Neuronal Hyperactivity Disturbs ATP Microgradients, Impairs Microglial Motility, and Reduces Phagocytic Receptor Expression Triggering Apoptosis/Microglial Phagocytosis Uncoupling

Abstract

Phagocytosis is essential to maintain tissue homeostasis in a large number of inflammatory and autoimmune diseases, but its role in the diseased brain is poorly explored. Recent findings suggest that in the adult hippocampal neurogenic niche, where the excess of newborn cells undergo apoptosis in physiological conditions, phagocytosis is efficiently executed by surveillant, ramified microglia. To test whether microglia are efficient phagocytes in the diseased brain as well, we confronted them with a series of apoptotic challenges and discovered a generalized response. When challenged with excitotoxicity in vitro (via the glutamate agonist NMDA) or inflammation in vivo (via systemic administration of bacterial lipopolysaccharides or by omega 3 fatty acid deficient diets), microglia resorted to different strategies to boost their phagocytic efficiency and compensate for the increased number of apoptotic cells, thus maintaining phagocytosis and apoptosis tightly coupled. Unexpectedly, this coupling was chronically lost in a mouse model of mesial temporal lobe epilepsy (MTLE) as well as in hippocampal tissue resected from individuals with MTLE, a major neurological disorder characterized by seizures, excitotoxicity, and inflammation. Importantly, the loss of phagocytosis/apoptosis coupling correlated with the expression of microglial proinflammatory, epileptogenic cytokines, suggesting its contribution to the pathophysiology of epilepsy. The phagocytic blockade resulted from reduced microglial surveillance and apoptotic cell recognition receptor expression and was not directly mediated by signaling through microglial glutamate receptors. Instead, it was related to the disruption of local ATP microgradients caused by the hyperactivity of the hippocampal network, at least in the acute phase of epilepsy. Finally, the uncoupling led to an accumulation of apoptotic newborn cells in the neurogenic niche that was due not to decreased survival but to delayed cell clearance after seizures. These results demonstrate that the efficiency of microglial phagocytosis critically affects the dynamics of apoptosis and urge to routinely assess the microglial phagocytic efficiency in neurodegenerative disorders.

Fig 1. Microglial phagocytic response during in vitro excitotoxic challenge.

(A) Representative epifluorescence (upper panels) and confocal (middle panel) images of the DG in hippocampal organotypic cultures treated with NMDA (50 μM) for 4 h, or fresh media for another 24 h. Normal or apoptotic (pyknotic/karyorrhectic) nuclear morphology was visualized with DAPI (white), microglia by the transgenic expression of fms-EGFP (cyan) and membrane permeability (characteristic of necrotic cells) by PI (red). High magnification inserts show a phagocytosed secondary apoptotic cell (pyknotic, PI⁺; left panel, arrow); primary apoptotic cells (pyknotic, PI^-), phagocytosed or not (arrow and arrow heads, respectively, central panel); and phagocytosed necrotic (nonpyknotic, PI⁺; red arrow, right panel). Scale bars, upper panel = 1 mm, lower panel = 30 μm. (B) Number of dead apoptotic (primary and secondary together) and necrotic cells in 200.000 μm³ of the DG in organotypic slices treated with NMDA. (C) Ph index in organotypic slices (% of apoptotic cells phagocytosed) treated with NMDA. The Ph index in vivo of animals at PND7 (the age at which the organotypic slices were cultured) and PND14 (equivalent to the age of the cultures after 1 wk in vitro) is shown in S1A and S1B Fig (D) Weighted Ph capacity of microglia (in parts per unit, ppu). (E) Histogram showing the Ph capacity of microglia (in % of cells). (F) Number of microglial cells. (G) Ph/A coupling (in fold-change) in organotypic slices treated with NMDA. Bars represent mean ± SEM. * indicates p < 0.05, ** indicates p < 0.01, and *** indicates p < 0.001 by Holm-Sidak posthoc test (after one-way ANOVA was significant at p < 0.05). Only significant effects are shown. Underlying data is shown in S1 Data.

