Inhibition of microglia overactivation restores neuronal survival in a mouse model of CDKL5 deficiency disorder

Background CDKL5 deficiency disorder (CDD), a severe neurodevelopmental disorder characterized by early onset epilepsy, intellectual disability, and autistic features, is caused by mutations in the CDKL5 gene. Evidence in animal models of CDD showed that absence of CDKL5 negatively affects neuronal survival, as well as neuronal maturation and dendritic outgrowth; however, knowledge of the substrates underlying these alterations is still limited. Neuroinflammatory processes are known to contribute to neuronal dysfunction and death. Recent evidence shows a subclinical chronic inflammatory status in plasma from CDD patients. However, to date, it is unknown whether a similar inflammatory status is present in the brain of CDD patients and, if so, whether this plays a causative or exacerbating role in the pathophysiology of CDD. Methods We evaluated microglia activation using AIF-1 immunofluorescence, proinflammatory cytokine expression, and signaling in the brain of a mouse model of CDD, the Cdkl5 KO mouse, which is characterized by an impaired survival of hippocampal neurons that worsens with age. Hippocampal neuron survival was determined by DCX, NeuN, and cleaved caspase-3 immunostaining in Cdkl5 KO mice treated with luteolin (10 mg/kg), a natural anti-inflammatory flavonoid. Since hippocampal neurons of Cdkl5 KO mice exhibit increased susceptibility to excitotoxic stress, we evaluated neuronal survival in Cdkl5 KO mice injected with NMDA (60 mg/kg) after a 7-day treatment with luteolin. Results We found increased microglial activation in the brain of the Cdkl5 KO mouse. We found alterations in microglial cell morphology and number, increased levels of AIF-1 and proinflammatory cytokines, and activation of STAT3 signaling. Remarkably, treatment with luteolin recovers microglia alterations as well as neuronal survival and maturation in Cdkl5 KO mice, and prevents the increase in NMDA-induced cell death in the hippocampus. Conclusions Our results suggest that neuroinflammatory processes contribute to the pathogenesis of CDD and imply the potential usefulness of luteolin as a treatment option in CDD patients. Supplementary Information The online version contains supplementary material available at 10.1186/s12974-021-02204-0.

Changes in neuronal morphology such as dendritic branching and stability of dendritic spines have been consistently reported in Cdkl5 KO mice [6,7]. Cdkl5 KO mice showed a reduction in dendritic length of cortical and hippocampal pyramidal neurons [6,11,15,16] and changes in the maturation and stability of dendritic spines, as well as in the density of PSD-95 dendritic clusters in several brain structures [10,[15][16][17][18].
Interestingly, recent evidence has shown that CDKL5, in addition to neuronal maturation and dendritic outgrowth, also affects neuronal survival [11,19,20]. Cdkl5 KO mice are characterized by an increased rate of apoptotic cell death in the hippocampal dentate gyrus that causes a reduction in the nal number of granule neurons [11] and by accelerated neuronal senescence/death during aging [21]. Moreover, hippocampal neurons of Cdkl5 KO mice exhibit increased susceptibility to neurotoxic/excitotoxic stress [19,20], indicating that absence of Cdkl5 increase neuronal vulnerability.
Neuroin ammatory processes are known to contribute to neuronal dysfunction and death. When overactivated in response to neuronal damage and genetic or environmental factors, microglia, the brain macrophages [22], cause widespread damage to neighboring neurons. Indeed, reactive microglia kill neurons by producing neurotoxic factors and pro-in ammatory molecules such as tumor necrosis factorα (TNF-α) and interleukin-1β and − 6 (IL-1β, IL-6) [23,24]. Remarkably, over-activated microglia have been described in several neurodegenerative diseases such as Alzheimer's, Parkinson's, and Huntington's diseases and Amyotrophic lateral sclerosis [25][26][27], suggesting that active neuroin ammation may account for the compromised neuronal survival observed in these pathologies. Interestingly, active neuroin ammation could account for the compromised brain development observed in neurodevelopmental disorders such as Down syndrome, autism-spectrum disorders (ASD), and Rett syndrome [28][29][30], by damaging synaptic connectivity.
Recently, a major cytokine dysregulation proportional to clinical severity, in ammatory status, and redox imbalance was evidenced in plasma from CDD patients [31,32], suggesting a subclinical chronic in ammatory status in children affected by this pathology. However, to date, it is unknown whether such an in ammatory state is even mirrored at the cerebral level and whether it can contribute to the pathophysiology of CDD.
Here, we show evidence of a microglia over-activation status in the brain of Cdkl5 KO mice, characterized by alterations in microglia cell number/morphology and increased pro-in ammatory gene expression. We found that microglia over-activation is already present in the postnatal period in Cdkl5 KO mice and worsens during aging. Importantly, by restoring microglia alterations, treatment with luteolin, a natural anti-in ammatory avonoid, counteracts hippocampal neuron cell death in both adult and aged Cdkl5 KO mice, and rescues NMDA-induced excitotoxic damage in Cdkl5 KO mice. These ndings highlight new insights into the CDD pathophysiology and provide the rst evidence that therapeutic approaches aimed at counteracting neuroin ammation could be bene cial in CDD.

