Pyramidal neurons (PyNs) of the cortex are highly susceptible to acute stroke damage, yet “lower” brain regions like hypothalamus and brain stem better survive global ischemia. Here we show for the first time that a “lower” neuron population intrinsically resists acute strokelike injury. In rat brain slices deprived of oxygen and glucose (OGD), we imaged anoxic depolarization (AD) as it propagated through neocortex or hypothalamus. AD, the initial electrophysiological event of stroke, is a front of depolarization that drains residual energy in compromised gray matter. The extent of AD reliably determines ensuing cortical damage, but do all CNS neurons generate a robust AD? During 10 min of OGD, PyNs depolarize without functional recovery. In contrast, magnocellular neuroendocrine cells (MNCs) in hypothalamus under identical stress generate a weak and delayed AD, resist complete depolarization, and rapidly repolarize when oxygen and glucose are restored. They recover their membrane potential, input resistance, and spike amplitude and can survive multiple OGD exposures. Two-photon microscopy in slices derived from a fluorescent mouse line confirms this protection, revealing PyN swelling and dendritic beading after OGD, whereas MNCs are not injured. Exposure to the Na+-K+-ATPase inhibitor ouabain (100 μM) induces AD similar to OGD in both cell types. Moreover, elevated extracellular K+ concentration ([K+]o) evokes spreading depression (SD), a milder version of AD, in PyNs but not MNCs. Therefore overriding the pump by OGD, ouabain, or elevated [K+]o evokes a propagating depolarization in higher gray matter but not in MNCs. We suggest that variation in Na+-K+-ATPase pump efficiency during ischemia injury determines whether a neuronal type succumbs to or resists stroke.
- pyramidal neuron
- whole cell recording
most stroke research has focused on injury to structures in the cerebral hemispheres, particularly neocortex, hippocampus, and striatum. Pyramidal neurons (PyNs) are the output cells of the cortex and have been the prototypical neurons studied in terms of stroke damage both in vivo (Centonze et al. 2001; Dijkhuizen et al. 1999; Koroleva and Vinogradova 2000; Memezawa et al. 1992; Murphy et al. 2008) and in live brain slices (Anderson et al. 2005; Jarvis et al. 2001; Jiang and Haddad 1992; Pisani et al. 2004; Tanaka et al. 2002; White et al. 2012). It is not well recognized that there is a decreasing rostro-caudal susceptibility of neurons to global brain ischemia (Aminoff 2007; Bureš and Buresova 1981; Martin 1999; Sieber et al. 1995). Is this because of regional differences in blood flow or the properties of the neurons? Few studies have investigated how well neurons of the diencephalon (including hypothalamus) and brain stem survive global ischemia. Here we ask whether PyNs of the neocortex necessarily represent typical CNS neurons in terms of susceptibility to ischemia. We compare two populations of neurons in brain slices of neocortex and of hypothalamus, making regional differences in blood flow irrelevant.
Specifically, we examine the susceptibility to anoxic depolarization (AD) induced by oxygen and glucose deprivation (OGD) in neocortical PyNs and hypothalamic magnocellular neuroendocrine cells (MNCs) of the supraoptic nucleus (SON). The extent of simulated ischemia in cortex and hypothalamus is identical because slices are employed. MNCs in the SON and paraventricular nucleus (PVN) produce the hormones vasopressin and oxytocin and discharge in response to osmotic stress, hemorrhage, parturition, and suckling (Li and Hatton 1997; Roper et al. 2003). MNCs are one of the best electrophysiologically characterized neurons caudal to the thalamus. Previous studies have demonstrated their resiliency to glutamate agonists in brain slices (Hu et al. 1992) and in vivo (Herman and Weigand 1986). Moreover, a study by Curras-Collazo et al. (2002) showed that MNCs could survive 48–72 h after stroke, while neocortex and striatum were damaged. Better residual blood flow to hypothalamus could be the explanation, so we examined whether MNCs are intrinsically resistant to metabolic stress in live coronal slices, independent of blood flow.
