Background

Fluoxetine (FLX), a selective serotonin reuptake inhibitor (SSRI), is widely used totreat depressive disorder; however, the cellular mechanisms underlying theantidepressant effect of FLX remain unclear. Findings from animal studies suggestthat adult neurogenesis in the brain is critically involved in this process [1]. It has been reported that chronic FLX treatment for 2–4 weeksresults in increased neurogenesis and cell proliferation in the adult dentate gyrus(DG) [2–4], a response that has been linked to the behavioral effects of FLX [3]. Furthermore, we recently demonstrated that chronic FLX treatment leadsto the generation of cortical gamma-aminobutyric acidergic (GABAergic) interneuronsfrom neural progenitor cells in adult mice [5]. Conversely, we have shown that chronic FLX treatment for more than6 weeks decreases neurogenesis in the subventricular zone of adult mice [6].

Besides its effect on adult neurogenesis, chronic FLX treatments cause“dematuration,” a reversal of the established state of maturation ofadult dentate granule cells [6–9], raising the possibility that a distinct form of synaptic plasticityunderlies the antidepressant effect of FLX. Dentate granule cells in FLX-treatedadult mice exhibit similarity with immature granule cells in non-treated mice interms of expressions of maturation cell markers (e.g., a decrease in calbindinexpression and an increase in calretinin expression) and electrophysiologicalcharacteristics (e.g., reductions of basal synaptic transmission and frequencyfacilitation of the synapses between DG and CA3, and reinstatement of high membraneexcitability) [7]. A juvenile-like state of granule cells has also been observed in theadult brains of some genetically-engineered mice strains [10], such as alpha calcium/calmodulin-dependent protein kinase II(αCaMKII) heterozygous knockout (HKO) mice [11–17], schnurri-2 (Shn-2) KO mice [18], and mutated synaptosomal-associated protein 25 knock-in (SNAP-25 KI)mice [19]. Interestingly, FLX-induced dematuration of neurons has also beenobserved in GABAergic interneurons of the basolateral amygdala [20]; FLX converts interneurons, in particular parvalbumin-positive (+) cells,a subclass of interneurons, to a more immature state. However, it remains unclearwhether FLX has any effect on cellular dematuration of interneurons in other brainregions.

In this study, we used immunohistological analysis to investigate the effect of FLXtreatment on GABAergic interneurons of the medial frontal cortex (mFC) and reticularthalamic nucleus (RTN). Specifically, we assessed whether FLX treatment demonstrateda dematuration effect on GABAergic interneurons by examining the expression ofperineuronal net (PNN), a marker of neuronal maturation, as well as the expressionof parvalbumin, calretinin, and somatostatin, which are markers for specificGABAergic interneurons.

Results

Chronic FLX treatment decreased the number of parvalbumin+ cells in thefrontal cortex

PNNs are reticular structures composed of extracellular matrix molecules, such aschondroitin sulfate proteoglycans, hyaluronan, and tenascin-R, and are expressedin the central nervous system [21]. The temporal pattern of PNN formation reportedly corresponds to theending of the critical periods in which synaptogenesis, synaptic refinement, andmyelination occur [21], thus suggesting that their formation coincides with neuronalmaturation. For this reason, PNNs are considered a marker of neuronal maturation [20, 22, 23].

Using PNN and parvalbumin stained sections, we first examined whether chronic FLXtreatment altered the staining pattern of parvalbumin+ cells in the mFC ofadult mice. FLX solution was intraperitoneally injected into mice at15 mg · kg-1 · day-1for 3 weeks. We chose the mFC (Additional file 1:Figure S1), because it extensively overlaps with regions referred to as theanterior cingulate cortex [24] and because it plays an important role in rodent emotional memory andbehavior associated with limbic regions, such as the amygdala and hippocampus [25–27]. The number of parvalbumin+ cells in the mFC was significantlydecreased in FLX-treated mice compared to vehicle-treated mice(Figure 1; p = 0.0049). There was nosignificant difference in the number of total PNN+ cells between FLX- andvehicle-treated mice, although the number of PNN+ cells tended to decreaseby chronic FLX treatment (p = 0.072). Chronic FLX treatmentsignificantly decreased the number of parvalbumin+/PNN+ cells(p = 0.00024). Using these data, we calculated the percentage ofparvalbumin+/PNN+ cells from the total number of parvalbumin+ cellsand found that FLX treatment decreased the percentage ofparvalbumin+/PNN+ cells to approximately 80% of the vehicle-treatedvalue (p = 0.00033).

