Alterations in Ca²⁺-Buffering in Prion-Null Mice: Association with Reduced Afterhyperpolarizations in CA1 Hippocampal Neurons

Abstract

Prion protein (PrP) is a normal component of neurons, which confers susceptibility to prion diseases. Despite its evolutionary conservation, its normal function remains controversial. PrP-deficient (Prnp^0/0) mice have weaker afterhyperpolarizations (AHPs) in cerebellar and hippocampal neurons. Here we show that the AHP impairment in hippocampal CA1 pyramidal cells is selective for the slow AHP, and is not caused by an impairment of either voltage-gated Ca²⁺ channels or Ca²⁺-activated K⁺ channels. Instead, Prnp^0/0 neurons have twofold to threefold stronger Ca²⁺ buffering and double the Ca²⁺ extrusion rate. In Prnp^0/0 neurons thapsigargin abolished the stronger Ca²⁺ buffering and extrusion, and thapsigargin or cyclopiazonic acid abolished the weakening of the slow AHPs. These data implicate sarcoplasmic/endoplasmic reticulum calcium ATPase in the enhanced Ca²⁺ buffering, and extrusion into the endoplasmic reticulum, which contains substantial amounts of PrP in wild-type mice. Altered Ca²⁺ homeostasis can explain several phenotypes identified in Prnp^0/0 mice.

Introduction

Prion protein (PrP) is a widely expressed plasma membrane-associated glycoprotein with highest levels of expression on neurons and glia. It is widely accepted that an abnormal isoform of this protein underlies transmissible spongiform encephalopathies such as bovine spongiform encephalopathy, scrapie, and Creutzfeldt-Jakob disease (Prusiner, 1998; Collinge, 2001). Despite the central role of PrP in these diseases, and its evolutionary conservation, its normal function remains unclear. A proposed role for PrP is metal ion homeostasis (Brown, 2002; Watt and Hooper, 2003) because it binds copper avidly although its functional consequences, if any, are unclear (Davies and Brown, 2008). Other proposed roles for PrP include protection from oxidative stress (Brown, 2002; Krebs et al., 2007), although this is contentious (Hutter et al., 2003; Jones et al., 2005), and signal transduction (Rieger et al., 1997; Chiarini et al., 2002). In the brain, PrP particularly localizes to synaptic membranes (Herms et al., 1999; Fournier et al., 2000), suggesting a role in neurotransmission. This hypothesis supported by several reports of subtle alterations in neuronal function in PrP-deficient (Prnp^0/0) mice. These alterations include mossy fiber reorganization (Colling et al., 1997), reduced long-term potentiation (LTP) (Collinge et al., 1994; Manson et al., 1995), decreased EPSPs (Carleton et al., 2001), and reduced synaptic inhibition (Collinge et al., 1994). A reproducible electrophysiological phenotype is a reduction in the slow afterhyperpolarization (AHP) (Colling et al., 1996; Herms et al., 2001; Mallucci et al., 2002; Fuhrmann et al., 2006). The slow AHP is a property found in many neurons, which is evoked by repetitive action potentials, can last for several seconds and reduces the likelihood of subsequent action potential firing (Lancaster and Adams, 1986). The slow AHP is mediated by a Ca²⁺-dependent K⁺ channel, the exact molecular identity of which remains elusive (Sah and Clements, 1999; Bond et al., 2004; Pedarzani et al., 2005). The observed weaker slow AHP in PrP-deficient mice has several potential mechanisms, including impaired function of either Ca²⁺ or K⁺ channels, disruption of their interaction, or altered Ca²⁺ homeostasis.

Herms et al. (Fuhrmann et al., 2006) suggested that the reduction in slow AHP is caused by an alteration in L-type voltage-gated Ca²⁺ channels (VGCCs) in CA1 hippocampal neurons. Whether this mechanism is sufficient to explain the observed alterations in Ca²⁺ signaling in diverse cell types remains unclear. In cerebellar granule neurons, basal levels and depolarization-induced Ca²⁺ signals are reduced without observed alterations in VGCC function (Herms et al., 2000). Prion protein expression has also been shown to have profound effects on Ca²⁺ signaling in Chinese hamster ovary cells (Brini et al., 2005).

In the present study, we examined the effect of PrP expression on the Ca²⁺ homeostasis of CA1 pyramidal neurons to find a mechanism that can explain the reduction in both Ca²⁺ signaling and the slow AHP. By use of patch-clamp recordings of individual hippocampal CA1 pyramidal neurons, we examined the mechanism of the reduction in slow AHP and correlated this with changes in Ca²⁺ signaling.

Materials and Methods

Origin and use of Prnp^0/0 mice.

Mice were bred at the University of Birmingham from progenitor mice supplied by the Medical Research Council Prion Unit (London, UK). Two groups of mice were used: homozygous PrP-null (Prnp^0/0) and wild-type controls (Prnp ^+/+). The PrP-null mice were generated from “Zurich 1” mice (Bueler et al., 1992), fully backcrossed onto an FVB/N background; control mice were inbred FVB/N mice bred in the same unit. All experiments and analysis were performed blind to the genotype of the mice.

Preparation and maintenance of acute hippocampal slices.

A total of 47 Prnp ^+/+ and 49 Prnp^0/0 age-matched male mice (3–9 weeks old) were anesthetized by intraperitoneal injection of a mixture of medetomidine (1 mg/kg) and ketamine (76 mg/kg), before being killed by cervical dislocation. The brains were rapidly removed and chilled (<3°C), in oxygenated, ice-cold sucrose-based cutting solution composed of (in mm) 189 sucrose, 26 NaHCO₃, 1.2 NaH₂PO₄, 2.5 KCl, 0.1 CaCl₂, 5 MgCl₂, and 10 glucose. The brains were cut along the midline, and 300 μm horizontal hippocampal slices were cut using an Integraslice (Campden Instruments, Loughborough, UK). Hippocampal slices were maintained at room temperature in an interface storage chamber containing 95%O₂–5%CO₂ oxygenated aCSF. The composition of the artificial CSF (aCSF) was (in mm) 135 NaCl, 16 NaHCO₃, 1.25 NaH₂PO₄, 3 KCl, 2 CaCl₂, 1 MgCl₂ and 10 glucose, pH 7.4.

