Tsc2 coordinates neuroprogenitor differentiation

Summary Neural stem cells (NSCs) of the ventricular-subventricular zone (V-SVZ) generate numerous cell types. The uncoupling of mRNA transcript availability and translation occurs during the progression from stem to differentiated states. The mTORC1 kinase pathway acutely controls proteins that regulate mRNA translation. Inhibiting mTORC1 during differentiation is hypothesized to be critical for brain development since somatic mutations of mTORC1 regulators perturb brain architecture. Inactivating mutations of TSC1 or TSC2 genes cause tuberous sclerosis complex (TSC). TSC patients have growths near the striatum and ventricles. Here, it is demonstrated that V-SVZ NSC Tsc2 inactivation causes striatal hamartomas. Tsc2 removal altered translation factors, translatomes, and translational efficiency. Single nuclei RNA sequencing following in vivo loss of Tsc2 revealed changes in NSC activation states. The inability to decouple mRNA transcript availability and translation delayed differentiation leading to the retention of immature phenotypes in hamartomas. Taken together, Tsc2 is required for translational repression and differentiation.


INTRODUCTION
4][5][6][7] Extracellular factors influence transcriptional programs and cellular identity, however, the role of post-transcriptional processes regulating which mRNAs are translated into functional proteins is less clear. 8][11] Tuberous sclerosis complex (TSC) is a genetic disorder affecting approximately 1 million patients worldwide. 12TSC is caused by inactivating mutations in TSC1 or TSC2. 13,146][17][18] TSC patients have numerous types of growths called hamartomas and malformations including in the brain. 194][25][26][27][28][29] As one would expect, surgical resection of epileptogenic tubers reduces seizure burden. 23,25,30Tubers can be detected during late gestation and are therefore thought to be a byproduct of abnormal development.Tubers are comprised of mislaminated cortical regions having cytomegalic excitatory neurons with hyperactive mTOR. 31Radial glia are embryonic NSCs that generate excitatory cortical neurons. 32,33Loss of TSC genes in radial glia is the most parsimonious explanation for how tubers form.5][36][37][38][39][40][41] In contrast, loss of Tsc1 from inhibitory neurons does not cause seizures. 42,43he cause of TANDs are less clear.A possible culprit is the additional hamartomatous growths and smaller microlesions found within the brain of TSC patients. 44Subependymal nodules (SENs) are hamartomas smaller than 1 cm situated between the striatum and ventricle near the V-SVZ. 45SENs occur in the vast majority of TSC patients. 46Histologically, they contain abundant DCX and Tuj1 positive immature neuroblasts dispersed within a glial fibrillary matrix along with sporadic NeuN, Iba1, Ki67, or Sox2/Pax6 positive cells. 47SENs can appear as early as $mid-gestation within the subependymal region and contain dysmorphic astrocyte-like cells that express vimentin, glutamine synthetase, GFAP, and filamin A. 48 Embryonic SEN-like lesions and postnatal SENs also contain occasional giant cells, a pathognomonic feature of TSC. 47,48Given the early onset and anatomical location of SENs, the cellular composition is not surprising.0][51] In nearly 10-30% of TSC cases, patients have subependymal giant cell astrocytomas (SEGAs) which may originate from continued SEN growth. 46SEGAs are often located near the caudate nucleus of the striatum and often arise in situ near the parenchyma V-SVZ border, and without disruption of the ventricular, ependymal lining. 48,52SEGAs also contain giant cells that express markers found in dysmorphic astrocytes but have a conspicuous and eccentrically localized nuclei with prominent nucleoli, are eosinophilic and can be multi-nucleated. 53Giant balloon-like cells having glial-like features are also present in late fetal SEGA-like growths, followed by the appearance of small fusiform cells including migrating neuroblasts and inhibitory interneurons. 48Eventually, SEGAs will contain sporadic inhibitory interneurons and potentially BIII tubulin positive neuroblasts. 47,48,52,54Such neuroblasts and neurons could theoretically be generated by the cell or origin of SEGAs.Thus, a hypothesis is that mutations arising in V-SVZ NSCs might cause SEGAs.Nestin, GLAST, and SOX2 which NSCs express, also are found in SEGAs suggesting that they may contain and/or originate from NSCs. 47,557][58][59] However, given the disease state, these transcripts could reflect aberrant transcription in non-NSC cells.
How loss of TSC genes causes brain hamartomas and effects NSC differentiation remains uncertain.In support of the hypothesis that loss of TSC genes in cells along the V-SVZ/striatal border causes subcortical hamartomas including SENs and SEGAs, nodules can be generated in mice by removing conditional Tsc1 alleles from postnatal V-SVZ NSCs or progenitors. 60,61Tsc1 knockout neuroprogenitors generate neuroblasts and neurons that are ectopically positioned. 60,61NSC proliferation and differentiation appear to be unaffected by loss of Tsc1 in vitro. 60n combination with loss of Tsc1, removal of PTEN from neonatal V-SVZ NSCs caused SEGA-like hamartomas which retain stem-like properties. 47Incidentally, loss of TSC genes in human organoid models appears to prevent neuronal differentiation. 62Yet inducible Tsc1 knockout in human cortical organoids prevents neuronal but enhances glial production. 63Thus, some TSC anatomical alterations might reflect an important role for TSC genes in fate decisions including gliogenesis.5][66] However, TSC2 mutations are responsible for the majority of SEGAs and studies have only recently started to examine the role of Tsc2 in V-SVZ NSCs and their capacity to generate different cell types. 67,68he following study utilizes a combination of in vivo genetic manipulation, whole transcriptome RNA-sequencing (RNA-seq), polysome profiling, and microscopy to determine what role, if any, Tsc2 has in V-SVZ NSCs and differentiation.Tsc2 mutant NSCs were confirmed to have increased mTOR pathway activity, altered translation factor mRNA, and stochastic mRNA translation.Changes in the amplitude and efficiency of translation were associated with the inability to down-regulate stem features in mutant Tsc2 cells.We propose that TSC gene mutations prevent the loss of stemness which may have implications in developing novel therapeutic approaches to treat patients.