Figure 1
figure 1

Decrease in the number of parvalbumin+ cells in the mFC.(A) Representative coronal images ofparvalbumin+ (magenta)/PNN+ (green) cells in the mFC of micetreated with vehicle (upper row) or FLX (lower row). Mice received FLXfor 3 weeks at 15 mg/kg/day. (B) Quantification of thenumber of parvalbumin+, PNN+, parvalbumin+/PNN+ cells, and theproportion of parvalbumin+/PNN+ cells in the total number ofparvalbumin+ cells (n = 4 mice each; 11-week-old).FLX, fluoxetine; PNN, perineuronal net; PV, parvalbumin.

Full size image

We also investigated whether FLX treatment affected the number ofparvalbumin+ cells in the RTN. In the RTN, in which almost all neurons areGABAergic, FLX treatment had no effect on the numbers ofparvalbumin+ (p = 0.54) and PNN+ cells(p = 0.36) (Figure 2C and D). However, wefound a significantly lower number of parvalbumin+ cells in thehippocampal CA3 region of FLX-treated mice compared to vehicle-treated mice(Figure 2A and B; p = 0.00063). Thenumber of PNN+ cells was also significantly decreased in the hippocampalCA3 region (Figure 2A and B; p = 0.040)but not in the RTN. FLX treatment also reduced the number ofparvalbumin+ cells in the basolateral amygdala (Additional file 1: Figure S2; p = 0.0058). Our results fromthe hippocampal CA3 region and basolateral amygdala corroborate previousfindings [20].

Figure 2
figure 2

Expression of parvalbumin and PNN in the hippocampal CA3 and RTN. (A,C) Double-staining for parvalbumin (magenta) and PNN (green) incoronal sections of the hippocampal CA3 region (A) and RTN(C). Mice received FLX for 3 weeks at 15 mg/kg/day.(B, D) The number of parvalbumin+, PNN+,parvalbumin+/PNN+ cells, and the proportion ofparvalbumin+/PNN+ cells in the total number ofparvalbumin+ cells were quantified in the hippocampal CA3 region(B) and RTN (D) (n = 4 mice each;11-week-old). Note that a decreased number of parvalbumin+ cellswere found in the hippocampal CA3, but not in the RTN. CA, cornuammonis; FLX, fluoxetine; PNN, perineuronal net; PV, parvalbumin; RTN,reticular thalamic nucleus.

Full size image

Chronic FLX treatment did not alter the numbers of calretinin+ andsomatostatin+ cells

Three main subgroups of GABAergic interneurons are found in the FC of adultrodents: parvalbumin+ cells, calretinin+ cells, andsomatostatin+ cells [28]. Thus, we next examined whether chronic FLX treatment decreased theexpression of calretinin and somatostatin in the mFC, RTN, hippocampus, andbasolateral amygdala. Chronic FLX treatment had no effect on the number ofcalretinin+ and somatostatin+ cells in these brain regions(Additional file 1: Figure S3 and Table S1). Inaddition, the calretinin+ and somatostatin+ cells in the mFC, RTN,hippocampus, and basolateral amygdala were not surrounded by PNNs (Additionalfile 1: Figure S3 and Table S1), which is consistentwith previous findings [29, 30]. Although calbindin was abundantly expressed in the RTN, FLXtreatment did not alter its expression (Additional file 1: Figure S3 and Table S1). This result suggests that FLXtreatment mainly affected parvalbumin+ interneurons, and not calretininand somatostatin, in the mFC, hippocampus, and basolateral amygdala. It alsohighlights that compared to calretinin+ and somatostatin+ cells,parvalbumin+ cells are more likely to be surrounded by PNNs.

