Approximately one-third of epilepsy patients are pharmacoresistant. Overexpression of P-glycoprotein and other multidrug transporters at the blood–brain barrier is thought to play an important role in drug-refractory epilepsy. Thus, quantification of regionally different P-glycoprotein activity in the brain in vivo is essential to identify P-glycoprotein overactivity as the relevant mechanism for drug resistance in an individual patient. Using the radiolabeled P-glycoprotein substrate (R)-[11C]verapamil and different doses of coadministered tariquidar, which is an inhibitor of P-glycoprotein, we evaluated whether small-animal positron emission tomography can quantify regional changes in transporter function in the rat brain at baseline and 48 h after a pilocarpine-induced status epilepticus. P-glycoprotein expression was additionally quantified by immunohistochemistry. To reveal putative seizure-induced changes in blood–brain barrier integrity, we performed gadolinium-enhanced magnetic resonance scans on a 7.0 tesla small-animal scanner. Before P-glycoprotein modulation, brain uptake of (R)-[11C]verapamil was low in all regions investigated in control and post-status epilepticus rats. After administration of 3 mg/kg tariquidar, which inhibits P-glycoprotein only partially, we observed increased regional differentiation in brain activity uptake in post-status epilepticus versus control rats, which diminished after maximal P-glycoprotein inhibition. Regional increases in the efflux rate constant k2, but not in distribution volume VT or influx rate constant K1, correlated significantly with increases in P-glycoprotein expression measured by immunohistochemistry. This imaging protocol proves to be suitable to detect seizure-induced regional changes in P-glycoprotein activity and is readily applicable to humans, with the aim to detect relevant mechanisms of pharmacoresistance in epilepsy in vivo.
Approximately one-third of patients suffering from epilepsy is pharmacoresistant, i.e., does not respond to an adequate antiepileptic drug therapy (Regesta and Tanganelli, 1999). Among patients with the most frequent epileptic syndrome, i.e., temporal lobe epilepsy, up to 70% are drug resistant (Kwan and Brodie, 2000). Cumulating evidence suggests that regional overactivity of efflux transporters at the blood–brain barrier (BBB) is one important mechanism contributing to the phenomenon of drug resistance by impeding therapeutically effective concentrations of antiepileptic drugs at their sites of action (“transporter hypothesis”) (Löscher and Potschka, 2005a). P-glycoprotein (Pgp), which is physiologically located at the luminal membrane of brain capillary endothelial cells, is currently the most widely studied multidrug transporter. Increased expression of Pgp has been found in epileptogenic brain specimens resected from patients with intractable epilepsy (Tishler et al., 1995; Lazarowski et al., 1999; Sisodiya et al., 2002). Moreover, status epilepticus (SE) or frequent spontaneous seizures in rodent models led to increased expression of Pgp, resulting in decreased brain levels of the antiepileptic drug phenytoin (Rizzi et al., 2002; van Vliet et al., 2007b; Bankstahl and Löscher, 2008), which has been characterized as a Pgp substrate (Potschka and Löscher, 2001; Baltes et al., 2007; Luna-Tortós et al., 2008).
Quantification of Pgp overactivity in epilepsy patients by in vivo imaging would be highly useful because altered treatment strategies, e.g., coadministration of a specific Pgp inhibitor or changeover to a nonsubstrate antiepileptic drug, could then be applied. Up to now, there is no validated diagnostic method that allows for in vivo measurement of regionally different Pgp expression and function at the BBB. Several studies demonstrated that positron emission tomography (PET) with 11C-radiolabeled Pgp substrates, such as (R)-[11C]verapamil (VPM), is a promising tool for in vivo investigation of Pgp function at the rat, monkey, and human BBB (Langer et al., 2007; Bankstahl et al., 2008; Liow et al., 2009; Seneca et al., 2009). Low brain uptake of high-affinity Pgp substrates such as VPM, however, limits their suitability as PET tracers for mapping regional overexpression in Pgp activity. We tried to overcome this drawback by performing PET scans after partial inhibition of Pgp. This results in sufficient brain activity uptake for PET imaging without complete Pgp blockade (Kuntner et al., 2010). Thus, regionally specific differences in Pgp expression and functionality after SE may become visible. In the present study, we evaluated whether small-animal PET imaging with VPM after partial Pgp inhibition by tariquidar (TQD) can be used to quantify regional changes in transporter function in the injured rat brain.
