Introduction

Thiamine is a co-factor for pyruvate dehydrogenase, α-ketoglutarate dehydrogenase, and transketolase. Although essential for the correct functioning of the Krebs cycle, no animals synthesize thiamine and it must be obtained from the diet. Skeletal muscle, heart, kidney, and liver can store thiamine; however, in humans, these stores are depleted in weeks under a diet deficient in thiamine, since the thiamine biological half-life is between 9 and 18 days (Pácal et al. 2014). The consequences of thiamine deficiency include beriberi and Wernicke encephalopathy, a neurodegenerative disease that causes loss of neurons in the central nervous system, mainly in the cerebellum (Butterworth 1993; Mulholland 2006; Pannunzio et al. 2000; Troncoso et al. 1981). Epidemiological evidence suggests that Wernicke encephalopathy is an under diagnosed disease, with an especially high prevalence among alcoholics and other undernourished individuals (Day and del Campo 2014; Galvin et al. 2010).

The brain is more vulnerable to thiamine deficiency than other tissues. Likely, this is due to the very high energy demand of the brain that makes it particularly susceptible to any impairment of energy metabolism. However, the link between thiamine deficiency and neuronal degeneration in the brain is still unclear. Several hypotheses have been raised, including increased oxidative stress, mitochondrial dysfunction, and activation of caspase 3-dependent apoptotic pathways (Chornyy et al. 2007; Kang et al. 2010; Yadav et al. 2010). One of the main triggers may be an overload of Ca2+. Previous studies demonstrated changes to Ca2+ permeable AMPA receptors leading to increased basal Ca2+ levels and Ca2+ dynamics when cultured cortical neurons were maintained in the presence of an inhibitor of the thiamine transporter (Lee et al. 2010).

Furthermore, the thiamine deficiency is a well-established model for the condition of oxidative stress (Wang et al. 2007) and knowing that NO may modulate Ca2+ channel expression in cerebellum granular neurons (Kim et al. 2004); we hypothesized that thiamine deficiency could modify CaV1.2 levels in cerebellum granular neurons. In addition, CaV1.2 blockers are easy to use for future in vitro and in vivo tests. However, the possibility of altered Ca2+ handling in cerebellum granular neurons during thiamine deficiency is currently unknown. Previously, we developed an in vitro preparation to study thiamine deficiency by maintaining cerebellum granular neurons on culture with thiamine-deficient media and studied effects on ion channels and electrophysiological parameters (Cruz et al. 2012; Oliveira et al. 2007). In the present paper, we used this preparation to measure the effects of thiamine deficiency on whole-cell Ca2+ currents and CaV1.2 protein levels in cerebellum granular neurons.

Materials and Methods

Cell Culture

The investigation was conducted according to the Guidelines for the Care and Use of Laboratory Animals published by the National Institutes of Health (USA) and the Institutional Ethics Committee for Animal Experimentation (CEUA-UFMG, protocol number 4/2010). Cells were cultured as previously described (Oliveira et al. 2007; Cruz et al. 2012). Unless stated otherwise, reagents were from Sigma-Aldrich. Briefly, newborn Wistar rats were killed by decapitation. Cerebellum were dissected and placed in Ca2+-free Hanks solution (0 °C) of the following composition (in mM): NaCl 136.9, KCl 5.3, KH2PO4 0.44, Na2HPO4 0.33, NaHCO3 4, and Glucose 5.5, pH 7.4 (adjusted with NaOH). Penicillin and streptomycin were added by including 1 % of an antibiotic stock solution containing 10.000 units/ml penicillin and 10.000 mg/ml streptomycin. The tissue was cut in small pieces (1–2 mm) and digested with trypsin (2.5 mg/ml) for 5 min. The trypsin solution was removed and the cerebellum fragments were washed three times in DMEM supplemented with 10 % fetal bovine serum (Cripion Biotechnology, Campinas, Brazil) and 1 % antibiotic solution. Cells were mechanically dissociated by passing the tissue fragments through three glass pipettes with decreasing tip diameters. Dissociated cells were centrifuged for one minute at 1000 RPM and then resuspended in thiamine (−) or thiamine (+) [0.004 g/l] DMEM supplemented with 10 % fetal bovine serum and 1 % antibiotics as above. The cells were plated at a density of 2 × 106 cells/cm2 onto poly-d-lysine-coated glass coverslips (18 × 18 mm) and maintained at 37 °C, 5 % CO2 for 1 h. After this time, the culture medium was exchanged and 0.4 μg/ml cytosine arabinoside (Ara-C) was added to prevent glial proliferation. Cells were maintained in an incubator at 37 °C in an atmosphere of 5 % CO2. Every three days, 800 μl of medium was removed from each dish and 1 ml of fresh medium (with Ara-C) was added. Cells were maintained in either thiamine (−) or thiamine (+) culture medium for 7–8 days and then used for Western blot or patch-clamp experiments.

