied from 0.091 to 0.394 (average 0.282 pH unit). This value of 0.282 is well below the accepted pK'a ±1 pH unit considered useful.
The reasons for the distinct in situ calibration parameters in different cellular domains when using SNARF-1 and BCECF as pH fluoroprobes are not immediately apparent. However, because the cytoplasmic microenvironment is different in terms of protein composition and viscosity, distinct diffusion mobility of fluorescent probes and distinct spectral properties are possible. Indeed, studies in which fluorescence recovery was performed after photobleaching have shown that the translational diffusion of intracellular BCECF is 6–10 times lower near the membrane and 4 times lower in the bulk cytoplasm than in water (68). However, the fluid-phase cytoplasmic viscosity in the absence of collisions or binding to cytoplasmic macromolecules is similar to the viscosity of water (42, 68). Our data indicate that there are significant differences in the pK'a of BCECF in different cellular regions. There are also important differences in molecular crowding within the different subcellular compartments, suggesting considerable diffusional heterogeneity for small metabolites, and thereby fluoroprobes, within different intracellular organelles (20). In addition, viscosity can alter the spectra of ion indicators (59). The increase in the pK'a of SNARF-1 by glycerol could be due to both an increase in the solution viscosity and a decrease in the solvent dielectric constant. The decrease in the dielectric constant promotes increases in the pK'a of weak acids and pH acid indicators and decreases in weak base salts and pH base indicators (3). The in vitro titration curves performed in the presence of glycerol and BSA confirmed that changes in both the viscosity and the amount of dye bound to protein induced relevant changes in Rmax and pK'a for SNARF-1. One limitation of changing the viscosity of the medium by increasing the glycerol concentration is the decrease in the ionic strength of the buffer, which alters the SNARF-1 pK'a. Therefore, in vitro calibration of fluoroprobes either by the addition of proteins or by changing the viscosity to mimic the intracellular milieu is an unreliable predictor of fluorescent indicator behavior in the heterogeneous intracellular environment. We therefore emphasize the need to perform regional in situ pH calibration curves to correct for different cytoplasmic microenvironments.
The in situ calibration parameters shown in Figs. 2 and 4 were measured in the presence of nigericin and valinomycin in high-K+ buffer. Under these conditions, pHex = pHcyt. This allowed us to clamp pHcyt at desired pH values. The data shown in these figures indicate that there was an increase in ratio as pHcyt increased from 5.5 to 8.5 in different cellular domains encompassing the soma to the filopodium. In this regard, both SNARF-1 and BCECF behaved as expected because ratio increased with increasing pH. Unfortunately, the in situ calibration parameters of these fluoroprobes (pK'a, Rmax, and Rmin) varied within different cellular regions in an unpredictable fashion, thus emphasizing the need to use in situ titrations from specific cellular regions to assign pHcyt values in those ROIs.
The relevance of in situ calibrations to the understanding of pHcyt in distinct cellular domains from soma to filopodium is highlighted by our data because the pK'a, Rmax, and Rmin parameters used to determine pHcyt from deprotonated-to-protonated SNARF-1 and BCECF ratios vary considerably in different cellular domains. Furthermore, the behavior of these parameters varies from cell to cell and from region to region. If we had used single global in situ calibration parameters (i.e., Rmax, Rmin, and pK'a), the pHcyt values in different cellular domains would have demonstrated the same behavior of the deprotonated-to-protonated SNARF-1 or BCECF ratio values. However, because the in situ calibration parameters were different for distinct cellular domains, the pHcyt values do not necessarily reflect the behavior of the deprotonated-to-protonated SNARF-1 or BCECF ratio values. (E.g., higher ratio values do not correlate with higher pHcyt.) Because the magnitude of these effects cannot be predicted, regional in situ calibration parameters are needed to assign pHcyt when using pH fluoroprobes.
The reasons underlying the different pHcyt values obtained with BCECF and SNARF-1 are not immediately apparent. We previously showed in 3T3 fibroblasts (21) that the pHcyt values reported for SNARF-1 tend to be more alkaline than those obtained using BCECF and fluorescence spectroscopy. Importantly, the pHcyt values determined using SNARF-1 are similar to those obtained using NMR spectroscopy in 3T3 fibroblasts (21). Thus different experimental approaches provide distinct pHcyt values in the same cell type. This emphasizes the need to use different approaches to establish the physiological significance of pHcyt values in any given phenomenon. Despite these differences in absolute pHcyt values, our data obtained using SNARF-1 and BCECF support our contention that there are pHcyt oscillations in glial cells.