Chronic FLX treatment did not alter the numbers of GABA+ cells andapoptotic cells

Although we had observed a significant decrease in the number ofparvalbumin+ cells in the mFC and hippocampus of FLX-treated mice, itremained unclear whether this decrease was reflective ofparvalbumin+ interneuron cell death or a decrease in the expression ofparvalbumin proteins in each interneuron. Thus, we performed immunofluorescencestaining for GABA to examine the effect of FLX treatment on the total numbers ofGABAergic interneurons in these regions. We found no difference in the number ofGABA+ cells in the mFC (Figure 3A and B;p = 0.82) and hippocampus (Figure 3C andD; p = 0.81) of FLX-treated mice compared to vehicle-treated mice.Furthermore, the fluorescence intensity of GABA in the mFC and hippocampus ofFLX-treated mice was similar to that in these regions of vehicle-treated mice(Figure 3B and D; mFC, p = 0.57;Hippocampal CA3, p = 0.53). We subsequently examined whether thenumber of apoptotic cells changed after FLX treatments. Terminaldeoxynucleotidyl transferase-mediated dUTP nick-end labeling (TUNEL) analysisrevealed that FLX treatments did not induce apoptotic cell death in the mFC andhippocampus (Additional file 1: Figures S4). Inaddition, we labeled the interneurons with 5-bromodeoxyuridine (BrdU), a markerof DNA synthesis, by intraperitoneal injection of BrdU into timed-pregnant miceevery 24 h from day 14 to day 20 of gestation. Thus, the interneuronsgenerated during the embryonic period contained BrdU. FLX treatment wascommenced 8 weeks after birth and continued for 3 weeks. FLX treatmentdid not alter the number of BrdU+ cells in the mFC or hippocampus;however, it significantly reduced the number of parvalbumin+/BrdU+ cells(Additional file 1: Figures S5 and S6). This resultsuggests that, in the cells generated during embryogenesis, parvalbumin proteinlevels in the mFC and hippocampus were reduced after FLX treatment. Takentogether, these findings suggest that the decreased number ofparvalbumin+ cells in the mFC and hippocampus may reflect a decreasedexpression of parvalbumin proteins in each interneuron.

Figure 3
figure 3

FLX treatment did not change the number of GABA+ cells in themFC and hippocampus. (A, C) Representative coronal images ofGABA+ cells in the mFC (A) and hippocampus (C) ofvehicle-treated (upper) and FLX-treated mice (lower). Mice received FLXfor 3 weeks at 15 mg/kg/day. (B, D) The number ofGABA+ cells (left graphs) and the fluorescence intensity of GABA(right graphs) were quantified in the mFC (B) and hippocampal CA3region (D) of vehicle-treated and FLX-treated mice(n = 4 mice each; 11-week-old). CA, cornu ammonis; FLX,fluoxetine; mFC, medial frontal cortex.

Full size image

Discussion

In this study, we examined the effect of chronic FLX treatment on PNN expression aswell as the expression of parvalbumin+, calretinin+, and somatostatin+ cellsin the mFC, hippocampus, basolateral amygdala, and RTN. Immunofluorescence analysisrevealed that FLX treatment decreases the number of parvalbumin+ cells, butnot that of calretinin+ or somatostatin+ cells, in the mFC, hippocampus,and basolateral amygdala. Our findings suggest that the decreased number ofparvalbumin+ cells reflects a decrease in parvalbumin protein expression ineach interneuron, and not apoptotic cell death of parvalbumin+ cells or adecrease in the total number of GABA+ cells. Furthermore, the percentage ofparvalbumin+/PNN+ cells was also decreased in the mFC. These findings suggestthat FLX treatment may have a dematuration effect on fast-spiking inhibitoryinterneurons in the mFC and hippocampus, which are immunoreactive for parvalbuminduring the mature status [31]. This pseudo-immature state of parvalbumin+ cells may account forthe antidepressant effect of FLX, in addition to, or as an alternative to,dematuration in the DG and amygdaloid neurons and increased neurogenesis in the DGand cortex.

