Genetic and Pharmacological Targeting of Heat Shock Protein 70 in

Research Article Cite This: ACS Chem. Neurosci. XXXX, XXX, XXX−XXX

pubs.acs.org/chemneuro

Genetic and Pharmacological Targeting of Heat Shock Protein 70 in the Mouse Amygdala-Kindling Model Eva-Lotta von Rüden,† Fabio Wolf,† Fabio Gualtieri,† Michael Keck,† Clayton R. Hunt,‡ Tej K. Pandita,‡ and Heidrun Potschka*,† †

Downloaded via KAOHSIUNG MEDICAL UNIV on November 20, 2018 at 08:55:24 (UTC). See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.

Institute of Pharmacology, Toxicology, and Pharmacy, Ludwig-Maximilians-University Munich, Koeniginstraße 16, D-80539 Munich, Germany ‡ Department of Radiation Oncology, The Houston Methodist Research Institute, 6550 Fannin Street SM8-024, Houston, Texas 77030, United States S Supporting Information *

ABSTRACT: Inflammatory responses involving Toll-like receptor signaling represent a key factor contributing to epileptogenesis. Thus, it is of particular interest to explore the relevance of toll-like receptor ligands and modulators, such as heat shock protein 70 (HSP70). Motivated by recent findings demonstrating an upregulation of HSP70 in a model of epileptogenesis, we analyzed the consequences of genetic and pharmacological targeting of HSP70 expression in a mouse kindling paradigm. Lack of inducible HSP70 resulted in increased prekindling seizure thresholds. However, at threshold stimulation the deficiency-promoted seizure spread, as indicated by an increased seizure severity. Subsequent kindling stimulations elicited more severe seizures in knockout mice, whereas endogenous termination of seizure activity remained unaffected with duration of behavioral and electrographic seizure activity comparable to that of wild-type mice. Interestingly, HSP70 deficiency resulted in enhanced microglia activation in the CA1 region. Next, we assessed a pharmacological targeting approach aiming to promote HSP70 expression. Celastrol treatment had no impact on kindling progression but reduced postkindling seizure thresholds and enhanced microglia activation in CA1 and CA3. In conclusion, the findings from HSP70-knockout mice support a protective role of HSP70 with an effect on microglia activation and spread of seizure activity. Unexpectedly, celastrol administration resulted in detrimental consequences. These findings should be considered as a warning about the general safety of celastrol as a drug candidate. The impact of pathophysiological mechanisms on the quality of celastrol effects requires comprehensive future studies exploring influencing factors. Moreover, alternate strategies to increase HSP70 expression should be further developed and validated. KEYWORDS: Celastrol, microglia, epileptogenesis, Hspa1a/b, seizure, Toll-like receptor

■

INTRODUCTION

an excessive inflammatory response following epileptogenic insults or in the epileptic brain. Therefore, we have recently explored a proteomic data set with analysis of different time points in an electrically induced post-status epilepticus model with development of spontaneous recurrent seizures. Analysis of hippocampus and parahippocampal cortex tissue revealed a pronounced regulation of inducible heat shock protein 70 (HSP70) in the early phase following the insult and during the latency phase.6 Taking into account that HSP70 has been described as a modulator and ligand of TLR2 and TLR4,9,10 these data provided the first evidence that HSP70 may

Numerous studies have demonstrated that Toll-like receptor (TLR) signaling can trigger glia activation during epileptogenesis as well as in the chronic epileptic brain.1−3 The resulting persistence of enhanced inflammatory signaling can critically contribute to epilepsy development following an insult and to ictogenesis in the epileptic brain.1,3 Thus, it is of particular interest to develop targeting approaches limiting TLR signaling and its consequences in order to prevent epilepsy or seizure development. Respective approaches have focused on high-mobility group box protein 1 as a wellcharacterized ligand of TLR.4,5 However, other ligands of TLR have been described, and these can additionally modulate TLR activation.6−8 Disregarding a potential role of these ligands might limit the efficacy of novel approaches aiming to prevent © XXXX American Chemical Society

Received: September 7, 2018 Accepted: November 5, 2018 Published: November 5, 2018 A

DOI: 10.1021/acschemneuro.8b00475 ACS Chem. Neurosci. XXXX, XXX, XXX−XXX

Research Article

ACS Chemical Neuroscience

Figure 1. Impact of Hspa1a and Hspa1b genetic deficiency on initial kindling parameters. Initial afterdischarge threshold (A), initial seizure severity (B), initial seizure duration (C), and initial afterdischarge duration (D) are depicted for KO and WT mice. The afterdischarge threshold and the seizure severity were significantly increased in KO in comparison to WT mice, whereas the seizure and afterdischarge duration were comparable in both groups. Statistical tests used: unpaired Student’s t test, if necessary with Welch’s correction, Mann−Whitney test. All data are given as mean ± SEM, * p < 0.05, Animal numbers in (A)−(D): WT mice n = 17, KO n = 25.