In response to OGD or ischemia, a wave of “ischemic” or “anoxic” depolarization (AD) propagates across gray matter at 2–4 mm/min (Anderson et al. 2005; Jarvis et al. 2001; Murphy et al. 2008). AD is characterized as a sudden and profound loss of membrane potential caused by failure of the Na+-K+-ATPase pump and represents the most reliable determinant of ensuing brain damage in stroke (Kaminogo et al. 1998). Upon AD onset, PyNs swell and their dendrites bead within a minute or so in slices (Andrew et al. 2007) and in vivo (Murphy et al. 2008). MNCs of the SON and PVN are reported to be osmosensitive cells that swell and shrink in response to physiological osmotic stress (reviewed in Bourque 2008), but there are no data indicating whether MNCs undergo AD or swell during ischemia. Using two-photon laser scanning microscopy (2-PLSM), we investigate whether PyNs and MNCs are resilient to swelling induced by simulated ischemia compared with neocortical neurons. We ask whether MNCs are capable of resisting acute ischemic injury and how they differ from the highly susceptible PyNs of the cortex. Using three independent experimental techniques [light transmittance (LT) imaging, whole cell current-clamp recording, 2-PLSM], we show that MNCs undergo depolarization in response to OGD. However, the response is delayed and weak compared with AD in neocortical PyNs, and so MNCs do not swell acutely. These hypothalamic neurons may intrinsically resist acute strokelike injury by virtue of a more resilient Na+-K+-ATPase pump that functions better under ischemic conditions.
MATERIALS AND METHODS
Brain slice preparation.
Sprague-Dawley rats (male, age 3–10 wk; Charles River, St. Constant, PQ, Canada) or Wistar-GFP rats (either sex) were decapitated by guillotine. After craniotomy, the olfactory bulbs and optic tracts were cut. The brain was quickly removed and immersed in ice-cold and oxygenated (95% O2-5% CO2) artificial cerebrospinal fluid (aCSF) composed of (in mM) 240 sucrose, 3.3 KCl, 26 NaHCO3, 1.3 MgSO4·7H2O, 1.23 NaH2PO4, 11 d-glucose, and 1.8 CaCl2. With a Leica 1000-T vibratome, 400-μm slices were cut in the sucrose aCSF through the coronal plane and then incubated in regular aCSF (equimolar NaCl replacing sucrose above) at 35°C for at least 1 h. Slices were then transferred to a recording/imaging chamber where they were submerged in flowing aCSF (3 ml/min) at 36 ± 0.5°C. The aCSF osmolality was 295 mosmol/kgH2O at pH 7.4.
Visually guided whole cell patch recordings were obtained with micropipettes pulled from borosilicate glass (outside diameter 1.2 mm, inside diameter 0.68 mm; World Precision Instruments) to a resistance of 3–6 MΩ. The internal pipette solution contained (in mM) 125 K-gluconate, 10 KCl, 2 MgCl2, 5.5 EGTA, 10 HEPES, 2 Na-ATP, and 0.1 CaCl2 (pH was adjusted to 7.3 with KOH). A junction potential of 14 mV was corrected prior to recording. Maximum depolarization values reached during AD were not corrected for a junction potential because of the high permeability of the membrane during this period. All recordings were acquired in current-clamp mode of an Axoclamp 2A amplifier and a Digidata 1322 A/D converter (Axon instruments). Clampex 10 software (Axon instruments) was used for data acquisition, with subsequent analysis with Clampfit 10 software. Low-pass filtering was with an external Bessel filter (LPF 202a; Axon Instruments) at 10 Hz. After whole cell recordings were obtained from pyramidal cells in layer V of neocortex or from MNCs in SON, slices were simultaneously imaged (below) while exposed to OGD for 10–12 min (cortical slices) or for 15 min (hypothalamic slices). The OGD aCSF was of a composition similar to control aCSF, except for substitution of 95% O2-5% CO2 bubbling of aCSF with 95% N2-5% CO2. In addition, 11 mM glucose was reduced either to 0 mM or 1 mM glucose with osmotic adjustment using NaCl. Occasionally, MNCs were exposed to multiple 15-min applications of OGD and/or newly acquired recordings were obtained after OGD in the same slice. aCSF for inducing SD contained 9.6 or 26 mM KCl replacing equimolar NaCl.
Imaging changes in light transmittance.
Layer V PyNs in neocortex (somatosensory region; 1.8–3.2 mm bregma) and MNCs in SON were visualized with near-infrared illumination and Dodt gradient contrast optics (Luigs and Neumann, Ratingen, Germany) through an upright microscope (Axoscope 2FS, Zeiss) with a ×40 immersion objective lens. Video images were captured with a cooled charge-coupled device (Hamamatsu C4742) using Imaging Workbench 6 software (Indec Biosystems). Each image of a video series consisted of 16 averaged frames acquired at 20 Hz. The first image of the series was the control transmittance (Tcont), which was subtracted from each of the subsequent images (Texp) in the series. The difference signal was normalized by dividing by Tcont, which varies across the slice depending on the zone sampled. For example, Tcont was lower in white matter than gray matter. This value was then presented as a percentage of the digital intensity of the control image of that series, that is, change in light transmittance (ΔLT) = [(Texp − Tcont)/Tcont ] × 100 = [ΔT/T]%. ΔLT was displayed with a pseudocolor intensity scale. The slice image in bright field was displayed with a gray intensity scale.