Single cell V-SVZ NSC Tsc2 knockout generates hamartomas
Electroporation of Cre recombinase (CRE) and green fluorescent protein (GFP) plasmids into neonatal mice having wild-type (wt) or conditional (f) Tsc2 alleles was performed to mimic somatic Tsc2 biallelic inactivation (mut) and to generate a model of TSC hamartomas and brains analyzed at P30 and P60 (Figures 1A-1C).CRE induced RFP expression and Tsc2 allele recombination (Figure 1D).Tsc2 wt/wt NSCs generated cells having a glial morphology that was confirmed by glutamine synthetase (GS) staining (Figure 1E).Tsc2 wt/wt glia had low mTOR pathway activity as indicated by eukaryotic initiation factor 4E binding protein (4E-BP) phosphorylation (Figures 1F and 1G).In contrast, Tsc2 mut/mut SVZ p4E-BP was elevated in comparison to controls (Figures 1H-1K).Tsc2 mut/mut NSCs also generated p4E-BP positive nodular hamartomas near the lateral V-SVZ and striatum (Figures 1I-1K; Figures S1A-S1S).Hamartomas had elevated p4E-BP in comparison to Tsc2 wt/wt striatal glia (Figures 1J-1L).Hamartomas could be categorized based on circularity which was measured by tracing the perimeter of lesions (Figure 1K).DNA counterstaining confirmed disorganized regions within the striatum (Figures S2A-S2R).105 hamartomas were quantified having an average size of 4,834 mm 2 (SEM = 314.508)and could be categorized as linear or round with round lesions being smaller (Figures 1M and  1N).Hamartomas frequently contained cytomegalic Neu-N positive neurons (Figures 1O and 1P).Neurons had hypertrophic dendrites with >7 primary dendrites originating from a cytomegalic soma with a round nucleus (Figures 1Q-1T).Tsc2 wt/wt striatum, in contrast, had fewer neurons (Figure S3A).Tsc2 wt/wt and Tsc2 mut/mut neurons were also found within the nucleus accumbens (Figures S3B and S3C) and olfactory bulb (OB) as expected. 69,70Tsc2 mut/mut striatal neurons were morphologically distinct from OB granule cells (GCs) and were larger than control striatal neurons (Figures S3D and S3E). 70][73][74] Tsc2 mut/mut striatal neurons had more dendrites than Tsc2 wt/wt neurons (Figures 1S and 1T).The presence of more Tsc2 mut/mut neurons in the lateral striatum was unlikely caused by defective migration since the ratio of mutant neurons in each region did not change (Figures S3B and S3D).Taken together, NSC Tsc2 deletion produced striatal hamartomas with heterotopic neurons.

Loss of Tsc2 in V-SVZ NSC alters transcripts encoding translational regulatory factors
V-SVZ Tsc2 wt/wt and Tsc2 mut/mut cells were subsequently cultured from Tsc2 wt/wt and Tsc2 mut/mut x Nestin-CRE-ER T2 neonatal mice and RNA sequencing performed to identify mechanisms that may be responsible for hamartoma and heterotopic neuron generation.Tsc2 mut/mut cultures contained cytomegalic cells with multinucleated giant cells (Figures S4A and S5).Multinucleated cells were also confirmed in the cerebral cortex and in dorsal astrocyte cultures (Figures S4B and S4C).Subculturing cells under NSC self-renewing conditions resulted in a homogeneous population of RFP positive NSCs (Figure S3A).Tsc2 mut/mut NSCs could self-renew slightly faster as monolayer cultures and could be grown as neurospheres in the absence of a substrate (Figures S6A-S6D).NSC subcultures were confirmed to have lost Tsc2 (Figures S4D and S4E).Whole transcriptome bulk RNA sequencing of sub-cultured NSCs maintained under self-renewing conditions confirmed expression of canonical NSC mRNAs including Vimentin, Olig1/2, TNC, Nestin, and a battery of Sox genes.Several transcripts present in the proposed originating cells of TSC SEGAs including Ttyh1, Ptgds, GFAP, GAD1, EGFR, and Ednrb were also present. 52Hierarchical clustering categorized NSCs into two groups, Tsc2 wt/wt and Tsc2 mut/mut NSCs (Figure 2A).RNA-seq identified 2,268 differentially expressed mRNAs based upon a P adjusted value < 0.00001, and more than log (base 2) two-fold enrichment.Most transcripts were downregulated, and 571 transcripts were increased (Figure 2B).Bioinformatic analysis using GeneMANIA and the most abundant differentially expressed transcripts pointed to a network of altered translation factors (Figure 2C).Indeed, Cytoscape analysis indicated that protein translation pathways including initiation and elongation as well as ribosome encoding transcripts were altered (Figure 2D).Gene ontology (GO) analysis revealed enrichment of pathways related to ribosome biogenesis and metabolism (Figure 2E).However, ribosome proteins (rps3, rpl26, rpS6) were often altered independently of mRNA transcript levels as determined by western blot analysis of precipitated protein pellets from the same RNA isolations, indicating that post-transcriptional mechanisms may be altered (Figure S4D).Nevertheless, comparison of the differentially expressed transcripts revealed overlap with transcriptomic profiles from TSC SEGAs (Figures S7A-S7D). 57,75,76Taken together, Tsc2 mutant NSCs had altered transcript levels of factors that regulate translation.
Translational programs that encode for ribosome biogenesis, metabolism, and development are altered in Tsc2 knockout V-SVZ NSCs To determine whether mRNA translation was altered, actively translating NSC mRNAs were subsequently identified by polyribosomal profiling.628 transcripts were differentially recovered with polyribosomes, 316 of which were elevated following loss of Tsc2 (Figure 3A, p-adjusted value < 0.0001).CTGF, which is increased in TSC patients and TSC mouse models, was among the top up-regulated genes. 77nflammatory processes were also significantly differentially affected.This is consistent with the inflammatory and reactive states often detected within TSC malformations.The extent that mRNA within polyribosomes reflected total RNA abundance was unclear.Transcriptomes of translatome-paired samples were also quantified.The ratio of polyribosomal to total mRNA was then determined as a measure of translational efficiency (Figure 3B).Translatome (polyribosome) levels positively correlated with transcript availability (Figure 3C).Transcripts that are differentially translated in Tsc2 mut/mut NSCs were more tightly correlated with their transcript availability (Figure 3C).However, not all transcripts whose mRNA availability changed were differentially translated.Nevertheless, the translational efficiency of several transcripts was altered in Tsc2 mut/mut NSCs (Figure 3D).The most abundant transcripts with differential translational efficiency included those that regulate translation (Figure 3E).In addition, transcripts with the greatest increase in translational efficiency included those that regulate developmental processes including telencephalon, forebrain, brain, head, and central nervous system development as well as differentiation and morphogenesis (Figure 3F).Taken together, these results confirm aberrant translation regulation in TSC NSCs and indicate that NSC developmental events may also be impacted during hamartoma formation.