Slices were transferred to a recording chamber, where they were submerged and continually perfused (4–5 ml/min) with warm (29–31°C) aCSF, bubbled with 95%O₂–5%CO₂. Slices were allowed to equilibrate in the aCSF solution for >10 min before recording. Slices were changed after ∼1 h.

Whole-cell recordings.

Whole-cell patch-clamp recordings were made from the somata of CA1 pyramidal neurons using infrared differential interference contrast on an Olympus (Tokyo, Japan) BX-51 upright microscope, fitted with a fluorplan 40×, 0.8 numerical aperture water-immersion lens (Micro Instruments, Long Hanborough, UK). CA1 neurons were visually selected based on appearance and the integrity of the apical dendrites (visible for ∼100 μm from the soma). Patch electrodes were pulled from borosilicate glass capillaries (outer diameter, 1.2 mm; inner diameter, 0.69 mm; Harvard Apparatus, Edenbridge, UK) using a P-97 horizontal puller (Sutter Instruments, Novato, CA) and filled with intracellular solution comprising (in mm) 135 KCH₃SO₄, 8 NaCl, 10 HEPES, 2 Mg-ATP and 0.3 Na-GTP; adjusted to pH 7.3 with KOH (∼285 mOsm), unless stated otherwise. For the VGCC studies, equimolar CsCH₃SO₄ replaced KCH₃SO₄ and 1 μm tetrodotoxin was added to the extracellular solution to reduce Na⁺ channel activation. For some VGCC studies, the standard aCSF was replaced by a tetraethylammonium chloride (TEA)-based extracellular solution containing (in mm) 148 TEA-Cl, 4 KCl, 2 CaCl₂, 1 MgCl₂, 10 HEPES, and 0.001 TTX, pH 7.3 with TEA-OH (298 mOsm). Patch electrodes filled with the intracellular solutions typically had resistances of 4–7 MΩ. Membrane potentials and currents were recorded using an NPI SEC-10L amplifier (Scientifica, Harpenden, UK), acquired and digitized at 10 kHz using a Power 1401 A–D converter (Cambridge Electronic Design, Cambridge, UK). Data were low-pass Bessel filtered at 1 kHz by either a Neurolog NL-125 or NL-135 filter unit (Digitimer, Welwyn Garden City, UK) before storage on a computer. Data acquisition and stimulus parameters were controlled using Signal software (version 2.16; Cambridge Electronic Design).

Epifluorescent Ca²⁺ measurements.

Hippocampal CA1 neurons were loaded, via the patch pipette, with the Ca²⁺ indicator dyes, Fura-2 or Fluo-3 as indicated in the text (Invitrogen, Eugene, OR). Epifluorescence measurements were performed using a dual photomultiplier tube (PMT) system (Cairn Research, Faversham, Kent, UK) mounted onto the microscope. The fluorophore was excited with light of an appropriate wavelength (Fura-2, 340 and 380 nm; Fluo-3, 488 nm), using a computer controlled Optoscan monochromator (Cairn Research). Emitted light was selected using either a 510 ± 20 nm filter for Fura-2 or a 535 ± 20 nm filter for Fluo-3 (Chroma Technologies, Rockingham, VT). The outputs of the PMTs were sampled at 200 Hz using Signal (Cambridge Electronic Design) and stored on the computer. Regions of interest (ROIs) were selected by use of adjustable rectangular field stops; one ROI was placed over the somatic region, and a second over the initial 25–75 μm segment of the apical dendrite. Changes in fluorescence were background subtracted using a ROI located close to the cell under investigation. Changes in intracellular Ca²⁺ concentration ([Ca²⁺]_i) were calculated from the Ca²⁺-dependent F ₃₄₀/F ₃₈₀ fluorescence ratio (R), according to Grynkiewicz et al. (1985) (Eq. 1). Calibration constants for Fura-2 were determined by patch clamping CA1 pyramidal neurons (seven neurons per intracellular solution per genotype) with the following intracellular solutions: R _min (Ca²⁺-free), 135 KCH₃SO₄, 8 NaCl, 10 HEPES, 2 Mg-ATP, 0.3 Na-GTP, and 10 BAPTA; R ₄₀₀ (400 nm free Ca²⁺), 135 KCH₃SO₄, 8 NaCl, 10 HEPES, 2 Mg-ATP, 0.3 Na-GTP, 9.9 BAPTA, and 6.6 CaCl₂, yielding a final intracellular Ca²⁺ concentration of 400 nm (calculated using the MaxChelator program, http://www.stanford.edu/∼cpatton/maxc.html); R _max (10 mm free Ca²⁺), 135 KCH₃SO₄, 8 NaCl, 10 HEPES, 2 Mg-ATP, 0.3 Na-GTP, and 10 CaCl₂. The R ₄₀₀ intracellular solution was used to confirm the value of the calibration constant. All intracellular solutions were adjusted to pH 7.3 with KOH. The Ca²⁺ dissociation constant for Fura-2 (K _d) was determined to be 247 nm by using the following equation (Grynkiewicz et al. (1985):

To minimize complications associated with diffraction of fluorescent light associated with alterations in cell depth, care was taken to ensure that the same criteria (cell soma; 15–25 μm, < 50 μm depth from slice surface) for cell selection were used to identify CA1 pyramidal neurons used for the Fura-2 calibration experiments. No systematic differences in the K _d were found between Prnp ^+/+ and Prnp^0/0 neurons (data not shown).