Two-photon laser scanning microscopy.
Neocortical and hypothalamic slices (400 μm thick) were taken from 30+-day-old C57 black mice of the B6.Cg-Tg (Thy1-YFP) 16Jrs/J strain and were prepared as described for rat slices. The mouse aCSF composition was similar but not identical to rat aCSF (see Andrew et al. 2007). These mice have a proportion of PyNs that express yellow fluorescent protein (YFP) (Feng et al. 2000). We used mice because there are no transgenic rats available with strongly fluorescent PyNs.
Hypothalamic slices (400 μm) were taken from 30+-day-old Sprague-Dawley transgenic rats expressing an arginine vasopressin (AVP)-enhanced green fluorescent protein (eGFP) fusion gene (Ueta et al. 2005), so ∼50% of MNCs were fluorescent to varying degrees depending on vasopressin content. Therefore only VP neurons were imaged, but both OX and VP neurons were recorded, as differentiated electrophysiologically (Stern and Armstrong 1995). No differences were apparent in their responses to OGD. An imaging chamber was mounted on a fixed stage of an upright Axioscope II FS microscope (Carl Zeiss, Jena, Germany). YFP+ and GFP+ neurons were imaged with appropriate filter sets and a Zeiss LSM 710 NLO meta multiphoton system coupled to a Coherent Ti:sapphire laser. Three-dimensional image stacks were taken at 3.0-μm increments with a Zeiss ×40 or ×63 water-immersion objective. Data acquisition and analyses were controlled by Zeiss LSM software.
Neurons were analyzed if they displayed stable resting membrane potentials and if the series resistance could be sufficiently compensated. Statistical significance was determined by an unpaired t-test, and all data are presented as means ± SD.
Pyramidal neurons and OGD.
Twenty PyNs were monitored by whole cell patch recording during simulated stroke (Fig. 1A). The patch pipette was visually placed within layer V, and the targeted PyNs were identified based on the cell's triangular shape and diameter of 15–20 μm (Fig. 1B1). The resting membrane potential of the 20 cells averaged −82 ± 2.5 mV. Mean input resistance and action potential amplitude were 82 ± 31 MΩ and 79 ± 6.6 mV, respectively (Table 1), as typically reported for PyNs in other studies (Williams and Stuart 2002).
Exposing PyNs to 10–12 min of OGD (n = 20) elicited an abrupt AD with a mean slope of 94.1 ± 40.1 mV/min and a mean onset time of 4.1 ± 0.8 min (Fig. 1, A and B2; Table 1). The maximal average membrane depolarization during AD was to 0 ± 2.3 mV. Upon return to control aCSF this slowly returned to a mean of 27% of the original membrane potential. After a single 10- to 12-min OGD exposure, subsequent intracellular recordings from the same neocortical slice could no longer be obtained, even up to 3 h after return of oxygen and glucose. Importantly, it was no longer possible to record evoked field potentials from layers II–III or V after OGD upon stimulation of layer VI (not shown).
LT imaging of the pyramidal cell field during whole cell recording demonstrated a sudden increase in ΔLT corresponding with AD onset in the single cell (Fig. 1B2, 3). PyNs slowly depolarized during initial OGD, followed by the fast component of the AD (Fig. 1B2). An immediate 8% increase in ΔLT (Fig. 1B3) coincided with the propagating AD wave front, which in this case reached the recorded PyN at ∼320 s of OGD, coinciding with the fast AD component (Fig. 1B1). Upon reintroduction of aCSF, membrane potential returned to only 32% of its original value after 20 min, with 70% recovery of input resistance (Fig. 1B2). Action potentials could not be evoked after OGD because of this maintained depolarization where Na+ channels were inactivated.
Magnocellular neuroendocrine cells and OGD.
Twenty-one MNCs in SON were similarly monitored with whole cell patch recording before, during, and after OGD (Fig. 2). The patch pipette was visually placed within SON, and an MNC was identified based on its ellipsoid shape and large diameter of 20–30 μm. The mean resting membrane potential was −65 ± 2.9 mV. Whole cell input resistance and action potential amplitude were 614 ± 103 MΩ and 92 ± 8.7 mV, respectively (Table 2). All recorded MNCs displayed characteristic frequency-dependent action potential broadening, linear current-voltage curves, and transient outward K+ current (A type) when depolarized from holding potentials more negative than −70 mV (not shown; Kolaj et al. 2000). Each of these properties is an MNC hallmark, so all recorded neurons in SON were considered MNCs.
