Nuclear PTEN deficiency causes microcephaly with decreased neuronal soma size and increased seizure susceptibility

Defects in phosphatase and tensin homolog (PTEN) are associated with neurological disorders and tumors. PTEN functions at two primary intracellular locations: the plasma membrane and the nucleus. At the membrane, PTEN functions as a phosphatidylinositol (3,4,5)-trisphosphate phosphatase and suppresses PI 3-kinase signaling that drives cell growth and tumorigenesis. However, the in vivo function of nuclear PTEN is unclear. Here, using CRISPR/Cas9, we generated a mouse model in which PTEN levels in the nucleus are decreased. Nuclear PTEN-deficient mice were born with microcephaly and maintained a small brain during adulthood. The size of neuronal soma was significantly smaller in the cerebellum, cerebral cortex, and hippocampus. Also, these mice were prone to seizure. No changes in PI 3-kinase signaling were observed. By contrast, the size of other organs was unaffected. Therefore, nuclear PTEN is essential for the health of the brain by promoting the growth of neuronal soma size during development.

In the nucleus, PTEN has been shown to control DNA damage responses and maintain genome integrity through interactions with nuclear AKT and p53 (7)(8)(9)(10)(11)(15)(16)(17)(18). PIP3, in addi-tion to PI3Ks and AKT, is also found in the nucleus, although the levels of PIP3 are considerably less (19). Therefore, the function of PTEN in the nucleus has been suggested to be independent of its PIP3 phosphatase activity, distinct from its function at the plasma membrane (11,20). Cytosolic and nuclear PTEN may take different conformations to undergo phosphatase-dependent and phosphatase-independent functions (21). It has been reported that nuclear PTEN displays protein phosphatase activity against other nuclear proteins or undergoes other types of protein-protein interactions (11,16). For example, the lipid and protein phosphatase activities of PTEN are dispensable for the function of nuclear PTEN in tumor suppression through regulating the anaphase-promoting complex (16). However, because these studies of nuclear PTEN have been primarily carried out in cell culture, the role of nuclear PTEN in vivo remains unknown.
The nuclear localization of PTEN is dynamic. The amount of PTEN in the nucleus changes in response to a wide range of cellular and environmental cues, such as cell cycle progression, development, oxidative stress, and neurotoxins (7)(8)(9)(10)(11). In neurons, where PTEN is highly expressed, PTEN transits in and out of the nucleus under physiological conditions, such as brain development, and under pathological conditions, such as ischemia, brain injury, and excitotoxicity (11,22). Monoubiquitination of lysine 13 has been identified as a critical step for targeting PTEN to the nucleus (23-25). Substitution of lysine 13 with arginine inhibits the localization of PTEN to the nucleus (25).
Loss of total PTEN (e.g. both nuclear and nonnuclear PTEN) in the brain leads to an increase in soma size, ectopic dendrites, and macrocephaly, as well as the formation of excess synapses because of increased PIP3 signaling via increased phosphorylation of AKT and mTOR at the plasma membrane (1, 26 -28). These previous studies have demonstrated the importance of total PTEN in vivo. A next overarching challenge in PTEN biology is to address the in vivo function of nuclear PTEN, separately from the function of PTEN at the plasma membrane. In the current work, using the CRISPR/Cas9 genome editing system, we specifically inhibited the nuclear localization of PTEN without affecting its plasma membrane localization in mice.

