FULL-LENGTH ORIGINAL RESEARCH

Decreased levels of active uPA and KLK8 assessed by [111In]MICA-401 binding correlate with the seizure burden in an animal model of temporal lobe epilepsy *Stephan Missault, *†Lore Peeters, *Halima Amhaoul, ‡§David Thomae, *Annemie Van Eetveldt, *Barbara Favier, *Anagha Thakur, §Jeroen Van Soom, ¶Asla Pitkanen,€ §Koen Augustyns, §Jurgen Joossens, ‡Steven Staelens, and *1Stefanie Dedeurwaerdere

Epilepsia, 58(9):1615–1625, 2017 doi: 10.1111/epi.13845

SUMMARY Objective: -type (uPA) and -related pepti- dase 8 (KLK8) are serine proteases that contribute to extracellular matrix (ECM) remodeling after brain injury. They can be labelled with the novel radiotracer [111In] MICA-401. As the first step in exploring the applicability of [111In]MICA-401 in tracing the mechanisms of postinjury ECM reorganization in vivo, we performed in vitro and ex vivo studies, assessing [111In]MICA-401 distribution in the brain in two animal mod- els: kainic acid–induced status epilepticus (KASE) and controlled cortical impact (CCI)–induced traumatic brain injury (TBI). Methods: In the KASE model, in vitro autoradiography with [111In]MICA-401 was per- formed at 7 days and 12 weeks post-SE. To assess seizure burden, rats were moni- tored using video-electroencephalography (EEG) for 1 month before the 12-week time point. In the CCI model, in vitro autoradiography was performed at 4 days and ex vivo autoradiography at 7 days post-TBI. Results: At 7 days post-SE, in vitro autoradiography revealed significantly decreased [111In]MICA-401 binding in hippocampal CA3 subfield and extrahippocampal tempo- ral lobe (ETL). In the chronic phase, when animals had developed spontaneous sei- zures, specific binding was decreased in CA3 and CA1/CA2 subfields of hippocampus, dentate gyrus, ETL, and parietal cortex. Of interest, KASE rats with the highest fre- quency of seizures had the lowest hippocampal [111In]MICA-401 binding (r = 0.76, Stephan Missault is a p ≤ 0.05). Similarly, at 4 days post-TBI, in vitro [111In]MICA-401 binding was signifi-À doctoral student at the cantly decreased in medial and lateral perilesional cortex and ipsilateral dentate gyrus. University of Antwerp. Ex vivo autoradiography at 7 days post-TBI, however, revealed increased tracer uptake in perilesional cortex and hippocampus, which was likely related to tracer leak- age due to blood–brain barrier (BBB) disruption. Significance: Strong association of reduced [111In]MICA-401 binding with seizure bur- den in the KASE model suggests that analysis of reduced levels of active uPA/KLK8 represents a novel biomarker candidate to be explored as a biomarker for epilepsy severity. However, limited BBB permeability of [111In]MICA-401 currently limits its application in vivo.

Accepted June 15, 2017; Early View publication July 19, 2017. *Experimental Laboratory of Translational Neuroscience and Otolaryngology, Faculty of Medicine and Health Sciences, University of Antwerp, Wilrijk, Belgium; †Bio-Imaging Lab, Faculty of Pharmaceutical, Biomedical and Veterinary Sciences, University of Antwerp, Wilrijk, Belgium; ‡Molecular Imaging Center Antwerp, Faculty of Medicine and Health Sciences, University of Antwerp, Wilrijk, Belgium; §Laboratory of Medicinal Chemistry, Faculty of Pharmaceutical, Biomedical and Veterinary Sciences, University of Antwerp, Wilrijk, Belgium; and ¶Department of Neurobiology, A.I. Virtanen Institute for Molecular Sciences, University of Eastern Finland, Kuopio, Finland 1Current address: UCB Biopharma S.P.R.L., Braine-l’Alleud, Belgium. Address correspondence to Stephan Missault, Experimental Laboratory of Translational Neuroscience and Otolaryngology, Faculty of Medicine and Health Sciences, University of Antwerp, Universiteitsplein 1, 2610 Wilrijk, Belgium. E-mail: [email protected] Wiley Periodicals, Inc. © 2017 International League Against Epilepsy

1615 1616

S. Missault et al.

KEY WORDS: Extracellular matrix, Urokinase-type plasminogen activator, Kallikrein-related peptidase 8, Epilepsy, Traumatic brain injury, SPECT.

