Huntingtin lowering impairs the maturation and synchronized synaptic activity of human cortical neuronal networks derived from induced pluripotent stem cells

Despite growing descriptions of wild-type Huntingtin (wt-HTT) roles in both adult brain function and, more recently


Introduction
Huntingtin (HTT) is a large, multifunctional scaffolding protein present in all cells throughout the body, from embryonic development to adulthood.It is particularly abundant in neurons and the testes (Mac-Donald et al., 1993;Marques and Humbert, 2013).In adult brain cells, HTT plays a crucial role in various cellular processes essential for neuronal survival, including transcription, DNA repair, mitophagy, autophagy, vesicular trafficking, endocytosis, and recycling (Saudou and Humbert, 2016).A growing body of research highlights the synaptic functions of HTT.This includes its roles in the recycling of synaptic vesicles through a clathrin-dependent process (Borgonovo et al., 2013), axonal transport of new synaptic vesicles to the synaptic knob (Bulgari et al., 2017;Gauthier et al., 2004b;Weiss and Troy Littleton, 2016), replenishing synaptic vesicles with neurotransmitters (McAdam et al., 2020), and removing cellular debris through autophagy (Ochaba et al., 2014;Rui et al., 2015) HTT also plays a role in spine morphology and functions (McKinstry et al., 2014;Wennagel et al., 2022).
HTT is primarily known for its polymorphic CAG repeat tract extension in its first exon.Expansion of this repeat causes the neurodegenerative disorder Huntington's disease (HD) (The Huntington's Disease Collaborative Research, The Huntington's Disease Collaborative Research, 1993).This mutation leads to the production of mutant HTT isoform (mut-HTT) with an abnormally long poly-glutamine stretch.HD is inherited in an autosomal dominant manner, suggesting that the HTT mutation confers toxic or dominant-negative gain-of-functions to mut-HTT, likely arising from misfolding and aberrant interactions with other cellular protein partners (Bates et al., 2015;Guo et al., 2018;Rubinsztein and Carmichael, 2003).The pathological hallmark of HD is the progressive and selective neurodegeneration of neurons, particularly affecting the medium spiny inhibitory projection neurons (SPNs) in the striatum and cortical projection neurons (Vonsattel et al., 1985;Hedreen et al., 1991).The mechanisms underlying the progressive neuronal dysfunction and death in these neurons are not fully understood but are believed to involve a cumulative impact of an extensive network of pathological molecular pathways (Ross and Tabrizi, 2011).
HD currently lacks a disease-modifying therapy, and treatment options focus on managing symptoms with limited effectiveness (Kumar et al., 2023).Studies in HD animal models demonstrate improvement of HD pathogenesis by HTT inactivation (Yamamoto et al., 2000) or reduction (Kaemmerer and Grondin, 2019).This leads to the exploration of various HTT-lowering strategies such as short hairpin RNAs (shRNAs) transduction via adenovirus-associated vectors (AAVs) and antisense oligonucleotides (ASOs) in HD patients (Caron et al., 2020;McColgan et al., 2023;Thomson et al., 2023).Most ongoing or recent clinical trials have focused on non-selective approaches that indiscriminately decrease both mut-HTT and wt-HTT.This raises concerns regarding the amount of wt-HTT necessary for normal striatal or cortical neuron development and function throughout adult life.Evidence suggest that wt-HTT loss in adult mice recapitulates HD-like features (Burrus et al., 2020), while wt-HTT depletion in HD mice worsens symptoms (Leavitt et al., 2006).Additionally, studies in developing mice and humans with wt-HTT loss-of-function mutations highlight its crucial role in neurogenesis and cortical development.Depletion of HTT at different stages of neural development in mice establishes the role of wt-HTT in neurogenesis and cortical development (Barnat et al., 2017;Dragatsis et al., 2000;Reiner et al., 2001).Conversely, loss-of-function compound mutations in wt-HTT alleles, resulting in a 90% reduction in HTT levels, cause significant neurodevelopmental defects in humans (Jung et al., 2021;Rodan et al., 2016).
In this study, we investigated the role of wild-type Huntingtin (wt-HTT) in maintaining healthy human cortical neurons.To do so, we took advantage of in vitro models of human cortical neuronal and neuro-glial networks derived from human induced pluripotent stem cells (hiPSCs).We examined how the loss of wt-HTT affects neuronal network formation, synaptic maturation, and homeostasis.Our findings demonstrate that wt-HTT in neurons is essential for the initial growth of neuritic arborization but not for its maintenance.In our in vitro model system, we thus established that wt-HTT plays a crucial role in the proper maturation of cortical neuronal networks in vitro, including the development of fully synchronized synaptic activity.Interestingly, we observed that the defects induced by the lowering of HTT in human cortical neurons are dose-dependent and cannot be compensated for by non-cell-autonomous astrocytic support.