Materials and Methods
Adult female Sprague Dawley rats (Harlan Nederland) were used for all experiments. As in our previous studies on pilocarpine-induced SE in rats, we used female rats, because they are easier to handle after SE and eliminate various drugs considerably more slowly than male rats (Löscher, 2007). The female rats were housed without males to keep them acyclic or asynchronous with respect to their estrous cycle (Kücker et al., 2010). We have shown previously that this avoids effects of estrous cycle on seizure susceptibility or severity (Wahnschaffe and Löscher, 1992). Animal housing facilities were kept at a temperature of 22 ± 1°C and a humidity of 40–70%. Rats had ad libitum access to food and water and were kept under a 12 h light/dark cycle. Before being used in the experiments, rats were allowed to adapt to the new conditions for at least 1 week. The study was approved by the institutional animal care and use committees, and all study procedures were performed in accordance with the European Communities Council Directive of November 24, 1986 (86/609/EEC). All efforts were made to minimize both the suffering and the number of animals used in this study.
Chemicals and drugs.
Unless otherwise stated, all chemicals were of analytical grade and obtained from Sigma-Aldrich Chemie or Merck and used without additional purification. Isoflurane was obtained from Baxter Vertriebs, TQD from Xenova, diazepam (Faustan) from Temmler Pharma, and dimeglumine gadoterate (Gd-DOTA; Dotarem) from Guerbet. Lithium chloride, methyl scopolamine, pilocarpine, Evan's Blue (EB), and chloral hydrate were dissolved in 0.9% saline. TQD was freshly dissolved on each experimental day in 2.5% aqueous dextrose solution. VPM was synthesized from (R)-norverapamil (ABX Advanced Biochemical Compounds) and [11C]methyl triflate as described previously (Brunner et al., 2005).
SE was induced by pilocarpine using a fractionated protocol as described in detail previously (Glien et al., 2001). Briefly, the evening before pilocarpine administration, 127 mg/kg lithium chloride were administered orally. Twelve to 14 h later, 1 mg/kg methyl scopolamine was injected intraperitoneally to reduce peripheral adverse effects of pilocarpine. After 30 min, 10 mg/kg pilocarpine was injected intraperitoneally every half hour until onset of generalized SE. Self-sustaining SE was terminated after 90 min by intraperitoneal administration of diazepam (up to 25 mg/kg).
PET experimental procedure.
Full details of the PET experimental procedure were described recently (Kuntner et al., 2010). Throughout the whole experimental procedure, animals were kept under isoflurane anesthesia. One hour before the first PET scan, each animal was implanted with microtubes (Kleinfeld) into the femoral artery and vein to allow repeated arterial blood sampling and administration of TQD and VPM, respectively. After surgery, animals were positioned on a μPET bed (animal cradle from Bruker BioSpin) that was kept at 38°C. A stereotactic holder attached to the bed consisting of ear plugs and a tooth bar was used to fixate the animals' head to ensure a reproducible position. A μPET Focus220 scanner (Siemens Medical Solutions) was used, which consisted of 168 detector modules providing a 7.6 cm axial and 22 cm transaxial field of view. Reconstructed image resolution (filtered backprojection) is 1.3 mm (full-width at half-maximum) in the central field of view and remains under 2 mm within the central 5-cm-diameter field of view. Before each baseline PET scan, a transmission scan using a 57Co point source was recorded over 10 min. List mode data were acquired for the defined time period with an energy window of 350–750 keV and 6 ns timing window.
