|Home | About | Journals | Submit | Contact Us | Français|
Regulation of intracellular pH (pHi) in neurons is crucial to maintain their physiological function. In the current study, newly-developed polydimethylsiloxane (PDMS) microfluidic devices were used to independently investigate pHi regulation in neuronal soma and neurites. Embryonic cortical neurons were cultured in PDMS microfluidic devices with soma growing in one chamber (seeded) and neurites extending through a set of perpendicular microchannels into the opposite parallel chamber (non-seeded). Neurons in the microchambers were characterized by the vital dye calcein-red, polarized mitochondria, neuronal specific β-tubulin (type-III), axonal Tau-1 protein, dendritic microtubule associated protein (MAP-2), and Na+/H+ exchanger isoform 1 (NHE-1). Neurites exhibited higher resting pHi than soma (7.16 ± 0.09 vs. 6.90 ± 0.15). The neurites had a proton extrusion rate 3.7-fold faster than in soma following NH4Cl prepulse-mediated acidification (p < 0.05). The difference in the pHi regulation rates between neurites and soma can be accounted for by the larger surface area to volume ratio in the neurites. Interestingly, inhibition of NHE-1 activity pharmacologically blocked the pHi regulation in soma and in neurites by ~ 70% (p < 0.05). Taken together, our study demonstrated that the microfluidic devices provide a useful tool to study neuronal pHi regulation in soma and their neurites. We conclude that NHE-1 plays an important role in regulation of pHi in both compartments.
Intracellular pH (pHi) can rapidly fluctuate in response to neuronal activity or pathological insults.1 The tight regulation of intracellular and extracellular pH (pHi and pHo) homeostasis is required for neuronal function in the central nervous system (CNS). Changes in pHi (even in a small range) affect the functions of multiple enzymes, ion channels, and other macromolecules, thus affecting synaptic transmission, neuronal excitability, and gap junction conductance.2–4 pHi may function as an important variable in regulating neurotransmitter release. 5,6 The Na+/H+ exchangers (NHEs) are acid extrusion proteins which catalyze the electrochemically neutral exchange of Na+ into the cell and H+ out of the cell at a ratio of 1 Na+: 1 H+.7 NHEs are extremely sensitive to pHi and characterized by the fact that they are allosterically activated by cytosolic H+, thereby promoting the rapid extrusion of acid when pHi drops below a threshold level. 7,8 The Na+/H+ exchanger 1 (NHE-1) is the most ubiquitous isoform in the CNS 9 and plays an important role in regulation of pHi in neurons and astrocytes.8 NHE1 activity is stimulated by growth factors, peptide hormones7, and pathological stimuli such as ischemic-hypoxia.10,11
We recently reported that pHi regulation rates in the dendrites of cultured cortical neurons grown in culture dishes were 330% higher then in the soma primarily because of the higher surface area to volume rates in dendrites.12 The traditional in vitro neuronal culture models using culture dishes can not be used to independently investigate differential pHi regulation between soma and neurites. Thus, the in vitro compartmentalized neuronal culture models make this type of study possible. Campenot first introduced a compartmentalized fluidic neuronal culture using a device consisting of a Teflon® divider attached to a collagen coated Petri dish with silicon grease.13 These devices required great skill to assemble and were difficult for imaging and prone to leakage.13 Recently, compartmentalized microfluidic devices have been made of biologically inert polydimethylsiloxane (PDMS) that allows neurites to grow into isolated fluidic environments.14 Such devices have been applied in studying of axonal injury and regeneration14, axon myelination15, and drug-induced axonal degeneration.16 Moreover, the enclosed microfluidic system provides a more physiologically relevant micro-environment (larger cell surface area to culture medium ratios) than culture dishes.17
In the current study, we developed an enclosed, compartmentalized microfluidic platform to examine the differential pHi regulation in neuronal soma and neurites. Cultured cortical neurons were characterized with vital dyes (calcein and Mitotracker) and immunocytochemistry staining. Resting pHi levels and pHi recovery rates were measured in neurites and soma independently. Our report suggests that NHE-1 plays an important role in neuronal pHi regulation.