Statistical analysis
All statistical analyses were performed using GraphPad Prism (GraphPad Software, Inc.) software.All experiments consisted of at least three independent replicates and were conducted blindly.The normality of the data distribution was tested by performing D'Agostino and Pearson test with threshold set at α = 0.05.When the n was too small to be analyzed by a normality test (e.g., Western blotting analyses), the normality was assumed.When comparing two groups, we used the unpaired two-tailed Student's t-test when the data were normally distributed or the Mann-Whitney test when the data were not normally distributed.When comparing multiple groups we used one-way analysis of variance (ANOVA) followed by Tukey's post hoc analysis when the data were normally distributed or Kruskal-Wallis test followed by a Dunn's post hoc analysis when the data were not normally distributed.When comparing multiple groups longitudinally, we used two-way analysis of variance (ANOVA) followed by Tukey's post hoc analysis; Data are expressed as means ± SEMStatistical significance: ns nonsignificant for P-value >0.05, *P-value ≤0.05, **P-value ≤0.01, *** Pvalue ≤0.001.

Human cortical neurons derived from iPSCs form neuronal networks displaying progressively synchronized synaptic activities
To investigate the role of HTT in the synapse maturation within human cortical neuronal networks in vitro, we generated cortical neuron progenitors from a human induced pluripotent stem cell (hiPSC) line derived from a healthy donor (GM25256).To direct the neuroepithelial cells towards a dorsal forebrain fate, we used two SMAD inhibitors to initiate neural induction and SHH inhibition, along with FGF2 and Wnt signals (Gribaudo et al., 2023).Over a 6-week period in vitro, we longitudinally studied the development of monolayered neuronal networks originating from these cortical precursors.Through phase-contrast images and immunostaining of neuronal and cortical markers, we tracked the progressive maturation of cortical progenitors into cortical networks (Fig. 1A-F, S1A & B).By week 5, >86% of these cell cultures are neuronal (86% HuC/D+, 97% MAP2+), with 60% of TBR1 cells and, with GABA+ inhibitory neurons constituting 10% of the cellular population (Fig. 1B&C).Between weeks 1 and 6, the maturation process involved a biphasic evolution of both the total number of neurons (Fig. 1D) and neurite morphology (Figs.1E&F).Initially, there was a decrease in the total neuron number (Fig. 1D) and in neurite branching per neuron (Fig. 1F) from week 3 to week 4 or week 1 to week 3, respectively.This was followed by a subsequent increase in the total number of neuron after week 4 and in neurite length per neuron and neurite branching per neuron after week 3.We proceeded with calciumimaging analyses to record the concurrent increase in synaptic activity, both at the individual neuron level and within the network.To track synaptic activities of individual neurons over time, we performed automated video microscopy weekly, examining several thousand neurons per cortical culture after exposure to Rhod2-AM, a red calcium chemical indicator (Fig. 1G&H).Quantification of active neurons per recording involved identifying the region of interest (ROI), approximately the size of a neuronal soma, displaying oscillatory variations in fluorescence intensity (Fig. 1G&I).While the total count of active neurons decreased by 20%, the proportion of active neurons increased from 46% at week 1 to over 58% beyond week 5 (Fig. 1I).Furthermore, the frequency and strength of observed calcium transients progressively increased over time, from <5 to >13 peaks per minute (Fig. 1J).Notably, synchronization of individual neuron activity, as quantified by the index of correlation (Ic) of ROI, significantly increased after week 4 (Fig. 1L).This progression delineates the transition from a network of human cortical neurons with unsynchronized, weak, and low frequency calcium transients before week 4 to a fully synchronized network of neurons with higher frequency and stronger calcium transients.
To evaluate the synchronized synaptic activity of entire cortical networks in our cultures, we used Cal-520, a green calcium chemical indicator, and examined the integration of calcium transients in neuronal cultures over a 10-min period.We analyzed recordings longitudinally from week 1 to week 6, extracting information on the frequency (Fig. 1N), amplitude (Fig. 1O), and variability of fluorescent intensity (Fig. 1P) in the cultures (Fig. 1M-P).As anticipated, we observed no oscillation in Cal-520 fluorescence in cortical populations matured 2 weeks or less, corresponding to the period with the lowest correlation index of these networks.Subsequently, beyond week 3, as the correlation index increased, we detected the integration of synchronized calcium transients, resulting in the modulation of fluorescent intensity with rhythmic oscillations (Fig. 1M).A longitudinal study of calcium oscillations revealed a twofold increase in frequency and a fourteen fold increase in amplitude between weeks 1 and 6 (Fig. 1N &  O).Conversely, the variability of these parameters, as indicated by the coefficient of variation (CV) of the time between two peaks (CV of peakto-peak time; CV P-P -Fig.1P) and the CV of the amplitude (CV AMP) of each oscillation (Fig. S2), decreased over time by 12% and 35%, respectively.These synchronized activities across the networks were sensitive to tetrodotoxin (TTX), a neuronal activity blocker, and demonstrated pharmacological sensitivity in both frequency and amplitude to GABAergic antagonists (bicuculline) as well as glutamatergic receptor antagonists (MK801, CNQX) (Fig. S3).Furthermore, the presence of human iPSC-derived astrocytes facilitated the morphological and functional differentiation of the iPSC-derived neurons in a dose-dependent manner (Fig. S.4).In summary, our experimental setup enables the establishment of synaptically active and coordinated networks consisting of a combination of excitatory and inhibitory human cortical neurons.

