CCL5 promotion of bioenergy metabolism is crucial for hippocampal synapse complex and memory formation

Abstract

Glucoregulatory efficiency and ATP production are key regulators for neuronal plasticity and memory formation. Besides its chemotactic and neuroinflammatory functions, the CC chemokine--CCL5 displays neurotrophic activity. We found impaired learning-memory and cognition in CCL5-knockout mice at 4 months of age correlated with reduced hippocampal long-term potentiation and impaired synapse structure. Re-expressing CCL5 in knockout mouse hippocampus restored synaptic protein expression, neuronal connectivity and cognitive function. Using metabolomics coupled with FDG-PET imaging and seahorse analysis, we found that CCL5 participates in hippocampal fructose and mannose degradation, glycolysis, gluconeogenesis as well as glutamate and purine metabolism. CCL5 additionally supports mitochondrial structural integrity, purine synthesis, ATP generation, and subsequent aerobic glucose metabolism. Overexpressing CCL5 in WT mice also enhanced memory-cognition performance as well as hippocampal neuronal activity and connectivity through promotion of de novo purine and glutamate metabolism. Thus, CCL5 actions on glucose aerobic metabolism are critical for mitochondrial function which contribute to hippocampal spine and synapse formation, improving learning and memory.

Fig. 1. Memory, cognitive ability and hippocampal neuron activity are impaired in mice lacking CCL5.

a The exploration tracks of WT and CCL5-KO mice in the box with objects in the novel object recognition (NOR) test. The exploration time (b) and preference for novel objects (c) were significantly lower in CCL5-KO mice. (Mann–Whitney test). d The movement track of WT and KO mice in the Barnes Maze (BM) on training day 4. The time to find target (e), missed target (f), and path length (g) were higher in CCL5-KO mice compared to age-matched WT mice; the time spent in target quadrant was less in CCL5-KO (h). (Day1 was the first day of training, which is not shown.) (Two-way ANOVA Bonferroni’s multiple comparisons test between two groups. Unpaired t-test was used to compare WT and KO at same day.) i, j Stimulation of hippocampal brain slices from both WT and CCL5-KO. The slope of fEPSPs was reduced in CCL5-KO compared to WT and marked reduction of LTP was found in CCL5-KO. (Two-Way ANOVA Bonferroni’s multiple comparison test, *p < 0.05; **p < 0.01; ***p < 0.001. WT: n = 10 slices of six mice, KO: n = 5 slices of two mice.). k, l Golgi staining identified neurite structure and spines; the neurite intersections and neurite length, analyzed by Sholl analysis, were reduced in KO mice. (Arrows point to the spines in WT and CCL5-KO mice) (k: Two-Way ANOVA, l: Unpaired t-test. WT: n = 14 of three mice, KO: n = 16 of three mice.) m Synaptic proteins −NR2B and PSD95 were reduced in hippocampal synaptosome fractions. Synaptophysin was used as a synaptosome control. (Unpaired t-test. n = 4–6.).

Fig. 2. Reintroducing CCL5 into hippocampus improved memory performance, synapse-related protein expression, and neuron connectivity in CCL5-KO mice.

a The scheme of experimental design. b The distributions of AAV-CCL5 and AAV-mCherry in mouse hippocampus. Hippocampal regional distributions and colocalization of CCL5 (green), mCherry (red), and neuron marker––NeuN (Cyan) are enlarged in (c). (Scale bar = 100 μm). d The preference for new objects increased in CCL5-KO mice receiving AAV-CCL5 after 2–3 months compared to mice receiving AAV-mCherry alone. (Paired t-test to compare mice before and after AAV injection. Mann–Whitney test for AAV-mCherry and AAV-CCL5 groups.) e, h The movement track of mice in the BM on training day 4. Target box labeled in black-circle in red quadrant and blue quadrant indicated the hidden box in the previous month. f, i The time to find the target in AAV-mCherry and AAV-CCL5 injected mice after 2 and 3 months; and (g, j) the mistarget number in both groups of mice. (Two-way ANOVA between groups. Unpaired t-test for single time point comparisons.) k The synaptic proteins phosphor-NR2B-S1303, PSD95, and GAP43 were increased in AAV-CCL5 injected mouse hippocampal synaptosome fractions. (Numbers indicate three separate mice; L: left and R: right hippocampus with AAV expression. Mann–Whitney test, n = 4–6.) l Color labels the regions of interest (ROI), including hippocampus (Hipp, Orange), dentate gyrus (DG, Red), CA3 (Cyan), medial prefrontal cortex (PFC, Yellow), and somatosensory cortex (SC, Green). m The FA values in Hipp, DG, CA3, and PFC regions were significantly increased in the AAV-CCL5 injected group compared to AAV-mCherry injected, age-matched mice without surgery (7-months old) and also before surgery (4-months old) groups (Dashed line show the average value in WT, in Fig. 5). (Mann–Whitney test.).