Analysis of endogenous Ca²⁺ buffering.

Calcium homeostasis was quantified by using the “added buffer method,” which relies on the ability of Ca²⁺-sensitive dyes, such as Fura-2, to compete with endogenous buffers and alter the Ca²⁺ dynamics (Neher, 1995; Helmchen et al., 1996; Palecek et al., 1999). Before patch breakthrough the buffering is mediated solely by the neuron's endogenous buffers; after breakthrough, the concentration of exogenous buffer increases until it reaches a point where virtually all Ca²⁺ that enters the cell is sequestered by the exogenous buffer. Ca²⁺ transients were evoked by a 50 ms depolarization (−65 to −5 mV) every 30 s after patch breakthrough. After establishment of the whole-cell recording configuration, the concentration of Fura-2 within the cell increased with time until [Fura2]_i equilibrated with [Fura2]_pipette (see Fig. 5 Ai,Aii). The buffering capacity of Fura-2 (κ_B) was calculated using the following equation:

where [B] and K _d are the concentration and dissociation constant of Fura-2, respectively (Helmchen et al., 1996).

Figure 5.

Endogenous Ca²⁺ buffering is enhanced in Prnp^0/0 neurons. Calcium buffering of CA1 neurons was assessed by monitoring calcium changes during Fura-2 loading. Ai, The soma of a Prnp ^+/+ CA1 pyramidal neuron was loaded with 100 μm Fura-2 (bottom). Measurements of depolarization-evoked somatic [Ca²⁺]_i transients were made every 30 s during loading (filled circles). [Ca²⁺]_i transients measured directly after patch excision and after full loading (open circles) are shown in the top. The Ca²⁺ transients decayed with time courses of 0.37 and 1.82 s, respectively. Aii, Ca²⁺ buffering analysis for a Prnp^0/0 neuron during loading with 100 μm Fura2 (bottom). The Ca²⁺ transients measured immediately after patch excision and after full loading (top, open square), decayed with time courses of 0.316 and 0.828 s, respectively. Bi, Inverse amplitude of Ca²⁺ changes plotted as a function of Fura-2 Ca²⁺ binding ratio (κ'_B) for individual cells. Negative x-axis intercepts were 25 (Prnp ^+/+; •) and 81 (Prnp^0/0; ○). Bii, Decay time constant (τ) plotted as a function of Fura-2 Ca²⁺ binding ratio (κ'_B). Negative x-axis intercepts were 10 (Prnp ^+/+) and 75 (Prnp^0/0), which is a measure of the ability of the neuron to buffer free Ca²⁺. Ci, Plot of mean endogenous binding Ca²⁺ binding ratio (κ_S) derived from inverse amplitude illustrating improved Ca²⁺ buffering in Prnp^0/0 neurons (means of 8 Prnp ^+/+ and 8 Prnp^0/0 cells). Cii, Plot of mean Ca²⁺ extrusion rate (γ) obtained from Equation 4 showing increased Ca²⁺ extrusion in Prnp^0/0 neurons. ***p < 0.01.

The amplitude (A) and decay rate (τ) of the Ca²⁺ signal are related to the buffering capacity of the endogenous buffer (κ_S) according to Equations 3 and 4 (Helmchen et al., 1996):

where Δ[Ca²⁺]_T is the increase in total Ca²⁺ (free and bound) and γ is the Ca²⁺ extrusion rate.

Equations 3 and 4 predict that increases in κ_B will reduce the amplitude and slow the decay kinetics of the Ca²⁺ transient. Both the inverse of the amplitude (A ⁻¹) and τ depend linearly on κ_B. The relationship between A ⁻¹ and κ_B allows estimation of the endogenous Ca²⁺ binding ratio (κ_S), as the negative abscissa intercept. κ_S then allows estimation of γ from Equation 4.

Added buffer measurements were performed on slices from mice over a narrow age range (postnatal days 42–55) to avoid any impact of age (Maravall et al., 2000).

Flash photolysis.

DM-nitrophen (Calbiochem, La Jolla, CA) was used as the caged compound for photorelease of Ca²⁺ and was dissolved directly in internal solution in which Mg-ATP was replaced with Na-ATP to avoid Mg²⁺ loading (Faas et al., 2005). DM-nitrophen (10 mm) was 60–70% loaded with Ca²⁺. The resulting free [Ca²⁺] was calculated to be ∼80 nm using the MaxChelator program. The lamp used for photolysis was a pulsed xenon arc lamp (Cairn Research); flash intensity was controlled by use of a variable capacitance system (20–4070 μF; 400V). The flash illuminated the entire field of view and discharged in <2 ms. For the flash photolysis experiments, it was necessary to use the longer wavelength indicator Fluo-3 (100 μm) to monitor [Ca²⁺]_i because the UV illumination required to excite Fura-2 fluorescence would also photolyse the DM-nitrophen. For the flash photolysis experiments, because Fluo-3 is not a ratiometric dye, changes in intracellular Ca²⁺ were not calibrated and are expressed as ΔF/F ₀.

Data analysis.

All values are expressed as mean ± SEM. Curve fitting was performed in Origin 8 (Silverdale Scientific, Stoke Mandeville, UK); statistical analysis was performed in SPSS 13 (SPSS, Woking, UK). Student's t tests and ANOVA were used as appropriate and the significant criterion was p < 0.05, unless otherwise stated.