Linear regression analyses of pHcyt values from soma to dendrite obtained in experiments similar to those shown in Tables 1 and 2 indicate that in 40% of cells, pHcyt tends to be 0.1–0.2 pH units more acidic in the dendrite than in the soma. A more alkaline pHcyt in the dendrite than in the soma may have physiological significance for the exocytotic process, because pH changes have been implicated in exocytosis. Specifically, the pH in secretory vesicles has been found to be more acidic in the soma than in the distal portion of dendrites of cultured sympathetic neurons (52). The bimodal pH distribution of these secretory vesicles may have significance for both the rapid acidification step needed for organelle acidification and the more alkaline pH needed for exocytosis. Our data show that 40% of glial cells exhibited a more acidic pHcyt at the dendrite than at the soma, consistent with the spatial distribution of secretory granules in neurons. In pituitary neuroendocrine cells, the release of prolactin via exocytosis appears to be regulated by changes in pH (2). In these neurosecretory cells, an increase in pH appears to regulate the exocytosis of both prolactin and signaling molecules contained in the dense core of secretory vesicles (2). Changes in intravesicular pH are needed because the proteins within the lumen are maintained in a stable, insoluble dense core that traps small neurotransmitters, granines, and other peptides (37, 66). Increases in pH within the vesicle allow disaggregation of the dense core material and facilitate the release of intravesicular components during exocytosis (27, 38). A more alkaline pHcyt near the dendrite terminus could facilitate this process by decreasing the magnitude of the pH gradient between the pHcyt and the secretory granule, thereby maintaining a more alkaline intravesicular pH with less energy expenditure. A more alkaline pHcyt could also facilitate the exocytotic process by solubilizing the components of the dense core even before the fusion of secretory granules with the plasma membrane during exocytosis.
In conclusion, using SNARF-1 or BCECF, we have identified distinct pHcyt oscillations in somata and filopodia in C6 glial cells. The future elucidation of the spatiotemporal pHcyt heterogeneities should expand knowledge of pH signaling in cells.
|
GRANTS
|
|---|
This work was partially supported by National Heart, Lung, and Blood Institute Grant R01 HL-65695, American Cancer Society Grant ACS-RPG-00-035-01-CNE, and Texas Higher Education Coordinating Board Advanced Research Program Grant THECB/ARP-010674-0012-2001 (to R. Martínez-Zaguilán) and by Consejo Nacional de Ciencia y Technológia Grant 0437P-N (México) and Universidad Autónoma de San Luis Potosí Grant C02-FAI-04-5.10 (to S. Sánchez-Armáss).
|
ACKNOWLEDGMENTS
|
|---|
We thank Karina Bakunts and Gloria Martínez for technical assistance and Dr. Fernando Toro and Maria Elena Dibildox for rheological measurements.
|
FOOTNOTES
|
|---|
Address for reprint requests and other correspondence: R. Martínez-Zaguilán, Dept. of Physiology, Texas Tech Univ. Health Sciences Center, 3601 4th St., STOP 6551, Lubbock, TX 79430-6551 (e-mail: Raul.MartinezZaguilan{at}ttuhsc.edu)
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
* S. Sánchez-Armáss and S. R. Sennoune contributed equally to this work. 
TOP
ABSTRACT
MATERIALS AND METHODS
, wavelength; a.u.f., arbitrary units of fluorescence. The comparison of the spectral shapes in 4 cellular domains selected from E indicates that domains 1 and 2 exhibit pHcyt more acidic than that in domains 7 and 8 (F) as shown by normalization of the spectra at the same scale (G). H: SNARF-1 steady-state ratio profile of 16 discrete cellular domains from the cell in A shows distinct deprotonated-to-protonated ratios from soma to filopodium. Data are representative of 11 cells. *P < 0.05 for domains 7 and 8 vs. domains 1 and 2, respectively.
pHcyt) were evaluated from the pHcyt waves determined under steady-state conditions in cells perfused with physiological saline solution (PSS) containing 24 mM HCO3– from experiments similar to those shown in