kallikrein-related peptidase 1 [KLK1]; hydroxynonenal, Key Points [HNE]; and acetylcholinesterase [AchE]). Recently, [111In] The radiotracer [111In]MICA-401 has a high affinity MICA-401 was evaluated for its capacity to image uPA • 6 for active urokinase-type plasminogen activator (uPA) activity in vivo by SPECT in two different cancer models. and kallikrein-related peptidases 4 and 8 (KLK4/8) Of interest, the best SPECT images were acquired at 4 days • Postmortem [111In]MICA-401 binding reveals postinjection due to a slow blood clearance of the tracer. decreased levels of active uPA/KLK8 in brain follow- Because the tracer can bind only to active enzyme, that is, ing status epilepticus and traumatic brain injury enzyme in which the catalytic site serine is available for Postmortem [111In]MICA-401 binding correlates with binding, tracer binding reflects levels of active uPA and • 6 seizure burden in a model of temporal lobe epilepsy KLK8. Little is known about the role of KLK4 in the brain; • In vivo application of [111In]MICA-401 is currently therefore we limit the interpretation of the results to uPA limited due to its incapacity to cross the intact blood– and KLK8. brain barrier The present study was designed to evaluate the capacity of [111In]MICA-401 to image the spatiotemporal dynamics of active uPA and KLK8 in the brain in the following two Epilepsy affects approximately 69 million people world- rodent models of acquired epilepsy: the kainic acid–induced wide.1 Despite >20 antiepileptic drugs on the market, 30% status epilepticus (KASE) model of mesial temporal lobe of patients remain in unsatisfactory seizure control.2 There- epilepsy (TLE) and the controlled cortical impact (CCI) fore, identification of epileptogenic mechanisms remains a model of posttraumatic epilepsy. Our in vitro data show a 111 major unmet need in neurology. reduction in [ In]MICA-401 binding in several injured Studies over the past 10 years have revealed involvement brain areas during the first postinjury week in both models, of extracellular proteases in the pathogenesis of epilepsy (re- which is even more pronounced and widespread in the viewed in Lukasiuk et al.3 and Pitkanen et al.4). These pro- chronic phase in the KASE model, when spontaneous sei- teases degrade various components of the extracellular zures have developed. Moreover, the reduced hippocampal 111 matrix, facilitating plasticity in both physiologic and patho- binding of [ In]MICA-401 correlated with the seizure bur- logic conditions. One family of extracellular proteases are den. serine proteases, which include urokinase-type plasminogen activator (uPA) and kallikrein-related peptidase 8 (KLK8), Methods also known as neuropsin or brain 1 (BSP1).5 Because these enzymes are located extracellularly, assess- Study design ment of their activity in vivo would provide an attractive tool Study design is shown in Figure 1. for monitoring the progression of extracellular plasticity. Our research group has recently reported the synthesis of a Animals novel activity-based single-photon emission computed Fifty-four male Sprague-Dawley rats were purchased tomography (SPECT) probe, [111In]MICA-401.6 The probe from Harlan Laboratories, the Netherlands. Upon arrival, was designed based on a diphenyl phosphonate scaffold, animals were single housed in a temperature- and humidity- which binds covalently to the alcohol group of the active site controlled environment on a 12 h light–dark cycle with serine of serine proteases. It also bears a benzyl guanidine standard food and water available ad libitum. Animals were moiety, which provides a high selectivity toward uPA,7 treated in accordance with the guidelines approved by the KLK8, and KLK4, and a polyethylene glycol (PEG) linker to European Ethics Committee (decree 2010/63/EU) and the connect the 1,4,7,10-tetraazacyclododecane-1,4,7,10-tetraa- Animal Welfare Act (7 USC 2131). Animal experiments cetic acid (DOTA)-chelator, which is used as a complexing were approved by the ethical committee of the University of agent for radioisotope indium-111 (111In). In vitro evaluation Antwerp, Belgium (ECD 2012-62). of the binding kinetics of [111In]MICA-401 revealed that the probe was highly selective toward uPA, KLK8, and KLK4 Kainic acid–induced status epilepticus (KASE) as compared to other related -like serine proteases Animals were about 8-weeks-old at the time of status (tissue-type plasminogen activator [tPA]; ; plas- epilepticus (SE) induction (mean SEM weight: Æ min; Factor Xa; , plasma [KLK]; 250.9 5.2 g). The KASE model was induced as Æ

Epilepsia, 58(9):1615–1625, 2017 doi: 10.1111/epi.13845 1617

uPA/KLK8 Correlate with Seizure Burden

Figure 1. Study design. In vitro autoradiography (ARG) with [111In]MICA-401 was performed in rats with kainic acid–induced status epilepticus (KASE) and in rats with controlled cortical impact (CCI)–induced traumatic brain injury, and their respective controls. KASE and healthy control rats were killed at 7 days (group 1) or 12 weeks post-SE (group 2). Neuronal loss was assessed with immunohistochemistry at both time points in KASE and healthy control rats. In the 12-week KASE group, electrodes were implanted at 6 weeks post-SE and 1- month continuous video-EEG monitoring was performed between the 8th and 12th week post-SE to assess the seizure burden. In CCI- and sham-injured rats, in vitro autoradiography was done at 4 days postinjury (group 3). Another cohort of CCI rats was injected with [111In]MICA-401 at 3 days postinjury and killed 4 days later for ex vivo autoradiography (group 4). To assess the contribution of blood– brain barrier (BBB) disruption to in vivo tracer uptake, we performed an in vivo BBB permeability assay at 3 days post-CCI injury (i.e., at the time of tracer injection; group 5). Epilepsia ILAE previously described8 with a few modifications. Briefly, but were not exposed to impact. Rats were sacrificed at animals received multiple low-dose subcutaneous injections 3 days (three CCI, three shams), 4 days (five CCI, five of kainic acid (A.G. Scientific, U.S.A.) until they reached shams), or 7 days postinjury (three CCI, three shams). SE (average total dose: mean SEM 16.2 5.4 mg/kg). Animals were killed at 7 daysÆ post-SE whileÆ undergoing Video-EEG epileptogenesis (11 KASE, 6 healthy controls) or at 12 weeks post-SE when they had already developed sponta- Electrode implantation neous recurrent seizures (SRS; 9 KASE, 6 healthy controls). Implantation of EEG electrodes was performed at Occurrence of SRS was confirmed with continuous video– 6 weeks post-SE in KASE rats as described previously.8 electroencephalography (EEG) monitoring (see below). Briefly, animals were anesthetized, and six epidural elec- trodes were implanted bilaterally: four recording electrodes Controlled Cortical Impact (CCI)–induced traumatic (two frontal, two parietal), and one reference and one brain injury (TBI) ground electrode (both occipital). The electrodes were fixed Rats about 8-weeks-old (mean SEM weight: into a plastic plug (Bilaney Consultants, Plastics One, Uni- 225.5 2.5 g) were anesthetized usingÆ isoflurane in oxy- ted Kingdom) and secured to the skull using dental cement gen (induction:Æ 5%, maintenance: 2.5%; Abbott, Belgium). (Simplex Rapid, Kemdent, United Kingdom; Durelon, 3M A craniotomy of 5 mm diameter was performed with a ESPE, U.S.A.). handheld trephine over the left parietal cortex (midway between bregma and lambda, bordering the lateral ridge). Recording Great care was taken to avoid damage of the underlying KASE rats were subjected to continuous video-EEG dura. CCI injury was performed with the Leica Impact One monitoring for 29 1 day between the second and third apparatus (Leica Biosystems, U.S.A.) using the following months post-SE. OneÆ animal was monitored only for 5 days parameters: flat tip of 3 mm diameter, impact angle 18 due to loss of the electrode assembly, and was excluded degrees, impact velocity 4 m/s, depth of penetration from the seizure analysis. Animals were connected to a digi- 2.5 mm, and dwell-time 500 msec. Following impact, the tal EEG acquisition system (Ponemah P3 Plus, Data craniotomy was sealed with a piece of plastic and skin Sciences International, U.S.A.) through a cable system as sutured. Sham-operated animals received the same surgery, described previously.8