Because 10–12 min of OGD caused minimal MNC depolarization, we increased the exposure time. Fifteen minutes of OGD eventually elicited a gradual depolarization (mean slope = 11.3 ± 3.0 mV/min between −40 and −20 mV) as shown in Fig. 2 and Table 2. The mean onset time at −40 mV was 9.8 ± 1.3 min. There was a concurrent reduction in action potential amplitude as the MNC reached a mean maximum depolarization of −19 ± 12.8 mV (Fig. 2; Table 2), where discharge ceased because of Na+ channel inactivation. Immediately upon return to control aCSF, there was a steep repolarization (Figs. 2⇓–4) at a mean rate of −65.1 ± 22.6 mV/min (Table 2). This represented rapid recovery never seen in PyN recordings (Fig. 1). Simultaneous imaging of the SON during whole cell recording demonstrated an increase in ΔLT during OGD corresponding with the onset of depolarization, although the ΔLT was notably slower and less dramatic than in neocortical layers. In Fig. 2, bottom, the LT increased gradually to 17%, coinciding with the slow depolarization to −19 mV (Fig. 2, top) recorded in an MNC within the imaged field. Upon cessation of OGD, ΔLT returned to near its original value, corresponding with repolarization of the recorded MNC. This was in contrast to imaging ΔLT in neocortical slices, where a distinct AD wave front passed by the pipette tip as seen at high (×40 objective) magnification (Fig. 1B1).
It proved easier to image the weaker and less distinct AD wave front in SON by imaging at low (×4 objective) magnification (n = 5; Fig. 3). During simultaneous whole cell recording (Fig. 3B), a diffuse AD front initiated at ∼474 s (arrows, Fig. 3A) and spread from the surrounding neuropil, before engaging the SON at ∼535 s. At that point, the recorded MNC gradually depolarized to a mean of −16 mV (Fig. 3B). The ΔLT during OGD again corresponded with the gradual depolarization in Fig. 3B, which returned to near baseline levels upon reintroduction of aCSF (Fig. 3C).
After a 15-min period of OGD, MNCs returned to 87% of their original resting potential compared with only 27% in PyNs after one 10- or 12-min OGD period (Table 3; P < 0.001). Furthermore, compared with PyNs, MNCs also displayed significantly greater percent recovery of action potential amplitude (P < 0.001) and input resistance (P < 0.05) after OGD. Notably, robust MNC recordings could be newly acquired and maintained in a slice previously exposed to either 15 or 30 min of OGD. Resting membrane potential, action potential amplitude, and input resistance were similar to those MNCs not exposed to OGD (n = 11, Table 4). In Fig. 4A, an MNC withstands an accumulated 45 min of OGD. Such multiple exposures were not possible in neocortical slices because PyNs could not recover from a single 10-min exposure to OGD.
Figure 4B illustrates that MNCs are not in some way protected from irreversible depolarization by inhibitory synaptic input. Superfusion of OGD aCSF with CaCl2 reduced from 1.8 to 0.45 mM blocks any residual synaptic input not removed by OGD itself, yet the depolarization remains intact, indicating that the response is intrinsic to the recorded neuron.
Two-photon laser scanning microscopy.
Fluorescent PyNs (in mice) and MNCs (in rats) were each imaged in live coronal slices in real time by 2-PLSM. There is a cytoplasmic component of VP-GFP that is not particulate, making the background cytoplasm clearly discernible in strongly positive MNCs. This is neither fixation nor a sectioning artifact because the cells are alive and located deep in the slice. As above, live neocortical slices from YFP mice were exposed to 10-min OGD while cortical neurons in layer V were imaged in real time. We sampled 60–120 μm deep into each slice so that imaged neurons were distant from the sliced surface. A total of 21 PyNs and their proximal dendrites were imaged just before and 10–20 min after the 10–12 min OGD period. As shown by the measured cross-sectional (XS) area of PyN somata, cell bodies swelled, as did their proximal dendrites (Fig. 5A). Dendrites further deteriorated, as evidenced by beading (Fig. 5A, inset), providing further morphological evidence of AD. As we have previously shown, there is minimal (if any) recovery of cell body volume or normal dendritic structure once AD has propagated across neocortical slices (Andrew et al. 2007). Changes in XS areas optically sectioned through the middle of each PyN cell body were measured off-line. Post-OGD PyN cell bodies were dramatically swollen by an average of 34.5 ± 11.0% (Fig. 5C).