Decreased percentage of PNN+ cells out of parvalbumin+ cells in the mFC

In the present study, we found a decrease in the number of PNN+ cells inthe hippocampal CA3 region and a decrease in the proportion ofPNN+/parvalbumin+ cells in the total number of parvalbumin+ cells inthe mFC, basolateral amygdala, and hippocampal CA3 region. It has previouslybeen demonstrated that chronic FLX treatment decreases the percentage ofPNN+/parvalbumin+ cells (from the total number ofparvalbumin+ cells) in the basolateral amygdala and hippocampal CA1 [20]; however, to our knowledge, such a finding has not been observed inthe FC until now. In contrast to our study, a previous study reported nodisruption to PNNs in the FC [20]. This difference may be attributable to the differences in FLXadministration, such as the dose of FLX(10 mg · kg-1 ∙ day-1 inthe previous study vs. 15 mg ∙ kg-1 ∙day-1 in this study) and administration method (drinking water inthe previous study vs. intraperitoneal injection in this study). Thisdiscrepancy in findings for PNN expression in the mFC should be examined infuture studies to determine whether it is a dose-dependent effect of FLX.

The formation of PNNs coincides with neuronal maturation in the central nervoussystem [20, 22, 23]. Thus, similar to other studies, we used PNN as a marker of neuronalmaturation. We found a significant decrease in PNN+ cells in thehippocampal CA3 region following FLX treatment; however, FLX treatment had lesseffect on the total number of PNN+ cells in the mFC. This is probablybecause the reduction of PNN happens specifically in the parvalbumin+ cells inthe mFC. Also, it may be due to the long life of PNN components. It haspreviously been demonstrated that the immunoreactivity of PNN components,tenascin and chondroitin sulphate proteoglycans, persists in vivo forat least up to 4 weeks and 14 months, respectively [32]. In this study, we used Wisteria floribunda lectin (WFA) todetect PNNs. WFA binds carbohydrate structures terminating inN-acetylgalactosamine linked to galactose, which are contained inchondroitin sulphate proteoglycans of PNNs [32]. Thus, this suggests that dematurated interneurons are stillsurrounded by PNNs or chondroitin sulphate proteoglycans after the disappearanceof parvalbumin proteins in each interneuron. Consequently, fast-spiking cells,in which parvalbumin proteins are diminished by FLX treatment, can be detectedby WFA.

No significant differences were observed in the numbers of parvalbumin+ orPNN+ cells in the RTN between vehicle- and FLX-treated mice. Currently, itis not clear why these numbers did not change after FLX treatment. However,possibly, this could be attributable to the origins of the GABAergicinterneurons present in the regions. Almost all the cortical, hippocampal, andamygdaloid GABAergic interneurons are derived from the ventricular zone of themedial and caudal ganglionic eminences [33, 34], while thalamic GABAergic interneurons originate from the ventricularzone of the third ventricles [35]. These results suggest the possibility that FLX treatment mightspecifically reduce parvalbumin protein levels in interneurons derived from theganglionic eminences. It will be interesting to examine this, as well as otherpossibilities, in future research.

Decrease in the number of parvalbumin+ cells in the mFC andhippocampus

In this study, we demonstrated that chronic FLX treatment did not induceapoptotic cell death of parvalbumin+ interneurons in the mFC andhippocampus. We also showed that the number of GABA+ cells in the mFC andhippocampus was not altered by FLX treatment. These findings suggest that thedecreased number of parvalbumin+ cells reflects a decrease in expressionof parvalbumin proteins in each cell.