Epilepsia, 58(9):1615–1625, 2017 doi: 10.1111/epi.13845 1618

S. Missault et al.

Analysis Tris–HCl buffer for 1 h at room temperature to assess total Video-EEG data were analyzed manually using NeuroS- receptor-ligand binding. To assess nonspecific binding of core 3.0 (Data Sciences International, U.S.A.). SRS were the radiotracer, consecutive brain sections were incubated identified as described previously.8,9 Briefly, an electro- under the same conditions with an excess amount (1 mM) graphic seizure was defined as an aberrant EEG signal with of UAMC-01162, the compound from which the radiolabel- an amplitude of at least three times the baseline signal and a ing precursor is derived, in addition to 50 nM [111In]MICA- duration of at least 5 s.10 The severity of the seizures was 401. Sections were washed with ice-cold Tris–HCl buffer determined by analyzing the corresponding video recording and distilled water to remove unbound tracer. After drying, and scored according to a modified scale of Racine.9 To sections were exposed on hyperfilm (Amersham Hyperfilm assess the seizure burden for each animal, we calculated the MP; GE Healthcare, Belgium) for 3 days, alongside a 14C total number of SRS per day. standard (American Radiolabeled Chemicals Inc., U.S.A.). The hyperfilms were developed manually and digitalized Radiochemistry using an Epson V700 photo scanner (Seiko Epson Corpora- Radiosynthesis of [111In]MICA-401 was performed as tion, Japan). recently described.6 The radiolabeling precursor, an uPA/ In the KASE model (20 KASE, 12 healthy controls), KLK8 inhibitor linked to a DOTA-chelator, was incubated specific binding of [111In]MICA-401 was assessed in CA1/ 111 with [ In]-InCl3 (PerkinElmer, Belgium) for 45 min at CA2 and CA3 subregions of hippocampus, dentate gyrus 60°C to achieve complexation of indium (111In). [111In] (DG), extrahippocampal temporal lobe (ETL), and parietal MICA-401 was synthesized with a 59% yield and a chemi- cortex. In the CCI model (five CCI, five shams), we ana- cal and radiochemical purity of >98%. The specific activity lyzed perilesional cortex (both medial and lateral to the was 15.44 GBq/lmol. lesion), contralateral cortex (regions symmetric to the med- ial and lateral perilesional cortex), and ipsilateral and con- Inhibitory potential of the radiolabeling precursor for tralateral CA1, CA3, and DG. All regions of interest (ROIs) serine proteases in rat were outlined manually in triplicate samples (sections with The inhibitory potential of the radiolabeling precursor for total binding and nonspecific binding separately). human uPA, KLK8, and a set of related trypsin-like serine Gray values were measured using ImageJ software proteases as well as mouse uPA has recently been (National Institutes of Health, U.S.A.). The mean gray val- described.6 Because there was a large difference in half ues were converted to doses of radioactivity (kBq/g 14C) by maximal inhibitory concentration (IC50) values for human interpolation from a standard curve that was derived from 14 111 and mouse uPA, we decided to assess the IC50 values for the C standard gray values. The specific binding of [ In] human and rat uPA and KLK8 of UAMC-01162, the com- MICA-401 was obtained by subtracting the nonspecific pound from which the radiolabeling precursor was derived, binding from the total binding. The average specific binding according to previously described protocols.7 Briefly, for the hippocampus proper was calculated by taking the recombinant rat uPA or KLK8 was incubated with an mean of the specific binding values for CA1/2 and CA3. enzyme-specific chromogenic substrate and different con- centrations of UAMC-01162. Inhibition of the enzymatic Ex vivo autoradiography reaction was evaluated through spectrophotometry. Three CCI-injured and three sham-operated rats were injected intravenously with [111In]MICA-401 (mean SEM 39.1 2.2 MBq) at 3 days postinjury. At 4 daysÆ Autoradiography postinjection,Æ animals were killed by decapitation and Tissue collection and preparation brains were resected, snap-frozen, and sectioned as Animals were killed by decapitation. Brains were imme- described earlier. This time point was chosen based on our diately resected and snap-frozen in 2-methylbutane on dry previous observations showing very slow blood clearance ice ( 35°C, 3 min). Brains were stored at 80°C. In the of the radiotracer.6 At 4 days postinjection, most of the sig- KASEÀ model, only the right hemisphere wasÀ snap-frozen nal in the blood has been cleared and tracer binding in vari- for the present analysis (kainate-induced lesion is symmet- ous organs can be more readily distinguished.6 Sections ric). Serial coronal cryosections (20 lm) were collected in were dried at room temperature and exposed on hyperfilm triplicate at two coronal levels: at 2.92 mm from bregma for 2 weeks along with a 14C standard. The film was devel- (sections containing lesion core andÀ dorsal hippocampus: oped manually and analyzed in the same way as the in vitro CCI and shams) and 4.80 mm from bregma (sections con- autoradiography films. ROIs included perilesional cortex taining ventral hippocampus:À KASE and healthy controls). and ipsilateral hippocampus.

In vitro autoradiography Immunohistochemistry Coronal brain sections were dried at room temperature To provide a potential explanation for the observed and incubated with 50 nM [111In]MICA-401 in 50 mM decrease in levels of active uPA/KLK8, a neuronal nuclei