Similar 2-PLSM experiments were carried out in transgenic rats where GFP expression is linked to vasopressin production. Imaging of single vasopressinergic MNCs revealed little or no change in cell body XS area during 15 min of OGD (Fig. 5B). In fact, MNC XS area actually decreased by an average of −3.4 ± 4.2% (n = 29). Compared with pre-OGD values, the degree of shrinking was minor compared with the amount of swelling by PyNs (Fig. 5C). The stability of MNC cell volume despite 15 min of OGD supported our intracellular data showing only minor metabolic stress to MNCs during and after OGD. Unlike PyN dendrites, most MNC dendrites are naturally varicose, so it was difficult to detect whether beading developed along distal MNC dendrites after OGD. However, two examples of proximal MNC dendrites are shown in Fig. 5B (small arrows), indicating no dendritic swelling or beading induced after OGD.
PyNs and MNCs exposed to elevated extracellular K+.
Like AD, spreading depression (SD) is a sudden wave of depolarization propagating across gray matter at 2–4 mm/min. SD is less prolonged and damaging than AD because the tissue is not as metabolically stressed. SD can be evoked in neocortex, striatum, and hippocampus by simply raising extracellular K+. We tested whether PyNs were more prone to SD compared with MNCs by elevating extracellular K+ concentration ([K+]o) to subthreshold levels with 9.6 mM K+ and to 26 mM K+ to elicit SD. In response to 5 min of 9.6 mM K+, PyNs depolarized to a mean of −71.2 ± 4.4 mV (n = 5) (Fig. 6A). Upon washing with control aCSF, they returned to their original baseline (Fig. 6A). In response to 26 mM K+, PyNs depolarized further, firing a series of action potentials followed by Na+ channel inactivation (Fig. 6A). The depolarization plateaued at an average of −47 ± 2.8 mV (n = 5). An additional rapid depolarization was then observed at ∼2 min of 26 mM K+ exposure (mean = 126.8 ± 27.6 s; n = 5) (Fig. 6A, right). This coincided with a simultaneous wave front of elevated LT propagating across the neocortical gray (not shown). Immediately after SD onset, the slice was washed with control aCSF, which returned the membrane potential to baseline.
Similar to PyNs, MNCs showed a small depolarization in response to 5 min of 9.6 mM K+, reaching a mean membrane potential of −58 ± 3.8 mV (n = 6; Fig. 6B). During 26 mM K+ exposure (5 min), the same MNCs reached an average plateau of −43.8 ± 2.0 mV. However, unlike PyNs, SD did not evolve. Rather, MNCs simply fired faster with some degree of spike inactivation (Fig. 6B, left). In response to 52 mM K+ (Fig. 6C), MNCs stopped firing as voltage-sensitive Na+ channels inactivated. Although membrane potential reached a plateau at an average of −34 mV, no SD was evoked. In support of this finding, no elevated LT front representing SD propagation was observed in hypothalamic slices (n = 9). Instead, the SON displayed only a diffuse increase in LT during 5 min of 26 or 52 mM K+ (not shown), likely representing a degree of neuronal and astrocytic swelling.
Inducing stronger depolarization in MNCs.
We attempted to evoke in MNCs a more PyN-like response involving consistent depolarization beyond −20 mV during OGD, either by closing K+ channels or by opening Na+ channels. This proved surprisingly difficult. Blockers of K+ channels such as 4-aminopyridine (4-AP), TEA, and Cs+ or blockers of calcium-activated K+ channels were ineffective (not shown). The Na+ channel activator veratridine induced strong bursts in MNCs, each leading to spike inactivation during a plateau, yet depolarization did not exceed −20 mV (not shown).
However, it was possible to consistently depolarize MNCs to near 0 mV by raising [K+]o to either 26 or 52 mM for 1.5–2 min during OGD (Fig. 7). This more closely simulates the neocortical environment, where [K+]o during AD can reach 60 mM. When [K+]o was added during the initial stages of OGD, MNCs simply depolarized, ceased firing, and reached a plateau depolarization of −34 mV, similar to SD (not shown). We therefore elevated [K+]o after MNCs had depolarized from OGD. This consistently depolarized MNCs to levels experienced by PyNs under OGD. Thirteen MNCs were exposed to either 26 or 52 mM K+ during OGD, beginning when firing stopped at ∼7 min of OGD. This drove MNCs to a mean peak of −5 ± 3.8 mV. The average recovery of membrane potential was 74%, with longer K+ exposure resulting in worse recovery. Action potential amplitude and input resistance were 69% and 52% of initial values, respectively. Therefore driving MNCs to near 0 mV did not impair their recovery from OGD. An additional four post-OGD recordings were newly obtained in slices exposed to 52 mM K+. Their resting membrane potentials, action potential amplitudes, and input resistances were similar to controls (not shown).
Simulating ischemia with ouabain treatment.