Fig 2. Microglial phagocytic response during in vivo acute and chronic inflammatory challenge.

(A) Experimental design and apoptosis in the DG of c57BL/6 fms-EGFP 1-mo mice injected systemically with LPS (1mg/kg; n = 5) or vehicle (saline; n = 4) 8 h prior to sacrifice. Apoptotic cells were identified by pyknosis/karryorhexis. Fig 2A was generated from data that was originally published as part of [9]. (B) Weighted Ph capacity of microglia (in parts per unit, ppu) in control and LPS mice. (C) Number of microglial cells in control and LPS mice. (D) Ph/A coupling in the 1-mo mouse hippocampus (in fold change) during acute inflammatory challenge. (E) Experimental design and representative confocal z-stacks of the DG of PND21 Swiss mice fed during gestation and lactation with a diet balanced (Ω3 bal; n = 7) or deficient (Ω3 def; n = 7) in the omega 3 polyunsaturated fatty acid, a diet that induces chronic inflammation in the hippocampus. Microglia were labeled with Iba1 (cyan) and apoptotic nuclei were detected by pyknosis/karyorrhexis (white, DAPI). Arrows point to apoptotic cells engulfed by microglia (M). Scale bars = 50 μm; z = 22.5μm. (F) Number of apoptotic (pyknotic/karyorrhectic) cells in mice fed with Ω3 balanced and deficient diets. (G) Ph index in the PND21 hippocampus (in % of apoptotic cells) in mice fed with Ω3 balanced and deficient diets. (H) Weighted Ph capacity of microglia (in ppu) in PND21 mice. (I) Histogram showing the Ph capacity distribution of microglia (in % of cells) in PND21 mice. (J) Total number of microglial cells (Iba1⁺) in PND21 mice. (K) Ph/A coupling in PND21 mice. Bars represent mean ± SEM. * indicates p < 0.05 and ** indicates p < 0.01 by one-tail Student´s t test. Underlying data is shown in S1 Data.

Fig 3. Microglial phagocytosis is impaired early (1 dpi) due to MTLE seizures in vivo.

(A) Hippocampal electroencephalographic recordings of mice injected in the ipsilateral side (I) with KA (50 nL, 20 mM) during status epilepticus (0 dpi) and during a spontaneous seizure occurring in the chronic phase of MTLE (49 dpi). The contralateral hippocampus (C) is shown for comparison purposes. (B) Representative confocal z-stacks of saline and KA (1 dpi) hippocampi labeled with DAPI (nuclear morphology, white), activated caspase 3 (act-casp3⁺, red, for apoptotic cells), and fms-EGFP (cyan, microglia). (C) Number of apoptotic cells (pyknotic/karyorrhectic and act-casp3⁺) in the septal DG (n = 3−9 per time point and treatment). The volume of the septal DG is shown in S3B Fig. (D) Representative confocal image of a nonphagocytosed apoptotic (pyknotic and act-casp3⁺, arrowhead) cell in the SGZ (orthogonal projection, left; and 3-D-rendered image, right). M, microglial cell body. (E) Representative 3-D-rendered confocal z-stack of apoptotic (pyknotic and act-casp3⁺) cells, phagocytosed (arrow) or not (arrowheads) in the septal DG of mice treated with KA at 1 dpi. M, microglial cell body. (F) Representative 3-D-rendered confocal z-stack of an apoptotic (pyknotic), nonphagocytosed cells (arrowhead) in the DG of mice treated with KA at 1 dpi. The arrow points to a semiengulfed apoptotic cell. M, microglial cell body. (G) Ph index in the septal DG (in % of apoptotic cells) after KA. Phagocytosis by astrocytes and neuroblasts is shown in S3C and S3E Fig. (H) Weighted Ph capacity of DG microglia (in ppu). (I) Histogram showing the Ph capacity distribution of microglia (in % of cells) in the DG. (J) Total number of microglial cells (fms-EGFP⁺) in the septal DG. Microglial density is shown in S3A Fig. (K) Ph/A coupling (in fold change) in the septal DG. (L) Histogram showing the distribution of the distance (in μm) of apoptotic cells (in %) to microglial processes. The average distance of apoptotic cells to microglia is shown in S3F Fig. Bars represent mean ± SEM except in L, where they indicate the sum of cells in each distance slot. * indicates p < 0.05, ** indicates p < 0.01, and *** indicates p < 0.001 by Holm-Sidak posthoc test after two-way ANOVA (H–K) or one-way ANOVA (C, G, where a significant interaction time x treatment was found) were significant at p < 0.05. Scale bars = 50 μm (B), 10 μm (D–F). z = 25 μm (B), 13.9 μm (D), 14.1 μm (E), 8.4 μm (F). Underlying data is shown in S1 Data.

