Cytoplasmic-predominant Pten increases microglial activation and synaptic pruning in a murine model with autism-like phenotype

Germline mutations in PTEN account for ~10% of cases of autism spectrum disorder (ASD) with coincident macrocephaly. To explore the importance of nuclear PTEN in the development of ASD and macrocephaly, we previously generated a mouse model with predominantly cytoplasmic localization of Pten (Ptenm3m4/m3m4).Cytoplasmic predominant Pten localization results in a phenotype of extreme macrocephaly and autistic-like traits. Transcriptomic analysis of the Ptenm3m4/m3m4 cortex found upregulated gene pathways related to myeloid cell activation, myeloid cell migration, and phagocytosis. These transcriptomic findings were used to direct in vitro assays on Pten wild-type and Ptenm3m4/m3m4 microglia. We found increased Iba1 and C1q expression with enhanced phagocytic capacity in Ptenm3m4/m3m4 microglia, indicating microglial activation. Moreover, through a series of neuron-microglia co-culture experiments, we found Ptenm3m4/m3m4 microglia are more efficient at synaptic pruning compared with wild-type controls. In addition, we found evidence for neuron-microglia cross-talk, where Ptenm3m4/m3m4 neurons elicit enhanced pruning from innately activated microglia. Subsequent in vivo studies validated our in vitro findings. We observed a concurrent decline in the expression of Pten and synaptic markers in the Ptenm3m4/m3m4 cortex. At ~3 weeks of age, with a 50% drop in Pten expression compared with wild-type levels, we observed enhanced activation of microglia in the Ptenm3m4/m3m4 brain. Collectively, our data provide evidence that dysregulated Pten in microglia has an etiological role in microglial activation, phagocytosis, and synaptic pruning, creating avenues for future studies on the importance of PTEN in maintaining microglia homeostasis.


Introduction
The gene encoding Phosphatase and TENsin homolog deleted on chromosome TEN (PTEN) is a well-recognized, syndromic risk allele for autism spectrum disorder (ASD), a neurodevelopmental disorder defined by deficits in two core symptom domains: social communication/interaction and restricted/repetitive behavior [1][2][3][4]. Carrying a germline PTEN mutation is also the molecular criterion for a diagnosis of PTEN Hamartoma Tumor Syndrome (PHTS), irrespective of clinical phenotype. PHTS (OMIM 601728) is an autosomal dominant, cancer predisposition syndrome, where patients present variably but generally have macrocephaly, benign hamartoma of all three germ layers, elevated risks for specific malignancies, and macrocephalic ASD, the latter occurring in about 20% of PHTS individuals [5][6][7]. Conversely,~17% of individuals with ASD and macrocephaly harbor a germline PTEN mutation [2]. Follow-up studies confirmed this genetic association, adjusting the weighted average for PTEN mutation in ASD individuals with coincident macrocephaly to~10% [4,[8][9][10]. These genetic data highlight PTEN as one of the most prevalent and penetrant ASD-risk genes.
To explore the importance of PTEN in the development of ASD and macrocephaly, we generated a mouse model with a knock-in mutation disrupting two of the four putative nuclear localization signals of Pten [11][12][13]. The consequent Pten m3m4 mouse is characterized by Pten expression predominantly restricted to the cytoplasm with substantial depletion from the nucleus. This cytoplasmic-predominant model presents with macrocephaly due to megencephaly, hypertrophy of neural somas, gliosis, and autistic-like behavior [12]. RNA sequencing of the Pten m3m4/m3m4 hemi-brain at 2 weeks of age (i.e., P14) and cortex at 6 weeks of age (i.e., P40) demonstrated that the Pten m3m4/m3m4 transcriptome is reflective of idiopathic autism with many of the same genes differentially expressed [13]. Earlier analyses of the transcriptome of the cortex from Pten m3m4/m3m4 mice at 6 weeks of age showed increased expression of genes associated with neuroinflammation and significantly decreased expression of genes associated with synaptic transmission [13].
Our earlier studies on the Pten m3m4 model identified activated microglia in the cortex and hippocampus of the Pten m3m4/m3m4 mouse [12]. Microglia, the resident immune cells of the brain, participate in synaptic pruning, an important neurodevelopmental process. During synaptic pruning, microglia utilize C1q as the regulator of complementmediated pruning to target immature synapses for engulfment and subsequent removal [14,15]. It has been demonstrated that altered complement function and C1q expression in microglia are associated with deficits in synaptic transmission and ASD as a consequence of inappropriate synaptic pruning [16][17][18][19][20]. In one such study, the authors show that C1q protein expression is elevated in the serum of children with ASD compared with age-matched controls, further stressing the necessity of proper complement function for neuronal plasticity [16]. There are no published data on the role of Pten as a regulator of microglia function and synaptic pruning. Based on the transcriptomic and preliminary in vivo findings, we hypothesize that constitutive Pten dysfunction predisposes microglia to aberrant activation, leading to exaggerated participation in synaptic pruning during neurodevelopment. Using both in vitro and in vivo models, we demonstrate that Pten activity in microglia is essential for regulating synapse targeting and engulfment, as well as microglial morphology throughout development.