Simultaneously with the first VPM injection, a dynamic 60 min PET scan was started. After this baseline scan, TQD was administered at the half-maximum effective dose (ED50) of 3 mg/kg (Kuntner et al., 2010) to control rats (n = 6) and rats that had exhibited an SE 48 h before (post-SE rats; n = 6). Additional control (n = 6) and post-SE (n = 5) rats received 15 mg/kg TQD for complete Pgp inhibition (Kuntner et al., 2010). Two hours after TQD administration, the second 60 min PET scan (inhibitor scan) was recorded (Fig. 1). During the first 3 min after radiotracer injection, arterial blood samples were continuously taken using preweighted 2 μl micropipettes, followed by additional 10 μl samples taken at 5, 10, 20, 30, 40, and 60 min after the beginning of the PET scan. Moreover, one blood sample of 600 μl was collected at 20 min (baseline scan) and 60 min (inhibitor scan). Radioactivity in the blood samples was measured in a one-detector Wallac gamma counter (PerkinElmer Life and Analytical Sciences), which was cross-calibrated with the PET camera. The removed blood volume was substituted with approximately the same volume of 0.9% saline containing 20 IU/ml of sodium–heparin. At the end of the inhibitor scan, isoflurane anesthesia was deepened, and rats were killed by drawing a terminal blood sample that was centrifuged for 5 min (4000 rpm, 25°C, Rottanta/TRC centrifuge) to obtain plasma. One aliquot of the plasma sample was directly used to assess metabolism of VPM using a previously described solid-phase extraction assay (Luurtsema et al., 2005; Abrahim et al., 2008). Another aliquot was used to quantify plasma concentrations of TQD as described previously (Wagner et al., 2009).
PET data analysis.
PET data from the 60 min dynamic scans were sorted into three-dimensional sinograms according to the following frame sequence: 8 × 5 s, 2 × 10 s, 2 × 30 s, 3 × 60 s, 2 × 150 s, 2 × 300 s, and 4 × 600 s. PET images were reconstructed by Fourier rebinning of the 3-D sinograms followed by two-dimensional filtered backprojection with a ramp filter, resulting in a voxel size of 0.6 × 0.6 × 0.8 mm3. A standard data correction protocol (normalization, attenuation, and decay correction) was applied. A calibration factor for converting units of PET images into absolute radioactivity concentration units was generated by imaging a phantom filled with a known concentration of VPM.
For analysis of PET scans before and after TQD administration, five brain regions of interest (ROIs) (cerebellum, frontal motor cortex, corpus striatum, thalamus, and hippocampus) were manually outlined on multiple coronal MR images of a naive female Sprague Dawley rat using PMOD (pixelwise modeling software; version 2.7.5; PMOD Group) and rigidly coregistered with PET images as described in detail recently (Kuntner et al., 2010). ROIs were assigned to the PET images and time–activity curves (TACs), expressed in units of kilobecquerels per milliliters, for the selected brain ROIs were extracted for compartmental modeling. In addition, TACs in individual brain ROIs were normalized to injected radiotracer dose and expressed in units of percentage injected dose per gram tissue (%ID/g). From the dose-normalized time activity curves, the area under the curve (AUC) (in %ID*h/g) was calculated using the OriginPro 7.5G software package (OriginLab Corp.).
Kinetic modeling of VPM.
Blood activity concentration data were corrected for radioactive decay and converted into plasma concentrations by multiplication with the mean plasma-to-blood activity ratio of all animals at 20 min after tracer injection. Because we found no differences in VPM metabolism between control and post-SE rats (see below), metabolite correction derived from six control rats was applied to activity concentration data at the 10, 20, 30, 40, and 60 min time points (Kuntner et al., 2010). A two-tissue four-rate-constant (Fig. 2) compartment model best fitted the VPM time–activity data in rat brain as described previously (Bankstahl et al., 2008). The rate constants K1 and k2 describe exchange of radioactivity between the plasma and the first tissue compartment, whereas k3 and k4 describe exchange of radioactivity between the first fast and the second slow tissue compartment (Fig. 2). The used nomenclature is in accordance with the “consensus nomenclature for in vivo imaging” (Innis et al., 2007).
Magnetic resonance imaging.
Six additional rats underwent T1- and T2-weighted baseline MR scans before and 48 h after SE as well as T1-weighted scans after administration of Gd-DOTA (0.5 mmol/kg, i.v.) to investigate whether severe SE-induced BBB leakage is detectable. MR imaging was conducted on a 7.0 tesla animal scanner with a 38-mm-volume coil (T10327V3) serving as both transmitter and receiver coil (Pharmascan 70/16; Bruker BioSpin). Images of coronal sections were acquired by using a T2-weighted multi-slice, multi-echo sequence with a 5500 ms repetition time and a 35 ms echo time, as well as a T1-weighted modified driven equilibrium Fourier transform sequence with a 4000 ms repetition time and a 3.5 ms echo time. The field of view used was 3.5 × 3.5 cm2 with a 256 × 256 matrix. To provide detailed anatomical structure, a thin-slice thickness of 0.8 mm was used.