Fig 4. Partial recovery of microglial phagocytic efficiency at 3 and 7 dpi is accompanied by multinuclearity and phagoptosis.

(A) Representative confocal z-stacks of saline and KA (3 dpi) hippocampi labeled with DAPI (nuclear morphology, white), activated caspase 3 (act-casp3⁺, red, for apoptotic cells), and fms-EGFP (cyan, microglia). At 3 dpi, KA led to an accumulation of apoptotic cells throughout the DG. (B) Orthogonal projection of a confocal z-stack from the hippocampus of a KA-treated mouse (3 dpi) showing a binucleated microglia cell (fms-EGFP⁺, cyan) undergoing division (phosphohistone 3, PH3, red) and phagocytosing an apoptotic cell (pyknotic by DAPI, white; A). Note the condensed chromosomes in the microglial nuclei, typical of mitosis (M1, M2). (C) Representative confocal z-stack of a large multinucleated phagoptotic microglia (fms-EGFP⁺, cyan) from the hippocampus of a KA mouse (3 dpi). Another example is shown in S4A Fig. (C1). 3-D-rendered image showing the continuum of EGFP through the microglial cytoplasm and within their nuclei. Up to eight nuclei were contained. (C2) Panel showing each nucleus individually. Nuclei 2, 3, and 5 showed condensed chromosomes, characteristic of an ongoing mitosis. Nuclei 4 and 7 were small but not pyknotic; they did not contain EGFP and thus were not microglial but rather surrounded by pouches of microglial cytoplasm. (C3) Orthogonal projections of nuclei 4 and 7 showing their complete engulfment by microglial processes (phagoptosis). (D) Ph index (in % of cells) in the septal hippocampus at 3 and 7 dpi (n = 3–7 per group). (E) Total number of phagocytosed apoptotic cells in the septal hippocampus after treatment with KA (3 and 7 dpi). (F) Total number of phagoptosed nonapoptotic cells in the septal hippocampus after treatment with KA (3 and 7 dpi). Nonphagoptosed cells were found in control animals. nd, not detected. (G) Proportion (in % of cells) between uni- and multinucleated microglia in the hippocampus of KA-treated mice at 3 and 7 dpi. (H) Weighted Ph capacity in the septal hippocampus after treatment with KA (3 and 7 dpi). (I) Weighted PhP (phagoptosis) capacity in the septal hippocampus after treatment with KA (3 and 7 dpi). Bars show mean ± SEM. * indicates p < 0.05, ** indicates p < 0.01, and *** indicates p < 0.001 by Student´s t test (G) or by Holm-Sidak posthoc test after two-way ANOVA (D, E) or one-way ANOVA (I) were significant at p < 0.05. Only significant effects are shown. Scale bars = 50 μm (A), 10 μm (B, C); z = 25 μm (A), 10.5 μm (C). Underlying data is shown in S1 Data.

Fig 5. Microglial phagocytosis impairment is unrelated to monocytes.

