DFP exposure induced dramatic phenotypic remodelling of microglia
To investigate the consequences of acute DFP poisoning on the physiology of microglia, we first made use of our well-established zebrafish model of DFP-intoxication  and the zebrafish transgenic line Tg[mpeg1:mCherryF], which enables live imaging of these cells in embryos from 3 days post-fertilization (dpf) onward  (Fig. 1A). In close agreement with previous data [18, 19], in 5 dpf larvae, almost all microglial cells showed a highly branched morphology, characteristic of the so-called “resting” state, with cells showing several long branches (Fig. 1B–B''). In contrast, in age-matched individuals exposed for 6 h to 15 µM DFP, microglial cells displayed a much more rounded morphology and a marked decrease in both the number and length of their branches (Fig. 1C–C'').
Precise measurements of several morphological parameters using the Imaris software (Bitplane) confirmed that DFP poisoning caused deep morphological changes of microglial cells. Their mean sphericity (Sp) (see Materials and methods) was significantly increased (0.79 ± 0.01 vs 0.63 ± 0.01, p < 0.0001) (Fig. 1D) and their average surface area (S) was decreased in DFP-treated larvae compared to that observed in controls (840 µm² ± 20 vs 1118 µm² ± 27, p < 0.0001) (Fig. 1E). In contrast, the average volume (V) of the cells was roughly similar in DFP-treated and control individuals (1654 µm3 ± 56 vs 1724 µm3 ± 58, p = 0.64) (Fig. 1F). In DFP-exposed larvae, microglia also displayed a reduced number of branches (NB) (0.6 ± 0.03 vs 4.5 ± 0.1, p < 0.0001) (Fig. 1G), a decreased mean branch length (ML) (17 µm ± 0.1 vs 21 µm ± 0.1, p < 0.001) (Fig. 1H), and thus a decreased mean total branch length (TL) (12.1 µm ± 1.3 vs 80.0 µm ± 2.1, p < 0.0001), compared to those in controls (Fig. 1I). A 3D Sholl analysis, a method used to quantify both the extent and complexity of cell branches (Imaris software, Bitplane), fully confirmed that DFP exposure induced major microglial phenotypic changes, with cells showing a decrease in both branch number and length (Fig. 1J).
To further characterize the consequences of acute DFP poisoning on microglial morphological changes, we performed a cluster analysis of these cells in control and DFPtreated larvae, based on the five morphological parameters that significantly changed following DFP exposure, namely Sp, S, NB, ML and TL, as indicated above (see Materials and methods). Results showed that microglial cells could be clustered into three distinct populations in controls (Fig. 1K). The largest cluster comprised 47.1% of the cells and included microglia showing highly branched morphology and low sphericity, likely corresponding to “resting” microglia. The smallest cluster, corresponding to 14.7% of the cells, represented microglia with a low process number and a high sphericity, likely corresponding to M1-type “activated” microglia. The third cluster, which contained 38.2% of the cells, comprised microglia displaying both an intermediate branch number and an intermediate sphericity. We refer to these cells as “intermediate” microglia. In contrast, only two main clusters were observed in DFP-treated larvae (Fig. 1K). The larger one, which contained 73.1% of the cells, included microglia resembling “activated” microglia. The other cluster, comprising 25% of the cells, contained microglia showing the “intermediate” phenotype as defined above. It is of note that fewer than 2% of the microglia showed a "resting" phenotype in DFP-exposed larvae, compared to 47% in controls, suggesting that DFP caused a massive, brain-wide microglial activation.
Because the above data indicate that microglia displayed deep morphological changes after 6 h of acute DFP exposure, we next investigated the dynamics of these changes using real-time confocal imaging on live 5 dpf Tg[mpeg1:mCherryF] larvae during 6 h of exposure to 15 µM DFP. In agreement with our previous data , prior to DFP addition, microglia were highly branched with several long processes that permanently scan their environment and neighbour cells, and during the first 2 h of DFP exposure, no significant morphological changes could be detected (Supplementary Fig. 1 and supplementary Video 1). In contrast, from 2.5 h of exposure, clear remodelling of the cells was observed (Supplementary Fig. 1 and supplementary Video 1), which included an increased sphericity and a decrease in the length of the processes. After 3.5 h of exposure, almost all the cells displayed a phenotype resembling “activated” microglia (Supplementary Fig. 1 and supplementary Video 1).
Visual examination of supplementary Video 2 suggested that DFP poisoning markedly increased the distance travelled by microglial cell bodies. To confirm this observation, we tracked the distances travelled by each cell body during an approximately 30-minute period (Supplementary Fig. 2A), before exposure (Supplementary Fig. 2B and Supplementary Video 1) and after 5.5 h of DFP exposure (Supplementary Fig. 2C and Supplementary Video 2). Measurements of track lengths indicated that microglial cell bodies moved more in DFP‑treated larvae, with a 0.86 ± 0.27 µm/minute increase compared to controls (p < 0.001) (Supplementary Fig. 2D).
DFP exposure induced microglia-mediated overexpression of inflammatory cytokines
Microglia phenotypic changes observed in DFP-exposed larvae were highly reminiscent of those of "activated“ M1-type microglia observed in human epileptic brains , but also in DFP-exposed rats .