Assessment of albumin extravasation.
EB is highly bound to albumin and can be visualized by fluorescence microscopy and is therefore used as a marker of albumin uptake into the brain (van Vliet et al., 2007a). To add a more sensitive method for regional assessment of BBB integrity, EB was administered under a short isoflurane anesthesia (50 mg/kg in 4 ml/kg, i.v.) in six control rats and nine rats 48 h after SE. Two hours later, the animals were transcardially perfused under deep chloral hydrate anesthesia with 4% paraformaldehyde in 0.1 m phosphate buffer, pH 7.4. Brains were removed, and 40-μm-thick coronal sections were cut on a freezing microtome. Slices were covered using a DAPI-containing mounting medium (Dianova) to facilitate regional orientation during fluorescence microscopy. Brain regions corresponding to PET ROIs were visually inspected to evaluate potential cellular EB uptake by a blinded experimenter.
Immunohistochemical labeling and computer-assisted quantification of Pgp expression.
For immunohistochemical analysis of Pgp expression, additional animals (control, n = 5; 48 h after SE, n = 8) were killed for staining of Pgp in brain-capillary endothelial cells. We used a previously published protocol to enhance signal-to-noise ratio for quantification of Pgp expression (Bankstahl and Löscher, 2008). Brains were removed immediately, snap frozen in precooled isopentane, and stored at −20°C until sectioning. Coronal brain sections (14 μm) at levels of +3.2, −2.3, −3.8, −5.8, and −9.6 mm relative to bregma according to Paxinos and Watson (2007) were cut in a cryostat (Microm) and thaw mounted on HistoBond slides (Marienfeld). All sections of one section level were stained simultaneously to reduce staining variability.
Analysis of the Pgp-labeled surface area in DAB-stained sections was performed by a computer-assisted system as described in detail previously (Volk et al., 2004). Depending on the size of each investigated region (cerebellum, frontal motor cortex, corpus striatum, thalamus, and hippocampus, further separated into CA1, CA3, dentate gyrus granule cell layer, and dentate hilus), 3–10 fields of 38.321 μm2 were chosen for analysis of Pgp expression. After standardized adjustment of light intensity and definition of a threshold for every region, the area of Pgp-positive labeling relative to the total area of the fields was measured per region and animal. Because of comparison of different brain section levels, results were normalized using the appropriate control group. In analogy to the PET scan analysis, cerebellum, frontal motor cortex, corpus striatum, thalamus, and hippocampal subregions (CA1, CA3, dentate hilus, and dentate gyrus) were chosen as ROIs for analysis. Mean values were calculated for ROIs that were present in multiple section levels. The mean of the hippocampal subregions was taken for the correlation analysis between outcome parameters of kinetic model analysis and Pgp expression.
Group differences of all parameters were analyzed by one-way ANOVA followed by two-tailed Student's t test using GraphPad Prism 5.0 software (GraphPad Software). Pearson's correlation coefficients were calculated for analysis of relationship between changes in modeling outcome parameters and Pgp expression. A p value ≤ 0.05 was considered significant.
Quantification of regional VPM brain uptake by small-animal PET
Before TQD administration, brain uptake was generally low, ranging from 1.94% ID*h/g in the hippocampus to 3.50% ID*h/g in the frontal motor cortex. Baseline brain uptake did not differ between groups, except for the cerebellar region of post-SE rats being reduced by 25.3% compared with controls (see Fig. 4A). Maximal Pgp inhibition with 15 mg/kg (Figs. 3, 4C) resulted in ∼5.5-fold increased brain uptake in the cerebellum compared with 12.1-fold increase in the thalamus relative to the baseline scan.