Hanks balanced salt solution (HBSS) was from Mediatech Cellgro (Manassas, VA). Neurobasal medium, B-27 supplement, MitoTracker green, calcein-AM, SYTO 60, and BCECF-AM were from Invitrogen (Carlsbad, CA). HOE 642 was a kind gift from Aventis Pharma (Frankfurt, Germany). Nigericin and antibody for MAP-2 were purchased from Sigma (St. Louis, MO). Antibody for β-tubulin (type-III) was obtained from Promega (Madison, WI). Tau-1 antibody was obtained from Santa Cruz Biotechnology (Santa Cruz, CA). SU-8 2 and SU-8 2001 were purchased from Microchem (Newton, MA). Polydimethylsiloxane (PDMS) was purchased from Dow Corning (Sylgard 184, Midland, MI). G116 antibody was a kind gift from Dr. Leong L. Ng (Leicester, UK).
The microfluidic devices were fabricated in PDMS using rapid prototyping and soft lithography.18,19 The layout of the microfluidic devices is shown in Figure 1 A, which consisted of two parallel chambers (seeded and non-seeded, 0.4 mm wide × 16 mm long × 0.25 mm tall) and bridging microchannels (30 µm wide × 200 – 300 µm long × 3 µm tall). Two corresponding transparency masks were printed from an EPS file using a 6,000 dpi high-resolution printer (Imagesetter Inc, Madison, WI). A dual-layered negative master was fabricated by patterning dual layers of the epoxy based photoresist SU-8 2 (one for the microchannels, one for the two chambers). The photoresist was spun on a silicon wafer at a spin speed of 1500 rpm for 40 sec to obtain an approximate height of 3 µm for the bridging microchannels. The photoresist SU-8 2100 was subsequently spun on the silicon wafer (1100 rpm for 40 sec, 0.25mm in height) to create the non-seeded and seeded chambers. Both mask layers were carefully aligned with a technique (error < 15 µm) as described recently.20 Microfluidic devices were fabricated by molding PDMS against this negative master. After degassing, PDMS (10:1 ratio base to catalyst) and the wafer were heated in a hot plate at 120 °C for 30 min. Inlets and outlets for fluid exchange in each chamber (1/8 inch in diameter) were made in the resulting PDMS microdevice (1.5 mm thick). PDMS microdevices were placed on poly-D-lysine coated glass coverslips (22 × 22 mm) and sterilized by autoclave.
Pure cortical neurons from mouse embryonic day 14–16 fetuses (E14-16) were prepared as described before.21 In brief, the cortices were removed from E14-16 fetuses and treated with 0.5 mg/ml trypsin at 37°C for 25 min. The cells were centrifuged at 300 g for 4 min at 4°C. The cell pellet was diluted in B-27 supplemented Neurobasal medium (2 %) containing 0.5 mM L-glutamine and penicillin/streptomycin (100 units/ml and 0.1 mg/ml, respectively). As shown in Figure 1 A, the cells were seeded at a density of 500 cells/mm2 in the seeded chamber by passive pumping22. Briefly, the seeded chamber was rinsed with Neurobasal medium by adding a drop (~20 µL) into the inlet and aspirating from the outlet (Figure 1 A). The non-seeded chamber was rinsed with the fresh medium similarly. Cells were seeded via passive pumping by adding ~ 10 µL of a 2×106/ml cell suspension to the inlet which then flowed into the seeded chamber due to a low surface tension created by the large drop of medium placed on the outlet. The microfluidic devices were placed in an incubator at 37°C (model 3130, Thermo Forma, Waltham, MA) with 5% CO2 and atmospheric air. A blank PDMS piece (~ 0.5 mm thick) was placed over the chamber ports on day 2 in vitro (DIV 2) to prevent medium evaporation. DIV 8–11 cultures were used in the study.
Neuronal cultures in the microfluidic device were incubated with 10 µM BCECF-AM in HC03−-free HEPES buffered medium at 37°C for 30 min. Microfluidic devices were then mounted on a temperature controlled (37°C) imaging chamber (Warner Instruments, Hamden, CT). The chamber was placed on the stage of the TE 300 inverted epifluorescence microscope and visualized with a 40× oil-immersion objective. The cells were excited every 1–20 s at 490 and 440 nm, and the emission fluorescence at 535 nm recorded. Images were collected using a Princeton Instruments MicroMax CCD camera. Regions of interest were drawn either in the cytosol of 20 cells (seeded side) or on neurites (non-seeded side) and analyzed with MetaFluor image-processing software. At the end of each experiment, the ratio of the background-corrected fluorescence emissions (F490/F440) was calibrated using the high K+/nigericin technique.23 In this study, the inherent error in determining pHi with the BCECF method is ~ 1.5% (supplemental Figure 1). Microchamber fluid replacement in these experiments was accomplished by placing a drop of solution with a 25 µl Hamilton syringe on the inlet of one microchamber and quickly drawing the fluid through the outlet with the corner of a Kimwipe®.
