Early-onset brain alterations during postnatal development in a mouse model of CDKL5 deficiency disorder

Mutations in the CDKL5 gene are the cause of CDKL5 deficiency disorder (CDD), a rare and severe neuro-developmental condition characterized by early-onset epilepsy, motor impairment, intellectual disability, and autistic features. A mouse model of CDD, the Cdkl5 KO mouse, that recapitulates several aspects of CDD symptomology, has helped to highlight brain alterations leading to CDD neurological defects. Studies of brain morphogenesis in adult Cdkl5 KO mice showed defects in dendritic arborization of pyramidal neurons and in synaptic connectivity, a hypocellularity of the hippocampal dentate gyrus, and a generalized microglia over-activation. Nevertheless, no studies are available regarding the presence of these brain alterations in Cdkl5 KO pups, and their severity in early stages of life compared to adulthood. A deeper understanding of the CDKL5 deficient brain during an early phase of postnatal development would represent an important milestone for further validation of the CDD mouse model, and for the identification of the optimum time window for treatments that target defects in brain development. In sight of this, we comparatively evaluated the dendritic arborization and spines of cortical pyramidal neurons, cortical excitatory and inhibitory connectivity, microglia activation, and proliferation and survival of granule cells of the hippocampal dentate gyrus in hemizygous Cdkl5 KO male ( (cid:0) /Y) mice aged 7, 14, 21, and 60 days. We found that most of the structural alterations in Cdkl5 (cid:0) /Y brains are already present in pups aged 7 days and do not worsen with age. In contrast, the difference in the density of excitatory and inhibitory terminals between Cdkl5 (cid:0) /Y and wild-type mice changes with age, suggesting an age-dependent cortical excitatory/inhibitory synaptic imbalance. Confirming the precocious presence of brain defects, Cdkl5 (cid:0) /Y pups are characterized by an impairment in neonatal sensory-motor reflexes.


Introduction
Loss-of-function variants in the X-linked cyclin-dependent kinase like 5 (CDKL5) gene (MIM 300203) underpin a severe developmental encephalopathy named CDKL5 Deficiency Disorder (CDD; OMIM 300203,300672).This rare genetic neurodevelopmental disorder has a worldwide incidence of approximately 1: 40,000-60,000 live births (Symonds et al., 2019), with higher frequency in females than in males (~4:1; (Lopez-Rivera et al., 2020)).The clinical manifestation of CDD usually begin during early infancy, with variable and complex combinations of symptoms including early-onset drug-resistant seizures (in the first days/months of life), and a severe neurodevelopmental impairment with major motor development delay such as generalized hypotonia and retardation of psychomotor development (already observed in the first few months of life).Cortical visual impairment, severe intellectual disability, and gastrointestinal and orthopaedic complaints are common (Bahi-Buisson and Bienvenu, 2012;Jakimiec et al., 2020;Olson et al., 2019).Currently, no cure is available for CDD patients; pharmacological treatments mostly address seizure management, and no effective treatments exist to ameliorate the cognitive and behavioral symptoms of Abbreviations: CDKL5, cyclin-dependent kinase-like 5; CDD, CDKL5 deficiency disorder; VGluT1, vesicular glutamate transporter 1; GABA, gamma-aminobutyric acid; VGAT, vesicular GABA transporter; AIF1, anti-allograft inflammatory factor 1; DCX, doublecortin; DG, dentate gyrus; GL, granular layer; SGZ, subgranular zone; EB2, microtubule end-binding protein 2; CNS, central nervous system.

CDD.
Studies in rodents have shown that Cdkl5 is expressed at low levels during embryonic stages, and that its expression rapidly increases postnatally in neurons.The peak of Cdkl5 expression during the postnatal period suggests that CDKL5 functions are required early on to ensure proper brain formation/function (Rusconi et al., 2008;Zhu et al., 2013).Several knockout (KO) mouse models for Cdkl5 have been developed to address how CDKL5 dysfunction leads to neurological defects in CDD (Amendola et al., 2014;Okuda et al., 2017;Wang et al., 2012).To date, studies on Cdkl5 KO mice have focused mainly on the adult brain phenotype.In particular, Cdkl5 KO mouse models exhibit several behavioral phenotypes that mimic CDD features, such as impaired learning and memory, social interaction, and motor coordination, together with increased stereotypy (Amendola et al., 2014;Fuchs et al., 2015;Fuchs et al., 2014;Ren et al., 2019;Wang et al., 2012).These defects are associated with neuroanatomical alterations: reduced neuronal branching, spine maturation, and connectivity observed in the cortex and in the hippocampal region of Cdkl5 KO mice (Amendola et al., 2014;Fuchs et al., 2015;Fuchs et al., 2014;Gennaccaro et al., 2021;Pizzo et al., 2016, Della Sala et al., 2016;Ren et al., 2019).In addition, Cdkl5 KO mice are characterized by an increased rate of apoptotic cell death in the hippocampal dentate gyrus and by an increased status of microglia overactivation, indicating that an absence of Cdkl5 increases neuronal vulnerability (Fuchs et al., 2014;Galvani et al., 2021;Gennaccaro et al., 2021;Loi et al., 2020).These studies present considerable evidence that a number of brain developmental deficits can be identified and that these deficits underlie the wellcharacterized phenotype in juvenile and adult Cdkl5 KO mice.Nevertheless, no studies are available regarding the presence of these brain alterations in Cdkl5 KO pups and unclear or untested issues remain regarding the development of symptomology and underlying mechanisms.This knowledge is of obvious importance for the identification of an optimum time window for treatments that target defects during brain development and for further validation of the mouse model, through the description of the altered crucial stages of early brain maturation in CDD.
Based on these premises, the goal of the current study was to establish whether defects in brain development, including dendritic complexity and spine maturation, excitatory and inhibitory connectivity, microglia activation, and neuronal survival, are already present in the brain of Cdkl5 KO pups, or, if not, to determine at what stage they appear and evolve during postnatal brain development.Since spine density reduction is a typical feature of the dendritic pathology that affects the brain of adult Cdkl5 − /Y mice, we conducted our analyses starting from 7-day-old Cdkl5 − /Y mice because dendric spines are generated approximately at the end of the first postnatal week in rodents (Eller et al., 1969;Juraska and Fifkova, 1979;Yuste and Bonhoeffer, 2004).As structural defects are reflected in altered behavior, Cdkl5 − /Y pups were tested for motor-reflex behavior responses, as an indication of impaired neurological development.