(A) CD45 staining in saline- and KA-injected mice at 3 dpi. Cell nuclei are shown in white (DAPI), microglia in cyan (fms-EGFP), and CD45 in red. In control mice, the expression of CD45 was dim, showing diffuse cytoplasmic inclusions within microglia. A CD45⁺ cell is shown engulfing an apoptotic cell (arrow, enlarged). In KA mice, CD45 had a higher and more widespread expression in all microglial cells, including a dividing cell (arrowhead, enlarged). A clear distinction between CD45^high and CD45^low cells was not evident. (B) Flow cytometry analysis of the expression of CD45 in fms-EGFP⁺ hippocampal cells from control and KA-treated mice. Gates for CD45^low (cyan) and CD45^high (red) were defined based on the distribution of the fms-EGFP⁺ cells in control (not injected) mice. 3 dpi after the KA injection, more cells were found in the CD45^high gate, although the fms-EGFP⁺ cells were in fact distributed along a continuum of CD45 expression, all of them with higher expression than control mice. At 7 dpi, the expression of CD45 returned to basal levels. The gating strategy is shown in S4B Fig. (C) Percentage of fms-EGFP⁺ cells that expressed low or high levels of CD45 in control or KA-treated mice determined by flow cytometry (n = 4 per group). (D) Experimental design and representative confocal z-stacks of the hippocampus of CCR2^-/- (CCR2 KO) mice and control WTs (C57BL/6) injected with KA (3 dpi). No obvious differences in the status epilepticus, neuronal damage, microglial morphology, nor in the DG volume (S4C Fig), or neutrophil infiltration were found (S4D and S4E Fig). (E) Number of apoptotic (pyknotic/karyorrhectic) in the septal DG in WT and CCR2 KO mice 3 dpi after KA (n = 4 per group). (F) Ph index in the septal DG (in % of apoptotic cells) in WT and CCR2^-/- mice 3 dpi after KA. (G) Multinuclearity in WT and CCR2^-/- mice. (H) Size of multinucleated cells in WT and CCR2^-/- mice. (I) Weighted Ph capacity in WT and CCR2^-/- mice. Note that the Ph capacity is higher than in our previous time course (Fig 4H), reflecting an increased number of apoptotic cells in this experiment compared to the previous one, possibly because it was performed in different animal facilities. (J) Weighted PhP (phagoptosis) capacity in the septal DG in WT and CCR2^-/- mice. Data are shown as mean ± SEM. * indicates p < 0.05, ** indicates p < 0.01, and *** indicates p < 0.001 by Holm-Sidak posthoc test, after one-way ANOVA was significant at p < 0.05; only significant interactions are shown. Scale bars = 20 μm (A), 50 μm (D); z = 14.7 μm (A), 12.6 μm (D). Underlying data is shown in S1 Data.

Fig 6. Long-term impairment of microglial phagocytosis in mouse and human MTLE.