Results

To understand the role of prion protein in Ca²⁺ signaling, we examined the relationship between the AHP and Ca²⁺ signaling using a combined electrophysiological and microfluorimetric approach. Previous studies have shown that Prnp^0/0 neurons have a reduced AHP (Colling et al., 1996; Herms et al., 2001; Mallucci et al., 2002; Fuhrmann et al., 2006). Whole-cell patch clamp recordings were made from the somata of CA1 pyramidal neurons and AHPs were evoked either by repetitive action potential firing or by direct activation of voltage-gated Ca²⁺ channels. Figure 1 A illustrates typical AHPs recorded from Prnp ^+/+ and Prnp^0/0 neurons in response to 50 action potentials (delivered at 20Hz). The slow AHP was significantly smaller in Prnp^0/0 neurons than in Prnp ^+/+ neurons (2.19 ± 0.13 mV, n = 17; 3.05 ± 0.18 mV, n = 19, respectively) (Fig. 1 A). When the slow AHP current (I _sAHP) was evoked by activation of voltage-gated Ca²⁺ channels by a depolarizing pulse in voltage-clamp mode, the I _sAHP was significantly smaller in Prnp^0/0 neurons than in Prnp ^+/+ neurons (85.5 ± 13.5 pA and 201.8 ± 25.3 pA, respectively) (Fig. 1 B). The apparent reduction in the medium AHP was attributable to the reduction in the underlying slow AHP (supplemental data part 2, available at www.jneurosci.org as supplemental material).

Figure 1.

A, Prnp^0/0 neurons possess a reduced slow afterhyperpolarization. Ai, A train of 50 action potentials at 20 Hz, evoked both a medium and slow AHP in Prnp ^+/+ (black traces) and Prnp^0/0 (gray traces) neurons. The medium AHP was defined by its rapid kinetics and was observed immediately after cessation of action potential firing. The slow AHP was defined by its slower kinetics and was smaller in Prnp^0/0 neurons at all frequencies tested (Aii). B, A 300 ms depolarization from −65 to +5 mV activated both a medium AHP current (I _mAHP), which displayed faster kinetics and was observed immediately after cessation of the depolarization and slow AHP current (I _sAHP) which displayed slower kinetics. Similar to the slow afterhyperpolarization, the I _sAHP was smaller in Prnp^0/0 neurons. ***p < 0.001.

We have confirmed that Prnp^0/0 neurons display a selective reduction in the slow AHP in CA1 pyramidal neurons. The slow AHP is mediated by Ca²⁺-dependent K⁺ channels that are activated by Ca²⁺ entry through VGCCs (Sah and Faber, 2002), suggesting several mechanisms for the reduction in the slow AHP amplitude. These include altered Ca²⁺ channel function (Fuhrmann et al., 2006), altered Ca²⁺ signaling (Herms et al., 2000) and altered K⁺ channel function.

Flash photolysis

To test the hypothesis that functioning of the K⁺ channel that is responsible for the slow AHP is altered in Prnp^0/0 mice, we used photorelease of calcium from DM-nitrophen to elevate intracellular [Ca²⁺] independently of Ca²⁺ channel activation. This method is well established in the investigation of the molecular basis of the slow AHP (Lancaster and Zucker, 1994; Sah and Clements, 1999).

Action potential firing (50 action potentials at 20 Hz), evoked slow AHPs in both Prnp ^+/+ and Prnp^0/0 neurons. Prnp ^+/+ neurons had a slow AHP of 2.44 ± 0.38 mV (n = 5) (Fig. 2 A, left). Ca²⁺ concentration, as reported by changes in Fluo-3 fluorescence, increased throughout the period of action potential firing reaching a peak fluorescence value of 1.24 ± 0.06 (Fig. 2 B, left). Prnp^0/0 neurons generated a significantly smaller slow AHP (1.24 ± 0.31 mV; n = 6; p < 0.05) (Fig. 2 A, left). Furthermore, Prnp^0/0 mice produced significantly smaller Ca²⁺ increases in response to action potential firing (peak fluorescence value, 1.10 ± 0.02, p < 0.05) (Fig. 2 B, left). Therefore, under the recording conditions necessary for flash photolysis studies, it is possible to replicate the reduced slow AHP amplitude in Prnp^0/0 neurons. In Prnp ^+/+ neurons, flash photolysis of DM-nitrophen evoked a rapid increase in Ca²⁺ concentration (peak value, 1.82 ± 0.14) (Fig. 2 B, right) and a concomitant membrane hyperpolarization of 3.23 ± 0.36 mV (Fig. 2 A, right). In Prnp^0/0 neurons, flash photolysis released similar amounts of Ca²⁺ as in wild-type neurons (peak value, 1.74 ± 0.07; p = 0.58); the decay rate of the Ca²⁺ signal did not significantly differ between Prnp ^+/+ and Prnp^0/0 neurons (τ_decay 2.73 ± 1.06 s, n = 5 and 2.55 ± 0.96 s, n = 6, respectively; p = 0.76). The hyperpolarization observed in response photorelease of Ca²⁺ was not significantly different in Prnp^0/0 mice (3.12 ± 0.66 mV; p = 0.89). The hyperpolarization activated by flash photolysis of caged Ca²⁺ was confirmed as a slow AHP, because it was inhibited by 5-hydroxytryptamine (10 μm). Moreover, the flash evoked hyperpolarization displayed similar kinetics to the action potential evoked AHPs (supplemental data, part 2, available at www.jneurosci.org as supplemental material).

Figure 2.

The reduced slow AHP is not caused by alterations in the underlying K⁺ channels. Whole-cell recordings were obtained using electrodes filled with internal solution including 100 μm Fluo-3 and 10 mm DM-nitrophen (Ca²⁺ cage). A, B, Prnp^0/0 neurons filled with this solution displayed a significantly reduced action potential evoked slow AHP (A, left, gray traces), associated with a significant attenuation of Ca²⁺ increases (B, left, gray traces). In contrast, flash photolysis evoked a robust and similar AHP in both Prnp ^+/+ and Prnp^0/0 neurons (A, right), associated with similar Ca²⁺ transients in the two genotypes (B, right). *p < 0.05.

