Long-term accumulation of microglia with proneurogenic phenotype concomitant with persistent neurogenesis in adult subventricular zone after stroke

Authors

  • Pär Thored,

    1. Laboratory of Neurogenesis and Cell Therapy, Section of Restorative Neurology, Wallenberg Neuroscience Center A11, University Hospital, Lund, Sweden
    2. Lund Strategic Research Center for Stem Cell Biology and Cell Therapy, Lund, Sweden
    Search for more papers by this author
    • Pär Thored, Ursula Heldmann, and Walace Gomes-Leal shared first authorship.

  • Ursula Heldmann,

    1. Laboratory of Neurogenesis and Cell Therapy, Section of Restorative Neurology, Wallenberg Neuroscience Center A11, University Hospital, Lund, Sweden
    2. Lund Strategic Research Center for Stem Cell Biology and Cell Therapy, Lund, Sweden
    Search for more papers by this author
    • Pär Thored, Ursula Heldmann, and Walace Gomes-Leal shared first authorship.

  • Walace Gomes-Leal,

    1. Laboratory of Neurogenesis and Cell Therapy, Section of Restorative Neurology, Wallenberg Neuroscience Center A11, University Hospital, Lund, Sweden
    2. Lund Strategic Research Center for Stem Cell Biology and Cell Therapy, Lund, Sweden
    3. Laboratory of Experimental Neuroprotection and Neuroregeneration, Department of Morphology, Biological Sciences Center, Federal University of Pará, Belém-PA, Brazil
    Search for more papers by this author
    • Pär Thored, Ursula Heldmann, and Walace Gomes-Leal shared first authorship.

  • Ramiro Gisler,

    1. Laboratory of Neurogenesis and Cell Therapy, Section of Restorative Neurology, Wallenberg Neuroscience Center A11, University Hospital, Lund, Sweden
    2. Lund Strategic Research Center for Stem Cell Biology and Cell Therapy, Lund, Sweden
    Search for more papers by this author
  • Vladimer Darsalia,

    1. Lund Strategic Research Center for Stem Cell Biology and Cell Therapy, Lund, Sweden
    2. Neuronal Survival Unit, Wallenberg Neuroscience Center A10, University Hospital, Lund, Sweden
    Search for more papers by this author
  • Jalal Taneera,

    1. Department for Diabetes and Endocrinology, Clinical Research Centre, Malmö, Sweden
    Search for more papers by this author
  • Jens M. Nygren,

    1. Lund Strategic Research Center for Stem Cell Biology and Cell Therapy, Lund, Sweden
    2. Hematopoietic Stem Cell Laboratory, Lund University, BMC B10, Lund, Sweden
    3. Division of Immunology, Department of Medical Experimental Science, Lund University, BMC D14, Lund, Sweden
    Search for more papers by this author
  • Sten-Eirik W. Jacobsen,

    1. Lund Strategic Research Center for Stem Cell Biology and Cell Therapy, Lund, Sweden
    2. Hematopoietic Stem Cell Laboratory, Lund University, BMC B10, Lund, Sweden
    3. Hematopoietic Stem Cell Laboratory, The Weatherall Institute of Molecular Medicine, University of Oxford, Oxford, England
    Search for more papers by this author
  • Christine T. Ekdahl,

    1. Laboratory of Neurogenesis and Cell Therapy, Section of Restorative Neurology, Wallenberg Neuroscience Center A11, University Hospital, Lund, Sweden
    2. Lund Strategic Research Center for Stem Cell Biology and Cell Therapy, Lund, Sweden
    3. Division of Clinical Neurophysiology, University Hospital, Lund, Sweden
    Search for more papers by this author
  • Zaal Kokaia,

    1. Lund Strategic Research Center for Stem Cell Biology and Cell Therapy, Lund, Sweden
    2. Laboratory of Neural Stem Cell Biology, Section of Restorative Neurology, Stem Cell Institute, University Hospital, Lund, Sweden
    Search for more papers by this author
  • Olle Lindvall

    Corresponding author
    1. Laboratory of Neurogenesis and Cell Therapy, Section of Restorative Neurology, Wallenberg Neuroscience Center A11, University Hospital, Lund, Sweden
    2. Lund Strategic Research Center for Stem Cell Biology and Cell Therapy, Lund, Sweden
    • Laboratory of Neurogenesis and Cell Therapy, Section of Restorative Neurology, Wallenberg Neuroscience Center, University Hospital, SE-221 84 Lund, Sweden
    Search for more papers by this author

Abstract

Neural stem cells (NSCs) in the adult rat subventricular zone (SVZ) generate new striatal neurons during several months after ischemic stroke. Whether the microglial response associated with ischemic injury extends into SVZ and influences neuroblast production is unknown. Here, we demonstrate increased numbers of activated microglia in ipsilateral SVZ concomitant with neuroblast migration into the striatum at 2, 6, and 16 weeks, with maximum at 6 weeks, following 2 h middle cerebral artery occlusion in rats. In the peri-infarct striatum, numbers of activated microglia peaked already at 2 weeks and declined thereafter. Microglia in SVZ were resident or originated from bone marrow, with maximum proliferation during the first 2 weeks postinsult. In SVZ, microglia exhibited ramified or intermediate morphology, signifying a downregulated inflammatory profile, whereas amoeboid or round phagocytic microglia were frequent in the peri-infarct striatum. Numbers of microglia expressing markers of antigen-presenting cells (MHC-II, CD86) increased in SVZ but very few lymphocytes were detected. Using quantitative PCR, strong short- and long-term increase (at 1 and 6 weeks postinfarct) of insulin-like growth factor-1 (IGF-1) gene expression was detected in SVZ tissue. Elevated numbers of IGF-1-expressing microglia were found in SVZ at 2, 6, and 16 weeks after stroke. At 16 weeks, 5% of microglia but no other cells in SVZ expressed the IGF-1 protein, which mitigates apoptosis and promotes proliferation and differentiation of NSCs. The long-term accumulation of microglia with proneurogenic phenotype in the SVZ implies a supportive role of these cells for the continuous neurogenesis after stroke. © 2008 Wiley-Liss, Inc.

