Nf1 mutation disrupts activity-dependent oligodendroglial plasticity and motor learning in mice

Neurogenetic disorders, such as neurofibromatosis type 1 (NF1), can cause cognitive and motor impairments, traditionally attributed to intrinsic neuronal defects such as disruption of synaptic function. Activity-regulated oligodendroglial plasticity also contributes to cognitive and motor functions by tuning neural circuit dynamics. However, the relevance of oligodendroglial plasticity to neurological dysfunction in NF1 is unclear. Here we explore the contribution of oligodendrocyte progenitor cells (OPCs) to pathological features of the NF1 syndrome in mice. Both male and female littermates (4–24 weeks of age) were used equally in this study. We demonstrate that mice with global or OPC-specific Nf1 heterozygosity exhibit defects in activity-dependent oligodendrogenesis and harbor focal OPC hyperdensities with disrupted homeostatic OPC territorial boundaries. These OPC hyperdensities develop in a cell-intrinsic Nf1 mutation-specific manner due to differential PI3K/AKT activation. OPC-specific Nf1 loss impairs oligodendroglial differentiation and abrogates the normal oligodendroglial response to neuronal activity, leading to impaired motor learning performance. Collectively, these findings show that Nf1 mutation delays oligodendroglial development and disrupts activity-dependent OPC function essential for normal motor learning in mice.

Neurogenetic disorders, such a s n eu ro fi br om atosis type 1 (NF1), can cause cognitive and motor impairments, traditionally attributed to intrinsic neuronal defects such as disruption of synaptic function.Activity-regulated oligodendroglial plasticity also contributes to cognitive and motor functions by tuning neural circuit dynamics.However, the relevance of oligodendroglial plasticity to neurological dysfunction in NF1 is unclear.
Here we explore the contribution of oligodendrocyte progenitor cells (OPCs) to pathological features of the NF1 syndrome in mice.Both male and female littermates (4-24 weeks of age) were used equally in this study.We demonstrate t h a t m i c e w i t h global or OPC-specific Nf1 heterozygosity exhibit defects in activity-dependent oligodendrogenesis and harbor focal OPC hyperdensities with disrupted homeostatic OPC territorial boundaries.These OPC hyperdensities develop in a cell-intrinsic Nf1 mutation-specific manner due to differential PI3K/AKT activation.OPC-specific Nf1 loss impairs oligodendroglial differentiation and abrogates the normal oligodendroglial response to neuronal activity, leading to impaired motor learning performance.Collectively, these findings show that Nf1 mutation delays oligodendroglial development and disrupts activity-dependent OPC function essential for normal motor learning in mice.
In the central nervous system (CNS), oligodendrocyte progenitor cells (OPCs) continuously proliferate throughout life, giving rise to oligodendrocytes, which are critical for myelination and the proper function of neural circuits [1][2][3] .In many brain regions, OPC division and differentiation are tightly regulated by neuronal activity, thus coupling adaptive myelination to neuronal signal propagation and network function 2,[4][5][6] .As such, disruption of activity-regulated OPC differentiation results in wide-ranging effects on multiple domains of cognition, including attention, learning and memory [6][7][8][9] .For example, genetic inhibition of OPC differentiation results in motor learning deficits on the complex wheel test 7,8 .
While neuronal activity-dependent regulation of oligodendroglial dynamics is essential for normal CNS function, the contribution of dysregulated activity-dependent oligodendrogenesis to neurological and neuropsychiatric disorders is just beginning to come to light.Prior studies have shown that activity-regulated OPC proliferation, Article https://doi.org/10.1038/s41593-024-01654-yadaptive responses, impairs oligodendroglial dynamics and results in motor learning deficits.
As expected for Nf1 WT mice 4 , optogenetically induced neuronal activity increased OPC proliferation ipsilateral to the site of optogenetic neuronal stimulation relative to the contralateral, nonstimulated side of the same mouse (Fig. 1e).During the 3 h following stimulation, newly proliferating OPCs have not yet differentiated into oligodendrocytes (Extended Data Fig. 1a); no microglial reactivity (Extended Data Fig. 1b-d) and few apoptotic cells (Extended Data Fig. 1e,f) were observed.As a control for surgical manipulation and blue light oligodendrogenesis and myelination are disrupted following chemotherapy, which contributes to chemotherapy-related cognitive impairment in mice 6 .Similarly, in rodent models of absence epilepsy, OPC proliferation, oligodendrocyte numbers and myelination are increased within the seizure network, and this aberrantly increased maladaptive myelination contributes to epilepsy progression such that genetic or pharmacological blockade of activity-regulated oligodendrogenesis decreases seizure frequency 10 .In another example of dysregulated oligodendroglial precursor proliferation leading to disease, OPCs can serve as a cell of origin for both low-and high-grade gliomas [11][12][13][14] .
The contribution of OPCs to both neurological dysfunction and gliomagenesis is particularly germane to neurofibromatosis type 1 (NF1), a cancer predisposition syndrome in which affected individuals are also prone to learning, behavioral and motor deficits.Patients with NF1 are born with a germline inactivating mutation in one copy of the NF1 gene (monoallelic or heterozygous NF1 loss) but may acquire a 'second-hit' mutation (biallelic NF1 loss) during development in susceptible cell types to induce glioma formation 15,16 .In addition to increased brain tumor risk, children with NF1 exhibit impairments in attention, learning, working memory, executive function, motor function and motor learning [17][18][19] , which could reflect abnormalities in adaptive myelination 4,6,7,9 .Support for dysregulated OPC function in the setting of NF1 derives from several studies: analysis of heterozygous Nf1-mutant mice reveals increased OPC density in the spinal cord 20 , while Nf1 genetic knockdown in zebrafish results in increased spinal cord OPC proliferation, density and migration 21 .Similarly, using the mosaic analysis with double markers (MADM) model, Nf1-null OPCs exhibit increased proliferation and decreased differentiation in vivo 22 .In this Article, we leveraged optogenetic and behavioral approaches coupled with numerous mouse strains harboring different NF1 patient germline Nf1 gene mutations and OPC-specific Nf1 loss to demonstrate that Nf1 mutation in OPCs disrupts their