Figure 2. Impact of Hspa1a and Hspa1b genetic deficiency on kindling progression. (A) Seizure severity, (B) seizure duration, and (C) afterdischarge duration are depicted for each day of stimulation. Hspa1a/b deficiency had a significant impact on the seizure severity with significant higher seizure scores on days 1 and 2 (A), whereas the seizure duration (B) and the afterdischarge duration were unaffected by the genotype (C). In line with the previous finding, the number of stimulations until elicitation of the first generalized seizure was significantly decreased (D) and the cumulative afterdischarge duration was not affected by the genotype (E). Representative images of EEG recordings in WT (F) and KO (G) mice of stimulation day two. The electrical trigger (first spike) is followed by an interference period of about 4 s, and then the typical electrical afterdischarge pattern of an epileptic seizure is visible. Statistical tests used: two-way ANOVA for repeated measurements and Bonferroni multiple comparison test, Mann−Whitney test, and unpaired Student’s t test. All data are given as mean ± SEM, * p < 0.05, Animal numbers: WT mice n = 8, KO n = 11.

NF-kB transcription factor with the consequence of reduced pro-inflammatory cytokine production.13,17−19 Considering the putative role of HSP70 in the generation of a hyperexcitable epileptic network and the contrasting data on its functional link with inflammatory signaling cascades, it is of particular interest to analyze the consequences of genetic and pharmacological targeting approaches in chronic epilepsy models. Earlier studies suggested a protective function of HSP70 overexpression as well as intracerebroventricular administration of purified HSP70 in animal models with chemically induced seizures.18,19 However, conclusions from these studies are limited, considering the use of a GABA

critically contribute to inflammatory responses and resulting hyperexcitability. Whereas general agreement exists that HSP70 acts as an important regulator of inflammatory responses,11,12 there is an ongoing debate about the quality of its effects.11,13,14 On the one hand, findings suggest that extracellular, peptide-free HSP70 can trigger activation of innate immune receptors such as TLR4, resulting in increased production and release of proinflammatory cytokines via the activation of NF-kB transcription factor.15,16 On the other hand, data have indicated that intracellular HSP70 induced by stress signals can also serve as a negative modulator of inflammation via decreasing B


Research Article


Figure 3. Impact of Hspa1a and Hspa1b genetic deficiency on microglia activation in kindled animals. In the CA1 region of the hippocampal formation microglia was significantly activated in KO animals in comparison to WT mice (A), whereas in CA3 (B) and in the hilus (C) microglia activation reached a comparable level. Representative histological images of WT and KO mice in hippocampal CA1 region at low (D, F) and high magnification (E, G). Higher magnification images show cell branching of positive stained microglia cells in WT (E) and in KO (G) animals with decreased complexity in arborization in KO mice indicating enhanced microglia activation. Statistical tests used: unpaired Student’s t test. All data are given as mean ± SEM, * p < 0.05. Animal numbers: n = 6. Abbreviations: Or, stratum oriens; Py, stratum pyramidale; Rad, stratum radiatum; alv, alveus. Scale bars: (D) and (F) 100 μm and (E) and (G) 25 μm.

raise an interest in the question as to whether celastrol might also exert beneficial effects in epilepsy paradigms such as the kindling model.

antagonist chemoconvulsant for seizure induction and the fact that HSP70 can act as a modulator of GABAergic neurotransmission.18−22 When we recently analyzed the consequences of HSP70 overexpression in an electrical kindling paradigm, transgenic mice exhibited altered prekindling seizure thresholds along with an impact on kindling progression.23 In the present study, we addressed the hypothesis that a genetic deficiency of HSP70 should increase seizure susceptibility and accelerate progression of seizure severity in an electrical kindling paradigm with repeated seizure induction. Thus, we used a genetic modified mouse line, which lack the encoding genes for HSP70: i.e., Hsp70.1 and Hsp70.3.24 The two genes are identical, and their functions are thought to be redundant; therefore, knockout of both genes was required. Moreover, using the same paradigm, we determined the effects of celastrol, which has been characterized as a neuroprotective compound with anti-inflammatory and antioxidant properties.25 Celastrol activates heat shock factor 1, a stress-inducible and DNA-binding transcription factor that plays a central role in the regulation of the heat shock response. Heat shock factor 1 mediates induction of Hspa1a/b gene expression including intracellular HSP70 upregulation.26,27 HSP70 induction by celastrol improves the outcome in different animal models of neurological and neurodegenerative diseases, including traumatic brain injury, brain ischemia, and multiple sclerosis.26,28−30 Respective findings

■

RESULTS AND DISCUSSION Hspa1a and Hspa1b Genetic Deficiency Increased Initial Seizure Thresholds. Determination of the initial afterdischarge threshold indicated that a genetic deficiency of Hspa1a and Hspa1b reduced seizure susceptibility in mice (p = 0.049, Figure 1A). Increased thresholds suggest a protective anticonvulsant effect of the KO with a limitation of mechanisms responsible for ictogenesis. However, once the threshold is reached with an electrographic and behavioral seizure response upon stimulation, the severity of induced seizures reaches higher scores in KO mice in comparison to WT mice (p = 0.0041, Figure 1B). In contrast, neither the duration of behavioral seizure activity nor the duration of afterdischarges was influenced by the genotype (seizure duration p = 0.7611, afterdischarge duration p = 0.3048, Figure 1 C−D). Analysis of seizure susceptibility following the kindling procedure did not reveal significant group differences (F(1,17) = 2.64, p = 0.1226, data not shown). Hspa1a and Hspa1b Genetic Deficiency Increased Seizure Severity at Supra-Threshold Stimulations. All KO mice exhibited generalized seizures during the first and C