INTRODUCTION

Insults to the adult brain, such as stroke and epileptic seizures, trigger proliferation of neural stem cells (NSCs) in its neurogenic areas, i.e., the subgranular zone (SGZ) in the dentate gyrus, and the subventricular zone (SVZ), lining the lateral ventricles (Abrous et al., 2005; Riquelme et al., 2008). The majority of the insult-generated neurons formed in the SGZ become dentate granule cells, similar to what occurs in the intact brain. In contrast, following stroke induced by middle cerebral artery occlusion (MCAO) in rodents, which causes cell loss in striatum and cerebral cortex, neuroblasts migrate towards the damage where a part of them develop the phenotype of the striatal neurons which died following the insult (Arvidsson et al., 2002; Parent et al., 2002; Yamashita et al., 2006; Zhang et al., 2004). Stroke-induced neurogenesis is long-lasting, new neurons being formed up to 1 year after the insult (Kokaia et al., 2006; Thored et al., 2006). Recent data suggest that neurogenesis after stroke also occurs in the human brain (Jin et al., 2006; Macas et al., 2006). These findings have raised the possibility that stimulation of neuronal replacement by endogenous neurogenesis could become of value for restoring function after stroke. However, although there is circumstantial evidence from animal studies that increased neurogenesis may be associated with improved recovery after stroke, definite proof for a causal relationship is lacking (Lindvall and Kokaia, 2008).

Microglia seem to have an important role in adult neurogenesis (Butovsky et al., 2005, 2006; Ekdahl et al., 2003; Monje et al., 2003; Schwartz et al., 2006; Ziv etal., 2006). Phagocytic microglia associated with status epilepticus or irradiation, or triggered by LPS administration, are detrimental for the survival of the new hippocampal neurons early after they have been born (Ekdahl et al., 2003; Monje et al., 2003). Consistent with a cytotoxic action of microglia also after cerebral ischemia, administration of the anti-inflammatory drug indomethacin improves the survival of stroke-generated neuroblasts in the striatum (Hoehn et al., 2005). Similarly, chronic delivery of minocycline, which reduces microglia activation, increases the number of new neuroblasts and mature neurons in the dentate gyrus after MCAO (Liu et al., 2007). However, recent experimental studies indicate that microglia may also be beneficial for neurogenesis. Despite chronic inflammation with increased numbers of activated microglia, seizure-generated new hippocampal neurons can survive for at least 6 months (Bonde et al., 2006). In agreement, chronically activated microglia are permissive to neuronal differentiation and survival in adult mouse SVZ cultures (Cacci et al., 2008). Microglia and microglia-conditioned medium rescue the in vitro formation of neuroblasts from SVZ stem cells, which otherwise is lost with continued culture (Aarum et al., 2003; Walton et al., 2006). In addition, activated microglia seem to modulate neurogenesis in the hippocampus of adrenalectomized animals (Battista et al., 2006). Hippocampal progenitor proliferation induced by enriched environment has been reported to be regulated by microglia through local interactions with T cells within the stem cell niche in the SGZ (Ziv et al., 2006). Following a stroke, activated microglia and reactive astrocytes express monocyte chemoattractant protein-1, which promotes the migration of neuroblasts to the damaged area (Yan et al., 2007). We hypothesized that microglia could be an important regulator of the stroke-induced, long-lasting neurogenesis in SVZ. However, whether alterations of numbers and morphological and functional phenotype of microglia in the SVZ occur after stroke is unknown.

Here, we have analyzed the microglial response in the SVZ after stroke induced by 2 h MCAO in rats. The objectives were three-fold: First, to determine the origins and quantify the numbers and the proportion of various morphological subtypes of microglia in SVZ at different time points after stroke, and compare with numbers and morphologies of microglia in the ischemically damaged striatum. Second, to explore whether T-cell-triggered conversion of microglia to MHC-II-expressing, antigen-presenting cells could play a significant role also for stroke-induced neurogenesis in the SVZ as reported for hippocampal neurogenesis (Ziv et al., 2006). Third, to clarify if stroke leads to long-term changes in the molecular phenotype of SVZ microglia, specifically increased numbers and proportion of microglia with proposed neuroprotective and neurogenesis-promoting actions.

MATERIALS AND METHODS

Animals and Induction of Stroke

All experimental procedures followed guidelines set by the Malmö-Lund Ethical Committee for the use and care of laboratory animals, and were conducted in accordance with European Union directive on the subject of animal rights. Under halothane anesthesia, the middle cerebral artery (MCA) was occluded on artificially ventilated male Wistar rats (body weight 300–320 g; Møllegaard Breeding Center, Copenhagen, Denmark). A nylon filament was inserted through the common carotid artery, into the internal carotid artery, past the origin of the MCA (Zhao et al., 1994), and left in place for 2 h. In sham-operated rats, the filament was advanced only a few mm inside the internal carotid artery. Physiological parameters (body temperature, arterial blood pressure, pO2, pCO2, and pH) were monitored during the entire surgery and kept within a predetermined physiological range.

Animals were perfused for microscopical analysis 2 weeks after sham surgery and 2, 6, and 16 weeks after MCAO (n = 6 in each group). Rats were also analyzed for lymphocyte recruitment 3 and 7 days poststroke (n = 5 and 7, respectively). Intraperitoneal injections of 5-bromo-2-deoxyuridine (BrdU, 50 mg/kg, Sigma Aldrich, St. Louis, MO) were given twice daily during weeks 1–2 or 7–8 and the animals (n = 6 in each group) were perfused 2, 6, and 8 weeks after MCAO, respectively. For gene analysis, rats were decapitated 1 and 6 weeks after MCAO (n = 3 at each time-point).

Adult, male CD45BL/6 (CD45.1) mice (Møllegaard) were lethally irradiated (Gy 9.75) and 4 h thereafter transplanted with 1000–1500 LSK stem cells from congenic female transgenic C57BL/6 (CD45.2) transgenic mice expressing GFP (Nygren et al., 2004). All mice were given sterile food and autoclaved acidified water and housed under pathogen free conditions. Ten weeks after bone marrow reconstitution, the animals were tested for donor derived multi-lineaged chimerism in the blood as previously described (Nygren et al., 2004), and chimeric mice were then subjected to 30 min of MCAO (n = 6) or sham surgery (n = 4), and transcardially perfused 5 weeks thereafter.