Single cell Tsc2 knockout followed by single nuclei RNA sequencing reveals changes in translation regulatory and metabolic transcripts and NSC transition states
Single nuclei RNA sequencing (snRNA-seq) of V-SVZs was performed to determine how altered Tsc2 mut/mut NSC transcriptomes and translatomes create hamartomas in vivo.Twenty-three cell clusters were identified (Figure 4A).Clusters 15, 8, 4, and 2 were mapped in graphical proximity based on cluster analysis thereby indicating similar identities (Figure 4A).Minor fluctuations in the percentage of clusters were noted, but Tsc2 mut/mut cells had similar cluster compositions to those of Tsc2 wt/wt conditions (Figure 4B).Cell clusters included NSCs denoted by shared markers ATP1a2 and Clu as well as Gpr37L1, Atp1b2, Mfge8, Gja1, S1pr1, and Plpp3 which were enriched in clusters 15, 8, and 4 (Figure 4C).Markers denoted a continuum of NSC activation states as previously reported with cluster 2 expressing AldoC, Sparcl1, Sparc, and Ckb which are characteristic of active NSCs (Figure 4C). 7Thus, as one might expect, some transcripts overlapped (Figure 4C).However, others were specific to activation state (Figure 4C).Additional clusters including astrocyte-like cells (Cluster 1-2) defined by higher Slc1a2, oligodendrocytes (Cluster 11 expressed Plp1 and Mbp), and numerous neural lineages (7, 10, 13, 18-20) were identified (Table S1).Two lineages expressed Ptgds, a marker of recently described caudal late interneuron progenitor cells (Cluster 11, oligo and Cluster 5 unknown) (Table S1).Ependyma (Cluster 16, denoted by Syne1) were additionally identified (Table S1).Tsc2 mut/mut and Tsc2 wt/wt transcripts were subsequently compared between the same clusters, for example we compared cluster 2 between the two genotypes (Figure 4D).Transcripts that represented cluster 2 NSC activation states were reduced in Tsc2 mut/mut NSCs.For example, cluster 2 transcripts, Ckb and AldoC, were significantly downregulated in Tsc2 mut/mut NSCs compared to Tsc2 wt/wt NSCs (Figure 4D).AldoC and Ckb were in fact reduced in all Tsc2 mut/mut NSC states reflecting a change in dynamics of NSC transitional states.In agreement, AldoC was one of the first identified down-regulated transcripts in SEGAs. 56Moreover, Sparcl1 down-regulation detected in snRNA-seq data were also confirmed in RNA-seq data as one of the most significantly down-regulated transcripts in cultured Tsc2 mut/mut NSCs.Tsc2 mut/mut NSCs had elevated Rpl39, Rplp1, and Rps21 expression whereas primed/activated Tsc2 mut/mut NSC clusters had reduced ribosomal transcripts.Dysregulation of mRNA transcripts also became more evident as cells advanced toward a more differentiated trajectory (Table S2).Incidentally, approximately 1% of cells were identified as choroid plexus epithelial cells.However, a slightly larger percentage of cells in Tsc2 mut/mut conditions clustered in this group.Tsc2 mut/ mut clusters up-regulated choroid plexus markers including Ttr which was also elevated in TSC transcriptomes (Figure 3A).Surprisingly, the Tsc2 mut/mut cluster 9 and choroid plexus cluster 14 had higher levels of Ptgds.Ptgds was previously identified as a marker of caudal late inhibitory progenitors implicated in the formation of TSC SEGAs.Ptgds is highly enriched in the choroid plexus as confirmed in the Allen Brain Atlas and MIT Single Cell Portal (data not shown).Nevertheless, snRNA-seq demonstrated that loss of Tsc2 alters transcripts that identify unique NSC activation states.
The mTOR pathway and translation are tightly controlled during neurogenesis.Active NSCs have high mTOR activity and newly generated neuroblasts have lower mTOR activity. 78We also found that NSCs have higher mTOR pathway activity in comparison to striatal glia.In contrast, mTOR pathway activity was elevated in striatal hamartomas.Thus, controlling mTOR regulation of translation programs might be critical for the maintenance of stemness.One would therefore hypothesize that downregulation of mTOR and translation in glia might help uncouple stemness transcript availability from translation of stem factors.However, when Tsc2 is mutated, stem factors may not be properly turned off.This could lead to ectopic generation of neurons as reported above.In support of the hypothesis that Tsc2 is required for  regulating neuroprogenitor-glia differentiation, hamartomas also contained newborn neuroblasts as indicated by the presence of Dcx (Figures 4E; Figures S8A and S8B).Moreover, hamartomas expressed and retained elevated Sox2, a marker of neuroprogenitors (Figures 4F; Figures S8C and S8D).Approximately half of all cells expressed glutamine synthetase confirming a glial component of hamartomas (Figures 4G; Figures S9A and S9B).Finally, hamartomas contained cells that expressed Nestin, which identifies NSCs (Figures 4H; Figures S9C and S9D).NeuN positive neurons were also enriched in striatal hamartomas (Figure 4I, Figures S9C, S9E, and S9F).Tsc2 mut/mut hamartomas were confirmed to slowly grow from P30 to P60 but were completely absent in Tsc2 w/w or Tsc2 wt/mut conditions (Figure 4J).In conclusion, hamartomas were characterized by elevated mTOR pathway activity, stem-like cells, and aberrant neurogenesis within the striatum (Figure 4K).Taken together, Tsc2 loss of function alters translation through numerous mechanisms and alters cellular differentiation.