Research Article


Figure 4. Impact of celastrol treatment on kindling progression in WT mice. Celastrol treatment in WT mice decreased the postkindling afterdischarge threshold in comparison with the initial afterdischarge threshold (A). Celastrol treatment exerted no impact on seizure severity (B), seizure duration (C), and afterdischarge duration (D). Moreover, the number of stimulations until elicitation of the first generalized seizure (E) and the cumulative afterdischarge duration (F) were comparable in vehicle- and celastrol-treated WT mice. Statistical tests used: two-way ANOVA for repeated measurements and Bonferroni multiple comparison test, Mann−Whitney test, and paired and unpaired Student’s t test. All data are given as mean ± SEM, * p < 0.05, Animal numbers: WT vehicle n = 8, WT celastrol n = 9).

determined comparable cumulative afterdischarge durations for WT and KO mice (p = 0.119, Figure 2E). Hspa1a and Hspa1b Genetic Deficiency Increased Microglia Activation in CA1 Region. We analyzed microglia activation in the CA1, CA3, and hilus regions of the hippocampal formation of kindled WT and KO mice (see Figure 3). For all of these regions we detected no statistical difference between both hemispheres and along the rostrocaudal axis of the hippocampus. Genetic Hspa1a/b deficiency reduced the ratio of total cell area/cell body area (p = 0.0225, Figure 3A), indicating an enhanced activation of microglia in the CA1 region. However, microglia activation remained unaffected in the CA3 region (p = 0.132, Figure 3B) and in the hilus (p = 0.3646, Figure 3C). Celastrol Reduced Postkindling Seizure Thresholds. Celastrol administration remained without effects on initial seizure thresholds in WT mice (p = 0.598, Figure 4A).

second kindling stimulations. As a consequence, mean seizure severity scores reached significantly higher levels at these stimulation days in comparison to WT mice (p < 0.01, Figure 2A). In the latter group, the majority of mice responded with focal seizures during the first two stimulation days (day 1 and 2: five out of eight). Generalized seizure activity was observed in some WT mice from the first stimulation day on. Toward the end of the stimulation phase, mean seizure severity scores were comparable in KO and WT mice. As further confirmation of these findings, the number of stimulations until the first generalized seizures proved to be reduced as a consequence of the genetic deficiency (p = 0.0039, Figure 2D). Analysis of the duration of behavioral as well as electrographic seizure activity during the kindling procedure did not reveal relevant group differences (seizure duration F(1,136) = 0.3488, p = 0.5626 and afterdischarge duration F(1,136) = 0.01262, p = 0.9119, Figure 2B,C). In line with this result, we D


Research Article


Figure 5. Impact of celastrol treatment on microglia activation in kindled mice. The ratio “total cell area/body cell area”, an index of microglia activation, was decreased in celastrol-treated WT mice in comparison to vehicle-treated WT mice in the CA1 (A) and CA3 (D) regions. This effect was abolished in vehicle- and celastrol-treated KO mice. In the hilus (G), celastrol treatment remained without effects. Representative histological images of celastrol- and vehicle-treated WT mice in the hippocampal CA1 region at low (B, E) and high magnification (C, F) and hippocampal CA3 region at low (H, J) and high magnification (I, K). Microglia cells of celastrol-treated animals (F, K) showed less cell branching in comparison to microglia cells of vehicle-treated animals (C and I), indicating an activated state of microglia cells. Statistical tests used: two-way ANOVA and Bonferroni multiple comparison test. All data are given as mean ± SEM, * p < 0.05. Animal numbers: n = 6. Abbreviations: Or, stratum oriens; Py, stratum pyramidale; Rad, stratum radiatum; GrDG, granule cell layer of the dentate gyrus; alv, alveus; SLu, stratum lucidum; DLG, dorsal nucleus of lateral geniculate body. Scale bars: (B, E, H, J) 100 μm; (C, F, I, K) 25 μm.

by reduced afterdischarge thresholds in WT mice with celastrol exposure (p = 0.0262, Figure 4A). In both groups, all mice except for one vehicle-treated mouse responded with generalized stage 4 or 5 seizures to the postkindling threshold stimulation. The duration of behavioral seizure activity and afterdischarges at threshold stimulation did not differ between groups (seizure duration p = 0.875, afterdischarge duration p = 0.9924, data not shown). Celastrol treatment in KO mice had no impact on kindling seizure

Moreover, seizure parameters at threshold stimulation proved to be comparable in WT mice receiving vehicle or celastrol injections (seizure severity p = 0.7854, seizure duration p = 0.7882 and afterdischarge duration p = 0.596, data not shown). The majority of animals responded with focal seizures (score 1−2) to threshold stimulation. The evaluation of thresholds following the kindling procedure revealed an increased seizure susceptibility reflected E