Loss of Huntingtin impairs dendritic arborization and blocks the synchronization of synaptic activity in human cortical neurons
Wild-type HTT and its interactome play crucial roles in the synaptic machinery governing exocytic neurotransmitter release (Barron et al., 2021).To investigate the potential impact of a non-allele-selective HTTlowering therapy on neurons of HD patients, we used our experimental setup to examine the effects of HTT loss on the formation and synaptic homeostasis of human cortical neuronal networks derived from human iPSCs.We aimed to deplete HTT protein by transducing young postmitotic cortical neurons with HTT-targeting (shHTT) shRNA lentiviruses four days after seeding the iPSC-derived neuron precursors (DIV4) (Fig. 2A).Western blot analysis confirmed a sustained reduction in the level of full-length HTT protein mediated by the shHTT viruses, reaching − 95% at week 5 post-exposure (day post exposure 31) (Fig. 2B).Given that HTT gene mutations disrupt axonal growth and branching in mice (Capizzi et al., 2022), we initially examined the impact of HTT loss on neuritic arborization of human iPSC-derived cortical neurons.We assessed MAP2-positive dendrites and GABA-positive neurite morphologies at week 5 (Fig. 2C-I).The number of MAP2-positive neurons in shHTT-treated cultures did not exhibit significant changes compared to shCTRL-treated ones (Fig. 2D).However, MAP2-positive dendritic arborization showed reductions in length and branching per neuron in shHTT-treated cultures (Fig. 2E&F).Similar findings were observed when analyzing GABA-positive neurites (Fig. 2 G-I), suggesting that the neuritic maturation of both excitatory and inhibitory neurons is affected by the reduction of HTT protein levels.
Considering that HTT plays a vital role in ensuring proper synaptic connectivity and function in striatal projection neurons in adult mice (Burrus et al., 2020), we proceeded to investigate how HTT loss affects the synaptic activity of individual human cortical neurons and the synchronized synaptic activity of the networks they establish in vitro.Calcium imaging recordings conducted at week 5, of individual neurons in both shHTT and shCTRL-transduced cortical cultures revealed that while the total number of active neurons decreased by only 13% in the absence of HTT (Fig. 2K), the frequency and burst strength of individual calcium oscillations were nearly halved in shHTT-treated cultures (Fig. 2L&M).Most notably, the synchronization of neurons was entirely disrupted in shHTT-treated cultures (Fig. 2N).Conversely, whole-well recordings of integrated synchronized calcium transients confirmed a complete cessation of network-wide, synchronized, and rhythmic oscillations in shHTT-treated cultures, while shCTRL-treated cultures remained unaffected (Fig. 2O-Q).
To corroborate these findings, we conducted a parallel series of experiments using antisense oligonucleotides (ASOs) to induce HTT reduction (Fig. S5).To achieve HTT-lowering levels comparable to those mediated by shRNA lentiviruses, we treated the cortical progenitors one day after seeding and every two weeks (DIV1, 14, 28) with 4 μM of HTTtargeting ASO (HTT_ASO) or non-targeting control ASO (CTRL_ASO).The HTT level in neuronal cultures treated with HTT-lowering ASO was reduced by 73% (Fig. S5B).Human cortical cultures treated with HTT_ASO displayed consistent outcomes across various aspects of our analyses compared to cultures treated with shHTT viruses.These similarities encompassed reduced MAP2 and GABA-positive neuritic arborization, a decreased frequency of individual calcium transients, and notably, a complete absence of network-wide synchronized calcium activity in HTT_ASO-treated cultures (Fig. S5 & S6).In summary, our findings indicate that the depletion of HTT in human iPSC-derived cortical neurons has negative effects on dendritic arborization and eliminates network-wide synchronized oscillatory activity within the neuronal network they establish.

Loss of synchronized synaptic activity of human cortical networks mediated by HTT loss is permanent and dose dependent
Our observation of impaired neuronal morphology and synaptic activity resulting from HTT loss in human cortical cultures may arise from either a delayed maturation of the neuronal network or a permanent intrinsic impairment of the neuronal cultures.To address this question, we conducted a longitudinal assessment of HTT reduction, starting from the earliest detectable activity and extending as long as the cultures remained healthy in vitro (from week 2 to week 6).Weekly monitoring of calcium transients in individual neuron synaptic activities revealed a lack of synchronization throughout the maturation process in cultures treated with shHTT viruses.In contrast, shCTRL treated cultures displayed a small but statistically significant increase in the synchronization index as early as week 3.This desynchronization in shHTT treated cultures persisted for at least six weeks (Fig. 3A-D & Fig. S7G-H).Conversely, we observed a reduction in the mean frequency (Fig. 3B) and correlation (Fig. 3D) of individual calcium transients in shHTT-treated neurons from week 5 onwards.Network-wide synchronized synaptic activity, assessed weekly, confirmed these observations (Fig. 3 E-G).ShHTT-treated cultures did not exhibit any oscillatory activity from week 1 to week 6, whereas shCTRL cultures displayed frequency and amplitude of oscillations with patterns similar to those observed in untreated cultures (Fig. 3F & G).These results were corroborated by analyzing human cortical neurons treated every two weeks (DIV1, 14, 28) with HTT-targeting ASOs (Fig. S7).Overall, our data suggest that the synaptic impairments we observed were unlikely to be caused by a delay in the maturation of the cortical precursors or the cortical neurons.