Article
https://doi.org/10.1038/s41593-024-01654-yexposure, we expressed eYFP instead of ChR2-eYFP in Nf1 WT mice and observed no change in OPC proliferation following blue light delivery (Extended Data Fig. 1g), indicating that the increase in Nf1 WT OPC proliferation (Fig. 1e) results from optogenetic ChR2 activation of cortical projection neurons and is consistent with our previous findings 4 .
In striking contrast, optogenetically induced neuronal activity did not increase OPC proliferation in OPC-specific Nf1 OPC-iHet or Nf1 OPC-iKO mice (Fig. 1e), demonstrating that both monoallelic and biallelic Nf1 inactivation abrogates the OPC proliferative response to neuronal activity.Notably, the overall density of proliferating OPCs is greater in Nf1 OPC-iKO mice (Fig. 1e) due to a generalized increase in total OPC density in Nf1 OPC-iKO , relative to Nf1 WT and Nf1 OPC-iHet , mouse brains (Extended Data Figs.1h and 2a,b).This biallelic Nf1 inactivation-induced increase in OPC density is consistent with previous findings demonstrating that Nf1-null OPCs exhibit increased proliferation 22 .Collectively, our findings reveal that Nf1-mutant OPCs lack the expected adaptive proliferative response to neuronal activity, and support a causative role for Nf1 in regulating the homeostatic density of OPCs, which is strictly maintained in the healthy brain 24 .

Deficiency in experience-regulated oligodendrogenesis
Since Nf1 mutation leads to OPC dysregulation, we next asked whether Nf1 inactivation impairs OPC responses to neuronal activity in the context of motor learning as measured using the complex wheel test.The complex wheel has unevenly spaced rungs (Fig. 2a), requiring motor learning in order for the mouse to remain on the wheel.As mice learn, they run increasingly faster for the duration of the observation period 7 .This motor skill learning task induces OPC proliferation and the generation of new oligodendrocytes, and this activity-regulated oligodendrogenesis is necessary for complex wheel motor learning 7,8 .
Since loss of either one or both Nf1 alleles in OPCs abrogates activity-regulated OPC proliferation (Fig. 1e), we hypothesized that Nf1 mutation might also impair experience-and activity-regulated oligodendrogenesis.To evaluate this possibility, Nf1 WT and Nf1 OPC-iKO mice were given EdU in their drinking water during the complex wheel test to trace OPC proliferation and differentiation (Fig. 2a).At the end of 7 days of complex wheel training, we analyzed the percentage of newly proliferated OPCs (EdU + /PDGFRα + cells) and newly generated oligodendrocytes (OLs; EdU + /ASPA + cells) in the cingulum, where dense axonal tracks of the motor cortex reside (Fig. 2b).Consistent with previous studies 7,8 , Nf1 WT mice following complex wheel training harbor increased numbers of new oligodendrocytes without any change in EdU-labeled OPC content at this time point (Fig. 2c), indicating that in healthy mice this motor learning experience increases oligodendrogenesis in the cingulum and that by this time point OPCs that proliferated in response to the motor learning paradigm have differentiated into oligodendrocytes.In striking contrast, at the end of the test, mice Article https://doi.org/10.1038/s41593-024-01654-ywith Nf1-null OPCs exhibit increased numbers of EdU-labeled OPCs, consistent with the overall increase in proliferating OPCs, and fewer new oligodendrocytes relative to their WT counterparts (Fig. 2d,e).No apoptosis (Extended Data Fig. 2c) after complex wheel training was detected at the time point examined to account for the observed reduction in new oligodendrocytes generated.Taken together, these data support a critical role for the Nf1 gene in OPC differentiation and demonstrate a deficit in experience-dependent oligodendrogenesis in Nf1-null OPCs.