Colony
The mice used in this work derived from the Cdkl5 KO strain in the C57BL/6N background developed in (Amendola et al., 2014) and backcrossed in C57BL/6J for three generations.Age-matched wild-type (+/Y) littermate controls were used for all experiments.The day of birth was designated as postnatal day zero (P0) and animals with 24 h of age were considered as 1-day-old animals (P1).Mice for experiments were produced by crossing Cdkl5 +/− heterozygous females with wild-type (+/Y) males and were karyotyped by PCR on genomic DNA as previously described (Amendola et al., 2014).Mice were housed three to five per cage and maintained in a temperature-(23 • C) and humiditycontrolled environment with a standard 12 h light/dark cycle, and were provided with standard mouse chow and water ad libitum.The animals' day-to-day health and comfort were controlled by the veterinary service.Experiments were performed in accordance with the Italian and European Community law for the use of experimental animals and were approved by Bologna University Bioethical Committee.In this study, all efforts were made to minimize animal suffering and to keep the number of animals used to a minimum.The number of animals used for each experimental procedure is specified in the figure legends.

Histological and immunohistochemistry procedures
Cdkl5 -/Y mice and +/Y littermates were anesthetized with 2% isoflurane (in pure oxygen) and sacrificed at postnatal day 7 (P7), 14 (P14), 21 (P21), and 60 (P60).Body weight was recorded prior to sacrifice.Brains were quickly removed and cut along the midline.Right hemispheres were Golgi-stained or quickly frozen and used for Western blot analyses (see description below).Left hemispheres were fixed 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 30μm-thick coronal sections and processed for immunohistochemistry procedures as described below.One out of 6 sections from the hippocampal formation and somatosensory cortex were used for immunohistochemistry for vesicular glutamate transporter 1 (VGluT1), vesicular GABA transporter (VGAT), anti-allograft inflammatory factor 1 (AIF1), Ki67 or doublecortin (DCX) following the protocol published in (Fuchs et al., 2014;Galvani et al., 2021;Pizzo et al., 2016).Nuclei were counterstained with Hoechst 33342 (Sigma-Aldrich).The primary and secondary antibodies used are listed in Supplementary Table 1.

Golgi staining
Hemispheres were Golgi-stained using the FD Rapid Golgi Stain TM Kit (FD NeuroTechnologies, Inc.) as previously described (Guidi et al., 2013).The hemispheres were cut with a cryostat into 100 μm-thick coronal sections.Sections were mounted directly onto super frost slides and air dried at room temperature.After drying, sections were rinsed with distilled water, stained in the developing solution of the kit and coverslipped with DPX mounting medium.

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 field images.Measurements were carried out using the Image Pro Plus software (Media Cybernetics, Silver Spring, MD, USA), by investigators blind to the animal's genotype.

Measurement of the dendritic tree
Dendritic trees were traced using custom-designed software for dendritic reconstruction (Immagini Computer, Milan, Italy), interfaced with Image Pro Plus as previously described (Guidi et al., 2013).Basal dendritic trees of Golgi-stained pyramidal neurons of layers II-III of the somatosensory cortex were traced live, at a final magnification of 500×, by focusing on the depth of the section.The operator starts with branches emerging from the cell soma and, after having drawn the first parent branch, goes on with all daughter branches of the next order in a centrifugal direction.At the end of tracing, the program reconstructs the number and length of individual branches, the mean length of branches of each order, and the total dendritic length.

Dendritic spine number and morphology
In Golgi-stained sections, dendritic spines located in the basal dendrites of pyramidal neurons of layers II-III were manually counted using a 100× oil immersion objective lens (Leitz microscope and objective with 1.4 NA).For each mouse, 10 dendritic segments of 10 μm were analyzed and spine density was expressed as number of spines per 10 μm.Based on their morphology, dendritic spines can be divided into two different classes according to their state of maturation: immature and mature spines.The number of spines belonging to the 2 different groups (immature spines: filopodium-like, thin, and stubby-shaped; mature spines: mushroom-and cup-shaped) was counted and expressed as a percentage.

Analysis of neuronal soma size
Starting from images of Golgi-stained sections of the somatosensory cortex, the soma size of pyramidal neurons (area and major axes) was quantified by manually tracing the outline and short and long axes of neuronal soma with the software Image Pro Plus (Media Cybernetics, Silver Spring, MD 20910, USA) at a final magnification of 500×.A total of 40 pyramidal neurons per genotype at each age (n = 10 neurons per mouse, n = 4 mice per genotype at each age) of layers II-III of the somatosensory cortex was measured and analyzed.

Cell density
The number of Ki-67-positive cells was counted in the subgranular, granular layer, and hilus of the hippocampal dentate gyrus and expressed as number of cells/μm 2 .DCX-positive cells were counted in the subgranular and granular layers of the dentate gyrus and expressed as number of neurons/100 μm.The number of Hoechst-positive nuclei in the upper granular layer of the dentate gyrus was manually counted and expressed as cells/μm 2 .

Morphometric microglial cell analysis
Starting from 20× magnification images of AIF-1-stained cortical slices, the number of AIF-1-positive cells in the somatosensory cortex was manually counted using the point tool of the Image Pro Plus software (Media Cybernetics, Silver Spring, MD, USA), and cell density was established as AIF-1-positive cells/mm 3 .AIF-1 positive microglial cell body size was manually drawn using the Image Pro Plus measurement function and was expressed in μm 2 .