Three main subgroups of GABAergic interneurons are found in the FC of adultrodents: parvalbumin+, calretinin+, and somatostatin+ cells [28]. On the other hand, based on the firing patterns for depolarization,cortical GABAergic interneurons are divided into three subgroups: fast-spiking,late-spiking, and regular-spiking/burst-spiking non-pyramidal cells [36]. It is widely accepted that almost all fast-spiking interneuronsexpress parvalbumin, whereas interneurons with the other types of spikingpatterns have calretinin and somatostatin [36]. In rodents, almost all fast-spiking inhibitory GABAergicinterneurons in both cortex [37] and hippocampus [38] are generated during the embryonic period, while the firstparvalbumin proteins appear at postnatal day 10 in the mouse cortex [39] and at postnatal day 7 in the hippocampus [40]. Moreover, using transcriptional and electrophysiological analyses ofa GFP knock-in mice, in which almost all fast spiking inhibitory interneuronsexpress GFP, fast-spiking inhibitory interneurons have been reported to maturebetween P10 and P25 [31, 41]. Thus, immature fast-spiking inhibitory interneurons in the cortex donot express parvalbumin mRNA. The present result suggests that FLX treatment mayconvert mature parvalbumin+ interneurons to a pseudo-immature state, i.e.,FLX treatment may cause “dematuration” ofparvalbumin+ interneurons (Figure 4).

Figure 4
figure 4

Dematuration of cortical fast-spiking inhibitory interneurons by FLXtreatments. Based on parvalbumin expression, the state of thedematurated fast-spiking inhibitory interneurons in the mFC ofFLX-treated mice were similar to that of fast-spiking inhibitoryinterneurons of non-treated mice between postnatal days 0 and 10(indicated in yellow). E, embryonic day; P, postnatal day.

Full size image

Implication of FLX-induced neuronal dematuration in the mFC, hippocampus, andamygdala

In adult mice, FLX treatment converts differentiated DG neurons to a moreimmature state [6–9]. Similar changes in DG neurons have been demonstrated in αCaMKIIHKO [11], Shn-2 KO [18], and SNAP-25 KI mice [19]; this phenomenon has been termed the “immature DG” [10]. In this study, we demonstrate for the first time that FLX treatmentmight also induce dematuration of parvalbumin+ interneurons in the mFC andhippocampal CA3 region, while our finding in the amygdala is consistent withthat of a previous study [20]. Consistently, the present study demonstrates that chronic FLXtreatment increases the expression of polysialic acid-neural cell adhesionmolecule (PSA-NCAM), which is a marker for immature neurons and a regulator ofneural plasticity [42, 43], in the mFC, hippocampus, and amygdala; this is in agreement withprevious findings [44, 45] (Additional file 1: Figure S7). Thissuggests that dematuration of parvalbumin+ interneurons is induced by FLXtreatment in the mFC, hippocampus, and amygdala, where neural plasticity mightbe enhanced by FLX treatment. In line with this, FLX treatment has been reportedto reinstate neural plasticity and promote the electrophysiological andbehavioral recovery of functions in the visual cortex of adult amblyopic rats [46]. In contrast, accelerated maturation of parvalbumin+ cells viaoverexpression of the neurotrophin brain-derived neurotrophic factor leads to areduced capacity for cortical neural plasticity [47, 48]. Thus, dematuration of parvalbumin+ interneurons in the mFC,hippocampus, and amygdala might reinstate synaptic plasticity that is reducedwith age and development, thereby potentially causing the antidepressant effectof FLX. Further studies are required to address the causal relationship betweendematuration of parvalbumin+ cells and enhanced neural plasticity.