PTEN K13R, D384V mice are born with microcephaly
To investigate the physiological function of nuclear PTEN, using the CRISPR/Cas9 genome editing system, we introduced two mutations into the PTEN gene in mice. The first mutation  cro ARTICLE substitutes lysine 13 to arginine. This substitution inhibits the ubiquitination required for the translocation of PTEN to the nucleus (24,25) (Fig. 1A). The second mutation targets the amino acid sequence from 380 to 386 in the C-terminal tail (Fig.  1A). This region consists of negatively charged aspartates and serine and threonine clusters that undergo phosphorylation. These negative charges and phosphorylation are important for closing the conformation of PTEN through intramolecular interaction with the membrane-binding regulatory interface (consisting of the catalytic and C2 domains) (29 -31). The combination of these two substitutions make PTEN more dynamic in terms of intracellular distribution and excludes PTEN from the nucleus (32). We injected a pair of single guide RNAs (sgRNAs) that are directed to these two regions of the PTEN gene into fertilized oocytes from standard B6SJLF1/J mice and transferred injected oocytes into pseudopregnant mice. We obtained 22 mice, and DNA sequenced their PTEN gene. We identified a founder mouse that carries substitutions of both lysine 13 to arginine and aspartate at 384 to valine (Fig. 1B). The founder line was backcrossed to C57BL/6J WT mice multiple times. We crossed heterozygous PTEN K13R, D384V mice to obtain homozygous PTEN K13R, D384V mice (hereafter PTEN K13R, D384V mice) and littermate control mice. PTEN K13R, D384V mice were born at an approximately Mendelian ratio and grew to normal body size (Fig. 1C).
To determine the impact of the K13R and D384V mutations on organ size, we measured the weights of multiple organs. Brain weight was significantly decreased in PTEN K13R, D384V mice of both sex ( Fig. 2A and Fig. S1). In contrast, other organs, including the liver, heart, thymus, kidney, lung, spleen, and prostate/seminal vesicle, showed comparable weights between mutant and control mice ( Fig. 2A and Fig. S1). Consistent with decreased brain weight, the area of the whole brain, as well as the cerebrum and cerebellum, were significantly decreased when measured in the top view (Fig. 2B). To further analyze subregions of the brain in PTEN K13R, D384V mice, we fixed the brains using cardiac perfusion of paraformaldehyde and cut sagittal sections in the midline. Hematoxylin and eosin (H&E) staining showed no gross changes in the histology of the PTEN K13R, D384V mice brain (Fig. 2, C and D). However, we found significant decreases in the area of the PTEN K13R, D384V cerebral cortex and cerebellum in the side-view (Fig. 2, C and E). The size of the PTEN K13R, D384V hippocampus was also significantly decreased in coronal sections (Fig. 2, D and E).
To determine when changes in brain size occur in PTEN K13R, D384V mice, we examined the weight of brains at different ages ranging from day 0 -8 months. We found that the brain in PTEN K13R, D384V mice is already smaller at birth and maintains its small size during adulthood (Fig. 2F). Because the development of the brain initiates in embryos and continues postnatally, these data suggest that PTEN K13R, D384V affects brain development during both embryonic and postnatal periods.
These small brain phenotypes in PTEN K13R, D384V mice are in sharp contrast to increased brain size in brain-specific PTEN knockout mice because of increased neuron size through enhanced PI3K signaling (1, 26 -28, 33-36).

Decreases in neuronal soma size and nuclear PTEN levels in PTEN K13R, D384V mice
To determine the effect of PTEN K13R, D384V on neuron size, we first examined the cerebellum and analyzed one of its major neurons, Purkinje cells (PCs), using confocal immunofluorescence microscopy. The soma of PCs forms a monolayer called the PC layer with highly extended dendritic arbors into the molecular layer. This stereotypical arrangement allowed us to quantitatively measure the density and size of PCs (37,38). Brain sagittal sections were cut in the midline and stained with antibodies to Car8, a cytosolic marker for PCs (37). The density of the PC soma was significantly increased along the PC layer of PTEN K13R, D384V mice, suggesting that the sizes of the soma of PCs decreased (Fig. 3, A and B). Indeed, image analyses of individual PCs showed that the size of their soma is significantly smaller in PTEN K13R, D384V mice (Fig. 3, E and F, Car8). Co-immunostaining with anti-PTEN antibodies also showed a similar decrease in soma size of PTEN K13R, D384V PCs (Fig. 3, E and F, PTEN). A significant decrease in the width of the molecular layer, which contains dendrites of PC, was also observed in PTEN K13R, D384V mice (Fig. 3, A and C). Images with greater magnification of the molecular layer suggest that Purkinje dendritic arbors are shorter in PTEN K13R, D384V mice (Fig. 3D).
Importantly, quantification of immunofluorescence images revealed significant decreases in the signal of PTEN in the nucleus relative to that of PTEN in the cytosol of PTEN K13R, D384V PCs (Fig. 3, G and H, PTEN). As a control, the relative signal intensity of Car8 between the nucleus and cytosol
The intracellular distribution of PTEN and somata sizes in neurons of the motor cortex (layers 3 and 4) and hippocam-pus (CA1 region) were measured using confocal immunofluorescence microscopy with antibodies to PTEN and a mature neuron marker, NeuN (Fig. 4, A and D). Anti-NeuN antibodies label the entire cytoplasm (39). Relative neuronal

