All animal experiments performed have been approved by and are in agreement with the guidelines of the Stony Brook University (SBU) Institutional Animal Care and Use Committee (IACUC). HMGB2 null breeding pairs on a C57BL/6 background were obtained from Dr. Lorenza Ronfani (Core Facility for Conditional Mutagenesis, San Raffaele Scientific Institute, Milan, Italy). The HMGB2 deficient mouse line was created by conventional gene targeting methods (Kokovay et al., 2010). Due to the sterility of the HMGB2 homozygous null males (Kokovay et al., 2010), two breeding strategies were employed for the following experiments: heterozygous females were crossed with heterozygous males or homozygous females crossed with heterozygous males. In instances where Mendelian ratios did not yield adequate numbers of HMGB2^+/+ controls (e.g., the second breeding scheme), age-matched controls were substituted from non-transgenic C57BL/6 breeding. The HMGB2^-/- animals are smaller in size, but do not initially appear grossly phenotypically different than wild-type animals.

Pups ranging in age from postnatal day 0 to postnatal day 2 were used for all cell culture experiments. Brains were extracted and placed in sterile Hank’s buffered salt solution (HBSS) containing 2% glucose. From each hemisphere—olfactory bulbs, hippocampi, midbrain, cerebellum, and meninges were dissected away and discarded. The remaining brain regions, VZ, SVZ, striatum, and minimal underlying cortex, were placed in 5 mL of HBSS supplemented with 2% glucose. The tissue was triturated in 1 mL of complete Neurocult neural progenitor cells (NPC) Basal Proliferation Media with the addition of 20 ng/mL of recombinant human epidermal growth factor (rh EGF), 10 ng/mL recombinant human basic fibroblast growth factor (rh bFGF), 0.2% heparin, and 5% penicillin/streptomycin. The tissue suspension was spun down at 1500 rpm at room temperature. The subsequent tissue pellet was resuspended in 200 μl of complete Neurocult NPC Basal Proliferation Media with the addition of 20 ng/mL of rh EGF, 10 ng/mL rh bFGF, 0.2% heparin, and 5% penicillin/streptomycin. The tissue was further triturated with a pipette tip until a uniform cell suspension was achieved. The cell suspension was filtered through a 40 μm nylon mesh cell strainer and plated on six-well plates precoated with 100 μg/mL poly-D-lysine (PDL, 1 h), and 10 μg/mL laminin (4 h). Following approximately 2–3 days of proliferative growth, the cells were replated in precoated 60 mm tissue culture plates in complete Neurocult differentiation media containing supplements and 5% penicillin/streptomycin.

Histones were obtained from cell culture lysates at various time points primarily employing the method described in Kriegstein and Alvarez-Buylla (2009). Briefly, cells were centrifuged at room temperature at 300 g and the pellets cleaned with 10 mL of sterile HBSS. The pellets were resuspended in 1 mL of hypotonic lysis buffer (10 mM Tris–Cl pH 8.0, 1 mM KCl, 1.5 mM MgCl2, 1 mM dithiothreitol (DTT), protease and phosphatase inhibitors; Kriegstein and Alvarez-Buylla, 2009) and rotated at 4°C for 30 min. Following this step, a nuclear pellet was obtained by spinning the hypotonically dissociated cells at 4°C, 10,000 g for 10 min. Following this step the NSPC nuclear pellet was resuspended in 400 μl of H₂SO₄ and rotated overnight at 4°C. Nuclear debris was removed by spinning the acid extracted preparation at 16,000 g for 10 min. A total of 132 μl of trichloroacetic acid was added to the supernatant to a 33% final concentration to precipitate the histones. A final spin step was used to pellet the purified histones and two subsequent washes with ice-cold acetone cleared away any remaining acid. The histones were dried for 20 min at room temperature and resuspended in 100 μl of ddH₂O. Concentrations of purified histone proteins were obtained via ImageJ densitometry measurements following 15% SDS-PAGE and subsequent coomassie staining. Immunoblots utilized the aforementioned percentage gels followed by transfer to polyvinylidene difluoride membranes. The following antibodies were used to probe for specific histone modifications: (1) Tri-Methyl-Histone H3 (Lys27), Rabbit monoclonal antibody (mAb) from Cell Signaling (#9733) at a concentration of 1:1000, (2) Di/Tri-Methyl-Histone H3 (Lys9), Mouse IgG1 mAb from Cell Signaling (#5327) at a concentration of 1:2000, (3) Tri-Methyl-Histone H3 (Lys4), Rabbit mAb from Cell Signaling (#9751) at a concentration of 1:1000, (4) Ubiquityl-Histone H2A (Lys119) Rabbit mAb from Cell Signaling (#8240) at a concentration of 1:2000, (5) Anti-Histone H3 Rabbit polyclonal from Abcam (ab1791) at a concentration of 1:2500, (6) Anti-Histone H2A Rabbit polyclonal from GeneTex (GTX129418).