Nf1 loss generates OPC hyperdensities via PI3K/ AKT activity
To determine whether monoallelic Nf1 inactivation also leads to deficits in experience-dependent oligodendrogenesis, we next analyzed OPC dynamics in Nf1 OPC-Het (Nf1 fl/+ ;Pdgfra::Cre) mice at the end of the 7-day complex wheel test.Interestingly, we observed focal areas containing increased OPC density (focal OPC hyperdensities) throughout the brains of Nf1 OPC-Het mice (Fig. 3a), a finding indicative of impaired control of OPC density 24 .These focal OPC hyperdensities were also present in Nf1 OPC-iHet mice (Nf1 fl/+ ;Pdgfra::Cre ER , tamoxifen injected at P24) not subjected to the complex wheel testing (Extended Data Fig. 2d,e), suggesting that formation of focal OPC hyperdensities is not a motor learning-driven event.OPC density within the focal OPC hyperdensities of OPC-specific Nf1-heterozygous mice is similar to that observed globally in the brains of Nf1 OPC-iKO mice (Extended Data Fig. 2e).In addition to the forebrain, focal OPC hyperdensities were also found in the hindbrains of Nf1 +/− mice, more often seen in the brainstem than the cerebellum (Extended Data Fig. 2f,g).These findings raise the intriguing idea that these regions of OPC hyperdensity, in which the normal OPC territorial boundaries 24 are not respected, represent areas in which OPCs have lost expression of the remaining functional Nf1 allele (Nf1-null OPCs).
Within these focal OPC hyperdensities, no changes in the density of microglia (Iba1 + cells), reactive microglia (CD68 + /Iba1 + cells) or reactive astrocytes (Cxcl10 + /Sox9 + cells) were observed (Extended Data Fig. 3a-c).We also did not detect senescent cells (p21 + cells) or apoptotic cells (TUNEL + cells) within these focal OPC hyperdensities (Extended Data Fig. 3d).The size of the focal OPC hyperdensities appears to increase with age (Fig. 3b-d).We found a transient increase in OPC proliferation in small focal OPC hyperdensities at 5 weeks, but not at 24 weeks, of age compared to regions with normal OPC density in the same brains (Extended Data Fig. 4a-d).Whereas the ability of OPCs to generate new oligodendrocytes (evidenced by EdU tracing) is reduced in the focal OPC hyperdensities relative to nonhyperdense areas (Extended Data Fig. 4e), the density of mature oligodendrocytes inside and outside of the focal OPC hyperdensities did not differ in adult mice (Fig. 3e).In contrast, mice harboring monoallelic inactivation of other tumor suppressor genes (for example, Trp53, Pten and Rb1) did not exhibit focal OPC hyperdensities (Extended Data Fig. 5), establishing this phenotype as unique to NF1.
As a GTPase-activating protein, one of the main functions of the NF1 protein (neurofibromin) is to negatively regulate RAS activity.Using mice in which constitutively active KRAS is targeted to oligodendroglial lineage cells (Kras LSL-G12D ;Olig2::Cre), KRAS hyperactivation phenocopies Nf1 loss (Fig. 3f), indicating that increased KRAS activity is sufficient to induce focal OPC hyperdensities.To determine whether KRAS is necessary for inducing focal OPC hyperdensities in heterozygous Nf1-mutant mice, we engineered Kras haploinsufficiency

Article
https://doi.org/10.1038/s41593-024-01654-y in the Nf1-mutant mice.Whereas Kras haploinsufficiency normalizes Nf1 mutation-induced RAS hyperactivation and does not induce OPC hyperdensities on its own (Extended Data Fig. 6a-c), Kras haploinsufficiency fails to reduce the size of the focal OPC hyperdensities in heterozygous Nf1-mutant mice (Extended Data Fig. 6c).Taken together, these findings suggest that oncogenic RAS hyperactivation in OPCs is sufficient to generate OPC hyperdensities but is not fully responsible for OPC hyperdensity formation in Nf1-mutant mice.These findings indicating sufficiency but not necessity prompted us to further explore the causative etiology underlying OPC dysfunction in Nf1-mutant mice.First, we leveraged a collection of Nf1-mutant mouse strains harboring different heterozygous NF1 patient-derived germline Nf1 gene mutations.Using this approach, we identified one line (Nf1 +/C383X ) that developed OPC hyperdensities similar to the Nf1-heterozygous mice engineered by inserting a neomycin cassette into exon 31 of the Nf1 gene (Nf1 +/neo ), while three other lines (Nf1 +/R1809C , Nf1 +/G848R and Nf1 +/R1276P ) did not exhibit OPC hyperdensities (Fig. 4a).In all of these NF1 model mouse strains, irrespective of OPC hyperdensity development, RAS activity is elevated [25][26][27] .Second, we examined PI3K-AKT signaling, which is also dysregulated in Nf1-mutant cells [28][29][30] .Examining AKT activity levels in brain lysates from WT and Nf1-mutant mice, we found that, relative to WT mice, AKT activity was increased by >2-fold in Nf1 +/neo and Nf1 +/C383X mice that form OPC hyperdensities but was largely not changed in Nf1 +/R1809C , Nf1 +/G848R or Nf1 +/R1276P mice that harbor few or tiny OPC hyperdensities (Fig. 4b).Consistently, in OPCs derived from human induced pluripotent stem (iPS) cells with NF1 mutations, only OPCs with the NF1 C383X , but not the NF1 R1809C , mutation exhibited elevated AKT activity relative to their WT controls (Fig. 4c).Third, we demonstrate that in vivo pharmacological PI3K/AKT inhibition using NVP-BKM120 reduced the OPC hyperdensity-affected area in the brains of Nf1 +/neo mice (Fig. 4d).Taken together, these data suggest that differential AKT activation underlies the development of OPC hyperdensities in Nf1-mutant mice.