The AHP in Prnp^0/0 neurons was smaller only when activated by Ca²⁺ entry through Ca²⁺ channels, but not by homogeneous Ca²⁺ elevation resulting from photolysis of intracellular DM-nitrophen. The direct activation of the K⁺ channels underlying the slow AHP reveals, for the first time, that they are both present and functioning correctly in Prnp^0/0 neurons. Therefore, alterations in K⁺ channel function cannot explain the observed reduction of slow AHP amplitude.

Ca²⁺ signaling

Several studies have implicated changes in Ca²⁺ homeostasis in the absence of PrP (Herms et al., 2000, 2001; Brini et al., 2005; Fuhrmann et al., 2006). The flash photolysis study (Fig. 2 B, left), suggested that Ca²⁺ homeostasis was altered in the present study. Here we have used single cell fluorimetry with Fura-2 specifically to examine the role of prion protein in Ca²⁺ signaling and buffering.

Resting Ca²⁺ levels in Prnp^0/0 neurons were lower in both dendritic (Prnp ^+/+, 148.2 ± 23.8 nm; Prnp^0/0, 62.2 ± 21.4 nm; p = 0.01) and somatic regions (86.9 ± 16.1 nm; Prnp^0/0, 39.5 ± 10.7 nm; p = 0.03) (Fig. 3 C).

Figure 3.

Ca²⁺ signaling is perturbed in Prnp^0/0 neurons. A, Prnp^0/0 neurons (gray traces) produce weaker Ca²⁺ transients, in both somatic (left) and dendritic regions (right), in response to 50 action potentials evoked at 20 Hz. B, Peak Ca²⁺ changes were significantly reduced across the frequency range (p < 0.001, ANOVA; •, Prnp ^+/+; ○, Prnp^0/0) for both somatic (left) and dendritic (right) regions. C, The resting Ca²⁺ level was significantly reduced in Prnp^0/0 neurons in both the somatic (left) and proximal dendritic (right) regions. Somatic and dendritic regions were not significantly different. *p < 0.05; **p < 0.01; ***p < 0.005.

Stimulating cells with 50 action potentials at 20 Hz evoked increases in intracellular Ca²⁺ in both the somatic (Fig. 3 A; left panel) and the proximal dendrite regions (Fig. 3 A, right panel), with the peak Ca²⁺ concentration occurring at the end of the spike train. The evoked peak [Ca²⁺] significantly increased with action potential frequency in both somatic and dendritic regions (Fig. 3) (ANOVA p = 0.01 and 0.03, respectively). ANOVA revealed that the peak Ca²⁺ increases in Prnp^0/0 neurons were smaller across a range of frequencies in both the soma and dendrites (p < 0.001). We chose to use 50 action potentials at 20 Hz as the standard stimulation protocol because the effective Ca²⁺ range for Fura-2 is ∼20 nm–2 μm and both 50 and 100 Hz resulted in Ca²⁺ increases in excess of 2 μm (Fig. 3 B). Ca²⁺ increases evoked by 50 action potentials at 20 Hz in the dendrites were smaller in Prnp^0/0 cells than in wild type (0.88 ± 0.11 μm, n = 10 and 1.37 ± 0.14 μm, n = 12, respectively; p < 0.01) (Fig. 3 A). The peak Ca²⁺ response was similarly smaller in the soma: 1.19 ± 0.10 μm, n = 16 in Prnp^0/0 and 1.41 ± 0.06, n = 21 in wild type (p < 0.005).

The observed alterations in Ca²⁺ signaling could be explained by changes in VGCC function, as has been reported previously for CA1 pyramidal neurons (Fuhrmann et al., 2006). We investigated this possibility by comparing the voltage dependence of VGCC currents and simultaneously measured Ca²⁺ transients from both Prnp ^+/+ and Prnp^0/0 neurons. Neurons were voltage clamped at −70 mV and 200 ms step depolarizations (−60 mV to + 30 mV; 10 mV increments) were applied every 30 s. No differences in the I–V relationships were observed between Prnp ^+/+ and Prnp^0/0 neurons whether in bicarbonate-buffered aCSF (p = 0.89) (Fig. 4 A,B) or HEPES-buffered aCSF (p = 0.69) (supplemental data part 4, available at www.jneurosci.org as supplemental material). Ca²⁺ changes evoked by step depolarizations, however, were significantly smaller in Prnp^0/0 than in Prnp ^+/+ neurons (1.37 ± 0.18 μm and 2.36 ± 0.32 μm, respectively; p = 0.03) (Fig. 4 A). We have therefore excluded the possibility that the alteration in slow AHP is caused by alterations in either K⁺-channel or Ca²⁺-channel function.

Figure 4.

Voltage-gated Ca²⁺ channels are unaffected in Prnp^0/0 neurons. A, Ca²⁺ current amplitude and kinetics were unaffected in Prnp^0/0 neurons in response to a 200 ms step depolarization. The Ca²⁺ changes, as monitored by 100 μm Fura-2, were significantly smaller in Prnp^0/0 neurons (bottom, gray trace) than in Prnp ^+/+ neurons (for statistics, see Results). B, Ca²⁺ current-voltage relationship is not altered in Prnp^0/0 neurons (•, Prnp ^+/+; ○, Prnp^0/0).

