Lipopolysaccharide‐induced alteration of mitochondrial morphology induces a metabolic shift in microglia modulating the inflammatory response in vitro and in vivo

Accumulating evidence suggests that changes in the metabolic signature of microglia underlie their response to inflammation. We sought to increase our knowledge of how pro‐inflammatory stimuli induce metabolic changes. Primary microglia exposed to lipopolysaccharide (LPS)‐expressed excessive fission leading to more fragmented mitochondria than tubular mitochondria. LPS‐mediated Toll‐like receptor 4 (TLR4) activation also resulted in metabolic reprogramming from oxidative phosphorylation to glycolysis. Blockade of mitochondrial fission by Mdivi‐1, a putative mitochondrial division inhibitor led to the reversal of the metabolic shift. Mdivi‐1 treatment also normalized the changes caused by LPS exposure, namely an increase in mitochondrial reactive oxygen species production and mitochondrial membrane potential as well as accumulation of key metabolic intermediate of TCA cycle succinate. Moreover, Mdivi‐1 treatment substantially reduced LPS induced cytokine and chemokine production. Finally, we showed that Mdivi‐1 treatment attenuated expression of genes related to cytotoxic, repair, and immunomodulatory microglia phenotypes in an in vivo neuroinflammation paradigm. Collectively, our data show that the activation of microglia to a classically pro‐inflammatory state is associated with a switch to glycolysis that is mediated by mitochondrial fission, a process which may be a pharmacological target for immunomodulation.

and was routinely greater than 99%. All incubations were performed at 37 o C in a 167 humidified atmosphere containing 5% CO 2 and 95% air. images were acquired as z-stacks with three angles and five phases in each plane and 194 the z-step between planes was 3.30 nm. SR-SIM processing was performed using the 195 Zeiss Zen software package. 3D rendering was done using Volocity 6 (Perkin-Elmer) and 196 figures were compiled using Photoshop CC software (Adobe Systems, San Jose, CA).

205
(ECAR) 206 Real-time measurements of oxygen consumption rates, and extracellular acidification 207 rates, a measure of lactate production, were performed on an XFe96 Seahorse 208 extracellular flux analyser (Seahorse Biosciences, North Billerica, MA (2-DG) was injected at a final concentration of 50mM to measure the non-glycolytic 240 acidification. Each step had three cycles; each cycle consisted of 3 min mixing, 2 min 241 incubation and 3 min measurement. All experiments were run in three replicates with 3-4 242 sample per replicates. Cell counts were used to normalize OCR and ECAR.

341
Mitochondrial morphology was examined in primary microglia cells cultured from Cox8-342 EGFP mice exposed to 50 or 100ng/mL LPS using 3D SR-SIM microscopy. The number 343 of fragmented mitochondria was significantly increased in microglia cells stimulated with 344 100ng/ml LPS for 24h (Fig 1c), and elongated and tubular mitochondria were decreased 345 compared with untreated controls (Fig 1g). These findings are in line with previous studies with 50ng/ml LPS for 24h (Fig 1b,g). whereas leak-driven OCR significantly increased with exposure to 50ng/mL of LPS (Fig. 358 2f-h). The ECAR parameters (glycolysis, glycolytic capacity and glycolytic reserve) were 20 increased following exposure to 50ng/ml LPS for 6-24hrs compared to controls  n). These results show that a moderate dose of LPS increases both OCR and glycolysis.