Colony and husbandry
The mice used in this work derive from the Cdkl5 null strain in the C57BL/6N background developed in [6] and backcrossed in C57BL/6J for three generations. Mice for experiments were produced by crossing Cdkl5 +/− females with Cdkl5 -/Y males and Cdkl5 +/− females with +/Y males [6]; animals were karyotyped by PCR on genomic DNA as previously described [6] and littermate controls were used for all experiments. P0, postnatal day zero, was designated as the day of birth and 24 h later mice were considered 1-day-old animals (P1). Mice were housed three to ve per cage and maintained in a temperature-(23 C) and humidity-controlled environment with a standard 12 h light/dark cycle, and provided with standard mouse chow and water ad libitum. Animals' day-to-day health and comfort were monitored by the veterinary service. All possible efforts were made to minimize suffering and the number of animals used. Experiments were carried out on a total of 89 Cdkl5 KO mice (Cdkl5 +/Y, n = 4; Cdkl5 -/Y, n = 4; Cdkl5 +/+, n = 26; Cdkl5 +/-, n = 55).

Treatments
All treatments were performed in the animal house at the same hour of the day.
Luteolin treatment -The dose of luteolin was chosen based on [33]. Mice were intraperitoneally (i.p.) injected daily with saline or luteolin (10 mg/kg in saline; Tocris) for 7 or 20 days. The day following the last treatment mice were sacri ced for histological and Western blot analyses.
NMDA treatment -Mice were treated with an intraperitoneal injection of NMDA (60 mg/kg; Sigma-Aldrich) in phosphate-buffered saline (PBS) after 7 days of luteolin treatment. Animals were sacri ced 24 h or 8 days after NMDA injection and immunohistochemical analysis was assessed as described below.
TATκ-GFP-CDKL5 protein treatment -Brain sections processed for AIF-1 immunohistochemistry derived from animals used in [34]. Brie y, mice were treated twice a day (morning and evening) with TATκ-GFP or TATκ-GFP-CDKL5 protein (50 ng/injection) directly injected into the carotid artery through a programmable pump [34].
Histological and immunohistochemical procedures Animals were anesthetized with 2% iso urane (in pure oxygen) and sacri ced through cervical dislocation. Brains were quickly removed and cut along the midline. While right hemispheres were quickly frozen and used for Western blot analyses, left hemispheres were xed via immersion in 4% paraformaldehyde (100 mM phosphate buffer, pH 7.4) for 48 h, kept in 15-20% sucrose for an additional 24 h and then frozen with cold ice. Hemispheres were cut with a freezing microtome into 30 µm-thick coronal sections and processed for immunohistochemistry procedures as described below. Brain sections from Emx1 KO mice used in [35] were processed for AIF-1 immunohistochemistry.
Cleaved caspase-3 staining -For cleaved caspase-3 immuno uorescence, one out of every six sections from the hippocampal formation was incubated overnight with a rabbit polyclonal anti-cleaved caspase-3 antibody (1:200; Cell Signaling Technology). The following day the sections were washed and incubated with an HRP-conjugated anti-rabbit secondary antibody (1:200; Jackson Immunoresearch, Inc.). Detection was performed using the TSA Cyanine 3 Plus Evaluation Kit (Perkin Elmer) and nuclei were counterstained with Hoechst 33342 (Sigma-Aldrich).
DCX immunohistochemistry -One out of every six free-oating sections from the hippocampal formation was incubated overnight with a goat polyclonal anti-doublecortin antibody (DCX, 1:1000, Santa Cruz Biotechnology, Inc.). The following day the sections were washed and incubated with a biotinylated antigoat IgG secondary antibody (1:200, Vector BioLabs, Malver, PA, USA) for 2 h and thereafter for 1 h with the VECTASTAIN®ABC kit (Vector BioLabs). Detection was performed using DAB kit (Vector BioLabs).