Fig. 3. CCL5 participates in brain glucose aerobic metabolism and mitochondrial functions.

a MicroPET images averaged over all mice from both WT and CCL5-KO, and the average regional [18 F] FDG uptake is presented as standardized uptake value (SUV). Regions of interest (ROIs), defined on a series of coronal sections of the co-registered PET-MRI image of the mouse brain, were used to determine [18 F] FDG uptake in bilateral medial prefrontal cortex (PFC) and hippocampus (Hipp). Precise localization of ROIs was aligned on the brain atlas to the matching structures on the brain images (PFC-orange; Hipp-blue). b PET images acquired from WT and CCL5-KO mice were compared. (** indicates p < 0.01, Kruskal–Wallis test compared with the WT group. n = 4 for each group.) The energy utilization rate for oxygen consumption (OCR) (c) and glycolysis - extracellular acidification rate (ECAR) (d) in WT and KO hippocampus and cortex regions were compared. (Mann–Whitney test, n = 6–8. NS: No significant difference.) e–g Electron microscopic images of mitochondrial structure revealed outer membrane abnormalities in KO mouse brains (f, arrowheads) and loss of inner cristae (arrows in g) compared to normal mitochondria in WT (e, arrowheads) (Scale bar = 5 μm). h Levels of mitochondrial proteins––COX and mitochondria biogenesis protein––PGC1α in CCL5-KO hippocampus synaptosome fractions but not in prefrontal cortex were lower than in WT. (Mann–Whitney test, n = 5–7. Numbers indicated independent mice.) i–l The ultrastructure of mitochondria in mouse hippocampal DG and CA1 regions after AAV-mCherry and AAV-CCL5 expression studied with electron microscopy. The outer membrane and mitochondrial size were increased in KO mice receiving AAV-CCL5 (k–l: Arrows point to the elongated mitochondria and dense inner cristae) compared to mice receiving AAV-mCherry (i–j: Arrowheads point to the loss of outer membrane and inner cristae.) (2–3 mice were analyzed and 20 pictures were taken for each brain area). m The mitochondrial proteins COX and TOM20 were increased on both sides of AAV-CCL5 injected mice in hippocampal and PFC synaptosome fractions. PGC1-α was increased in AAV-CCL5 expressing hippocampus but not in PFC. (Numbers indicated separate mice; L: left and R: right hippocampus with AAV. Mann–Whitney test, n = 4–6.) n The respiratory activity of mitochondria in mouse brain tissues was measured after mice were injected with AAV-mCherry or AAV-CCL5 for 3 months. The OCR and ECAR increased in AAV-CCL5 expressing hippocampal DG, CA1, and cortex regions. (Dashed line point to the average value in WT-mCherry in Fig. 5; Mann–Whitney test, n = 6.) o The diagram of pentose phosphate pathway (PPP) and purine metabolism-related molecules and enzymes. p The key enzyme for PPP––NADPH activity in WT and KO mice hippocampus. (Mann–Whitney test, n = 10.) q Quantitative PCR analysis of the de novo purine metabolism enzyme 5’-neucleotide (5’-NT) and salvage purine synthesis enzyme adenosine deaminase (Ada) and hypoxanthine-guanine phosphoribosyl transferase (HPRT) between WT and KO mice or CCL5-KO mice receiving AAV-CCL5 or mCherry (r). (Data was normalized to contralateral side and analyzed by unpaired t-test.).