Transcriptomic data analysis
The cortical transcriptome of 6-week-old Pten m3m4 mice (GSE59318) mimics that of idiopathic autism, as described in detail in Tilot et al. [13]. We performed PANTHER GO classification on the list of differentially expressed genes in the cortex of Pten homozygous mutant and wild-type mice. Genes annotated as related to biological process and cellular components of interest were identified by immune system processes (GO: 0002376) and the synapse (GO: 0045202), respectively. Next, we performed a "core analysis" using Ingenuity Pathway Analysis (IPA) software (Qiagen, Redwood City, California). The core analysis was mined for significantly enriched pathways relevant to microglia activity, migration, and function, which were used to inform further in vivo and in vitro experiments.

Animals
Generation and characterization of the Pten m3m4 mouse on a CD1 background has been described previously [12]. The Pten m3m4 mutation is located within exon 7 of Pten and consists of five nucleotide substitution mutations, resulting in four nonsynonymous and one synonymous amino acid changes in the third and fourth putative nuclear localization sequences of Pten [11,12]. Genotyping was performed on genomic DNA from clipped toes per the Jackson Laboratory protocol using modified PCR primers. Wild-type allele primers: mPten-F5, 5′-TGGCAGACTCTTCATTTCTGT GGC-3′, and mPten-R6, 5′-ACTTCTTCACAACCACTT CTTTCAAC-3′. Mutant allele primers are mPten-F3, 5′-TACCCGGTAGAATTTCGACGACCT-3′, and mPten-R6, 5′-ACTTCTTCACAACCACTTCTTTCAAC-3′. Mice were maintained on a 14:10 light:dark cycle with access to food and water ad libitum. The room temperature was maintained between 18 and 26°C. Animals were euthanized via CO 2 asphyxiation followed by cervical dislocation. All experiments were not blinded but randomized and conducted under protocols approved by the Institutional Animal Care and Use Committee (IACUC) at Cleveland Clinic.
We utilized roughly equal numbers of male and female mice for our experiments. We show in a previous study no significant differences between Pten m3m4/m3m4 male and female gross cellular phenotype so all available samples were utilized accordingly [12]. These data are further supported by western blot experiments which denote the sex of each biological replicate used.

Cell culture
Mixed glia were obtained by trypsinization of P2 cortices followed by plating on poly-D-lysine coated T-75 culture flasks. Mixed glia cultures, were maintained in DMEM with 10% FBS and 1% Penicillin and Streptomycin (Pen/Strep). Once the mixed glia cultures reached DIV 10, they were agitated at 170 RPM for 1 h in order to isolate primary microglia. Isolated microglia were seeded on poly-D-lysine coated glass cover slips and used for immunofluorescent (IF) staining and phagocytic assays at DIV 3 post-shaking.