The following Qiagen RT² Profiler 84-gene arrays were utilized: Mouse Polycomb & Trithorax Complexes (PAMM-506Z). The arrays were loaded with cDNA reverse transcribed from RNA which was obtained from 24-h differentiation cultures of HMGB2^+/+ and HMGB2^-/- NSPCs. An Applied Biosystems StepOnePlus Quantitative Real-Time PCR (qRT-PCR) machine served as the PCR platform. Data were analyzed initially on the Qiagen web-based analysis software followed by Microsoft Excel (Supplementary Figure 1).

Total RNA was obtained from differentiating HMGB2^+/+ and HMGB2^-/- NSPCs at days 1, 3, and 5. RNAbee (Amsbio) was used for RNA extraction followed by purification via the Qiagen RNeasy Mini Kit (74104). Subsequently cDNA was reverse transcribed using the High-Capacity cDNA Reverse Transcription Kit (Life Technologies—4368814). GAPDH was used as housekeeping gene. For gene expression analysis, the relative quantitation method was used (ΔΔCt) with GAPDH as an internal control. All primer sequences are listed in Supplementary Table 1.

HMGB2^+/+ and HMGB2^-/- NSPCs were differentiated for 3 or 7 days in eight-well chamber slides (Fisher-Mediatech 177445) precoated with 100 μg/mL PDL (1 h), and 10 μg/mL laminin (4 h) in a tissue culture incubator at 37°C. HMGB2^+/+ and HMGB2^-/- NSPCs were differentiated for 3 or 7 days in eight-well chamber slides (Fisher-Mediatech 177445) precoated with 100 μg/mL PDL (1 h), and 10 μg/mL laminin (4 h) in a tissue culture incubator at 37°C. Individual chambers were blocked for 1 h with phosphate buffered saline (PBS) containing 5% normal goat serum, 1% bovine serum albumin (BSA), and 0.3% Triton-X100. Antibody solutions were made up of PBS containing 1% BSA, and 0.03% Triton-X100. Immunocytochemistry/immunofluorescence antibodies are listed in Supplementary Table 2.

P1 wild-type (WT) and HMGB2 knock-out (KO) brains were collected and passively fixed in 4% paraformaldehyde (PFA), whereas p21 WT and KO brains were perfused with 4% PFA and postfixed with 4% PFA for 12 h. After fixation, brains were put in 30% sucrose until they were no longer floating, and then frozen in OCT via dry ice. Cryostat sectioning at 20 μm thickness was done, and slides were stored in -80°C.

For staining, slides were allowed to get to room temperature, and then washed in PBS. Blocking was done for 2 h at room temperature with 5% serum and 0.3% Triton X-100. Primary antibodies were diluted with 0.01% BSA and 0.3% Triton X-100, and incubated at 4°C overnight or 2 h at room temperature. Slides were washed in PBS, and then incubated with Alexa fluorophore secondary antibodies at 1:1000 dilution for 1 h at room temperature. After another PBS wash, slides were cover slipped with 4′,6-diamidino-2-phenylindole (DAPI) Fluoromount.