Immunohistochemistry

After transcardial perfusion with 4% ice-cold phosphate-buffered paraformaldehyde (PFA), brains were postfixed in PFA overnight, cryoprotected in 20% sucrose, and then cut coronally in 30-μm thick sections on powdered dry ice. Before staining using diaminobenzidine (DAB), free-floating sections were quenched for 20 min in 3% hydrogen peroxide and 10% methanol, and before staining with BrdU, sections were incubated in 1 M HCl at 65°C for 10 min and at room temperature for 20 min. After preincubation with appropriate normal sera, sections were incubated with rabbit anti-Iba1 (1:1000; Wako Chemicals, Osaka, Japan) overnight. The sections were then incubated with one of the following antibodies: mouse anti-ED1 (1:200; Serotec, Oxford, UK), mouse anti-Mac-2 (1:100, Abcam, Cambridge, UK), mouse anti-CD45 (1:50; BD Biosciences, San José, CA), rat anti-BrdU (1:100; Oxford Biotechnology, Oxford, UK), mouse anti-MHC-II (1:100; Serotec), mouse anti-CD86 (1:50; Serotec), mouse anti-CD43 (1:200; Serotec), mouse anti-CD8 (1:100; Serotec), mouse anti-CD4 (OX35; 1:100, Serotec), mouse anti-TCR (1:100; Serotec) and mouse anti-CD43 (1:200; Serotec), mouse anti-insulin-like growth factor-1 (IGF-1, 1:50; R&D Systems, Minneapolis, MN), goat anti-IGF-1 (1:50; Santa Cruz Biotechnology, Santa Cruz, CA), mouse anti-NeuN (1:100; Temecula, CA) and goat anti-doublecortin (Dcx, 1:400; Santa Cruz Biotechnology, Santa Cruz, CA). The stainings were visualized by incubation for 2 h with Cy3-conjugated donkey anti-rabbit antibody (1:200; Jackson Immunoresearch, West Grove, PA) and biotinylated secondary antibodies (1:200; Vector, Burlingame, CA) followed by Alexa 488-conjugated streptavidin (1:200; Molecular Probes, Eugene, OR), for 2 h for double stainings, or with avidin-biotin-peroxidase complex for 1 h followed by treatment with DAB (0.5 mg/mL). Finally, sections were rinsed, counterstained with Hoechst (0.25 μL/mL; Molecular Probes), mounted, and cover-slipped.

All stainings were performed with the proper positive and/or negative controls. When using the lymphocyte markers (CD43, CD86, CD4, and TCR), tissue from mouse thymus was used as positive control. All antibodies were incubated with normal sera obtained from the animal species from which the secondary antibody was produced.

Microscopical Analysis

All assessments were performed by an observer blind to the treatment conditions. The number of immunoreactive cells was counted in both SVZ and striatum at 40× magnification using an epifluorescence microscope in four coronal sections at +1.60, +1.00, +0.48, and −0.26 mm from bregma. All cells were counted in the SVZ, both ipsi- and contralaterally. In the striatum, cells were counted in 12 continuous fields using a 0.0625 mm2 square grid (Thored et al., 2006). Three fields were counted along the medio-lateral and four along the dorso-ventral extension of the striatum, the most medial fields lining the SVZ. The percentage of the different morphological subtypes of microglia was based on analysis of 100 cells in each animal in both SVZ and striatum.

Co-expression of Iba1 and various phenotypic markers and BrdU, as observed using epifluorescence microscopy, was validated using a confocal laser-scanning microscope (Leica). Cells were considered double-labeled if labeling with relevant morphology was seen throughout the extent of the nucleus for nuclear markers, or if a cytoplasmic marker surrounded a nuclear marker, when viewed in x-y, x-z, and y-z cross-sections produced by orthogonal reconstruction from z-series (z-step, 1 μm) taken with a 63× objective.

Preparation of Total RNA, cDNA Synthesis, and Relative RNA Quantification

Subventricular zone was dissected and immediately frozen in liquid nitrogen. Samples were homogenized using the Tissue Roptor (Qiagen, Dellre, Germany) in a vessel containing the proper amount of QIAzol (Qiagen), followed by total RNA (totRNA) purification using the RNeasy Lipid Tissue Mini Kit (Qiagen) in accordance with manufacturer's protocol. Quantity and contamination degree of totRNA was investigated using the Nano Drop analyzer (Nano Drop Technology, Dellre, Germany). The cDNA was generated by annealing 0.5 μg of totRNA to 2.5 μM of Oligo DT Primers (Takara Bio) in the presence of 10 mM of dNTP Mixture (Takara Bio) in 10 μL of RNase free water. Reverse transcriptase reactions were performed with 0.5 μL PrimeScript™ RTase (for two steps; Takara Bio) in the manufacturer's buffer supplemented with 20 U RNase inhibitor (Takara Bio) in a total volume of 20 μL. Quantitative RT-PCR was carried out using the Power SYBR Green PCR Master Mix (Applied Biosystems) and the StepOne™ Real-Time PCR System (Applied Biosystems). The threshold cycles (Ct) for the endogenous control Actinβ RNA and the target signals were determined and the relative RNA quantification was calculated using the comparative Ct method as 2−Ct were Ct is Ct(target) − Ct(Actinβ). Samples were run in triplicates for each investigated transcript. The analysis was performed using the following primers:

Actinβ sense: 5′-GCCCTAGACTTCGAGCAAGA; Actinβ antisense: 5′-AGGAAGGAAGGCTGGAAGAG; IGF-1 sense: 5′-GAACAGAAAATGCCACGTCA; IGF-1 antisense: 5′-GGAAATGCCCATCTCTGAAA. Actinβ and IGF-1 were amplified by 45 cycles at a Tm of 60°C.

Statistical Analysis

Comparisons were performed using one-way analysis of variance (ANOVA) with Bonferroni-Dunn post-hoc test, or with paired or unpaired t-test. Data are presented as means ± SEM, and differences considered significant at P < 0.05.

RESULTS

Stroke Causes Activation of Microglia with Different Time Course in SVZ and Striatum

The 2 h MCAO gave rise to extensive tissue damage in the striatum and overlying parietal cortex (Arvidsson et al., 2002; Thored et al., 2006). In NeuN-stained sections, neuronal loss was observed in the lateral and caudal parts of striatum ipsilateral to MCAO, whereas the most medial and dorsal parts were often spared. We found migration of Dcx+ neuroblasts from SVZ into the damaged striatum at all time points (Fig. 1B), in agreement with our previous report (Thored et al., 2006). In contrast, Dcx immunoreactivity was confined to the SVZ and to very few striatal neurons on the contralateral side (Fig. 1A) and in sham-operated animals.

Figure 1.