Impaired adaptive oligodendrogenesis in OPC hyperdensities
To determine whether OPCs within the focal OPC hyperdensities of Nf1 OPC-Het mice exhibit the same activity-dependent defects seen in Nf1 OPC-iKO mice, we quantified new OPCs and new oligodendrocytes in the cingulum of Nf1 OPC-Het mice at the end of the complex wheel test (Fig. 3a).Whereas the OPCs in regions of normal OPC density behave similarly to Nf1 WT OPCs, OPCs within focal OPC hyperdensities display impaired oligodendrogenesis at the end of the complex wheel test (Extended Data Fig. 7a) similar to Nf1-null OPCs (Fig. 2d,e).
Since the experience-dependent oligodendrogenesis deficits within focal OPC hyperdensities of Nf1 OPC-Het mice are similar to those observed in Nf1-null OPCs, we postulated that these hyperdense OPCs probably exhibit Nf1 loss of heterozygosity, resulting in reduced levels of Nf1 mRNA and protein.Unfortunately, few reagents (probes and antibodies) currently exist to accurately quantitate Nf1 RNA/protein by in situ hybridization or immunohistochemistry.We first evaluated multiple commercially available or laboratory-generated Nf1 probes and antibodies without success.We next performed spatial transcriptomic analysis, which allows for evaluation of regional RNA expression, in OPC hyperdensities relative to areas lacking these hyperdensities in heterozygous Nf1-mutant mice (Extended Data Fig. 7b).While Nf1 mRNA copy numbers in the brain were too low to evaluate differential expression, we detected increased expression of four genes (Ttr, Enpp2, Rarres2 and Ecrg4) in the focal OPC hyperdensities relative to regions lacking focal OPC hyperdensities (Extended Data Fig. 7c).Many of the candidates we identified in the spatial transcriptomics analysis have been previously implicated in oligodendroglial lineage function.First, transthyretin (Ttr) is expressed by OPCs and has been reported to promote both OPC proliferation and differentiation 31 .Second, ECRG4 augurin precursor (ECRG4) is a hormone-like peptide that induces OPC senescence 32 .Third, ectonucleotide pyrophosphatase/phosphodiesterase 2 (ENPP2) induces the differentiation of OPCs from Olig2 + precursor cells in the developing zebrafish hindbrain 33 .Notably, ENPP6, another member of the ENPP family, serves as a marker of motor learning (complex wheel)-induced oligodendrogenesis 8 .Immunofluorescence (IF) and in situ hybridization validation of the spatial transcriptomic data revealed increased areas of Ecrg4 (IF), Enpp2 (IF) and Ttr (in situ hybridization) signal in regions of focal OPC hyperdensities within Nf1 +/− mouse brains relative to brain regions lacking focal OPC hyperdensities (Extended Data Fig. 7d,e).
Taken together, these findings demonstrate that Nf1 loss leads to the development of focal OPC hyperdensities, within which OPCs exhibit defective experience-dependent oligodendrogenesis and dysregulated oligodendroglial gene expression relevant to OPC function.

Nf1-mutant mice exhibit delayed oligodendroglial development
After observing that Nf1 loss leads to defective experience-dependent oligodendrogenesis, we next asked whether baseline oligodendrogenesis is affected by Nf1 loss in OPCs using EdU lineage tracing.In both 1-month-old and 4-month-old Nf1 OPC-iKO mice, we observed increased OPC proliferation and reduced differentiation relative to Nf1 WT mice, with smaller differences in the 4-month-old mice (Extended Data Fig. 8a,b).We next examined oligodendroglial lineage progression in OPCs isolated from Nf1 OPC-iKO and Nf1 WT pups.Using a standard OPC differentiation assay, Nf1 loss (both monoallelic and biallelic) in OPCs decreases oligodendrogenesis in vitro (Fig. 5a,b).
Given the impaired OPC differentiation phenotype observed in Nf1-mutant OPCs, we quantified oligodendrocytes in developing and adult mice.At 1 month of age, Nf1 OPC-iKO mice exhibited reduced overall oligodendrocyte density compared to the Nf1 WT mice (Fig. 5c).This difference was no longer evident at 4 months of age (Fig. 5d).These findings indicate an impairment in developmental oligodendrogenesis that compensates by adulthood and suggests that the increased overall number of OPCs in Nf1 OPC-iKO mice may gradually compensate for their reduced capacity for oligodendrogenesis.Concordantly, by 4 months of age, we did not observe differences in myelination (number of myelinated axons and myelin sheath thickness) in the cingulum or corpus callosum of Nf1 OPC-iKO mice, as assessed by electron microscopy (Extended Data Fig. 8c-e), nor did we find differences in myelin or density of oligodendrocytes of adult Nf1 +/− mice (Extended Data Fig. 8f,g).The volume of the cingulum and corpus callosum was equivalent in Nf1-mutant and WT mice at all time points examined (Extended Data Fig. 8d,h).
Collectively, no gross deficits in baseline oligodendrocyte density or myelination were observed in Nf1-mutant mice at the age at which the complex wheel test was performed.We postulate that normal myelination is achieved by adulthood in Nf1-mutant mice, whereas activity-regulated oligodendrogenesis remains impaired.These data suggest that baseline abnormalities in OPC proliferation and differentiation contribute to the OPC/oligodendrocyte phenotype observed in Nf1-mutant mice following testing on the complex wheel and that failure of the activity-related response means that neuronal activity does not overcome this oligodendrogenesis deficit.