Electrophysiological Recordings

Macroscopic currents were recorded at room temperature (24–28 °C) with the whole-cell configuration of the patch-clamp technique (Hamill et al. 1981) using an EPC-9 amplifier (HEKA Instruments, Germany) connected to a PC computer. The holding potential was −80 mV unless otherwise stated. Patch-clamp pipettes were made of soft glass capillaries (Perfecta, Brazil) using a vertical 2-stage puller (Narishige, Japan) and were backfilled with an internal solution containing (in mM): CsCl 120; MgCl2 1; EGTA 10; HEPES 10; Mg-ATP 2; and pH 7.2 (adjusted with CsOH). Open tip patch pipette resistances were 3–5 MΩ. The protocols were generated and data acquired with Pulse software (HEKA Instruments, Germany) at an acquisition frequency of 20 kHz and low-pass filtered at a cut-off frequency of 2.9 kHz. Series resistance was less than 15 MΩ and compensated by 75 %. Leak subtraction was performed online using a P/4 protocol. Leak current was <25 pA. Access resistance was verified before and after each recording and those cells that showed more than 20 % change in the access resistance were discarded from the analysis. Junction potentials were not corrected.

To obtain giga-ohm seal in neurons, we used an external solution containing (in mM): NaCl 140; KCl 5.4; CaCl2 1.8; MgCl2 0.5; NaH2PO4 0.33; Glucose 11; HEPES 5; and pH 7.4 (adjusted with NaOH). After achieving whole-cell configuration, the cell was superfused at 6 μl/s with an external solution containing (in mM): TEA-Cl 130; MgCl2 2; BaCl2 10; HEPES 10; Glucose 10; and pH 7.2 (adjusted with TEA-OH). The external and pipette solutions were designed to completely suppress Na+ and K+ currents, leaving only the Ba2+ currents (IBa) through voltage-dependent Ca2+ channels.

Stimulation Protocols

To record Ba2+ currents, two protocols were applied: slow ramps and square wave pulse protocols. In both of these protocols, cells were held at −80 mV. Ramps were from −80 mV to +50 mV over 100 ms. The ramp protocol was repeated 12 times, once every 15 s, starting immediately after achieving whole-cell. Following the series of ramps protocols, we applied square wave protocol consisted of test pulses of 70 ms duration ranging from −50 to 50 mV in 5 mV steps, followed by a test pulse to 0 mV to measure current inactivation. For I–V and G–V analysis, we only included cells that exhibited less than 10 % rundown from the beginning to end of the ramp protocols. Between step depolarizations, cells were held for 15 s at −80 mV to allow full recovery from inactivation.

Immunoblotting

Cerebellum neuronal cultures were maintained in thiamine (+) or thiamine (−) culture media for 7 days and then lysed in Triton buffer (in mM: NaCl 150, Tris–HCl 50, 1 % Triton X-100, and pH 7.4) containing protease inhibitors (1 mM AEBSF and 10 µg/ml of both leupeptin and aprotinin). 100 µg of total cellular protein for each sample were subjected to SDS-PAGE, followed by electroblotting onto nitrocellulose membranes. Membranes were blocked with 10 % non-fat milk in wash buffer (in mM) NaCl 150, Tris–HCl 10, 0.05 % Tween-20, and pH 7.4 for 1 h and then incubated with either rabbit anti-CaV1.2 (1:200; Millipore #MAB13170) or rabbit anti-actin (1:1000, Sigma-Aldrich) antibodies in wash buffer containing 3 % non-fat milk for 2 h at room temperature (24–26 °C). Membranes were rinsed three times with wash buffer and then incubated with secondary peroxidase-conjugated anti-rabbit IgG antibody diluted 1:5000 in wash buffer containing 3 % non-fat milk for 1 h. Membranes were rinsed three times with wash buffer and incubated with ECL Western blotting detection reagents.