Epilepsia, 58(9):1615–1625, 2017 doi: 10.1111/epi.13845 1619

uPA/KLK8 Correlate with Seizure Burden

(NeuN) staining was performed to evaluate the extent of matched-pairs signed-rank test was used to investigate pos- neuronal loss in the KASE model. Coronal brain sections sible differences between ipsilateral and contralateral hemi- adjacent to the brain sections used for in vitro autoradiogra- spheres of CCI-injured rats. Spearman’s rank correlation phy were acclimatized to room temperature, fixated with was used to investigate the relationship between specific 4% paraformaldehyde, and rinsed with phosphate-buffered binding of [111In]MICA-401 and (1) neuronal loss in KASE saline (PBS). Sections were treated with 3% H2O2 and 5% rats and (2) number of SRS per day in chronic KASE rats. normal donkey serum (NDS) containing 0.5% Triton X- Statistical significance was set at p ≤ 0.05. 100. Sections were incubated overnight with mouse anti-rat NeuN antibody (1:2,000; clone A60; Merck Millipore, Bel- Results gium) in 1% NDS. The following morning, sections were rinsed and incubated for 1 h with horse radish peroxidase– High affinity of UAMC-01162 for rat uPA and KLK8 conjugated donkey anti-mouse immunoglobulin (IgG) anti- UAMC-01162, the compound from which the radiolabel- body (1:500; Jackson ImmunoResearch Laboratories, Inc., ing precursor is derived, has a high affinity for both rat uPA United Kingdom) in 1% NDS. Sections were rinsed again and KLK8. The IC values were 21 1 nM for rat uPA 50 Æ and incubated with 3,30-diaminobenzidine for 10 min. The and 15 2 nM for rat KLK8 (Table 1). Æ reaction was stopped with dH2O and sections were gradu- ally dehydrated and coverslipped. Decreased specific binding of [111In]MICA-401 in the Images were obtained with a NanoZoomer-XR slide KASE model scanner (Hamamatsu, Japan) equipped with a 209 objective In vitro [111In]MICA-401 autoradiography of normal rat and analyzed with ImageJ software. Images were trans- brain revealed a prominent binding in several brain regions, formed to 8-bit images, and a threshold was set to select including cerebral cortex, hippocampus, amygdala, and specifically stained neuronal nuclei from background stain- caudate putamen, whereas binding in the thalamus was low. ing (Fig. S1). The area fraction, that is, the % area of the At 7 days post-SE, while KASE rats were undergoing ROI that consists of NeuN-stained nuclei (as set by the epileptogenesis, specific binding of [111In]MICA-401 was threshold), was calculated for the following ROIs: CA1/ reduced in extrahippocampal temporal lobe (ETL, defined CA2 and CA3 subregions of hippocampus, granule cell as amygdala, piriform cortex, and entorhinal cortex; layer of the dentate gyrus (DG), extrahippocampal temporal p ≤ 0.01) and CA3 subregion of hippocampus (p ≤ 0.05) as lobe, and parietal cortex. A higher area fraction reflects a compared to healthy controls (Fig. 2A,C). At 12 weeks higher neuronal density in this ROI. Furthermore, the area post-SE, when all KASE rats had SRS, [111In]MICA-401– (in mm2) of each ROI was measured in ImageJ. All ROIs specific binding was reduced in the ETL and parietal cortex were outlined manually in triplicate samples. (p ≤ 0.001), CA1/CA2 subregion of hippocampus, and den- tate gyrus (p ≤ 0.05). We also found a trend toward Assessment of blood–brain barrier (BBB) permeability decreased binding in CA3 (p ≤ 0.1; Fig. 2B,D). Moreover, One group of CCI-injured (n = 3) and sham-operated rats (n = 3) was injected intravenously with 2% Evans blue (in 0.9% saline, 4 ml/kg) at 3 days postinjury. Evans blue was allowed to circulate for 1 h, after which the ani- Table 1. Inhibitory potential of UAMC-01162 and the mal was perfused transcardially with 100 ml of ice-cold radiolabeling precursor MICA-401 for several serine PBS when under deep anesthesia (60 mg/kg Nembutal, proteases

Ceva Sante Animale, Belgium; intraperitoneally) to IC50 (lM) remove the blood from the blood vessels. Following per- UAMC-01162 (1) MICA-401 (2) fusion, brains were resected and snap-frozen as described Human uPA 0.007 0.022a earlier. During sectioning, photographs were taken for Rat uPA 0.021 visual inspection of Evans blue extravasation. Evans blue Human tPA 38.05 >10a has a very high affinity for serum albumin, which cannot Human thrombin 30 >10a a cross the intact BBB. In case of increased BBB perme- Human 27 >10 Human Factor Xa 35.2 >10a ability, Evans blue (bound to albumin) can extravasate Human plasma KLK 10.16 >10a into the brain parenchyma. Human KLK1 >10 >10a Human KLK4 0.012 0.036a Statistics Rat KLK4 0.044 0.082 a All data were analyzed using nonparametric tests in Human KLK8 0.050 0.033 Rat KLK8 0.015 0.034 GraphPad Prism 5. The Mann-Whitney U test was used to Human HNE >10 >2.5a compare specific binding and tracer uptake between KASE Human AChE >20 >20a and controls or CCI and sham rats, or between KASE and aVangestel et al.6 controls of two different time points. The Wilcoxon

Epilepsia, 58(9):1615–1625, 2017 doi: 10.1111/epi.13845 1620

S. Missault et al.

ABB

C D * * * *

Control 7 d KASE 7 d* Control 12 w KASE 12 w

EF

GH

Control 7 d KASE 7 d Control 12 w KASE 12 w I J

Figure 2. Decreased in vitro binding of [111In]MICA-401 and neuronal cell loss in the kainic acid–induced status epilepticus (KASE) model. (A) At 7 days post-SE, when rats were undergoing epileptogenesis, specific binding of [111In]MICA-401 was decreased in the extrahippocampal temporal lobe (ETL) and hippocampal CA3 subfield of KASE rats (n = 11) compared to healthy controls (n = 6). (B) At 12 weeks post- SE, when KASE rats had spontaneous recurrent seizures, specific binding of [111In]MICA-401 was decreased in most brain areas, including ETL, hippocampal CA1/CA2 subfield, dentate gyrus, and parietal cortex. (C, D) Representative in vitro autoradiographs of total [111In] MICA-401 binding in healthy control and KASE rats at 7 days (C) and 12 weeks post-SE (D). The regions of interest (ROIs) are drawn in yellow. Dark red asterisk, parietal cortex; green asterisk, CA1/CA2; orange asterisk, CA3; purple asterisk, dentate gyrus; blue asterisk, ETL. (E) Neuronal cell loss was observed in hippocampal CA1/CA2 and CA3 subfields, parietal cortex, and ETL of KASE rats at 7 days post-SE. (F) Neuronal loss was observed in CA3 of KASE rats at 12 weeks post-SE. (G, H) Representative images of NeuN staining in KASE and healthy control rats at 7 days (G) and 12 weeks post-SE (H). The ROIs are drawn in yellow. (I) Specific binding of [111In] MICA-401 was significantly lower in KASE rats at 12 weeks post-SE compared to 7 days post-SE in parietal cortex and CA1/2. (J) The area fraction of NeuN staining was significantly higher in KASE rats at 12 weeks post-SE compared to 7 days post-SE in parietal cortex. Statistical significance: *p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001 (compared to age-matched control group). Data are presented as boxplots. Epilepsia ILAE