Endogenous Ca²⁺ buffering capacity

We used the “added buffer” technique (Neher, 1995) to investigate whether the PrP-dependent differences in Ca²⁺ signaling, found in this and other studies (Herms et al., 2000; Brini et al., 2005; Fuhrmann et al., 2006), could be explained by changes in endogenous Ca²⁺ binding capacity. Diffusion of Fura-2 from the patch pipette provided controlled loading of Ca²⁺ buffer into the neurons, with simultaneous recording of calcium responses triggered by somatic depolarization (50 ms duration −65 to +5 mV, every 30 s). During Fura-2 loading, the concentration of exogenous Ca²⁺ buffer progressively increases and eventually dominates; the relationship of the Ca²⁺ response with the concentration of Fura-2 allows estimation of Ca²⁺ buffering extrusion (Eqs. 3, 4) (Neher, 1995; Helmchen et al., 1996; Palecek et al., 1999).

Fura-2 concentration was estimated from changes in the fluorescence intensity recorded at the isobestic wavelength (358 ± 3 nm in our system), measured every 30 s after establishment of a gigaohm seal (Fig. 5 Ai,Aii). The time course of Fura-2 loading was best fit by a single exponential, and did not differ between wild-type (time constant, 2.8 ± 0.4 min, n = 8) and Prnp^0/0 neurons (time constant, 2.6 ± 0.2 min, n = 8) (Fig. 5 Ai,Aii, supplemental Fig. 1B, available at www.jneurosci.org as supplemental material). The concentration of Fura-2 was converted to Fura-2 binding ratio (κ'_B) according to the method of Helmchen et al. (1996). Initial series resistance varied between 8 and 12 MΩ; cells were excluded from buffering analysis if the resistance changed by >20% during the recording period. Somatic Ca²⁺ transients were evoked by a 50 ms depolarization every 30 s. Ca²⁺ increases evoked by this stimulation protocol were stable throughout the duration of the Ca²⁺ buffering experiments (supplemental Fig. 1C, available at www.jneurosci.org as supplemental material).

Analysis of the amplitude and the decay of the Ca²⁺ transients from eight wild-type and eight Prnp^0/0 neurons during loading with 100 μm Fura-2 showed that both inverse amplitude and decay rate depended on κ'_B and were best fit by a linear regression (Fig. 5 B, supplemental Fig. 1D, available at www.jneurosci.org as supplemental material). The integral of the Ca²⁺ change was independent of κ'_B (similar to Helmchen et al., 1996) (supplemental Fig. 1Diii, available at www.jneurosci.org as supplemental material). The intercept of the regression line with the ordinate (κ'_B = 0) provides an estimate of both the Ca²⁺ change and kinetics in the absence of exogenous buffer. The negative abscissa intercept provides an estimate of the endogenous binding capacity (κ_S) (Helmchen et al., 1996). κ_S estimated from the inverse amplitude of the Ca²⁺ increase (A ⁻¹) (Fig. 5 Bi) showed that Prnp^0/0 neurons had a greater Ca²⁺ buffering capacity (Prnp ^+/+: 38 ± 10; Prnp^0/0: 89 ± 11, p < 0.005). Estimates of the Ca²⁺ signaling profile in the absence of exogenous buffering showed that the peak increase was significantly smaller in Prnp^0/0 neurons than in Prnp ^+/+ neurons (0.51 ± 0.05 μm and 1.10 ± 0.13 μm, respectively, p < 0.001). The decay rate at κ'_B = 0 was significantly slower in Prnp^0/0 neurons than wild type (0.37 ± 0.03 s and 0.22 ± 0.03 s, respectively, p < 0.005).

The Fura-2 loading experiments also allow estimation of κ_S by analyzing the depolarization evoked fluorescence decrements (ΔF ₃₈₀). κ_S can be obtained by expressing the fluorescence decrements ΔF ₃₈₀ as a function of κ'_B (Helmchen et al., 1996). Estimates of κ_S suggested that Prnp^0/0 neurons (78 ± 13) have greater buffering capacity than wild-type neurons (38 ± 6, p < 0.05) (data not shown), consistent with the results from the inverse amplitude method.

The above Ca²⁺ buffering experiments are based on a one-compartment model and rely on the assumption that the decay of the Ca²⁺ signal is described by a single extrusion rate (Eq. 4). This assumption is supported by the observation that the decay of the Ca²⁺ transient was described by a single exponential (Fig. 5). Furthermore, the decay rate depended linearly on κ_B as predicted by Equation 4 (Fig. 5 Bii). The extrusion of Ca²⁺ (γ) is a function of multiple mechanisms including active Ca²⁺ transport across the plasma membrane and uptake into intracellular organelles including the endoplasmic reticulum (ER) and mitochondria. Estimating γ from Equation 4 gave an extrusion rate for Prnp ^+/+ neurons of 123 ± 16 s⁻¹ (n = 8). In contrast, the extrusion rate for Prnp^0/0 neurons was significantly faster (240 ± 23 s⁻¹; n = 8; p < 0.001) (Fig. 5 D), suggesting that Prnp^0/0 neurons have an enhanced extrusion mechanism.

We explored the mechanisms responsible for the increased Ca²⁺ buffering capacity and the consequent reduction in slow AHP in Prnp^0/0 mice. Ca²⁺ buffering mechanisms in neurons include: extrusion across plasma membrane, binding to mobile and immobile cytoplasmic buffers and uptake into intracellular stores such as mitochondria and endoplasmic reticulum (Neher, 1998). The ER functions as a major Ca²⁺ uptake mechanism in neurons (Neering and McBurney, 1984; Sabatini et al., 2002; Usachev et al., 2006). This functional role for the ER coupled with the observations that PrP is widely expressed in the ER (Holscher et al., 2001), and that the ER has been implicated in the aberrant response of Prnp^0/0 neurons to oxidative stress (Krebs et al., 2007), led us to hypothesize that changes in the function of the ER may underlie the observed alterations in Ca²⁺ homeostasis. We examined the contribution of Ca²⁺ uptake into the ER by inhibiting the major ER Ca²⁺ uptake mechanism, the sarcoplasmic/endoplasmic reticulum pump Ca²⁺-ATPase (SERCA), with either thapsigargin or cyclopiazonic acid (CPA).