Quantification of VGluT1 and VGAT immunoreactive puncta
Images from the somatosensory cortex were acquired using a LEICA TCS SL confocal microscope (LEITZ; Leica Microsystems, Wetzlar, Germany; objective 63×, NA 1.32; zoom factor = 8).Three to four sections per animal were analyzed; in each section three images from the regions of interest were captured and the density of individual puncta exhibiting VGAT or VGluT1 immunoreactivity in layers II-III was evaluated as previously described in (Pizzo et al., 2016).The number of VGAT or VGluT1 immunoreactive puncta was expressed per μm 2 .

Western blotting
For Western blot analysis tissue samples of the cortex were homogenized in RIPA buffer and quantified using the Bradford method, as previously described (Galvani et al., 2021).Equivalent amounts (50 μg) of protein were subjected to electrophoresis on a Bolt™ 4-12% Bis-Tris Plus gel (Life Technologies Corporation, Carlsbad, CA, USA) and transferred to a Hybond ECL nitrocellulose membrane (GE Healthcare Bio-Science, Piscataway, NJ, USA).The primary and secondary antibodies used are listed in Supplementary Table 1.The densitometric analysis of digitized Western blot images was performed using Chemidoc XRS Imaging Systems and the Image LabTM Software (Bio-Rad); this software automatically highlights any saturated pixels in the Western blot images in red.Images acquired with exposition times that generated protein signals out of a linear range were not considered for the quantification.

Behavioral testing 2.6.1. Surface righting
The righting reflex was tested using the Surface Righting Test in P5 mice, since this reflex appears in rodents at around this age (Heyser, 2004).As described in (Feather-Schussler and Ferguson, 2016) protocol, the pups were held on their backs on a cotton sheet; after 5 s they were released, and the time required to return to the prone position was recorded.A total time of 60 s was given for each trial, for a total of three trials.

Negative geotaxis
The Negative Geotaxis reflex was tested in P7 pups since this reflex appears in rodents at around this age (Heyser, 2004).As described in the (Feather-Schussler and Ferguson, 2016) protocol, the pups were placed heads downward on a 45 • inclined plane and held in this position for 5 s.After this time, the pups were released, and the time they took to turn their faces upward was recorded.A total time of 120 s was given for each trial, for a total of three trials.

Cliff aversion
The Cliff Aversion test was performed on P8 mice following the protocol described in (Feather-Schussler and Ferguson, 2016).Pups were placed on top of a flat, elevated ledge, with only their forepaws and snout positioned over the edge.The total time taken for the pups to turn away from the cliff and move their paws and snout away from the edge was recorded.The test was repeated for a total of three trials.

Statistical analysis
The reported results are expressed as means ± standard error (SEM).The GraphPad Prism (version 7) software was used for data analysis.All datasets were analyzed using the ROUT method (Q = 1%) for the identification of significant outliers and the Shapiro-Wilk test for normality testing.Datasets with normal distribution were analyzed for significance using the two-tailed Student's t-test or two-way ANOVA, followed by Fisher's LSD post hoc test or Newman-Keuls multiple comparisons test, as specified in the figure legends.A probability level of p < 0.05 was considered to be statistically significant.The confidence level was taken as 95%.A descriptive statistic of genotype and age factors and Post-Hoc comparisons is given in Supplementary Table 2.

Developmental changes in body weight in Cdkl5 − /Y and Cdkl5 +/Y mice
To assess whether defects in brain development due to loss of Cdkl5 are already present at an early stage of life and to monitor their evolution with age, we analyzed Cdkl5 − /Y mice at different ages (7-day-old, 14-day-old, 21-day-old, and 60-day-old mice).We first evaluated whether Cdkl5 − /Y mice showed a significant difference in body weight compared to their wild-type counterparts.As previously reported in young and adult Cdkl5 KO mice (Amendola et al., 2014), no genotyperelated differences in body weight were observed between Cdkl5 -/Y and wild-type mice at any of the analyzed ages (Table 1), showing evidence of similar physiological growth and no difference in food intake or animal well-being between Cdkl5 − /Y and wild-type mice.In both genotypes, after an initial weight gain of +70 ±2.5% from P7 to P14, weight increased by a further +62 ±2.0% as mice progressed to adulthood (P60, Table 1).