For the pre-pulse treatment, either soma (the seeded-chamber) or neurites (non-seeded chambers) were subjected to an acid load by transient application (5 min) of 30 mM NH4+/NH3 solution.24 NH4+/NH3 solutions were prepared by replacing 30 mM NaCl in the HCO3−-free HEPES-buffered solution with an equimolar concentration of NH4Cl. Application of NH4+/NH3 solution caused pHi to rise as NH3 gas diffused quickly into the cell and combined with H+ to form NH4+. After 5 min, the NH4+/NH3 solution was replaced with HCO3−-free HEPES-buffered solution, causing NH3 to leave the cell and trapping H+ in the cell. This is the classic method of placing an artificial acid load on cells to characterize pHi regulation.25 Interestingly, we observed that the pre-pulse treatment applied to one microchamber (soma or neurites) did not cause either alkalization or acidification in the opposite chamber (data not shown). This suggests that we are able to investigate pHi regulation in these compartments (ie. between soma and neurites), separately. Before returning cells to normal HCO3−-free HEPES-buffered solution to acidify the cells, the image acquisition interval rates was increased to 1 HZ in order to monitor pHi recovery. pHi recovery rate was determined from the slope of a fitted linear regression fit to the initial linear portion of the trace after NH4+/NH3 application (5–30 sec).24
Resting pHi was averaged from the first ~2 min incubation of cells with HCO3−- HEPES-buffered normal solution before the pre-pulse was performed. In the experiments with specific NHE-1 inhibitor HOE 642, either soma or neurites were incubated with HOE 642 (1 µM) for 10 min prior to and then throughout the NH4+/NH3 pre-pulse.
To determine surface area to volume ratios in pure neuron cultures growing in microchamber devices, neurons were loaded with 0.5 µM calcein-AM green (a vital dye) and 5 µM SYTO 60 (nucleus dye) for 30 min at 37°C. The coverslips were then placed in the perfusion chamber on the stage of a Leica DMIRE2 confocal microscope and visualized with a 100× 1.4 NA apochrome oil-immersion objective. An ~15 µm (neurites) to 30 (somata) µm thick image stack (150–250 slices at 512 by 512 pixels) was collected sequentially (ex. 488 nm argon laser line, em. 500–545 nm and ex. 543 HeNe laser, em 544–677) and imported into ImageJ (Version 1.41, NIH). A cellular region (somata, nucleus, or process) was defined and the surface area was calculated by summing the product of the region perimeter by the distance between each image section (0.12 µm). The volume of the region was calculated with the region area and intra-section distance. The somata volume was corrected by subtracting the calculated volume for the nucleus volume. In other experiments, neurons in microchambers were further characterized by duel-labeling with 0.5 µM calcein-AM red and 0.2 µM MitoTracker green, a mitochondrial-selective fluorescent dye, for 30 min. Calcien-AM and Mitotracker images were sequentially captured with a Leica DMIRE2 confocal microscope using a 20× objective and 543 nm HeNe and a 488 nm argon lasers respectively.