Nuclear PTEN-deficient mice
PTEN levels at the nucleus and the cytosol were again found to be significantly decreased in PTEN K13R, D384V mice (Fig. 4, A-C). Also, immunostaining of both PTEN and NeuN showed that these neuronal somata were significantly smaller in PTEN K13R, D384V mice (Fig. 4E). Therefore, the K13R and D384V mutations decrease levels of PTEN in the nucleus and reduce the size of neurons in the cerebellum, cortex, and hippocampus in vivo.
In addition to neurons, glial cells in PTEN K13R, D384V mice were examined via immunofluorescence microscopy of the cerebellum using antibodies to NeuN and Aldh1L1, a glial marker that labels Bergmann glial cells in the cerebellum (40,41). The number and distribution of Aldh1L1-positive cells in controls and PTEN K13R, D384V mice were similar (Fig. 5A). The hippocampus was also analyzed using antibodies to NeuN and GFAP, another glial marker (40,41). Again, no gross changes in the pattern of GFAP-positive glial cells were found in PTEN K13R, D384V mice (Fig. 5B).
As with the neurons, immunostaining of the cerebellum with antibodies to PTEN and Aldh1L1 showed that PTEN levels in the nucleus significantly decreased in Aldh1L1-positive glial cells in PTEN K13R, D384V mice, but the total PTEN levels in these glial cells were lower than those in the PCs in both controls and PTEN K13R, D384V mice (Fig. 5, C and D). Because GFAP-positive glial cells in the hippocampus expressed much lower levels of PTEN (Fig. 5E), a reliable comparison of PTEN levels in the nucleus and cytoplasm was impossible.
Although the livers in PTEN K13R, D384V mice did not appear to change in size ( Fig. 2A and Fig. S1), nuclear PTEN levels were significantly decreased in their hepatocytes (Fig. S2). It appears that the K13R and D384V mutations universally affect PTEN localization in the nucleus, but the physiological consequences of PTEN mislocalization depend on cell type.

PTEN K13R, D384V mice show susceptibility to seizure
Seizures have been observed in patients with diseases linked to PTEN mutations, such as autism spectrum disorder (4 -6, 42). Because no spontaneous seizure was observed in PTEN K13R, D384V mice, we tested susceptibility to seizure using a well-established paradigm in which seizure is induced by intraperitoneal injection of a GABA receptor antagonist, pentylenetetrazol (PTZ) (43,44). Littermate control mice started showing seizures at 2-3 min after injection of a standard amount of PTZ (75 mg/kg body weight) and subsequently recovered from the seizure (Fig. 6A). In contrast, PTEN K13R, D384V mice showed a much shorter latency before starting seizure upon PTZ injection (Fig. 6A). Remarkably, most PTEN K13R, D384V mice (7 of 10) died within 5 min after PTZ injection (Fig. 6B). We did not observe PTZ-induced death of control mice. Therefore, these data suggest that

PTEN K13R, D384V mice display normal PI3K signaling
To determine whether decreased levels of nuclear PTEN affect PI3K signaling, we examined the levels of phosphorylation of the serine/threonine kinase AKT at serine 473 and threonine 308, two major downstream events in PI3K signaling (45). Activated AKT by these phosphorylation events, in turn, phosphorylates and activates a crucial downstream transcriptional factor, forkhead box protein O1 (FOXO1). The activation of AKT also leads to phosphorylation of S6 ribosomal protein through mTORC1. Using Western blotting of the isolated cerebellum, cortex, and hippocampus, we found that the phosphorylation levels of AKT (Ser-473 and Thr-308), FOXO1 (Ser-256), and S6 (Ser-240/244) were not changed in PTEN K13R, D384V mice (Fig. 7, A and B). As controls, the abundance of PTEN was comparable in PTEN K13R, D384V and control mice. Consistent with previous studies showing that mutations in the C terminus of PTEN affect phosphorylation of the serine/threonine cluster in this region (25,32), we observed decreased PTEN phosphorylation in Western blotting with antibodies that recognize PTEN phosphorylation of serine 380, threonines 382 and 383, and serine 385 (Fig. 7, A and B). These data suggest that nuclear PTEN is not involved in the PI3K signaling pathway, different from the function of PTEN at the plasma membrane. These findings support previous findings showing that the K13R mutation does not affect the enzymatic activity of PTEN or its recruitment to the plasma membrane; therefore leaving PI3K signaling to be intact (25,32).