Neuroblasts migrate from SVZ towards the striatal damage after stroke. A, B: Dcx+ neuroblasts migrating from the SVZ towards the injury at 6 weeks after 2 h MCAO. C: Confocal image of migrating, Dcx+ neuroblasts in striatum. Orthogonal reconstructions from confocal z-series presented as viewed in x-z (bottom) and y-z (right) planes; Scale bar is 50 (A, B) and 15 μm (C). [Color figure can be viewed in the online issue, which is available at www.interscience.wiley.com.]

We wanted to explore if the microglial response after MCAO involved the neurogenic area, i.e., SVZ. The number of Iba1+ cells, which include both activated and quiescent or “surveying” (Hanisch and Kettenmann, 2007) microglia was first quantified in the SVZ (see Fig. 2). At all time points analyzed following the insult (2, 6, and 16 weeks), the number of Iba1+ cells was markedly higher in the SVZ ipsilateral to MCAO when compared with the contralateral side and sham-operated animals (Fig. 2P). No alterations in microglia numbers were detected in SVZ contralateral to MCAO.

Figure 2.

Microglia activation is long-lasting in the SVZ ipsilateral to stroke. A–L: Photomicrographs showing distribution of Iba1+ (B, F, J), ED1+ (C, G, K), and Iba1+/ED1+ microglia in the ipsilateral (D, H, L) or contralateral (A, E, I) SVZ at 2, 6, and 16 weeks after 2 h MCAO. M–O: Confocal images of Iba1+/ED1+ microglia in SVZ showing Iba1 (M) and ED1 (N) immunoreactivity separately or as merged image (O). Orthogonal reconstructions from confocal z-series are presented as viewed in x-z (bottom) and y-z (right) planes. Scale bar, 25 μm (A–L) and 10 μm (M–O). LV, lateral ventricle; Str, striatum. P–R: Total number of microglia (Iba1+ cells, P), and number (Iba1+/ED1+ cells, Q) and percentage (ED1+ of Iba1+ cells, R) of activated microglia in the ipsilateral and contralateral SVZ after sham surgery or at 2, 6, and 16 weeks after 2 h MCAO. Means ± SEM, n = 6 for each time-point. *, P < 0.05 compared with contralateral side, paired t-test, or to sham or other time-point, one-way ANOVA with Bonferroni-Dunn post-hoc test. [Color figure can be viewed in the online issue, which is available at www.interscience.wiley.com.]

We then counted phagocytic microglia, expressing both Iba1 and ED1 in the SVZ (see Fig. 2). The numbers of Iba1+/ED1+ cells in SVZ ipsilateral to MCAO were elevated compared with the contralateral side at all time points (Fig. 2Q). Also the percentage of phagocytic microglia was higher in the ipsilateral SVZ (Fig. 2R). Taken together, our data shows that concomitantly with the continuous striatal neurogenesis after stroke, there is a long-lasting increase of microglia numbers and activation in the SVZ ipsilateral to the ischemic damage.

In contrast to the SVZ, where microglia numbers peaked at 6 weeks (Figs. 2P,Q), the Iba1+ and Iba1+/ED1+ cells in the peri-infarct striatal area reached maximum numbers already at 2 weeks after MCAO (see Fig. 3). The percentage of phagocytic microglia was high at all time points (Fig. 3O). Few microglia were observed in the contralateral striatum and in sham-operated animals (Fig. 3M–O). These data are consistent with the classical descriptions of a peak of microglia reaction in the damaged striatum within the first 2 weeks after stroke (Clark et al., 1993; Lehrmann et al., 1997; Morioka et al., 1993). However, our findings indicate a different time course for the stroke-induced microglial response in the SVZ.

Figure 3.

Microglia activation occurs early in the striatal peri-infarct area. A–L: Photomicrographs showing distribution of Iba1+ (B, F, J), ED1+ (C, G, K), and Iba1+/ED1+ microglia in the ipsilateral (D, H, L) and contralateral (A, E, I) striatum (Str) at 2, 6, and 16 weeks after 2 h MCAO. Scale bar, 25 μm. M–O: Total number of microglia (Iba1+ cells, M) and number (Iba1+/ED1+ cells, N) and percentage (ED1+ of Iba1+ cells, O) of activated microglia in the ipsilateral and contralateral striatum after sham surgery or at 2, 6, and 16 weeks after 2 h MCAO. Means ± SEM, n = 6 for each time-point. *, P < 0.05 compared with contralateral side, paired t-test, or to sham or other time-point, one-way ANOVA with Bonferroni-Dunn post-hoc test. [Color figure can be viewed in the online issue, which is available at www.interscience.wiley.com.]

Microglia Exhibit Different Morphological Phenotypes in SVZ and Damaged Striatum After Stroke

We then wanted to compare the morphological characteristics of microglia in the SVZ with those of striatal microglia at different time points after stroke. Microglia were classified into ramified, intermediate, amoeboid, or round phenotypes (Fig. 4A–D) in SVZ and striatum ipsilateral to MCAO using the morphological criteria described by Lehrmann et al. (1997). These morphological phenotypes are believed to represent different steps of microglial activation, which can be correlated with distinct functional states, ramified microglia signifying resting and round microglia typically the most activated ones (Streit et al., 1999). In the SVZ of sham-operated animals, the majority of microglial cells were ramified, but a portion of them exhibited intermediate (15%) and a small percentage amoeboid (2%) or round (1%) morphologies (Fig. 4E). The proportion of microglia with different morphologies was similar in the intact striatum (Fig. 4E).

Figure 4.

Microglia exhibit different morphological phenotypes in SVZ compared with striatal peri-infarct area. A–D: Photomicrographs showing examples of microglia classified into four different phenotypes: ramified, intermediate, amoeboid and round. E: Percentage of microglia with different morphological phenotypes of total number of Iba1+ cells in the ipsilateral SVZ and peri-infarct striatum at 2, 6, and 16 weeks after 2 h MCAO or sham surgery. F: Table summarizing the statistical comparison between percentage of microglia of different morphological phenotypes within the SVZ at 2, 6, and 16 weeks after 2 h MCAO. Means ± SEM, n = 6 for each time-point. *, P < 0.05 compared with sham, unpaired t-test. #, P < 0.05 SVZ compared with striatum, paired t-test, x, P < 0.05 compared with the other time points, one-way ANOVA. [Color figure can be viewed in the online issue, which is available at www.interscience.wiley.com.]