(A) Representative confocal images of the DG of saline- and KA-injected mice at 4 mpi showing the nuclei (with DAPI, in white) and microglia (Iba1⁺, in cyan). Note the gross dispersion of the DG in KA injected mice (S4F Fig). The number of apoptotic cells in control and KA-treated mice at 4 mpi is shown in S4G Fig. (B) Upper panel: representative confocal z-stack of an apoptotic cell (pyknotic, with DAPI, in white; arrowhead) located nearby a hypertrophic reactive astrocyte (rA; visualized with nestin-GFP⁺, in green) and a microglial cell (M; Iba1⁺, in cyan) at 4 mpi after KA. Lower panel: representative confocal z-stack of an apoptotic cell phagocytosed by microglia at 4 mpi after KA. A representative image of phagocytosis by a reactive astrocyte at 4 mpi after KA is shown in S4H Fig. (C) Ph index in the DG (% of apoptotic cells engulfed). (D) Histogram showing the distribution of the distance (in μm) of apoptotic cells to microglia at 4 mpi after KA (in %). (E) Density of microglial cells (in cells/mm³). (F) Microglial volume (in % of volume of DG occupied). (G) Representative confocal tiled image of a slice of the human hippocampus from an MTLE patient showing cell nuclei (with DAPI, white), neuronal nuclei (NeuN⁺, magenta), and microglia (Iba1⁺, cyan). (H) Representative confocal image of a nonphagocytosed apoptotic cell (pyknotic, with DAPI) adjacent to a microglial process (Iba1⁺) in the hippocampus of an MTLE patient. (I) Representative confocal image of phagocytosis by a ball-and-chain mechanism in the hippocampus from an individual with MTLE. The apoptotic cell (pyknotic, with DAPI in white; arrow) was engulfed by a terminal branch of a nearby microglia (Iba1⁺, cyan). The right panel shows an orthogonal projection of the same cell, where the 3-D engulfment is evident. (J) Representative confocal z-stack of phagocytosis by an aster mechanism in the hippocampus from an individual with MTLE. The apoptotic cell (pyknotic, with DAPI in white; arrow) was engulfed by a mesh of processes from many surrounding microglia (Iba1⁺, cyan; M). The right panel shows an orthogonal projection of the same cell. (K) Representative confocal z-stack of a granule neuron in the DG (NeuN⁺, magenta; arrow) targeted by the processes of several surrounding microglia (Iba1⁺). Nuclei are shown in white (DAPI). The right panel shows an orthogonal projection of the same neuron directly targeted the processes of up to three microglia (M). Another example is shown in S8A Fig and further data in S1 Table. (L) Ph index in the human DG (% of apoptotic cells engulfed). (M) Density of microglial cells (in cells/mm³) in the DG of three hippocampal samples from human MTLE patients. (N) Histogram showing the distribution of the distance of apoptotic cells (in %) to Iba1⁺ microglial processes in the DG of MTLE patients (n = 21 cells from 3 patients). (O) Microglial volume (in % of volume of DG occupied) in the three hippocampal samples from individuals with MTLE. Bars represent mean ± SEM (C, E, F), the individual values of all the pooled cells for each patient (L), the average values for measures in different z-stacks for each patient (M, O), or the sum of cells in each distance slot (D, N). ** represents p < 0.01 by Student´s t test (C, E, F). Scale bars = 50μm (A, K), 10 μm (B, H, I), 1 mm (G), 20 μm (J). z = 25 μm (A), 6.6 μm (B, upper panel), 12.7 μm (B, lower panel), 2.8 μm (H), 2.6 μm (I), 5.2 μm (J), 12 μm (K). Underlying data is shown in S1 Data.

Fig 7. Early phagocytic impairment is related to reduced expression of phagocytosis receptors and reduced motility.

(A) Experimental design and expression of phagocytosis and purinergic receptors by RTqPCR in FACS-sorted microglia from control and KA mice at 1 dpi (n = 3 from 8 pooled hippocampi). HPRT was used as a reference gene. (B) Experimental design and representative projections of 2-photon microscopy images of microglia at t0 (cyan) and 15 min later (magenta) from the DG of controls and KA-treated mice (1 dpi). (C) Motility of microglial processes by 2-photon microscopy in acute slices from CX3CR1^GFP/+ mice after in vivo injection of KA (1 dpi; n = 4–5 cells from 3–4 mice per group). (D) Retraction and protraction of microglial processes by 2-photon microscopy in acute slices from CX3CR1^GFP/+ mice after in vivo injection of KA (1 dpi). (E) Experimental design and representative projections of 2-photon images of microglia at t0 (cyan) and 13.5 min (magenta) in the cortex of controls and KA-treated mice (1 dpi). (F) Motility of microglial processes by 2-photon microscopy in the living cortex of CX3CR1^GFP/+ mice after the injection of KA (1 dpi; n = 6 cells from 3 mice per group). (G) Retraction and protraction of microglial processes by 2-photon microscopy in the living cortex of CX3CR1^GFP/+ mice after the injection of KA. Bars represent mean ± SEM. * indicates p < 0.05, ** indicates p < 0.01, and *** indicates p < 0.001 by Student´s t test (A, C, D). Scale bars = 20 μm (B), 50 mm (E). z = 50 μm (A), 40 μm (B). Underlying data is shown in S1 Data.