Research Article


kindled mice.32−34 In addition, the influence on the progression of seizure severity reflecting the development of a kindling hyperexcitable network can be assessed.32,34,35 Without previous exposure to induced seizures, lack of inducible HSP70, encoded by the two genes Hspa1a−/− and Hspa1b−/−, resulted in an increase in seizure thresholds, demonstrating that the genetic deficiency exerts protective effects preventing ictogenesis. However, when thresholds were reached, the deficiency promoted seizure spread, as indicated by a significantly increased seizure severity. In contrast, endogenous termination of seizure activity remained unaffected by loss of HSP70 with a duration of behavioral and electrographic seizure activity comparable to that of wild-type mice. The inverse conclusion suggests that HSP70 may increase seizure susceptibility in individuals who, without a pre-existing neurological disease, are exposed to a seizureprovoking situation. On the other hand, our findings indicate that inducible HSP70 may have the potential to limit the spread of seizure activity, thereby protecting from seizure generalization. This conclusion was further supported by data from the subsequent suprathreshold kindling stimulations, which induced more severe seizures in the KO mice. On the other hand, seizure thresholds proved to be higher in KO mice, indicating a decreased seizure susceptibility and rendering KO mice less prone to seizure initiation. Our present findings seem to be in line with an earlier study reporting a delayed progression of seizure severity in a pentylenetetrazole-induced chemical kindling procedure in HSP70 overexpressing mice.18 Moreover, Ekimova et al.19 reported that intracerebroventricular administration of purified HSP70 reduces the severity of NMDA- or pentylenetetrazoleinduced acute seizures. As stated in the Introduction, we have recently completed an analysis in mice overexpressing the human inducible HSP70 isoform HSPA1A.23 Respective data are in partial conformity with our present findings, as the transgenic mice exhibited lowered prekindling seizure thresholds but exhibited generalized seizures early during the kindling paradigm with subsequent stimulations. It needs to be considered that we have analyzed the overexpression of one human HSP70 isoform in this previous study versus the knockout of both endogenous mouse HSP70 isoforms in the present study. Thus, one cannot necessarily expect the exact opposite outcome when comparing both studies. Interestingly, lack of inducible HSP70 resulted in an enhanced microglia activation state in the CA1 hippocampal subregion. Strong evidence has previously been obtained by several studies that excessive microglia activation along with increased formation of pro-inflammatory cytokines can directly affect synaptic signaling, excitability, and seizure susceptibility.36 For instance, interleukin-1beta and tumor necrosis factoralpha can affect excitability mediated by phosphorylation of the NMDA receptor subunit NR2B or by alterations in the expression of NMDA and AMPA receptor subunits, respectively.37 Thus, it is likely that the impact of HSP70 genetic deficiency on the spread and generalization of kindled seizure activity is mediated by increased inflammatory signaling. In conclusion, our findings rather support a negative regulatory impact of HSP70 on neuroinflammation rather than a pro-inflammatory action. In agreement with such a role, Kim and Yenari13 described that intracellular upregulation of HSP70 can reduce the generation and release of proinflammatory factors including NF-KB, matrix metalloproteinases, reactive oxygen species, and the pro-inflammatory

thresholds (see Supplementary Figure 1 in the Supporting Information). Celastrol Treatment Remained without Effect on Kindling Progression in WT Mice. Throughout the kindling procedure vehicle- and celastrol-treated WT mice exhibited comparable mean seizure severity scores (F(1,120) = 0.09508, p = 0.7621; Figure 4B). In line with these findings, the number of stimulations until the first generalized seizure did not differ between the treatment groups (p = 0.8407, Figure 4E). Neither the duration of behavioral seizure activity nor the duration of electrographic seizure activity indicated relevant group differences (seizure duration F(1,120) = 1.904, p = 0.1879; afterdischarge duration F(1,120) = 2.076, p = 0.1702; Figure 4C,D). This result received further confirmation by the fact that the cumulative afterdischarge duration proved to be in the same range in vehicle- and celastrol-treated mice (p = 0.1702, Figure 4F). On the basis of these findings, we expected no impact of celastrol treatment in KO mice. To confirm this hypothesis, we analyzed the effects of celastrol in KO mice. As expected, celastrol treatment remained without any relevant effects (for details see Supplementary Figure 1 and supplementary text in the Supporting Information). Celastrol Increased Microglia Activation in WT but Not in Mice with Hspa1a and Hspa1b Genetic Deficiency. Celastrol treatment had an impact in CA1 and CA3 regions of the hippocampal formation with a decreased total cell area/body cell area ratio indicating an increase in microglia activation in these regions (CA1, two-way ANOVA F(1,20) = 11.0, p = 0.0035, Bonferroni multiple comparison test p < 0.01, Figure 5A−C; CA3 ,two-way ANOVA F(1,20) = 5.26, p = 0.0328, Bonferroni multiple comparison test p < 0.05, Figure 5A−F,H−K). This effect was abolished in vehicle- and celastrol-treated KO mice (CA1, p > 0.05, Figure 5A; CA3, p > 0.05, Figure 5D), suggesting that this effect depends on HSP70 signaling. In the hilus, celastrol remained without impact on microglia activation (F(1,20) = 0.403, p = 0.5328; Figure 5G). Genetic and pharmacological targeting of HSP70 in the mouse kindling paradigm provided novel insights into the impact of HSP70 on ictogenesis and seizure spread. Moreover, our findings contribute to the state of knowledge regarding the role of HSP70 as a modulator of microglia activation in neurological diseases. We confirmed a morphological change of microglia cells from the nonactivated into the activated state. However, so far we do not know if the expression or signaling profile of these microglia cells is more of a pro-inflammatory or is more of an anti-inflammatory nature. Further investigations in this model and in these KO mice are needed to determine how HSP70 is affecting microglial activation during epilepsy development. Finally, the study indicates that celastrol may also exert proinflammatory detrimental effects probably depending on the pathophysiological cellular and molecular alterations characterizing the disease state. Knockout mice lacking both Hspa1a−/− and Hspa1b−/− represent a valuable tool to address the functional role of inducible HSP70 in in vivo disease models.24,26,31 We selected the kindling paradigm for assessment of the consequences of the genetic deficiency for several reasons. As demonstrated in previous studies, analysis of kindling data can provide valuable and comprehensive information about the impact of a genetic deficiency on seizure thresholds, seizure spread, and seizure termination, allowing comparison between naive and fully F