In the context of non-allele-selective therapies targeting the reduction of HTT levels, a major concern arises from potential cellular dysfunctions resulting from even a partial loss of wt-HTT functions.In our neuronal model, we addressed this concern by investigating synaptic activity of the network, aiming to determine the minimal reduction in HTT protein levels required to disrupt synchronization among cortical neurons.Unlike shHTT lentiviruses, HTT_ASOs offer a suitable approach for this investigation as they decrease concentrations uniformly in all cells, allowing for dose-dependent reduction of HTT in our neuronal cultures.We used a total-HTT quantification assay based on Homogeneous Time Resolved Fluorescence (HTRF) on week 5 cultures treated every two weeks with HTT_ASO at concentrations ranging from 0 to 10 μM.This assay demonstrates ample sensitivity to quantify soluble forms of HTT within the same microplate wells used for calcium imaging analyses.The dose-dependent decline in HTT levels at week 5 exhibited an IC50 of 0.03 μM (Fig. 4A).Correspondingly, calcium imaging recordings of individual cells revealed a dose-dependent decrease in all synaptic parameters, with IC50 values ranging from 0.6 μM to 0.1 μM (Fig. 4B-G).
Conversely, whole-well recordings of integrated synchronized calcium transients confirmed a HTT dose-dependent loss of network-wide rhythmic oscillations, with IC50 values for frequency and amplitude ranging from 0.4 μM to 1.3 μM respectively (Fig. 4I-J).Overall, our model intriguingly demonstrates that as HTT levels decrease, the activity of individual neurons gradually diminishes, as indicated by the number of active cells, the frequency and strength of calcium transients.Interestingly, while the decrease in the frequency of active neurons is progressive, the network synchronized activity remains unaffected in neurons exposed to HTT_ASO concentration (≥1 μM), causing more than two third of HTT loss but collapses beyond this concentration (Fig. 4D).

Astrocytes enhance human cortical network maturation but cannot rescue phenotypes mediated by HTT loss
Astrocytes play a crucial role in supporting neuronal maturation during brain development and in adult brain systems (Stevens, 2008).Previous studies have demonstrated that co-culturing human iPSCderived astrocytes with neurons enhance the maturation of the latter (Kuijlaars et al., 2016).We sought to investigate whether iPSC-derived astrocytes could mitigate the phenotypic effects resulting from HTT neuronal, but not astroglial, loss in human neuron-glia networks.We derived human astrocytes from the GM25256 iPSC line using a protocol adapted from Lundin et al. (Lundin et al., 2018b).This protocol avoids the use of cytokines and supplement such as CNTF and serum that induce astrocytic reactivity.To specifically examine the impact of neuronal HTT loss, we applied HTT-lowering viruses to cortical neuron precursors and introduced human astrocytes only at DIV7 after removal of any remaining infecting particles (Fig. 5A).
Although cultures exclusively derived from cortical neuron precursors yielded <1.55% GFAP+ astrocytes, introducing iPSC-derived astrocytes in a 1:3 astrocyte/neuron ratio resulted in a neuronastrocyte co-culture containing 26% astrocytes (Fig. S4C&D).In this co-culture setup, we confirmed the pro-maturation activity of astrocytes on the morphology and synaptic activity of cortical neurons, observing a significant increase in neurite length and branching, along with increased synchronized network-wide activity (Fig. S4).We initially evaluated the impact of neuronal HTT loss on the neuritic arborization of neurons in the presence of astrocytes by analyzing MAP2-positive dendritic neurite morphologies at week 5 (Fig. 5B-E).Unlike in pure neuronal culture, the number of MAP2-positive neurons in shHTTtreated neuroglial co-cultures was 2-fold higher than in shCTRL-treated co-cultures (Fig. 5C).Conversely, shHTT-treated co-cultures exhibited less complex dendritic arborization both in length and branching per neuron (Fig. 5D & 5E).
We proceeded to evaluate the impact of neuronal HTT loss on both the synaptic activity of individual cortical neurons and the synchronized synaptic activity of astrocyte-neuron networks.Overall, the effects of neuronal HTT loss on astrocyte-neuron co-cultures mirrored those observed in neuronal cultures following HTT loss.This included a reduction in the total number of active neurons (Fig. 5G), alongside reductions in the frequency and burst strength, intensity and duration of individual calcium oscillations (Fig. 5H & Fig. S8).Within neuronastrocyte co-cultures, synchronization of neuronal activity was completely abolished in shHTT-treated neurons, as indicated by a low correlation index value (Fig. 5J).Consequently, whole-well recordings of integrated synchronized calcium transients demonstrated a complete cessation of network-wide synchronized rhythmic oscillations in shHTTtreated co-cultures, while shCTRL-treated co-cultures remained unaffected (Fig. 5L&M, respectively).These findings imply that human astrocytes are incapable of mitigating the intrinsic impairments caused by neuronal HTT loss in human cortical neurons.