In vitro immunofluorescent staining
We cultured microglia or neurons on poly-D-lysine (PDL)coated cover slips until DIV 14. Microglia were washed with ice-cold PBS and fixed in ice-cold methanol for 2 min. This was followed by three washes for 5 min each with icecold PBS. We then permeabilized the microglia with 0.03% Triton X-100 dissolved in PBS for 4 min. Next the cultures were blocked with 10% normal goat serum for 1 h at room temperature, followed by incubation with primary antibody: Pten

Quantification of functional synapses
We fixed neuron/microglia co-cultures and stained for the synaptic markers, Synaptophysin at 1:500 dilution of primary antibody, and Psd-95 at 1:250 dilution of primary antibody once the neurons had reached DIV 14. Neuron/microglial cultures were imaged using confocal microscopy (Leica Biosystems, Richmond, Ilinois) and analyzed using the Volocity v6.3.0 (Quorum Technologies Inc., Puslinch, Ontario, Canada) to count functional synapses (co-localized Syn and Psd-95) along individual neurites. Functional synapse number was then normalized to the length of the respective neurite.

Phagocytosis assay
We plated primary microglia at a density of 1 × 10 5 in a 12well dish with PDL-coated coverslips for 48 h in a 37°C cell incubator with 5% CO 2 and 100% humidity. Next, we blocked 1 μm fluorescent beads (#L1030, Sigma-Aldrich) in FBS for 1 h at 37°C at a ratio of 1:5 v/v. Florescent beads were diluted with DMEM to reach a final concentration of 0.01% (v/v). Microglial culture media was replaced with 250 μl DMEM containing beads, and incubated for 1 h at 37°C in a cell incubator. Cultures were washed thoroughly five times with ice-cold PBS and fixed in ice-cold methanol prior to immunofluorescent staining for Iba1 (1:500, #019-19741, Wako).

Immunofluorescent quantification
We captured images of brain sections as well as cells grown in vitro as confocal images using a Leica TCS-SP8-AOBS inverted confocal microscope (Leica Microsystems, GmbH, Wetzlar, Germany). Brain section, primary neuron, and/or microglial co-cultures were imaged with a minimum of n = 5 biological replicates. We used ImageJ software to measure area and intensity of the stain and calculated integrated density of brain images. In addition, ImageJ was used to measure area of stain per microglia in vitro and in vivo to assess morphological changes beyond "bushy" and "amoeboid" [21]. Primary neuron/microglial co-cultures were analyzed using the Volocity 3D imaging software in order to count the number of functional synapses and normalize these values to neurite length. Due to the low abundance of primary microglia resulting from our isolation protocol, we pooled microglia from various cultures where appropriate. These microglia were maintained in culture and stained accordingly. We captured a minimum of five images per cultured microglia genotype. Finally, we quantified these images using ImageJ to measure signal, area and integrated density.

Statistical analysis
Sample size was determined according to our previous studies utilizing the Pten m3m4 model [12,13,22]. These studies were conducted in a non-blinded manner with sample utilization being randomized. Statistical tests are justified according to the distribution of the data, and also if data meets the assumptions of each analysis. We analyzed normally distributed data using a one-way analysis of variance (ANOVA) or Student's t-test, where appropriate (Graph Pad Prism 8). After performing a one-way or twoway ANOVA when appropriate, we performed a post-hoc Tukey-Kramer analysis (F). When data were not normally distributed, we performed non-parametric analyses including Mann-Whitney U and Kruskall-Wallis tests (H), where appropriate (Graph Pad Prism 8). P values that are less than 0.05 were considered statistically significant. We calculated effect sizes as the mean difference with 95% confidence intervals using the R package DABESTR [23]. In addition, Spearman rho R correlation matrix was used to determine correlations between data plots when appropriate (Graph Pad Prism 8).