Developmental changes in dendritic morphology in Cdkl5 − /Y and Cdkl5 +/Y mice
In order to establish whether defects in dendritic morphology are M. Tassinari et al. already present at the earliest phases of brain development, we examined the basal dendritic tree of the pyramidal neurons in layer II/III of the somatosensory cortex (Fig. 1A) of Cdkl5 − /Y mice.We first examined the number and mean length of segments, and the total length of the dendritic tree.Development of the dendritic tree consists of an increase in dendritic length and in the number of branches which leads to an increase in total dendritic length; dendrites branch more profusely and are spatially more expanded (Fig. 1B).Total dendritic tree size increased rapidly during the second postnatal week (from day 7 to day 14) in both wild-type and Cdkl5 − /Y mice (Fig. 1C; +/Y +77.2±9.9%,− /Y +107±12.9%),due to an increase in both number and length of dendritic branches (Fig. 1D,C).The dendritic tree underwent a further increase to adulthood (day 60) (Fig. 1C; +/Y +38.7±11.04%,− /Y +47.7±3.01%,from day 14 to day 60).The number of dendritic segments per cell did not change during this time (Fig. 1D), but there was an increase in dendritic length (Fig. 1E).It is worth noting that, at all ages, Cdkl5 -/Y mice had a reduced number (P7 -37±6.5%,P14 -23±3.3%,P21 -15±1.4%,P60 -23±1.8%)and length (P7 -30±2.7%,P14 -28 ±5.7%, P21 -16±5.5%,P60 -19±1.2%) of dendritic branches compared with wild-type (+/Y) mice (Fig. 1D,E), and a consequent reduction in total dendritic length (Fig. 1C).
In order to analyze dendritic morphology in detail we examined each dendritic order separately.As expected, there was an age-related increase in the maximum number of dendritic orders that, from a value of 5 in P7 and P14 wild-type mice, reached a value of 6 in P21 and P60 wild-type mice (Supplementary Fig. 1).In P7 and P14 mice there was no genotype-related difference in the maximum number of dendritic orders (Supplementary Fig. 1), while in P21 and P60 Cdkl5 -/Y mice the branches of the last order (6) were missing (Supplementary Fig. 1).Regarding the mean length of dendritic branches of individual orders, we found few differences between wild-type and Cdkl5 − /Y mice (Supplementary Fig. 1A-D).The only genotype-related differences regarded the mean length of branches of order 4 at P7 and of order 5 at P21 and P60, that were significantly smaller in Cdkl5 − /Y mice in comparison with wild-type mice (Supplementary Fig. 1A-D).Regarding the number of dendritic branches of individual orders, we found significant differences, or a downward trend, in the number of ramifications in Cdkl5 − /Y mice compared with wild-type mice, starting from order 2 (Supplementary Fig. 1E-H).Taken together, these results show early-appearing defects in the dendritic architecture of the pyramidal neurons of Cdkl5 − /Y mice, represented by an overall reduction in the number and length of dendritic branches.
As an index of neuronal maturation (Miller and Peters, 1981), we evaluated the size of the cell body of the sampled pyramidal neurons (Fig. 1F).We found a difference between Cdkl5 − /Y and wild-type mice in the long and short axes and in the surface area at P7, but not at more mature ages (Fig. 1G,H).The long axis was significantly smaller in Cdkl5 − /Y mice in comparison with their wild-type counterparts, with a trend toward a reduction also in the short axis (Fig. 1G), and this, consequently, was also the case for the surface area (Fig. 1H).At all ages the ratio between the long and short axis was no different in Cdkl5 -/Y mice in comparison with wild-type mice (Fig. 1I).Pyramidal neurons of both genotypes had an immature shape at P7 (Fig. 1I), as indicated by the smaller body size (Fig. 1G,H) and their more ovoid cell body (LA/SA closer to 1; Fig. 1I).At older ages, the soma of the pyramidal neurons of both Cdkl5 -/Y and wild-type mice had a more pyramidal, mature shape (Fig. 1I).

Developmental changes in spine density and maturation in Cdkl5
− /Y and Cdkl5 +/Y mice Spine density reduction is a typical feature of the dendritic pathology that affects the brain of adult Cdkl5 -/Y mice (Fuchs et al., 2018;Fuchs et al., 2015;Trazzi et al., 2016).In addition, Cdkl5 -/Y mice exhibit a deficit in dendritic spine structure and stabilization (Della Sala et al., 2016;Fuchs et al., 2018;Fuchs et al., 2015;Ricciardi et al., 2012;Trazzi et al., 2016).There is no information, however, regarding the onset of this defect.To this purpose, we examined the number of spines on the basal dendritic tree of the pyramidal neurons of the somatosensory cortex of Cdkl5 − /Y mice (Fig. 2A,B).No differences between P7 Cdkl5 − /Y and wild-type mice were present in spine density, indicating that during the early establishment of dendritic spines there are no genotyperelated differences in their density (Fig. 2A,B).Differently, during the age-dependent increase in dendritic spine density, differences in the number of spines were present when comparing Cdkl5 − /Y and wildtype mice (Fig. 2A,B).Cdkl5 -/Y mice had a reduced spine density compared with their wild-type counterparts (Fig. 2A,B; P14 -7.8±1.2%,P21 -4.3±0.7%,P60 -8.3±1.0%).In wild-type mice the dendritic spine density on the cortical pyramidal neurons undergoes an age-related increase until adulthood (P60).Notably, from day 21 to day 60 this developmental increase in spine number was not present in Cdkl5 -/Y mice, as spine density did not differ significantly over time (Fig. 2A,B).
Next, we categorized dendritic spines according to their morphologies to test whether the defect in spine formation in Cdkl5 -/Y mice might be accompanied by an impairment in spine maturation represented by an overabundance of filopodia/thin and stubby spines.Therefore, similar to previous studies (Hering and Sheng, 2001), we categorized dendritic spines as filopodia-like and thin, stubby, mushroom-like, and cup spines.We found no differences in the percentage of mature spines between P7 Cdkl5 -/Y and wild-type mice (Fig. 2A,C).The number of mature spines increased rapidly during the second postnatal week (from day 7 to day 14) in wild-type mice (Fig. 2C; +/Y +59±2.7%),but less so in Cdkl5 -/Y mice (Fig. 2C;-/Y +22±9.9%),which led to a reduction in the percentage of mature spines in P14 Cdkl5 -/Y (− 52±3.2%) in comparison with wild-type mice (Fig. 2C).The lower percentage of mature spines was still present in P21 (− 35±3.0%) and P60 (− 32±1.5%)Cdkl5 -/Y mice versus wild-type mice (Fig. 2A,C), indicating that the reduction in spine density was accompanied by alterations in spine maturation.

Developmental changes in glutamatergic and GABAergic axon terminals in Cdkl5 − /Y and Cdkl5 +/Y mice
Next, we investigated synaptic organization in the cortex of Cdkl5 − /Y mice by analyzing the localization of the vesicular glutamate transporter 1 (VGluT1), which is a marker of cortico-cortical glutamatergic axon terminals (Wojcik et al., 2004), and that of the vesicular GABA transporter (VGAT), a marker of GABAergic axonal terminals (Chaudhry et al., 1998).
The density of VGluT1-positive puncta increased rapidly during the second postnatal week (from day 7 to day 14) in both wild-type and Cdkl5 − /Y mice (Fig. 2D,F; +/Y +63.3±2.52%,− /Y +58.5±2.22%).Differently, while the density of VGluT1-positive puncta did not change with age in wild-type mice, it underwent a further increase during the third week (from day 14 to day 21) in Cdkl5 -/Y mice (Fig. 2D,F 2D,F).
Taken together, these data indicate a dysregulation of both glutamatergic and GABAergic terminal development in the cortex of Cdkl5 − /Y mice.