Image acquisition and measurements
Fluorescence images were taken with an Eclipse TE 2000-S microscope equipped with a DS-Qi2 digital SLR camera (Nikon Instruments Inc.). A light microscope (Leica Mycrosystems) equipped with a motorized stage and focus control system and a color digital camera (Coolsnap-Pro; Media Cybernetics) were used for neuronal tracing and to take bright eld images. Measurements were carried out using

Microglia cell isolation
Microglial cells were isolated following the protocol published in [37]. Brie y, wild-type (+/+) and Cdkl5 KO (+/-) female mice were anesthetized with 2% iso urane (in pure oxygen) and transcardial perfused with PBS. Brains were transferred to ice-cold phosphate buffered saline (PBS; without Ca2 + and Mg2', with NaHCO 3 , 0.75 g/l, Hepes buffer 10 mM, pH 7.4), freed of meninges, minced in serum-free Dulbecco with sieved tissue derived from 2 mice brain was mixed with 8.6 ml of cold isotonic Percoll (Formerly GE Healthcare Life Sciences; Fisher scienti c) in PBS and centrifuged (2000 rpm at 4°C for 20 min). Pellet was washed in PBS and nally resuspended in 1 ml of TRI reagent (Sigma-Aldrich, St. Louis, MO, USA) and stored at -80°C or processed for RNA extraction.

RNA isolation and RT-qPCR
Microglia cell RNA isolation was performed using the Direct-zol RNA MiniPrep kit (Zymo Research) and cDNA synthesis was achieved with 1 µg of total RNA using iScript™ Advanced cDNA Synthesis Kit (Bio-Rad, Hercules, CA, USA) according to the manufacturer's instructions. Real-time PCR was performed using Sso Advanced Universal SYBR Green Supermix (Bio-Rad) in an iQ5 Real-Time PCR Detection System (Bio-Rad). We used primer pairs (table S1) that gave an e ciency close to 100%. GAPDH (glyceraldehyde 3phosphate dehydrogenase) used as a reference gene for normalization in the qPCR. Relative quanti cation was performed using the ΔΔCt method.

Statistical analysis
Statistical analysis was performed with GraphPad Prism (version 7). Values are expressed as means ± standard error (SEM). The signi cance of results was obtained using Two-tailed unpaired t-test and oneway or two-way ANOVA followed by Fisher's LSD post-hoc test as speci ed in the gure legends. A probability level of P < 0.05 was considered to be statistically signi cant. The con dence level was taken as 95%.