361
Exposure to 100ng/mL of LPS for 6h resulted in an increase in basal OCR, ATP-linked 362 OCR and leak-linked OCR compared to controls . In contrast, there was a 363 significant decrease in basal OCR and ATP linked OCR at 24h after 100ng/mL LPS (Fig. 364 2r,s,u). FCCP-induced maximal OCR and SRC significantly decreased at 24h 100ng/mL 365 LPS . Glycolytic parameters increased with 100ng/ml LPS exposure for 3-24h 366 compared with controls ( Fig.2w-y). The overall decrease in OCR and increase in ECAR 367 parameters with 100ng/ml LPS for 24h indicates a metabolic switch from OXPHOS to 368 glycolysis.  high (100 ng/ml) dose of LPS induced an increase in fragmented mitochondria (Fig. 3b). 377 We examined the effect of pharmacologically blocking mitochondrial fission in LPS-378 exposed microglia cells cultured from Cox8/EGFP mice by pre-treatment with 25 µM 1 pre-treatment and normalized mitochondrial morphology (Fig. 3c). Mdivi-1 treatment 382 before LPS exposure reduced the number of fragmented mitochondria and increased the 383 number of tubular and elongated mitochondria to control levels (Fig. 3d) .

388
Mdivi-1 pre-treatment in cells exposed to LPS (100ng/ml for 6h) exhibited a decrease in 389 the level of basal respiration and ATP-linked OCR to control levels compared to LPS 390 treated cells (Fig. 4c-d). Conversely, Mdivi-1 treatment in cells exposed to 100ng LPS for 391 24h led to an increase in basal and ATP-linked OCR compared to non-treated LPS 392 exposed cells . Mdivi-1 treatment also increased FCCP-induced maximal OCR 393 at 24h and leak-driven OCR compared to LPS exposed cells at both time points  f). Administration of Mdivi-1 in combination with LPS normalized the spare respiratory 395 capacity (Fig. 4g). ECAR measurements showed that glycolysis and glycolytic capacity 396 was significantly reduced to control levels in Mdivi-1 treated cells at 6 and 24h 100ng/ml

399
To show how LPS activation was inducing an inflammatory reaction in the primary 400 microglia and to test whether this was effected by Mdivi-1 we measured cytokine and 401 chemokine response in microglia conditioned media after treatment with of LPS and or 402 Mdivi-1 (supporting information Fig. S1 and S2). As expected both doses, of LPS led to 403 a significant up-regulation of essentially all cytokines and chemokines compared to 404 controls. In general there was much higher cytokine production in microglia exposed to

413
Succinate is a well-established pro-inflammatory metabolite that is known to accumulate 414 during LPS induced macrophage activation (Mills et al., 2016) but the role of succinate 415 during microglia activation needs further investigation. We found that LPS (100ng/ml) 416 resulted in a significant increase of succinate (Fig.6a)

Mdivi-1 treatment attenuated LPS induced increase of mitochondrial membrane
437 potential 438 Our data suggest that after LPS (100ng/ml) exposure for 24h microglia mainly depended 439 on glycolysis for energy production. Therefore, we investigated the mitochondrial 440 membrane potential using the mitochondrial membrane potential probe JC-1 in these 441 conditions. We found that there was a consequent elevation of mitochondrial membrane 442 potential and treatment with Mdivi-1 significantly reduced mitochondrial membrane 443 potential (525/565 nm) ratio compared to LPS treated group (Fig. 8). immunomodulation, but had no effect on IGF1 gene expression, and only partly recovered

518
Supporting our in vitro data mdivi-1 was able to reduce the expression of genes 519 associated with classically pro-inflammatory genes, and the anti-inflammatory activation 520 state, which is associated with the in vivo inflammatory reaction.

521
Previous work with BV2 demonstrated that LPS causes an inhibition of OXPHOS 522 (Voloboueva, Emery, Sun, & Giffard, 2013). However, this study used a very high dose 523 of LPS (1µg/ml) which is shown to elicit mitochondrial toxicity (Ahn et al., 2012). We demands. Glycolysis may also facilitate in cytokine production by producing intermediate 547 metabolites (Mills et al., 2016). A previous study found that glycolysis was required to 548 produce optimal IFN-γ during T cell activation and is translationally regulated by the 549 binding of the glycolysis enzyme GAPDH to IFN-γ mRNA (Chang et al., 2013). with LPS for 24h. The data are for at least 12 cells per condition in three independent experiments. Bar graphs expressed as mean ± SEM. ***P ≤ 0.001; student-t test calculating the difference between control and LPS treated groups.