DISCUSSION
We have shown that Tsc2 is required for neuroprogenitor differentiation.9][80][81] One potential reason for tightly controlling mTORC1 activity is the need to decouple mRNA availability and translation during the transition from stem states.Here it is demonstrated that lateral V-SVZ neuroprogenitors generate glia with repressed mTORC1 activity.The reduction of mTORC1 pathway activity in cortical astrocytes has been previously documented and the study here extends this finding to striatal astrocytes. 39We found that loss of Tsc2 prevented the downregulation of mTORC1 activity in a subset of cells within the striatum.These cells were located within nodular growths.The heterogeneous cellular morphologies and hence composition led us to propose that these growths represent hamartomas that occur in the neurodevelopmental disorder TSC.TSC hamartomas include subependymal giant cell astrocytomas (SEGAs) that form along the striatum and V-SVZ.Occasional hamartomas appeared to protrude into or invade the ventricular wall.Hamartomas were characterized by abnormal heterotopic clusters of morphologically heterogeneous cells which included neurons.Hamartomas had characteristically elevated mTORC1 activity.Therefore, this model allowed us to examine the molecular and cellular underpinnings of brain hamartoma formation.
The presented study uncovered a transcriptional signature of dysregulated translational components in Tsc2 null NSCs.Transcriptional changes and the cellular morphology of NSCs was similar to a seminal finding that proposed mice could be used to model human TSC giant cells. 82That model demonstrated that homozygous but not heterozygous Tsc2 mutant NSCs from mice have robust transcriptional changes and defects in neural differentiation. 82Subsequent studies used conditional Tsc1 deletion in embryonic NSCs (GFAP, Emx1, and Nestin promoter driven CRE) which caused cytomegaly, macrocephaly, hypomyelination, and gliosis with seizures. 35,37,38,83hGFAP-CRE Tsc2 deletion produced subtle effects on dorsal NSCs but was reported to increase GFAP in the V-SVZ which may reflect the presence of striatal hamartomas as reported here. 35Tsc1 deletion by Emx1-Cre also increases Pax6 NSCs and BrdU labeling, but leads to NSC exhaustion, and increased GFAP positive cells. 38BrdU labeling at E15.5 resulted in fewer BrdU positive Ki67 negative cells in the Emx1-Cre x Tsc1 model and the authors hypothesized this may indicate abnormal differentiation. 38In a congruent line of thought, several groups have proposed that mTORC1 regulation of translation may be important for V-SVZ NSCs.For example, mTORC1 was shown to be active in transit amplifying cells (TACs) but not in NSCs and that inhibition of mTORC1 causes a reversible quiescent phenotype that depletes TACs. 79However, a conflicting report indicated that dividing NSCs also have active mTORC1 which is required for TAC generation and that up-regulation of mTORC1 activity through 4E-BP1 phosphorylation induces differentiation. 80On the other hand, expression of constitutively active mTOR in NSCs induces massive Hif1a dependent apoptosis. 84Nevertheless, these have focused on dorsal NSCs because of their relationship to cortical development and cortical tubers which may not be applicable to postnatal lateral and ventral V-SVZ NSCs.
The role of V-SVZ NSCs in TSC has been previously examined in rodents as they may be a cell of origin for neurodevelopmental diseases and brain tumors which stems from the fact that brain growths are often highly heterogeneous. 85Single-cell Tsc1 deletion from NSCs was used to test the hypothesis that V-SVZ inhibitory NSCs generate SENs and SEGAs. 61Similar experiments used the same approach of conditional Tsc1 mice crossed to tamoxifen-inducible Nestin-CRE or Ascl1-CRE mice. 60Tsc1 deletion caused nodules to form.Nodules had minor increases in pS6 and neuronal heterotopias.However, SEGAs were not reported.Homozygous deletion of a second gene, Pten, along with Tsc1, in neonatal NSCs, generated SEGA-like lesions. 47The Pten/Tsc1 deletion model faithfully recapitulates most aspects of SEGAs and provides one of the most compelling examples that rodents can be used to model human brain disorders.However, whether the requirement for Pten deletion represents a technical hurdle to using mice or reflects a yet to be discovered genetic link to SEGAs is unclear.The model presented here differs in that only Tsc2 was removed and has the additional advantages of low mortality, no seizure activity allowing for distinguishing seizure-dependent and seizure-independent changes, and focal hamartomas which mimics that seen in patients and therefore affords the tracking of cells and delineation of cell autonomous and intercellular effects of loss of Tsc2.Nevertheless, these results argue that rodents can generate hamartoma-like lesions and that deletion of TSC genes in neonatal V-SVZ NSCs can sufficiently model TSC.
Hamartomas in the model described here contained neurons having a striatal-like morphology which vastly differed from GC-like morphologies of the OB.Occasional granule cell-like neurons were present in the rostral migratory stream (RMS) and V-SVZ and could represent those destined for the OB.In agreement, neurons are scattered along the RMS, mirroring results using Tsc1 or Pten V-SVZ deletion. 61,86owever, neurons in the striatum in the current report had large dendrite arbors that differ greatly from Tsc2 mut/mut OB GCs. 70It is a plausible hypothesis that micro-hamartomas not detected by conventional clinical imaging techniques could contribute to TSC-associated neurological disorders. 44Given the location within the striatum, it could be that TSC neurons could modulate motor activity or reward pathways. 87Notably, balloon-like giant cells that were multinucleated and cytomegalic were also identified within the striatum.These cells were slightly Neu-N positive with moderate pS6 staining.Similar cells were occasionally detected within the cerebral cortex.These cells represent, in our opinion, the most compelling case that loss of Tsc2 in murine models recapitulates TSC-specific patient cellular phenotypes.It is noteworthy that giant cells were often found among clusters of cells with astrocyte morphology.9][90][91] Notch signaling appears to be critical in reprogramming striatal astrocytes. 90Notch signaling is altered in TSC. 92,93Indeed, loss of Tsc1 has been shown to alter Notch and alter the progression of NSC-like cells in lymphangioleiomyomatosis and along the neural tube. 93Thus, loss of TSC genes leading to hyperactive mTORC1 might be hypothesized to prevent NSC-astrocyte programming.Most patient therapies are directed toward inhibiting mTORC1 and alleviating neurological symptoms.However, promoting terminal differentiation could in theory reduce the growth and development of TSC hamartomas.
V-SVZs were cultured, and cytomegalic multinucleated giant cells were found adhering to culture plates thereby supporting in vivo observations of giant cells.Tsc2 wt/wt vs. Tsc2 mut/mut NSCs had consistent transcriptomic differences.Notable differences included changes in translational regulatory and ribosome encoding transcripts.However, attempts to confirm changes at the protein level were often fruitless with unpredictable changes.We hypothesized that stochastic changes in mRNA translation caused by mTORC1 phosphorylation of translational machinery might be involved.Thus, delineating whether transcriptomic changes are a direct effect of loss of Tsc2 or secondary to changes in cellular homeostasis was not feasible in this study, although future studies could answer such a question by treating cells with mTORC1 inhibitors and examining consequences.
Changes to p4E-BP, pS6, and ribosome transcripts and ribosome proteins supported that translation is altered in Tsc2 mut/mut NSCs.Indeed, a unique translatome was confirmed by polyribosomal profiling.However, this in theory could reflect changes in the abundance of the total mRNA pool.Yet, when the total mRNA availability was corrected for, notable differences in translational efficiency were found.These results support the notion that the translational efficiency of specific transcripts is altered.Interestingly, translational efficiency of mitochondrial respiration and of translational regulators themselves were noted.This is consistent with the well-established role of TSC-mTOR pathway in controlling translation and specifically in regulating mitochondrial transcript translation.Thus, translation is subject to multi-tier regulation by Tsc2.
The capacity for TSC1/2 regulation of NSC differentiation appears conserved in human NSCs.Organoids that were genetically engineered to lose one or both alleles of TSC1 or TSC2 produced more astrocytes at the expense of neurons. 63However, loss of both alleles was required in this study.A recent manuscript described Tsc2 heterozygous organoids generated from passage 40-50 induced pluripotent stem cells. 52rganoids developed small protrusions after 220 days in culture.Protrusions had cells with mTOR pathway activity, EGF receptor expression, and a high proliferative index.ScRNA-seq of organoids provided evidence that protrusions shared markers of caudal late gestation inhibitory interneuron progenitors (CLIPs).The CLIP cell origin differs from bulk sequencing of RNA from human SEGAs.7][58][59] Microarrays identified $50 differentially expressed transcripts in SEGAs including a reduction in AldoC which were reduced in our model. 56Another study used bulk RNA-seq of SEGAs and found 3,693 differentially expressed transcripts which included those found by microarray. 57Cellular deconvolution software identified immune cell transcripts as highly differentially expressed supporting the need to perform single cell resolution sequencing.SENs and SEGAs were enriched in immature neuron, NSC, and astrocyte markers.Analysis of an additional 19 SEGAs revealed that $10,000 transcripts were differentially expressed (4,621 increased and 4,779 decreased). 58$1,000 of transcripts were shared between these RNA-seq studies.Differentially expressed transcripts include Vimentin, Olig1, Olig2, Nestin, Sox genes, Ednrb, Egfr, Ttyh1, Ptgds, Nkx6-, and Nkx2, Hopx, Emx2, Prox1, Pdgfra, and Nr2f1 (Coup-TFI) indicating the presence of NSCs.However, GFAP, SCGN, or Coup-TFII were not.These latter transcripts are critical because their absence suggests a lateral or medial ganglionic eminence origin.A lateral or medial ganglionic eminence origin differs from the report that cultured TSC organoids and concluded a human specific caudal ganglionic eminence origin. 52oreover, the Pten-Tsc1 deletion mouse model provides strong support for the origin of SEGA-like lesions from V-SVZ NSCs. 47In addition, a more recent manuscript indicated that Tsc2 knockout using NKX2.1 CRE resulted in SVZ abnormalities consistently in the ventral SVZ and that this is because the ventral SVZ NSCs have more mTOR activity. 68The NKX2.1 CRE is expressed in a wide range of progenitors beginning at embryonic day 10.5 and effects cells that produce inhibitory neurons within the pallidal telencephalon including the medial ganglionic eminence and OB.Taken together, these results support that an NSC-like cell originating from the lateral or medial ganglionic eminences is enriched in SEGAs.Nevertheless, given the contradictory reports in clinical literature of markers and cell types, it could be that there are multiple paths to forming a SEGA depending upon the originating cell type mutation.
SnRNA-sequencing of our mouse model confirmed NSCs as a cohort of cells that could be parcellated by activation state.Our findings highly overlapped with and hence were inspired by analysis of Troy positive V-SVZ NSCs. 7The preparations of our samples, although enriched in NSCs, also included a wide range of additional cell types which were also altered.Whether the changes reflect cell-autonomous or non-cell autonomous effects will require further experimentation.Nevertheless, loss of Tsc2 decreased many NSC activation markers.Tsc2 loss could cause precocious differentiation into other cell types.Alternatively, Tsc2 loss could delay differentiation through the retention of stem cell transcripts.Hamartomas retained stem proteins in support of the latter hypothesis.In conclusion, these results prove that stemness is not lost and therefore differentiation is incomplete following V-SVZ NSC Tsc2 deletion.These results demonstrate the evolutionarily conserved importance of Tsc2 in regulating neuroprogenitor differentiation.