In the peri-infarct striatum, there was a dramatic decrease of ramified microglia at 2, 6, and 16 weeks after MCAO (Fig. 4E). Conversely, the proportion of amoeboid and round microglia was increased at all time points, with maximum at 2 weeks. When compared with striatum, ramified microglia were much more frequent in the SVZ ipsilateral to MCAO at 2, 6, and 16 weeks, whereas the proportion of amoeboid and round microglia was clearly lower (Fig. 4E). The percentage of microglia with intermediate morphology was markedly increased at 6 and 16 weeks in both areas, and did not differ between SVZ and striatum. When comparing the percentage of the various phenotypes within the SVZ at the three time points after stroke (Fig. 4F), we found that ramified microglia were transiently decreased at 6 weeks after stroke. In contrast, the percentage of both intermediate and amoeboid microglia peaked at 6 weeks. The percentage of microglia with round morphology did not differ between the time points (Fig. 4F). Thus, the overall majority of microglia in the SVZ presented a ramified or intermediate morphology and, in contrast to peri-infarct striatal microglia, never became excessively activated. Taken together, these findings provide morphological evidence of a differential state of activation of microglia in SVZ and injured striatum after stroke.

Microglia in SVZ Originate Partly from Bone Marrow, Proliferate Early, and Survive Long-Term After Stroke

We next wanted to determine the contribution of resident microglia and bone-marrow-derived macrophages to the microglial response in the SVZ after MCAO. First, we evaluated the expression of CD45, a putative marker for myeloid cells (Donnou et al., 2005), in the Iba1+ cells in SVZ. The number of Iba1+/CD45+ cells, which mainly exhibited intermediate morphology, was increased in ipsilateral SVZ at 2, 6, and 16 weeks after MCAO compared with contralateral side and sham-operated controls (Fig. 5A–D,I–K,R). About 30–50% of the total number of Iba1+ cells expressed CD45 at these time points. About 30–50% of the total number of Iba1+ cells in SVZ expressed CD45 at these time points, and we found similar percentages (ranging from 30 to 55%) of Iba1+/CD45+ cells in the ipsilateral striatum. Thus, recruitment of bone marrow-derived macrophages contributes to the long-term increase in the number of microglia inside the SVZ after stroke.

Figure 5.

Microglia in SVZ originate from resident microglia and from blood after stroke. A–H: Photomicrographs showing distribution of Iba1+ (B, F), CD45+ (C), Mac2+ (G), Iba1+/CD45+ and Iba1+/Mac2+ microglia in the ipsilateral (D, H) and contralateral (A, E) SVZ at 2 weeks after 2 h MCAO. I–Q: Confocal images of Iba1+ microglia in ipsilateral SVZ co-expressing CD45, Mac2, or BrdU showing Iba1 (I, L, O), CD45 (J), Mac2 (M), or BrdU (P) immunoreactivity separately or as merged images (K, N, Q). Orthogonal reconstructions from confocal z-series are presented as viewed in x-z (bottom) and y-z (right) planes. Scale bar, 25 μm (A–H), 5 μm (I–Q). LV, lateral ventricle; Str, striatum. R–U: Photomicrographs showing distribution of Iba1+ (R) and GFP+ (S), microglia in the ipsilateral SVZ and striatum 5 weeks after stroke in mice with reconstituted, GFP+ bone marrow. Arrow and arrowhead point to a singe labeled Iba1+, and a double-labeled Iba1+/GFP+ cell, respectively. R–U: Number of blood-derived (Iba1+/CD45+ cells, U) microglia, number (Iba1+/Mac2+ cells, V) and percentage (Mac2+ of Iba1+ cells, X) of presumed resident microglia, and number of proliferated microglia (Iba1+/BrdU+ cells as assessed with BrdU injections during weeks 1–2 (left) or 7–8 (right), Y) in the ipsilateral and contralateral SVZ after sham surgery or at 2, 6, 8, and 16 weeks after 2 h MCAO. Means ± SEM. n = 6 for each time point. *, P < 0.05 compared with contralateral side, paired t-test, or to sham or other time-point, one-way ANOVA with Bonferroni-Dunn post-hoc test. [Color figure can be viewed in the online issue, which is available at www.interscience.wiley.com.]

To provide further evidence that recruitment of bone marrow-derived cells to the SVZ occurs after stroke, we subjected lethally irradiated mice that had been reconstituted with bone marrow from transgenic mice expressing GFP to MCAO. Five weeks after the insult, we found substantial numbers of bone marrow-derived, GFP+ cells in the SVZ ipsilateral to the insult (see Fig. 5). Large numbers of GFP+ cells were also observed in the ipsilateral striatum. In contrast, only scattered GFP+ cells were detected in the SVZ and striatum of sham-operated mice. The bone marrow-derived, GFP+ cells in the SVZ and stroke-damaged striatum also expressed Iba1. These macrophages were intermingled with Iba1+/GFP- cells, most likely representing resident microglia. A recent report has suggested that the lectin Mac-2, also known as galectin-3, is expressed by resident microglia but not by bone-marrow-derived macrophages up to 3 days after MCAO (Lalancette-Hebert et al., 2007). We tested this potential marker for resident microglia and evaluated the expression of Mac-2 in Iba1+ cells in the SVZ after MCAO (Fig. 5E–H,L–N,S–T). At all analyzed time points, there were increased numbers of Iba1+/Mac-2+ cells in the SVZ ipsilateral to MCAO when compared with contralateral side and sham. However, only about 8–10% of the total number of Iba1+ cells in the SVZ co-expressed Mac-2 at 2, 6, and 16 weeks after MCAO (Fig. 5T), displaying mainly intermediate morphology. Similar percentages of Iba1+/Mac-2+ cells were found in the stroke-damaged striatum, ranging between 3 and 14% at the different time points (data not shown). It seems most likely, therefore, that the Iba1+/Mac-2+ cells constitute a subpopulation of resident microglia in SVZ after stroke.

We finally analyzed how much cell proliferation contributed to the increased microglia numbers in SVZ early and late after stroke. The mitosis marker BrdU was injected during weeks 1–2 or 7–8 after MCAO and the rats were killed at the end of injections. The number of Iba1+/BrdU+ microglia in SVZ ipsilateral to the damage was markedly increased compared with the contralateral side and sham-operated controls directly after BrdU injections during the first 2 weeks after the insult (Fig. 5O–Q,U). In contrast, in animals injected with BrdU during weeks 7 and 8, Iba1+/BrdU+ cells in SVZ did not differ from controls. Thus, proliferation of microglia accumulating in the SVZ occurs early after the insult. In animals injected with BrdU during the first 2 weeks after MCAO, the number of Iba1+/BrdU+ cells in SVZ did not change between 2 and 6 weeks (Fig. 5U). Our finding suggests long-term survival of this population of microglial cells.