Nf1 loss in OPCs results in motor skill learning deficits
Since Nf1 loss leads to deficient experience-dependent oligodendrogenesis (Fig. 2d,e and Extended Data Fig. 7a), we next sought to determine whether OPC-specific Nf1 loss causes impaired motor learning in adult mice.Before evaluating motor learning, we first assessed baseline motor function and found that the Nf1 OPC-Het mice lack abnormalities in overall motor function, including stride length, paw intensity and swing speed during normal gait (Extended Data Fig. 9a).This was important   Focusing on complex wheel performance (measured as running velocity) as a function of running distance, we found that all Nf1 OPC-Het mice, when grouped together, exhibit similar motor performance as their Nf1 WT counterparts (Fig. 6a).As the hyperdense OPC areas in Nf1 OPC-Het mice exhibit impaired experience-dependent oligodendrogenesis (Extended Data Fig. 7a), we correlated focal OPC hyperdensity size and location with mouse motor performance.Nf1 OPC-Het mice with more focal OPC hyperdensities within the cingulum exhibit lower running speeds at the end of the complex wheel test (6.5 and 7 km in running distance) (Fig. 6b,c).This finding indicates that the degree of focal OPC hyperdensities in subcortical motor projections inversely correlates with motor learning, consistent with the result described above demonstrating that experience-dependent oligodendrogenesis is impaired in these focal OPC hyperdensities (Extended Data Fig. 7a).
Given that OPCs within focal OPC hyperdensities of Nf1 OPC-Het mice behave like Nf1-null OPCs, we evaluated the hypothesis that biallelic Nf1 inactivation in OPCs impairs motor learning.Baseline gait analyses demonstrated that Nf1 OPC-iKO mice do not differ from Nf1 WT mice with respect to swing speed, stride length or paw intensity (Extended Data Fig. 9c).During the complex wheel test, both Nf1 WT and Nf1 OPC-iKO mice increased their running velocity.However, Nf1 OPC-iKO mice exhibited slower velocities than Nf1 WT controls toward the end of the test (6.5-and7-km running distance; Fig. 6d,e), illustrating impaired motor learning concordant with the impaired experience-dependent oligodendrogenesis shown above (Fig. 2d,e).Taken together, these results establish that OPC-specific Nf1 biallelic inactivation leads to motor learning deficits.

Discussion
In this study, we demonstrate that heterozygous Nf1-mutant and Nf1-null OPCs lack appropriate proliferative responses to neuronal activity and Nf1-deficient OPCs exhibit impairments in experience-induced There is a negative correlation between the percent area covered by focal OPC hyperdensities and motor performance.Linear regression R 2 = 0.8174 (6.5 km) and 0.8761 (7 km).d, Toward the end of the complex wheel test (6.5-7km, e), Nf1 OPC-iKO mice (orange, N = 8) exhibited motor learning deficits relative to the Nf1 WT (magenta, N = 10) mice.e, The velocity of individual mice in d at 6.5 and 7 km.Two-way ANOVA with Bonferroni test.*P = 0.0332 (6.5 km) and 0.0372 (7 km).TAM, tamoxifen injection; CW, complex wheel.Data shown as mean ± s.e.m.; each point represents one mouse (c and e); two-sided (a, d and e).