Recent findings have led to the hypothesis that problems in informationprocessing within neural networks, rather than altered chemical balance, mayaccount for the mechanism underlying depression [49, 50]. Thus, antidepressant drugs may induce changes in neuronal morphologyand connectivity, gradually improving neuronal information processing andrecovering mood. Indeed, volume changes in the hippocampus, mPFC, or amygdalaare found both in patients with depression and in animal models of depression [51, 52]. Previous studies, as well as the present one, have identified someof the effects of FLX on the brain, which include increased adult neurogenesisin the DG [2] and cortex [5], decreased adult neurogenesis in the SVZ [6], dematuration of neurons in the DG [7], amygdala [20], and mFC. Most events occur in the FC and limbic system.Interestingly, it has become increasingly clear that network dysfunction in thePFC and limbic system, including the hippocampus and amygdala, is involved inthe pathophysiology of depressive disorder [24, 53, 54]. Therefore, neuronal dematuration and adult neurogenesis in theseregions may play important roles in the mechanism of action of antidepressantdrugs like FLX. In addition, some of the adverse effects of FLX [55], such as aggression, violence, and psychosis, might be mediated bythe dematuration of fast-spiking inhibitory interneurons in the mFC. Aggressionand violence are associated with deficits in the prefrontal cortex of humans [56, 57], where activation of GABAergic interneurons decreases [57]. Dematuration of fast-spiking inhibitory interneurons might decreaseinhibitory transmission of the interneurons, which in turn could evokeaggression and violence. It should be noted that, in post-mortem brains ofpatients with schizophrenia, the number of parvalbumin+ interneurons [41, 58] and PNN+ cells [59] is decreased in the prefrontal cortex. This dematuration ofparvalbumin+ fast-spiking interneurons by FLX treatment may be related tothe antidepressant-induced psychosis and agression observed in clinical settings [55, 60]. Future studies will need to address the behavioral significance ofthe FLX-induced dematuration effect on fast-spiking inhibitory interneurons inthe mFC.

Conclusions

The present study demonstrates that chronic FLX treatment reduces parvalbuminproteins and PNNs in GABAergic interneurons in the mFC, suggesting that FLX inducesjuvenile-like state of fast-spiking inhibitory interneurons in mFC. This effect ofFLX on parvalbumin+ cells in the mFC might account for the therapeuticmechanism of this antidepressant drug and/or some of its adverse effects.

Methods

Animals and antidepressant treatment

Adult male C57BL/6 J mice (Charles River Laboratories Japan, Yokohama,Japan), that were 2 months old at the start of our experiments, were used.All animal experiments were approved by the Institutional Animal Care and UseCommittee of Fujita Health University, based on the Law for the Humane Treatmentand Management of Animals (2005) and the Standards Relating to the Care andManagement of Laboratory Animals and Relief of Pain (2006). Every effort wasmade to minimize the number of animals used.

Treatment with FLX (LKT Laboratories, St. Paul, MN) was performed as previouslydescribed [5]. Briefly, FLX solution was intraperitoneally injected into micebetween 10:00–11:00 a.m. every day for 3 weeks. The appropriateFLX concentration (15 mg ∙ kg-1 ∙ day-1)was determined for each body weight before injection. Mice were fixed at6 h after the last injection of FLX. Chronic FLX treatment at thisconcentration remarkably decreased the expression of calbindin in the DG(Additional file 1: Figure S8;p = 0.00046), as previously reported [6, 7].

BrdU labeling

BrdU injection was performed as previously described [14]. Briefly, the BrdU (Sigma-Aldrich, St. Louis, MO) stock solution wasprepared in distilled water with 0.007 N NaOH at 20 mg/ml and storedat −20°C until use. In order to label GABAergic interneurons of theembryonic cerebral cortex with BrdU, timed-pregnant mice were intraperitoneallyinjected with BrdU solution (100 mg/kg body weight) dissolved in phosphatebuffered saline (PBS) every 24 h from day 14 to day 20 of gestation. Afterbirth, the mice were bred for 2 months before subsequently receiving FLXinjections for 3 weeks at a concentration at 15 mg ∙kg-1 ∙ day-1. The mice were deeply anesthetizedand transcardially perfused with 4% paraformaldehyde in 0.1 M phosphatebuffer (PB), pH 7.4. For staining BrdU staining, sections were pretreatedwith HCl as previously described [14].