PDMS microfluidic devices were removed from the glass coverslips and the cells on the coverslips were gently rinsed with PBS (pH 7.4) and fixed with 4% paraformaldehyde in PBS.24 After rinsing, cells were incubated with blocking solution (10% normal goat serum, 0.4% Triton X-100, and 1% bovine serum albumin in PBS) for 1 h at room temperature. Cells were incubated for 1 h at 37°C with polyclonal G116 NHE-1 antibody (1:50), mouse monoclonal MAP-2 antibody (1:200), monoclonal β tubulin (type-III) antibody (1: 200) or rabbit polycolonal Tau-1 antibody (1:100). After rinsing in PBS, cells were incubated with goat anti-mouse Alexa 488-conjugated IgG (1:100; Invitrogen), goat anti-rabbit Alexa 546 conjugated IgG (1:100; Invitrogen, for confocal microscopy) or goat anti-rabbit Alexa 594 conjugated IgG (1:100; Invitrogen, for epifluorescence microscopy) for 1 h at 37°C. 1 µg/mL To-pro-3 iodide was used to stain nuclei for confocal image collection. Fluorescence images were captured by the Nikon TE 300 inverted epifluorescence microscope (20×) using a Princeton Instruments MicroMax CCD camera and MetaMorph image-processing software. Negative controls were incubated with secondary antibodies only. Identical digital imaging acquisition parameters were used in both negative control and experimental images. Images of Tau-1 (20×) were captured with a Leica DMIRE2 confocal microscope using an 543 nm HeNe laser for Alexa Fluor 546, and a 640 nm GreNe laser for To-pro-3 iodide signals.
Paired Student’s t-test was used for comparisons between groups and analysis of variance with the Bonferroni post-test for multiple comparisons (SigmaStat, Systat Software, Point Richmond, CA). When comparison groups did not have normal distribution or equal variance, a Mann-Whitney Rank Sum Test was used. A P-value of < 0.05 was considered as significantly different.
Neurons were grown in the PDMS microfluidic system as described in Methods. DIV 8–11 cultures were loaded with the vital dye calcein red and MitoTracker green (Figure 1 B). Both soma and neurites retained calcein-red, indicating intact plasma membrane integrity of neurons in the culture (Figure 1 B, a). Figure 1 B (a, b) also illustrate that the neurites grew across the bridge-channels and protruded into the non-seeded chamber. Concurrently, neuronal cultures were stained with the mitochondrial dye MitoTracker-green, which reacts with thiol groups of proteins and forms conjugates in the mitochondria, thus labeling mitochondria independent of their membrane potential. As shown in Figure 1 B (c, d, arrows), punctate MitoTracker-green signals were detected in soma and neurites. Interestingly, a high level of mitochondria was distributed in neurites in the non-seeded chamber (punctuate, rod-shape individual mitochondria in d, f). In contrast, Mitotracker-green signals were less evident in the neurites in the bridging channels, suggesting that MitoTracker green does not diffuse quickly within the intracellular space or through the bridging channels (Figure 1 B, c). An overlay of the calcein red and MitoTracker-green images revealed that the distal neurites expressed less calcein red (Figure 1 C). Taken together, these data illustrate that cortical neurons (in the seeded chamber) were able to grow and extend their neurites through the micro-bridging channels into the non-seeded chamber. Mitochondria appear to be enriched in the distal neurites in a similar manner as in vivo.
Neurons cultured in the PDMS microfluidic devices were further characterized with immunofluorescence staining assays. First, expression of three neuronal marker proteins [β-Tubulin (type-III), Tau-1 and MAP-2] in the cultures was examined. As shown in Figure 2 A, β-Tubulin was abundantly expressed in neuronal soma and throughout the neurites. Moreover, dendritic marker protein MAP-2 was expressed in soma and dendritic neurites in the bridging channels and non-seeded chamber (Figure 2 B). In one occasion, small immature neurons migrated through the bridging channel and grew in the non-seeded chamber (arrow, Figure 2 B). In Figure 2 C, D, the axonal marker protein Tau-1 was expressed in the neurites growing into the non-seeded chamber, but with less immunointensity in neuronal soma (in the seeded-chamber) or in the neurites of the bridging channels. Negative control studies for these proteins were conducted by omitting the primary antibodies and illustrated no fluorescence signals (insets of Figure 2 A, B, C). These findings demonstrate that cortical neurons grow in the microfluidic devices with neurites (both axons and dendrites) extruding into the non-seeded chambers.
Next, we evaluated expression of an important pHi regulatory protein NHE-1. As shown in Figure 2 E, F, NHE-1 was present in soma of the seeded-chamber and neurites in the non-seeded chambers. Negative control studies were shown in insets of Figure 2 E, F. No or low fluorescence signals in these controls were detected under the same data acquisition settings.