Fig 8. Phagocytosis impairment is not directly mediated by glutamate receptors on microglia.

(A, B) Experimental design for RTqPCR expression of KA, NMDA, AMPA and metabotropic, receptor subunits in acutely purified microglia (FACS-sorted) from the hippocampus and the cortex of 2 mo mice (n = 4 samples of 8 pooled hippocampi and cortices each). The relative expression was compared to a positive control, a PND8 hippocampus, except for Grm6, where the retina from a 2-mo mouse was used. L27A was used as a reference gene. Amplification plots and denaturing curves for each target gene are shown in S9 Fig. (C) Experimental design and representative projections of confocal z-stacks of organotypic slices from fms-EGFP mice treated with vehicle (control) or KA (1 mM) for 6 h. The number of apoptotic cells, Ph capacity, and number of microglia in the DG is shown in S10A–S10C Fig. (D) Ph index in the DG organotypic slices (in % of apoptotic cells). (E) Ph/A coupling (in fold-change) in organotypic slices treated with KA. (F) Experimental design to test the effect of KA on microglial phagocytosis in vitro. Primary cultures were pre-treated with KA (1 mM) for 2 h prior to adding apoptotic NE-4C cells (treated with 5 μM CM-DiI for 25 min and 10 μM staurosporine for 4 h). NE-4C cells were left in the culture for another 3 h in the presence or absence of KA. (G) Representative confocal z-stacks of fms-EGFP⁺ microglia phagocytosing apoptotic CM-DiI⁺ NE-4C cells. (H) Percentage of phagocytic microglia in cultures (n = 2 independent experiments in triplicate). Bars represent mean ± SEM. ** indicates p < 0.01 by Student´s t test (H). Scale bars = 30 μm (C, G). z = 6.3 μm (F, J). Underlying data is shown in S1 Data.

Fig 9. Seizures trigger ATP-mediated microglial activation and impairment of phagocytosis.

(A) Experimental design used to induce seizures in acute hippocampal slices with an epileptogenic cocktail that included high K⁺, low Mg²⁺, and 4-AP in ACSF, in the presence or absence of the broad P2X receptor antagonist BBG (5 μM). Seizure activity was recorded in CA1, where it had higher amplitude than in the DG. Top, extracellular recording (mV) and bottom, simultaneous microglia patch clamp recording (pA) before and after seizure induction in the absence or presence of BBG. BBG did not alter seizure frequency or amplitude (S10K–S10M Fig). (B) Patch-clamp currents (in pA) induced in microglia by the seizure activity after the epileptogenic cocktail in the absence (control, n = 18 cells) or presence of the P2X antagonist BBG (n = 11 cells). (C) Latency (in minutes) of the currents induced in microglia by the seizure activity. (D) Experimental design and representative projections of confocal z-stacks of fms-EGFP organotypic slices treated with vehicle (control) or ATP (300 μM and 1 mM) for 4 h. Normal or apoptotic nuclear (pyknotic/karyorrhectic) morphology was visualized with DAPI (white) and microglia by the transgenic expression of fms-EGFP (cyan). The high magnification inserts show single images of apoptotic cells phagocytosed by microglia (arrows) or not phagocytosed (arrowheads). (E) Number of apoptotic cells in a 200.000 μm³ volume containing the DG in organotypic slices treated with ATP (n = 3 slices per group). (F) Ph index in organotypic slices (in %). (G) Weighted Ph capacity (in ppu) in organotypic slices. (H) Histogram showing the Ph capacity of microglia distribution (in % of cells) in organotypic slices treated with ATP. (I) Number of microglia in organotypic slices. (J) Ph/A coupling (in fold-change) in organotypic slices. Bars represent mean ± SEM. * indicates p < 0.05, ** indicates p < 0.01, and *** indicates p < 0.001 by Student´s t test (B) or by Holm-Sidak posthoc test after one-way ANOVA (E–J) was significant at p < 0.05. Scale bars = 30 μm. z = 6.3 μm. Underlying data is shown in S1 Data.