Imaging was done on a SP8X Leica Confocal Microscope. Images shown are 20 μm z-stacks.

For comparisons between groups of two an unpaired, two-tailed t-test was employed. For multiple comparisons within a group a one-way ANOVA was used followed by either Holm–Sidak multiple comparison post hoc analysis, or Bonferroni’s post hoc multiple comparisons test. GraphPad Prism and Microsoft Excel were used for all statistical analyses. A 95% confidence interval is used for all figures indicating initial significance at p < 0.05 marked by single asterisk “^∗” and greater significance at p < 0.01 marked by double asterisks “^∗∗”. Standard error of the mean (SEM) is the calculation underlying all graph error bars, n refers to the number of biological replicates used in each experiment.

Monolayer NSPC cultures from HMGB2^+/+ and HMGB2^-/- p0–p2 pups (isolated from primary dissections of SVZ) were grown in media promoting cellular proliferation (bFGF and rh EGF) until the plates were ∼85% confluent. The cells were replated in 60 mm dishes containing differentiation media. Following 24 h of differentiation the cells were harvested and total RNA was extracted for 84-gene analysis via the Qiagen qRT-PCR array system—specifically looking at genes encoding key PcG and TrxG subunits (Supplementary Figure 1). In the PcG and TrxG arrays key PRC1 and PRC2 subunits (EED, SUZ12, CBX3, BMI1) were revealed to be downregulated in the HMGB2^-/- NSPCs following 1 day of differentiation (Supplementary Figure 1). Although many other complex macromolecular assemblies have built-in redundancies making the elimination of one component tolerable, these key PcG genes encode critical pieces of PRC1 and PRC2 complexes (Gonzales-Roybal and Lim, 2013). To further refine the analysis following the initially shallow Qiagen screen, we sought to verify which of these components was downregulated at days 1 and 3 of NSPC differentiation using qRT-PCR. Primary NSPCs were plated; after 2 days of proliferative growth they were exposed to differentiation conditions through the removal of growth factors from the media. RNA was extracted at 1 and 3 days following the induction of differentiation (Figure 1). Our findings indicate that EED, an important member of PRC1 and PRC2 is significantly downregulated by 50% in HMGB2^-/- NPCs at day 1 of differentiation (Figure 1A), and by 20% at day 3 of differentiation (Figure 1B). The expression of the remainder of the genes examined, SUZ12, BMI1, and CBX3, was not significantly modified (indicating that while the array produced important leads, it would only be an initial step in the analysis). Lower EED expression was evident in p1 HMGB2^-/- SVZ tissue compared to the expression in HMGB2^+/+ animals, although the difference was not statistically significant, possibly reflecting EED protein stability (Supplementary Figure 2).

Collectively these results point to a critical role for the chromatin protein HMGB2 in the dynamic regulation of EED, a key polycomb complex component, and potentially the expression of cell fate genes regulated by PcG.

EED is a core component of the PRC2 complex catalyzing the methylation of histone H3 at lysine 27 (K27; Lamiable et al., 2010). The presence of tri-methylation on H3K27 can function as signal to recruit PRC1 complexes to H3K27me3 marks, and thus initiate the process of gene silencing and chromatin condensation. EED also acts to stabilize H2A ubiquitin E3 ligase activity (Gruber et al., 2011). It is therefore responsible for the coordination of PRC1 and PRC2 activities, which are critical to the repression of specific neuronal and astroglial genes (Gruber et al., 2011; Lim and Alvarez-Buylla, 2014) and the onset of stem cell differentiation (Lamiable et al., 2010). To examine whether global levels of H3K27me3 were altered in HMGB2^-/- NSPCs we utilized the monolayer NSPC culture system from which we acid-extracted histone proteins at days 1 and 3 of differentiation (Figures 2A,B). H3K27me3 levels were visualized using immunoblots; they remained the same 1 day following the initiation of differentiation in the HMGB2^+/+ and HMGB2^-/- NSPCs, however, they were significantly reduced at day 3 of differentiation in the HMGB2^-/- cultures (Figures 2C,D).