Modeling

Simulations were performed in the Neuron modeling environment (Hines and Carnevale 1997). The original cerebellum granular neuron model was published by D’Angelo et al. (2001). Recently, we described how this model could be modified to reflect the effects of thiamine deficiency on A-type K+ currents (for details, Cruz et al. 2012). Here, we extend our modifications of the model to account for the 50 % increase in HVA Ca2+ currents. We modeled two situations, normal HVA Ca2+ current density (control) and 1.5 times the normal density of HVA Ca2+ current, as recorded experimentally in thiamine-deficient neurons. As we described in detail previously (Cruz et al. 2012), the functional effects on excitability can be probed using simulated synaptic stimulation. We examined two aspects of excitability: refractory periods (defined here as an increase in threshold for a second EPSC following a previous suprathreshold EPSC) and temporal summation of EPSPs (defined here as a decrease in threshold for a second EPSC following a previous subthreshold EPSC). In this approach, the model cell was stimulated with pairs of synaptic inputs (conductances) that were based on the kinetics previously describe for mossy fiber inputs to granular neurons (Sola et al. 2004). The interval between pairs of synaptic conductance was varied systematically from 1.5 to 160 ms. At each time interval, refractoriness and temporal summation were measured. To measure refractoriness, the amplitude of the first synaptic input was set to 1.5 times threshold (i.e., 1.5 times the minimum synaptic input necessary to generate an action potential) and then threshold was determined for the second synaptic input as the minimum synaptic conductance needed to generate an action potential. To measure temporal summation, the amplitude of the first synaptic input was set to 90 % of threshold and then threshold for the second synaptic input was determined as for refractoriness. In both cases, the threshold synaptic conductance of the second stimulus was measured as a function of the interpulse interval and normalized relative to its values for infinitely long intervals.

Analysis

Electrophysiology Data

Currents were analyzed using routines developed in the Python and C++ programming languages written by DCML. Ca2+ current amplitudes were defined as the maximal inward current during ramp or step depolarization protocols. Ca2+ conductance was determined by fitting Eq. 1 to current–voltage (I–V) data.

$$i\left( v \right) = g_{ \hbox{max} } \times \left( {V - E_{\text{Ca}} } \right) \times \left( {1 + e^{{\frac{{ - \left( {v - v_{\text{a}} } \right)}}{{k_{\text{a}} }}}} } \right)^{ - 1},$$
(1)

where g max is the maximum Ca2+ conductance, E Ca is the Ca2+ current reversal potential, and (~+50 mV) determined individually for each cell as the intersection between the linear rising phase of the I–V relationship and the v axis. v is the test potential, v a is the voltage required for half-maximal activation, and k a is the steepness of the voltage dependence.

Conductance (g) as a function of membrane potential (v) was calculated according to Eq. 2:

$$g\left( v \right) = i/\left( {v - E_{\text{Ca}} } \right),$$
(2)

where E Ca is the Ca2+ current reversal potential (see above). The derived G–V data was then used to adjust Eq. 3 to determine g max, the half activation potential (v a), and the steepness of the voltage dependence (k).

$$g\left( v \right) = g_{ \hbox{max} } \left( {1 + e^{{\frac{{ - \left( {v - v_{\text{a}} } \right)}}{{k_{\text{a}} }}}} } \right)^{ - 1}.$$
(3)

To obtain the voltage for half inactivation, the normalized current was fitted by a modified Boltzmann equation (Eq. 4):

$$i\left( v \right) = \left( {1 - p} \right)/\left( {1 + e^{{\frac{{\left( {v - v_{i} } \right)}}{{k_{i} }}}} } \right) + p,$$
(4)

where, i is the current at the normalized test pulse, p is the fraction of the current that inactivated, v is the voltage of the applied pre-pulse, v i is the half inactivation potential, and k i is the steepness of the voltage dependence.