Developmental changes in the density and morphology of microglia cells in Cdkl5 − /Y and Cdkl5 +/Y mice
Adult Cdkl5 KO mice are characterized by increased microglial activation (Galvani et al., 2021;Medici et al., 2022), suggesting the involvement of inflammatory processes in the CDD pathophysiology.Young 20-day-old heterozygous female Cdkl5 KO mice showed increased microglial body size compared to their wild-type counterparts of the same age, suggesting that microglia overactivation is an early onset defect (Galvani et al., 2021).To assess microglia overactivation, we analyzed microglial cell status by evaluating the density and analyzing the morphology of microglia (AIF-1-positive cells) in the cortex of all age groups in Cdkl5 − /Y mice.In the early postnatal stage (first two weeks), microglial cells invade the brain from specific areas, such as the supraventricular corpus callosum (Hirasawa et al., 2005).These invading ameboid microglia migrate into the brain parenchyma where they ultimately mature into smaller body ramified microglia (Hirasawa et al., 2005).As development proceeds, the resident microglia decrease in number (Kim et al., 2015).As expected, in both Cdkl5 -/Y and wild-type mice we found an increase in microglia cell density from day 7 to day 14 (Fig. 3A,B; +/Y +225±32%, − /Y +150±8.1%)and a significant decrease in the density of microglial cells with age (Fig. 3B; from day 14 to day 60, +/Y -39±2.3%,− /Y -24.5±1.5%).The minor decrease in microglial cells with age in Cdkl5 -/Y mice likely subtends the higher density of microglia found in P60 Cdkl5 − /Y mice in comparison to their wild-type counterparts (+19.4±2.4%;Student's t-test p < 0.006; Fig. 3B) (Galvani et al., 2021).Regarding microglial body size, a decrease in size, as an index of microglial maturation, was present in both Cdkl5 − /Y and wild-type mice from day 7 to day 14 (Fig. 3A,C; +/Y -35.8±0.5%,− /Y -36±0.7%).Interestingly, microglial cells had an enlarged body size in all ages of Cdkl5 -/Y mice compared to their wildtype counterparts of the same ages (Fig. 3C,D; P7 +23.6±4.6%,P14 +13.8±1.2%,P21 +12.8±1.5%,P60 +12.6±1.1%),indicating that in the absence of Cdkl5, the microglia adopted a larger soma shape, typical of a state of activation.Comparative analysis of microglia body size frequency distribution in Cdkl5 − /Y mice versus age-matched wild-type mice confirmed the high percentage of microglia cells with a large body size in Cdkl5 − /Y mice at each analyzed age, as showed by the rightshifted distribution compared to that of wild-type mice (Fig. 3D).While in the cortex of P7 mice of both genotypes the population of microglial cells had a greater body size variety (Fig. 3D), at later developmental stages microglia cells displayed cell bodies ranging in size from 40 to 120 μm 2 (Fig. 3D).

Developmental changes in hippocampal neurogenesis in Cdkl5 − /Y and Cdkl5 +/Y mice
Hippocampal neurogenesis plays a role in the maintenance and function of the dentate gyrus and hippocampal circuitry and consequently of cognitive functions (Kozareva et al., 2019).In terms of the process of neurogenesis, Cdkl5 KO mice are characterized by an increased rate of apoptotic cell death of postmitotic granule precursors in the hippocampal dentate gyrus that causes a reduction in the final number of granule neurons (Fuchs et al., 2014).In order to establish whether the lack of Cdkl5 affects progenitor cell proliferation/survival in the hippocampal dentate gyrus (DG) at early developmental ages, immunohistochemistry for Ki-67, an endogenous marker of actively proliferating cells, and for doublecortin (DCX), a microtubule-associated protein expressed by differentiating (postmitotic) granule neurons, was performed on hippocampal slices.A quantitative analysis of the density of Ki-67-positive cells in the hippocampal dentate gyrus (granular layer (GL), subgranular zone (SGZ) and Hilus) showed that the pool of proliferating precursors in the DG rapidly decreased with age (Fig. 4A,  B).Cdkl5 − /Y mice had the same density of Ki-67 labeled cells as their wild-type counterparts of the same age, indicating no genotype-related difference in the number of proliferating precursors (Fig. 4A,B).As previously reported (Nicola et al., 2015), DCX-positive cells were not detectable at the earlier postnatal time point (P7 mice) since the DCX was poorly expressed in the newborn granule cells and was diffused throughout the DG (Supplementary Fig. 2).Quantification of DCXpositive cells in the GL of the DG showed that the density of newborn cells was notably lower in Cdkl5 − /Y mice at all analyzed ages compared to their wild-type counterparts (Fig. 4C,D; P14 -12.7±3.4%,P21 -13.3 ±1.7%, P60 -29.5±9.1%),indicating that loss of Cdkl5 affects survival of postmitotic neurons very early on during postnatal hippocampal    Values in B, D, and F are represented as means ± SEM. * p < 0.05, ** p < 0.01, *** p < 0.001 as compared to the wild-type condition of the same age; # p < 0.05, ## p < 0.01, ### p < 0.001, in the comparison between Cdkl5 -/Y mice of different age groups; $ p < 0.05, $$ p < 0.01, $$$ p < 0.001 in the comparison between Cdkl5 +/Y mice of different age groups.Newman-Keuls test after two-way ANOVA.Abbreviations: DG = dentate gyrus; GL = granular layer; n = number of mice; P = postnatal day; SGL = subgranular layer.development.
To establish the impact of the reduced density of newly-formed granule neurons in Cdkl5 − /Y mice on the overall granule cell number, we stereologically evaluated the granule cell number in mice.At all ages, we found a reduced granule cell density in Cdkl5 − /Y mice compared to their wild-type counterparts (Fig. 4E,F; P7 -7.7±1.2%,P14 -9.7±0.6%,P21 -8.5±0.6%,P60 -8.6±0.3%),suggesting that a larger number of dying cells was already present in the GL of the DG of Cdkl5 − /Y mice at a very early age.