Research Article


effects that might explain the results obtained. However, we faced severe technical problems and despite intense efforts did not succeed in establishing a method for expression analysis of mouse HSP70. In the present study, we have adopted the treatment protocol, including dosing and administration intervals, from Eroglu et al.,26 who reported a relevant improvement of the in vivo consequences of neurological dysfunction in a mouse traumatic brain injury model. In line with this finding, the authors described that celastrol limited the number of CD11bpositive cells infiltrating the injured site.26 When comparing our present findings with these and other outcomes of celastrol exposure in rodent models of neurological disease, one has to conclude that the impact of celastrol on microglia and inflammatory responses might differ tremendously depending on the disorder, its pathophysiological mechanisms and alterations, and the model used. However, the techniques and methods applied should also be considered as a contributing factor. In this context it is emphasized that we have applied an approach assessing the morphological state of microglial cells providing information about the functional state. This method has previously been applied and validated by Hovens et al.44 to analyze microglia morphology as a marker for microglia activation in Iba1-stained brain sections. As a matter of course, a respective method will provide data with a different informative value as counting of microglial cells. Regarding the data by Eroglu et al.,26 an additional bias by a putative impact of celastrol on CD11b-positive peripheral macrophages cells entering the brain via the compromised blood−brain barrier also needs to be taken into account. As we analyzed microglia morphology on the basis of Iba1 immunohistochemistry and not cytokine expression levels, we finally cannot conclude about the nature of microglia activation. Further studies should address the question if celastrol-induced microglia activation triggers the expression of pro- or anti-inflammatory cytokine expression. Nevertheless, our present findings raise concerns regarding the general safety of celastrol as a drug candidate for neurological disorders, as it may promote a detrimental activation of microglial cells and may under specific circumstances lower seizure thresholds. The findings imply that the potential for anti- and pro-inflammatory effects of celastrol requires careful further assessment: in particular, exploring the pathophysiological factors influencing the cellular outcome in different disease states. Considering the safety profile of celastrol, it would be interesting to test additional HSP70 inducers in this animal model, e.g. 15-BGP or 17-AAG, and to analyze their impact on epileptogenesis. In this context, it would be desirable to test specific HSP70 agonists. However, the development of specific HSP70 agonists is still in its infancy and faces many challenges. Initial efforts in the development of specific inducers have focused on small molecules and on modulating regulatory miRNAs.45,46 In conclusion, our findings in mice with HSP70 genetic deficiency support a protective role of HSP70 on the basis of its impact on activation of microglial cells and on the spread and generalization of seizure activity. These data generally suggest a protective potential of HSP70 targeting approaches. However, celastrol administration unexpectedly resulted in detrimental consequences, including an increase in microglia activation. These findings should be considered as a warning about the general safety of celastrol as a drug candidate. The

cytokines interleukin-1beta and tumor necrosis factor-alpha. According to findings by Ferat-Osorio et al.,38 these effects might be mediated by a TLR4 antagonistic action of HSP70. The authors described that HSP70 negatively regulates the production of pro-inflammatory cytokines in blood monocytes exposed to TLR agonists.38 However, here we did not address the question of downstream inflammatory actions of microglia cells. As microglia cells can be activated to produce antiinflammatory cytokines, e.g. interleukin-4 or -10, or can be activated to produce pro-inflammatory cytokines, e.g. interleukin-1beta and tumor necrosis factor-alpha, or even a mixture of both, the analysis of inflammatory mediators would help to unveil the question if HSP70 triggers a pro-inflammatory signaling cascade in this model. This has to be investigated in detail in future studies. For future studies, a pharmacological approach with selective inhibition of HSP70 should be investigated to confirm our findings in KO mice. This would address questions regarding developmental adaptations, which may be present in the HSP70 KO mice and would provide further information about HSP70 as a potential target for the treatment of epilepsy. The fact that lack of HSP70 triggered activation of microglia along with an increased seizure severity in the kindling paradigm further motivated us to assess a pharmacological targeting approach aiming to promote HSP70 expression and its anti-inflammatory protective function. On the basis of the outcome of several in vivo studies reporting protective effects,26,28−30 we selected celastrol for the pharmacological study. Celastrol is a triterpene, which is extracted as a natural plant component from Tripterygium wilfordii Hook.39 Its antiinflammatory potency has already provided a basis for its use in traditional medicine approaches.40 As stated above, its antiinflammatory effects have at least been partially attributed to potentiation of HSP70 expression.41 Celastrol administration did not exert relevant effects on prekindling seizure susceptibility or on seizure severity progression during kindling acquisition. Following the kindling procedure, seizure susceptibility was increased with thresholds lower than those determined in vehicle-treated control mice. Thus, in contrast with our hypothesis, only a very limited effect of the test compound was observed in the kindling model, and this effect rather suggests a detrimental influence on seizure thresholds in animals with previous exposure to repeated seizure induction. These data provide the first evidence that celastrol might even trigger ictogenesis in individuals with epilepsy, history of an epileptogenic insult, or other predispositions. In line with this in vivo result, post mortem analysis revealed an enhanced activation of microglia cells in the hippocampal CA1 and CA3 regions as a consequence of subchronic celastrol administration. Again, this result was unexpected on the basis of the series of previous studies in different rodent models of neurological diseases. Neuroprotective and beneficial effects of celastrol have for instance been reported in models of Alzheimer’s disease, brain ischemia, traumatic brain injury, and multiple sclerosis.26,29,30,42,43 In several of these studies, evidence for an anti-inflammatory effect of celastrol exposure has been obtained.26,30,42 In this context, we like to emphasize that the analysis of HSP70 expression levels induced by celastrol treatment would have been of interest for this study. The repetitive administration of celastrol and the consecutively increased HSP70 expression levels might have induced significant side G