Discussion
Previous studies using a variety of PTEN knockout mice have clearly demonstrated the physiological importance of PTEN (46 -48). Because PTEN functions in different intracellular locations, it remains to be determined the specific in vivo functions of the nuclear and membrane PTEN. In this study, we generated and characterized PTEN K13R, D384V mice in which the amount of nuclear PTEN is significantly decreased. Importantly, this new mouse model showed normal PI3K signaling, and therefore the function of PTEN at the plasma membrane was not disturbed.
Nuclear PTEN deficiency predominantly affected the brain and caused microcephaly with decreased size of neuronal soma in the cerebellum, cerebral cortex, and hippocampus. These phenotypes were surprising to us because previous studies have shown that the loss of PTEN results in hypertrophy and mac-

Nuclear PTEN-deficient mice
rocephaly (1, 26 -28, 33-36). The increased neuron and brain sizes in these PTEN knockout mice resulted from increased PI3K signaling at the plasma membrane in the absence of the PIP3 phosphatase function of PTEN. Along with these previous studies, our findings provide a new model that PTEN at the plasma membrane and the nucleus plays opposite roles in the growth of neurons in the brain. We expect that if one could specifically block PTEN function in PIP3 signaling at the plasma membrane, without affecting nuclear PTEN function in the brain, neuron and brain sizes would further increase compared with the loss of total PTEN.
Because we observed no changes in PI3K signaling in PTEN K13R, D384V mice, nuclear PTEN regulates neurons through a mechanism distinct from the PI3K signaling pathway. Studies have shown that nuclear PTEN functions in DNA repair, genome integrity, and cell cycle progression (7)(8)(9)(10)(11)(15)(16)(17)(18). It appears that inhibition of these activities might not necessarily result in decreased neuron size or microcephaly. Further studies are necessary to better understand the mechanisms underlying the physiological consequence of nuclear PTEN deficiency. Our PTEN K13R, D384V mice would provide a useful tool to address this key question. PTEN K13R, D384V mice were extremely sensitive to the GABA receptor antagonist PTZ. PTZ induced dramatic, lethal seizure in PTEN K13R, D384V mice. Spontaneous seizure has been observed in brain-specific PTEN knockout mice and human patients with PTEN mutations (26,33). Therefore, it appears

Nuclear PTEN-deficient mice
that defects in both nuclear and membrane PTEN contribute to seizure susceptibility. It would be interesting to analyze seizure phenotypes in mice where the recruitment of PTEN to the plasma membrane is selectively inhibited to define specific mechanisms of this PTEN function to seizure.
Although the amounts of PTEN in the nucleus are significantly decreased in PTEN K13R, D384V mice, these mice are not completely devoid of PTEN in the nucleus. We also noticed that the extent of decreases in nuclear PTEN levels varies depending on neuron types. Therefore, it is likely that multiple mechanisms recruit and maintain PTEN in the nucleus in addition to ubiquitination of lysine 13. These mechanisms may be cell type specific. Additional physiological functions of nuclear PTEN would be revealed if one could further decrease the amount of PTEN in the nucleus by simultaneously targeting such multiple mechanisms underlying the nuclear localization of PTEN.
In summary, our findings advance the current understanding of the function of nuclear PTEN in vivo. Although human diseases associated with PTEN deficiency are complex because of the multiple localizations of PTEN in cells, dissecting PTEN functions by blocking each PTEN localization would provide insights into the underlying disease mechanisms. In future studies, our new mouse model will enable us to further define the role of PTEN in the nucleus.

