Introduction

Minocycline is a second-generation tetracycline antibiotic that inhibits microglial activation and elicits pleiotropic effects, including anti-inflammatory, antioxidant and anti-apoptotic actions in several neurodegenerative disorders [1]. Moreover, minocycline is effective as add-on therapy in schizophrenia [2]. The mechanism of action of minocycline in schizophrenia is still unknown; the inhibition of over-activated microglia occurring as result of stress or pathogen-associated molecular patterns has been hypothesized as possible therapeutic mechanism [1]. In addition, the interaction with glutamate receptors may be important: minocycline attenuated hyperlocomotion and sensorimotor gating deficits (correlates of positive symptoms of schizophrenia) and MK-801-triggered hyperdopaminergia [3]. The relation with glutamatergic pathways appears complex, since minocycline counteracts neurotoxicity induced by glutamate/NMDA via inhibition of microglia activation [4].

Psychosis triggered by NMDA receptor antagonists is accompanied by severe neurotoxicity in the retrosplenial cortex of adult rats [5] and widespread apoptosis in the early postnatal brain [6]. The latter is followed by protracted schizophrenia-like behavioral abnormalities with onset at peripubertal stages, NMDAR antagonist treatment at/around postnatal day 7 (P7) representing, meanwhile, a widely employed developmental model of schizophrenia [7].

Here we aimed to investigate the potential neuroprotective effect of minocycline on MK-801-induced apoptosis as assessed by cleaved caspase-3 expression at P7. To our knowledge, this was not examined previously and would be important for the clinical profile and neurodevelopmental effects of this potential new antipsychotic drug.

Materials and methods

Animals and drug treatment

Experiments were performed using C57BL6/N mice purchased from Charles River as breeding pairs (Sulzfeld, Germany). Animals were supplied with food and water ad libitum. For determination of caspase-3 expression pattern in young animals, 7-day-old mice (n = 6 for each drug) were treated with either: (1) vehicle (0.9 % NaCl, 5 ml/kg i.p.), (2) minocycline (40 mg/kg i.p.), (3) MK-801 (0.5 mg/kg i.p.) or (4) minocycline (40 mg/kg i.p.) 30 min before MK-801 (0.5 mg/kg i.p.), and killed 8 h later. Brains were removed, postfixed in the same fixative for 24 h and cut into 50 µm coronal sections, as described previously [8]. All experiments were approved by the German Committee on Animal Care and Use and were carried out in accordance with the local Animal Welfare Act and the European Communities Council Directive of November 24, 1986 (86/609/EEC).

Immunohistochemistry and cell counting

Free-floating sections were incubated with the anti-caspase 3 (polyclonal rabbit; Cell Signaling, Beverly, MA, USA) (diluted 1:1000), and the staining was visualized using Nickel-3,3′-diaminobenzidine (DAB), as described previously [9–11]. Cell counting was performed blind to the treatment using a light microscope (LEICA TCS-NT). Quantification of caspase-3-immunoreactive cells was performed at 40× magnification in regions with high expression in the brain: cingulate and retrosplenial cortex, hippocampus, nucleus accumbens and thalamus, according to standard description of each region in the developing brain [12]. For each animal, cell counts in 3–4 sections were performed bilaterally. The average value across all sections for each animal was determined. Afterwards, the average density for each treatment group was calculated and compared statistically. For all experiments, the number of caspase-3-expressing cells represents the mean ± SEM of 6 animals per treatment.

Statistical analysis

Statistical analyses were performed using the statistical program SPSS 23 for Windows. In all experiments, the mean number of caspase-3-expressing cells (±SE) was estimated for each treatment group, and differences between groups were determined by one-way ANOVA followed by Bonferroni post hoc tests, with p < 0.05 considered statistically significant.

Results

We found that, surprisingly, treatment with minocycline alone already triggered significant and extensive caspase-3 expression in numerous brain regions. Significant caspase-3 expression following treatment with MK-801, minocycline, or their combination was visible especially in the cingulate cortex (Fig. 1a–c), subiculum of the hippocampus (Fig. 1d–f), nucleus accumbens (Fig. 1g–i) and retrosplenial cortex (Fig. 1j–l). Interestingly, a strong caspase-3 staining was noticed in layer I of the motor cortex, representing putative transient Cajal-Retzius cells (Fig. 1j–l). Such caspase-3-expressing cells were not observed in the retrosplenial cortex and were therefore not quantified. In addition, a high number of caspase-3-positive cells were found in all treatment designs in the lateral thalamus (data not shown). The quantitative analysis (Fig. 2) revealed similar levels of caspase-3 induction in some regions, such as the cingulate cortex and nucleus accumbens following MK-801 versus minocycline treatment (Fig. 2a). In the retrosplenial cortex, however, MK-801 induced a significantly higher caspase-3 expression than minocycline alone, which did not increase significantly in addition the level of caspase-3 expression in combination with MK-801 (Fig. 2a). Interestingly, minocycline induced a strong caspase-3 expression outnumbering that triggered by MK-801 in the subiculum (Fig. 2b), whereas in other hippocampal areas the level of caspase-3 induction was similar after MK-801 compared to minocycline treatment (Fig. 2b). In contrast to the robust expression triggered by all drugs used, only few caspase-3-expressing neurons were quantified in saline-treated pups (Fig. 2).