In neocortex, blocking the Na+-K+-ATPase pump by exposure to 100 μM ouabain for 10 min evoked an AD-like response similar to OGD (Fig. 8A). Again, PyN depolarization approached 0 mV and no PyNs (n = 4) regained membrane potential, even after up to 50 min in control aCSF. In SON, 100 μM ouabain for 10 or 15 min resulted in a slow, incomplete depolarization of MNCs (n = 6; Fig. 8B) similar to OGD, reaching a mean plateau of −10 ± 2.4 mV. MNCs then recovered up to 82% of their membrane potential, after 40- to 50-min recovery with aCSF, unlike the terminal PyNs. MNCs also recovered 65% and 68% action potential amplitude and input resistance, respectively. MNCs were unable to rapidly repolarize upon reintroduction of control aCSF (compare MNCs in Fig. 8B with Figs. 2⇑–4). This delay in recovery compared with OGD likely reflects a slow washout of ouabain bound to the Na+-K+-ATPase pump compared with the rapid reinfusion of oxygen and glucose in control aCSF.
Intrinsic properties of PyNs and MNCs during OGD.
The vulnerability of PyNs to OGD has been well documented in cortical brain slices (Anderson et al. 2005; Andrew et al. 2007; Jarvis et al. 2001; Jiang and Haddad 1992; Tanaka et al. 1997) and in the ischemic core in vivo (Dijkhuizen et al. 1999; Koroleva and Vinogradova 2000; Memezawa et al. 1992; Murphy et al. 2008). We have previously shown that OGD for 10 min eliminates the evoked field potential in layers II/III of the neocortical slice and that this requires AD propagating through the recorded region (Anderson et al. 2005). The present study demonstrates that these highly susceptible PyNs are not prototypical. By comparison, MNCs in hypothalamus are remarkably resistant to strokelike stress. They recover their membrane potential, action potential amplitude, and input resistance after one (or several) 15-min exposures to OGD. This represents functional recovery not possible in neocortex, where, unlike SON, no new recordings could be obtained in a slice after OGD.
We found that despite a longer OGD period (15 min vs. 10–12 min for PyNs) MNCs recovered 87% of their resting membrane potential compared with 27% by PyNs. Multiple action potentials could be evoked in all but two MNCs after OGD, a feat not possible in PyNs, where fast Na+ channel inactivation (Catterall 2000) predominated. Mean percent recovery of whole cell input resistance was significantly (P < 0.05) better in MNCs after OGD (Table 4). In both cell types this parameter varied after AD because of slice movement as neurons and astrocytes swell (Andrew et al. 2007; Risher et al. 2009). To further demonstrate MNC resiliency to OGD, 11 MNCs were either maintained or newly acquired after multiple 15-min OGD exposures, for cumulative exposures of up to 30 min. Remarkably, resting membrane potentials, input resistance, and action potential amplitudes were similar to those in MNCs not previously exposed to OGD. In contrast, no additional recordings could be obtained from PyNs after a single 10-min OGD insult to the neocortical slice. Our findings confirm that PyNs are acutely injured after OGD and do not recover in neocortical slices. Such neuronal vulnerability to AD is characteristic of “higher” structures in the cerebral hemispheres such as hippocampus (Tanaka et al. 1997, 2002) and striatum (Centonze et al. 2001; Pisani et al. 2004). Few studies have investigated the diencephalon (including hypothalamus) and brain stem. Work by Curras-Collazo et al. (2002) examined various brain regions following a three-vessel occlusion model in rat, assessed by the degree of India ink penetration after ischemia. Despite similar reductions in blood volume observed in caudoputamen (53%) and the MNC nuclei SON (46%) and PVN (45%), no histological damage was seen in the latter nuclei 48–72 h after stroke. A substantial collateral blood supply was suggested as a protective mechanism for MNCs in vivo. However, our study demonstrates MNC resiliency to OGD in the brain slice preparation, which is independent of blood supply.
Changes in light transmittance are small in SON vs. cortex during AD.
Imaging during AD showed that PyNs generate a robust front of elevated LT that propagates across gray matter at 2–4 mm/min, as documented both in vivo (Murphy et al. 2008) and in brain slices (Anderson et al. 2005; Jarvis et al. 2001). The elevated LT wave front in neocortical slices was sharp and of high amplitude. At the upstroke of this signal, PyNs displayed a sudden steep depolarization as the AD wave passed by the pipette tip (Basarsky et al. 1998; Jarvis et al. 2001). PyNs invariably underwent a near-complete loss of membrane potential, followed by minimal recovery. After AD, evoked field potentials from neocortex or hippocampus are permanently lost (Anderson et al. 2005; Jarvis et al. 2001). On the other hand, MNC recovery likely results from the slow onset and weaker depolarization in hypothalamus, as inferred from K+-sensitive electrode recording from lateral hypothalamus adjacent to the SON (Bureš and Buresova 1981) as well as from LT imaging (Brisson et al. 2009). MNC depolarization coincided with a mild and diffuse increase in the LT signal imaged in SON during OGD, rather than a moving front. Other factors likely contribute to the ease of AD propagation in cortex, such as low levels of myelination and higher densities of neurons and astrocytes, leading to higher levels of released [K+]o (Bureš et al. 1974; Shibata and Bureš 1974; Somjen 2001). However, it takes both raising [K+]o and OGD to depolarize MNCs to near 0 mV, whereas either treatment alone is effective in PyNs. This implicates a Na+-K+-ATPase pump in MNCs that functions better under high metabolic stress. Is there other supporting evidence?