Developmental changes in EB2 phosphorylation in the brain of Cdkl5
− /Y and Cdkl5 +/Y mice Recent studies have suggested that CDKL5 plays a role in microtubule dynamics through its direct phosphorylation of microtubuleassociated proteins such as end-binding 2 (EB2) protein (Baltussen et al., 2018), which may represent a potential molecular basis for the dendritic pathology exhibited by Cdkl5 KO mice.In order to determine whether EB2 phosphorylation is dynamically regulated along with Cdkl5 throughout critical periods of brain development, we examined the protein levels of Cdkl5, phospho-EB2 (Ser22), and total EB2 at the various developmental mouse ages.While, as expected, Cdkl5 protein levels were undetectable at every age in Cdkl5 − /Y mice cortex, in wildtype mice Cdkl5 levels slightly increased in adulthood (P60; Fig. 5A,B).Interestingly, phospho-EB2 levels were developmentally regulated, showing a substantial reduction between day 7 and day 14 in both wildtype and Cdkl5 − /Y mice (Fig. 5A,C, +/Y -71.6±2.9%,− /Y -81.3 ±2.1%).The comparison of EB2 phosphorylation between Cdkl5 − /Y and wild-type mice showed lower phospho-EB2 levels in Cdkl5 − /Y mice brains at all ages (Fig. 5A,C; P7 -58.6±3.5%,P14 -59±7.9%,P21 -53.4 ±14%, P60 -88.2±4.9%).Interestingly, the percentage of reduction in EB2 phosphorylation in Cdkl5 -/Y mice in comparison with their wildtype counterparts was higher in adult mice (P60), the age at which Cdkl5 is more greatly expressed, confirming a direct correlation between Cdkl5 activity and EB2 phosphorylation levels.Total levels of EB2 protein remained unchanged between day 7 and day 14 for both Cdkl5 − /Y and wild-type mice showing instead a decreasing trend to P60 (Fig. 5A,D).

Early appearance of abnormal motor-reflex behavior responses in Cdkl5 − /Y mice
Movement and coordination difficulties are the most common signs, along with delays in reaching motor developmental milestones in CDD children.To investigate the possible functional impact of the brain defects revealed in Cdkl5 − /Y pups, we tested mice for motor ability (Surface Righting test) and labyrinth reflexes (Cliff Aversion test), as well as for strength and motor coordination (Cliff Aversion and Negative Geotaxis tests) (Fig. 6A).All these developmental tests require perception of body orientation in space.Righting reflex, negative geotaxis, and cliff drop aversion tests revealed a poorer performance of Cdkl5 − /Y pups in comparison with their wild-type counterparts (Fig. 6B-D), indicating that developmental structural defects reflect altered behavior.