Article
https://doi.org/10.1038/s41593-024-01654-yoligodendrogenesis.In the brains of OPC-specific heterozygous Nf1-mutant mice, we identified focal OPC hyperdensities where OPCs within these hyperdensities exhibit impaired experience-induced oligodendrogenesis.Moreover, Nf1 loss in OPCs alone is sufficient to cause motor learning deficits.Collectively, our findings provide a mechanistic connection between oligodendroglial plasticity and cognitive function in a neurogenetic disorder, raising several key points relevant to NF1 and OPC function.First, neuronal activity is known to regulate OPC proliferation, oligodendrogenesis and myelination, which in turn mediates optimal circuit dynamics and several domains of neurological function, including attention, motor function, motor learning, spatial learning and memory consolidation [6][7][8][9] .Our group has previously demonstrated that optogenetically induced neuronal activity in the motor planning region (premotor, M2) increases OPC proliferation and oligodendrogenesis, leading to increased myelination of the cingulum and corpus callosum and improved motor function 4 .Using a similar in vivo optogenetic paradigm herein, we found that neither heterozygous Nf1-mutant nor Nf1-null OPCs exhibit proper neuronal activity-induced proliferation, suggesting that Nf1 mutation disrupts the cellular/molecular mechanisms used by OPCs to sense and/or respond to activity-dependent signals.Since neuronal activity-regulated oligodendroglial responses support adaptive myelination to fine-tune circuit dynamics, the failure of Nf1-mutant OPCs to respond to neuronal activity implies that compromised oligodendroglial plasticity may partially contribute to the learning difficulties common in individuals with NF1.
In children with NF1, diffusion tensor imaging studies of white matter reveal differences in fractional anisotropy within the corpus callosum and cingulum compared to the control group; such differences were detected only during childhood (1-12 years of age) and not at adolescent ages 34 .These findings suggest delayed myelination during childhood that catches up later in adulthood.Supporting this hypothesis, we found reduced oligodendrocyte numbers in young (1-month-old) Nf1-mutant mice that normalize by adulthood.It is important to note that oligodendroglial cells also play myelin-independent roles, including axonal pruning 35 , synaptic pruning 36,37 , potassium buffering 38 and antigen presentation 39 ; Nf1 mutations in OPCs could also affect these noncanonical oligodendroglial functions.
In the experience-dependent motor learning paradigm (complex wheel test), OPC-specific heterozygous Nf1-mutant mice show normal motor learning, whereas OPC-specific Nf1-null mice exhibit impaired motor learning.One plausible explanation for the motor learning difference between the heterozygous Nf1-mutant and Nf1-null conditions is that the complex wheel test-induced neuronal activity in the motor cortex (for example, both M1 and M2) is sufficient to induce some adaptive oligodendroglial changes in heterozygous Nf1-mutant but not in Nf1-null oligodendroglial cells.Notably, these two different models display baseline differences in their OPC populations: heterozygous Nf1-mutant mice show comparable OPC density to WT controls outside of hyperdense foci, whereas Nf1-null mice exhibit higher OPC density globally throughout their brains.It should be noted that not all newly generated oligodendrocytes derive from new OPC proliferation but by direct OPC differentiation without prior proliferation 8 .Additionally, existing oligodendrocytes can remodel myelin (for example, myelin sheath length) in response to neuronal activity 40,41 , underscoring the numerous mechanisms by which neuronal activity can modulate myelination; we have not assessed such potential myelin remodeling aspects of myelin plasticity in this study.
Activity-regulated oligodendroglial responses contribute to several cognitive functions, many of which are impaired in people with NF1, including attention, learning and memory [6][7][8][9] .Our findings show that mice with Nf1-mutant OPCs exhibit impaired responses to neuronal activity and consequent defective motor learning.While prior cell type-specific genetic mouse modeling studies demonstrated that Nf1 loss in neurons causes spatial learning deficits 42 , it was unclear whether other neurological processes that oligodendroglial plasticity modulates (for example, attention, short-term memory and spatial learning) are similarly disrupted by Nf1 mutation in OPCs.In this regard, we observed that larger focal OPC hyperdensities within the cingulum, which is critical for motor cortex function, correlate with poorer motor learning.These findings suggest a model where several neuronal circuits may be partially disrupted by the focal OPC hyperdensities, creating a potential threshold for the development or progression of neurological deficits.
The elucidation of an oligodendroglial-based mechanism for cognitive impairment in NF1 also suggests new potential therapeutic avenues focused on restoring adaptive oligodendroglial responses.Coupled with previous work demonstrating that a TrkB agonist, which induces oligodendrogenesis, rescues chemotherapy-induced impairment in oligodendroglial plasticity and cognitive function 6 and that clemastine, which promotes oligodendrogenesis, rescues social isolation-induced deficits in adaptive myelination and social avoidance 43 , these findings support the concept that therapies that promote OPC differentiation could potentially be leveraged to treat cognitive deficits in individuals with NF1.
Second, the germline NF1 mutation (monoallelic inactivation) affects all cell types in the body of individuals with this neurogenetic disorder; however, subsequent second-hit events (for example, NF1 loss of heterozygosity) involving the one remaining functional NF1 allele (biallelic inactivation) in specific cell types contribute to many NF1 clinical manifestations.Depending on the cell types affected, monoallelic and biallelic NF1 inactivation can both be pathogenic.Using genetically engineered mice to model NF1, previous studies revealed that monoallelic inactivation of Nf1 in inhibitory neurons leads to deficits in spatial learning 42 .In contrast, the NF1-associated corpus callosum enlargement was observed only with biallelic, but not monoallelic, Nf1 inactivation in neural stem cells 44 .It is thus important to study the function of NF1 in a cell type-specific manner and to investigate the effects of both monoallelic and biallelic NF1 inactivation.In this study, we found that Nf1 OPC-iKO (biallelic NF1 loss) mice exhibit impaired motor learning, while motor learning performance in Nf1 OPC-Het (monoallelic NF1 loss) mice depends on the size and location of the focal OPC hyperdensities.Our findings also underscore the contribution of biallelic Nf1 inactivation to more severe CNS deficits associated with NF1.
The discrete appearance of focal OPC hyperdensities in heterozygous Nf1-mutant mouse brains is reminiscent of the T2 hyperintensities (focal areas of signal intensity, FASI) detected on magnetic resonance imaging of children with NF1 (ref.45).Prior histological analyses of the brains of three patients with FASI uncovered vacuolar changes suggestive of myelin disruption 46 .However, the cellular identity of FASI remains inconclusive so far.The OPC density and the experience-dependent oligodendrogenesis deficits of OPCs within focal OPC hyperdensities of heterozygous Nf1-mutant mice are similar to the behavior of Nf1-null OPCs, suggesting that OPCs within focal OPC hyperdensities probably exhibit Nf1 loss of heterozygosity.In support of this idea, OPC-specific Nf1 loss of heterozygosity induced in heterozygous Nf1-mutant mice by Cre-mediated chromosomal recombination (MADM) results in regional increases in OPC proliferation 22 .
Third, the finding that some, but not all, germline Nf1 mutations result in the formation of OPC hyperdensities suggests differential effects of the mutation on oligodendroglial lineage biology.In this respect, there are Nf1 mutation-specific effects, which are not accounted for by neurofibromin regulation of RAS, as all Nf1 mutations examined lead to increased RAS activity.In contrast, the germline Nf1 mutations associated with OPC hyperdensity formation result in increased PI3K/AKT signaling, suggesting additional functions of neurofibromin in maintaining PI3K/AKT homeostasis independent of RAS hyperactivation alone.Importantly, these observations dissociate OPC hyperdensity from optic glioma formation in that some strains that do not generate focal OPC abnormalities still form optic Article https://doi.org/10.1038/s41593-024-01654-ygliomas (R1278X) 47 .Future investigation will be required to define the mechanism by which germline Nf1 mutations differentially regulate PI3K/AKT activation in OPCs.
Last, NF1 is both a neurological disorder and a cancer predisposition syndrome.It is conceivable that the focal OPC hyperdensities represent preneoplastic regions at risk of transforming into gliomas.To this end, inactivating both Nf1 and Trp53 transforms OPCs into high-grade gliomas 14 .It is therefore possible that Nf1 inactivation primes OPCs for neoplastic transformation by increasing proliferation and decreasing oligodendrogenesis, while Trp53 inactivation is required to inhibit the senescence program in Nf1-null OPCs 22 and facilitate gliomagenesis.Given that adult patients with NF1 have a higher chance of developing high-grade gliomas than observed in the general population 48,49 , it is likely that these focal OPC hyperdensities serve as a preneoplastic pool of glioma-initiating cells that transform into glioma when mutations in other glioma driver genes (for example, Trp53) co-occur.