We then independently determined pHi regulatory responses in soma of the seeded chambers and neurites in the non-seeded chambers. pHi recovery rate was measured following the NH3/NH4+ prepulse-induced acidification. In the absence of HCO3−, pHi regulation largely reflects NHE-mediated H+ efflux.26 As shown in Figure 3 A, when neuronal somata in the seeded chamber was exposed to 30 mM NH3/NH4+, pHi rose rapidly as NH3 diffused into the cell and combined with H+ to form NH4+ (a–b) and then declined slowly (b–c). Removal of the NH3/NH4+Cl solution from the seeded chamber caused somata pHi to decrease because the newly formed NH4 rapidly left cells as NH3, trapping H+ inside of cells (c–d). Neurons were able to quickly restore somata pHi at a rate of 1.25 ± 0.29 pH units/min < 3 min and returned to their basal levels (6.90 ± 0.15 pH). In some cases, there was an alkaline overshoot (Figure 3 A). In the presence of the potent NHE-1 inhibitor HOE 642 (1 µM), somata pHi recovery rate was significantly reduced (0.35 ± 0.13 pH units/min, p < 0.05, Figure 3 A, C) and the alkaline overshoot was blocked.
After NH3/NH4+ prepulse-induced acidification, the neurites exhibited much faster pHi recovery kinetics (4.59 ± 0.86 pH units/min, Figure 3 B, C). Inhibition of NHE-1 with HOE 642 reduced the recovery rate to 1.47 ± 0.14 pH units/min in neurites (p < 0.05, Figure 3 B, C). Moreover, a more alkaline basal pHi (7.16± 0.09) was detected in the neurites than in soma (Figure 3 D).
The apparent higher pHi recovery rates in the neurites could result from significantly larger A/V ratios in the neurites. A/V ratios were determined in somata (arrow) and neurites (arrow-head) as illustrated in Figure 4 A, B. Figure 4 C shows that the A/V ratio in the neurites is 2.8 times higher than that of soma. Interestingly, the A/V ratios of soma and neurites were higher in the microfluidic devices than in cells cultured on coverslips in traditional plastic 35 mm culture dishes (Figure 4 C).12 Although digital 3-D reconstruction from 2-D data sets will tend to underestimate surface areas by smoothing the images, this estimate of A/V ratios served as approximation for correcting pHi regulation rates.
Consequently, we corrected the pHi recovery rate for the A/V ratios in soma and neurites (Table-1). After the correction, the differences in H+ extrusion rates in the two regions were abolished. Once the soma pHi recovery rates were corrected for A/V ratio they were comparable to the corrected rates that were observed in neurons in 35 mm culture dishes (0.30 ± 0.07 vs. 0.30 ± 0.01 pH units/min respectively) 12. However, the A/V-corrected pHi recovery rates in neurites grown in microfluidic devices (0.40 ± 0.08 pH units/min) remained significantly higher than in neurites grown in the culture dishes (0.19 ± 0.04, p < 0.05).12
Our results showed viable neuronal growth in the seeded-chamber with neurites expanding through the bridging channels and into the non-seeded chamber. In most cases, the calcein signals were colocalized with the MitoTracker signals, indicating the intact integrity of plasma membrane and mitochondria. Damaged neurons would not retain either dye. Taken together, these results demonstrate that we have developed a viable compartmentalized microfluidic system for culturing neurons.
Microfluidic devices with micro-bridging channels lengths greater than 200 µm exhibit almost exclusively axonal growth and a few dendrites in the non-seeded chambers at DIV 14.14 In contrast, both dendrites and axons extend through 150 µm long bridging channels at DIV 14. 14 In our study, we used 200–300 µm long bridging channels. We observed abundant expression of neuronal specific β-III tubulin and dendritic protein MAP-2 in both soma and neurites, while the neurites in the non-seeded chambers of DIV 8–11 neuron cultures exhibited abundant expression of the axonal marker protein Tau-1. Because both axons and dendrites were present in the non-seeded chambers in the current study, we are unable to differentiate between the two with regard to pHi measurement in the non-seeded chamber. Therefore, in the non-seeded channels we referred them as neurites.