Results
The Pten m3m4 neural transcriptome is enriched in neuroinflammation networks, including microglial activation and phagocytic activity Our earlier studies have revealed a neuroinflammatory signature in the cortical transcriptome of 6-week-old Pten m3m4/m3m4 mice [4]. Using this platform, we found a significant increase in expression of complement-related proteins, and genes implicated in synaptic pruning in Pten m3m4/m3m4 mice compared with Pten WT/WT controls (Table 1). In addition, numerous inflammation-related and synapse-related genes also showed differential expression (Supplementary Tables 1 and 2). These results highlight dysregulation of mediators of immune and synaptic function in Pten m3m4/m3m4 mice.
Subsequent gene enrichment and pathway analyses (Ingenuity Pathway Analysis [IPA] software) on the transcriptomic data from the cortices of 6-week-old Pten m3m4/m3m4 mice, identified significant enrichment of pathways defined as "activation of myeloid cells", "phagocytosis", "cell movement of phagocytes" (Fig. 1a, Supplementary Fig. 3a, b) and phagosome formation (Fig. 1b). Collectively, these analyses strongly suggest activation of microglia and increased phagocytosis.
Pten m3m4/m3m4 microglia show evidence of enhanced synaptic pruning activity To evaluate synaptic pruning capacity of Pten mutant microglia we co-cultured primary neurons and microglia from Pten WT/WT , Pten WT/m3m4 , or Pten m3m4/m3m4 mice in differing combinations for 72 h. Co-localization of pre-and post-synaptic markers, Syn and Psd-95, respectively, was used to determine presence of functional synapse and normalized to neurite length (Fig. 2c, d). Upon analysis of our microglia-neuron co-cultures we found the most dramatic decrease in synapse number with Pten m3m4/m3m4 microglia co-cultures compared with Pten WT/WT microglia co-cultures (mean difference of synapses per 10 µm = 0.74; F = 57.52, p value ≤ 0.0001) and also compared with neuronal cultures without microglia (mean difference of synapses per 10 µm = 1.62; F = 57.52, p value ≤ 0.0001) (Fig. 2c, d). Subsequent analysis shows the highest numbers of functional synapses per 10 µm of neurite length were observed when no microglia were present in the neuronal cultures of all   Fig. 4c, d). We confirmed that the decrease in synapse number in the co-culture was due to microglial-pruning, and not due to neuronal death by co-staining for NeuN and Cleaved Caspase-3. No increase in neuronal death was observed ( Supplementary Fig. 5a). These data show that neurons from Pten m3m4/m3m4 mutants form more synapses compared with Pten WT/WT neurons as shown in earlier studies. However, in a systemic germline Pten m3m4/m3m4 mutant system, our data suggest cross-talk between neurons and microglia, resulting in significantly increased synaptic pruning in Pten m3m4/m3m4 mutants compared with Pten WT/WT mice.