Neurons were loaded with 100 μm Fura-2 for 10 min to allow the dye to equilibrate between the patch pipette and cell (Fig. 5 A). Then, using the standard stimulation protocol of 50 action potentials delivered at 20 Hz, changes in slow AHP, basal and peak Ca²⁺ levels were monitored while the slice was perfused with 3 μm thapsigargin. Before application of thapsigargin the slow AHP, basal Ca²⁺ levels and peak Ca²⁺ change were significantly smaller in Prnp^0/0 neurons (Fig. 6 A). Application of thapsigargin, increased the amplitude of the slow AHP, and both basal and peak Ca²⁺ changes within 4 min (Fig. 6 A), and abolished the difference between the responses from Prnp^0/0 and control neurons (Fig. 6 B,C). To confirm that the reversal of the slow AHP and Ca²⁺ signaling was the result of inhibition of SERCA, we examined the effect of a second inhibitor of SERCA, CPA, on the amplitude of the slow AHP. Similar to thapsigargin (Fig. 6 A), CPA did not affect the amplitude of the slow AHP in Prnp ^+/+ neurons (control, 2.5 ± 0.3 mV; 10 μm CPA, 2.3 ± 0.3, n = 5; p = 0.42) (supplemental Fig. 2, available at www.jneurosci.org as supplemental material). In contrast, the amplitude of the slow AHP in Prnp^0/0 neurons, was initially significantly smaller than in Prnp ^+/+ neurons (control, 1.5 ± 0.3 mV; p < 0.05). Application of 10 μm CPA increased the amplitude of the slow AHP to 2.4 ± 0.5 mV (n = 5), which did not significantly differ from Prnp ^+/+ neurons in the presence of CPA (p = 0.9). Similarly, CPA (10 μm) increased the amplitude of I _sAHP in Prnp^0/0 neurons (control 43.5 ± 5.9 pA, CPA 105.8 ± 24.8 pA, n = 4, p < 0.05), but not in Prnp ^+/+ neurons (control 127.4 ± 17.8 pA, CPA 108.1 ± 9.1 pA).

Figure 6.

Block of SERCA by thapsigargin abolishes the observed changes in slow AHP and Ca²⁺ signaling Prnp^0/0 neurons. A, Bath application of thapsigargin (3 μm) increased slow AHP amplitude (top), peak Ca²⁺ levels (middle), and basal Ca²⁺ levels in Prnp^0/0 neurons (open circles), but not Prnp ^+/+ neurons (filled circles). B, C, Representative traces of voltage response (B) and Ca²⁺ changes (C) evoked by 50 action potentials at 20 Hz; control responses are shown in the left panel, thapsigargin treated responses are illustrated in the right panel (Prnp^0/0 in gray; Prnp ^+/+ in black). *p < 0.05; **p < 0.01; ***p < 0.005.

These data suggest that the observed differences from Prnp^0/0 neurons are the result of an alteration in SERCA activity. To examine the role of SERCA in Ca²⁺ buffering further, we used the added buffer method described above. Neurons were pretreated with thapsigargin (3 μm) for 10 min before establishing a whole-cell recording and performing an added-buffer experiment. In thapsigargin naive neurons, the I _sAHP, basal Ca²⁺ and peak Ca²⁺ levels were significantly smaller in Prnp^0/0 neurons than control (Fig. 7 A, Table 1). In contrast, these responses in thapsigargin pretreated Prnp^0/0 neurons were not significantly different from Prnp ^+/+ responses (Fig. 7 A, Table 1). Thapsigargin-naive Prnp^0/0 neurons showed a significantly greater κ_S (estimated from the inverse amplitude) than control neurons (80 ± 9 and 37 ± 7, respectively; n = 11) (Fig. 5). Inhibition of SERCA by thapsigargin did not affect κ_S in control neurons (28 ± 5; n = 8; p = 0.37). In contrast, inhibition of SERCA by thapsigargin significantly shifted the κ_S in Prnp^0/0 neurons (19 ± 4; n = 7; p < 0.001) (Fig. 7 B,C), toward that observed in control neurons. Calculation of the extrusion rate revealed that pretreatment with thapsigargin reversed the increase in extrusion rate observed in Prnp^0/0 neurons (thapsigargin naïve, 240 ± 23 s⁻¹; thapsigargin, 51 ± 11 s⁻¹; p < 0.001). In contrast, thapsigargin did not alter the extrusion rate in Prnp ^+/+ neurons significantly (thapsigargin naïve, 123 ± 16 s⁻¹; thapsigargin, 84 ± 10 s⁻¹; p = 0.08).

Figure 7.

Block of SERCA by thapsigargin abolishes the observed changes in slow AHP, Ca²⁺ signaling, and Ca²⁺ buffering in Prnp^0/0 neurons. To examine the role of the ER in buffering, the added buffer experiments were repeated in neurons that were either under control conditions (Tg naive) or pretreated with thapsigargin (3 μm). A, In Prnp^0/0 neurons, the reduction in both slow AHP (Ai, light gray traces) and Ca²⁺ signaling (Aii, light gray traces) was abolished by pretreatment with thapsigargin (3 μm; Ai, Aii, dark gray traces). B, C, Concomitant with these changes, the increase in the buffering capacity was also abolished. Buffering capacity was estimated from the inverse amplitude because blockade of SERCA alters Ca²⁺ decay kinetics. p < 0.005.

Table 1.