Loss of Huntingtin disrupts the synaptic coordination of human cortical networks
Since HTT-lowering therapies target adult patients, we aimed to explore the impact of HTT loss in human cortical neurons integrated in active and synchronized neuronal networks.To achieve this, we replicated our HTT-lowering experiments exposing cortical neurons only at week 3 (Fig. 6A).Western blot analysis confirmed a 45% reduction in full-length HTT protein levels fourteen days after exposure (week 5, dpe14) to HTT_ASO (Fig. 6B).Unlike the morphological changes observed when reducing HTT during earlier stages of neuronal maturation (Figs. 2 & 3), lowering HTT in more mature networks did not affect the neuritic arborization of human iPSC-derived neurons.Specifically, both the length and branching of MAP2-positive dendrites and GABA-positive neurites remained unchanged in neuronal cultures treated with HTT_ASO only at week 3 (Fig. 6C-J).
We next explored the effects of this delayed HTT loss on the synaptic activity of individual mature neurons and on the network-wide synchronization of these activities.We evidenced the alterations of synaptic activity of individual neurons and of their synchronization caused by HTT-loss only after week 3 mirrored those observed when it was triggered during the first week of neuronal cultures (Fig. 6K).These changes encompassed a decrease in the total number of active neurons (Fig. 6L), as well as in the frequency and burst strength of individual calcium oscillations (Fig. 6M&N).The synchronization of neuronal activities was significantly reduced although not entirely abolished, in cultures treated with HTT_ASO at week 3 (Fig. 6O).This is likely caused by the more limited extent of HTT lowering achieved when neurons are treated at week 3 than during the first week of culture.Whole-well recordings of integrated synchronized calcium transients corroborated the loss of network-wide, synchronized, and rhythmic oscillations in shHTT-treated cultures, while shCTRL-treated cultures remained unaffected (Fig. 6P-R).Similar finding were obtained when lowering HTT levels with shHTT lentiviruses (Fig. S.9).

Proteomic signature of wt-HTT loss in human cortical neurons
To further explore the implications of HTT reduction induced by shHTT and HTT_ASO, we compared the proteome of cortical neuron cultures derived from wt-iPSCs.These cultures were exposed either to shHTT or shCTRL lentiviruses at DIV 4, or to 4 μM of either CTRL_ASO or HTT_ASO at DIV 1, 14, and 28.Protein extracts were collected at week 5 (DIV35) and analyzed by mass spectrometry to identify differentially expressed proteins (DEPs).The mass spectrometry analysis revealed distinct proteomic profiles for each treatment group (shHTT, shCTRL, HTT_ASO and CTRL_ASO) (Fig. 7A).Label-free quantification of HTT peptides confirmed a significant reduction in HTT protein levels compared to control, falling below the detection threshold, in all or all but one shHTT and HTT_ASO samples, respectively.We identified DEPs through pairwise comparisons between shRNA and ASO-treated samples using a threshold of adjusted P-value ≤0.05 and an absolute fold change ≥1.25 (Fig. 7C-E; Table S1).We found 133 overlapping proteins in the shRNA and ASO treatment lists, with 83 proteins displaying upregulation and 50 downregulation in both HTT-lowering conditions.Gene set enrichment analysis (GSEA) of these 133 commonly identified DEPs highlighted significant enrichment for Gene Ontology (GO) terms linked to neuronal projections, axons and axonogenesis (Fig. 7F-H) (Chen et al., 2013;Kuleshov et al., 2016;Xie et al., 2021).This finding is consistent with the observed changes in neuronal morphology observed through MAP2 and GABA immunostaining, suggesting potential alterations in neuronal connectivity.Interestingly, while GSEA of only the 83 upregulated DEPs confirmed an enrichments of those DEPs for GO terms linked to neuronal projections, axons and axonogenesis, the GSEA of only the 50 downregulated DEPs additionally revealed a significant enrichment for GO terms linked to axonal transport (Fig. S10).Several of the most differentially expressed proteins we identified are involved in neurotransmitter release at the synapse (CADPS, CASKIN1, SYT4, SYT11: (Kabachinski et al., 2016;Hsueh, 2006;DeBello et al., 1993)) or have been linked to excitatory /inhibitory balance, a key determinant of network synchronization (UBR-1, SYNGAP1: (Ozkan et al., 2014, Li et al., 2023, 1)) (Fig. S11).Furthermore, GSEA against public gene expression databases (GEO) revealed an enrichment of the 133 DEPs in genes that are differentially expressed in the cortex of HD patients and the striatum of R6/1 HD mice.The highest enrichment score was observed for genes differentially expressed in mouse primary cortical neurons treated with topotecan, a topoisomerase 1 inhibitor.Topotecan reduces the expression of very long genes in mouse cortical neurons (King et al., 2013)  HD (Shekhar et al., 2017).The proteomic signature of wt-HTT loss in our human neurons are consistent with the observed morphological changes and suggests an indirect impact on synaptic function.Similar to numerous studies conducted in mice, our proteomic study supports the notion of a loss-of-function component in HD pathogenesis.