Fig 10. ATP impairs microglial phagocytosis in vivo.

(A) Representative confocal z-stacks of saline, 100 mM ATP and 100 mM ATPγS (2 hpi) DG labeled with DAPI (nuclear morphology, white), activated caspase 3 (act-casp3⁺, red, for apoptotic cells), and fms-EGFP (cyan, microglia). Arrow points to a phagocytosed apoptotic cell, whereas arrowheads point to nonphagocytosed apoptotic cells. Activated-caspase 3 puncta within microglia are labeled with a round-ended arrow. (B, H) Experimental designs (B, 100 mM of ATP and ATPγS, 2 h; H, 10 and 100 mM ATP, 4 h; n = 3–4 per group) and number of apoptotic (pyknotic/karyorrhectic and act-casp3⁺) in the septal DG (n = 3–4 per group). No changes in the volume of the DG were found in either experiment (S11C Fig). (C, I) Ph index in the septal DG (in % of apoptotic cells). (D, J) Weighted Ph capacity of hippocampal microglia (in ppu). (E, K) Histogram showing the Ph capacity distribution of microglia (in % of cells) in the septal DG. (F, L) Total number of microglial cells (fms-EGFP⁺) in the septal DG. (G, M) Ph/A coupling (in fold change) in the septal DG. Bars represent mean ± SEM, * indicates p < 0.05, ** indicates p < 0.01, and *** indicates p < 0.001 by Holm-Sidak posthoc test after one-way ANOVA were significant at p < 0.05. Scale bars = 50 μm, z = 11.9 μm (control, ATP), 9.8 μm (ATPγs). Inserts are single plane images of the corresponding confocal z-stacks. Underlying data is shown in S1 Data.

Fig 11. Microglial phagocytic impairment leads to delayed clearance of apoptotic cells at 1 dpi.