Pten m3m4/m3m4 mice have reduced expression of synaptic markers and increased microglia activation concurrent with decreased Pten expression in the forebrain
The in vitro data presented above show that Pten m3m4/m3m4 microglia have enhanced phagocytosis and synaptic pruning capacity with increased C1q expression. These findings were further validated by our Pten m3m4/m3m4 murine model characterized by a systemic cytoplasmic predominant Pten.
Western blot analyses showed that Pten expression is significantly reduced in the hemibrains of Pten m3m4/m3m4 mice compared with those of Pten WT/WT controls at P2 (relative expression = 0.63, p value = 0.002) and P8 (relative expression = 0.52, p value = 0.002) (Fig. 4a-c) and later ages ( Supplementary Fig. 1a-e). As the Pten m3m4/m3m4 mouse continues to age, Pten expression progressively declines in the cortex until reaching its nadir at P40 (relative expression = 0.36, p value = 0.001) (Fig. 4a). Interestingly, it is also at this time point, the terminal age for this mouse model, we observe a significant decrease in expression of Syn (relative expression = 0.62, p value = 0.02) and Psd-95 (relative expression = 0.60, p value = 0.01) (Fig. 4a-c). This is concordant with increased expression of C1q (relative expression = 4.6, p value = 0.001) (Fig. 5g) in the Pten m3m4/m3m4 cortex. In order to determine if a correlation existed between the trends in Pten, Syn, Psd95, and C1q expression over time, we performed a Spearman rho R correlation matrix. We find there to be a correlative expression between Pten and Psd95 (r = 0.90), as well as inverse correlative expression between Pten and C1q (r = 0.70), C1q and Syn (r = 0.60), and C1q and Psd95 (r = 0.90) ( Supplementary Fig. 2a, b). In addition, we find that only at P40 that a strong correlation exists between decreased expression of Pten and decreased expression of the synaptic marker Syn (r = 0.94, p value = 0.01). In addition, in Pten m3m4/m3m4 hippocampus, we observe decreases in Syn and Pten expression at the P40 time point relative to Pten WT/WT controls by Western analysis (Supplementary Fig. 6a, b, c, e). Syn expression on the surface of hippocampal CA3 neurons of Pten WT/m3m4 and Pten m3m4/ m3m4 mice actually showed a significant increase in expression by IF analyses during early development at age  P14 (Mean Syn: Pten WT/WT = 1.05 ± 0.13 vs. Pten m3m4/m3m4 = 1.85 ± 0.41, p = 0.01 (Fig. 4d, e). However, by P40 Syn expression in the CA3 neurons is significantly decreases in Pten m3m4/m3m4 mice compared with their wild type littermates (Mean Syn: Pten WT/WT = 0.96 ± 0.33; Pten m3m4/m3m = 0.47 ± 0.04, p value = 0.03) (Fig. 4d, e). To further investigate the increased synapse number in hippocampus at P14 followed by significantly decreased expression at P40, we looked at the presence and activation status of the microglial pruning machinery in the hippocampus. We found that at P14, there is no significant increase in activated microglia (Fig. 4d, e). In contrast at P40, Pten m3m4/m3m4 mice showed a significant increase in microglia activation compared with Pten WT/WT mice as estimated by activated morphological changes and increased cell area at P40 (Iba1 area: Pten WT/WT = 3.62 ± 0.89 vs. Pten m3m4/m3m4 = 10.41 ± 1.70, p value = 0.001) (Fig. 4d, e), as well as by Iba1 expression levels (relative expression: 1.90 ± 0.44; F = 0.7, p value = 0.002) (Supplementary Fig. 6d). In addition, C1q expression was also found to be significantly increased in Pten m3m4/m3m4 mice compared with Pten WT/WT at P40 both in the hippocampus (relative expression: 1.78, p value = 0.004) (Supplementary Fig. 6d). Western blot analysis showed no statistically significant increase in C1q expression during earlier developmental time points (Fig. 5g,  Supplementary Fig. 1a-e).
These data suggest a correlation in vivo between activation of microglia, increased expression of C1q and decrease in synaptic proteins in the brain, but it is only when the Pten m3m4/m3m4 mouse reaches P40, near its terminal age and coincidentally the nadir of Pten expression do we observe the largest effect.
Pten m3m4/m3m4 microglia become activation at P21 in mouse cortex To understand the role of microglial activation and its temporal effect on synapse numbers in the brain, we analyzed microglial activation across development using IF. No significant differences in microglial cell area were observed at P2, P8, or P14 in Pten WT/m3m4 or Pten m3m4/m3m4 mouse cortex compared with wild-type cortex (Fig. 5a-f).
Collectively, our in vivo data suggest that dysregulated Pten expression is correlated with changes in microglial activation status and a subsequent decrease in number of synapses at later ages. Although the Pten m3m4 mouse is predisposed to increased synapses consistent with observations in neuron-specific Pten mutant models, the decrease in synapses begins to take place by P21, a correlated time point when Pten levels plunge below 50% of wild-type mice (Fig. 4a). Incidentally, synaptic pruning has been shown to peak in mice also at P21 [24]. We show that dysregulated feedback cues between microglia and neurons involving C1q expression may play some role in the exaggerated decrease in synapses in Pten m3m4 mutants.