Limitations of the study
The study reported here demonstrates the importance of Tsc2 in differentiation.While it is tempting to generalize these findings to TSC, the applicability is less certain.Mouse models generated by conditional deletion of Tsc1 or Tsc2 have contributed greatly to our understanding of TSC pathogenesis including with regards to brain, heart, and kidney manifestations.However, despite the evolutionary conservation of the biochemical pathways, there remain considerable gaps in our knowledge in how TSC pathogenesis occurs in vivo in patients and the applicability of these rodent models.Models that use homozygous loss of function mutations appear appropriate since over 70% of mutations reported lead to loss of TSC protein expression and loss of heterozygosity occurs in nearly all growths.Despite detailed histopathological characterization and use of bulk techniques such as RNA-seq to determine the composition of human TSC SEGAs, the cellular composition remains unclear.While expansion of the extracellular matrix, immune infiltration, and calcification are well documented which could be sufficient to cause growths, whether additional changes in the cellular composition also occurs is unknown.In the absence of consensus for the composition and hence cellular origins, what the appropriate control is for TSC SEGAs is unclear.This has caused extraordinary challenges in comparing SEGA characteristics.For example, SEGAs often are found within lateral ventricles indicating they could arise from the choroid plexus.Should one therefore compare SEGA data to control choroid plexus is unclear.Whether loss of TSC genes in NSCs, as occurs in our model, reflects the bona fide cell of origin in patients is currently unanswerable.This is particularly relevant because of the distinct origins of another lesion in TSC patients called lymphangioleiomyomatosis.Moreover, the literature has conflicting histopathological characterization of the cell types in SEGAs. 19One potential reason for these differences is that many cell types may be able to generate SEGAs.SEGA cellular heterogeneity therefore might reflect ontogenetic heterogeneity.Indeed, while SEGAs are most frequently surgically removed from the foramen of Monro near the subfornical organ, SEGAs and SENs frequently initiate along other parts of the walls of the lateral ventricles including striatum.
Another limitation to the study presented here is that cells were cultured in defined media components.These conditions include exposure to exogenous ligands such as EGF.Cell culture conditions are highly unlikely to reflect in vivo conditions where factors such as EGF are released in gradients during specific time periods.Moreover, SEGAs are not aggressive malignancies and do not grow at excessively fast rates.Thus, pushing NSCs into a state of cell division would appear to have limited applicability.Nevertheless, given the protracted rate of human brain development, it could be argued that the events identified in the current study mimic events of early TSC hamartoma formation when NSCs are known to divide.Finally, technical aspects of the genetic manipulations performed remain a limitation to our study.The current study removed both TSC genes at the same time.This contrasts to the many patients born with one functional and one mutant allele.Thus, the complex interplay between heterozygous and homozygous mutant cells is not modeled.Nevertheless, the genetic manipulation does mimic what can occur in mosaic TSC patients and non-TSC patients that have brain lesions caused by TSC mutations which include SEGAs as well as focal cortical dysplasia.