Research Article


>15% body weight were euthanized due to termination criteria. Because the animals continuously lost weight with ongoing drug administration during the kindling period, it was necessary to stop the kindling process along with celastrol applications in the middle of the experiment for 4 days and then again for 2 days to stabilize the animals’ body weight. Another adverse effect of the celastrol application was a retraction of the flanks directly after celastrol application. Electrode Implantation. A bipolar Teflon-isolated stainless steel electrode with a diameter of 280 μm was stereotactically implanted in the right amygdala of all mice as described earlier by von Rüden et al.32 For anesthesia and analgesia mice received injections of chloral hydrate (400 mg/kg in a volume of 10 mL saline i.p., Merck KGaA, Darmstadt, Germany), bupivacaine 0.5% with epinephrine 0.0005% (5 mL/kg s.c., Jenapharm, Mibe, GmbH, Brehna), and 30 min prior to surgery and 24 h after surgery meloxicam (1 mg/kg, s.c. Metacam, Boehringer-Ingelheim, Ingelheim, Germany). For the exact placement of the electrode, mice were fixed in a stereotactic frame (TSE Systems GmbH, Bad Homburg, Germany) and the stereotactic coordinates in millimeters relative to Bregma were applied as follows: WT and KO AP −1.4, L +3.3, DV −4.8. Kindling: Experimental Design. The kindling process was initiated after a 2 week recovery period. Prior to the kindling procedure WT mice were randomized (randomizer.org) and distributed to either the nonkindled control (n = 17) or the kindling group (n = 17). We handled animals from the nonkindled control group exactly in the same way as animals from the kindling group, except for the electrical stimulations. After determination of the initial naive afterdischarge threshold, animals (WT (n = 17) and KO (n = 25)) were distributed to groups on the basis of stratified randomization (randomizer.org) considering their afterdischarge threshold to ensure equivalent mean afterdischarge thresholds for the two groups (randomized group allocation from a low-, intermediate-, and highthreshold group). With this procedure the mice were distributed to groups receiving either vehicle (WT mice n = 8 and KO n = 11) or celastrol treatment (WT mice n = 9 and KO n = 14). We further checked that the bodyweight of the animals in the respective groups was homogeneously distributed. During kindling acquisition, the experimenter was unaware of the animals’ genotype. We performed electrical stimulations at the same time of the day (1 pm to 5 pm) to minimize the impact of circadian rhythms. We transferred the mice to the laboratory 30 min prior to the electrical stimulation for habituation to the experimental environment. All electrical stimulations were performed in a glass aquarium (40 × 35 × 40 cm). Behavioral motor responses were video monitored and the electroencephalogram was recorded with LabChart 7 (ADInstruments Ltd., Hastings, United Kingdom). Two days after the determination of the initial afterdischarge threshold in naive mice, mice received vehicle/celastrol injections, and 24 h later, we evaluated a second afterdischarge threshold. In addition, 6 h prior to the determination of the second afterdischarge threshold, animals received another vehicle/celastrol injection to test the impact of celastrol on the afterdischarge threshold. During kindling acquisition, animals were injected with vehicle/celastrol once daily 6 h prior to the electrical stimulation. Following the determination of the initial afterdischarge threshold, the amygdala was electrically stimulated once daily (nine stimulations in total). As we were only interested in kindling progression rates and did not aim for experiments in fully kindled mice, we did not continue stimulation once all vehicle-treated WT mice exhibited at least five consecutive generalized stage 4 or 5 seizures; thus, mice were not fully kindled. We made the observation that mice do not reproducibly exhibit the typical described decrease in afterdischarge threshold after the kindling process. In contrast, the kindled state of rats is more stable and therefore they reach more reliably the significant decrease in afterdischarge thresholds. The final electrical stimulation was scheduled 24 h prior to the determination of the postkindling afterdischarge threshold. Ten mice (5 WT and 5 KO, all kindled and celastrol treated) were euthanized in the course of the experiment due to alterations in their

impact of pathophysiological mechanisms on the quality of celastrol effects requires careful and comprehensive future studies exploring influencing factors. Moreover, alternate strategies to increase HSP70 expression should be further developed and validated. Finally, we like to emphasize that followup studies focusing on the expression profile and expression levels of downstream signaling molecules are necessary to draw a final conclusion if microglia activation via the HSP70 signaling cascade has a protective or detrimental impact on epileptogenesis.