Epilepsia, 58(9):1615–1625, 2017 doi: 10.1111/epi.13845 1621

uPA/KLK8 Correlate with Seizure Burden at 12 weeks post-SE, specific binding of [111In]MICA-401 401 in hippocampus proper and a daily seizure frequency in in CA1/CA2 and parietal cortex was lower than at 7 days KASE rats at 12 weeks post-SE (Fig. 3). post-SE (p ≤ 0.01; Fig. 2I). There were no significant dif- ferences between healthy controls of 7 days and 12 weeks. Decreased specific binding of [111In]MICA-401 in the CCI-injury model Specific binding of [111In]MICA-401 is not correlated In vitro autoradiographic evaluation of [111In]MICA-401 with neuronal loss binding in CCI-injured rats revealed reduced specific bind- We investigated whether the decrease in [111In]MICA- ing in medial and lateral perilesional cortex and ipsilateral 401 binding was related to neuronal loss in the KASE dentate gyrus as compared to the corresponding regions in model. KASE rats exhibited a significantly lower area frac- sham-operated rats at 4 days postinjury (p ≤ 0.01; Fig. 4A, tion of NeuN staining in CA1/2 and CA3 subfields of hip- B). Specific binding of [111In]MICA-401 was significantly pocampus, parietal cortex (p ≤ 0.01), and ETL (p ≤ 0.05) decreased in both medial and lateral perilesional cortex of at 7 days post-SE compared to healthy controls, indicating CCI-injured rats compared to the corresponding contralat- extensive neuronal cell loss (Fig. 2E,G). At 12 weeks post- eral cortical regions (p ≤ 0.01). Moreover, a trend toward a SE, KASE rats had a significantly lower area fraction of decreased binding was observed in the ipsilateral CA1 sub- NeuN staining in the CA3 subfield (p ≤ 0.05) as well as a field and dentate gyrus of CCI-injured rats as compared to trend for a lower area fraction in the granule cell layer of the the contralateral CA1 subfield and dentate gyrus, respec- dentate gyrus (p ≤ 0.1; Fig. 2F,H). At the same time point, tively (p ≤ 0.1). a trend for an increased area was observed for the granule cell layer of the dentate gyrus (p ≤ 0.1; data not shown). At In vivo uptake of [111In]MICA-401 after CCI-injury 7 days post-SE, KASE rats had a significantly lower area Ex vivo autoradiography of [111In]MICA-401 showed an fraction of NeuN staining in parietal cortex compared to increased uptake of the radiotracer in perilesional cortex 12 weeks post-SE (p ≤ 0.01), as well as a trend for a lower and hippocampus in CCI rats as compared to shams at area fraction of NeuN staining in CA1/2 (p ≤ 0.1; Fig. 2J). 7 days postinjury (Fig. 4C,D). In addition, tracer accumula- There were no significant differences between healthy con- tion was observed in the ventricles of both sham and CCI trols of 7 days and 12 weeks. There was no correlation rats. Similarly, extravasation of Evans blue was observed in between neuronal cell loss and [111In]MICA-401 binding in the perilesional area in another cohort of CCI rats at 3 days brain regions in any of the studies at any time point (data not postinjury (i.e., the time point of tracer injection; Fig. 4E). shown). Evans blue was also perceived in the ventricles of both sham and CCI rats. Specific binding of [111In]MICA-401 correlated with epilepsy phenotype in KASE rats at 12 weeks post-SE Discussion All KASE rats experienced SRS in the chronic stage. They had on average mean SEM 10.8 2.9 SRS per The objective of this study was to investigate the capacity day (range: 0.6–19.9 SRS perÆ day). A negativeÆ correlation of the novel activity-based SPECT probe [111In]MICA-401 was found between the specific binding of [111In]MICA- to image brain alterations in active uPA and KLK8 in two rodent models of acquired epilepsy. In this study, we observed a (sub)acute decrease of [111In]MICA-401 binding in the temporal lobe in a temporal lobe epilepsy (TLE) model and in the perilesional cortex and dentate gyrus in a traumatic brain injury (TBI) model. This decrease in [111In] MICA-401 binding was even more pronounced and wide- spread when animals had developed SRS in the TLE model. Moreover, the hippocampal decrease in [111In]MICA-401 binding was correlated with the severity of the epilepsy phe- notype. Because [111In]MICA-401 can bind only to active uPA/ KLK8, [111In]MICA-401 binding is an indirect measure of uPA/KLK8 activity. Figure 3. 111 In vitro [ In]MICA-401 binding is correlated with the seizure bur- 111 den in chronic kainic acid–induced status epilepticus (KASE) rats. Decreased [ In]MICA-401 binding after epileptogenic The KASE rats with the highest daily number of seizures had the insults largest decrease in [111In]MICA-401 binding in the hippocampus In the in vitro autoradiography assays we observed a 111 proper. Statistical significance: *p ≤ 0.05. decrease in [ In]MICA-401 binding and thus levels of Epilepsia ILAE active uPA and KLK8 in several relevant brain regions after

Epilepsia, 58(9):1615–1625, 2017 doi: 10.1111/epi.13845 1622

S. Missault et al.

A B * * * * * * *

sham CCI

CD * *

sham CCI

E

sham CCI Figure 4. Decreased in vitro binding of [111In]MICA-401 and increased in vivo uptake of [111In]MICA-401 and Evans blue extravasation in con- trolled cortical impact (CCI)–injured rats. (A) Specific binding of [111In]MICA-401 was decreased in medial and lateral perilesional cortex and ipsilaterally in the dentate gyrus of CCI rats (n = 5) compared to shams (n = 5) at 4 days postinjury. (B) Representative in vitro autoradiographs of total [111In]MICA-401 binding in sham-operated and CCI-injured rats at 4 days postinjury. The regions of interest (ROIs) are drawn in yellow. Blue asterisk, medial perilesional cortex; dark red asterisk, lateral perilesional cortex; light red asterisk, med- ial contralateral cortex; turquoise asterisk, lateral contralateral cortex; green asterisk, CA1; orange asterisk, CA3; dark brown asterisk, dentate gyrus. (C) Increased radiotracer uptake was observed with ex vivo autoradiography in perilesional cortex and ipsilateral hip- pocampus of CCI-injured but not sham-operated rats injected with radiotracer 4 days earlier. (D) Representative ex vivo autoradio- graphs of [111In]MICA-401 accumulation in sham-operated and CCI-injured rats. Tracer uptake can be observed in perilesional cortex and ipsilateral hippocampus of CCI-injured rats (arrow), but not in shams. ROIs are drawn in yellow. Dark red asterisk, perilesional cor- tex; blue asterisk, ipsilateral hippocampus. (E) Extravasation of Evans blue is visible in the perilesional area of CCI rats (arrow), but not in shams. Statistical significance: **p ≤ 0.01. Data are presented as boxplots. Epilepsia ILAE