(A) Experimental design used to analyze the survival of 3 do cells after the injection of saline (n = 7) or KA (n = 8) in mice. (B) Representative confocal z-stacks of the DG of control and KA-injected mice (1 dpi). The damage induced by KA was evidenced by the presence of cells with abnormal nuclear morphology (DAPI, white), and the altered morphology of microglia (fms-EGFP⁺, cyan). (C) Representative confocal images of 3 do apoptotic (pyknotic, DAPI, white) cells labeled with BrdU (red; arrows) in the SGZ of the hippocampus of saline and KA-injected mice at 1 dpi. In the saline mouse, the BrdU⁺ apoptotic cell, next to a cluster of BrdU⁺ cells, was phagocytosed by a terminal branch of a nearby microglia (fms-EGFP, cyan), whose nucleus was also positive for BrdU. In the KA mouse, the apoptotic BrdU⁺ cell was not phagocytosed by microglia. A nearby apoptotic cell (BrdU^-; arrowhead) was partially engulfed by microglia. (D) Total number of live 3 do BrdU⁺ cells (nonapoptotic) in the septal hippocampus after treatment with KA. The total number of 3 do and 8 do BrdU⁺ cells by a single BrdU injection in saline and KA-injected mice is shown in S13A and S13B Fig. (E) Total number of apoptotic 3 do BrdU⁺ cells in the septal hippocampus after treatment with KA. (F) Percentage of 3 do BrdU⁺ cells that re-enter cell cycle, assessed by their colabeling with the proliferation marker Ki67 after treatment with KA. Representative confocal z-stacks of BrdU/Ki67 cells are found in S13C Fig. (G) Percentage of apoptotic BrdU⁺ cells over total apoptotic cells in the septal hippocampus. (H) Estimated clearance of apoptotic cells in the septal hippocampus. The total number of apoptotic BrdU⁺ (from E) present in the tissue was added to the number of estimated apoptotic BrdU⁺ cells that had been cleared. In saline mice, this number was calculated using the clearance time formula shown in Methods with a clearance time of 1.5 h [9]. As the total number of cells should be identical in saline and KA mice, the number of cleared apoptotic cells in KA mice was calculated as the difference between the total (in saline) and the number of present apoptotic cells (in KA). From here, we calculated a new clearance time using the same formula as in saline mice, of 6.3 h. (I) Linear regression analysis of the relationship between apoptosis and phagocytosis (Ph index) in saline and KA-injected mice (6 hpi and 1 dpi). (J) Experimental design used to compare SGZ apoptosis induced by KA at 1 dpi in young (2 mo) and mature (6 mo) mice. (K) Representative epifluorescent tiling image of the hippocampus and surrounding cortex of 2 and 6 mo mice injected with KA at 1 dpi stained with the neuronal activation marker c-fos. The same pattern of expression was found in young and mature mice throughout the DG, CA2, CA1 and the above cortex. (L) Representative confocal z-stacks of the apoptotic (pyknotic, white; act-casp3⁺, red) cells in the SGZ of the hippocampus of 2 mo and 6 mo mice injected with KA (1 dpi). The microglial phagocytosis impairment was similar in the two age groups (S13D Fig). (M) Total number of apoptotic cells in the SGZ of 2 and 6 mo mice treated with saline or KA (1 dpi; n = 4–5 per group). Bars show mean ± SEM. * indicates p < 0.05, ** p < 0.01, and *** p < 0.001 by Student´s t test (E, G) or by Holm-Sidak posthoc test after one-way ANOVA (M) was significant at p < 0.05. Scale bars = 50 μm (B), 20 μm (C), 500 μm (K), 25 μm (L). z = 14 μm (B), 12.6 μm (C, sal), 15.4 μm (C, KA), 25 μm (L). Underlying data is shown in S1 Data.

Fig 12. Phagocytosis impairment correlates with inflammation in mouse MTLE.

(A) RTqPCR quantification of a panel of pro- and anti-inflammatory cytokines in the hippocampus of mice injected with saline or KA over a time course. Expression was normalized with the reference gene L27A and expressed as fold change (FC) over the saline injected mice (dashed line). The expression of the cytokines was linearized by a logarithmic transformation; only significant interactions are shown. (B) Linear regression analysis of the relationship between the average expression of tissue cytokines (IL-1β, IL-6, MIC, TGFβ) and average Ph index in KA-injected mice along the time course. The Ph index explained a large percentage of the variation of the expression of these cytokines, although significance at p = 0.05 was only reached for MIC-1 likely due to the small number of time points analyzed. (C) RTqPCR quantification of a panel of pro- and anti-inflammatory cytokines in FACS-sorted fms-EGFP⁺ microglia. In A and C data are shown as mean ± SEM. a, * indicates p < 0.05; b, ** indicates p < 0.01; and c,*** indicates p < 0.001 by Holm-Sidak posthoc test compared to their respective time point controls, after one-way ANOVA was significant at p < 0.05. Underlying data is shown in S1 Data.

Fig 13. Microglial phagocytosis/apoptosis coupling in health and disease.

In physiological conditions as well as during excitotoxicity and inflammation, microglial phagocytosis is tightly coupled to apoptosis due to “find-me” signals released by apoptotic cells, such as ATP. Microglia display a combination of three adaptation strategies to boost their phagocytic efficiency: recruit more phagocytic cells, increase the phagocytic capacity per cell, and/or increase the number of cells. In contrast, neuronal hyperactivity induced by seizures leads to a widespread release of ATP, among other possible signals, and interferes with the ability of microglia to recognize and engulf apoptotic cells, resulting in their delayed clearance. Because phagocytosis is actively anti-inflammatory, the phagocytosis impairment was associated with the production of proinflammatory, epileptogenic cytokines.