Discussion
This study demonstrates that most of the structural deficits of the brain in Cdkl5 KO mice appear early during postnatal development (summarized in Supplementary Table 3).The earliest observed changes in 7-day-old mice include a reduction in the dendritic arbor of the pyramidal neurons, a deregulation in excitatory and inhibitory synaptic terminals, an increase in microglia activation, and a decrease in the survival of newborn hippocampal granule cells.A deficit in dendritic spines is not present in 7-day-old Cdkl5 − /Y mice, but a reduction in dendritic spine density and maturation was observed in 14-day-old Cdkl5 − /Y mice.These early defects are mostly retained during further stages of brain development, and do not worsen or improve with age, as Cdkl5 − /Y and wild-type mice are characterized by similar agerelated changes during postnatal brain development (Supplementary Table 4).Confirming the early presence of brain defects, Cdkl5 − /Y pups (5-8 day-old mice) are characterized by an impairment in neonatal reflexes.
Dendritic arborization is significantly reduced in cortical pyramidal neurons from adult Cdkl5 KO mice (Amendola et al., 2014;Ren et al., 2019).In the current study, we provide novel information regarding the appearance of dendritic abnormalities in the somatosensory cortex of a mouse model of CDD, the Cdkl5 − /Y mouse.We found that at the end of the first week of age the dendritic tree of the pyramidal neurons of Cdkl5 − /Y mice has a reduced number of branches and a reduced length.An analysis of the number and length of branches of each order showed differences between P7 Cdkl5 − /Y and wild-type mice.These defects in dendritic complexity are retained during further stages of development, indeed, young (P14), juvenile (P21), and adult (P60) Cdkl5 − /Y mice have cortical pyramidal neurons with a hypotrophic dendritic arbor in comparison with wild-type mice.These defects do not worsen with age, as seen in the similar age-related increase in the number and length of dendritic branches in both Cdkl5 − /Y and wild-type mice.
The process of dendritogenesis is controlled by an intrinsic genetic program and extracellular signals (Urbanska et al., 2008).The mechanisms whereby the absence of Cdkl5 affects dendritic development remain elusive, although it is known that CDKL5 is not only required, but is sufficient to promote dendrite growth.It was previously shown that CDKL5 overexpression in SH-SY5Y neuroblastoma cells induces neurite outgrowth (Valli et al., 2012) and when overexpressed in cultured neurons, CDKL5 increased the total dendritic length (Chen et al., 2010), confirming that CDKL5 regulates a signaling pathway that controls dendrite growth.A recent study utilized iPSC derived neurons from patients with distinct mutations to show that loss of CDKL5 led to a decrease in phosphorylation of the CDKL5 substrate EB2.As a microtubule regulator, EB2 phosphorylation status was shown to be critical for microtubule dynamics within dendrites (Baltussen et al., 2018), suggesting that EB2 phosphorylation levels were involved in dendritic outgrowth.In support of this, as previously reported (Baltussen et al., 2018), we found a reduction of phospho-EB2 levels in the mouse cortex with age, which links EB2 phosphorylation to the developmental time of neurite outgrowth.Specific studies are required to determine the role of phospho-EB2 and, possibly, other pathways in the regulation of dendritic development at different developmental time points in the Cdkl5 KO brain.
The cell bodies of pyramidal neurons change markedly as pyramidal neurons mature.Not only does the shape of the cell body transform from a spheroid to a more pyramidal shape, but the size of the soma increases substantially in the first few weeks after birth (Miller and Peters, 1981).Interestingly, pyramidal neurons of the somatosensory cortex had a soma with a smaller size in P7 Cdkl5 − /Y mice versus their wild-type counterparts, subtending a reduced neuronal maturation in the absence of Cdkl5 which also involves the soma size of the pyramidal neurons.However, differently from the dendritic arbor, the soma size defect changes over time and it is not retained during further stages of development.In Cdkl5 − /Y mice, probably due to some form of compensation that takes place between P7 and P14, the cell body of Cdkl5 − /Y pyramidal neurons no longer differs from that of wild-type counterparts of the same age.
Alterations in spine density and maturation correspond with aberrant brain function observed in various neurodevelopmental and neuropsychiatric disorders (Khanal and Hotulainen, 2021).Various studies in vitro and in vivo converge regarding the role of CDKL5 in controlling dendritic spine morphology and excitatory functions (Barbiero et al., 2019).Previous evidence showed differences in dendritic spine density and maturation of cortical pyramidal cells in juvenile (P27-P28) and adult Cdkl5 KO in comparison with wild-type mice (Della Sala et al., 2016;Ren et al., 2019).Neurons that are devoid of CDKL5 are characterized by an increased number of filopodia-like protrusions and a reduced number of mushroom-shaped spines (Della Sala et al., 2016;Ricciardi et al., 2012).The enlarged spine heads, which are distinguished by their mushroom-shaped spines, correlate with increased synaptic strength.Importantly, an alteration in dendritic spine formation was also observed in neurons derived from induced pluripotent stem cells reprogrammed from patient fibroblasts, suggesting that spine and synaptic abnormalities are also likely to be present in patients (Ricciardi et al., 2012).
We found no differences between P7 Cdkl5 − /Y and wild-type mice in the density of dendritic spines as well as in the percentage of mature spines.In contrast, an evaluation of basal dendritic spines in Cdkl5 − /Y mice at later ages showed genotype-related differences.Spines are dynamic structures and their shape, size, and density change with developmental age and depend upon the balance among spine initiation, stabilization, and pruning rate.The evidence that in the cortex of Cdkl5 − /Y mice new spines are normally generated (spine initialization) but fail to stabilize (spine maturation), and are eliminated (spine pruning) at an abnormally high rate (Della Sala et al., 2016) may explain our finding that in 7-day-old Cdkl5 − /Y neurons, which have mainly immature spines (86%), differences in spine density are still not present.
Taken together our results show that, while very precocious defects are present in the basal dendritic arbor of pyramidal neurons in Cdkl5 − /Y mice, defects in spine density and maturation start to appear slightly later in development.This suggests that defects in spine development due to lack of Cdkl5 are temporally shifted in relation to the process of dendritogenesis.
Absence of CDKL5 profoundly affects not only the number of spines, the postsynaptic sites for excitatory synapses, but also the excitatory and inhibitory presynaptic terminals (Pizzo et al., 2016).As expected, since synaptogenesis occurs largely during the first postnatal period in rodents, with the greatest increases in synaptic density during the phase of the greatest synaptogenesis (P11-P16 in the mouse), we found an increase in the density of excitatory and inhibitor terminals from P7 to P14 in both Cdkl5 − /Y and wild-type mice.However, the development of the number and stabilization of synaptic terminals was not the same in Cdkl5 − /Y and in wild-type mice.A higher number of putative excitatory terminals labeled for VGluT1 had previously been observed in the cortex of adult Cdkl5 − /Y mice compared to their wild-type counterparts (Pizzo et al., 2016).In agreement with this finding we found a higher density of excitatory terminals in the cortex of Cdkl5 − /Y mice at all analyzed ages.However, the difference in the density of excitatory terminals between Cdkl5 KO and wild-type mice increases with age.This could be attributed to an increase in callosal excitatory synaptic inputs with ageindeed, a functional hyperconnectivity across hemispheres has been described in Cdkl5 KO mice (Awad et al., 2023) and/or to a compensatory mechanism established to counteract reduced postsynaptic functionality (number of mature spines).In addition to changes regarding the glutamatergic terminals, we found an increased density of GABAergic terminals in Cdkl5 − /Y mice.Our finding that in the wildtype cortex the density of inhibitor terminals increased from P7 to P14, and then decreased by 44% to P21 and adult levels, is in line with the finding that in the rat visual cortex symmetrical synapses (putatively inhibitory synapses) decline significantly following the second/third week after birth (Blue and Parnavelas, 1983).This decline was also evident in the somatosensory cortex of the mouse, but occurred after P32 (De Felipe et al., 1997), indicating that the inhibitory synapses are affected by the pruning process.Interestingly, we found an increased density of GABAergic axonal boutons in the cortex of Cdkl5 − /Y mice during postnatal development at P7 and P21 and in adulthood; however, fewer inhibitory terminals were present at P14, in comparison with their age-matched wild-type counterpart.We can hypothesize that, in the absence of Cdkl5, the period of the high rate of production of new inhibitory synapses does not occur or lasts for a briefer time, inhibiting the transient increase in the number of inhibitory synapses.The mechanisms responsible for the age-related increased density or pruning of inhibitory terminals in Cdkl5 − /Y mice are presently unclear.However, CDKL5 was reported to be expressed at high levels in GABAergic interneurons (Rusconi et al., 2008), suggesting a possible involvement of Cdkl5 in the kinetic of synaptogenesis of the inhibitory neurons.A study aimed at evaluating the inhibitory and excitatory postsynaptic currents will be an important future step to determine age-related synaptic function alterations in the absence of CDKL5.
Microglia activation, associated with an increase in cell body size, as well as with cytokine alterations, has recently been described in Cdkl5 KO mice (Galvani et al., 2021).Our results highlight the presence of a generalized status of microglia overactivation in the brain of the Cdkl5 − /Y mouse, that is already present at a very early stage of life.As previously described (Lawson et al., 1990;Perry et al., 1985), we observed that during postnatal brain development in the mouse microglia gradually accumulate, dramatically increasing in number between P7 and P14; by P14, the entire cortical parenchyma is filled with ramified microglia.Microglia make essential contributions to CNS development.
They regulate neurogenesis and neuronal migration in addition to facilitating neural connectivity by regulating axon outgrowth, myelination, and remodeling synaptic connections (Frost and Schafer, 2016;Li and Barres, 2018;Sato, 2015;Tremblay et al., 2010).Interestingly, in the developing brain, microglia interact specifically with neuronal synapses, thus influencing synaptic maturation and pruning (Parkhurst et al., 2013;Tremblay et al., 2010).Our previous finding that inhibition of neuroinflammation through treatment with luteolin improves dendritic pathology in (3-4-month-old) Cdkl5 +/− female mice (Tassinari et al., 2022) strongly suggests that an inflammatory state is associated with the pathophysiology of the adult CDD brain.The early arrival of microglia in the developing brain, together with the presence of an activated phenotype in the absence of Cdkl5, suggests a possible developmental microglia-neuron crosstalk that may affect neuronal maturation and survival in the Cdkl5 null brain at a very early age.
The majority of the immature granule neurons, derived from primary progenitor cells located in the subgranular layer of the dentate gyrus (DG), are subjected to a selection process, during which they are either recruited into function or eliminated (Biebl et al., 2000).Cdkl5 loss causes hypocellularity of the DG due to the increased apoptotic cell death of immature granule neurons (DCX+ cells) that overrides the proliferation of their progenitors (Fuchs et al., 2014).Here we found that loss of Cdkl5 already decreases survival of postmitotic neurons very early on in postnatal hippocampal development.There is evidence that activated microglia, through the integrin CD11b in cooperation with the immunoreceptor DAP12, trigger developmental apoptosis of hippocampal neurons via the production of superoxide ions (Wakselman et al., 2008).It is therefore possible that Cdkl5 loss-induced microglia activation could also lead to excessive immature newborn neuronal death.Future studies aimed at evaluating the effect of the inhibition of neuroinflammation during the early phase of brain development in Cdkl5 KO mice will help to clarify whether microglia activation underlies early postnatal structural deficits in the Cdkl5 null brain.
Our finding that most of the reported brain alterations appear to be already well established in 7-day-old Cdkl5 pups suggests a very early onset of these defects.However, we are aware that there are limitations to our study.First, we did not identify at what stage, before 7 days of age, the brain defects appear in Cdkl5 − /Y mice.Second, the behavioral analysis was carried out within a limited time window (5-8 days).Therefore, even if this analysis supports the presence of early behavioral alterations, it does not allow us to identify the onset of behavioral defects and their temporal progression.Considering that CDKL5 is expressed in foetal brains both in humans and mice (Hector et al., 2016), future studies of brain development/function during perinatal and prenatal ages in Cdkl5 KO mice will provide indications regarding the onset and progression of brain alterations.In addition, although most CDD patients are females who are heterozygous for CDKL5 deficiency, in this first comparative study of brain development in the early postnatal period through adulthood we used, as a CDD disease model, Cdkl5 − /Y male mice.A study aimed at characterizing brain development in Cdkl5 +/− females will be an important future step that will help to clarify the complex CDD pathophysiological processes in a context in which the mosaicism of Cdkl5 expression could influence the phenotypic outcome.