Generation of PTEN K13R, D384V mice using CRISPR/Cas9
All of the work with animals was conducted according to the guidelines established by the Johns Hopkins University Committee on Animal Care and Use. To engineer the PTEN gene in mice, sgRNA-encoding sequences (5Ј-AGATCGTTAGCA-GAAACAAA-3Ј to target Lys-13 and 5Ј-ATTCTCTGGAT-CAGAGTCAG-3Ј to target the C-terminal region) were cloned into the BbsI site of pX330ϩT7 and amplified from pX330ϩT7 with a leading T7 promoter by PCR. These sgRNAs were in vitro transcribed using the HiScribe T7 Quick High Yield RNA Synthesis Kit (New England Biolabs) and purified using the MEGAclear Kit (Ambion). Cas9 mRNA was in vitro transcribed using NotI-linearized pX330ϩT7 and the mMESSAGE mMACHINE T7 Ultra Kit (Ambion), and purified by LiCl precipitation. The homology-directed repair oligos (5Ј-TTCTTC-AGCCACAGGCTCCCAGACATGACAGCCATCATCAAA-GAGATCGTTAGCAGAAACAGAAGGAGATATCAAGA-GGATGGATTCGACTTAGACTTGACCTGTATCCATTT-CTGCGGCTGT-3Ј for Lys-13 and 5Ј-ATCAGACTTTTGTA-ATTTGTGAATGCTGATCTTCATCAAAAGGTTCATTC-TCTGGATCAGCGTCAGCGGCGTCAGCATATCTATAA-TGATCAGGTTCATTGTCACTAACATCTGGAGTCACA-GAAGTTGAACTGCT-3Ј for the C-terminal region) were purchased from Integrated DNA Technologies (4 nM ultramer). Pronuclear injections of zygotes from B6SJLF1/J mice (The Jackson Laboratory, stock no. 100012) were performed at the Johns Hopkins University Transgenic Facility using a mix of Cas9 mRNA, two sgRNA, and two homology-directed repair oligos in injection buffer (10 mM Tris-HCl, 0.1 mM EDTA filtered with 0.2-m pore size). Two combinations of concentrations were used: 100 ng/l Cas9 mRNA, 50 ng/l each sgRNA, 100 ng/l each repair oligo and 25 ng/l Cas9 mRNA, 12.5 ng/l each sgRNA, 25 ng/l each repair oligo. The embryos were cultured at 37°C in the CO 2 incubator for 2 h and then transferred into the oviducts of pseudopregnant ICR females (25 embryos per mouse) (Envigo). Twenty-two pups were obtained, and their genotypes were analyzed by DNA sequencing using the following primers: 5Ј-GCCAAGTCCAGAGC-CATTTC-3Ј and 5Ј-CGATCTAGAAATGCGCCCAG-3Ј for K13R and 5Ј-GAAAAGCAGTGCCCTTCAGA-3Ј and 5Ј-GCAGTGCCCTTCAGAATTCA-3Ј for the C-terminal region. We identified one founder mouse that carried mutations at both

Nuclear PTEN-deficient mice
the Lys-13 and the C-terminal region (K13R, D384V). This line was backcrossed to C57BL/6J WT mice (The Jackson Laboratory, stock no. 000664) multiple times before use.

Histology and immunofluorescence microscopy
Mice were anesthetized by intraperitoneal injection of Avertin (200 mg/kg) and fixed by cardiac perfusion of ice-cold 4% paraformaldehyde in PBS (40 -50 ml), as previously described (37,38). The brains were dissected and further fixed in 4% paraformaldehyde in PBS for 2 h at 4°C. The samples were further incubated in PBS containing 30% sucrose overnight and frozen in O.C.T. compound (Fisher Scientific, 23-730-571) in a Tissue-Tek Cryomold. Frozen tissue blocks were sectioned using a cryostat (Leica, CM 3050S, Nussloch, Germany) and mounted on Superfrost Plus Microscope Slides (Fisher Scientific, 12-550-15).
For histology, frozen sections were H&E stained at Johns Hopkins School of Medicine Pathology Core. The samples were observed using a microscope (Olympus, model BX51) equipped with a DP-70 color camera and 10ϫ (0.3 numerical aperture) UIS2 objectives.

Statistical analysis
The Student's t test was used to compare PTEN K13R, D384V and littermate control mice in all of the experiments.

Western blotting
The cerebellum, cortex, and hippocampus were harvested from mice, flash frozen in liquid nitrogen, and homogenized in RIPA buffer (Cell Signaling Technology) containing phosphatase inhibitor mixtures 2 and 3 (Sigma, P5726 and P0044). Lysates were centrifuged at 14,000 ϫ g for 10 min, and the supernatants were collected. Protein concentrations were determined using a Bio-Rad protein assay. Proteins were separated by SDS-PAGE and transferred onto Immobilon-FL (EMD Millipore, Billerica, MA). After blocking in 3% BSA/PBS/ Tween 20 for 1 h at room temperature, the blots were incubated with primary antibodies: Cell Signaling Technology; 9461), phospho-S6 (Ser-240/244; Cell Signaling Technology; 5364), and GAPDH (Thermo Fisher; MA5-15738). Immunocomplexes were visualized using a PharosFX Plus Molecular Imager (Bio-Rad) with fluorescently labeled secondary antibodies. Band intensity was determined using NIH ImageJ software (49).