After partial Pgp inhibition with 3 mg/kg (Figs. 3, 4B), cerebellar brain uptake of post-SE rats was decreased by 30.3% compared with controls, whereas brain uptake in the frontal motor cortex was increased by 35.9%. Group differences determined with partial Pgp inhibition in the five investigated ROIs disappeared after maximal Pgp inhibition (15 mg/kg) (Figs. 3, 4C). Therefore, this dose level was not further analyzed.
Compartmental modeling of PET data was performed for groups with partial and without Pgp inhibition. Without Pgp modulation by TQD, significant group differences in model outcome parameters were not only observed in the cerebellum but also in the thalamus (Fig. 5, Table 1). A decrease in distribution volume VT and influx rate constant K1 values of 33.2 and 26.8%, respectively, was detected in the cerebellum. In the thalamus, VT was decreased by 20.4%, whereas the efflux rate constant k2 was increased by 39.0%.
After partial Pgp inhibition, group differences became more apparent both in terms of number and extent (Fig. 5, Table 1). We detected VT decreases in cerebellum (−41.5%) and thalamus (−19.0%) of post-SE rats. In contrast, we found a VT increase of 18.6% in the frontal motor cortex. The influx rate constant K1 reflected the VT group differences only in the cerebellum in terms of a 41.0% decrease of post-SE rats. The K1 values of the four other brain regions were comparable. It is noteworthy that regional K1 values after 3 mg/kg TQD were well below rates of regional cerebral blood flow (CBF) in brain tissue of naive rats under isoflurane anesthesia [e.g., regional cerebral blood flow of ∼3.5, 4.1, and 5.5 ml · g−1 · min−1 in hippocampus, cortex, and thalamus, respectively (Hendrich et al., 2001)], indicating that in individual brain regions functional Pgp activity was still present to limit the extraction of VPM. Importantly, partial Pgp inhibition resulted in markedly increased k2 values in cerebellum (+70.7%), thalamus (+64.5%), and hippocampus (+46.6%) of post-SE rats compared with control rats. Analysis of k3 values did not reveal any group differences between control and post-SE rats neither with nor without Pgp inhibition, whereas a global decrease of k4 was observed in post-SE rats after partial Pgp inhibition. Nevertheless, partial Pgp inhibition led to globally increased k3 values in both control and post-SE rats (Table 1).
VPM metabolism did not differ significantly between control and post-SE rats. At 20 min after tracer injection during baseline scan, the percentage of unchanged VPM and its lipophilic [11C] metabolites was 66.1 ± 3.9% (n = 5) in control rats and 70.7 ± 4.0% (n = 6) in post-SE rats. At the end of the inhibitor scan (3 mg/kg TQD), these values amounted to 70.5 ± 13.1 and 77.0 ± 6.3%, respectively. Plasma levels of TQD at 180 min after administration of the 3 mg/kg dose were higher in post-SE rats (0.718 ± 0.074 mg/ml) compared with control rats (0.537 ± 0.090 mg/ml).
Quantification of Pgp expression by immunohistochemistry
Pgp was only stained in brain capillary endothelial cells but in neither neurons nor astroglia under the conditions of the immunohistochemical protocol used for this study. Figure 6A illustrates representative immunohistochemical Pgp stainings in cerebellar (a, b), thalamic (c, d), and hippocampal (e, f) brain-capillary endothelial cells, illustrating a clear SE-induced increase of Pgp-labeled area. The Pgp-labeled area was significantly higher in the cerebellum (+75.9%), thalamus (+39.7%), CA1 (+51.5%), CA3 (+64.7%), and dentate gyrus (+45.0%) but remained unchanged in the other investigated regions (Fig. 6B). The increased regional expression of Pgp determined by immunohistochemistry perfectly matched the increased efflux rate constant k2 values determined by PET (compare Fig. 5F, Table 2). A correlation analysis between SE-induced increases in Pgp expression and in outcome parameters of kinetic model analysis reached significance only for k2 values after partial Pgp inhibition (r2 = 0.80; p = 0.04) (Fig. 6C).