Results
Increased microglial activation in the brain of Cdkl5 KO mice.
To investigate whether in ammatory processes could be involved in the pathophysiology of CDD, we counted the number and analyzed the morphology of microglia (AIF-1-positive cells) in the hippocampus and cortex of male (-/Y) and female (+/-) Cdkl5 KO mice and wild-type (+/Y, +/+) littermates. We found an increase in the number of microglial cells in both the analyzed brain regions of -/Y and +/-Cdkl5 KO mice in comparison with their +/Y, and +/+ counterparts ( Fig. 1A-C). Moreover, microglial cells in Cdkl5 KO mice presented an enlarged body size ( Fig. 1D and Fig S1A,B) and reduced roundness of the cell body (Fig. 1E) compared to wild-type counterparts (Fig. 1B,D,E). Together, these data indicate that in the absence of Cdkl5 microglia adopted a bigger, more irregular soma shape, typical of a state of activation [39].
Importantly, replacement of CDKL5 protein through a systemic injection of a TATκ-GFP-CDKL5 fusion protein [34] reversed microglial activation in Cdkl5 KO mice. We found a lower number of microglial cells ( Fig. 2A) with a smaller body size (Fig. 2B) in the hippocampus and cortex of TATκ-GFP-CDKL5-treated Cdkl5 -/Y mice compared to Cdkl5 -/Y mice treated with a TATκ-GFP control protein, indicating the reversibility of the in ammatory phenotype due to the absence of Cdkl5.
Non-cell autonomous microglial activation in the absence of Cdkl5 In order to investigate whether microglial activation in Cdkl5 KO mice is a cell-autonomous effect, we rst evaluated Cdkl5 expression levels in puri ed microglial cells. Real time and western blot analyses showed very low mRNA levels (Fig. 3A) and undetectable protein levels (Fig. 3B) of Cdkl5 in microglial cells compared to cortical extracts of wild-type mice, suggesting that Cdkl5 function is of minor relevance in these cells. Next, we evaluated the microglia activation status in Emx1 KO mice, a conditional Cdkl5 KO mouse model carrying Cdkl5 deletion only in the excitatory neurons of the forebrain, but not in microglial cells [6,35]. Similarly to Cdkl5 KO mice, activation of microglial cells, increased number and body size of microglial cells (Fig. 3C), was present in the hippocampus of Emx1 KO mice, suggesting a non-cell autonomous microglial over-activation in the absence of Cdkl5.
Luteolin, a naturally occurring polyphenolic avonoid, is a potent microglia inhibitor that possesses antioxidant, anti-in ammatory, and neuroprotective effects both in vitro and in vivo [40,41]. Since the majority of CDD patients are heterozygous females [42], we tested the e cacy of an in vivo treatment with luteolin on microglial over-activation in the heterozygous female mouse model of CDD (+/-). Threemonth-old Cdkl5 KO female mice (+/-) were injected daily with luteolin (i.p. 10 mg/Kg) for 7 or 20 days. While both short and long term treatments with luteolin did not affect microglia cell number in the cortex of Cdkl5 +/-mice (Fig. 4A,C), a 7-day treatment was su cient to recover microglia body size at the wildtype level in both cortex (Fig. 4B,C) and hippocampus (Fig. S2).