Discussion
Huntington's disease is a genetic neurodegenerative disorder for which several gene and drug therapies targeting mutant-HTT are under exploration (Bhat et al., 2023;Kumar et al., 2023).Although therapies aiming to lower both wild-type and mutant-HTT levels indiscriminately have recently been tested in HD patients and are still under investigation (Estevez-Fraga et al., 2023), the role of wt-HTT in maintaining neuronal health in adults and during development remains poorly understood.In this study, we demonstrate the value of cortical neuronal networks derived from human pluripotent stem cells in addressing this question.Our research provides evidence that the loss of the wt-HTT isoform in otherwise healthy neurons disrupts the maturation of neuritic arborization and impedes the establishment of network-wide synchronized synaptic activity.Notably, we observe that this latter alteration depends on the dosage of wt-HTT in our in vitro model.Therefore, our findings suggest that therapies aimed at indiscriminately lowering mutant and wt-HTT isoforms levels might compromise the health of targeted neurons and potentially exacerbate the contribution of loss-of-function mechanisms to neuronal pathology in HD patients.
The coordinated activity of neuronal networks, characterized by synchronized and oscillatory synaptic firing, is a defining feature of many neuronal systems, including those within the cortex (Gansel, 2022;Uhlhaas et al., 2009).This coordinated firing plays a critical role in human brain development, particularly in the formation of the adultlike six-layered cortex (Molnár et al., 2020).This synchronized firing emerges at the cellular level, through the dynamic interplay of recurrent excitatory and inhibitory connections between co-active neurons (Gansel, 2022).Both rodent primary cortical neurons and human iPSCderived cortical neurons are capable of forming networks exhibiting such recurrent excitatory and inhibitory connections.These networks develop synchronized and oscillatory activities (Bodai and Marsh, 2012;Cornelissen et al., 2013;Gribaudo et al., 2019;Kuijlaars et al., 2016;Odawara et al., 2016;Verstraelen et al., 2014;Woodruff et al., 2020).Recording synchronized network-wide calcium transients provides a powerful tool for assessing neuronal health and maturation.This technique is sensitive and can detect responses to pharmacological synaptic challenges, as well as subtle changes in neuronal physiology, such as those induced by the accumulation of phosphorylated alpha-synuclein fibrils (Gribaudo et al., 2019).To investigate how the loss of wt-HTT affects synaptic activity and the overall health of human neurons, we developed a high-throughput, high-content microplate-based in vitro model that replicates human cortical network maturation and synaptic activity.This model takes advantage of cryopreserved human cortical neuron progenitors cultured either alone or with astrocytes, both derived from wild-type hiPSCs.Our findings confirm previous reports, demonstrating progressive morphological and functional differentiation of iPSC-derived neurons (Odawara et al., 2016).Calcium level recordings revealed a progressive increase in spontaneous synaptic activity within a growing proportion of neurons over time.Individual neurons exhibited increasingly faster and stronger calcium oscillations that synchronized across the network as development progressed.These network-wide synchronized activities were sensitive to pharmacological synaptic challenges, such as TTX or antagonists of GABA or glutamate receptors.Furthermore, the presence of human iPSC-derived astrocytes enhanced both the morphological and functional differentiation of the iPSC-derived neurons (Kuijlaars et al., 2016).Overall, these features support the reliability of this cellular model for exploring the role of HTT at the human cortical synapse.
Huntingtin is a protein essential for normal brain development and function.Loss of Htt disrupts the differentiation of mouse neuroblasts into mature neurons across various brain regions, including the striatum, cortex and thalamus (Reiner et al., 2001).This disruption involves alterations in cellular polarity and division orientation, ultimately affecting the transition from multipolar to bipolar morphology (Godin et al., 2010).Later in development, Htt loss alters the morphology of cortical neurons, leading to shorter and less complex dendrites (Barnat et al., 2017).In addition, Htt depletion in mouse cortical neurons disrupts the shape and function of dendritic spines (McKinstry et al., 2014) through a cofilin-mediated control of the actin cytoskeleton (Wennagel et al., 2022).Htt also influences the shape and structure of nuclei in striatal neurons (Burrus et al., 2020).Similarly, mutations causing HD negatively affect neuronal morphology.In particular, the microtubules in axonal growth cones are disrupted due to the downregulation of NUMA1.As a consequence, Htt loss limit axonal growth during mouse development (Capizzi et al., 2022).Fewer studies have explored the effects of HTT loss in human neurons.Wild-type HTT depletion affects spindle orientation and disrupts self-organization properties of human