MR imaging with Gd-DOTA as contrast agent
This was performed in six rats before and 48 h after SE. Figure 7A exemplarily shows T1-weighted MR images of an individual rat in which the choroid plexus and pituitary gland were the only regions with an obviously enhanced signal after Gd-DOTA injection. These observations suggest that there was no major leakage of BBB that may explain the observed group differences in brain activity uptake in the PET experiments after SE. This conclusion is substantiated by the fact that, after complete Pgp inhibition by TQD, VPM uptake into the brain did not differ between controls and post-SE rats (Fig. 4C). However, as shown in Figure 7A, T2-weighted images after SE reveal a striking signal enhancement in hippocampus and piriform cortex as well as clearly reduced CSF space, which was attributable to cell edema, indicating severe pathological changes in the brain after SE.
Assessment of albumin extravasation
This was performed in six controls and nine rats 48 h after SE. In part of the SE rats, the visual inspection of brain slices resulted in differentially pronounced red fluorescence in neurons of the dentate hilus, CA1, and CA3, as well as in thalamic subregions but not in cerebellum, frontal motor cortex, or corpus striatum. Striking interindividual differences were found: one rat showed a high amount of fluorescent neurons and circumscribed extracellular fluorescence in hippocampal and thalamic subregions, in five rats intraneuronal fluorescence was found in the same brain regions but in lower extent (Fig. 7B), and in three SE rats no fluorescence at all was detected. Albumin extravasation did not occur in any of control animals.
Using a novel imaging protocol composed of VPM–PET scans after half-maximal Pgp inhibition by TQD in control rats and rats 48 h after SE, we were able to quantify group differences in regional Pgp function. This is the first study investigating Pgp activity with VPM–PET in rats after a severe brain insult that can lead to development of epilepsy in animal models as well as in humans. We also demonstrate that changes in outcome parameters of kinetic model analysis correlate with disease-induced changes in Pgp expression.
Our major findings are as follows. First, after partial Pgp inhibition, the extent and number of SE-induced group differences within outcome parameters of kinetic model analysis clearly increased. Second, changes in Pgp function are notably mirrored by both changes in the efflux rate constant k2 and Pgp expression levels, which correlate significantly. Third, increased Pgp expression and activity after SE occurs not only in primary epileptogenic regions, such as the hippocampus, but also in the thalamus and the cerebellum.
Seizure-induced overexpression of Pgp
Because accessibility to brain tissue of epilepsy patients is very limited, animal models are mandatory tools for studying mechanisms of pharmacoresistant epilepsy (Löscher and Potschka, 2005a) and essential for the neurobiological evaluation of imaging protocols. By using Pgp immunohistochemistry, we recently demonstrated an increase in Pgp expression at the BBB of rats 48 h after SE in hippocampal subregions (Bankstahl and Löscher, 2008), which we extended to brain regions that are often not the focus in basic epilepsy research. Importantly, we and others found decreased brain levels of the anticonvulsive Pgp substrate phenytoin in parallel with increased Pgp protein or Mdr1 mRNA levels shortly after SE as well as in chronically epileptic animals (Rizzi et al., 2002; van Vliet et al., 2007b; Bankstahl and Löscher, 2008), further supporting the assumption that the detected overexpression of Pgp has functional consequences. It has been hypothesized that Pgp overexpression could also be induced by prolonged SE in humans (Sisodiya and Thom, 2003; Iannetti et al., 2005; Bankstahl and Löscher, 2008). In apparent line with this possibility, coadministration of verapamil acting as competitive Pgp inhibitor reversed resistance to phenobarbital or phenytoin in two patients with long-lasting refractory SE (Iannetti et al., 2005; Schmitt et al., 2007).
Imaging changes in Pgp function
A new study design: VPM–PET after half-maximal inhibition of Pgp
Because Pgp overexpression in the epileptic focus region is considered as one important mechanism of pharmacoresistance in epilepsy patients (Löscher and Potschka, 2005b), a validated method for in vivo detection of regionally increased Pgp activity would be a valuable tool for predicting the individual risk of an epilepsy patient not responding to pharmacological treatment with antiepileptic drugs that are Pgp substrates. Although several radiotracers have been used to image cerebral Pgp function (Zoghbi et al., 2008; Syvänen et al., 2009; Kreisl et al., 2010; la Fougère et al., 2010), radiolabeled verapamil is still the most widely studied one (Pike, 2009). However, because VPM is a high-affinity Pgp substrate, its brain uptake is very low, so that small differences in regional Pgp activity cannot easily be measured with this radiotracer (Kuntner et al., 2010). Because complete transporter inhibition would result in VPM brain uptake driven only by passive diffusion, partial Pgp inhibition before tracer administration could be an encouraging paradigm. Therefore, we used the ED50 of the third-generation Pgp inhibitor TQD (Kuntner et al., 2010).