Microglia Partly Adopt the Phenotype of Antigen-Presenting Cells but Scarcely Interact with T Cells in SVZ After Stroke

Recent experimental evidence has suggested that interaction between microglia-like, MHC-II+ dendritic cells, and T lymphocytes is important for the maintenance of hippocampal neurogenesis (Ziv et al., 2006). We wanted to determine if such an interaction could occur also in the SVZ after stroke. We therefore analyzed the expression of the antigen-presenting cell marker, MHC-II, in Iba1+ microglia in both SVZ and striatum at different time-points after MCAO (see Fig. 6). In sham-operated animals, about 1% of Iba1+ cells were also MHC-II+ in the SVZ but only 0.1% in the striatum. This finding may suggest that SVZ but not striatal microglia exhibit significant levels of constitutive MHC-II activation. We found increased numbers and percentage of Iba1+/MHC+ cells (ranging between 4 and 6% of total Iba1+ cell numbers) in the SVZ ipsilateral to MCAO but not on the contralateral side from 2 to 16 weeks poststroke (Fig. 6P). The Iba1+/MHC-II+ cells predominantly displayed an intermediate morphology (Fig. 6M). The pattern of expression of Iba1+ cells that were positive for CD86, another antigen-presenting cell marker, was similar to that expressing MHC-II (not illustrated).

Figure 6.

Microglia in SVZ partly adopt the phenotype of antigen-presenting cells after stroke. A–L: Photomicrographs showing distribution of Iba1+ (B, F, J), MHC-II+ (C, G, K), and Iba1+/MHC-II+ microglia in the ipsilateral (D, H, L) and contralateral (A, F, I) SVZ and striatum at 2, 6, and 16 weeks after 2 h MCAO. LV, lateral ventricle; Str, striatum. M–O: Confocal images of Iba1+/MHC-II+ microglia in SVZ showing Iba1 (M) and MHC-II (N) immunoreactivity separately or as merged image (O). Orthogonal reconstructions from confocal z-series are presented as viewed in x-z (bottom) and y-z (right) planes. Scale bar, 50 μm (A–H), 25 μm (I–L), 6 μm (M–O). P–Q: Percentage of MHC-II+ cells of total number of Iba1+ microglia in SVZ (P) and striatum (Q) at 2, 6, and 16 weeks after 2 h MCAO or sham surgery. Means ± SEM, n = 6 for each time point. *, P < 0.05 compared with contralateral side, paired t-test, or to sham or other time-point, one-way ANOVA with Bonferroni-Dunn post-hoc test. [Color figure can be viewed in the online issue, which is available at www.interscience.wiley.com.]

In the striatal peri-infarct area, there was a major increase of Iba1+/MHC-II+ cells, mainly displaying amoeboid and round morphologies (see Fig. 6). At 2 weeks after MCAO, 65% of the Iba1+ cells expressed MHC-II (Fig. 6D,Q). The percentage decreased thereafter, but remained elevated at all time-points compared with sham and contralateral side (Fig. 6Q). Also the increase of Iba1+/CD86+ cells was maximum at 2 weeks after MCAO and then declined (data not shown).

MHC-II is involved in the mechanism of antigen presentation to lymphocytes, and interaction between microglia and lymphocytes has been suggested to be important for neurogenesis (Butovsky et al., 2006; Ziv et al., 2006). CD43 was used to label lymphocytes and assess the extent of lymphocyte recruitment to SVZ and damaged striatum after stroke (see Fig. 7). There was only very minor lymphocyte recruitment to SVZ at all time points after MCAO. The CD43+ cells were virtually absent in the SVZ and only single cells could be detected in individual animals (Fig. 7A,B). In the peri-infarct striatum, some CD43+ cells were found at all time points, with a peak at 3 days after (Fig. 7B,C). The same distribution of lymphocytes was also observed using the other markers (CD4, CD8, and TCR; not illustrated). To conclude, our results argue against the possibility that interaction between lymphocytes and microglia plays any major role for maintenance of long-term neurogenesis after stroke.

Figure 7.

Recruitment of lymphocytes is restricted to damaged striatum and peaks in peri-infarct area within first week after stroke. A–C: CD43+ lymphocytes are virtually non-detectable in the SVZ ipsilateral to MCAO or sham surgery (A, B) at all time-points studied (3 and 7 days, and 2, 6, and 16 weeks), but present in peri-infarct striatum (B). No lymphocytes are present in the striatum of sham-operated animals (A). Inset in B is shown in higher magnification in C. LV, lateral ventricle; Str, Striatum. Scale bar, 50 μm (A, B), 25 μm (C). [Color figure can be viewed in the online issue, which is available at www.interscience.wiley.com.]

Microglia in SVZ Exhibit Proneurogenic Molecular Phenotype After Stroke

We wanted to further explore the molecular phenotype of SVZ microglia after stroke to reveal potential mechanisms by which these cells could support long-term neurogenesis. Ipsi- and contralateral SVZ was dissected from rats 1 and 6 weeks after MCAO and the gene expression level of IGF-1, which has been proposed to be characteristic of microglia with neuroprotective/proneurogenic phenotype (Ziv et al., 2006), was assessed using quantitative real time PCR (qRT-PCR). Previous studies have shown that IGF-1 is an important diffusible mediator of neural progenitor proliferation in the hippocampus and SVZ proliferation (Yan et al., 2006), and is released by astrocytes (Åberg et al., 2000, 2003) as well as microglia after stroke (Lalancette-Hebert et al., 2007). Our qRT-PCR study showed that the expression level of IGF-1 messenger RNA (mRNA) in the SVZ was dramatically increased (around 16-fold upregulation) 1 week after MCAO (Fig. 8T). Moreover, the transcription of the IGF-1 gene was maintained in an upregulated fashion 6 weeks after stroke, when the mRNA levels were 11 times higher ipsi- than contralaterally. This finding suggests elevated expression levels of the gene after stroke, possibly due to increased cell numbers. Using immunohistochemistry, we then investigated whether the persistently high IGF-1 gene levels in SVZ tissue after stroke were associated with increased numbers of microglia cells expressing the protein at 2, 6, and 16 weeks after MCAO (see Fig. 8). Consistent with the gene expression data, we found marked increases in the number of Iba1+/IGF-1+ microglia in the ipsilateral SVZ at all time-points, with maximum at 6 and 16 weeks (Fig. 8P). The same staining pattern was observed using two different IGF-1 antibodies. At 16 weeks, the percentage of Iba1+ cells also expressing IGF-1 reached 5%, which was 10-fold higher than in SVZ of sham animals (Fig. 8Q). All IGF-1+ cells in SVZ were Iba1+, indicating that IGF-1 expression at the time-points studied was confined to microglia. In the peri-infarct striatum, the time course for changes of Iba1+/IGF-1+ cells was different from that in SVZ, with maximum numbers at 6 weeks and tapering off thereafter (Fig. 8R,S). Taken together, these findings indicate that there is a long-lasting, marked increase of microglia with proneurogenic IGF-1+ phenotype in SVZ concomitant with the continuous, long-term neuroblast production after stroke.