Electroporation and tamoxifen injections
Neonatal mice were electroporated as previously described. 105Mouse pups were injected with equal concentrations and volumes of DNA plasmids diluted in phosphate buffered saline (PBS) with 0.1% fast green.Electroporated plasmids are as follows; CAG-CRE (Plasmid #13775, Addgene) and CAG-GFP (Plasmid #11150, Addgene). 94,95DNA was injected into the lateral ventricles and delivered using a borosilicate glass micropipette generated from pulled capillary tubes.Borosilicate capillary tubes were pulled with a P97 Sutter micropipette puller.Tweezer electrodes (model 520; BTX) were rinsed in 0.9% saline solution and swept over the head of neonatal pups using five, 100-volt square pulses of 50 ms duration with 950-ms intervals that were applied using a pulse generator (ECM830; BTX).Nestin-CRE-ER T2 mouse pups were weighed and injected with approximately 20 mg/g (Z)-4-Hydroxytamoxifen (Sigma Aldrich, #H7904) and 2.2 mg/g EdU (Invitrogen, #C10337).

Image analysis
Images (320) of RFP positive cells were uploaded to FIJI (ImageJ 1.5 g).The freehand selection tool was used to trace a region of interest (ROI) on electroporated and non-electroporated cells in the same Z section and a mean gray value measured to quantify the staining intensity of phospho-4EBP and normalized.Ratios of staining in electroporated and non-electroporated cells were compared for RFP positive cells in Tsc2 w/w and Tsc2 mut/mut conditions to account for immunohistochemical variation.All values were normalized to the average intensity of the ratio of electroporated divided by non-electroporated cells in Tsc2 w/w conditions.
For lesion analysis, lesions were defined in images (320) as containing approximately eight or more overlapping RFP positive cells present within the striatum and in an individual Z section.RFP+ cells were counted in the lesion and assessed for either Sox2, p4EBP, DCX, GS, or NeuN on a per cell basis.Circularity and area were measured in maximum intensity projections.The area of the lesions was then circumscribed using the freehand selection tool and the area and circularity quantified.Hamartomas were considered Type 1 when circularity values were 0.269-0.531and Type 2 when circularity values were 0.024-0.23.
Images (320) were used to measure dendrite morphology.Images were uploaded as described above and simple neurite tracer plug-in was used to trace dendrite processes of RFP positive cells.Sholl analysis was performed at 10 mm intervals to quantify dendrite arborization using the Sholl plug-in.The total number of dendritic crossings were calculated by taking the sum of crossings at 10 mm intervals for each traced neuron and averaging the total number of crossings per neuron in each condition.The freehand selection tool was also used to trace the soma of neurons to measure cell size in Tsc2 w/w and Tsc2 mut/mut conditions.
Measurements were made on original image files without enhancements.Figures were created by converting LIF files into TIFF format images using FIJI.Images were uploaded into Photoshop (Adobe, version 2022).Merged images containing multiple fluorophores were subject to brightness and contrast changes and RBG levels were optimized.Changes were applied across the entirety of each image.For individual-colored images, the RBG images were duplicated, and individual colors were removed.To make black and white images, the RFP channel was converted to gray scale and inverted.

Total RNA isolation
Trizol was added to NSCs grown on 100 mm plates.NSC lysate was removed, triturated, and placed on ice in 1.5 mL Eppendorf tubes.Chloroform was added to tubes after freezing at À80 C and vigorously shaken by hand for 15 s.Samples were incubated for 5 min at room temperature and then centrifuged at 12,000 3 g at 4 C.The upper aqueous phase was placed in a fresh tube and protein was subject to chloroform methanol precipitation for Western blot analysis.The aqueous phase was used to precipitate RNA by adding isopropanol and incubated overnight at À20 C. Samples were centrifuged for 10 min at 12,000 3 g at 4 C. Supernatant was removed and pellets were washed with 75% ethanol.Samples were spun at 7,500 3 g for 5 min at 4 C. Supernatant was removed and pellet air dried and resuspended in dH 2 O.Samples were incubated with DNase I and buffer RDD.Samples were then subject to Qiagen MinElute Column clean up on an RNeasy MinElute spin column.Briefly, buffer RLT is added to samples and mixed followed by addition of 100% ethanol.Samples were mixed and transferred to RNeasy MinElute spin column in a collection tube.Samples were centrifuged for 15 s at 8,000 x g twice.Columns were placed into a fresh 2 mL collection tube and 80% ethanol is added to MinElute column.Samples are centrifuged at 8,000 3 g for 2 min and spin column is placed into a new tube.Samples were subject to an additional centrifugation and RNA is eluted in RNase free water and subject to quantitative and qualitative analyses.