Fig. 4. CCL5 activation of the PI3K/Akt pathway promotes synaptogenesis and ATP generation in hippocampal neurons.

a The schematic of CCL5 and AICAR treatments in cultured hippocampal neurons. Hippocampal neurons were treated with CCL5 (0, 10, 50, 100, 500 pg/ml) and AICAR (25 mM) after 4 days in culture for 3 days. b Synapses labeled by synaptophysin (red) in hippocampus neurons (Tau in green), which was increased by CCL5 treatment. c AICAR treatment reduced synaptic structures; CCL5 treatment blocked AICAR-induced synapse reduction. (DAPI labeled nuclei in blue, Scale bar = 50 μm). The quantification of synapse number––synaptophysin (d) and intensity (e) under CCL5 and AICAR treatments. (One-way ANOVA). The co-localization of pre-synaptic protein––synaptophysin and post-synaptic protein––PSD95 upon CCL5 (f), and AICAR alone, or AICAR plus CCL5 (g). h The quantification of synaptic complex density upon CCL5 and AICAR (Mann–Whitney test for AICAR treatment and one-way ANOVA for CCL5 treatment). i–j The protein levels of synaptic proteins PSD95, GAP43, and synaptophysin and the mitochondrial protein––COX with CCL5 administration (i) and AICAR plus CCL5 (j). (k, l) CCL5 reduced the elevation of phosphor-AMPKα by AICAR and increased cellular ATP levels (m, n) in primary hippocampal neurons. (m–n: one-way ANOVA; n: unpaired t-test, AICAR compared to control. n = 4–6.) (o) CCL5 treatment activated PI3 Kinase signaling molecules––p110α, p85-p55, GSK3β to Akt in a dose dependent matter. p Pan-PI3K inhibitor––Wortmannin and P110α specific inhibitor––BYL179 blocked CCL5 induced PI3K pathway activation. q Synapse localization of Akt S473 (green) under normal neuron basal medium––control compared to treatments with CCL5, AICAR, the CCL5 antagonist––MetCCL5, the PI3K inhibitor––Wortmannin, or BYL179 in primary hippocampal neurons. Tau (red) labeled neuritis.

Fig. 5. Expressing CCL5 in WT mouse hippocampus enhanced mouse memory performance.

a The preference for a new object in the NOR test was increased in AAV-CCL5 injected WT mice after 2 months. (Mann–Whitney test). b The movement track of mice in the Barnes Maze (BM) on training day 4 in AAV-mCherry and AAV-CCL5 injected mice. (Target box labeled in black-circle in red quadrant and blue quadrant indicated the hidden box in last month.) c Mice spent less time in finding targets and made fewer mistakes, mistargets in the BM task, 2 months after receiving AAV-CCL5. (Two-ways ANOVA). d The levels of neuron plasticity related proteins NR2B and GAP43 in mouse hippocampus synaptosome fractions were increased in AAV-CCL5 injected right (R) hemisphere but not PSD95, (f) PGC1-α and COX. (Numbers indicate independent mouse. R: right, injection side, L: left, contralateral side.) e The in vivo structural changes were evaluated by MRI-DTI. The FA value in DG, CA3, and PFC significantly increased in AAV-CCL5 injected WT mice after 3 months compared to AAV-mCherry injected mice, but not in whole hippocampus and somatosensory cortex (SC). (Mann–Whitney test, n = 6–7.) (g, h) The respiratory activity in OCR increased in the hippocampal DG and CA1 in WT mice but not ECAR-glycolysis. (Mann–Whitney test, n = 6.) (i) AAV-CCL5 administration increased 5ʹ-NT gene expression in WT mouse hippocampus and reduced Ada gene expression but HPRT was not changed. (Mann–Whitney test, n = 4–6.NS: No significant difference.).