Figure 8.

Microglial expression of IGF-1 is increased in SVZ after stroke. A–L: Photomicrographs showing distribution of Iba1+ (B, F, J), IGF-1+ (C, G, K), and Iba1+/IGF-1+ microglia in the ipsilateral (D, H, L) and contralateral (A, E, I) SVZ and striatum 2, 6, and 16 weeks after 2 h MCAO. LV, lateral ventricle; Str, striatum. M–O: Confocal images of Iba1+/IGF-1+ microglia in SVZ showing Iba1 (M) and IGF-1 (N) immunoreactivity separately or as merged image (O). Orthogonal reconstructions from confocal z-series are presented as viewed in x-z (bottom) and y-z (right) planes. Scale bar, 25 μm (A–L), 6 μm (M–O). P–S: Number (Iba1+/IGF-1+ cells, P, R) and percentage (IGF-1+ of Iba1+ cells, Q, S) of IGF-1-expressing microglia in the ipsilateral and contralateral SVZ or striatum after sham surgery or at 2, 6, and 16 weeks after 2 h MCAO. T: Changes in IGF-1 mRNA expression levels in ipsilateral when compared with contralateral SVZ at 1 and 6 weeks after 2 h MCAO as determined by qRT-PCR. Means ± SEM. n = 4 (P–S) and 3 (T) for each time point. *, P < 0.05 compared with contralateral side, paired t-test, or to sham or other time-point, one-way ANOVA with Bonferroni-Dunn post-hoc test. [Color figure can be viewed in the online issue, which is available at www.interscience.wiley.com.]

DISCUSSION

Here, we show that the persistent production of new striatal neuroblasts from SVZ stem/progenitor cells (Thored et al., 2006) is accompanied with a long-lasting accumulation of activated microglia in the SVZ up to at least 16 weeks after stroke. At all time-points studied (2, 6, and 16 weeks), the number of activated microglia in SVZ after stroke was higher ipsilaterally when compared with contralateral side or to sham-operated animals, with maximum activation at 6 weeks postinsult. In contrast, in the striatal peri-infarct area, microglia were most numerous already at 2 weeks and their numbers then tapered off. These latter findings agree well with the classical descriptions that the inflammatory response in general and microglial activation in particular is maximum in striatum and cerebral cortex during the first 2 weeks after stroke and declines thereafter (Clark et al., 1993; Lehrmann et al., 1997; Morioka et al., 1993; Zhang et al., 1997). In addition, SVZ microglia exhibited a lower degree of activation, as evidenced by predominantly ramified and intermediate morphologies. Striatal microglia, in contrast, were more activated and often amoeboid and round. The SVZ microglia were also characterized by a proneurogenic molecular phenotype. Taken together, our findings indicate that microglia activation in the peri-infarct area and in the neurogenic area, i.e., the SVZ is differentially regulated after stroke.

Previous studies have shown early activation of resident microglia and late invasion of blood-borne macrophages in the damaged area after stroke (Schroeter etal., 1997). Depletion of blood-borne macrophages with clodronate does not affect the number of ED1+ cells up to 3 days after photothrombosis, but the number of ED1+ phagocytes is considerably lower in clodronate-treated animals at late time points, indicating that hematogenous macrophages are secondarily recruited after focal ischemia (Schroeter et al., 1997). Studies using green fluorescent protein (GFP) transgenic, bone marrow chimeric mice support that microglial activation precedes and predominates over macrophage infiltration after MCAO (Schilling et al., 2003, 2005). In mice transplanted with bone marrow from a GFP+ donor, hematogenous GFP+ macrophages are observed from 2 days after MCAO, with maximum numbers at 7 days and decline thereafter (Schilling et al., 2003, 2005). However, GFP-microglia are activated already at day 1 and constitute the majority of microglia during the first weeks after MCAO (Schilling et al., 2003, 2005).

We provide evidence that the increased number of SVZ microglia after stroke was due to both proliferation of resident microglia and inflow of blood-derived cells. We used CD45 as a marker for blood-derived macrophages, which are considered to express high levels of this molecule (Goings et al., 2006). Up to at least 6 weeks after stroke, there was a progressive increase in the number of Iba1+/CD45+ cells in SVZ, which suggests that infiltration of blood-borne cells partly underlies the increase in SVZ microglial numbers. We also demonstrated the occurrence of Iba1+/GFP+ cells in the SVZ of stroke-damaged, irradiated mice reconstituted with bone marrow from transgenic GFP-expressing mice. This finding confirmed that recruitment of blood-derived macrophages and not upregulation of CD45 in resident microglia explained the elevated numbers of Iba1+/CD45+ cells in SVZ. However, not more than 50% of Iba1+ cells co-expressed CD45 at any time-point, which indicates that resident microglia constitutes a considerable portion of SVZ microglia after stroke. It is conceivable that the early and marked proliferation of Iba1+ cells, which occurred during the first 2 weeks after stroke and declined thereafter, lead to an increase of the population of resident microglia in SVZ. These newly generated microglia showed good survival and most likely contributed to the long-term elevation of SVZ microglia numbers. It is a problem that no reliable marker for resident microglia has been identified. Recent findings using GFP-chimeric mice indicated that within the first 72 h after ischemic injury, Mac-2 was preferentially expressed by resident microglia (Lalancette-Hebert et al., 2007). We observed expression of Mac-2 in about 8–10% of total Iba1+ cells at 2, 6, and 16 weeks after MCAO. Our results suggest that a subpopulation of activated resident microglia in the SVZ co-expresses Mac-2. However, Mac-2 is unlikely to be a marker for all resident microglia after stroke. In summary, although our findings show that both resident and blood-borne microglia contribute to SVZ microglia after stroke, the relative importance of the different origins could not be determined.