Con rming microglia over-activation, we found signi cantly higher levels of phosphorylated STAT3, a key promoter of microglial cell pro-in ammatory phenotype [43,44], in brain homogenates of Cdkl5 +/-mice in comparison with their wild-type counterparts (Fig. 4D-F). A 7-day treatment with luteolin restored phospho-STAT3 levels to those of wild-type mice (Fig. 4D,F). No differences in total STAT3 levels were observed in treated or untreated Cdkl5 +/-mice in comparison with their wild-type counterparts (Fig. 4E,F).
Accordingly with activation of STAT3 signaling and microglia over-activation, we found increased expression of molecules involved in the microglial neuroin ammatory response, such as IL-1β and IL-6 cytokines, and TNFα or microglial markers (CX3CR1 and AIF-1) (Fig. 5A,B); the increased expression was recovered by treatment with luteolin in Cdkl5 +/-mice (Fig. 5A,B).
Treatment with luteolin restores survival and maturation of newborn cells in the dentate gyrus of Cdkl5 +/-mice Loss of Cdkl5 impairs survival and maturation of newborn hippocampal neurons [11,16]. In order to evaluate the e cacy of in vivo treatment with luteolin on the survival rate of new neurons, we assessed the number of doublecortin (DCX)-positive cells in the dentate gyrus (DG) of untreated Cdkl5 +/-and Cdkl5 +/+ mice and Cdkl5 +/-mice treated with luteolin for 7 days. We found that treatment with luteolin restored the number of DCX-positive granule neurons in Cdkl5 +/-mice (Fig. 6A,B). To determine whether increased proliferation rate underlies the positive effect of luteolin on newborn neuronal number, we counted proliferating cells immunostained for Ki-67, an endogenous marker of actively proliferating cells. We found that luteolin-treated Cdkl5 +/− mice had the same number of Ki-67 labelled cells as untreated Cdkl5 +/− and Cdkl5 +/+ mice (Fig. 6C). This evidence indicates that the higher number of DCX-positive cells in treated Cdkl5 +/− mice is not due to an increase in proliferation rate.
Importantly, luteolin treatment improved impaired dendritic development in Cdkl5 +/-mice. Quanti cation of the dendritic tree of DCX positive cells showed that Cdkl5 +/− mice had a shorter dendritic length than wild-type mice (Fig. 6A,D); a 7-day treatment with luteolin improved this defect (Fig. 6A,D).
It was reported that luteolin treatment increases brain levels of the brain-derived neurotrophic factor (BDNF) [33], which is necessary for neuronal survival and maturation [45]. We examined mature BDNF levels in cortical homogenates of Cdkl5 +/-and Cdkl5 +/+ mice and in Cdkl5 +/-mice following administration of luteolin for 7 days. While levels of BDNF were similar in Cdkl5 +/-and Cdkl5 +/+ mice treated with the vehicle, treatment with luteolin increased levels of BDNF by about 50% (Fig. 6E,F).
Microglia activation with age in Cdkl5 +/-mice To assess whether microglia over-activation is already present at an early stage of life and to monitor its evolution with age, we analyzed microglial cell status in the brains of Cdkl5 KO mice at different developmental stages (young: 20-day-old, adult: 3-month-old, and middle-aged: 11-month-old mice). We found that an increase in microglial cell number and soma size was already present in the cortex and hippocampus of young Cdkl5 +/-mice compared to their wild-type counterparts of the same age (Fig. 7A,B). A signi cant decrease in the density of microglial cells with age was present in both wild-type and Cdkl5 +/-mice compared to their 20-day-old