Comparison of slow AHP and Ca²⁺ signalling changes in thapsigargin naive and thapsigargin pretreated Prnp ^+/+ and Prnp^0/0 neurons

	Tg naive		Tg pretreated (3 μm)
	Prnp +/+	Prnp0/0	Prnp +/+	Prnp0/0
I sAHP (pA)	38.9 ± 4.8	10.6 ± 3.2**	37.6 ± 5.9	39.5 ± 8.3
Basal Ca2+ (nm)	166.8 ± 13.0	121.7 ± 8.3*	152.2 ± 9.7	161.7 ± 15.2
Peak Ca2+ (μm)	0.71 ± 0.09	0.37 ± 0.05*	0.70 ± 0.11	0.81 ± 0.15

I _sAHP was evoked by a 50 ms somatic depolarization. Values are mean ± SEM. *p < 0.05; **p < 0.01.

The endoplasmic reticulum provides a substantial mechanism for sequestration of action potential induced Ca²⁺ entry as blocking SERCA slows the decay of the Ca²⁺ signal (Sabatini et al., 2002). In the present study, inhibition of SERCA by thapsigargin resulted in an approximate twofold (1.92 ± 0.14; n = 10) slowing of the decay of the Ca²⁺ signal in Prnp ^+/+ neurons. In contrast, inhibition of SERCA in Prnp^0/0 neurons resulted in a significantly greater slowing of the Ca²⁺ decay time constant (2.84 ± 0.34; n = 13; p = 0.04), supporting the hypothesis that Ca²⁺ buffering is enhanced in Prnp^0/0 neurons as a result of enhanced SERCA activity. Similar results were observed when 300 nm thapsigargin was applied to neurons (data not shown).

The observation that the reduction in the slow AHP in Prnp^0/0 neurons could be rapidly reversed by inhibition of SERCA by either thapsigargin or CPA, in parallel with changes in the endogenous buffering, suggested that the κ_s could be used to predict the amplitude of the afterhyperpolarization. A plot of κ_s versus I _sAHP amplitude confirmed a linear relationship (Fig. 8 A) (regression significant at p < 0.001). This relationship was maintained between genotypes and also for neurons that were pretreated with thapsigargin. A plot of I_mAHP amplitude versus κ_s revealed no correlation (Fig. 8 B) (regression, p = 0.21).

Figure 8.

Slow afterhyperpolarization amplitude is correlated with endogenous buffering capacity. A, Plot of slow AHP amplitude versus endogenous buffering capacity (κ_s) reveals a linear relationship (coefficient of determination 0.54; p < 0.001). B, The medium AHP was not correlated to endogenous buffering capacity (coefficient of determination 0.022; p = 0.21). Filled symbols, Prnp ^+/+; open symbols, Prnp^0/0; circles, thapsigargin naive; triangles, pretreated with 3 μm thapsigargin.

Discussion

Previous studies of the functional role of PrP have reported that PrP-deficient (Prnp^0/0) mice have smaller slow afterhyperpolarizations (Colling et al., 1996; Herms et al., 2001). This reduction in Ca²⁺-activated K⁺ current could arise from impairments of: Ca²⁺ channels, K⁺ channels, or Ca²⁺ signaling. We identified the mechanisms responsible for the reduced AHP using a combined microfluorimetric and electrophysiological approach in CA1 pyramidal neurons.

We found no differences in Ca²⁺ currents (Herms et al., 2000; cf. Fuhrmann et al., 2006) and conclude that the weaker slow AHP must have another cause. To date, deficits in K⁺ channel function have not been directly examined. If the slow AHP deficit is caused by alteration in K⁺ channel function, this may arise because of alterations in the K⁺ channel itself or to alterations in the putative Ca²⁺ sensor for the slow AHP, hippocalcin (Tzingounis et al., 2007). Direct activation of the Ca²⁺-dependent K⁺ channels by photorelease of caged Ca²⁺ evoked comparable AHPs in both Prnp ^+/+ and Prnp^0/0 pyramidal neurons, despite the latter having significantly smaller action potential evoked slow AHPs. We therefore conclude that neither disruption of K⁺ channels nor alteration in the expression of the Ca²⁺ sensor is responsible for the reduction in AHP amplitude.

The most likely explanation remaining is that Prnp^0/0 neurons have altered Ca²⁺ homeostasis, as suggested previously (Herms et al., 2000). In the present study, we found a decrease in Ca transients evoked by depolarization. We used the added buffer method to investigate this in detail (Neher, 1995; Fierro and Llano, 1996; Helmchen et al., 1996) and found that the observed alterations in Ca²⁺ signaling can be explained by increases in intracellular buffering and clearance of Ca²⁺. The observation that the SERCA inhibitor, thapsigargin, abolished the increases in both buffering (κ_S) and clearance (γ) in Prnp^0/0 neurons suggested that the differences were caused by changes in the ER. The role of SERCA is to pump Ca²⁺ into the ER, resulting in Ca²⁺ clearance, but it also acts as a Ca²⁺ buffer (Higgins et al., 2006); both of these functions are inhibited by thapsigargin or CPA (Tadini-Buoninsegni et al., 2008). We conclude that changes in SERCA can explain increases in both buffering and clearance in the Prnp^0/0 neurons.

Aberrant ER function has been suggested previously to underlie alterations in the response of Prnp^0/0 neurons to oxidative stress (Krebs et al., 2007). Altered SERCA function may explain other phenotypes in Prnp^0/0 mice, for instance impaired LTP (Collinge et al., 1994): SERCA activity has been implicated in the induction of LTP (Harvey and Collingridge, 1992; Bardo et al., 2006), perhaps related to its effects on dendritic spine function (Scheuss et al., 2006).

In summary, we have shown that Prnp^0/0 neurons possess a reduced slow afterhyperpolarization that is a consequence of increased Ca²⁺ buffering in the endoplasmic reticulum resulting from increased SERCA activity. We propose that the increased Ca²⁺ buffering provides a mechanism for several previously reported phenotypes in prion null neurons.