Immunoblotting Analysis

Non-saturated immunoreactive CaV1.2 and actin bands were quantified by scanning densitometry from total cellular homogenate protein content of each sample. Immunoband intensity was calculated using ImageJ software (NIH), and the intensity of the pixels of CaV1.2 band was normalized by the intensity of pixels of actin band and reported as percentage.

Statistical Tests

Ca2+ current I–V relationship are based on data from at least 5 independent cell cultures. Data are expressed as the mean ± SEM. Statistical testes were done using SigmaPlot 11.0 (Systat Software, Inc). Probability values of p < 0.05 were considered to be statistically significant.

Results

To determine the consequences of thiamine deficiency on voltage gated Ca2+ channels, we performed whole-cell voltage clamp experiments using Ba2+ as the charge carrier under conditions that isolate Ca2+ channel current. During ramp protocols, peak current in thiamine-deficient cells was significantly greater than in controls [Thiamine (+): −64.5 ± 3.13 pA/pF, n = 21; Thiamine (−): −94.2 ± 7.65 pA/pF, n = 15; data not shown]. To study the I–V relationship in more detail, we applied step depolarizations ranging from −50 to 50 mV. As shown in Fig. 1, thiamine deficiency had two marked effects on Ca2+ channel currents: an increase in peak current density and a positive shift in the voltage dependence of activation. When measured during a step depolarization to 0 mV (Fig. 1a, b), Ca2+ currents were −62.7 ± 11.6 pA/pF in control cells and −103 ± 11.6 pA/pF in thiamine-deficient cells [n = 5 for thiamine (+) and n = 7 for thiamine (−)]. To determine the increase in conductance, we adjusted Eq. 1 (see “Materials and Methods”) to the I–V data (curves in Fig. 1b). Based on the best-fit parameters, Ca2+ channel conductance was increased 55 % by thiamine deficiency [g max = 1.39 nS/pF in thiamine (+) vs. 2.15 nS/pF in thiamine (−)].

Fig. 1
figure 1

Thiamine deficiency increases Ca2+ current density and shifts the voltage dependence for activation in cerebellum granular neurons. a Representative current traces recorded in response to 0 mV depolarization in thiamine (+) (Ctrl; gray line) and thiamine (−) (black line) conditions. b I–V relationships of Ca2+ channel current (measured using Ba2+ as charge carrier). Curves represent the best fit of Eq. 1 (see “Materials and Methods” section) to the data [n = 5 for thiamine (+) and n = 7 for thiamine (−)]. c Normalized G–V relationships obtained from the data in (b). Curves represent the best fit of Eq. 2 (see “Materials and Methods” section) to the data. d Steady-state inactivation. Curves represent the best-fit to Eq. 3 (see “Materials and Methods” section), *p < 0.05

Full size image

To quantify the effects of thiamine deficiency on the voltage dependence of activation, we converted the I–V data shown in Fig. 1b to normalized G–V by dividing the amplitude of the measured current by the driving force based on the Ca2+ current reversal potential and then normalizing to the largest conductance value (Fig. 1c). By fitting Eq. 3 (see “Materials and Methods” section) to these data, we determined that thiamine deficiency caused a positive shift of 6.6 mV in the voltage required for half-maximal activation [v a = −19.2 mV in thiamine (+) cells vs. −12.6 mV in thiamine (−) cells]. In contrast, thiamine deficiency did not modify the steepness of the G–V relationship [k a = 5.8 mV in thiamine (+) cells vs. 6.2 mV in thiamine (−) cells].

To determine if thiamine deficiency altered the voltage dependence of steady-state inactivation, we measured peak Ca2+ channel current available after holding the cell for 70 ms at depolarized potentials (Fig. 1d). We observed no significant difference in the potential for half inactivation [v i  = –23.0 mV in thiamine (+) cells vs. −22.9 in thiamine (−) cells].

To estimate the amount of L-type Ca2+ channel protein, immunoblotting was performed on cerebellum granular neurons under thiamine (+) and thiamine (−) conditions. Figure 2 shows CaV1.2 L-type Ca2+ channel protein is increased in neurons maintained in thiamine (−) conditions [CaV1.2/Actin = 0.29 ± 0.02 in thiamine (+) cells vs. 1.08 ± 0.18 in thiamine (−) cells], a 2.7-fold increase.