Bains et al. (2001) showed that a K+ current (ID) in MNCs of PVN was responsible for resisting strong depolarization evoked by an NMDA receptor agonist, in contrast to adjacent susceptible parvocellular neurons. Inhibition of ID with 4-AP increased depolarization and resulted in death of MNCs. Indeed, MNCs of the SON and PVN are resistant to systemic and intracranial injections of glutamate analogs, in contrast to neocortex and adjacent hypothalamic neurons (Bains et al. 2001; Herman and Weigand 1986; Hu et al. 1992; Schwob et al. 1980). However, glutamate receptor activation is not required for the generation or propagation of AD in adult cortical slices (Anderson et al. 2005; Jarvis et al. 2001; Joshi and Andrew 2001; Muller and Somjen 2000; Tanaka et al. 1997) or in vivo (Murphy et al. 2008; Nellgard and Wieloch 1992). Also, although 4-AP slightly depolarized neurons (∼5 mV), it does not increase MNC susceptibility to AD induced by OGD (unpublished observations). Thus it is unlikely that MNCs resist OGD simply by possessing K+ channels that open during depolarization.
Two-photon laser scanning microscopy confirms MNC protection.
Studies of PyNs show dramatic cell swelling as a result of AD during OGD together with dendritic beading and spine loss indicating neuronal damage in both mouse (Joshi and Andrew 2001; Zhang and Murphy 2007) and rat (Andrew et al. 2007; Obeidat et al. 2000). We confirmed this PyN damage in the present study, but, in sharp contrast, MNCs maintained their cell volume or slightly shrank after 15 min of OGD. This provided independent experimental evidence that, unlike PyNs, MNCs resist AD and the swelling from ischemic damage that results. It is likely that lack of swelling and dendritic beading reflects the lower level of metabolic stress that MNCs experience during OGD because they do not undergo a robust AD. The reversible increase in LT we observed in SON during OGD likely involved the swelling of astrocytes that dissipated by the time of measurement at 15 min after OGD as seen in neocortex (Risher et al. 2009).
MNCs do not generate spreading depression in response to elevated [K+]o.
During elevation of aCSF [K+] to 9.6 mM, both PyNs and MNCs mildly depolarized by 5–10 mV. With 26 mM [K+]o both neuronal types depolarized to −47 and −43 mV, respectively, approximating the equilibrium potential for this [K+] in neurons. PyN discharge led to spike inactivation and then the rapid depolarization of SD. An increase in LT corresponded to the SD front that is well documented during elevated [K+]o in cortical slices (Anderson and Andrew 2002). In contrast, SD is not normally evoked in slices of “lower” brain regions such as hypothalamus (this study) and brain stem of the adult rat (Richter et al. 2008). Neonatal (but not adult) rat brain stem will undergo SD induced by severe hypoxia (Funke et al. 2009), although this is not equivalent to SD evoked by K+ because there is additional metabolic stress. During exposure to 26 or 52 mM K+, we found that MNCs did not undergo electrophysiological SD (n = 6), nor was an LT front imaged indicating SD propagation, an event consistently observed in neocortex (Anderson and Andrew 2002). Only a diffuse increase in LT was recorded in SON during elevated [K+]o.
Elevating [K+]o during OGD.
Under ischemic conditions the SD-like event (i.e., AD) becomes prolonged and recovery is more difficult. Once initiated, the strong AD by PyNs resists recovery in slices. MNCs remain less depolarized during 26 mM K+ (−43 mV), 52 mM K+ (−34 mV), or OGD (−19 mV) compared with PyNs, which approach 9 mV in all three cases, again implying a difference in the potency of the Na+-K+-ATPase pump.
A higher peak of [K+]o released by depolarizing neurons during AD will increase the strength of AD, which is reciprocal, that is, strong AD promotes K+ release and vice versa. Only by briefly raising [K+]o near the end of OGD (thereby further driving this positive feedback) were we able to push MNCs to 0 mV, similar to PyNs, but even then MNCs recovered, driven by their Na+-K+-ATPase pump, in sharp contrast to the terminally depolarized PyNs. We propose that intrinsic differences in isoform distribution and/or activation kinetics of the Na+-K+-ATPase may provide a mechanism as to why MNCs better survive OGD, as discussed below.