telencephalic neural or neuroepithelial cells derived from pluripotent stem cells (Lopes et al., 2016;Louessard et al., 2024;Ruzo et al., 2018).The most compelling evidence for HTT crucial role in brain development comes from rare cases of individuals with hypomorphic wt-HTT.While mild reductions of wt-HTT (15 to 50%) are not associated with developmental problems or adult phenotypes (Ambrose et al., 1994;Jung et al., 2021), severe reductions, below approximately 10% of normal due to compound heterozygous mutations on each HTT allele, result in a severe neurodevelopmental disorder (LOMARS) distinct from HD (Jung et al., 2021;Rodan et al., 2016).Interestingly, these authors found a lower-than-expected frequency of such damaging loss-of-function mutations in the HTT gene in the general population (Genome Aggregation Database) (Jung et al., 2021).Our own findings support a developmental role of wt-HTT in neurons.We observed morphological alterations in both excitatory and inhibitory neurons matured with reduced HTT levels.These alterations affected the extension and branching of neurites.Conversely, our proteomic analyses of these cells revealed significant changes in levels of proteins involved in axonal growth, (E) MAP2+ branch point total count per neuron, n = 31-35, unpaired, two-tailed, Student t-test, C: t = 6.123, df = 61, p < 0.001; D: t = 11.18,df = 65, p < 0.001; E: Mann-Whitney test, unpaired, two-tailed, U = 87, p < 0.001.(F-J) HTT-dependent properties of calcium transients in week 5 human cortical neurons transduced with shCTRL or shHTT lentiviruses and subsequently co-cultured with astrocytes, measured by video microscopy: (F) Variation of Rhod-2 fluorescence intensity of each ROI at week 5 displaying calcium oscillations.Comparative quantification of (G) the number of active neurons per field, (H) the mean number of calcium transients per minute, (I) the mean strength of transients, and (J) the correlation index of transients of all active neurons, n = 14-16, unpaired, two-tailed, Student t-test, G: t = 3.886, df = 28, p < 0.001; H: t = 4.010, df = 28, p < 0.001; I: t = 10.39,df = 28, p < 0.001; J: t = 49.51,df = 28, p < 0.001.(K-M) HTT-dependent properties of synchronized calcium transients in human cortical neuronal networks transduced with shCTRL or shHTT lentiviruses at DIV4, co-cultured with astrocytes at DIV7 and recorded at week 5 by whole well recording of Cal-520 calcium indicator fluorescence measured by a kinetic plate reader: (K) Representative 60-s traces from the 10min recordings of calcium intensity oscillations in wells, (L) Peaks per minute of network oscillations, (M) Mean amplitude of peaks of network oscillations, n = 23-27, n = 23 Mann-Whitney test, unpaired, two-tailed, L: U = 0, p < 0.001; M: U = 0, p < 0.01.For all panels, mean and SEM are shown, when applicable, data are normalized to median of shCTRL-treated samples, n indicates the number of cultures per condition in at least three independent experiments, *** P-value <0.001.axonal guidance, cell adhesion and vesicular trafficking, although we did not detect change in NUMA1 levels.This suggests a potential mechanism for HTT influence on neuronal morphology, involving semaphorin (SEMA6A/D), their receptors (PLXNA4), ephrin signaling (EPHA4) and vesicular trafficking (KALRN).Interestingly, SEMA6D and PLXNA4 proteins are differentially expressed in HD-iPSC-derived striatal neurons (Tshilenge et al., 2023).

(caption on next page)
The importance of HTT in synapse formation and function becomes evident through the synaptic effects observed upon HTT depletion in mouse neurons.In the developing mouse cortex, wt-Htt loss accelerates the formation and maturation of excitatory synapses, while creating abnormally shaped dendritic spines.This combination increases excitatory connections between cortical and striatal neurons (McKinstry et al., 2014).In adult striatal neurons of the indirect pathway, Htt depletion reduces the number of inhibitory synapses by half in the globus pallidus.Conversely, it increases the number of synapses in this region when targeting Htt loss in the striatal neurons of the direct pathway (Burrus et al., 2020).Overall, HTT is involved in various functions crucial to synapse formation, function and homeostasis (Barron et al., 2021).This includes the regulation of expression, posttranslational modifications, transport, distribution, as well as endoand exocytosis of pre-and post-synaptic proteins and receptors, including BDNF (Gauthier et al., 2004a), its receptor TrkB (NTRK2) (Liot et al., 2013), AMPA and NMDA glutamate receptors (GRIA1 and GRIA2) (Huang et al., 2011;Wennagel et al., 2022), GRIN2B (Kang et al., 2019), PSD95 (DLG4) (Parsons et al., 2014), NFkB (NFKB1) (Marcora and Kennedy, 2010) and Synaptotagmin (SYT2) (Culver et al., 2012).Here, we show that HTT depletion in human cortical neurons alters the protein level of many of these proteins including BDNF, GRIA1, GRIA2, PSD95, SYT11 and SYT4 under at least one of our HTT-lowering conditions.We propose that these changes underlie the reduced synaptic activity of individual neurons and the disruption of synchronized activity within neuronal networks following HTT depletion.