Fig. 2
figure 2

Immunoblotting for L-type Ca2+ channel in cerebellum granular neurons (CGN). CGN cultures were maintained in thiamine (−) [n = 6] or thiamine (+) [n = 6] conditions and then lysed to quantify L-type Ca2+ channels using anti-CaV1.2 antibody (see “Materials and Methods” section). *p < 0.05

Full size image

Cerebellum granular neurons receive synaptic inputs from mossy fibers, and temporal summation of mossy fiber EPSPs is required to bring the granule neuron to threshold (Chadderton et al. 2014). To determine the impact of increased Ca2+ channel current on granular neurons firing properties, we modified an existing biophysical model and simulated action potential firing evoked by synaptic stimulation. The biophysical model was based on one published by D’Angelo et al. (2001) and modified to include the observed effects of thiamine deficiency on A-type K+-currents (Cruz et al. 2012) and HVA Ca2+ currents (this study). The model cell was stimulated with synaptic conductance based on fast mossy fiber excitatory post-synaptic currents (Sola et al. 2004). We examined two aspects of excitability: refractory periods (an increase in threshold for a second EPSC following a previous suprathreshold EPSC) and temporal summation (a decrease in threshold for a second EPSC following a previous subthreshold EPSC). These two parameters were determined as a function of the interval between paired stimuli (Paired-pulse interval, Fig. 3). As previously reported (Cruz et al. 2012), the model predicts decreased refractoriness when the effect of thiamine deficiency on A-type K+ current was included (Fig. 3a, black vs. gray lines). The model predicts little additional effect of increased HVA Ca2+ current (Fig. 3a, solid vs. dashed lines). The major effect of increased Ca2+ current was a small increase in the absolute refractory period (Fig. 3a, rightward shift in the position of the symbol). As shown in Fig. 3b, increased HVA Ca2+ current had essentially no effect on temporal summation, presumably because the first subthreshold EPSC does not significantly activate these channels.

Fig. 3
figure 3

Computational analysis of the effects of HVA Ca2+ current conductance on refractory periods and temporal summation. a Refractory periods. The model neuron was stimulated with pairs of synaptic inputs with intervals ranging from 1.5 to 160 ms. The conductance of the first synaptic input was set to 1.5 times action potential (AP) threshold, and the threshold of the second synaptic input to generate an action potential was measured as a function of paired-pulse interval. The symbol indicates the absolute refractory period. Black lines represent control thiamine (+) K+ currents, gray lines represent modified A-type K+ current as reported in Cruz et al. (2012). Solid lines represent control Thiamine (+) HVA Ca2+ currents, broken lines represent Thiamine (−) modified HVA Ca2+ currents. b Temporal summation. Threshold of a second synaptic conductance was measured when the first synaptic conductance was set to 0.9 times AP threshold. Same color code as for (a). For all cases, thresholds are normalized the control values (Black solid lines) at the longest interval

Full size image

Discussion

In previous studies, we have reported changes to voltage gated channels caused by thiamine deficiency (Cruz et al. 2012; Oliveira et al. 2007). In this paper, we show that thiamine deficiency causes an increase of HVA Ca2+ protein levels of HVA Ca2+ channel and increased the size of Ca2+ currents in cerebellum granular neurons. These data imply a larger influx of Ca2+ during each action potential and could result in Ca2+ overload. Abnormally high Ca2+ entry in neurons has been associated with neurodegenerative conditions (Tymianski et al. 1993; Szydlowska and Tymianski 2010; Mattson 2007) and learning and memory impairments (Kawamoto et al. 2012; Oh et al. 2010; Veng et al. 2003). Several studies have shown that intracellular Ca2+ deregulation participates in important pathological processes in Alzheimer’s disease (AD). In particular, β-amyloid accumulation may induce intracellular Ca2+ enhancements, leading to neuronal cell dysfunction and death (Davidson et al. 1994; Kim and Rhim 2011; MacManus et al. 2000; Ueda et al. 1997; Gleichmann and Mattson 2011).