Conclusion
Oligodendroglial plasticity is critical for proper neurological function in the healthy brain, and we now demonstrate that adaptive OPC responses are disrupted by NF1 mutations in the neurogenetic disorder NF1, which impairs oligodendroglial dynamics and results in motor learning deficits.

Online content
Any methods, additional references, Nature Portfolio reporting summaries, source data, extended data, supplementary information, acknowledgements, peer review information; details of author contributions and competing interests; and statements of data and code availability are available at https://doi.org/10.1038/s41593-024-01654-y.
To measure Ttr area, the threshold was adjusted to Ttr fluorescence signal.To get consistent data, the same threshold value was used for all images.To quantify the number of reactive astrocytes, Cxcl10 and Sox9 double-positive cells were counted.Only the 4′,6-diamidino-2-phenylindole covered with over four dots of Cxcl10 and Sox9 signal was counted as a reactive astrocyte.For both Ttr and reactive astrocyte experiment, fewer than four Pdgfra + cells or more than six Pdgfra + cells in the region of interest were considered as nonhyperdensity or OPC hyperdensity areas, respectively.

Complex wheel test
Mice (15 weeks of age) were individually housed with the complex wheel with water (supplemented with 0.2 mg ml −1 EdU) and food.The training sessions for the complex wheel test are conducted over 7 days.The animals have free access to the wheel, water and food during the entire session.Complex wheels with dimensions previously described were constructed from laser-cut acrylic and assembled in laser-cut acrylic housings 7 .Wheels were mounted on stainless-steel axles with polytetrafluoroethylene bearings.Rung arrangements were configured as previously described 7 .To measure wheel speed, infrared distance sensors were mounted above each wheel to identify the movement of individual rungs across the sensor path during the wheel's rotation.Analog output from these sensors was logged along with timestamps using an Arduino.Up to six wheels were active simultaneously per experiment, and the speed of Nf1 WT mice does not differ among wheels, indicating similar performance of the wheels.Further hardware specifications are available upon request.
Logged data from the Arduino-based monitoring system were processed in Matlab.The rotation of individual wheel rungs across the sensor path (rung intercepts) was identified via Matlab's 'findpeaks' https://doi.org/10.1038/s41593-024-01654-yfunction.The data were then filtered to periods of active wheel rotation, defined as periods where rung intercepts occurred at less than 2-s intervals.One full revolution of each wheel was identified by a series of rung intercepts equal to the total number of rungs per wheel.The cumulative distance traveled by each mouse over the total number of revolutions in each experiment was calculated using the circumference of each wheel.Using a sliding window of size equal to the number of rungs per wheel, wheel speed was estimated by dividing the wheel circumference over the total time elapsed within a given sliding window.These values were then initially smoothed by moving average over 100-revolution intervals.Subsequently, average velocities were calculated and reported over a defined distance interval, such that mouse-to-mouse comparisons were made on the basis of total distance traveled.The code for complex wheel test analyses is available at Zenodo (https://doi.org/10.5281/zenodo.10864194).