Discussion
This study demonstrates that germline Pten mutation that disrupts Pten nuclear localization results in changes in microglial morphology, activation, and function. Using IPA, we re-analyzed the published Pten m3m4/m3m4 neural transcriptome [13], finding enriched networks related to neuroinflammation and synaptic transmission (Fig. 1a, b). We were able to bring these pathway-based predictions to in vitro and in vivo validation. Initially, we utilized a series of co-culture experiments with microglia and neurons from Pten WT/WT and Pten m3m4/m3m4 genotypes. Our results demonstrate that cytoplasmic-predominant Pten in Pten m3m4/m3m microglia leads to a greater propensity to prune synapses (Fig. 2), enhanced phagocytic capacity, and increased expression of C1q (Fig. 3). We next confirmed these in vitro findings using the Pten m3m4/m3m mouse, showing decreased expression of synaptic proteins and increased microglia activity in the cortex and hippocampus ( Fig. 4 and Supplementary Fig. 6d). This study is the first to demonstrate a function for Pten in regulating not only microglial activation but also synaptic pruning.
In this study, we demonstrate the first microglial pathology subsequent to a germline Pten mutation arising from cell-autonomous mechanisms. Other Pten models have observed similar activated microglial morphologies but fail to pursue any inquiry into Pten function in microglia, making them the last neural cell type requiring Pten function to be characterized [25,26]. Here, utilizing in vitro culture techniques we were careful to separate the Pten m3m4 microglia from neuronal and glial influences and still demonstrate a microglial pathology (Figs. 2 and 3). Although there is a vast body of literature demonstrating that microglial-dependent synaptopathies are associated with ASD-like behaviors in mice [24], this study does not demonstrate a direct cause-and-effect relationship between our reported microglia pathology and the ASD-like characteristics of the Pten m3m4 model [4,12,13]. These observations are purely associative, and it is unknown if the mutant microglia have an impact on neuronal function and the ASD-like phenotypes reported in our model.
Although this study of the effect of Pten disruption in microglia is unique, there have been other studies on the effect of Pten disruption in myeloid-derived cells, the cellular precursors of microglia. These studies partially corroborate our findings. For instance, when Pten is knocked out in myeloid-derived cells (MDCs), they show increased PI3K signaling, elevated G-CSF secretion, and enhanced proliferation [27]. In addition, Pten knockout in MDCs increases phagocytosis of apoptotic cells and differences in cytokine production [28].
Interestingly, there are also some parallels between the microglial phenotypes in the Pten m3m4 model and the microglial phenotypes associated with neurodegenerative disorders, such as Alzheimer's disease [29,30]. The activation of microglia and increased synaptic pruning are a feature of many models of neurodegeneration. Collectively, these data suggest that Pten functions as a negative regulator of microglial activity in a cell-autonomous fashion. This is manifested most clearly in our in vitro studies of microglia morphology, Iba1 expression, and phagocytosis activity. Additional studies will be required to elucidate the undoubtedly numerous pathways by which Pten disruption contributes to microglial pathology.
In addition, we observed increases in the expression of molecules related to complement and innate immune function: C1qa-c, C3ar1, Cx3cr1, Itgam, Itgb2, and Trem2. This finding is often reported both in ASD and neurodegenerative models with activated microglia [24,30,31] ( Table 1). The overexpression of C1q is the most telling of these molecules given its status as a critical mediator of synaptic pruning (Fig. 3a, b). This study is the first reported correlation between Pten and C1q. Thus, our work suggests Pten participates in the regulation of synaptic pruning via complement signaling. Another study associated Pten disruption in an MKPOSE model to changes in complement expression, but the genetic complexity of that model previously obfuscated what is now a clear molecular connection [32]. Future studies will be important to dissect out how Pten function regulates C1q expression and determining if there are subsequent physiological consequences relevant to synaptic transmission, neuronal plasticity, and autism behavior.