■

METHODS

Animals. All animal experiments were approved by the responsible government of Upper Bavaria (license number 55.2-1-54-2532-5815). The study was performed in accordance with the EU directive 2010/63/EU and with the German Animal Welfare Act and in compliance with the ARRIVE guidelines. Breeding pairs of C57BL/6 (in the following wild-type (WT)) mice were purchased from Envigo (Blackthorn, United Kingdom) and homozygous breeding pairs of Del(Hspa1b-Hspa1a)1Dix (in the following Hspa1a−/−/Hspa1b−/−-knockout (KO)) mice backcrossed into the C57BL/6 strain were obtained from the the Department of Radiation Oncology of the Houston Methodist Research Institute.24 Male WT mice (n = 39) and male KO (n = 30) mice in an age range of 14−18 weeks with a body weight range of 24−35 g were used for the experiments. All mice were kept under controlled environmental conditions (temperature, 20−24 °C; humidity, 45−65%; 12 h dark/light cycle), fed with standard laboratory food (Ssniff R/M Haltung, ssniff Spezialdiäten GmbH, Soest, Germany), and offered tap water ad libitum. Experimental animals were housed individually in Type II Makrolon cages enriched with poplar wood bedding (Lignocel Select, Altromin Spezialfutter GmbH & Co. KG, Lage, Germany), polycarbonate houses, and nesting material (Nestlets, Ancare, Bellmore, NY, USA). All efforts were made to minimize pain or discomfort and to limit the number of the animals used in the study. Genotyping. The genotype of the KO mice was evaluated by PCR reaction twice during the experiments: shortly after weaning and after killing the mice. For genotyping of KO mice, a Kit Phire Tissue Direct PCR Master Mix (ThermoFisher Scientific, Waltham, MA, USA) was used following the manufacturer instructions (forward primer, CCCCACTTCCACGAGAATTTAC; reverse primer, TGTATTGCACGTGGGCTTTATC; annealing temperature, 68.7 °C; cycles, 40; TProfessional Basic Thermocycler (Biometra GmbH, Goettingen, Germany)).24 The genotyping was based on the absence of an intragenetic sequence that is deleted along with the Hspa1a and Hspa1b genes.24 Celastrol Treatment. Celastrol (10-hydroxy-2,4a,6a,9,12b,14ahexamethyl-11-oxo-1,2,3,4,4a,5,6,6a,11,12b,13,14,14a, 14b-tetradecahydropicene-2-carboxylic acid, CAS No. 34157-83-0, C0869 SIGMA, Sigma-Aldrich Chemie GmbH, Taufkirchen, Germany) is a phytochemical triterpene from the thunder god vine (Tripterygium wilfordii Hook F).39 Celastrol causes a hyperphosphorylation of heat shock factor 1 and increases the binding of heat shock factor 1 and other heat shock elements.47 As a consequence, celastrol increases Hspa1a/b gene expression via the heat shock factor 1 cascade.48 Therefore, it acts as an indirect HSP70 agonist. Celastrol was dissolved in 5% ethanol, 0.1% Cremophor EL (Sigma-Aldrich, Taufkirchen, Germany), and 94.4% NaCl to a final concentration of 0.1 mg/mL and a pH of 6.5. The celastrol solution was freshly prepared every day. Celastrol or vehicle (5% ethanol, 0.1% Cremophor EL and 94.4% NaCl) was injected once daily 6 h before the electrical kindling stimulation with a volume of 10 mL/kg body weight and a dosage of 1 mg/kg (i.p.). The pretreatment time of 6 h and the application dosage of 1 mg/kg were based on previous studies.26,47,49,50 As celastrol has negative effects on appetite,51 we carefully observed the weight of the animals. Animals with a continuous weight loss of H


Research Article

ACS Chemical Neuroscience general health condition (1 WT and 4 KO) or due to weight loss caused by celastrol injections (4 WT and 1 KO). These mice were excluded from all analysis. Experimental Details: Kindling Acquisition. Before the beginning of the kindling process, the initial afterdischarge threshold was evaluated twice following an ascending stair-step protocol (initial current of 8 μA, increase by 20% of the previous current, 1 min intervals) using an HSE Type 215E12 SV1 stimulator (Hugo Sachs Elektronik, Harvard Apparatus, March-Hugstetten, Germany) as previously described by von Rüden et al.32 First, we determined initial afterdischarge thresholds in naive mice. Afterward, mice were distributed to groups on the basis of stratified randomization (randomizer.org) considering their afterdischarge threshold to ensure equivalent mean afterdischarge thresholds for the two groups and a second determination was performed 3 days later with vehicle/ celastrol pretreatment. The afterdischarge threshold was defined as the current triggering electrographic seizure activity, with spikes being twice as high as the basal electroencephalogram and lasting at least 5 s. For daily kindling stimulations, the amygdala was electrically stimulated with 700 μA (1 ms, monophasic square-wave pulses, 50 Hz for 1 s). We decided to choose a stimulation current of 700 μA on the basis of the range of the initial afterdischarge seizure thresholds. In our laboratory we made the experience in several previous experiments that it makes no difference in the response to the first stimulation following threshold determination and in the kindling rate, if individualized stimulation strength (20% above initial threshold) or the suprathreshold stimulation with 700 μA is used. The only difference was that kindling with 700 μA resulted in a more stable fully kindled state. For every electrical stimulation the seizure severity was scored according to a slightly modified Racine scale:52 (1) immobility along with facial movements, (2) head nodding and more severe facial clonus with chewing movements, (3) unilateral forelimb clonus, (4) bilateral forelimb clonus with or without rearing, and (5) bilateral fore- and hindlimb clonus with or without rearing along with loss of balance. Scores 1 and 2 were considered as focal seizure activity, score 3 was the transition state from focal to generalized seizure activity, and scores 4 and 5 indicated generalized seizure activity. In addition, for each seizure event, we measured the seizure duration as visible motor seizure activity with a digital timer and the afterdischarge duration was recorded with Lab Chart 7 set selection. Moreover, we computed the cumulative afterdischarge duration (sum of all afterdischarge durations throughout the daily electrical stimulations). Following the kindling acquisition, we evaluated the postkindling afterdischarge threshold using the same ascending stair-step procedure as described above. First, we evaluated the impact of Hspa1a/b knockout in the kindling paradigm (kindled, vehicle-treated WT and KO), and second, we analyzed the effect of the Hspa1a/b agonist celastrol (kindled, vehicle- or celastrol-treated WT and KO). Tissue Preparation and Immunohistochemistry. Directly after postkindling afterdischarge threshold determination, animals were euthanized with pentobarbital (600 mg/kg, i.p., Narcoren, Merial GmbH, Halbergmoos, Germany) and transcardially perfused with 4% paraformaldehyde (PFA) in 0.1 M phosphate buffer (pH 7.4). Brains were postfixed for 24 h in 4% PFA, cryoprotected with a 30% sucrose solution, embedded in Tissue-Tek, and stored at −80 °C. Coronal sections (40 μm, six series) were cut on a cryostat (HM560M, Microm International GmbH, Walldorf, Germany) and stored at −80 °C in a cryoprotectant solution (glycerol and 0.1 M phosphate buffer, pH 7.4, 1/1 in volume). One of the series was stained with thionine acetate (9937, Carl Roth, Germany) to verify the correct position of the electrode in the amygdala. For evaluation of activated microglia, we performed an immunohistochemical staining of the ionized calcium binding adaptor molecule 1 (Iba1), a microglia-/macrophage-specific calcium-binding protein. For antigen retrieval, free-floating brain sections were exposed to sodium citrate buffer (10 mM, pH 6.0) in a water bath at 80 °C for 30 min. Endogenous peroxidase was quenched by