Transmission electron microscopy
Mice were killed and perfused with PBS followed by the Karnovsky's fixative (2% glutaraldehyde and 4% paraformaldehyde in 0.1 M sodium cacodylate).The samples were kept in Karnovsky's fixative for more than 3 weeks, then post-fixed in 2% osmium tetroxide (EMS 19100) for 2 h at room temperature, washed three times with ultrafiltered water and then en bloc stained 1% uranyl acetate (EMS 541-09-3) overnight at 4 °C.Samples were then dehydrated in graded ethanol (30%, 50%, 75% and 95%) for 15 min each at 4 °C; the samples were then allowed to equilibrate to room temperature and were rinsed in 100% ethanol two times, followed by propylene oxide (EMS 20401) for 15 min.Samples were infiltrated with EMbed-812 resin (EMS 14120) mixed 1:1 with propylene oxide for 2 h followed by 2:1 EMbed-812/propylene oxide for 2 h.The samples were then placed into EMbed-812 for 2 h and then placed into TAAB capsules filled with fresh resin, which were then placed into a 65 °C oven overnight.Sections were taken at 80 nm on a Leica Ultracut S (Leica) and mounted on 100 mesh Cu grids (EMS FCF100-Cu).Grids were contrast stained for 30 s in 3.5% uranyl acetate in 50% acetone followed by staining in 0.2% lead citrate for 2 min.Samples were imaged using a JEOL JEM-1400 transmission electron microscope at 120 kV, and images were collected using a Gatan Orius digital camera.
The g-ratio, defined as the axonal diameter in its short axis divided by the diameter of the entire fiber in the same axis (axonal diameter/ axonal diameter + myelin sheath), was measured using ImageJ.A total of 54-209 myelinated axons were quantified for each animal from 2-10 ×6,000 transmission electron micrographs.The number of myelinated axons was quantified from 16-25 ×6,000 transmission electron micrographs.

Western blot
Snap-frozen whole mouse brains or iOPC pellets were lysed in 200 µl RIPA buffer (Fisher) supplemented with protease and phosphatase inhibitor cocktails (Cell Signaling Technology).Thirty micrograms of total protein of each sample was analyzed by western blot using antibodies, including rabbit anti-phospho-AKT T308 (Abcam ab38449, 1:500), rabbit anti-AKT (Cell Signaling Technology, 9272S, 1:1,000) and mouse anti-α-tubulin (Cell Signaling Technology, 3873S, 1:5,000).The membranes were imaged using a Li-Cor Odyssey Fc system and protein band intensities were analyzed using Li-Cor Image Studio Software (version 2.0).Relative expression of phospho-AKT T308 was calculated after normalization to total AKT levels and using α-tubulin as a loading control.A minimum of four independent brain samples or independently generated iOPC pellets were used for each genotype.

Stereology
White matter volume was measured with the Cavalieri method by marking grid points over an area of interest and calculating with the Cavalieri Estimator in Stereo Investigator (v2023.1.2),as described previously 58 .

CatWalk gait analysis
Mice (15 weeks of age) were tested on the CatWalk system (Noldus) before the complex wheel test.The test was performed as previously Extended Data Fig. 7 | Spatial transcriptomic analysis of focal OPC hyperdensities.(a) Immunohistochemistry revealed an increased percent of new OPCs (left, the number of EdU + /PDGFRα + cells divided by the number of EdU + cells) and decreased percent of new OLs (right, the number of EdU + /ASPA + cells divided by the number of EdU + cells) in the HD areas Nf1 OPC-Het mice (yellow, N = 5), relative to the non-HD areas of Nf1 OPC-Het mice (blue, N = 4) and Nf1 WT (white, N = 4) mice.*, P = 0.0255 (new OPC), 0.0272 (new OL).Brown-Forsythe ANOVA test with Dunnett's T3 multiple comparisons (F = 5.812 [new OPC], 7.001[new OL]).(b) Representative image of a 24w-old Nf1 +/-mouse brain used for spatial transcriptomics.Left, PDGFRα (green) immunohistochemistry revealed focal OPC hyperdensities within the corpus callosum (blue, inset; orange represents non-hyperdensity area).Right, areas with focal OPC hyperdensities (blue) and those without (orange) are marked in the Loupe Browser for regional differential gene expression analyses.Scale bar, 1 mm.(c) Regional differential gene expression analyses of brains from four 24w-old Nf1 +/-mice revealed four genes increased in the focal OPC hyperdensities relative to the areas lacking hyperdensities in the corpus callosum.The p-value was adjusted using Benjamini-Hochberg correction for multiple tests in the Loupe Browser 6.0.0.(d) Immunofluorescence of Nf1 +/-brains revealed increased ECRG4 and ENPP2 expression within corpus callosum focal OPC hyperdensities (HD) relative to areas lacking hyperdensities (non-HD).The area of ECRG4/ENPP2 signal (green) positively correlates with the area of PDGFRα signal (red, linear regression).To measure ECRG4, ENPP2 and PDGFRα area, the threshold was adjusted to cover the fluorescence signal.The same threshold value was used for all images.Four images were collected for each animal, representing focal OPC hyperdensities and areas lacking hyperdensities.Unpaired t test with Welch's correction.* P = 0.

Fig. 6 |
Fig.6| Nf1 inactivation in OPCs results in motor learning deficits.a, Nf1 OPC-Het (Nf1 +/fl ;Pdgfra::Cre) mice (green, N = 8) showed a similar motor skill learning curve as Nf1 WT mice (magenta, N= 8).Two-way ANOVA with Bonferroni test.Comparisons between Nf1 WT and Nf1 OPC-iHet mice at each distance are not significantly different (P > 0.05).b, The shaded area represents the cingulum analyzed for the percent focal OPC hyperdensity coverage in c. c, The velocity of Nf1 OPC-Het mice at the end of the complex wheel test (6.5 and 7 km) is plotted against the percent area of cingulum covered by focal OPC hyperdensities.