SE and CCI-injury. Although several studies have reported ), which was increased during the first 72 h post-SE an increase in uPA and KLK8 expression and activity fol- but had already declined at 100 h post-SE.14 Our study lowing epileptogenic insults, most of these used electrical focused on the expression of active proteases at later time stimulation of the amygdala or hippocampus to trigger points post-SE. Based on these studies, we hypothesized epileptogenesis.11–16 On the other hand, most studies that that the decreased [111In]MICA-401 binding might be used KASE to evoke epileptogenesis showed primarily that explained by neuronal loss. Consequently, we assessed the the basal neuronal expression of these proteases was lost extent of neuronal loss in the KASE model. Indeed, neu- due to neuronal degeneration in some brain regions, in addi- ronal loss was observed in several brain regions that exhib- tion to an overexpression of uPA and KLK8 by, respec- ited a decreased [111In]MICA-401 binding. However, tively, astrocytes and oligodendrocytes.17–19 Masos and although the decrease in levels of active uPA/KLK8 clearly Miskin did not report a decreased neuronal expression of progressed in KASE rats, with only a few regions affected at uPA following KASE, but they used a rather low dose of 7 days post-SE and many more regions at the chronic time KA (i.e., 12 mg/kg, in mice), resulting in relatively mild sei- point, no such spatiotemporal profile was observed for the zures during SE and possibly less pronounced neurodegen- neuronal loss in these animals. Extensive neuronal loss was eration compared to the other studies.14 Moreover, they already overt in all investigated brain regions except the focused mainly on the expression of uPA mRNA (not granule cell layer of the dentate gyrus at 7 days post-SE and

Epilepsia, 58(9):1615–1625, 2017 doi: 10.1111/epi.13845 1623

uPA/KLK8 Correlate with Seizure Burden was not more pronounced or widespread at the chronic time following epileptogenic insults. Indeed, lack of uPA in mice point. At the chronic time point, a trend for a decreased area resulted in more neurodegeneration and less neurogenesis fraction of NeuN staining was seen in the granule cell layer after KASE,17,28 as well as in a bigger lesion and a worse of dentate gyrus, but at the same time, a trend for an behavioral outcome following CCI injury.29,30 However, in increased area of the granule cell layer of the dentate gyrus this study, no correlation was observed between uPA/KLK8 was observed. The decreased area fraction in the dentate expression and neuronal loss. Other studies have shown that gyrus may not necessarily reflect true cell loss, but might uPA is necessary for synaptic remodeling, repair, and recov- point to granule cell dispersion, which would indeed result ery following different kinds of injury.31–34 A pronounced in a less compact and broader granule cell layer of the den- decrease in uPA expression following KASE might result in tate gyrus. No correlation could be established between the the failure to initiate the appropriate repair mechanisms, level of [111In]MICA-401 binding and neuronal loss in all leading to a worse behavioral phenotype. However, uPA investigated brain regions at any given time point. We must deficiency did not have an effect on epilepsy phenotype fol- therefore conclude that neuronal degeneration does not fully lowing KASE or epileptogenesis following CCI injury in explain the observed decrease in levels of active uPA and mice.28,35 On the other hand, deficiency of KLK8 did result KLK8. in increased seizure susceptibility to kainic acid.36,37 It has Another possible explanation for the decreased levels of been shown that KLK8 regulates the balance between exci- active uPA and KLK8 is decreased neuronal activity due to tation and inhibition in hippocampus and inhibits pyramidal hypometabolism, which is known to occur following SE20– neuron hyperexcitability through its regulation of parvalbu- 26 and CCI injury27 and which affects more brain regions min-expressing interneuron activity via neuregulin 1 - than just those that are marked by severe neuronal loss. A receptor tyrosine-protein kinase ErbB4 signaling.37,38 Thus decreased neuronal activity can result in decreased expres- a decrease in KLK8 expression following an insult may lead sion and activity of uPA and KLK8 as well as of upstream to decreased activity of c-aminobutyric acid (GABA)ergic activators of these proteases, which all contribute to a lower interneurons and increased pyramidal neuron activity, binding of [111In]MICA-401. However, although cerebral resulting in increased seizure activity. We hypothesize that glucose hypometabolism is widespread in the acute phase a more pronounced decreased expression of KLK8, rather after SE, persistent hypometabolism in the chronic phase is than uPA, may be responsible for the increased seizure fre- less pronounced and less widespread than in the acute quency in KASE rats observed in this study. phase.20–23,25,26 On the contrary, we observed a distinct pro- gression of uPA/KLK8 decrease in the KASE model. Hypo- [111In]MICA-401 does not penetrate the intact blood– metabolism can therefore not completely explain the brain barrier observed progressive decrease in levels of active uPA and Our ex vivo autoradiography assay revealed an opposite KLK8. trend compared to the in vitro autoradiography assay. An increased tracer uptake was observed in perilesional cortex Hippocampal decrease in [111In]MICA-401 binding is and ipsilateral hippocampus of CCI rats in the ex vivo autora- correlated with seizure outcome in a TLE model diography assay, whereas a decreased binding of the radio- Of interest, the hippocampal decrease in [111In]MICA- tracer had been established in perilesional cortex and 401 binding was correlated with the number of seizures in ipsilateral dentate gyrus in the in vitro autoradiography assay. the chronic phase of the KASE model. Our results also show We confirmed that these brain regions exhibit a disrupted that this decrease in levels of active uPA and KLK8 is BBB with the Evans blue BBB permeability assay and could already present in the temporal lobe in the early phase of the therefore infer that we were most likely merely observing model. It would therefore be interesting to have a radio- local BBB leakage, which allowed regional accumulation of tracer that can be used for in vivo assessment of uPA/KLK8, the otherwise nonpermeable radiotracer in these brain to investigate whether the early decrease in levels of active regions. Indeed, the radiotracer has an unfavorable logD7.4 uPA and KLK8 correlates with chronic seizure burden. This ( 2.73 0.01) for BBB penetration,6 which limits its would allow us to investigate whether uPA/KLK8 can serve inÀ vivo application.Æ The radiotracer may be useful to imple- as a prognostic biomarker for the disease outcome following ment in animal models with a consistent BBB injury where SE. Whether the widespread decrease in levels of active an altered activity of uPA/KLK8 is to be expected. However, uPA and KLK8 is the cause or the consequence of the sei- one cannot compare results with healthy controls. One can zures or simply an epiphenomenon cannot be inferred from only compare the difference of active uPA/KLK8 levels this study. More research is warranted to investigate the pre- between subjects of the disease model, preferably after cor- cise relationship of the uPA/KLK8 decrease and the SRS. recting for possible interanimal variability in BBB disrup- Regardless of the exact sequence of events, a decrease of tion. Another possibility for implementing the radiotracer in active uPA and KLK8 will affect the extracellular matrix assays of active uPA/KLK8 levels in vivo, is to disrupt the and contribute to the pathophysiology of the disease. It has BBB using, for example, mannitol or focused ultrasound to been suggested that uPA plays a neuroprotective role enhance the delivery of the radiotracer into the brain. Tracer