There was a noticeable lack of MitoTracker-labeled mitochondria within the micro-bridging channels. This most likely resulted from restricted access of the MitoTracker dye to the mitochondria within the micro-bridging channels. To reach mitochondria within the micro-bridging channel, a dye must either enter the bridging-channel and then be taken up by the cell or enter the cell and then be transported through the cytoplasm to reach mitochondria in the neurites of the bridging-channel. Movement of the dye depends on molecular diffusion which is driven by concentration gradients of the dye. Upon entering the cell, MitoTracker is concentrated in nearby mitochondria where it covalently binds to proteins and would be unavailable to reach mitochondria located further away. However, we can not rule out that the low density of mitochondria in the micro-bridging channel determined with MitoTracker staining could be the result of its shallow height (3 µm) which would limit mitochondrial congregation.
The high fluidic resistance of the micro-bridging channels produces a small but sustained flow between the seeded and non-seeded chambers that counteracts diffusion. The fluidic isolation between the two chambers can last for 20 h owing to the high fluidic resistance of the micro-bridging channels.14 Thus, the fluidic resistance allows us to measure pHi regulation in soma and neurites independently. We found that neurites exhibited more alkaline resting pHi than in soma (7.16 ± 0.09 vs. 6.90 ± 0.15). The neurites were able to change [H+] 3.7-fold faster than soma following NH4Cl prepulse-mediated acidification (p < 0.05). Interestingly, inhibition of NHE-1 activity pharmacologically blocked the pHi regulation in soma and neurites by ~ 70% (p < 0.05). Moreover, the presence of NHE-1 isoform protein was detected in both neuronal soma and neurites, supporting its role in the pHi regulation. Investigating the mechanisms involved in pHi regulation in soma and dendrites is important to understand how the cell counteracts the robust intracellular acidification that occurs during synchronous neural activity.27 In addition, we have recently reported that the selective vulnerability of dendrites to ischemic insults may involve overstimulation of NHE-1 activity.12
The more alkaline basal pHi observed in neurites suggests that increased NHE1 function could have significant metabolic consequences under physiological and pathological conditions. For example, NHEs are major intraterminal pHi regulators on GABAergic presynaptic nerve terminals and play an important role in the neuronal excitability by increasing inhibitory tones.1 It is recently reported that NHE1 accumulates at the leading edge of neurites where it alkalizes growth cones and regulates membrane protrusion and directional motility.28 On the other hand, NADPH oxidase is inhibited by low pHi, thus, increased NHE1 activity could increase formation of reactive oxygen species and potentiate reactive oxidative damage in neurites. 29
We found that the A/V ratio in the neurites is 2.8 times higher than that of soma grown in microfluidic devices. In addition, the A/V ratios of soma and neurites were higher in the microfluidic devices than in the traditional plastic plates. The reason for the difference in A/V ratios between neurons grown in culture dishes vs. microfluidic devices is not clear. It could be because of the difference in “effective culture volume” (ECV) between the two.30 The ECV is a measure of the cell’s control over its local environments and a function of the magnitude of mass transport along each axis, the effects of diffusion and convection, and the degree of protein adsorption to surfaces. 30 In vivo, extracellular molecules such as secreted growth factors have a crucial role in determining cell shape.31 These signals would be lost through dilution in culture dishes. We speculate that the neuronal cell shape and A/V ratios in microscale cultures result from extracellular microenvironments which are not present in culture dishes.
In addition, the significantly higher NHE-1 activity in the neurites in microfluidic devices than in the culture dishes could also result from the density of NHE-1 expression or post-translational regulation (phosphorylation state of the protein). Further investigation is needed to define the influence of the culture microenvironment on pHi regulation through NHE-1.
In summary, we reported here that dissociated pure cortical neurons were cultured in the microfluidic devices with neurites grown across a 200–300 µm bridging-channels. The microfluidic devices allowed us to successfully examined pHi in the two compartments independently. We detected that neurites exhibited more alkaline resting pHi than in soma. The neurites had a pHi regulation rate 3.7-fold faster than in soma following acidification. These findings suggest that the NHE1’s role in maintaining alkaline resting pHi in neurites may regulate function of NMDA receptor activity, voltage-gated Ca2+ channel, and synaptic vesicle-mediated exocytosis.
This work was supported in part by NIH grants R01NS38118 and R01NS048216 (D. Sun), and an AHA Established-Investigator award grant N0540154 (D. Sun); MS RG-4054-A-8 (S.Y. Chiu).