Polyribosome profiling
Polyribosome profiling was performed similar to as previously described. 106Primary NSC cultures were passaged once prior to isolations. 100 mg/mL (66.6uL) cycloheximide (CHX, VWR, #94271) was added to each dish and incubated at 37 C for 15 min.The cells were dissociated in 500 mL lysis buffer (0.25 M sucrose, 50 mM Tris/HCl (pH7.5, Sigma Aldrich, #93363), 5 mM MgCl 2 (Sigma Aldrich, #M1028), 25 mM KCl (Sigma Aldrich, #P9333), 200 mg/mL CHX (VWR, #94271), 1X HALT protease inhibitor (Thermo Scientific, #87785), 1 mM dithiothreitol (DTT, Enzo, #ALX-280-001-G005), 100 U/mL recombinant RNAsin ribonuclease inhibitor (Promega, #N2511), scraped into a 1.5 mL microcentrifuge tube, and incubated for 15 min on ice.Lysates were homogenized five times with a Dounce homogenizer.50 mL was removed and 150 mL TRIzol LS reagent (Ambion, #10296010) was added for RNA Total isolation.The remaining lysates were centrifuged for 10 min at 500 3 g at 4 C. 10% sodium deoxycholate and 10% NP-40 was added, and the samples were incubated on ice for 30 min.Prepared lysates were loaded onto 10 mL continuous sucrose gradients that consisted of 17.5-50% sucrose (as well as 30 mM Tris-HCl (pH7.5, Sigma Aldrich, #93363), 30 mM MgCl 2 (Sigma Aldrich, #M1028), 600 mM NaCl (Fisher Scientific, #BP358-10), 200 mg/mL CHX (VWR, #94271), 2 mM DTT (Enzo, #ALX-280-001-G005), and UltraPure DEPC treated H 2 O (Invitrogen, #750023).The gradients were balanced using lysis buffer and were centrifuged for 1.5 h at 280,000 3 g in an SW-41 Ti swinging bucket rotor with the acceleration speed set to max and the deceleration speed set to coast.Seven 1080 mL fractions were then removed from the top of the sucrose gradients leaving 3 mL of the heaviest sucrose fractions.TRIzol LS reagent (9 mL) (Ambion, #10296010) was added and stored at À80 C overnight.The solutions were then split into two equal parts and 1.2 mL of chloroform (Sigma Aldrich, #372978) was added to each tube.The tubes were incubated for 3 min before centrifugation at 12,000 3 g for 15 min at 4 C.The aqueous layer was then removed and 3 mL isopropanol (Fisher Chemical, #A416-1) and 1 mL GlycoBlue co-precipitant (Invitrogen, #AM9516) was added.The tubes were then moved to the À80 C freezer to precipitate overnight.Samples were then centrifuged for 10 min at 12,000 x g at 4 C and the supernatants discarded.Pellets were resuspended in 6 mL 75% EtOH (Sigma Aldrich, #E7023) and vortexed before centrifugation for 5 min at 7,500 3 g at 4 C.The supernatants were discarded, and the pellets allowed to air dry for 8 min before resuspension in 50 mL UltraPure Distilled water (Invitrogen, #10977-015).The samples were incubated for 15 min at 57 C in a water bath before being recombined into one polysomal RNA fraction each.Chloroform (Sigma Aldrich, #372978) was added to each cell lysate that was collected during polysomal profiling as a total RNA fraction.The tubes were incubated for 3 min before centrifugation at 12,000 3 g for 15 min at 4C.The aqueous layer was then removed and 100 mL isopropanol (Fisher Chemical, #A416-1) and 1 mL GlycoBlue co-precipitant (Invitrogen, #AM9516) was added.The tubes were then moved to the À80 C freezer to precipitate overnight.Samples were centrifuged for 10 min at 12,000 3 g at 4 C and the supernatants discarded.Pellets were resuspended in 200 mL 75% EtOH (Sigma Aldrich, #E7023) and vortexed before centrifugation for 5 min at 7,500 3 g at 4 C.The supernatants were discarded, and the pellets allowed to air dry for 8 min before resuspension in 50 mL UltraPure Distilled water (Invitrogen, #10977-015).The samples were incubated for 15 min at 57 C in a water bath before library prep.
Single Nuclei RNA sequencing bioinformatics BCL files from the sequencer were demultiplexed and converted to FASTQ files using the mkfastq command and default parameters within the Cell Ranger 7.0.1 pipeline.FASTQ files were preprocessed for quality and length and aligned to Mus musculus reference genome GRCm38 (mm10) using the counts command with -force-cells parameter set to 60000 to account for spurious cell identity in one of the samples.This number was derived by calculating the median and rounding to the nearest 10000 from the unaffected samples part of the sequenced set.The output from the counts command was exported into R for quality control, multi-sample integration, cell-type clustering, cell-type marker identification and differential expression within each cluster using the Seurat v4 single cell analysis platform 108 and previously described methods. 109,110Differential expression lists were curated by removing mitochondrial encoded transcripts.The bash and R code for Cell Ranger and Seurat pipeline, respectively can be found here: https://github.com/vshanka23/tsc_scrna_feliciano.