Epilepsia, 58(9):1615–1625, 2017 doi: 10.1111/epi.13845 1624

S. Missault et al. accumulation, as well as Evans blue, was observed in the setting up the CCI model. Human and rat KLK8 and KLK4 were kindly ventricles of both sham and CCI rats, which is due to the donated by Prof. Viktor Magdolen, Klinische Forschergruppe der Frauen- klinik der TU M€unchen Klinikum rechts der Isar. We are grateful to UCB higher permeability of the blood–cerebrospinal fluid barrier Biopharma and to Georges Mairet-Coello, Mathieu Schmitt, and Tania at the choroid plexus as compared to the BBB.39 Deprez for the use of the slide scanner. This study was supported by ERA- Finally, this study underscores the importance of perform- NET NEURON/Research Foundation Flanders GA00913N, the Special Fund for Research of the University of Antwerp (BOF-UA), the Industrial ing in vitro autoradiography experiments in combination Research Fund of the University of Antwerp (IOF-UA), the Agency for with in vivo or ex vivo assays, which might otherwise lead to Innovation by Science and Technology (IWT) with a Strategic Basic incorrect interpretations of the in vivo/ex vivo assays. Research (SBO) grant (ChemProTools) and the Hercules Foundation Flan- ders with medium-scale research infrastructure. The Laboratory of Medici- nal Chemistry belongs to the Department of Pharmaceutical Sciences and is Limitations of the study a partner of the Antwerp Drug Discovery Network (www.ADDN.be). Ste- In the current study, we used healthy male age-matched phan Missault has a PhD fellowship of the Research Foundation Flanders (11K3714N/11K3716N). Sprague-Dawley rats as controls for the KASE rats at the chronic time point, rather than controls that underwent elec- trode implantation and video-EEG monitoring like the Disclosure KASE rats. Although we used only epidural electrodes and None of the authors has any conflict of interest to disclose. We confirm no depth electrodes, some damage to the dura and possibly that we have read the Journal’s position on issues involved in ethical publi- the superficial parietal cortex may have occurred in the cation and affirm that this report is consistent with those guidelines. KASE rats with concomitant alterations in the extracellular matrix, which could have contributed to the observed References decreases in levels of active uPA/KLK8. However, in this study we clearly showed that decreases in uPA/KLK8 were 1. Ngugi AK, Bottomley C, Kleinschmidt I, et al. Estimation of the bur- den of active and life-time epilepsy: a meta-analytic approach. Epilep- already present in the early phase post-SE when neither sia 2010;51:883–890. KASE nor control rats underwent electrode implantation. 2. Eadie MJ. Shortcomings in the current treatment of epilepsy. Exp Rev Moreover, in the chronic phase post-SE, decreased levels of Neurother 2012;12:1419–1427. 3. Lukasiuk K, Wilczynski GM, Kaczmarek L. Extracellular proteases in active uPA/KLK8 were observed in all investigated brain epilepsy. Epilepsy Res 2011;96:191–206. regions, which is unlikely due solely to the presence of 4. Pitkanen A, Ndode-Ekane XE, Lukasiuk K, et al. Neural ECM and epi- epidural electrodes. In addition, we observed that the lepsy. Prog Brain Res 2014;214:229–262. 5. Shiosaka S. Serine proteases regulating synaptic plasticity. Anat Sci Int chronic KASE rats had no neuronal loss in parietal cortex at 2004;79:137–144. the investigated level, indicating that the electrodes did not 6. Vangestel C, Thomae D, Van Soom J, et al. Preclinical evaluation of induce pronounced damage. [111 In]MICA-401, an activity-based probe for SPECT imaging of in vivo uPA activity. Contrast Media Mol Imaging 2016;11:448–458. 7. Joossens J, Ali OM, El-Sayed I, et al. Small, potent, and selective dia- ryl phosphonate inhibitors for urokinase-type plasminogen activator Conclusions with in vivo antimetastatic properties. J Med Chem 2007;50:6638– 6646. Our study showed a (sub)acute decrease in levels of 8. Amhaoul H, Hamaide J, Bertoglio D, et al. Brain inflammation in a active uPA/KLK8 in the temporal lobe in a TLE model and chronic epilepsy model: evolving pattern of the translocator protein in perilesional cortex and dentate gyrus in a TBI model. In during epileptogenesis. Neurobiol Dis 2015;82:526–539. 9. Amhaoul H, Ali I, Mola M, et al. P2X7 receptor antagonism reduces the TLE model, this decrease in active uPA/KLK8 was even severity of spontaneous seizures in a chronic model of temporal lobe more pronounced and widespread in the chronic phase of epilepsy. Neuropharmacology 2016;105:175–185. the model, when SRS had developed. Moreover, the hip- 10. Dedeurwaerdere S, van Raay L, Morris MJ, et al. Fluctuating and con- 111 stant valproate administration gives equivalent seizure control in rats pocampal decrease in [ In]MICA-401 binding correlated with genetic and acquired epilepsy. Seizure 2011;20:72–79. with the severity of the epilepsy phenotype in this disease 11. Gorter JA, Van Vliet EA, Rauwerda H, et al. Dynamic changes of pro- model, which suggests that analysis of reduced levels of teases and protease inhibitors revealed by microarray analysis in CA3 and entorhinal cortex during epileptogenesis in the rat. Epilepsia active uPA/KLK8 might represent a novel biomarker candi- 2007;48(Suppl. 5):53–64. date for epilepsy severity. In vivo applications of the radio- 12. Lahtinen L, Lukasiuk K, Pitkanen A. Increased expression and activity tracer are currently limited due to its inability to cross the of urokinase-type plasminogen activator during epileptogenesis. Eur J Neurosci 2006;24:1935–1945. intact BBB. Different strategies to improve the delivery of 13. Lukasiuk K, Kontula L, Pitkanen A. cDNA profiling of epileptogenesis the radiotracer into the brain are under consideration. This in the rat brain. Eur J Neurosci 2003;17:271–279. will allow us to determine whether the (sub)acute decrease 14. Masos T, Miskin R. mRNAs encoding urokinase-type plasminogen activator and plasminogen activator inhibitor-1 are elevated in the in active uPA/KLK8 can predict the epilepsy phenotype in mouse brain following kainate-mediated excitation. Brain Res Mol animal models of acquired epilepsy. Brain Res 1997;47:157–169. 15. Okabe A, Momota Y, Yoshida S, et al. Kindling induces neuropsin mRNA in the mouse brain. Brain Res 1996;728:116–120. 16. Momota Y, Yoshida S, Ito J, et al. Blockade of neuropsin, a serine pro- Acknowledgments tease, ameliorates kindling epilepsy. Eur J Neurosci 1998;10:760–764. 17. Cho E, Lee KJ, Seo JW, et al. Neuroprotection by urokinase plasmino- We thank Krystyna Szewczyk, Philippe Joye, and Ingmar Stuyver for gen activator in the hippocampus. Neurobiol Dis 2012;46:215–224. their excellent technical support and Dr. Tamuna Bolkvadze for her help in