We observed differences in time course of activation and in morphological and molecular phenotype of microglia between SVZ and the striatal peri-infarct area. These findings indicate that the long-term microglia response to stroke in the SVZ is specific and not merely part of a general reaction in the ischemically damaged hemisphere. The factors responsible for the microglia activation in the SVZ after stroke, resulting in migration and proliferation of these cells, are unclear. Recently, we observed that in the present stroke model, SVZ is acutely and transiently hypoxic, but not exhibiting any increased cell death after 2 h MCAO (Thored et al., 2007). Hypoxia can trigger microglia activation (Ábrahám et al., 2001). ATP and ADP released by damaged cells are also important inducers of microglial activation and proliferation (Davalos et al., 2005; Haynes et al., 2006; Nimmerjahn et al., 2005). Hypothetically, damaged cells in the SVZ or adjacent striatum could release nucleotides triggering microglia activation and proliferation. Interestingly, in parallel experiments we have found upregulation of the gene for CXCL10 in the SVZ at both 1 and 6 weeks after stroke. This chemokine has been shown to control microglia migration by acting on the CXCR3 receptor on the microglia themselves (Rappert et al., 2004). The long-lasting aggregation of microglia in the SVZ could partly be due to stroke-induced, persistent elevation of CXCL10 expression.

After stroke, microglia in the SVZ displayed a different morphological profile when compared with striatal microglia. In the SVZ, 80–90% of microglia were ramified or had an intermediate morphology at all time points and only 10–20% of the cells were amoeboid or round. In contrast, 40–90% of striatal microglia were amoeboid or round, with maximum at 2 weeks after MCAO, and the remaining ones were mainly intermediate with few ramified microglia. It is conceivable that the fact that SVZ microglia avoided excessive activation, characterized by amoeboid and round morphologies, provides evidence for a more neuroprotective role when compared with the presumed neurotoxic action of striatal microglia (Butovsky et al., 2006; Shaked et al., 2004; Yrjanheikki et al., 1999). In accordance with this hypothesis, Battista et al. (2006) reported that moderate microglial activation was positively correlated with hippocampal neurogenesis in adrenalectomized animals. It should be pointed out, though, that studies in a model of prion disease (Perry et al., 2007) have indicated that microglia can switch to a phenotype contributing to neuronal damage without morphological change (Perry et al., 2007). Thus, to determine whether microglia have neurotoxic or neurogenic functions, morphological analysis seems insufficient and requires molecular characterization.

We found increased levels of IGF-1 mRNA at 1 and 6 weeks and elevated numbers of IGF-1+ microglia at 2, 6, and 16 weeks in the SVZ, with IGF-1 expression in about 5% of SVZ microglia at 16 weeks after MCAO. Several lines of evidence indicate that the IGF-1 secreted by the aggregated microglia in the SVZ demonstrated here could be an important regulator of stroke-induced neurogenesis. First, IGF-1 can influence several steps of cell genesis in CNS including the proliferation (Kalluri et al., 2007), survival (Camarero et al., 2003), and differentiation (Otaegi et al., 2006) of neural progenitors. Second, Yan et al. (2006) have provided immunocytochemical evidence that the neural progenitors in the SVZ express the IGF-1 receptor. Third, intraventricular infusion of an antibody against IGF-1 during the first week after the insult decreased stroke-induced progenitor proliferation in SVZ (Yan et al., 2006). Because Yan et al. (2006) detected IGF-1+ reactive astrocytes and increased IGF-1 gene expression in the ischemic cortex but no IGF-1+ cells in the SVZ at day 3–4 after MCAO, they proposed that IGF-1 mediated stroke-induced progenitor proliferation through diffusion from cerebral cortex to SVZ. Our data indicate that IGF-1 is produced locally in SVZ and can directly influence the NSCs.

In the dentate gyrus, enriched environment stimulates neurogenesis and leads to increased numbers of microglia expressing MHC-II and IGF-1 and recruitment of T lymphocytes (Ziv et al., 2006). Lack of T cells reduced progenitor proliferation, which was restored by T-cell replenishment. Ziv et al. (2006) therefore hypothesized that interaction between lymphocytes and microglia is important for the regulation of hippocampal neurogenesis, in particular progenitor proliferation. Consistent with the findings of Ziv et al. (2006), we found here that after stroke, SVZ contained increased numbers of microglia expressing MHC-II and IGF-1, i.e., molecules that are believed to be associated with a neuroprotective microglial phenotype. However, very few lymphocytes were detected in the SVZ at all time points in contrast to the peri-infarct area where lymphocytes were readily found during the first days after MCAO. Thus, our data argue against the idea that T-lymphocyte-microglia interaction in the neurogenic niche in SVZ is of importance for long-term neurogenesis after stroke. It should be emphasized though, that a large number of molecules besides IGF-1 have been reported to promote SVZ neurogenesis after stroke (Lindvall and Kokaia, 2008) including, e.g., VEGF (Sun et al., 2003; Wang et al., 2007), GDNF (Kobayashi et al., 2006) erythropoietin (Wang et al., 2004), FGF (Leker et al., 2007), and Notch ligands (Androutsellis-Theotokis et al., 2006). The neurogenic response after stroke is, therefore, most likely not regulated only by a single molecule, but its different steps orchestrated by many factors. Our data are consistent with the idea that IGF-1 is one of these proneurogenic factors.

The long-term aggregation of microglia with proneurogenic and neuroprotective molecular and morphological phenotypes in the SVZ after stroke implies a supportive role for the continuous neuroblast production. To definitely establish a causal relationship, genetic mouse models for conditional ablation of this specific population of microglia need to be developed. Importantly, the present findings raise the possibility that targeting inflammatory mechanisms and microglia phenotype regulation may be developed into a novel therapeutic strategy to optimize regenerative responses after stroke.

Acknowledgements

The Lund Stem Cell Center is supported by a Center of Excellence grant in Life Sciences from the Swedish Foundation of Strategic Research.

Ancillary