Regional changes in Pgp function
In line with our hypothesis, no AUC group differences could be detected after complete Pgp inhibition (15 mg/kg TQD), i.e., the efflux transporter was chemically “knocked out,” resulting in equally high tracer uptake in all analyzed brain regions (Fig. 4C). Additionally, before Pgp modulation, we observed low brain uptake of radioactivity, demonstrated by low VT values for both groups of rats (Fig. 5A). The cerebellar decrease in VT and influx rate constant K1 48 h after SE seen in the baseline scans could be interpreted as very high Pgp activity in this brain region. This interpretation is supported by our immunohistological results, in which increase in Pgp expression after SE was highest in the cerebellum (Fig. 6B). Moreover, in scans after half-maximal Pgp inhibition, an even more pronounced cerebellar decrease in VT and K1 and an increase in the efflux rate constant k2 could be observed in 48 h post-SE rats, again pointing to an increased transporter activity in cerebellum (Fig. 5). This was the case despite the fact that the increased TQD levels measured in the post-SE rats would act in the opposite direction.
Reduced influx or increased efflux?
Increased expression of Pgp at the BBB would be expected to increase the efflux of Pgp substrates from the brain (Löscher and Potschka, 2005a). SE-induced increase in Pgp expression was indeed associated with significant changes of kinetic modeling parameters in affected brain regions. K1 and k2 describe exchange of radioactivity at the luminal membrane of the BBB. The second, deeper compartment described by k3 and k4 is seen only after Pgp inhibition (Bankstahl et al., 2008; Wagner et al., 2009), but the distinct physiological correlate remains unclear so far.
Among all calculated PET parameters, k2 values after partial Pgp inhibition in post-SE rats correlated best with the immunohistological data (Fig. 6F). It is still a matter of debate which of the two rate constants for transport of Pgp-substrate tracers across the BBB is most affected by changes in Pgp function.
Two different modes of action for Pgp have been suggested, i.e., influx hindrance (resulting in low K1) and efflux enhancement (resulting in high k2). First, a vacuum cleaner model has been suggested (Higgins and Gottesman, 1992), resulting in influx hindrance. In this model, substrates are transported back from the lipid layers of the luminal cell membrane into the blood before they reach the cytoplasm. Second, substrates could be transported from the cytoplasm into the blood. It has also been suggested that Pgp may act by both mechanisms simultaneously (Stein et al., 1994; Sharom, 1997; Higgins and Linton, 2004). Before Pgp inhibition, brain uptake of VPM is very low, which consequently results in low influx and efflux rate constants (K1, k2). Probably, most VPM does not reach the cytoplasm and is directly transported back into the blood. However, after partial Pgp inhibition, a considerable amount of VPM enters the first compartment of the brain and is only then sufficiently available for efflux out of the brain by the second mode of action. This could finally lead to an increase in efflux rate k2, as observed in this study, despite partial inhibition of Pgp.
In this respect, it is interesting to note that several previous studies in healthy volunteers or rats found that pharmacological inhibition of Pgp caused an increase in the influx rate constant K1 but no change in the k2 parameter (Liow et al., 2009; Muzi et al., 2009; Wagner et al., 2009; Kreisl et al., 2010; Kuntner et al., 2010). However, Bauer et al. (2009) showed that age-dependent increase in VPM VT in healthy subjects was mirrored by a decrease in k2 but not by an increase in K1. In patients suffering from drug-resistant unilateral temporal lobe epilepsy, Langer et al. (2007) found increases in VPM k2 ipsilateral to the epileptic focus. Hence, for translation of our paradigm to epilepsy patients, both influx and efflux rate constants should be taken into account.