Epilepsia, 58(9):1615–1625, 2017 doi: 10.1111/epi.13845 1625

uPA/KLK8 Correlate with Seizure Burden

18. He XP, Shiosaka S, Yoshida S. Expression of neuropsin in oligoden- susceptibility and acquired epilepsy phenotype but reduces sponta- drocytes after injury to the CNS. Neurosci Res 2001;39:455–462. neous exploration in mice. Epilepsy Behav 2015;42:117–128. 19. Tomizawa K, He X, Yamanaka H, et al. Injury induces neuropsin 31. Wu F, Catano M, Echeverry R, et al. Urokinase-type plasminogen acti- mRNA in the central nervous system. Brain Res 1999;824:308–311. vator promotes dendritic spine recovery and improves neurological 20. Goffin K, Van Paesschen W, Dupont P, et al. Longitudinal microPET outcome following ischemic stroke. J Neurosci 2014;34:14219–14232. imaging of brain glucose metabolism in rat lithium-pilocarpine model 32. Minor KH, Seeds NW. Plasminogen activator induction facilitates of epilepsy. Exp Neurol 2009;217:205–209. recovery of respiratory function following spinal cord injury. Mol Cell 21. Guo Y, Gao F, Wang S, et al. In vivo mapping of temporospatial Neurosci 2008;37:143–152. changes in glucose utilization in rat brain during epileptogenesis: an 33. Seeds N, Mikesell S, Vest R, et al. Plasminogen activator promotes 18F-fluorodeoxyglucose-small animal positron emission tomography recovery following spinal cord injury. Cell Mol Neurobiol study. Neuroscience 2009;162:972–979. 2011;31:961–967. 22. Jupp B, Williams J, Binns D, et al. Hypometabolism precedes limbic 34. Karagyaur M, Dyikanov D, Makarevich P, et al. Non-viral transfer of atrophy and spontaneous recurrent seizures in a rat model of TLE. BDNF and uPA stimulates peripheral nerve regeneration. Biomed Epilepsia 2012;53:1233–1244. Pharmacother 2015;74:63–70. 23. Lee EM, Park GY, Im KC, et al. Changes in glucose metabolism and 35. Bolkvadze T, Rantala J, Puhakka N, et al. Epileptogenesis after trau- metabolites during the epileptogenic process in the lithium-pilocarpine matic brain injury in Plau-deficient mice. Epilepsy Behav 2015;51:19– model of epilepsy. Epilepsia 2012;53:860–869. 27. 24. Park GY, Lee EM, Seo MS, et al. Preserved hippocampal glucose 36. Davies B, Kearns IR, Ure J, et al. Loss of hippocampal serine protease metabolism on 18F-FDG PET after transplantation of human umbilical BSP1/neuropsin predisposes to global seizure activity. J Neurosci cord blood-derived mesenchymal stem cells in chronic epileptic rats. J 2001;21:6993–7000. Korean Med Sci 2015;30:1232–1240. 37. Kawata M, Morikawa S, Shiosaka S, et al. Ablation of neuropsin-neur- 25. Shiha AA, de Cristobal J, Delgado M, et al. Subacute administration of egulin 1 signaling imbalances ErbB4 inhibitory networks and disrupts fluoxetine prevents short-term brain hypometabolism and reduces hippocampal gamma oscillation. Transl Psychiatry 2017;7:e1052. brain damage markers induced by the lithium-pilocarpine model of epi- 38. Tamura H, Kawata M, Hamaguchi S, et al. Processing of neuregulin-1 lepsy in rats. Brain Res Bull 2015;111:36–47. by neuropsin regulates GABAergic neuron to control neural plasticity 26. Zhang L, Guo Y, Hu H, et al. FDG-PET and NeuN-GFAP immunohis- of the mouse hippocampus. J Neurosci 2012;32:12657–12672. tochemistry of hippocampus at different phases of the pilocarpine 39. Cipolla MJ. The Cerebral Circulation. San Rafael (CA): Morgan & model of temporal lobe epilepsy. Int J Med Sci 2015;12:288–294. Claypool Life Sciences; 2009. 27. Park BN, Yoon JK, An YS. Bone marrow mesenchymal stem cell transplantation in acute brain trauma. Improvement of brain glucose metabolism in a rat model. Nuklearmedizin 2013;52:192–197. Supporting Information 28. Lahtinen L, Ndode-Ekane XE, Barinka F, et al. Urokinase-type plas- minogen activator regulates neurodegeneration and neurogenesis but Additional Supporting Information may be found in the not vascular changes in the mouse hippocampus after status epilepti- cus. Neurobiol Dis 2010;37:692–703. online version of this article: 29. Morales D, McIntosh T, Conte V, et al. Impaired and trau- Figure S1. Example of a NeuN-stained brain section and matic brain injury in mice. J Neurotrauma 2006;23:976–984. setting of the threshold. 30. Rantala J, Kemppainen S, Ndode-Ekane XE, et al. Urokinase-type plasminogen activator deficiency has little effect on seizure

Epilepsia, 58(9):1615–1625, 2017 doi: 10.1111/epi.13845