To confirm that the remodelling of microglia observed in DFP-treated larvae does reflect an M1-like inflammatory type of activation of these cells, we next investigated, by qRT-PCR analysis of whole-body RNAs, the expression levels of transcripts encoding a set of pro‑inflammatory (Il1β and Il8) and immuno-modulatory cytokines (Il4), before and at different time points during a 6-hour exposure to either 1% DMSO or 15 µM DFP. In close agreement with the results obtained in preclinical models of DFP intoxication [22–24], our data first revealed a massive expression of il1β (fold change (fc): 413 ± 67, p < 0.0001) and il8 (fc: 44 ± 6, p < 0.0005), together with a significantly increased expression of il4 transcripts (fc: 2.9 ± 0.4, p < 0.001) after 6 h of DFP exposure (Fig. 2A). Moreover, precise timing of the expression of these RNAs during DFP exposure indicated that significantly increased expression of il8 (fc: 1.6 ± 0.2, p < 0.05) and il1β (fc: 10.4 ± 2.4, p < 0.01) was observed after 1 h and 2 h of exposure, respectively, prior to increasing gradually after 3 h (fc: 3.3 ± 0.7, p < 0.001 and fc: 18.5 ± 3, p < 0.001), 4 h (fc: 26 ± 9, p < 0.001 and fc: 217 ± 28, p < 0.001) and 5 h of exposure (fc: 82 ± 9.7, p < 0.001 and fc: 125 ± 15.79, p < 0.001) (Fig. 2). In contrast, increased expression of il4 transcripts was delayed compared to that of il1β and il8, starting after 5 h of exposure to DFP (fc: 5.3 ± 0.56, p < 0.001) (Fig. 2).
To verify that the massive overexpression of pro-inflammatory cytokines observed in DFP-exposed larvae did reflect brain inflammation, we next investigated, by qRT-PCR analysis of RNAs extracted from dissected brains, the expression levels of transcripts encoding the same set of cytokines (Il1β, iI8, and Il4), in larvae exposed for 6 h to either DMSO or DFP. Results confirmed that a 6-h DFP exposure induced massive expression of il1β (fc: 189 ± 37, p < 0.01) and il8 (fc: 42 ± 19, p < 0.05), and increased expression of il4 transcripts (fc: 2.9 ± 0.7, p < 0.05) in the brain of exposed larvae (Fig. 3), confirming that DFP exposure induced bona fide brain inflammation.
Kinetics of neuronal activity in DFP-exposed larvae
We previously showed that larvae exposed to 15 µM DFP displayed massive neuronal hyperactivation from 1 h to 1.5 h of exposure to DFP . To further investigate this point, we studied, by qRT-PCR analysis of RNAs extracted from larvae at different time points of DFP exposure, the temporal expression profile of fosab, an immediate early gene (IEG) whose expression is an early, sensitive marker of neuronal activation, especially epileptiform seizures . A significantly increased expression of fosab transcripts was detected in larvae exposed to DFP for 1 h, (fc: 2.89 ± 0.4, p > 0.01), which then gradually increased over the next 5 h (fc: 21.7 ± 2.9, p < 0.001) (Fig. 4). This result suggests, in agreement with our previous calcium imaging data and the results in rodent models [8,14,23] that neuronal hyperactivation in zebrafish larvae is an early consequence of DFP poisoning that starts as early as 1 h post-exposure.
Inflammatory cytokines and neuronal activity in DFP-treated larvae without microglia
Two glial cell types mediate brain inflammation, including that induced by acute DFP exposure: microglial cells and astrocytes [26,27]. Accordingly, we next undertook to assess the role played by microglia in the neuroinflammatory process induced by DFP exposure. To this end, we analysed the expression levels of the same three cytokine RNAs in larvae fully devoid of microglia as the result of morpholino-oligonucleotide-mediated inactivation of the pU.1 gene, hereafter referred to as pU.1 morphants . Results indicated that il1β and il8 transcripts were still overexpressed in pU.1 morphants exposed for 6 h to DFP, albeit at markedly lower levels than observed in their wild-type counterparts (fc: 134 ± 44.3 vs 413 ± 94, p < 0.01, and fc: 15.7 ± 4.7 vs 44 ± 6, p < 0.001 respectively), suggesting that while microglial cells are important players in DFP-induced neuroinflammation, other cells are also involved in the process. In contrast, no overexpression of il4 was detected in pU.1 morphants exposed to DFP (fc: 1.2 ± 0.2, p = 0,81) (Fig. 5), supporting the hypothesis that microglia were the main cell type overexpressing this cytokine following DFP exposure.
DFP-induced neuronal hyperactivation was markedly reduced in larvae without microglia
It has long been known that brain inflammation creates an environment that favours neuronal hyperexcitation and epileptogenesis . Therefore, we next set out to evaluate the consequences of microglia activation and subsequent inflammation on the neuropathological processes induced by DFP poisoning. For this purpose, we used pU.1 morphants lacking microglia and showing reduced inflammatory response to DFP to study the consequences of the absence of microglia on DFP-induced neuronal activation as revealed by fosab transcript expression.
The results indicated that fosab RNAs were still overexpressed in pU.1 morphants exposed to DFP albeit at significantly reduced levels when compared to that observed in their wild-type counterparts (fc: 7.3 ± 1.5 vs 22.2 ± 2.9, p < 0.001) (Fig. 6), suggesting that DFP-induced neuronal hyperactivation was markedly reduced in larvae without microglia.