Na+-K+-ATPase pump efficiency probably varies with neuronal type.
Once depolarized, PyNs could not repolarize after 10 min of ouabain treatment, even with up to 50-min recovery in control aCSF. Similar to OGD, no action potentials could be evoked and no additional recordings could be obtained in the same slice. In contrast to PyNs, MNCs markedly recovered after 10- or 15-min ouabain exposure, but repolarization was much slower than after OGD. This likely reflects the slow washout of ouabain, which has a half-life of 45 min (Tobin and Brody 1972) and highlights the importance of Na+-K+-ATPase pump recovery for efficient repolarization. Several mechanisms could contribute to MNC recovery from pump failure, including activation kinetics, isomer distribution, second messengers, or regulation of permeate ions (Blanco and Mercer 1998).
The one way to mimic the electrophysiological time course, the elevated [K+]o profile, the time course of elevated intracellular Ca2+ concentration, the AD propagation rate, and the post-AD neuronal injury is by inhibiting the Na+-K+-ATPase pump (Grotton and Lo 2010; Jarvis et al. 2001; Tanaka et al. 1997). In this way ouabain treatment effectively reiterates OGD, producing strong AD in cortex (Balestrino et al. 1999; Jarvis et al. 2001; Tanaka et al. 1997) but weak AD in hypothalamus, as shown here. It has been assumed that CNS neurons in general are vulnerable to ischemia, so the idea that Na+-K+-ATPase pump efficiency varies among brain regions has yet to be explored.
PyNs, the output cells of cortex, have been the prototype for CNS neurons undergoing stroke injury, yet whole cell recordings demonstrate a significant resilience to OGD by MNCs compared with susceptible PyNs. Specifically, PyNs undergo a rapid and near-complete membrane depolarization during OGD (AD), while MNCs depolarize slowly, generate a weak AD, and resist complete depolarization during OGD. Moreover, MNCs repolarize immediately upon cessation of OGD, permitting survival after one or more OGD exposures. Unlike in neocortex, robust recordings from newly patched MNCs are readily obtained after an OGD exposure. Two-photon microscopy shows that MNCs resisted cell swelling and dendritic beading in response to OGD, unlike PyNs. By imaging ΔLT we show that hypothalamic slices experience much weaker AD than neocortex, which can help explain why lower brain regions like hypothalamus and brain stem (Brisson and Andrew 2011) better survive global ischemia (Aminoff 2007; Sieber et al. 1995). Increased myelination and decreased cell density are possible reasons for the weak or absent propagation of an SD-like signal in these brain regions, yet raising [K+]o to as high as 52 mM does not evoke SD in MNCs, so resistance to strong depolarization is also a major factor. Exposure to the Na+-K+-ATPase inhibitor ouabain (100 μM) induces AD similar to OGD in both cell types. Therefore overriding the pump by OGD, ouabain, or elevated [K+]o evokes a propagating depolarization in higher gray matter but not in MNCs. Further work is required to determine whether “lower” neurons simply lack channels to generate AD. This seems unlikely because the ionic fluxes underlying AD are large and nonspecific, involving a sudden loss of membrane selectivity where ions run down their concentration gradients (Muller and Somjen 1998; Radar and Lanthorn 1989; Somjen 2001; Tanaka et al. 1997). A more likely hypothesis is that Na+-K+-ATPase pump isoforms function better under metabolic stress in lower neurons. Our study demonstrates that the propensity to immediately shut down in the face of ischemia is an intrinsic property of the cortical PyN. Furthermore, this is not simply the “default” strategy of stressed CNS neurons. Indeed, shutdown of lower brain regions would be a poor survival strategy.
This work was supported by the Heart and Stroke Foundation of Ontario (Grant T-4478).
No conflicts of interest, financial or otherwise, are declared by the author(s).
Author contributions: C.D.B. and R.D.A. conception and design of research; C.D.B. performed experiments; C.D.B. and R.D.A. analyzed data; C.D.B. and R.D.A. interpreted results of experiments; C.D.B. and R.D.A. prepared figures; C.D.B. and R.D.A. drafted manuscript; C.D.B. and R.D.A. edited and revised manuscript; C.D.B. and R.D.A. approved final version of manuscript.
We thank the laboratory of Dr. Yoichi Ueta, University of Occupational and Environmental Health, Japan, for generously providing the founders of our transgenic rats. Thanks also to Patti Storey for technical expertise.
- Copyright © 2012 the American Physiological Society