Western blot
Lysates from NSCs subject to Trizol RNA isolation were further processed by chloroform methanol isolations.Proteins were placed into Laemmli buffer and heated to 95 C for 5 min.Samples were resolved by electrophoresis on 10% polyacrylamide precast mini-Protean gels and transferred to polyvinylidene difluoride (PVDF) membranes.Blots were cut to probe upper vs. lower parts of the same blot for different proteins.Blots were rinsed in Tris-buffered saline (TBS-T, 0.1% Tween 20) for 5 min at room temperature and subsequently blocked in 5% weight/volume blotting grade block in TBS-T for 1 h at room temperature.Samples were incubated for 1 h at room temperature or overnight at 4 C with the following antibodies from Cell Signaling Technology at a 1:1,000 dilution: Tuberin (D93F12, Cat#4308), ATF4 (D4B8, Cat# 11815), Actin (13E5, Cat# 4970), phospho-RPS6 (D68F8, Cat# 5364), RPL26 (D8F6, Cat#5400), RPS3 (D50G7, Cat# 9538), and RPS6 (5G10, Cat# 2217).Blots were washed three times each for 10 min in TBS-T, samples were incubated for 1 h at room temperature with donkey or goat anti-rabbit antibodies in blocking buffer and washed for 15 min in TBS-T and visualized using a Bio-Rad Chemidoc MP imaging system.PVDF membranes were stripped for at room temperature using Restore Western Blot Stripping Buffer according to manufacturer's recommendations (#21059, Thermo Fisher Scientific).p values less than 0.05 were considered significant unless otherwise noted.No methods were used to determine whether the data met assumptions of the statistical approach.Quantification and statistics for single cell RNA sequencing (Figures 4A and 4B, N = 4 mice per genotype, Seurat), polyribosome profiling (Figures 3A-3D, N = 4 cultures per genotype, DeSeq2 and tRanslatome, p adj. <0.00001), bulk RNA sequencing (Figures 2A and 2B, N = 3 cultures per genotype, DeSeq2, p adj. <0.00001), and GO term analysis (Figure 2E, N = 4 mice per genotype; Figures 3E and 3F, N = 4 mice per genotype, p < 0.05) was determined using the indicated programs for bioinformatic analyses.All experiments were performed on 6-17 mice per condition per time point.N (number of mice) and n (number of cells or hamartomas) are listed where applicable.Error bars are reported as standard error mean.

Figure 2 .
Figure 2. NSC Tsc2 deletion alters ribosome biogenesis (A) Upper-Timeline of NSC culturing and RNA sequencing.Lower-Heatmap and hierarchical clustering of mRNA transcripts from Tsc2 w/w and Tsc2 mut/mut NSCs.(B) Example of topmost increased and decreased mRNA transcripts from Tsc2 w/w and Tsc2 mut/mut NSCs.(C) Network analysis of differentially expressed transcripts from Tsc2 w/w and Tsc2 mut/mut NSCs.Green circles highlight proteins critical for translation.(D) Translation-related genes in top 100 most up-regulated genes by Tsc2 deficiency (red).(E) G.O. analysis indicating top biological processes represented in differentially expressed transcripts.See also Figures S4-S7.

Figure 3 .
Figure 3. NSC Tsc2 deletion alters mRNA translation dynamics (A) Upper-Timeline of NSC culturing and RNA sequencing.Lower-Heatmap of topmost differentially translated polyribosomal mRNAs.(B) Heatmap of topmost differentially translated mRNAs.(C) Scatterplot of mRNA transcript availability compared to translatome abundance.Each diamond represents an mRNA transcript plotted on the x axis which represents the total mRNA abundance (transcriptome) and the y axis represents the polyribosome abundance (translatome).A gray diamond indicates there is no change between Tsc2 wt/wt and Tsc2 mut/mut .A blue diamond indicates changes to the total mRNA abundance only without translational changes.A yellow diamond indicates changes only to the translatome.A green diamond denotes changes for both transcriptome and translatome that occur in the same direction (i.e., more total mRNA correlates with more polyribosome mRNA = homodirectional).A red diamond reflects changes in the opposite direction (more total abundance but less polyribosome abundance).(D) Histogram of transcript and translatome abundance and changes.(E) G.O. analysis of most highly expressed mRNAs with altered translation efficiency.(F) G.O. analysis of mRNAs having the most highly increased translational efficiency.Green highlights those terms mentioned in text.

Figure 4 .
Figure 4. Tsc2 mutant neuroprogenitors exhibit altered transitional states (A) U-MAP clustering of V-SVZ microdissected cells from four mice for each genotype (Tsc2 wt/wt and Tsc2 mut/mut ) and represented as a composite.Data represent an N of 8 mice.(B) Percentage of cells represented in respective clusters.(C) Summary of top 10 transcripts (Identifiers) enriched in the indicated clusters 15, 8, 4, and 2. Each row indicates a cluster type and their representative activation states (left) with cluster 15 being the least activated and cluster 2 being the most activated.See Table S2 for additional clusters and respective differentially enriched transcripts.(D) Each row represents the indicated cluster (15, 8, 4, and 2).Each cluster (row) contains the differentially expressed transcripts in Tsc2 mut/mut mice compared to Tsc2 wt/wt mice.Each listed transcript is significantly decreased in the Tsc2 mut/mut condition.Note that the transcripts also happen to be important identifiers of clusters listed in C. For reference, cluster numbers were placed at the bottom of D to identify the clusters that the transcripts are identifiers of.The decrease in transcripts reflect changes in activation states, most notably, a reduction in transcripts found in cluster 2 indicating a change in activation dynamics.Columns represent the clusters that the transcripts are enriched in.(E-I) 20x images of CRE and GFP electroporated brains with Tsc2 mut/mut RFP (red) and DCX (E, blue), Sox2 (F, blue), Glutamine Synthetase (GS) (G, blue), Nestin (H, blue), or NeuN (I, blue) stained hamartomas and nodules.

TABLE
d RESOURCE AVAILABILITY B Lead contact B Materials availability statement B Data and code availability d EXPERIMENTAL MODEL AND SUBJECT DETAILS B Animals B Primary NSC culture d METHOD DETAILS B Polymerase chain reaction (PCR) B Electroporation and tamoxifen injections B Immunohistochemistry B Image analysis B Total RNA isolation B Polyribosome profiling B RNA sequencing and bioinformatics B FACS and Single Nuclei RNA sequencing B Single Nuclei RNA sequencing bioinformatics B Western blot d QUANTIFICATION AND STATISTICAL ANALYSIS