Immunohistological analysis

Fixation and immunofluorescence staining were performed as previously described [5]. Briefly, mice were deeply anesthetized with chloral hydrate(245 mg/kg, intraperitoneally) and transcardially perfused with 4%paraformaldehyde in 0.1 M PB. The brains were dissected, immersedovernight in the same fixative, and transferred to 30% sucrose in PBS for atleast 3 days for cryoprotection. All brain samples were mounted inTissue-Tek (Miles, Elkhart, IN), frozen, and cut coronally into 50-μm-thickcoronal sections, using a microtome (CM1850, Leica Microsystems, Wetzlar,Germany). Sections were stored in PBS containing sodium azide (0.05%, w/v)at 4°C until use. After washing in PBS for 1 h, they were preincubatedwith PBS-DB (4% normal donkey serum [Vector Laboratories, Burlingame, CA]and 1% BSA in PBS) for 2 h at room temperature. The sections wereincubated at 4°C for 48 h or at room temperature overnight with theindicated primary antibodies. After washing in PBS for 1 h, the sectionswere incubated at room temperature for 1 h with secondary antibodies. ForPNN staining, the sections were incubated with biotinylated WFA (1:200;Sigma-Aldrich) at 4°C for 48 h or at room temperature overnight. Afterwashing in PBS for 1 h, the sections were incubated with Alexa Fluor 488conjugated to streptavidin (10 μg/ml; Life technologies, Carlsbad, CA)for 1 h at room temperature. After washing in PBS containing Hoechst 33258for nuclear counterstaining for 1 h, the sections were mounted on glassslides coated with 3-aminopropyltriethoxysilane and embedded with Permafluor(Thermo Scientific, Pittsburgh, PA). Confocal laser-scanning microscopy (LSM700; Carl Zeiss, Oberkochen, Germany) was used to obtain images of the stainedsections.

Antibodies

The following primary antibodies were used: mouse monoclonal antibodies forcalbindin (1:2000, Sigma-Aldrich), parvalbumin (1:2000, Sigma-Aldrich), PSA-NCAM(clone 2-2B mouse IgM; 1:200, Millipore, Billerica, MA); rat monoclonal antibodyfor BrdU (1:100, Abcam, Cambridge, MA); and rabbit polyclonal antibodies forcalretinin (1:500, Life technologies), GABA (1:1000, Sigma-Aldrich), andsomatostatin (1:1000, Bachem, Bubendorf, Switzerland). The following secondaryantibodies were used: goat anti-mouse IgG Alexa Fluor 488 (1:200, LifeTechnologies), goat anti-mouse IgG Alexa Fluor 594 (1:200, Life Technologies),goat anti-mouse IgM Cy3 (1:200, Millipore), goat anti-rabbit IgG Alexa Fluor 594(1:200, Life Technologies), and goat anti-rat IgG Alexa Fluor 594 (1:200, LifeTechnologies).

TUNEL staining

TUNEL staining was performed according to the manufacturer’s instructions(in situ cell death detection kit, Roche, Mannheim, Germany).

Ischemia treatment

Global ischemia was induced as previously described [61]. Briefly, after anesthesia, both common carotid arteries weretransiently occluded with clamps for 10 min. Control animals were treatedidentically, except for the occlusion of common carotid arteries. The mice wereallowed to survive for 2 days after ischemia and were then killed byperfusion.

Quantification of labeled cells

The mFC region was determined according to the mouse brain atlas [62]. Quantification analysis was performed as previously reported [5]. Briefly, analysis was performed using a confocal microscope equippedwith a 40 × objective lens (Plan-NEOFLUAR, NA = 0.75, CarlZeiss) and a pinhole setting that corresponded to a focal plane thickness ofless than 1 μm. To exclude false-positives due to the overlay ofsignals from different cells, randomly selected areas were analyzed by movingthrough the entire z-axis of each area. Cells were counted under the live modeof confocal scanning. For quantifying the fluorescence intensity ofimmunostained images, we used the ImageJ software. The region of interest of theacquired images was traced, and optical densities were obtained from at leastthree sections per mouse. Background intensity was subtracted using nonstainedportions of each section. Slides were coded and quantified by a blindedindependent observer. Data were analyzed by one-way ANOVA. The error bars in thefigures represent SEM.