counterparts (Fig. 7A,B). Surprisingly, while the difference in the number of microglial cells was maintained between Cdkl5 +/-and wild-type mice at 3 months of age, in middle-aged mice there was no longer a difference (Fig. 7A,B). In contrast, increased microglial body size in Cdkl5 +/-mice was present in all three age groups compared to their wild-type counterparts of the same age (Fig. 7A,B). Interestingly, an age-dependent worsening of microglial activation, and, therefore, microglial body size, was observed in both middle-aged Cdkl5 +/-and Cdkl5 +/+ mice (Fig. 7A,B).
Luteolin treatment restores neuron survival in middle-aged Cdkl5 +/-mice Microglial cell changes compatible with their activation have been documented in aging [46,47] and it has been suggested that they contribute to the brain decline in pathological conditions [48,49]. Recent evidence showed an age-dependent decreased hippocampal neuron survival in middle-aged Cdkl5 KO mice, paralleled by an increased cognitive decline [21].
To explore the possibility that microglia over-activation in aged Cdkl5 KO mice could underlie the higher neuronal loss, we assessed the e cacy of treatment with luteolin in counteracting neuronal loss in 11month-old Cdkl5 +/-mice (Fig. 8A). Treatment with luteolin reduced microglia body size in the hippocampus of middle-aged Cdkl5 +/-mice at even lower levels than those of wild-type mice of the same age (Fig. 8B). Importantly, the reduced number of Hoechst-positive nuclei (Fig. 8C) and NeuN-positive cells (Fig. 8D) in middle-aged Cdkl5 +/-mice was strongly improved by treatment with luteolin (Fig. 8C,D).
Treatment with luteolin prevents NMDA-induced excitotoxicity in the hippocampus of Cdkl5 +/-mice.
To investigate whether microglia over-activation has a causative role in the increased neuronal susceptibility to excitotoxic stress in Cdkl5 KO mice [19,20], we pre-treated Cdkl5 +/-mice for 7 days with Luteolin before NMDA (60mg/kg) intraperitoneal injection (Fig. 9A). Animals were sacri ced 1 day or 8 days after NMDA administration (Fig. 9A). As expected, microglia activation increased in the hippocampus of both Cdkl5 +/-and Cdkl5 +/+ mice after NMDA treatment (Fig. 9B). Nevertheless, after NMDA excitotoxic stimulation, the somal volume of microglial cells in Cdkl5 +/-mice was higher than that of NMDA-treated wild-type mice (Fig. 9B). Importantly, luteolin treatment was able to counteract both basal and NMDA-induced microglial activation in Cdkl5 KO mice, bringing microglial soma size back to that of the untreated wild-type mouse condition (Fig. 9B).
Neuronal death was assessed 1 day after NMDA administration using immunohistochemistry for cleaved caspase-3 and 8 days after using Hoechst staining and immunohistochemistry for NeuN. In the CA1 layer of the hippocampus, NMDA-treated Cdkl5 +/-mice showed a higher number of cleaved caspase-3 positive cells (Fig. 9C,D) and a lower number of Hoechst-positive nuclei (Fig. 9E) and NeuN-positive cells (Fig. 9F,G) in comparison with NMDA-treated wild-type mice, indicating increased cell death in Cdkl5 +/mice after the excitotoxic stimulus. Importantly, pre-treatment with luteolin reduced cell death at 24 h after NMDA treatment in Cdkl5 +/-mice (Fig. 9C,D), thus preventing neuronal loss in the hippocampal CA1 region (Fig. 9E-G).