incubating the sections with 0.45% H2O2 for 60 min. Unspecific binding sites were blocked with 10% donkey serum and 2% bovine serum albumin in TBS. We then incubated the brain sections with a polyclonal rabbit anti-Iba1 primary antibody (1:10000, Wako Cat. No. 019-19471, RRID:AB_2665520, Osaka, Japan) overnight at 4 °C, with a biotinylated goat anti-rabbit secondary antibody (1:500 (Jackson ImmunoResearch Laboratories Cat. No. 111-065-003, RRID:AB_2337959), West Grove, USA) for 1 h at room temperature and with peroxidase-labeled streptavidin (1:1400, Vector Laboratories Cat. No. PK-4000, RRID:AB_2336818), USA) for 1 h at room temperature. Subsequently, the diaminobenzidine reaction was performed for 1 min. Sections were mounted, air-dried, and coverslipped with Entellan (Merck, Darmstadt, Germany). Negative controls omitting the primary antibody were prepared in parallel. Evaluation of Activated Microglia. An operator unaware of the animals’ genotype and treatment condition evaluated microglia activation as described by Hovens et al.44 using a semiautomatic image analysis method in both brain hemispheres in the cornu ammonis regions 1 and 3 (CA1 and CA3) and in the hilus of the hippocampal formation. Therefore, images of five brain sections (−1.22, −1.7, −2.3, −2.8, −3.4 mm caudal to bregma) were captured as ×400 magnification with an Olympus BH2 microscope with a single chip charge-coupled device (CCD) color camera (Axiocam; Zeiss, Göttingen, Germany), and an AMD Athlon 64 Processor based computer with an image capture interface card (Axiocam MR Interface Rev.A; Zeiss, Göttingen, Germany). In the CA1 and CA3 regions we analyzed two visual fields (1388 × 1040 pixels) and in the hilus one visual field (1388 × 1040 pixels) with ImageJ software (ImageJ 1.51 version, https://imagej.nih.gov/ij). The color deconvolution plugin (vector H DAB) described by Ruifrok and Johnston53 with selection of color 2 (brown) was applied for further analysis. We utilized the threshold function with the default mask (a variation of the IsoData algorithm54) to measure the total cell area, referring to the total Iba1 expression. To detect the total cell body area (radius >35 μm) a size filter (>150 pixels) was used and the threshold was lowered by 40 points. In the following, the ImageJ analyze particle function was applied to measure the total cell area and the cell body area of all microglia. As activated microglia cells transform from the nonactivated state (cells with a small cell body and long, starlike processes) into phagozytic cells with an enlarged cell body and fewer small cell processes,55 the different morphologies can be used to distinguish different activation states of microglia cells. On the basis of this fact, we analyzed microglial activation as described by Hovens et al.:44 microglia activation = (total cell area)/(cell body area) Thus, a high value of the ratio (total cell area)/(cell body area) reflects fewer activated cells (many cells with a small cell body), whereas a low value refers to strong microglia activation (many cells with a large cell body). Statistics. For the statistical analysis of group differences, we used GraphPad Prism 5.04 for Windows (GraphPad Prism Software, San Diego, CA, USA). We analyzed seizure severity, seizure duration, and the afterdischarge duration during the course of kindling by two-way ANOVA for repeated measurements and compared individual stimulation days with a Bonferroni multiple comparison test. Selected group comparisons of the bodyweight change were determined by a Mann−Whitney U test. We used paired Student’s t tests to compare pre- and postkindling parameters for individual groups. Activation of microglia was analyzed after log transformation of the raw data by unpaired Student’s t test for comparison of strain differences and twoway ANOVA followed by the Bonferroni posthoc correction for the analysis of treatment effects. We considered a p value

Genetic and Pharmacological Targeting of Heat Shock Protein 70 in

Recommend Documents