A critical question regarding non-selective HTT-lowering therapies for HD revolves around determining the threshold level of wt-HTT loss that brain cells can endure without compromising their health and network function.The extent of wt-HTT reduction is contingent upon the selectivity of the drug for the mutant allele, as well as the actual dosage received by the neuron.Various HTT-lowering strategies tested in HD rodent models have reported widely differing reductions in Htt levels in the cortex and/or striatum of mice, ranging from approximately 30% to 90%.These strategies encompass treatments with different HTTlowering agents such as AAV5-miHTT in Hu128/21 (Caron et al., 2020), or Q175 mice (Thomson et al., 2023), HTT-targeting ASOs (Kordasiewicz et al., 2012) or the RNA-splicing modulator Branapalm in BacHD mice (Liu et al., 2023).Results from the GENERATION-HD1 trial (NCT03761849) (McColgan et al., 2023) reported mutant-HTT reductions in cerebrospinal fluid ranging from approximately 30% to 50%, depending on the regimen of Tominersen injection.Since Tominersen does not discriminate between wt-HTT and mutant-HTT pre-mRNA, it is likely that both wt-HTT and total-HTT levels are reduced by a similar percentage in the cerebrospinal fluid.
In our study, we applied non-allele-selective HTT-targeting ASO (HTT_ASO) to our in vitro model of human cortical neuronal networks to achieve dose-dependent reductions of wt-HTT in neurons.This methodology enabled us to establish a correlation between the remaining wt-HTT levels in neurons and alterations in synaptic activity at both the individual neuron and network levels.Interestingly, while both the average frequency of calcium transients in individual neurons and network-wide oscillations decreased proportionally with HTT reduction, the loss of correlation between individual neurons shifted towards higher levels of HTT reduction.Specifically, despite a 40% reduction in wt-HTT already leading to a significant decrease in the frequency of neuronal calcium transients in our in vitro model, synchronized activity between neurons remained unaffected.We observed only in conditions causing HTT lowering by more than two-thirds a collapse in synchronized synaptic activity within the neuronal networks.
While the precise levels of remaining wt-HTT and total HTT in striatal or cortical neurons of patients treated with the highest Tominersen dose are currently unknown, our data suggest a potential risk of nearing the threshold for synaptic impairment observed in our in vitro model.This study highlights the potential risks associated with excessive loss of wt-HTT during HTT-lowering therapies for HD.Non-selective approaches targeting both mutant and wt-HTT isoforms may inadvertently disrupt healthy neuronal network activity.Our findings emphasize the importance of meticulous titrating HTT-lowering therapies in neurons to minimize the impact of wt-HTT loss-of-function.

Declaration of competing interest
None to declare.

Fig. 6 .
Fig. 6.Loss of Huntingtin in already matured neuronal cultures impairs dendritic arborization and blocks the synchronization of synaptic activity in human cortical networks.(A) Schema of ASO treatment and maturation of human cortical neuronal networks derived from iPSCs in vitro, treated with CTRL_ASO or HTT_ASO 3 weeks (DIV21) post seeding.(B) Western blot and quantification of HTT protein levels normalized to βIII-tubulin levels in week 5 neuronal cultures n = 5, unpaired, two-tailed, Student t-test t = 11.63,df = 8, p < 0.001.(C & I): Representative immunostaining at week 5 of neuronal soma and neurites (MAP2, green or GABA, yellow), counterstained with DAPI to highlight nuclei, Scale bar: 100 μm.(D) Quantification of MAP2+ cells, (E) MAP2+ total neurite length per neuron, and (F) 001. (O-Q) HTT-dependent properties of synchronized calcium transients in human cortical neuronal networks treated with CTRL_ASOs or HTT_ASOs at DIV21 and recorded at week 5 by whole well recording of Cal-520 calcium indicator fluorescence measured by a kinetic plate reader: (O) Representative 60-s traces from the 10-min recordings of calcium intensity oscillations in wells with human cortical neuronal networks treated with CTRL_ASO or HTT_ASO.(P) Peaks per minute of network oscillations, (Q) Mean amplitude of peaks of network oscillations, P: n = 71, Mann-Whitney test, unpaired, two-tailed, K: U = 29, p < 0.001; Q: n = 71 unpaired, two-tailed, Student t-test.t= 27.49,df = 140, p < 0.001.For all panels, mean and SEM are shown, data are normalized to median of CTRL_ASO-treated samples, n indicates the number of cultures per condition in at least three independent experiments, ** P-value <0.01; *** P-value <0.001, ns: not significant.(For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)Investigation, Formal analysis, Data curation, Writingreview & editing.Morgane Louessard: Writingoriginal draft, Resources, Methodology.Noëlle Dufour: Writingoriginal draft, Resources.Chloé Baroin: Formal analysis, Investigation, Writingreview & editing.Aurore de la Fouchardière: Writingoriginal draft, Software, Formal analysis.Laurent Cotter: Writingoriginal draft, Methodology, Conceptualization.Hélène Jean-Jacques: Writingoriginal draft, Methodology, Investigation.Virginie Redeker: Writingoriginal draft, Software, Resources, Methodology, Investigation, Formal analysis, Data curation, Writingreview & editing.Anselme L. Perrier: Writingoriginal draft, Visualization, Validation, Supervision, Software, Resources, Project administration, Methodology, Investigation, Funding acquisition, Formal analysis, Data curation, Conceptualization, Writing review & editing.