Cerebellum granular neurons express several different types of Ca2+ channels. In a previous study using very similar experimental conditions, Randall and Tsien (1995) distinguished five pharmacologically distinct HVA Ca2+ channel currents in rat cerebellum granular neurons: nimodipine-sensitive L-type Ca2+ current, ω-conotoxin-GIVA-sensitive N-type Ca2+ current, ω-Agatoxin-IVA-sensitive P- and Q-type Ca2+ currents, and R-type current that is resistant to all of the above. Of these five components, the L-type component accounts for about 15 % of the total. In the present study, we observed a 2.7-fold increase in the levels of CaV1.2 L-type Ca2+ channel protein in cells from thiamine-deficient cultures. Based on the results of the aforementioned study (Randall and Tsien 1995), such a 2.7-fold increase in the L-type component would result in a 41 % increase in the total HVA Ca2+ current. It thus appears that the observed increase in CaV1.2 protein can explain most, but not all, of the observed increase in HVA Ca2+ channel current. The remaining increase in HVA Ca2+ channel current may be due to increased expression of other alpha subunits beside CaV1.2. One caveat to this calculation is that Western blot experiments measured total cellular protein, whereas the electrophysiological recordings detect only functionally channels in the plasma membrane. If a significant fraction of the increased CaV1.2 protein level does not form functional channels, our calculation will overestimate its role in the alterations observed in cells from thiamine-deficient cultures.

The signaling pathway whereby thiamine deficiency results in increased Ca2+ channel protein levels is unknown. Future experiments should explore these pathways and determine if they are conserved in different regions of the brain and neurodegenerative conditions. In a previous study, Randall and Tsien (1995) observed the density of HVA Ca2+ channel current in cerebellum granular neurons as a function of time in culture. They reported very low Ca2+ currents after 1 day in culture, with a progressive increase until day 5, at which point Ca2+ currents stabilized. We observed Ca2+ currents at a single time point of 7 days in culture because our previous results indicate that 7 days is sufficient for the in vitro effects of thiamine deficiency to stabilize (Cruz et al. 2012; Oliveira et al. 2007). We thus do not know at which stage during the maturation of the cell culture Ca2+ currents in cells from thiamine-deficient cultures begin to differ significantly from control cultures. Possibly, the differences in the pattern of expression of HVA Ca2+ channels in cells from thiamine-deficient cultures as compared to control cultures may change at different stages of culture maturation.

Neurodegeneration caused by thiamine deficiency occurs primarily in encephalic areas, especially the cerebellum (Mulholland 2006; Troncoso et al. 1981; Pannunzio et al. 2000; Butterworth 1993). Previous studies have linked thiamine deficiency to Ca2+ deregulation in the thalamus (Hazell et al. 1998) and cortical neurons (Lee et al. 2010), however both used a pharmacological approach to develop thiamine deficiency. Here, we describe, for the first time, Ca2+ deregulation in cerebellum granular neurons caused by thiamine deficiency, without any pharmacological interference. Ca2+ homeostasis is a pivotal process in a cell that appears to be finely tuned by several systems including an array of voltage gated Ca2+ channels and Ca2+ buffering mechanisms (Berridge et al. 1999; Berridge et al. 2000; Bootman et al. 2001; Oh et al. 2013; Maravall et al. 2000). Previously, using the same in vitro approach to study ion channels in granule neurons, we reported a decrease in A-type K+ current caused by thiamine deficiency. In that study, computation modeling predicted an increase in excitability (Cruz et al. 2012). In the current study, computation modeling indicates that the increase in Ca2+ current caused by thiamine deficiency does not compensate for the effects of decreased K+ conductance. Future studies should test the model predictions using recordings in cerebellar brain slices.

Thiamine deficiency is a neurodegenerative condition affecting the energy metabolism. Several neurodegenerative diseases show energy metabolism impairments and Ca2+ signaling alterations. This link between energy metabolism and Ca2+ homeostasis open a broad field in the search for new targets in neurodegenerative diseases.

Conclusion

In conclusion, thiamine deficiency augments CaV1.2 protein levels and increases Ca2+ current density in cerebellum granular neurons which probably leads to increase in Ca2+ entry building up a scenario predisposing to neurodegeneration.