Fig. 1
figure 1

Caspase-3 induction in various brain areas by treatment with minocycline, MK-801 or their combination at P7: the cingulate cortex (a–c), subiculum (d–f), nucleus accumbens (g–i) and retrosplenial cortex (j–l). Note the intense caspase-3 expression in layer I of the retrosplenial and parietal cortices (j–l). Mino minocycline, ac anterior commissure, cc corpus callosum, rsp retrosplenial cortex, mo motor cortex

Full size image
Fig. 2
figure 2

Quantitative analysis of the caspase-3 expression at P7 in the cingulate and retrosplenial cortex, as well as in the nucleus accumbens (a) and in the CA1, CA3, dentate gyrus and subicular regions of the hippocampus (b). Statistical differences determined by one-way ANOVA followed by Bonferroni post hoc tests: *p < 0.001 versus Mino + MK-801, **p < 0.001 versus Mino, § p < 0.001 versus MK-801, ns not significant). All groups were highly significant compared to saline (not depicted). Mino minocycline, Retrosp. Cx. retrosplenial cortex, Cingulate Cx. cingulate cortex

Full size image

Discussion

Here we show that minocycline triggers widespread caspase-3 expression in the early postnatal mouse brain, in a pattern that resembled but was not identical to that exhibited by NMDAR antagonists. Caspase-3 (apopain) is a protease activated by the upstream caspase-8 and caspase-9 that plays a crucial role in the initiation of the “death cascade” in mammals [13]. Immunohistochemical analysis of caspase-3 expression represents a commonly used method to quantify apoptosis also under physiological conditions in short-lived types of cells [14]. The validity of caspase-3 mapping as marker of apoptosis was demonstrated also in the P7 neuronal apoptosis paradigm: The pattern of caspase-3 immunoreactivity closely resembled the pattern of argyrophilic neurodegeneration and co-localized with TUNEL (terminal deoxynucleotidyl transferase-mediated dUTP nick-end labeling)-positive neurons following MK-801 administration at P7 [15]. Moreover, caspase-3 activation is indispensable for the induction of cortical apoptosis by NMDAR blockade at P7 [16].

Surprisingly and against the hypothesis to represent an antipsychotic drug, minocycline did not reduce apoptosis triggered by NMDAR antagonist treatment; instead, it even exacerbated cell death induced by MK-801. These results appear intriguing, considering the neuroprotective effect of minocycline in several pathological situations, especially since it inhibited caspase-3 upregulation in an (adult) animal model of Huntington disease [17]. However, at early postnatal stages, minocycline was reported to worsen neuronal loss following hypoxic-ischemic brain injury [18] or deprivation of thalamocortical neurons [19] and to induce cell death in the somatosensory cortex [20].

Therefore, one possible explanation for these opposite effects of minocycline on neuronal survival could be that its pro-apoptotic action is restricted to early postnatal stages. The first postnatal week represents in rodents a developmental period, during which the brain is particularly vulnerable [21], also to the excitotoxic effect of NMDA [22]. This may reflect the dual role of NMDAR in regulating cell death. NMDAR-induced responses were postulated to depend on the receptor location: Stimulation of synaptic NR2A (GluN2A)-containing NMDAR, acting primarily through nuclear Ca2+ signaling, confers neuroprotective properties, whereas stimulation of extrasynaptic NR2B (GluN2B)-containing NMDAR promotes cell death [23]. However, at P7, the pro-apoptotic effect of NMDAR antagonists is mediated by the GluN2A [24] and not by GluN2B subunit [8]. The contrasting effects on cell death mediated via NMDAR were shown to depend on the activation of different intracellular signaling cascades [23]. Therefore, it is possible that minocycline interacts differentially with these (or similar) pathways at perinatal versus adult stages, resulting in different/opposite effects on cell death. Interestingly, we found in several brain regions (cingulate cortex, nucleus accumbens, CA1, CA3 and dentate gyrus of the hippocampus) similar levels of apoptosis following MK-801 and minocycline treatment; in these areas, the combined treatment of both substances appears to result in a “cumulative” effect (Fig. 2). However, in other areas, such as the retrosplenial cortex or the subiculum, the expression levels induced by each substance differed significantly and the effect of the combined treatment reflected rather the contribution of one substance than a cumulative effect of both drugs. Therefore, the pro-apoptotic effect of minocycline may be, at least in some brain areas, the result of NMDAR-independent pathway. A possibility could be that it results from the interaction of minocycline with other glutamate receptors, like AMPA receptors (AMPAR) [25]. However, unlike NMDAR antagonism, blockade of AMPAR or of other glutamate receptors like mGluR5 did not induce any cortical apoptosis at P7 [6, 26]. Another explanation could be that a yet unknown stage-specific effect on microglia may underlie the pro-apoptotic action of minocycline described here. Additionally, we cannot exclude species-specific effects, i.e., selective high vulnerability of perinatal C57Bl/6 mice to the action of minocycline.

The results presented here—although obtained in an animal model—may have also implications regarding the clinical use of minocycline during early development/pregnancy. The same deleterious effects as reported for NMDAR antagonists were shown for several anesthetics, most likely due to their action on NMDAR [27]. This raised concerns regarding neurodevelopmental risks of the use of these substances for pediatric anesthesia. Finally, it should be noticed that the developmental pro-apoptotic effect found here appears to distinguish minocycline from other substances with antipsychotic profile: The classical neuroleptic drug haloperidol did not induce any apoptosis in the rodent brain at P7 [6]. Moreover, pre-treatment with olanzapine significantly reduced the morphological and long-term behavioral abnormalities induced by perinatal NMDAR blockade [28].

The molecular mechanisms underlying the pro-apoptotic effect of NMDAR antagonists and minocycline, as well as possible interactions between them, need to be unraveled by future studies. It would be important to find out, if minocycline induces, as NMDAR antagonists, neurotoxicity also in the adult brain and if early postnatal neurotoxic effects are accompanied by long-term deficits with relevance for neuropsychiatric disorders to assure a safe clinical use.