Possible limitations: CBF and integrity of the BBB
On the one hand, the unexpected absence of decreased VT values in hippocampus could to some extent be attributable to spillover of radioactivity from adjacent structures, such as choroid plexus, which showed high radioactivity uptake as it was also reported for human subjects (Langer et al., 2007). On the other hand, it cannot be ruled out that part of the differences between control and post-SE rats is related to regional changes in CBF after SE. However, given the magnitude of the differences and their correlation with regional Pgp expression levels, it seems unlikely that these changes were only related to CBF. Because regions with higher blood flow would have greater radioactivity uptake, this could at least partially explain why increased hippocampal Pgp levels after SE did not result in decreased VT of VPM after partial Pgp inhibition in our study. Notably, a recent study by Choy et al. (2010) revealed clear increases of CBF in hippocampus 2 d after pilocarpine-induced SE. Unfortunately, only one additional brain region corresponding to our ROIs, i.e., thalamus, was investigated in this study, without finding SE-mediated changes in CBF (Choy et al., 2010). Additional studies are needed to reveal possible flow dependency of VPM uptake in additional brain regions.
To clarify whether changes in activity uptake in post-SE rats may be dependent on transient opening of the BBB, rather than on differences in Pgp activity, we performed T1-weighted MRI scans after injection of Gd-DOTA, which is a paramagnetic MRI contrast agent used routinely for evaluation of BBB lesions in human patients. In our study, T1-weighted MR scans after Gd-DOTA injection did not reveal major BBB breakdown in any ROI as exemplarily shown in Figure 7A. This is in line with previous MR imaging studies in which no or only circumscribed breakdown of BBB 2–3 h after SE in rodents has been described (Bouilleret et al., 2000; Roch et al., 2002; Hsu et al., 2007; Immonen et al., 2008), which disappeared 6–24 h after SE (Roch et al., 2002; Immonen et al., 2008). Furthermore, low VT values in VPM baseline scans without marked regional differentiation argue against unselective BBB opening (Fig. 5A).
A recent study detected increased permeability of the rat BBB after SE and spontaneous recurrent seizures, measured by uptake of fluorescence-labeled albumin into the brain (van Vliet et al., 2007a). In line with the results of this study, we found limited uptake of EB, a marker of albumin uptake, in two of the five PET ROIs, i.e., hippocampus and thalamus, and only in 66% of post-SE rats. Our finding that, after chemical knock-out of Pgp by TQD, VPM uptake is the same in controls and SE rats (Fig. 4C) also argues against any severe impairment of the BBB 48 h after SE.
In summary, we established a VPM small-animal PET protocol that is suitable to quantify changes in Pgp activity in distinct brain regions after a brain insult that often leads to epilepsy in humans. Two recent pilot studies in healthy volunteers indicate that the PET paradigm described in this paper could be translated to human subjects (Wagner et al., 2009; Bauer et al., 2010). Our protocol is currently used to analyze regional Pgp activity in pharmacoresistant rats and will be expanded to epilepsy patients. Hopefully, it may ultimately serve as a predictive tool for detection of Pgp-mediated pharmacoresistance in epilepsy.
This work was supported by funding from the European Community Seventh Framework Programme (FP7/2007-2013) under Grant Agreement 201380 (“Euripides”) and from the Austrian Science Fund project “Transmembrane Transporters in Health and Disease” (SFB F35). We are grateful to Dr. Joan Abbott, Dr. Alexander Hammers, and Dr. Matthias Koepp for critical reading of this manuscript and helpful suggestions. We thank Gloria Stundner (Austrian Institute of Technology), Dr. Thomas Fillip, and Maria Zsebedics (Seibersdorf Laboratories) for their skilful assistance with animal handling and Severin Mairinger, Bernd Dörner, and Florian Bauer as well as the staff of the radiochemistry laboratory (Seibersdorf Laboratories) for their continuous support. Thomas Flanitzer is gratefully acknowledged for help with data analysis and Dr. Rudolf Karch for advice on compartmental modeling issues.
The authors declare no competing financial interests.
- Correspondence should be addressed to Dr. Jens P. Bankstahl, Department of Pharmacology, Toxicology, and Pharmacy, University of Veterinary Medicine Hannover, Buenteweg 17, 30559 Hannover, Germany.