Conclusions
Taken together, the current findings show that dendritic pathology, the main cause of behavioral defects such as motor and cognitive disability, is detectable very early on in the brain of Cdkl5 − /Y pups.Similarly, a decreased survival of postmitotic neurons that causes hypocellularity of the DG appears early in the postnatal developing brain, probably compromising the function of the hippocampal circuit.Both humans and rodents undergo extensive postnatal neurological development, and have a relatively immature CNS at birth (Zeiss, 2021).The primary developmental events occurring postnatally in humans are synapse production, myelination, and neuronal and synaptic pruning.In general, the same processes occur in rodents.For example, maximum brain growth occurs after birth both in humans and rodents, and a significant amount of synaptogenesis continues postnatally.Although postnatal development of CNS is not identical between humans and rodentsthere are examples in which the human is more developed at birth than rodents, such as neurogenesis in the dentate gyrus (Zeiss, 2021) rodents appear to be an appropriate model for evaluating genotype-related effects on postnatal brain development in humans (Watson et al., 2006).Therefore, even if knowledge of the timeline of dendritic pathology in CDD patients is still missing, the current findings strengthen the suitability of the mouse model of CDD as a tool to test early therapeutic interventions aimed at ameliorating brain development and, therefore, intellectual disability in children with CDD.

Fig. 6 .
Fig. 6.Motor-reflexes in Cdkl5 KO pups.A: Timeline of when mice were behaviorally tested.The Surface Righting test was performed on postnatal day 5 (P5), the Negative Geotaxis test on postnatal day 7 (P7), and the Cliff Aversion test on postnatal day 8 (P8).B: Surface righting reflex in Cdkl5 -/Y mice (n = 8) and wild-type littermates (+/Y n = 18).Histograms show the average time required for pups placed in the supine position to right themselves into the prone position.C: Abnormal negative geotaxis reflex in Cdkl5 -/Y mice (n = 9) and control littermates (+/Y n = 22).Histograms show the latency of pups to turn and orient themselves facing up the slope.D: Cliff aversion reflex in Cdkl5 -/Y mice (n = 9) and control littermates (+/Y n = 14).Graphs show the time required for pups to complete the reflex.The results in B-D are presented as means ± SEM. * p < 0.05, ** p < 0.01, *** p < 0.001 (two-tailed Student's t-test).Abbreviations: n = number of mice; P = postnatal day.

Table 1
Body weight.