Here, we show the synaptic pruning by Pten m3m4/m3m4 microglia is due both to cell-autonomous and non-cellautonomous mechanisms as outlined by our current working model (Supplementary Fig. 7a-c). It is evident that there is an additive effect on pruning from the combination of Pten m3m4/m3m4 neurons with Pten m3m4/m3m4 microglia, illustrated by both the increase in phagocytosis and the increase in C1q foci in these co-cultures (Figs. 2g, and 3c). It is important to note that although we observe decreased nuclear Pten in the brain of the Pten m3m4 mouse, we also observe decreased cytoplasmic Pten [12]. We believe this to be, in part, due to decreased Pten stability, as well as increased proteasomal degradation as a result of the inability of Pten to evade cytoplasmic ubiquitination. In light of this, it is important to note that both decreased stability and nuclear localization have been observed in a subset of human mutations involving Pten from ASD patients [33,34]. Therefore, the Pten m3m4 mouse is currently the best in vivo mouse model for this type of mutation.
In addition, further work is required to dissect whether the enhanced pruning is because of the increase in dendritic branching (i.e., more "weak" synaptic connections) in the Pten m3m4/m3m4 neurons or because of an unknown "eat me" signal [14,15]. As discussed earlier, we observed that Pten m3m4/m3m4 neurons cultured in vitro have more synapses compared with Pten WT/WT and Pten WT/m3m4 ( Supplementary  Fig. 4a, b). These data are supported by the work of others who have shown Pten deletion in neurons results in increased synaptic spine density and dendritic arborization [35][36][37].
Interestingly, in vivo, we observe a significant decrease in Syn expression at P21, and by P40, this effect is even greater with significantly decreased Psd-95 expression (Fig. 4b, c). While in vitro, we see significant decreases in synapses at DIV 14 when co-cultured with microglia for 7 days. Perhaps these subtle differences in the timing of decreased synapses may be due to the microenvironment of our neuron-microglia co-culture system, which may not be as complex as the in vivo context [38,39]. These difference in the timing of synaptic loss may also be partially due to a lack of the milieu of exogenous cellular factors, which also must undergo temporal changes, that would only be present in vivo. The literature has demonstrated that synaptic pruning peaks in mice at P21 [15], when we first see significantly decreased Syn expression and increased microglia activation in the Pten m3m4/m3m4 model (Figs. 4b and 5a, e). We would like to note, that although there is not a significant change in microglia morphology in the Pten m3m4/m3m4 until P21, we did observe some microglia activated in the cortex at P14, which is concordant with synaptic pruning in our DIV 14 neuron-microglia co-cultures (Fig. 5a). Despite the primary neuron cultures only having 7 days to incubate with microglia, they show signs of significant synaptic pruning already, suggesting that pruning occurs earlier in vitro than in vivo. This could be explained by one of two non-mutually exclusive contexts: (1) priming of primary microglia due to culture conditions prior to isolation and incubation with primary neurons; or (2) exogenous regulators (other cells and their signals) that are lacking in the in vitro system. The direct mechanism of how synaptic pruning in microglia is initiated by P21 and subsequently repressed remains unclear, however our data suggest that Pten signaling may be a key regulator of this process.
Another avenue to consider in this study and moving forward is that the Pten m3m4 model is not a conventional "knockout", rather it is a "knock-in" germline mutant. This is an innovative aspect to our study since the germline model more accurately recapitulates the systemic effects of Pten dysfunction observed in our PTEN germline mutationpositive patients with ASD. It would be interesting to investigate other human-PTEN-ASD mimicking mutant mouse models with respect to Pten structure, function, subcellular localization, and overall phenotype, including the effects on neurons, astrocytes, oligodendrocytes, and microglia. Perhaps some mutations would present in a "spectrum" of minor to severe phenotypes as described in clinical cases of ASD with respect to behavior, and perhaps even to cellular severity of the phenotype.
This study identifies a new role for Pten in influencing microglia activity, especially in the context of synaptic pruning, a normal neurodevelopmental process that supports a healthy synaptic architecture. Although this study is limited in that it cannot cleanly distinguish between the effects of cell-autonomous and exogenous factors on synaptic pruning, it identifies an important role for Pten in microglia, where Pten acts to guard against neuroinflammation, preventing aberrant activation and enhanced phagocytosis. Moreover, this observed microglia pathology may be relevant to neurodevelopmental and synaptopathic phenotypes such as autism.