We also evaluated another repressive histone modification, H3K9me3, which is laid down by a different epigenetic complex that does not include EED (Margueron and Reinberg, 2011). No changes were observed between HMGB2^+/+ and HMGB2^-/- NPCs at differentiation (Supplementary Figure 3), suggesting HMGB2 may be specifically involved in the regulation of EED function. The mechanism regulating this putative negative feedback loop remains to be elucidated.

To assess the differentiation properties of perinatal NSPCs monolayer cultures were induced to differentiate in eight-well chamber slides for 3 days and 7 days, followed by immunostaining with cell-type specific antibodies. At day 3 of differentiation significantly more GFAP⁺ astrocytes were present in the HMGB2^-/- wells compared to WT, while neuronal populations, characterized by Tuj1⁺ staining, remained at near undetectable levels (Figure 3). Oligodendrocyte differentiation was not assessed at this timepoint, as the differentiation process to mature oligodendrocytes is not complete, and not all NG2⁺ cells become oligodendrocytes. Interestingly at day 7 of differentiation the percentage of Tuj1⁺ cells increased in the HMGB2^-/- wells (Figure 4). At this time point there was no difference in the numbers of GFAP⁺ astrocytes between the two genotypes; immunostaining for CC1 revealed a decrease in oligodendrocytes in the HMGB2^-/- wells. These data point to a possible role of HMGB2 loss on the neurogenic to gliogenic fate transition that occurs in perinatal NSPCs (Figure 4; Corley and Kroll, 2015). In addition, our previous experimental findings in the adult HMGB2^-/- mouse brain corroborate these perturbed perinatal cell fate transitions with a greater proportion of adult newborn olfactory bulb interneurons present in the adult HMGB2^-/- animal (Hirabayashi et al., 2009). This also is supported by the observation that there are apparent structural differences in CNS anatomy in the HMGB2^-/- mice. Fifty percent of the animals (n = 14) displayed enlarged ventricles at 10 weeks of age, while only 1 of 10 of HMGB2^+/+ mice exhibited any detectable ventricular enlargement at that age (Supplementary Figure 4). Therefore, the effects of HMGB2 depletion may be additive throughout development and into the adult mouse brain.

To further assess the differentiation properties of neural stem cells (NSCs) lacking HMGB2, p1 and p21 HMGB2^+/+ (WT) and HMGB2^-/- (KO) brains were sectioned and imaged (Figures 5A–C). As mentioned earlier, neurogenesis switches to give rise to glial cells (Corley and Kroll, 2015). Glial (PSA-NCAM⁺/NG2⁺) and neuronal (PSA-NCAM⁺/NG2^-) progenitors were counted along the SVZ, with p1 HMGB2 KO mice having increased numbers of progenitors (Figure 5D). There was no change in the immature glial cell population (NG2⁺/PSA-NCAM^-). Additionally, astrocytes were further examined. Immature (Aldh1l1⁺) and mature (GFAP⁺) astrocytes were evaluated in both the striatum and cortex (Figures 5B,C). A trend toward an increase of immature astrocytes in p1 KO was evident compared to WT, but by p21 there was no change between WT and KO mice (Figure 5E). There was no difference in the mature astrocyte population at p1, as GFAP expression is low and mature astrocytes are just now starting to form by p1 (Figure 5F; Mohn et al., 2008; Ming and Song, 2011). However, by p21, there was a decrease in the numbers of mature astrocytes in the medial striatum of KO mice compared to WT mice (Figure 5F). These data further support a role of HMGB2 in regulating the neurogenic/gliogenic fate, and support a prolonged neurogenic phase.