Discussion
The lack of effective therapies for CDD stresses the urgency with which pathogenic mechanisms underlying the disorder need to be identi ed. Our results highlight, for the rst time, the presence of a generalized status of microglia over-activation in the brain of a mouse model of CDD. We found alterations in microglial cell morphology and number, increased levels of AIF-1 and proin ammatory cytokines, and increased STAT3 signaling in the brain of Cdkl5 KO mice. Remarkably, treatment with luteolin (a natural anti-in ammatory avonoid) is able to recover impaired neuronal survival and maturation in Cdkl5 KO mice, suggesting that a hyperactive state of microglia plays a causative role in the CDD phenotype.
Recently, a cytokine dysregulation, proportional to clinical severity and redox imbalance, was found in children affected by CDD [31,32]. Results included increased Tumour Necrosis Factor (TNF)-α, Interleukin (IL)-1β, and Interleukin (IL)-6 in the peripheral blood of children with CDD. These in ammatory cytokines could signal in ammatory changes in the brain that could, in turn, greatly impact neurodevelopment and neural function in CDD. Our nding of an over-activation of microglial cells with increased TNF-α, IL-1β and IL-6 levels in the brain of Cdkl5 KO mice is in line with the results observed in CDD patients, and suggests the involvement of neuro-in ammatory processes in the pathophysiology of CDD.
Microglia activation, associated with an increase in cell body size, as well as cytokine alterations in the peripheral blood, has been recently described in neurodevelopmental disorders such as autism spectrum disorders (ASDs), Down syndrome, and Rett syndrome [28-30, 50, 51]. Similarly to Rett syndrome [52], the mechanism by which absence of Cdkl5 induces microglia over-activation appeared to be non-cell autonomous. By using Emx1 KO mice, a conditional Cdkl5 KO mouse model which does not carry Cdkl5 deletion in microglial cells [6,35], we found that microglia over-activation is independent of microgliaspeci c loss of Cdkl5 expression. This nding suggests that microglial activation in the Cdkl5 KO brain may be attributable to neuronal loss of Cdkl5. Accumulating evidence indicates the presence of bidirectional microglia-neuron communication in the healthy and diseased brain [53]. In the healthy brain, microglia exhibit an actively repressed 'surveying' phenotype that is dependent on a dynamic crosstalk between microglia and neurons [54]. It has been proposed that the removal of this neuronal-derived inhibitory control represents a type of danger signal for microglia, indicating that neuronal function is impaired and leads to alterations in microglia morphology and function. The chronic activation of microglia may, in turn, cause reduced neuronal maturation and survival through the release of potentially cytotoxic molecules such as proin ammatory cytokines [23,24]. Our nding that inhibition of microglia over-activation by luteolin restores survival and maturation of newborn neurons in the hippocampal dentate gyrus of Cdkl5 KO mice is in line with this hypothesis. Though luteolin treatment had no effect on the number of microglial cells, it restored microglia body size and shape, and, importantly, proin ammatory cytokines and P-STAT3 levels. Differently from a recent study that demonstrated the effect of luteolin treatment in increasing cell proliferation in a mouse model of Down syndrome [33], we did not observe an increased number of Ki-67-positive cells in the hippocampal dentate gyrus of Cdkl5 KO mice. This discrepancy can be explained by a selective effect of luteolin on the Ts65Dn mouse, as the authors themselves did not observe a luteolin-dependent proliferation increase in wild-type mice. The pro-survival effect of microglial inhibition was con rmed by the restoration of the age-dependent decreased hippocampal neuron survival in middle-aged Cdkl5 KO mice. This is in agreement with recent studies that suggest that luteolin plays a role in counteracting age-induced microglia activation in aged mice [55,56].
The neuroprotective effect of luteolin might also be associated with its effect on increasing BDNF levels.
Recent ndings reported that luteolin increases the expression of BDNF in the cerebral cortex and hippocampus of mice [33,57]. Similarly to these ndings, we found increased BDNF levels in the brain of luteolin treated Cdkl5 KO mice. Since BDNF plays an important role in the survival and development of neurons [58], it is reasonable to hypothesize that, in Cdkl5 KO mice, the luteolin-dependent activation of the BDNF pathways contributes to hippocampal neuronal survival and maturation.
As previously reported in male Cdkl5 KO mice [19,20], we found here that hippocampal neurons of heterozygous female Cdkl5 KO mice showed an increased vulnerability to excitotoxicity. In ammatory processes, including activation of microglia and production of proin ammatory cytokines, such as TNF-α, IL-1β, and IL-6, are associated with excitotoxic stimuli [59,60]. Similarly, we observed an increased microglial over-activation in response to NMDA treatment. Importantly, the microglia activation was higher in Cdkl5 KO mice than in wild-type mice. Luteolin pre-treatment recovered the increased NMDAinduced cell death in hippocampal neurons of Cdkl5 KO mice. Because the increased cell death of hippocampal neurons observed in response to NMDA treatment correlates with an increased microglial over-activation, the bene cial effect of luteolin in neuronal survival can be, at least partially, ascribed to the inhibition of microglia activation. Luteolin was observed to exert a similar neuroprotective activity against kainic acid-induced brain damage in mice [61].

Conclusions
Microglia remain to be controversial cells within the CNS, with both bene cial and detrimental roles, especially in the context of disease pathology. Our nding that treatment with luteolin recovers neuronal survival and maturation in Cdkl5 KO mice supports the theory that microglia exert a harmful action in the CDKL5-null brain. Recent evidence strongly supports the role of neuroin ammation in the pathophysiology of human epilepsy [62]. In support of this it has been shown that pre-treatment with luteolin signi cantly reduces the frequency of drug-induced seizures in animal models of epilepsy [61, 63-65]. Since epilepsy is one of the hallmark features of CDD, it may be hypothesized that luteolin has a bene cial effect on this feature. Therefore, the suppression of microglia-mediated in ammation may be considered as an important strategy in CDD therapy.

Consent for publication
Not applicable.

Availability of data and materials
The datasets analyzed during the current study are available from the corresponding author on reasonable request.

Competing interests
The authors declare that they have no competing interests.